Windows Are Absurd

[Note that this article is a transcript of the video embedded above.]

Have you ever thought about windows? Like really considered the idea? If we hadn’t been exposed to them our entire lives, it might seem like kind of a bizarre concept. On the face of it, windows are about light. But it’s more esoteric than that, especially now that we can make our own light. We want to be inside, but not all the way inside. So we put holes in the wall. It’s a heck of a trade-off: you get light and a view of the outdoors, but it comes at a cost. Not just the cost of the glass itself, but also the structural changes they require and their effects on ventilation, sound, cooling, heating, safety, and security inside a building.

As simple as they seem, windows are a technology of sorts. And like a lot of building science, we’ve slowly been making progress, readjusting those tradeoffs and giving us the benefits without so many of the costs. I love to pick a relatively innocuous part of the built environment and really take a close look. So, I’ve built a few demonstrations in the garage to show you how far the science of glazing has come. I’m Grady, and this is Practical Engineering.

Glass is not a good substitute for an exterior wall, in many respects. If you take some wall away and replace it with a window, you’re not trading like-for-like. For one, you never accidentally throw a baseball through a brick wall. But it’s especially true when it comes to heat. And this is a pretty big deal because we collectively spend vast sums of money, resources, effort, and energy on maintaining a comfortable thermal climate inside our buildings. Almost all the time, in almost all the places, the temperature outside of a building is different from the temperature we want it to be inside. But the laws of physics (also all the time in all the places) are constantly working to equalize those temperatures.

We usually categorize the flow of heat into three separate modes. Conduction is heat moving through solid objects, atoms or molecules that are tightly packed, transmitting energy to their neighbors directly by vibrating against each other. Convection is heat transfer via the movement of a fluid like air or water. Radiation is heat transfer through electromagnetic waves. Since it is literally light, radiation doesn’t need a medium. All three play a role in the walls of buildings. And of course, it works both ways: sometimes it’s too hot outside and you want to keep that heat out. Sometimes it’s too cold and you want to keep the heat in. Any movement of heat from inside to out or vice versa is a flow of power, just like we measure with electricity. If you have 500 watts of heat flowing out of your house, you essentially need a 500 watt heater inside to make up for the loss and maintain a steady temperature. And just like voltage is a difference in potential that drives electrical current, heat flow is driven by a difference in temperature. The bigger that difference, the greater the flow of heat through a given wall.

Continuing the electrical metaphor, insulation is the generic term we use to describe materials that resist the transmission of heat from one side to the other. It makes sense that we put it in our walls. In the US, we use R-value as a measure of this resistance. This is kind of a cursed unit of measurement, so I’m going to gloss over the scientific definition. But, a typical exterior wall assembly will usually have an R value between 10 and 25, depending on climate and how new the building is. A simple pane of glass has an R value of about 1. Let me show that to you.

I built this box of foam insulation in my garage. Inside, I have a 25 watt ceramic heat emitter. I can close this up and wait for the temperature inside to stabilize. When that happens, we can assume that the heat going into the box is the same as the heat flowing out.

Okay, this has pretty much settled out, and you can see that the difference between the interior and exterior probes is about 43 degrees Fahrenheit, I think that’s around 24 celsius. This is R-5 foam, soif you do the math for a perfectly sealed box with these dimensions that should be right around 50 F, 28 C, um, so accounting for the imperfections in my setup, like this, uh, terrible tape seal, that’s actually pretty close. Um, so now I’m going to swap this lid out with a pane of glass, and we’ll set this going again.

Okay, this has settled out. And you can see that the temperature inside the shop hasn’t really changed, but the temperature inside the box has gone down a lot compared to the foam lid. The difference is about 22 F, which I think is around 13 C. So, just replacing one side of the box with glass instead of foam, it takes a much smaller difference in temperature to move that same 25 watts out of the box.

I’m sure you see the issue here. If we want windows in our buildings, it comes at a cost of thermal efficiency. For every square foot or square meter of exterior wall that we swap with single pane windows, heat will move through that space roughly 10 to 25 times faster. So, we rarely use single-pane glazing anymore. Almost all modern windows use two, and sometimes even three panes of glass with a sealed air gap between them. Sometimes this gap is filled with a heavy gas like argon or krypton that has lower conduction and convection. There are even vacuum insulated panels that have no air inside, eliminating convection entirely. It seems like a simple change, but it makes a pretty big difference.

I happen to have a double-pane glass unit, so we can try the experiment again.

Okay, this demo has settled out now, and you can see that the temperature difference is right at 30 F, 17 C. Remember with the foam lid it was about 43 degrees and with the single pane it was around 23 degrees. So this is basically right in the middle. It’s not as good as when there was no window, but it’s a lot better than the single pane of glass.

This varies a lot with details of the frame and coatings. But very generally, modern double-pane windows can double to quadruple the insulative value of a single pane of glass, and that can make a huge difference in a building with a lot of them.

This idea of a sealed air gap between panes comes with one problem. If you’ve ever brought a bag of chips on an airplane, you know what I’m talking about. If the glass manufacturer is at a low elevation and the window’s going to be installed somewhere much higher, you end up with an overpressure situation between the panes compared to the barometric pressure. That can cause them to bow out and even fracture. In some cases, manufacturers will actually dial in the pressure for the altitude of the final destination. Another option is to install a capillary tube that equalizes the pressure inside and out. These are very narrow diameter and long tubes meant to reduce the amount of moisture that diffuses between the panes, which would otherwise lead to condensation inside the unit.

Insulating glass panels help reduce heat conduction and convection, but that’s not the only way heat gets into and out of a building. The whole point of a window is to let in light, and we already learned that heat can move as electromagnetic waves. Of course, if we block visible light through a window to avoid heat transfer, that kind of defeats the purpose. But radiation also happens at other wavelengths on the electromagnetic spectrum. Specifically, for the temperatures we deal with on a day-to-day basis, most of the thermal radiation happens in infrared: wavelengths longer than visible light but shorter than microwaves. Even within that range, there are big differences in how the waves interact with matter. That’s particularly true for windows.

Typical stuff at ambient temperatures like a sun-baked driveway or human body, emit thermal radiation mostly in the far-infrared spectrum. This is the stuff you can see on a normal infrared camera. And even though glass is designed to let visible light through, it is almost completely opaque to longwave infrared light. Try looking through a window with an infrared camera and you just see the window itself, not anything beyond.

It’s cool to think that if our eyes worked a bit differently - if they were tuned to a slightly different part of the electromagnetic spectrum - a glass window would be entirely useless. And it seems like it would be a good thing for our thermal problem, preventing heat from transferring through. Unfortunately it’s not that simple.

Glass isn’t just opaque to far-IR; it’s also highly emissive. That means it mostly absorbs thermal radiation rather than reflecting it. So, the radiation doesn’t come directly through, but it does warm the glass, which can then just conduct through the pane and reradiate it where you don’t want it to go, whether inside or out. It’s basically radiation with extra steps.

On the shorter wavelength part of the spectrum we call near IR, it’s a little different. The main source of near IR radiation is the sun, and like visible light, that radiation can pass directly through most glass. This is how greenhouses work. The shortwave radiation passes right through where it can warm the stuff inside. But the stuff inside radiates longwave IR, which can’t pass back out through the glass. So, windows are kind of a two-way half-door for far infrared (since they absorb and reradiate it in both directions) and a one-way door for near infrared, but we can do better than that, mostly through low-emissivity or low-e coatings.

These coatings essentially convert a pane of glass into a mirror for IR. And you can see the difference really well with an IR camera. You still get some reflections with an uncoated pane, but the difference is obvious when you put a coated pane next to an uncoated one. But this gets a little tricky, especially when it comes to near IR. If you live in a cold climate and your main use of energy is heating, you’re happy to invite as much sunlight as you can get inside. You don’t want to reflect the near IR. The opposite is true in warm climates. Fortunately, manufacturers can tune the coatings according to the climate. You can reflect shortwave radiation if you’re worried most about cooling, or you can let it in to gather more heat inside. Window manufacturers also sometimes vary which face is coated. In warmer climates, you can coat the inside of the outer pane, called “surface 2,” reflecting the IR before it crosses the gap to keep it from warming the air between the panes. In cooler climates, you can put it on “surface 3” to reflect the long-waveshort-wave IR back into your house without it having to cross the air gap.

Making windows more thermally efficient isn’t the only use for glass coatings. If you’re like me, there’s a good chance you have the Merlin app installed on your phone. I think it’s a bit of an inevitability that once you reach a certain age, you start getting interested in birds. Unfortunately, one of the largest human-caused sources of bird mortality is actually windows. At certain angles, windows work essentially like mirrors, reflecting the trees and sky, and birds (bless their hearts) can’t always tell the difference. It can happen the other way too. Sometimes a bird will see something inside a building or all the way through two windows that they want to fly toward, not recognizing that there’s solid glass in the way. For you and me, walking into a glass pane like a sliding door is embarrassing, but for a bird flying at much higher speeds with tiny, hollow bones, it is often fatal.

McCormick Place Convention Center on the lakeshore in Chicago may be the most famous case study. In an average year, monitors would find around 1,000 dead birds around the building from collisions with the windows. In October 2023, nearly 1,000 birds died in a single night. The owner finally decided to address the problem head on, installing a pattern of dots on all the glass surrounding the building. The solution cut the number of bird strikes by 95 percent.

Awareness about bird safe glass is continuing to grow, and so are the number of solutions. Many glass manufacturers have bird-safe products that work in a similar way. The goal is to break the illusion of glass being a mirror or imperceptible as a solid surface, so most of the fixes involve patterns or designs on the glass. I think it looks pretty cool, actually. And because many bird species can see into the ultraviolet spectrum, you can even get dots and patterns that are practically invisible to humans, but still perceptible to birds to keep them from crashing into windows. Some cities are already including requirements for bird safe glass into codes, particularly in major migration corridors and for buildings that use large, uninterrupted facades of glass.

This is a really common design for tall buildings and skyscrapers. Ever since I was a kid, I have always been impressed with this kind of facade. From the outside, it looks like a sleek, continuous column of glass. And it’s pretty interesting how it’s actually done. These are called walls, specifically curtain walls. But they’re also not walls in the traditional sense, at least when it comes to bearing loads. We typically think of a building’s exterior walls as the primary load path - the way forces from the building are transferred to the foundation. With curtain walls, that’s not true, and that name kind of gives it away. In buildings with curtain walls, the load bearing happens through interior columns. The exterior walls just hang there. Not that this needs another analogy, but it’s pretty much like the building is wearing a coat.

This has a lot of advantages. A curtain wall is relatively lightweight, reducing structural loads on a building. It’s also thin, maximizing space in a building where it comes at a premium. They’re also unitized: you get very large pieces assembled in a factory so they go up quickly on a construction site. And, of course, you get a lot of light into the building when the entire exterior wall is a window. But how do you hold the glass in place?

The process of installing glass, called glazing, is not like installing most other materials. You can’t use traditional fasteners like nails or screws. Glass is brittle, and the sharp point forces would induce too much stress. If you look at the windows of your house, the glass panes are usually sandwiched inside a frame with softer material that both seals against the glass to keep out the weather and spreads out the holding forces across a larger area along the edges of the glass. That’s sometimes called captured glazing. It’s done on large buildings with expansive glass facades too, but it’s not the only way.

When you really want the outside of a building to look like only glass with no mullions or framing elements protruding beyond the panes, you can use structural silicone glazing, where the sealant does some or all the work to hold the glass in place. In 4-sided structural glazing, the glass panels are adhesively bonded on all four edges. You have nothing on the exterior except smooth glass.

If you’re used to the silicone sealants sold at the hardware store, this might seem absolutely crazy to you. I definitely wouldn’t trust a tube of bathroom caulk to hold thousands of pounds of glass hundreds of feet in the air above sidewalks and roads full of people. To be fair, these systems do usually have some kind of mechanical clips or toggles that surreptitiously prevent panes from coming entirely loose if the sealant fails, but also, structural silicone is very different from caulk.

Just to demonstrate this, I have a glass block. First I’m going to attach this to a metal plate using typical silicone caulk you might use in the bathroom.

I’m starting with a 45-pound weight on the hook, which is probably a bit ambitious. And of course, the camera died right when that dropped, but it held 45 lbs for just about a minute before it dropped. And part of that’s cause, even after 2 weeks of curing, you can see, only the caulk around the edges actually cured. This stuff is just not designed for heavy weights like this.

Now I’ll try it with structural silicone. This stuff comes from a few manufacturers in a variety of strengths, curing systems, and material compatibilities. After I let this cure for a week, which is when the manufacturer says it reaches 50 percent of its strength, I tried my little demo again.

Wow. It’s holding 45 pounds just fine.

So I added this small metal anvil just for good measure, and it’s still holding fine.

Just a silly little experiment like this can give you a lot more confidence that those big insulated glass units on tall buildings aren’t going anywhere. But the engineering of windows doesn’t stop at holding them in place. We also have to think about what happens if they break.

The most typical glass we think about with picture frames and windows is actually annealed glass. The manufacturing process is pretty impressive. The molten material is allowed to spread out and float atop a bed of molten tin, making it extremely flat and very consistent in thickness. Then the glass is allowed to cool slowly, which relieves internal stresses. That makes it possible to cut along clean lines and even drill holes in the material. But it also means that when annealed glass breaks, it usually fractures into large, razor-sharp shards that can be extremely dangerous. I think most people have enough experience with the issue that there’s hardly a need at all to say that broken glass is a serious hazard. And in many cases, it’s a simply unacceptable hazard: a slip in the shower that breaks the door, a toddler on a glass tabletop, even pushing too hard on a glass door. There are plenty of documented cases where a simple accident became fatal simply because of how annealed glass becomes jagged and razor-sharp when it fractures.

One of the main solutions to the problem is tempering the glass. If you heat glass and cool it down quickly, often using a high-pressure blast of air, you can get the outside to cool a little quicker than the inside of the glass. The outside faces shrink. Then when the inside cools, it shrinks as well, pulling the outside faces of the glass into compression. If you use a polarizing filter, sometimes you can even see the stresses in tempered glass. This is a tricky process because if the cooling happens unevenly, the glass will crack or break. I’ve discovered this myself by accidentally setting a glass dish on the stove burner. But if you do it right, tempering glass makes it much stronger than typical annealed glass. Since the outer faces are in compression, cracks are less able to propagate.

But even if it does break, tempered glass has a big advantage. Instead of the large shards, you tiny, relatively harmless pieces. These can still have sharp edges, but just being so much smaller means they’re much less able to injure someone. They have a lot less mass when they fall, and the edges are shorter, so it’s more like angry gravel and less like deadly knives and spears. That’s why tempered glass is often called safety glass, and lots of building codes require it in situations where broken glass can pose a safety hazard like glass doors, large windows close to the ground, and in bathrooms and showers. Most vehicles also use tempered glass on the side windows to reduce the risk of injury during a crash.

One disadvantage of tempered glass is that it can’t be modified after the fact. All that built-up stress in the material makes sure it shatters into tiny pieces, but it also prevents you from cutting or drilling holes. Tempering has to be the final step. The other disadvantage also comes from its main advantage. Because tempered glass breaks in a relatively safe way, it doesn’t offer much security. And really it’s true for annealed glass too. Windows aren’t just a thermal weak point in buildings; they are also a security weak point. Why would a criminal struggle to break through a reinforced door if they can just quickly smash a window? That’s where laminated glass comes in.

You can glue individual panes of glass together where they still retain almost perfect optical clarity. But even better than that, you can sandwich a tough plastic interlayer between those panes of glass, creating a composite material. And it behaves a lot differently when it breaks. Now you don’t have shards separating at all. Even completely cracked apart, the glass stays adhered to the plastic instead of flying apart. And more importantly, this is still a barrier even after it breaks. I’m not getting my hand through there. Laminated glass is ideal for overhead applications, where a break could otherwise rain down shards on people below. And it’s perfect for commercial buildings, especially on ground floors, to make it more challenging for someone to break in through a window or glass door. Laminated glass is also standard on vehicle windshields to offer more protection from a road hazard that could otherwise break through the glass and injure the driver or passengers.

What’s kind of crazy is that I’m just scratching the surface of modern window technology, not to mention the glass itself, which can get a lot more complicated for applications like phone screens. But these are all innovations that are kind of hiding in plain sight, and to me at least, they make the view through a window just a little bit sweeter.

Whatever Happened to Fire Escapes?

[Note that this article is a transcript of the video embedded above.]

The idea is simple: If there’s a fire in a tall building that blocks the main stairwell, people need an alternative way to evacuate, preferably one that is somewhat protected or at least far away from where the fire is happening. If you’re in a one or two story building, it’s pretty simple: just get out of any window or door. You may get hurt a little, but you’ll still be away from the fire. In the denser cities of the early United States, though, as the buildings grew taller, the options for evacuation became more complicated. High-profile, horrific building fires were a nearly constant plague in factories, tenements, office buildings, and theaters. The external fire escape seemed obvious to many: Bolt a metal staircase to the outside of the building with a landing below a window on each floor.

Fire escapes became an element of American culture in many ways beyond simple safety equipment. They serve as dramatic devices in films and plays, provide a scenic place for urbanites to get some fresh air, and they impart a quintessential character on the skylines of our oldest cities. But it turns out that we’ve pretty much never agreed on how well they actually work (if at all) or whether they were ever the right solution to the problem. And while many older buildings still have the original ones installed and maintained, they’ve slowly started disappearing. The story of their decline is a fascinating case study in how building codes actually work. I’m Grady and this is Practical Engineering.

An important thing to keep in mind about any safety devices in buildings, including elaborate metal contraptions that serve as fire escapes: owners don’t just install this stuff out of the goodness of their hearts. And I don’t mean that as a dig on the owners. It’s just very difficult to weigh the costs and benefits of safety devices that, honestly, might never even be called on to function over the lifetime of the building. The economics of real estate just don’t incentivize going above and beyond, or really much consideration at all of low-probability, dangerous situations. Building safety is rarely a primary factor in purchasing decisions for tenants, so it doesn’t hit the bottom line on a landlord’s balance sheet. It sounds like cold-hearted capitalism, and in many ways it is, but of course, there’s another side to this coin. You could cram so many sprinklers and fire alarms into a building that there’s no actual space to live or work in, and nobody could afford it even if they could squeeze in.

So obviously, there is some reasonable line between no safety at all and overkill. And it doesn’t make sense to put the burden of drawing that line on every individual building owner or even every individual architect and engineer who designs them. This is the entire purpose of building codes: they’re an (ideally) well-informed consensus around what’s quote-unquote “worth it.” It’s society’s way of taking responsibility for the safety of our constructed environment.

That’s exactly how fire escapes came into being in the first place. In major American cities like New York, Boston, and Chicago, we began building tall, multi-occupancy residential buildings around the mid-1800s. And they weren’t like the apartments and flats that are the baseline standard today. Multi-family housing, called tenements, were almost exclusively built for poor immigrants and laborers. They were designed for efficiency, cramming as many units as possible into a single building. And as early as the 1860s, cities started recognizing the issue of fire safety. Smoke and flames seek open vertical shafts where hot air can rise quickly, and the stairways in tenement buildings provided the perfect conduit. With lots of people, combustible materials, and often a single open stairway, these buildings were dangerous fire traps.

In February 1860, a bakery in the basement of a six-story New York City tenement caught fire. Flames quickly spread upward, but the fire department’s ladders could only reach up to the fourth floor. Ten people lost their lives, because there was no way to escape the building. The press coverage that followed catalyzed a public outcry for reform. The fire forced the city into action. Working with the state, they passed “An Act to Provide Against Unsafe Buildings in the City of New York,” which required a fireproof stairway or some means of exterior escape on all tenement buildings. The law was retroactive, meaning even existing buildings were required to comply. And that decision shaped the design of fire escapes perhaps more than anything else in their history.

Reconfiguring a stairway inside an existing building is an enormous endeavor. It’s costly, time consuming, and disruptive, if it’s even feasible at all. For many building owners, that just wasn’t realistic. The alternative was much simpler and cheaper. So that’s what most did. In reality, enforcement of the new law was lax, so many buildings didn’t comply with the new rule at all, but where buildings were upgraded, it was almost exclusively to facilitate egress on the outside.

The rules for alternative egress routes would be extended to other types of buildings over time. But they weren’t very specific about what form those escape structures should take or if a “structure” was even necessary at all. So inventors got pretty creative. All types of rope and chain ladders, chutes, slides, slings, baskets, and tubes were proposed. For a time, New York City building officials accepted just about any type of escape device as compliant with the law. One of my favorites is the parachute hat you could keep stored in a closet by the window, along with rubber overshoes to soften the inevitable hard landing. I doubt anyone took this seriously, even at the time, but you have to wonder if the inventors actually tested some of these wacky ideas.

The other large cities mostly followed New York, requiring some kind of alternative escape route on tall buildings in the event of a fire. And eventually the classic iron or steel ladder-and-balcony system coalesced as the default solution to fire escapes in the early 1900s. It definitely seems like a better system than some of the others being proposed at the time, but not everyone was convinced.

Even before the skeleton-style fire escapes became somewhat standardized, Americans were trying to apply some scientific thinking to the problem. You have to realize that building regulations are, by necessity, kind of onerous. If people were doing these things already, there would be no need for the rules. Practically every aspect of the urban environment looks the way it does because of the codes enshrined by some committee or group of professionals in what’s usually a fairly opaque process. They’re blanket requirements that often affect every single building, they’re expensive to comply with, and they have wide-ranging effects that literally shape the city. So it’s important that we get this stuff right.

One of the first technical considerations of the modern fire escape was carried out by the Franklin Institute in 1881 after the Randolph Street mill fire in Philadelphia trapped 35 people inside. Their conclusion was that external egress paths, in other words, the bolt-on fire escapes had, quote, “comparatively little value.” They recognized the fact that it was the only practical choice in some situations, but that internal stairways, properly constructed, afforded the best means of escape.

Decades later, their conclusion and recommendations were made clearer in one of the most famous and also worst industrial disasters in American history. In 1911, the Triangle Shirtwaist garment factory caught fire in the top floors of the Asch Building in New York City. The building was out of compliance with the exit laws of the time. The doors in the narrow, winding staircases opened inward, making them hard to open in a crush of panicked people trying to escape, and one of them was reportedly locked. The building had an exterior fire escape, but it buckled under the weight of fleeing garment workers, dropping them to the ground below. Ultimately more than 140 people died in the tragedy, most of them women and girls. Among the many reforms that resulted was showing how a fire escape could do more harm than good.

This debate continued for decades. One side said, essentially, we should not be using these external fire escapes. A report in the National Fire Protection Association’s quarterly journal called them, quote, “a pitiful delusion,” saying that they “contributed to the principal element of tragedy to all fires where panic resulted.” The worry was that the widespread use of these iron escapes was just making it harder to enact regulations that would be more functional. That organization eventually released a model building code that recommended prohibiting outside stairs as fire escapes in 1923, but it wasn’t widely adopted.

The other side said that, since external escapes are already so widely used, we should be focused on making them better. The 1949 National Building Code, which was largely developed by insurance companies, took a more pragmatic approach, allowing exterior escapes but regulating them with minimum sizes. Basically they made them into full outside stairways as robust and large as the ones inside. But by then, the list of disadvantages of the traditional fire escape had just grown too long:

Architects didn’t like them because they cluttered up a building’s facade, sometimes even blocking views. They could make burglaries easier, giving anyone nefarious access to the upper floors of a building. They required a lot of maintenance to prevent rust and maintain stability. No one really practiced using them, so they were rarely the first inclination in an actual emergency. Even if they were used, the bottom stairway, which was often hinged up to avoid intrusions, was sometimes inoperable from lack of regular use. They weren’t suitable for many types of buildings, like hospitals, where getting people through a window or down a steep staircase is a challenge. And, as nice as they sometimes are for tenants to have a little balcony, it is difficult to preserve their function as an actual egress route. It’s unused space, after all, so people naturally take advantage of them, putting furniture or using them for storage, blocking that path if it’s ever needed.

The only real advantage is that they didn’t take away from the expensive interior floorspace, but that just wasn’t enough of a benefit anymore. When New York City updated their building codes in 1968, exterior exits were officially prohibited for new construction. Many other large cities across the country had either already banned them or would follow soon after. What took their place? The exact thing the Franklin Institute proposed in 1881, nearly a century earlier. I have to read this quote because of how prescient it was:

“As a general rule, applicable to all buildings, the main reliance should be placed upon stairways in daily use, for the obvious reason that the occupants of a building instinctively seek such an escape in case of fire, that they are familiar with it, and that, being in daily use, it is reasonably certain to be unobstructed and to have the approaches to it clear.”

Even more than that, they suggested that stairways be enclosed in brick, built of incombustible materials, and have self-closing doors to keep out smoke. And that’s practically a perfect description of the modern tower stairway used in tall buildings today. And you might not even notice it. All the safety features are a lot less conspicuous than a spindly iron assembly bolted to the outside of the building, but we’ve just found that it’s so much safer. And it really took these gradual improvements over decades to reach this consensus: safer consumer products, fire rated building materials, sprinklers, alarms, and even reliable air-conditioning systems that make windowless, interior, smoke-proof stair towers possible.

And that’s kind of the story of lots of building codes. They have this weird momentum. For one, they normally only apply to new construction, so they’re kind of like planting a tree. We don’t see the benefits until years later. But once they are in place and the industry evolves to accommodate those regulations, there’s a lot of resistance to getting rid of them later. So it can be really hard to imagine the repercussions a specific code will have years into the future. Here’s a perfect example:

The Center on Building in North America recently completed a report on elevators, finding that they cost at least three times as much as those in other developed countries around the world. One of the reasons for this is just that they’re much bigger in America. And the reason for that is one fire inspector in Glendale, Arizona submitted a proposal to the International Code Council to require elevators fit a horizontal stretcher, erroneously stating on the form that there would be no cost impact. At face value, this is sensible. If you have a medical emergency high up in a tall building, it would be ideal if the first responders could use the elevator to get you down.

The amendment slipped through without close scrutiny, and now, if you want to put an elevator in your building in the US, it has to be a big one. But the result of that change wasn’t only that our elevators got bigger, but also that more and more building owners now just decide it isn’t worth the cost at all, opting for walkup designs instead. So the change actually led to fewer accessible buildings, and ultimately hurt the very people with mobility issues the rules were originally intended to protect.

In a similar way, New York City’s first fire egress law in 1860 applied to existing buildings as well as new construction, essentially ensuring that the prevailing solutions were going to be bolt-on retrofits to the outside of buildings, and that momentum carried for practically the next century and a half. Writing building regulations is a messy process, and there really is rarely a single “right” answer. There are just tradeoffs that affect every building differently and ripple through space, time, and the economy. There’s something nostalgic and charming about the old iron fire escapes across American skylines, but I think it’s probably best that they’ve finally gone the way of the parachute hat.

The Hidden Engineering of Pressure Regulators

[Note that this article is a transcript of the video embedded above.]

If you have a fluid-filled system of pipes in your life, whether liquid or gas, (and who among us doesn’t?) there’s a very good chance that it passes through a simple device at some point on its journey to you. This device is almost unbelievably reliable for a purely mechanical system, and it has changed very little since the mid 1800s. So reliable that there’s a good chance you’ve probably never serviced or replaced one and maybe never even noticed one, despite them controlling so many aspects of our everyday lives. Of course, I’m talking about pressure regulators. But don’t let the jargon bore you, because these things are fascinating. They’re basically Victorian-era mechanical computers, and I cut one in half so we can see how it works. I’m Grady and this is Practical Engineering.

“Control theory” is the branch of engineering that we use to describe managing dynamic systems, including the flow of fluids in pipes. I have a bunch of videos and demonstrations of just how dynamic those systems can get. A fundamental idea in this field is that, to garner any amount of control, you need some kind of feedback. And this is not a complicated idea.

Say I want to control the pressure in my garden hose. I can put a pressure gauge on it, look at that gauge, and adjust the valve until I hit my setpoint. If something changes, like someone flushing all the toilets in the house simultaneously, I’m the feedback loop. I look at the gauge and make the change to get the pressure back to where it’s supposed to be.

In fact, this exact situation (more or less) contributed to the pressure regulation equipment that we know and love today. The legend goes that in 1876, a massive fire broke out in Marshalltown, Iowa. William Fisher, a city engineer, spent all day and all night adjusting the throttle on steam-driven pumps by hand to manage the water pressure in the system to help the firefighters. Exhausted by the effort, he went on to develop the constant pressure pump governor, a precursor to the modern pressure regulators that are absolutely ubiquitous today. And I really mean that. Let’s take a little tour.

One of the easiest regulators to find is on an air compressor. You generally want the reservoir as full as possible, which means pressurizing it to a level higher than what you would actually want out of the hose. Every air tool has its own maximum pressure, so you have a knob like this so that, no matter how much higher the pressure in the tank is, you get a consistent and controllable pressure out. If you use pressurized tanks of gas like oxygen, argon, or propane - exact same thing. You’re almost always going to see a regulator on top to control the pressure leaving the tank. Maybe you have a natural gas connection to your house. In most cases, residential plumbing and appliances are designed for very low pressures, like a half a psi or about 30 millibar. That’s great for getting gas from your basement up to your kitchen, but it’s hard to get gas to flow long distances at those pressures, so the lines feeding houses are usually at pressures quite a bit higher. You don’t want high pressure explosive gas in the walls of your house, so it has to be regulated down at the meter. That’s the pancake shaped device you often see outside. Even a standard pressure cooker has a regulator on top. A weight on top of a small pipe balances the steam pressure inside, providing only enough release to maintain a constant pressure inside.

It’s not just gases either. The pressure in your water main can be too high for residential plumbing, so you might have a pressure reducing valve on your water service line. Most internal combustion vehicles have regulators that manage fuel pressure between the pump and injectors. And, of course, there are countless industrial applications of pressure regulators used in factories, power plants, and more. If you can find a pipe anywhere in the world, there’s a good chance that, no matter what’s in it, somewhere along it is a pressure regulating device. By the way, the stakes associated with pressure regulation are extremely high, particularly when it comes to natural gas.

In 2018, the Merrimack Valley in Massachusetts saw over a hundred structures damaged by fire and explosions, 22 people injured, and 1 dead all as part of a single incident. It all came down to a mistake made during a pipe replacement project that kept the regulators from working correctly. This was a system where pressure was regulated down at a district level instead of each individual meter. The mistake sent natural gas into homes and businesses at pressures way above what the plumbing was designed to handle, ultimately resulting in one of the worst natural gas disasters in American history. I covered the whole story in a video a while back if you want to learn more after this.

Here’s the thing: it’s not that complicated to reduce the pressure in a stream of fluid. Basically any kind of obstruction to the flow will do it. A simple way to do it is to put a flat plate with a hole inside the pipe. But a graph will show you why it’s not quite that easy. Let’s assume you have a constant pressure on the inlet side. If you graph the outlet pressure as a function of flow rate through the pipe, you don’t get a flat line, but a curve. And, critically, when there’s no flow, the pressure on the outlet side is the same as the inlet. There’s no reduction at all. If you let the pressure on the inlet vary, things get even more complicated. It’s easy to see why a static device, like an orifice plate, is not a very good regulator. There’s no feedback and no control. You definitely get a lower pressure in some situations, but if you need a consistent pressure that doesn’t exceed some maximum level, this is not going to work.

Early gas regulators were bulky contraptions, but actually pretty simple. You could suspend an iron bell in a tank of water. A cast iron cone was attached to the top of the bell, sliding inside the inlet pipe. If the pressure inside the bell rose, it would float upward, pulling the cone too. The higher the cone is, the more restriction you get on the inlet pipe, decreasing the flow to maintain a consistent pressure leaving the device. It’s a pretty clever invention, but not entirely practical. The water level had to be maintained; it could freeze or get gross; the metal corrodes. And importantly, when it failed, it didn’t fail safely. If the bell sprung a leak or the counterweight cable broke, the cone would fall downward, fully opening the inlet.

Modern regulators have a few features that improve on the original idea, and I happen to have a natural gas regulator so we can take a look inside. This is a used regulator that probably came from a large commercial building or a light industrial setting. And it’s actually built by Fisher Controls, the company William Fisher started after his firefighting pump throttling experience. Not a sponsor, but I like to think he would appreciate us cutting it up to learn more about it. I tried to be strategic about this to allow a look inside without it completely falling apart.

From the outside, it kind of looks like gas would make a straight shot through, but when you cut it open, you can see that there's a separation here where the regulator connects to the line. I have it set where the discharge is pointed down. Gas has to pass through this valve to make it to the discharge side, and you can see that, past the valve, the discharge side is connected to this chamber in the main body of the regulator. Inside the chamber is this flexible membrane called the diaphragm sandwiched between the two sides of the housing. It’s a little floppier than usual, since I cut the whole thing in half, but hopefully you can still see how this works.

This regulator has a stiffening plate attached to the diaphragm that acts against a spring at the top. The spring is a little too stiff for me to show you the full range of motion, so I’m going to take the seat off just to demonstrate. Let’s say there’s no demand for gas downstream. In that case, the pressure in the discharge line will build up, pushing the diaphragm upward. The diaphragm is connected to this lever, which is connected to a poppet, which pushes up against an orifice to close the valve, preventing gas from flowing.

Let’s say someone opens a valve downstream, like a stove or a heater. As the gas flows out of the system, the pressure in the discharge line will fall, reducing the pressure on the diaphragm. The spring at the top will push the diaphragm down, lowering the lever, and opening the poppet so that gas can start flowing. If the demand increases, the pressure will drop further, lowering the diaphragm and opening the valve even more. And this system will constantly adjust to the downstream pressure, throttling the valve to keep it consistent - a completely mechanical control loop maintaining equilibrium. Any difference in the setpoint and actual downstream pressure creates a proportional movement of the diaphragm and poppet valve. And it’s adjustable too: The compression of the spring at the top can be increased or decreased, which allows you to dial in the exact pressure the regulator will supply.

This is just so impressive to me. It’s a dead simple idea, but it does such an important job. But one of the difficulties, especially with natural gas, is that, like all mechanical devices, there’s some friction in the system. I mentioned that the downstream pressure of natural gas is pretty low. This regulator has an outlet range of about 1.5 to 3 psi above ambient air pressure, or about 100 to 200 millibar. Force is pressure times area. If the area of the diaphragm was small, the total force from the gas pressure acting against the spring would be practically indistinguishable within that range, especially when you consider the friction of the lever and valve. That’s why the diaphragm in natural gas regulators is so big. Even small changes in pressure create large difference in force, so you get more sensitivity, and the valve positions are more closely tied to the actual changes in pressure.

You might see an issue with this design though: For the valve to open wider to allow more flow, the diaphragm must move down. For the diaphragm to move down, the pressure holding it up (the downstream pressure) must drop. Engineers call this droop, which I love. But there is still some variability in the downstream pressure. Pressure is tied to the valve position, so it’s necessary that it be allowed to fluctuate some. It will never be rock solid in this model. If you need that, the solution is usually a pilot-operated regulator. In this design, the downstream pressure is connected to a tiny, ultra-sensitive pilot regulator, and that regulator basically uses the higher-pressure inlet gas to move the main valve. In this way, you can go from 0 percent to 100 percent flow with almost no change in downstream pressure.

Regulators can also be sensitive to inlet pressure. You can see on my model that the inlet pressure acts against the spring to open the valve. Of course the valve is a lot smaller than the diaphragm, so the effect isn’t as big, but there’s still a relationship between inlet pressure and outlet pressure, which isn’t always ideal. A lot of regulators work the opposite way, where the inlet pressure acts to close the valve. If you use a regulator on a tank, this can cause the counterintuitive issue of discharge pressure spiking as the tank empties, since the inlet to the regulator isn’t pushing as hard to close the valve. If you want to reduce this sensitivity, you can use a two stage regulator where you drop the pressure in steps. Let the first stage handle the coarse reduction, providing a more consistent inlet pressure to the second stage which can then keep the discharge pressure rock steady.

One thing this regulator doesn’t do is fail closed. If this diaphragm rips, the outlet pressure won’t be able to push it upward to close the valve. So we have to account for that potential in other ways. Lots of gas systems will use a secondary, redundant regulator set to a slightly higher pressure that will take over if the primary fails. There is also a circuit breaker equivalent for gas systems called an overpressure shut-off or slam-shut. This model uses another option: an internal relief valve.

Say the pressure on the discharge end somehow got too high. Maybe something got stuck in the valve, keeping it from fully closing. Or maybe the discharge line was exposed to sunlight, expanding the gas inside. In this case, the diaphragm can bottom out and act against this secondary spring, lifting off this plate. Gas is allowed to escape through a hole in the center of the diaphragm into the top half of the casing and out of this vent hole. And here we have another valve called a flapper. It can open inward to balance the pressure inside the regulator. And it can open outward if the relief valve activates, letting the excess pressure escape. The regulator would normally be mounted like this so the vent points downward, keeping rain out. And it has a screen so bugs don’t make a home inside.

Obviously, this has some tradeoffs. This regulator has to be mounted outside or be attached to a ventilation pipe running outdoors to make sure it’s not releasing gas into a closed space. Even so, you don’t necessarily want to vent a bunch of natural gas outside. But because of the odorant that’s added to it, the idea is that someone would notice pretty quickly that some part of the system is malfunctioning and shut the line down for repairs.

Like every part of engineering, it’s a game of tradeoffs: pressure versus flow, capacity versus cost, accuracy versus redundancy, and safety here versus safety there. I just love that there’s stuff like this out there, pretty much anywhere you’re willing to look, doing an essential job that few people even consider, and that their basic function really hasn’t changed in centuries. Samuel Clegg, one of the early engineers in natural gas systems had this to say about the pressure regulator: “Its use is nowhere sufficiently appreciated. Had it been a complicated piece of machinery, or expensive in its first cost and after application, objections to its adoption would not have been surprising; but it is perfectly simple: its action is certain and unvarying, and its first cost inconsiderable.” Nearly 200 years later, I couldn’t have put it any better myself.

Every US Electrical Outlet Explained

[Note that this article is a transcript of the video embedded above.]

I love the periodic table of the elements. I love it because it reveals the deeper order of what seems like an otherwise wildly disparate collection of atoms with different physical forms, chemical properties, and nuclear stabilities. I love it because, even before we actually found the elements that fit into each box, we knew that something did and could even predict some things about those elements before they were ever discovered. And finally, I love it because it’s a bit messy. Not everything lines up perfectly, and in some ways, it’s still a work in progress. In many ways, human-created standards follow that same form, and I want to try and convince you that they deserve the same affection.

Let me present the periodic table of standard North American electrical connections. Isn’t it beautiful? I’m fascinated by stuff like this: a diversity of needs and purposes put into a relatively nice, neat order. But why do we need so many? And where do any of these actually get used? Well, I’ve spent the past month reading just about everything I could find on electrical plugs and receptacles to figure those questions out, and I even have a few of them here so I can show you what I learned. I’m Grady, and this is Practical Engineering.

Electricity is something we really don’t want to be proprietary. It’s one thing if your charger doesn’t work on your buddy’s cell phone. It’s another thing entirely when you have to rewire your house because you bought a different brand of toaster. The National Electrical Manufacturers Association, or NEMA, was founded in 1926 as a coalition of companies making electrical equipment. Their members realized that life would be better with some standards, so that any company making an electrical device could be reasonably confident that the people who might want to buy that device would be able to use it, and more importantly, use it safely.

This didn’t happen overnight. It took a diverse group of manufacturers, engineers, and testing labs to form a consensus around the system we use today. And it’s far from a perfect system. My friends Mehdi and Alec have covered receptacle-related topics on their channels, including the merits and disadvantages of the NEMA designs. But it works pretty well. Well enough that the NEMA connector standards have been adopted not just in the US, but all of North America, Central America, parts of South America, Japan, Taiwan, the Philippines, and beyond.

Here’s that table again. You probably noticed that every type of plug and receptacle has its own special number. They seem a bit arcane at first glance, but it’s actually a handy naming scheme that’s pretty straightforward to understand. The first number is the configuration that defines the combination of voltage rating, wire count, and grounding style. These numbers are a bit arbitrary, but they kind of represent a certain class of receptacles and plugs. For example, NEMA 1 receptacles are rated for 125 volts and have just 2 poles (a hot and neutral) with no ground. The NEMA 1-15 was the classic North American outlet until the 1960s, and you still see these in older buildings. Lots of devices made today can still use them, especially low-voltage equipment like chargers, and, critically, those without external metal parts. If an energized wire inside the device comes loose and contacts the case, there’s still an insulating barrier protecting someone from being shocked. The reason NEMA 1 receptacles are mostly a thing of the past is what could happen when equipment didn’t have that protection.

If a device with a metal enclosure or exposed metal parts had an energized wire come loose, that metal would be energized too. But, critically, it might not create a short circuit. With nowhere for current to flow, the device could just sit there, indefinitely dangerous, until someone happened to touch it, allowing current to flow through them to a lower potential. The ground wire we see in nearly all plugs and receptacles today fixes that specific hazard. Bonding exposed conductive elements and connecting them to ground makes sure that if they somehow become energized, current will flow, a short circuit will form, and protective devices like breakers will activate.

Today we use the NEMA 5 standard for the vast majority of receptacles and plugs. Even if you’ve never heard of NEMA or seen the other plugs on the periodic table, you’re almost certainly familiar with this design. They have a 125 volt rating to handle the standard 120 volt service for most electrical devices with a little buffer. They have an energized pole, called the hot; a neutral pole to provide a return path, and a separate ground return that is bonded to the neutral line in the main electrical panel. The ground pin on most outlets is round instead of flat, and that’s the reason why nearly all electrical outlets kind of look like they’re screaming. Or at least they do to me.

One thing about NEMA 5, and actually most of the NEMA configurations, is that the outlets have polarity. On the NEMA 5-15, the neutral slot is a bit wider than the hot, making it so the plug can only go in one way. In function, polarity often doesn’t matter for AC circuits. Current travels in both directions, so the equipment inside the device can’t really tell the difference. And some devices, like switch-mode power supplies, don’t care which direction they’re plugged in. Both blades are the same size. For safety, though, a lot of devices do. You really don’t want heating elements, motor coils, and circuit boards energized and waiting for a ground. It’s less hazardous to put the switch on the hot wire so that nothing beyond the cord is energized until it’s turned on. Enforcing polarity at the plug prevents “switched neutrals” along with other issues like electrical noise.

The NEMA 5-15 plug and outlet were designed to be backward compatible with the older 1-15 standard. 1-15 plugs work just fine in the modern 5-15 outlets, and there are quite a few interesting compatibility cases like that in the NEMA standards. For example, the “15” in 5-15 refers to the current rating. Nearly every household device and appliance that runs on 120 volts is designed so that it never draws more than 15 amps, and actually, if the device is meant to run for more than 3 hours continuously, like a space heater, it can only draw 80% of that (which is 12 amps if you’re keeping score at home).

That limit is obviously fine for most household appliances. But, especially in commercial spaces, it’s not quite enough power for certain devices like kitchen mixers, treadmills, copy machines, and power tools. Of course, we could just change the codes to require 20-amp circuits everywhere, but that has huge implications: larger circuit breakers, heavier-gauge wiring, and more expensive receptacles. And in many cases, it’s just not necessary. So instead, NEMA created a different receptacle and plug for 120-volt, 20-amp circuits, the 5-20. I have a bunch of these in the studio. You can see they have that T shape on the neutral slot. And 20-amp devices have the neutral blade rotated 90 degrees on the plug. But here’s the backward compatibility: regular 15-amp plugs fit into the 5-20 receptacle as well.

NEMA 5 has 30 and 50 amp receptacles too, although they aren’t used very often these days because of a quirk about the historic availability of voltage. Today, split phase electrical service is basically standard for residential power. You get two 120-volt hot lines which can be used individually for smaller circuits or combined to get 240-volts for circuits that need more oomph. In the early 20th century, 240-volt service wasn’t always available, so you have these very-high-current 120-volt receptacles that could power heavy commercial cleaning equipment like floor burnishers and blowers, kitchen equipment like warming cabinets and steam tables, and large shop tools like table saws and compressors. Also, not all portable generators run at 240-volts, so older models used the larger NEMA 5 receptacles as well. These are still available and installed in places where, for whatever reason, a higher-voltage circuit is hard to come by. But in most cases, the more power-hungry devices are going to run on 240-volts. That brings us to NEMA 2.

Like NEMA 1, these are ungrounded receptacles, but instead of a hot and neutral, they have two hots. Each is 180 degrees out of phase with its neighbor, so you get 240-volts across them, handled with a little cushion by the 250-volt rating. There were 20 and 30 amp receptacles, but, also like NEMA 1, these are mostly obsolete now that a ground is required by code. They’ve been replaced with NEMA 6, which has 15, 20, 30, and 50-amp receptacles and plugs. Of course, with double the voltage, you also get double the power compared to the NEMA 5 equivalents at the same current rating.

The 6-15 is common for window or wall-mounted air conditioners. The 6-20 is used for heavier-duty air conditioners plus commercial kitchen equipment and shop tools. The 6-30 is used with large heaters, kilns, and heavy power tools. The 6-50 is kind of the standard welder outlet, plus it’s pretty common these days for level 2 EV chargers, capable of delivering nearly 10 kilowatts of continuous power through the receptacle.

Like NEMA 5, the NEMA 6 has some backward compatibility, allowing 6-15 plugs to fit into 6-20 receptacles. This is kind of clever, but it doesn’t work all the way up the different current ratings. Of course a 50-amp outlet could easily handle a 15-amp device. And it would certainly be possible to design a series of outlets where each successive jump in current rating allowed those smaller devices to plug in. But there are two main reasons why they don’t:

One is practicality. The blades on plugs aren’t all the same thickness. Designing a single receptacle slot that can safely grip both a thin, 15-amp blade and a massive 50-amp one would make manufacturing more difficult and increase the chances of developing loose connections inside the receptacle over time. Two is safety: circuit breakers are sized to protect everything downstream, including the plug and the appliance cord. If a thin cord on a low-current device develops an internal short, the resistance of that thin wire itself will cap the fault current so that a larger breaker might take much longer to trip or not trip at all. That could allow the wire to reach high enough temperatures to start a fire. Of course you don’t want a high-current device plugged into a lower-current-rated circuit, but if you trace out the things that can go wrong, it turns out that you also don’t want lower-current devices plugged into a high-capacity circuit. So, the plugs and outlets are designed to prevent both cases, except for the 15 and 20 amp situation, where the current is close enough that a breaker should still work as intended.

240 volts are useful to supply more power at the same current rating, but of course it comes at a cost. Higher voltage means more potential, literally, for arcs to occur. Equipment designed to handle the higher voltage needs better insulation and more careful design. Take a clothes dryer for example. You want the extra voltage for the power-hungry heating elements, but all the other stuff inside (like timers, controllers, and clocks) can easily run on 120 and those lower-voltage components are more affordable. That’s where NEMA 10 came in. You get three poles: two hots and a neutral. In that way, you get dual voltage: 240 between the hots and 120 between each hot and neutral. Of course, NEMA 10 receptacles also lack a ground connection, so they’re mostly obsolete. Plenty of houses still have them installed for clothes dryers and kitchen ranges, but since the 1990s, they’ve been supplanted with the NEMA 14 configuration.

This is the most widely-used 240-volt standard in North America today. It’s versatile, providing both voltages. And there are a full range of current capacities, allowing you to design a circuit that’s well-suited for a device, from 15 all the way up to 60 amps.

The 14-15 is pretty rare. I couldn’t even find someone making the receptacle. The 14-20 is also not that common. Some food service equipment uses this like certain coffee makers. The warmers rely on 240 volts while the fans and timers run on 120. Same with some jobsite heaters and specialized laboratory equipment.

The 14-30 is the standard residential electric clothes dryer plug and is often used for EV chargers. Some server and mainframe equipment uses it as well. The 14-50 is the standard residential cooking range and oven plug. It’s also widely used for EV chargers and pretty common at RV campgrounds as well. The 14-60 is more of a commercial or industrial receptacle, used for large kitchen appliances and distribution of power at events like concerts.

Single phase electrical service covers nearly all residential and lots of commercial buildings. But, the grid runs on three phases and it’s pretty common for larger commercial buildings and essentially all industrial facilities to have three-phase service. It’s particularly useful for devices that use large motors. And of course, if you have the service, you’re going to need receptacles and plugs for those devices, or at least the ones that aren’t hard-wired. NEMA 11 was the standard for up to 250V with receptacles and plugs ranging from 15 to 50 amps. Those have been replaced by the new NEMA 15, again because of grounding requirements. And this is going to almost always be relatively specialized industrial devices: woodshop and machining tools, laboratory testing equipment, grinders, pumps, dust collectors, heavy welders, plasma cutters, and so on. It’s not stuff most people see in everyday life, and in many cases, each receptacle is going to be custom-installed for a specific piece of equipment. And since hard-wiring equipment directly to the service panel is typically the default, that makes receptacles like these even more rare. You really only see them in places that need a high degree of modularity, allowing for rapid reconfiguration of workspaces like jobsites, certain manufacturing facilities, and short life-cycle equipment that needs to be easily swapped out.

There are two main three-phase service classes used in most commercial and industrial buildings in the US. The most common is 208 volts phase to phase, which uses the NEMA 15 configuration. There’s also 480 volts phase to phase, but like I mentioned before, you can get a lower voltage between phase and neutral (in this case, 277 volts). So NEMA 7 has plugs and receptacles specifically for using just one phase from buildings wired with 480-volt, three-phase service. A lot of commercial and industrial lights use these receptacles, like warehouses, factories, and arenas, making them easy to swap out without hard-wiring. Commercial ventilation and air conditioning systems use them too.

And just like the dual-voltage 240-volt plugs, there are also dual-voltage three-phase plugs, delivering equipment with all three hot phases plus a neutral so different components can run at different voltages. NEMA 18 has receptacles for 208-volt service, although they don’t have a ground, so they’re mostly obsolete. There are no straight-blade plugs that have replaced NEMA 18. Aligning and inserting a 5-blade plug would be tricky and take a lot of force. And I’ve kind of buried the lede here only talking about the straight-blade NEMA standards. The reality is that a large number of the NEMA receptacles and plugs have an equivalent locking version. These use curved blades that twist inside the receptacle so they can’t be easily pulled out. Actually the locking versions are more common than the straight-blade equivalents in many cases, especially when it comes to portable generators, jobsite equipment, and events where things are always moving around. If your vacuum cleaner unplugs itself because you’ve gone too far into the hallway, that’s usually not a big deal, but if a three-phase 600 volt plasma cutter does the same thing, you can get serious damage from arcing. That’s why the locking standards extend beyond the voltage ratings of the straight-blade ones up to three-phase 600-volt circuits. They even have receptacles for 400-hertz power used in aerospace, submarine, and military systems.

Of course, sometimes the standards make themselves. When it comes to RVs and travel trailers, (from what I can gather) the industry had already developed a 120-volt, 30-amp receptacle before NEMA formalized its catalogue of standards. Instead of forcing an entire industry to retool, NEMA just adopted what everyone was already using, calling it the TT-30. TT for travel trailer and 30 for the current capacity. In function, it’s not any different than the NEMA 5-30 receptacle and plug, but you’ll almost never see one of those, because the TT-30 is far more common. It’s a face only an outlet enthusiast could love.

I haven’t really talked about the smaller versions of the locking connectors used where space is an issue. And there are even more specialized standards like ship-to-shore power, aircraft, and military uses. Of course, when you look beyond NEMA, there are way more standards out there. But I feel like this is enough to get you excited about the weird, wide world of electrical receptacle standardization. There are all kinds of practical considerations that make it much more complicated than just a 2D chart with voltage on one side and current on the other. Just like the periodic table of the elements, the NEMA connection standards are a bit messy. And that’s what I love about them.

Stopping the Unstoppable

[Note that this article is a transcript of the video embedded above.]

On the morning of September 29, 2016, a New Jersey Transit commuter train rolled into Hoboken Terminal during rush hour with about 250 passengers on board, and it just… didn’t stop. The train reached the end of the track, overrode the bumping post, and slammed into the terminal wall at about 21 miles per hour. Debris fell onto the platform, killing one person, and 110 passengers and crew were injured.

Just three months later, essentially the same scenario played out again, this time at Atlantic Terminal in Brooklyn. A Long Island Railroad train crashed into the end of a terminal track. The lead car actually rode up and came to rest on top of the concrete platform. 108 people were injured, but thankfully, there were no fatalities.

In both investigations, the NTSB reached a simple conclusion: the trains didn’t stop because the engineers fell asleep. They were diagnosed with sleep apnea after the crashes, a chronic disease that leads to sleepiness and fatigue. But the NTSB also pointed to the absence of safety devices or systems that could have intervened to stop each train before the collision.

When you picture a railroad, it probably feels endless. Rails stretch for miles in both directions, with plenty of room for mistakes to unfold. But every line has an end. No matter how good the brakes are and no matter how experienced and well-trained the crew is, any system that depends on an individual, biological, fallible human, is eventually going to fail.

That’s why end-of-track protection is such a unique engineering problem. You might say that those bumping posts at Hoboken and Atlantic Terminal failed, but in reality, they were never meant to “catch” a fully powered passenger train at speed. They were a last-resort backstop, designed for low-speed situations. Train brakes have a lot of redundancy, and failures are rare, but those brakes only work if they are applied in some way. If it doesn’t happen, a static bumping post doesn’t offer much protection, and may even make things worse. End-of-track protection seems like a simple problem, but the solutions are pretty complex. And of course, I built a little model in the garage so we can take a closer look at them. I’m Grady, and this is Practical Engineering.

In many ways, stopping a train is a pretty elementary physics problem. You have the train’s mass, its speed, and an elementary equation: one half mass times velocity squared. That gives you the system's kinetic energy. And in order to bring a train to a stop, all that energy has to leave the train and go somewhere. But this can vary by orders of magnitude. A single empty freight car inching along in a yard may only reach kinetic energies on the order of a hundred kilojoules. A tram or streetcar experiencing a low-speed overrun, like a partial brake failure, may take a few hundred kilojoules to stop. For light rail, you’re around half a megajoule to stop one moving at terminal approach speed. Heavier commuter rail trains may be 5 to 10 times that. And heavy-haul freight trains can climb into the hundreds of megajoules even at relatively low speeds.

You can see why the design of end-of-track devices, often known as bumping posts or buffer stops, is deceptively complicated. The ability to bring a train to a halt in an overrun scenario depends on both the mass and speed of the train. Ideally, if a train is approaching the end of the line, it’s not moving at track speed, and that is the assumption that most designs are based on. That’s why simple bumping posts are so common. They’re like a door stop for a train, just static devices that provide a hard stop at the end of the line. Let me show you an example:

This is my train stopping simulator, which is kind of just an excuse to play with a pneumatic piston. At one end, is that piston, and I can use a regulator to adjust the force this applies. For my “train,” I’m using a little cart on a length of V-rail with some weights.

Let’s set up a little bumping post and see what happens. No surprises here. At low speeds, the train stops nicely against the post. Let’s try it at a higher speed. The collision is a little more violent. And actually, we can quantify that. You may have noticed the accelerometer at the top of my train and the laptop with squiggly graphs in the background. Here’s the data from that smoothed out a bit from the raw readings. Again, at low speeds, it’s not a big deal. It takes about 5 gs to bring the train to a stop. At higher speeds, it was maxing out my accelerometer at 16 g.

It makes sense why you see static bumping posts where only low-speed overruns are expected. They’re cheap; they’re simple; they don’t take much maintenance. But obviously, this is not a great solution for larger trains, and especially not for situations where speeds are higher. If you think back to our equation, where is all that energy going? Obviously, not all of it is leaving the train, at least in my demo. You can see it being bounced back a bit. The rest of it goes into heat and sound generated by the impact, vibrations into the table and track, and some plastic deformation of the post itself. It’s a bit hard to see, but it did bend over a bit in the high-speed tests. There’s really no energy dissipation built into the device. It just resists until something physically breaks.

There are some situations where that might be okay. I kind of dove into this without really taking the time to discuss what safety really means. That’s because there is no one answer in the case of a terminal track. The consequences of an overrun can vary a lot, and so can the consequences of a violent stop. On an industry track or yard, an end-of-track device may just be there to save the inconvenience of having to lift a railcar back onto the tracks. Nothing's moving very fast, so a rigid stop isn’t likely to damage the rolling stock. In passenger terminals, or places where buildings or public spaces lie beyond the end of the track, that calculus changes. You may have people on the train, and you might have people in its path. The design choices between stopping a train in an instant and limiting the violence of that stop are entirely a case-by-case decision. Even given a set amount of energy to dissipate, the requirements for acceleration will vary. So, there’s another simple equation to consider.

The stopping distance of a vehicle from an initial velocity is inversely proportional to the acceleration (or in this case deceleration). They’re on a see-saw. If you want to limit accelerations, the force that an end-of-track device applies to a train has to happen not at a single point in space but over some amount of distance along the track. Sometimes that happens whether you want it to or not. “Telescoping” is the term used to describe when railcars crush into or over one another. Like I said, the kinetic energy has to go somewhere. For freight trains in a true emergency situation, some amount of equipment damage might be acceptable. I can simulate that using a soda can. You can see that the accelerations go down a lot when you’re willing to accept some deformation. You can probably also understand why that’s rarely an ideal configuration, though.

One of the simplest alternatives is just to let the bumping post slide. And I can show you exactly what that looks like. I’ve modified my little demonstration here by adding some springs along the bolts that clamp this bracket down. Let’s run it again to see what happens. You can see that now, instead of an instantaneous hard stop, there’s some stopping distance. Looking at the graph, the peak acceleration is a lot lower, so this is much less violent for the train. And, unlike the hard stop, it’s easy to see where the energy of the train is being dissipated. It’s in the friction between the post and the track.

This is a pretty common arrangement in the real world, where limiting deceleration matters to protect passengers and equipment. The post or buffer stop is equipped with brake shoes that slide along the rail, burning off that energy through friction. Theoretically, this provides a constant level of deceleration that is independent of the train's mass or velocity. The friction is only a function of the clamping force and the properties of the brakes and rails. In the real world, though, that’s tricky to get right. You can see in my demo that there’s a spike in acceleration when the cart first hits the stop. That’s because it has to overcome the static friction first, before the lower kinetic friction takes over. You can also see the deceleration is a little more erratic. That’s the so-called stick-slip action of metal-on-metal sliding. You also have environmental factors like rust or moisture on the rail that can change the force curve. And of course, once something like this gets used, it has to be reset or, depending on the intensity of the collision, replaced, which can be a big job. There is another solution that addresses these challenges.

I’ve replaced the end-of-track device on my demo again, this time using a dashpot snubber. This cylinder that forces air though an adjustable hole to provide a nice smooth force curve. Let’s try it at a slow speed first. I’ve adjusted the stiffness to work well for this situation. Now let’s try a faster run. It looks a lot smoother. And you can see in the accelerometer data that it is. Here’s the rigid stop, the sliding friction run, and the dashpot run all together. The sliding friction had the lowest acceleration, which makes sense because it used up the longest stopping distance. But the damper provides a more controlled and consistent stop, especially when considering the range of train speeds and sizes. And it resets itself after the impact.

That’s why devices like this are most often used for passenger railways where control matters most. Usually, they are hydraulic cylinders that use oil to provide smooth deceleration across a range of train masses and speeds. Of course, using a hydraulic buffer stop doesn’t solve the distance problem. A device like this doesn’t really buy you more space. And in fact, if you’re trying to stop a high-speed train without extreme accelerations, you often need a lot more stopping distance than what’s possible with hydraulic buffers. In that case, sliding friction stops are really the best solution. You can see in my demo that it’s possible to bottom out the dashpot, and after that, it just becomes a much more expensive static bumping post. Hydraulics also add another layer of complexity and maintenance, so these are generally more expensive than sliding friction stops. And actually, you can combine the two into a hybrid system that uses the hydraulics for slower speed impacts… if you remember to tighten the clamps. Let’s try that again. This setup is kind of the best of both worlds. You get the control of the dashpot, while still allowing for sliding friction to dissipate extra energy in extreme situations.

There is another end of track safety solution at the other end of the complexity and maintenance spectrum: the earth mound, essentially a big pile of dirt. Never underestimate the power of a pile of dirt. If there is a large drop-off, important building, or some other critical situation beyond the end of the track, sometimes you don’t want to rely on a modest bumping post or buffer to make absolutely sure a train stops. A simple pile of earth doesn’t do much to protect the train, but it will definitely protect whatever’s behind the pile of earth, and sometimes that’s the thing that matters most.

Of course, there’s a lot more to this than the demos I’ve shown. Unlike the automatic "knuckle" couplers common in North America, many train cars are linked using chains. Managing slack in the train requires shock absorption on the cars themselves, and these devices are also known as buffers. There are also situations where you need to prevent a train from reaching a certain section of track that isn’t at the end of a line. A derailer is a device that forces train wheels off the track using a wedge-shaped metal block. Again, it’s likely to cause damage to the rolling stock, but there are situations where safety is worth that cost. A controlled derailment at low speed is far better than a runaway car endangering workers, crossing a busy road, or colliding with a high-speed passenger train on the main line.

There are also non-structural solutions to train overruns. If buffers and derailers are the "brawn," Positive Train Control is the "brains.” These systems use digital logic to take over train operations if a human operator makes a mistake or is incapacitated. PTC relies on GPS, trackside sensors, and onboard computers to "know" exactly where the train is, how fast it’s going, and what the conditions are ahead. If the system detects that the train is exceeding its authority, the PTC will automatically apply the brakes.

When you’re moving big heavy stuff around, and especially when that big heavy equipment has people inside or nearby, there is a ton of thought and engineering that goes into the safety systems, their redundancy, and the unavoidable physics tradeoffs between energy, acceleration, and space. Whether it’s the elegant, constant pressure of a hydraulic piston or the raw, grit-and-metal friction of a sliding stop, buffer stops and bumping posts are the last line of defense against the unimaginable energy in modern trains. Digital innovations like Positive Train Control are helping to ensure these devices never have to be used, but as long as we rely on rail (and I hope we rely on it more in the future), the "simple" physics of stopping will always be a complex balancing act.

Are Beavers Actually Good at Dams?

[Note that this article is a transcript of the video embedded above.]

When I was a kid, my dad would take us camping almost every year in the San Juan Mountains of Colorado. We would hike along the valleys and fish in the ponds that dotted the little creeks like pearls on a string. I remember one time I was fishing and I heard this monstrous SLAP in the water. I looked up just in time to catch a beaver diving under the surface. And it just clicked right then. These little ponds weren’t part of the landscape. At least, not like the rest of it, formed by weather and geology. They were constructed. I was an interloper into an entire engineered ecosystem that industrious rodents had built to their own design.

Is that experience why I became a civil engineer? Wouldn’t it be funny if that were true? If I just really thought beavers were cool and wanted to grow up and become the human equivalent of one? It’s actually kind of the opposite, and not just because that beaver scared me when it slapped its tail. When I became a dam engineer, beavers became my professional rival. They gained their knowledge over millions of years of evolution. I gained mine from a handful of years in a university. Without a single blueprint, degree, or bulldozer, the beavers were showing me up.

I don’t work on dams anymore, but this is still something that keeps me up at night. I’ve spent a lot of time considering the differences between the professional practices that have culminated from centuries of human progress and the innate behaviors of these cute but unusual creatures. There are some surprising similarities, some glaring discrepancies, and a growing admission around the world that the rodents might have a better dam system. I’m Grady, and this is Practical Engineering.

It’s not unusual for animals to build stuff. There are lots of interesting examples of bioengineering out there, from intricately woven webs, to towering termite nests, to sprawling prairie dog towns, and even underwater crop circles. What’s impressive about what beavers do is the scale of it. They’re a keystone species because their presence affects the ecosystem so significantly. Beavers’ ability to reshape the landscape to fit their needs gives them an enormous advantage. The historic range of the two beaver species on earth covers pretty much the entirety of the continental US and Canada in North America, plus a huge area of Eurasia. They can live almost anywhere because (just like us), if a habitat doesn’t suit them, they just fix it so it does.

Not all beavers build dams, but a lot of them do. And their dams are not really that complicated. It usually starts with some branches or rocks at the base to kind of anchor things in place. Then it’s just a pile of interlocking sticks and logs. Finally, they place mud, leaves, and grass along the upstream face of the pile to make it more and more watertight. If you’re familiar with dam engineering, you probably recognize this design. It’s basically a zoned embankment dam.

Just like humans pile and compact soil (and sometimes rock) to create reservoirs, beavers do it with wood and mud. And the structural behavior is very similar. Branches interlock like soil and rock particles, creating a sort of friction that binds them into a stable mass. No cementing substance required. Not only that, beavers also use different materials to do different jobs. Logs and branches are large and lock together well. They work perfectly to form the bulk of the structure, but a large mass of just trees is highly permeable. Mud and leaves are less permeable, but they wouldn’t form a stable structure on their own. Applied as a layer on the upstream face of a mass of logs and branches, they are an ideal way to “plug up the holes” and seal up the structure. Our dams work in exactly the same way.

Embankment dams are the most common type of dam in the world. They’re not as glamorous as the concrete behemoths, but they work well, and they’re often more cost effective to build. And it is very common that they’re built in zones. Just like the beaver’s branches, the bulk of an embankment dam uses a shell material where the specifications are looser. This soil can often be sourced nearby, reducing transportation costs, and it doesn’t need to be sorted or separated so carefully since its primary job is to add mass. It’s okay if water seeps through the shell, because we also have a core. This is the impermeable zone, made from clay, concrete, or asphalt, that stops the water from flowing through just like the beaver’s mud and leaves. Some human dams even put this layer on the upstream face, just like the beavers, although most zoned embankment dams have it in the center where it’s more protected from eroding away.

So equal points awarded for design philosophy. In fact, I think it’s a pretty good example of convergent evolution of professional practice. The environmental pressures of resources that vary in cost and function made embankments the crabs of the dam world. But despite our similarities, humans and beavers are very different in how we build them.

The first part of human dam construction is usually a process of diversion. You have to get that water out of the way so you can build in the dry. On some sites it’s as simple as digging a channel. For the more famous projects in big rocky canyons, there were cofferdams and massive bypass tunnels. Beavers don’t have this luxury. They just build right in the current, and this takes some ingenuity. Often the work is seasonal, taking advantage of dryer parts of the year. The choice of stream also matters. Beavers don’t impound enormous rivers like we do. They prefer smaller creeks in gentle valleys. They also don’t stop the water on day one. The order of the construction makes each next step possible. First the rocks and heavy longs that will stay put in flowing water. Then the matrix of branches that lock together while still allowing the water to flow through. Only when the skeleton is there do they start placing the mud and finer debris to slowly raise the level of the water.

I love this in a weird way - I think because it’s how I remember building stuff as a kid. It’s obvious what the next step is, so you don’t really need an elaborate plan. It’s kind of adaptable: you just see if one step needs extra effort here or there. Beavers are testing their construction every moment as they go. In that way, they really aren’t exactly true engineers. They have no need to predict how something will behave. They just build and fix and fix and build until it works, and there’s something admirable in that.

It’s not to say that beaver dams don’t fail though. In fact, they’re relatively fragile structures. Sometimes the mud layer loses its integrity, allowing water to resume flow through the embankment. Given enough time or a strong enough flow, this can break the dam apart and eventually wash it out. Sometimes the stream carves a path around the dam, blowing out the riverbanks on the sides. Many failures result from floods that simply overwhelm the dam. Every once in a while, those failures are dramatic on a human scale. In 1984, a failed beaver dam washed out a railroad embankment, causing an Amtrak passenger train to derail in Vermont, ultimately killing five people and injuring 150 more. A similar situation happened in Ontario in 1994 and in Michigan in 2003.

Beavers show almost no consideration of margins. They use cheap, continuous labor as their factor of safety. A beaver dam is a dynamic structure. Its "sensor array" is the sound of running water. Beavers are super sensitive to it. When a leak happens, that audible cue triggers an automated biological repair loop. It works perfectly for continuous maintenance, but not for rare floods. Beavers design for the present, accepting, whether intentionally or just through ignorance, that when the river rises, their work will be ruined.

Humans don’t have the luxury of ignorance, at least in modern times. I guess it wasn’t that long ago that we did. In the late 1800s, a bunch of wealthy elites modified an abandoned dam in Pennsylvania so that their hunting and fishing resort would have a nice lake full of stocked fish. When a relatively modest flood came in 1889, the dam overtopped and washed out, sending a wall of water downstream to Johnstown, killing more than 2000 people. None of the members of the South Fork Fishing and Hunting Club ever admitted fault or were held responsible for the disaster.

I have a whole chapter about the story in my new book, Disasters By Design: How Engineering Failures Shaped the Modern World. Pre-order link below if you want to pick up a copy.

We remember the Johnstown Flood and the other catastrophic dam failures that preceded and followed it. And we, collectively as a society, don’t really tolerate those kinds of risks anymore. In fact, when you look at loading conditions for civil infrastructure, dams are designed for probably the most extreme events of any other type of structure. Many dams are built to withstand the probable maximum flood, which is literally the most severe flood we can conceive of for a given location. That’s not so much to protect the dams themselves and the resources that went into building them as it is to avoid the consequences to all the people and structures downstream if they fail. Once again, equal points to beavers and human dam engineers on this one. Beavers, with their little beaver brains, are necessarily less thoughtful about the consequences of their actions. But also their dams, almost always being fairly small and located in undeveloped areas, don’t usually have serious consequences when they fail. Besides a few isolated calamities by the beavers, almost all the damage and destruction caused by dam failures has resulted from human-built structures. But failure isn’t the only cause of damage and destruction when it comes to dams.

Whether human or beaver-built, dams represent a substantial change to the local hydrology and ecology. We’re kind of used to thinking of this in a binary way: something is either good or bad for the environment. But when you get into the specifics, it gets a lot more complicated than that. The quote-unquote “environment” is extremely dynamic, constantly changing, and what’s good for one part of it isn’t always what’s good for all of it. And "good" isn't always an objective baseline; it's a value judgment based on which outcome we choose to prioritize, and beavers really highlight that complexity.

It seems like a paradox at first glance: Especially in modern times, beaver dams get a ton of credit for environmental uplift and restoration while human engineers get flooded with red tape and potential regulatory penalties for doing the exact same thing. There are three big differences that explain the discrepancy.

One is permeability. Beaver dams are messy, porous, dynamic structures. They act more like a sieve than a plug. Because they’re usually small, and built from mud and sticks, they cause much smaller changes to the flow of sediment, nutrients, and fish from one side to the other, especially when viewed on the larger scale of a whole regional ecosystem.

Two is dynamism. Beaver dams are temporary, as we already discussed. They are fragile structures frequently abandoned or washed away. When a beaver family moves on or their dam fails, the drained pond turns into a beaver meadow, which is a relatively rare and important type of habitat for plants and animals. Healthy stream systems go through these successions where the landscapes evolve and flourish, where a permanent human-built dam can lock them into stagnation or steady degradation.

Three is lateral connectivity. Because beaver dams are fairly short and don’t have spillways or floodgates, they just push water laterally into the floodplain, creating a patchwork of meadows and wetlands. These slow, shallow flows help improve water quality, helping to process pollutants. But they only slow the water down to a point. Large floods that are important for sediment and nutrient transport downstream aren’t inhibited by beaver dams. On the other hand, human-made structures trap much more of the sediment and nutrients behind them, starving the downstream area. I have a whole video on this topic if you want to learn more.

But you can see why, in many cases, beaver activity is in sync with what we generally consider to be river restoration or uplift. They augment processes that are net positives like improving water quality, increasing the complexity of habitat, improving biodiversity, and facilitating groundwater recharge. Many biologists and environmental professionals introduce beavers as a form of environmental restoration. We even build fake beaver dams to achieve those benefits when the messy process of relying on rodents to do it isn’t feasible, called Beaver Dam Analogues.

At this point, human dams seem to be clearly on the wrong side of the dam engineering spectrum of good and evil. Beavers aren’t all good though. Because they change the landscape and ecology around them so substantially, they can be a serious nuisance. There are lots of cases where beavers flood areas we really don’t want flooded like pastures and agricultural crops. They clog up culverts, cause roads to overtop with water, damage infrastructure, weaken riverbanks with burrows, accelerating erosion, and they cut down an awful lot of trees. In Patagonia, beavers were brought for their fur in 1946. They’ve since become an invasive species, and with no natural predators, they’re completely altering the subantarctic forests.

If it weren’t so tragic, it would be kind of funny. Because the primary reason beavers build their dams is totally different from the reason we build them. Human reservoirs serve a lot of purposes: water conservation, flood control, hydropower, and recreation. For beavers, it’s protection. On land, wolves, bears, eagles, can catch a vulnerable beaver with ease. It’s a whole other story in the water. Beavers’ webbed back feet, paddle tail, breath-holding ability, and lodges with underwater entrances are stalwart defences for predators. A pond is like a castle wall, or really a moat for a slow, awkward waddling rodent. So, a pond in Patagonia is of little use to a beaver because there’s nothing to hide from; it’s just that creating them is in their DNA. They hear water and they have to dam it.

This all creates a pretty interesting irony: Humans generally build dams with grand, intentional goals: to power cities, secure water supplies for millions of people and crops, and to protect ourselves from floods, potentially saving thousands of lives. But the road to noble infrastructure is paved with degraded ecosystems and unintended consequences.

Beavers don't care about the bigger picture. They aren't trying to save the planet, filter groundwater, or build a sanctuary for migratory birds. They’re just trying not to get eaten. Their engineering intent is entirely selfish. And yet, because their structures are built within the boundaries of natural rhythms (porous, temporary, and connected to the floodplain), their purely selfish engineering accidentally adds to the richness, complexity, and biodiversity of a river system, at least if it happens in the right setting.

I joke about beavers being my professional rivals, but the truth is that they’re also kind of mentors. I give the little guys a nod of respect. We have the blueprints, the bulldozers, and the professional rigor. But with millions of years of practice over us, the rodents still have a lot they can teach us about how to reshape a landscape.

The Grid That Doubles the Strength of the Ground

[Note that this article is a transcript of the video embedded above.]

The Port of Long Beach in Southern California is the second-busiest container seaport in the United States, a major gateway for goods flowing to and from the west coast. Together with its next door neighbor, the Port of Los Angeles, roughly half a TRILLION dollars worth of trade moves through the port every year. To keep up with the staggering growth of shipping demand, the port is in a nearly constant state of improvement and expansion.

In the early 2000s, Long Beach was expanding the container storage yard at the Pier T Marine Terminal, but engineers hit a problem. The area they were expanding into was an old, disused dry dock, much lower than the rest of the yard. They were going to need a lot of fill to bring it up. There aren’t many more difficult paving challenges than a container terminal. You have constant heavy traffic of cranes, forklifts, and yard trucks, and the local fill materials that are available are junk: soft, water-logged silt scraped from the bottom of the ocean that has the structural integrity of a cake in the rain.

Traditionally, you would have to dig out all the muck, then haul in a literal mountain of expensive backfill, a process that would cost millions of dollars and take years. Instead, the engineers turned to a solution that looks more like a giant plastic accordion than a structural foundation. This 3D network did more than just hold the soil in place; it physically transformed those mushy dredged spoils into a high-capacity platform capable of supporting 100-ton machines and rows and rows of heavy container stacks.

The prefix “geo” gets a lot of work in the world of engineering. Just tack it on something semi-technical-sounding, and there’s a good chance there’s a product out there. Geotextile, geogrid, geofoam, and more. Today I want to show you how geocells work, and of course I built a little model in the garage so you can see it in action. I’m Grady and this is Practical Engineering.

When you think about engineering, it’s easy to conjure the typical materials we use: steel, concrete, bolts, gears, electrical components, pipes, beams, and so on. The first thing to come to mind typically isn’t dirt. Most geotechnical engineers prefer the alternative four-letter words of “rock” or “soil” to describe their medium of choice. But whatever they call it, they don’t really get to choose what it is. Where most engineers work with materials that adhere to stringent specifications, geotechnical engineers work with the materials nature gives them, which, by the way, are often terrible. So much of our built environment completely depends on our ability to predict and improve the behavior of soil and rock below our feet. And those materials are so entirely different from really anything else that engineers work with, particularly in how they fail.

When you walk on the ground, you don’t really even consider that it can quote-unquote “fail” under load. But just like a bridge has a maximum weight limit, every patch of dirt, sand, clay, or loam has a specific amount of pressure it can take before it physically gives up. Of course, most soils can squish a little bit, what we normally call settlement. But it gets worse than that. Soil can’t just disappear under a vertical load, it doesn’t crush like concrete or bend like a steel beam. Instead, it shears. The friction between the soil particles isn’t enough to resist the pressure in the soil, so they slide along each other in a plane. Generally, it looks like this: the soil moves down, then out, then up, causing anything on top to sink or tip.

This is called a bearing capacity failure. It’s one of the most important failure modes of soil, and it has to be considered when designing just about any type of structure (since everything sits on the ground). Engineers for buildings, dams, retaining walls, and of course, roadways have to contend with this limitation.

Spreading the load is the simplest fix. If you’re wearing high heels on a muddy lawn, you sink; if you put down a wide piece of plywood and stand on that, you don't. In engineering, this means building wide concrete footings to distribute the weight over a larger area of the ground. The problem is that concrete is expensive. For many structures, spread footings make a lot of sense. But there’s a reason not every roadway is paved with it.

Another option is just to replace the subgrade. If the dirt is mushy, you dig it all out and haul it away. Then you bring in so-called, “select-fill” an ambiguous term that really just means whatever material the engineer has determined will be able to support the load. For roadways, it’s usually high-quality, angular crushed stone. But you don’t want just large chunks of gravel; you need some finer particles to help lock everything together. That’s generally called “road base,” and it does most of the heavy lifting when it comes to both paved and unpaved roads. But how much of it you need is an important decision. The stress from a load spreads out with depth, dissipating as it reaches downward. The layer of road base has to be sufficiently thick that the remaining pressure from that “stress cone” is low enough for the native soil to support it without failing. The weaker the subgrade, the thicker the base. Sometimes that can be many feet, which requires an enormous amount of digging, hauling, placing and compacting.

Footing and subgrade replacement are good solutions in many cases, but there are situations where it makes more sense to bring outside reinforcements to the soil we already have. These are often known as geosynthetics: humanmade materials meant to make the ground work harder. Geosynthetic option 1 is geotextiles, industrial strength fabrics that go down like a big bedsheet. These are used across the construction industry, mainly to act as a physical boundary to keep individual gravel or rock particles from sinking into the mud underneath. In some situations, they’re used as filters to let water flow through without carrying soil with it. But they don’t have a lot of strength, and more importantly, they don’t really engage with the layer above or below. Geotextiles can move relatively independently of the soil, meaning if the loads are heavy enough, the whole thing will still rut and sag.

Option 2 is geogrid, which is exactly what it sounds like: a stiff plastic mesh used to reinforce layers of soil. And this basically works in the exact same way as the steel reinforcement used in concrete. The benefit over geotextile is pretty clear. Not only is the material much stronger, but the grid allows it to grab on to the material above and below. One of the first videos on this channel was about reinforced earth retaining walls. I built a cube of sand with layers of fiberglass window screen, and it could hold up one of my car’s wheels. A lot of the retaining walls you see in the world, especially on highway projects, use technology just like this. It really is pretty remarkable what it can do for even weak soils. But, that demo had multiple layers of it, pointing toward one of the limitations of geogrid: it’s inherently 2D. It can grab the soil near it, but the further away from the grid you get, the less it’s able to help. If you need to reinforce a thicker layer, that’s where option 3 comes in: geocells.

If geotextile is a sheet and geogrid is a mesh, geocell is a 3D structure. It’s basically formed from plastic strips welded together in a somewhat-honeycomb shape. And these strips don’t just engage with the backfill like geogrid; they encapsulate it. I feel like this is pretty intuitive. If I drop a heavy weight on this unconfined sand, it sinks in pretty deep. But if you confine the sand to a small area, it you get a lot less sinking. Like I mentioned before, the soil can’t just disappear under a heavy load, it has to shear and shift in order to fail, but when it’s confined, there’s nowhere for it to go.

If you put these confinement cells right next to each other, they start acting like the foundation footing we talked about earlier. It’s spreading the load out a lot more so, even with heavy vertical forces, you get a lot lower stress on the potentially weak soils below. And the advantages of this are a big deal. Road base may just look like dirt, but it’s actually a fairly expensive and specialized material. It has to be quarried, crushed, sorted, and sometimes hauled over a long distance. Not every place on earth has a rock or gravel quarry nearby that can produce the stuff in large quantities. Just through the added strength alone, geocells can significantly reduce the thickness of a base layer you need, but it’s more than that. Cellular confinement systems are a lot more flexible when it comes to the material properties of the infill. You’re no longer relying on that perfect interlocking of angular crushed rock and fines to distribute the load. The plastic does a lot of the work, potentially opening the door to use cheaper base materials or even using what’s already there

This isn’t really a new thing. The technology was pioneered by the US Army Corps of Engineers in the 1970s during the Vietnam war when there was a need to quickly build landing strips and roads on soft soils. Even NASA has done research on using similar technology to build stuff on the moon. One idea is filling sandbags with regolith, making them into a sort of building block, but there is research happening right now on the use of actual geocells. Maybe they would be lunarcells? When you consider how light this stuff is, and how compact it can be for transport, it just makes sense that it might be a good fit for extraterrestrial infrastructure.

Geocells can improve bearing capacity, but there’s another failure mode where they also shine. I released a video about washboarding on unpaved roadways. You’ve almost certainly experienced this yourself at some point: a dirt or gravel road has somehow formed rhythmic corrugations that jostle your vehicle and may even cause you to lose traction. I built this circular track with a motorized arm and wheel to simulate the phenomenon in real time. It's really fascinating how this works, and I definitely encourage you to check that video out after this. The one thing I didn’t really dig deep into in that video is potential solutions. Well it turns out that one of those solutions is geocells. So I got my trusty contraption back out. Let’s see if it really works.

There’s not a lot of scientific rigor in this demonstration. I just looked at a few academic papers where engineers have done these tests in a lab and tried to recreate the setup they used. So I wasn’t really sure how much to scale my geocells down. But I’m pretty proud of this design. I had them printed in TPU which is a flexible material kind of like the real thing. In the real world, the cells can bulge a little bit under load, engaging the strength of the material and the passive resistance of all the neighboring cells. Let’s see how my model geocells do on the track.

If you watched the previous video, you know that washboard formation is a very speed-dependent process. When the tires are moving slowly, it just leaves a rut, not a cyclical corrugation. But there’s kind of a threshold speed where it really starts to happen. In my model without the geogrid, that was about 35 rpm. I have a 10-to-1 gearbox between the motor and arm. So let’s start there.

I let this run for a while, maybe a hundred rotations, and you can see there’s basically no change in the profile of the track. Not even a rut. There’s a little bit of movement at some of the geocells, but this loose sand is basically a worst-case scenario when it comes to an unpaved roadway. There’s practically no situation where you would use loose sand for the surface of a road, except maybe on an actual beach, and there’s a reason tow trucks make good money in beach towns.

Even in this worst-case material, there’s nothing even approaching the look of a washboard. So, let’s bump the speed up a bit. This is about as fast as I trust this homebuilt contraption to go. You can see that it’s still remarkably stable. After quite a few rotations, one of the geocell sections is starting to “float” a bit, but normally this would be underneath a layer of cover. I left the geocells at the top so you could see them better, and because it’s kind of a worst case scenario for the test. Compare the high speed footage between the unreinforced and the geocell setup. You can see there’s no skip and push of the particles with the geocells. The sand stays mostly confined in that grid. I hope this demo doesn’t feel mundane. It’s a little hard to show at scale, but this is a tiny amount of plastic. It’s almost negligible, at least compared to the weight of the sand in the track. But it completely transformed the behavior of the system. A small amount of plastic makes a huge difference.

A while back, I made a video on geofoam and lightweight fills used on roadways projects to build tall embankments without overloading the foundation soil underneath. A lot of the feedback I got on that video was about the environmental considerations of putting styrofoam in the ground. People were just uncomfortable with that idea, and I totally get it. It’s something we definitely need to be thoughtful about. But it’s not something I really worry about with geosynthetics, and especially geocells. Most of the products out there are made of HDPE, the same material that we use for a lot of underground piping and conduits. It’s a relatively inert material, especially when it’s not exposed to UV light.

And actually, when you look at the totality of the situation, there are times when geocells provide a net environmental benefit. They can substantially reduce the amount of excavation needed at a site. They expand the options for materials that can be used as road base. They reduce haul distance for specialized fills because you can often use local materials instead. In some situations, you might even forgo pavement like asphalt altogether when it otherwise would have been justified, because now you can build a road that’s just as strong and maintenance free for a lower cost. Geocells can also use cleaner fill with less fines, making it more permeable compared to something like an asphalt parking lot, reducing the amount of impervious cover that contributes to flooding. Since you're more able to use the native materials, you also are less likely to generate truckloads of muck that you need to ship somewhere as well. And there are so many applications, from roadways to retaining walls, protecting steep slopes from wind and rain erosion, and of course, shipping terminals.

Geocells are not magic, and there are plenty of times where this technique isn’t the right choice, but they fit in a really interesting niche. The only reason engineering exists is because, in the real world, technical decisions aren’t simple. Engineers have a job because you have to weigh all these competing factors of cost, maintenance, material availability, traffic volume, and so on, and find the right balance between them. And I’m fascinated by innovations like this that find a little area of that multi-dimensional state space where all the constraints can be met in a clever way. Geocells are that for all kinds of geotechnical projects. Sometimes it’s not all about more massive concrete slabs or the brute strength of steel, but just a little bit of clever geometry and some dirt.

What's Under Your Feet in New York City?

[Note that this article is a transcript of the video embedded above.]

New York City is unlike any other city in the United States. It’s the most densely populated area in the country by far, absolutely packed with buildings, business, homes, and people. But for all that you can see walking the streets or flying overhead, there’s a whole other world of infrastructure that makes the city possible below the surface. What if you could peel back the paving and soil to see what one of my favorite authors, David Macaulay, called the city’s “massive root system.” Let’s take a tour underneath the Big Apple. I’m Grady and this is Practical Engineering.

Let’s pick a generic Manhattan intersection to start this journey. You’ve got cars, buses, bikes, buildings, sidewalks, traffic lights, fire hydrants, hotdog stands, manholes, and more. But open it up and there’s the whole other world below the surface. We’ll start with the water lines.

If you think about it, water pipes could run overhead like electrical wires. Digging trenches is a lot of work, after all. And fixing underground pipes is pretty disruptive to streets. Of course, New York isn’t unique in having them run underground. It’s standard practice across most of the world for a lot of simple reasons: water is heavy, so continuous support along the length of a pipe is a structural convenience. Water also freezes, so putting pipes underground below the frost line prevents them from freezing in the cold winters. It also protects them from a whole host of natural and human-caused hazards like car crashes and rogue parade balloons. That’s important because a broken water main can be a big problem.

When you see a huge rooster-tail of water spraying from the street, it’s easy to wonder why we need water mains to run at such high pressures. Of course, pressure helps move water through pipes to all the individual places where it’s needed across the city. But maybe more importantly, that pressure pushing out keeps contamination from getting in. You want any crack, hole, or break in a water main to be a one-way street. If anything’s moving from one side of the pipe wall to the other, it’s pretty important that it happens from in to out.

Like electricity, water lines typically run in somewhat of a grid pattern. This adds redundancy, providing multiple paths for water to reach a destination so that taking a line out of service doesn’t disrupt the flow to residents. It also makes sure that all the water in the pipes is constantly flowing. If you build a water distribution system like the branches of a tree, you end up with a lot of dead-ends where water can slow down or even stagnate, making it unsafe to drink.

New York City’s water system is famously gravity fed, with most of the source water coming from upstate at a higher elevation. It also requires no filtration because the source watersheds are fiercely protected to keep contamination out. But the City doesn’t just assume things are good. Dotted throughout the streets are more than 900 water sampling stations that let officials collect and test the quality of the water at the end of the distribution system to make sure it’s safe to consume.

If you could peel back the soil and look at the city’s water distribution system, you’d see water mains down nearly every street; shutoff valves used to isolate individual lines for maintenance or repairs, connections to street and wall hydrants where firefighters can hook up their engines, and service lines that tap into the mains to supply each individual building. You’ll notice that few utilities run under the buildings themselves. The main reason is that we need to be able to access them to fix them if needed. The other reason is that buildings often have their own underground structures, specifically piles, piers, or drilled shafts that serve as their foundation. I have a whole video on deep foundations if you want to learn more after this.

Unlike water pipes, it is pretty typical to see electrical distribution lines running overhead on utility poles everywhere across the globe. You won’t see this in most parts of New York City, though. Roughly 85 percent of the electrical lines are underground. Part of it’s about looks: lines clutter up the space and require dedicated rights of way that limits the use of that space. Another part is safety: keeping people and vehicles free and clear of distribution level voltages. And, of course, there’s reliability. When a heavy storm takes out an above-ground utility pole in a suburban neighborhood, the ensuing power outage is an inconvenience. That same outage in Manhattan could affect a lot more people.

There’s a lot of confusion about underground electrical service. You can kind of divide the grid into three distinct categories defined by voltage ranges: there’s transmission (where power moves over very long distances at hundreds of thousands of volts), distribution (where it’s carried throughout a populated area at a a few thousand volts), and finally service (the voltage at the plug). Putting service lines underground is pretty straightforward. You might even have an underground line at your house running to a lamp or a detached garage. Putting transmission lines at hundreds of thousands of volts underground is a pretty extreme engineering challenge because of insulation, heat buildup, and capacitance. Undergrounding distribution lines lies somewhere in the middle.

One of the big upsides of running lines above ground is the availability of air. Air is free and it works pretty well as an insulator if you keep enough space around energized conductors. You only need actual insulators at the pole. Putting lines at tens of thousands of volts underground requires pretty expensive insulation that prevents arcs to ground or other phases and resists the effects of water, a hazard that is inevitable for every underground utility.

We often call an electrical interconnection a “grid,” but that term mostly applies to the high-voltage bulk power system covering whole states or countries. It’s not really a good description at a city scale. Most urban areas use what’s called a radial system for distributing electricity, which is more akin to branches of a tree than a mesh. For a single-family residential home, you might share a transformer with a few houses. That connects to the distribution feeder, and you can follow that line all the way back to the substation. Each feeder is essentially a one-way dead end for power flow. There may be a crossover somewhere for redundancy, but it’s not an inherent part of the radial architecture. In New York City, it’s totally different.

Throughout the five boroughs, New York City operates about 70 separate so-called “secondary networks,” each of which is served by somewhere between 8 and 28 feeder lines from an area substation. Rather than individual transformers that serve one or two buildings, there are network transformers dotted around the city, usually in concrete vaults belowground, each connected to one of the redundant feeders from the substation. Because they’re underground, these transformers have to be capable of operating while fully submerged in water.

Those transformers drop the voltage to the service-level where a grid of conductors spread out to all the buildings in the area. These are true networks, actual grids of service-level voltage with multiple redundant pathways for energy to take (not like the branches of a tree at all). And there’s another way it’s not quite like the rest of North America.

A typical service transformer in the US gives you “split phase power”. It takes one phase from the grid and provides two energized lines we call hots. Each hot leg has a voltage sine wave between neutral that is 180 degrees out of phase. So between one hot and the neutral, you get 120 volts. That’s a typical wall outlet. Larger appliances and EV chargers use both hot lines to get 240 volts. In New York City, the service networks are different. They use a three-phase system just like the rest of the grid. Each phase is offset by 120 degrees. Most individual apartments or houses get two of the three phases. So in the city, you still get 120 volts from hot to neutral, but you only get 208 volts between the hots. Most large appliances are designed to work on both 240 and 208 volts in the US because of this mixing and matching with single phase and three phase power.

Larger buildings like skyscrapers usually get their power at a higher voltage directly from the feeder and use their own transformers on maintenance floors to provide service throughout the building. So ConEdison maintains basically two power grids, each underground, one for the feeders between 13,000 and 27,000 volts and one for low voltage service. They both run through ducts that travel below streets and sidewalks and are serviced in the thousands of underground manholes and vaults throughout the city. With everything protected underground with lots of redundant paths, New Yorkers enjoy one of the most reliable electrical services in the country, but obviously, that reliability comes at a price. The “service network” architecture is one reason why New York City has some of the highest electricity prices in America. But wires aren’t the only way New Yorkers get power.

Looking back at our generic intersection, you’ll see something that is not present in most other cities, and certainly not at the scale we see here. New York City has a district heating network, offering steam as a public utility. It’s not as extensive as the power system, with pipes mainly confined to Manhattan, but it is the largest of its kind in the world by far. There are about 1,500 customers with steam service delivered through the underground lines. It’s mainly used for heating buildings and for hot water, but it can also be used for sterilization in hospitals, cleaning dishes at restaurants, for presses in dry cleaning facilities, and somewhat counterintuitively, for air conditioning through steam-driven compressors.

Of course, there are a lot of engineering challenges with running steam pipes underground. Thermal expansion is a big one. If you shut one of these lines down, it changes in temperature by roughly 300 degrees Fahrenheit or 170 degrees Celsius. You need regular expansion loops or heavy-duty slip joints that can absorb the physical movement created by those huge swings in temperature. You also have to deal with condensation. Some of that steam naturally condenses into water, and if you don’t get it out of the pipe, that liquid can act like bullets inside the high speed steam flow. I have a video on that from way back in the day if you want to learn more. Steam traps can discharge condensate while keeping the steam inside the pipe. Occasionally you get a steam leak or just a spot where groundwater is coming into contact with the hot pipes, creating a cloud of steam through a manhole to the street. ConEd puts these orange smokestacks to divert the vapor away from the public until they can fix the underlying issue. Those steam lines are insulated to keep the heat in, and usually deeper than other utilities to avoid heating up the surface or the other lines.

One of those lines you definitely don’t want to heat up is natural gas. A large percentage of buildings and households in New York City use natural gas for heating, cooking, and hot water. Some of the gas lines are low-pressure mains that connect directly to buildings through the meter. The newer lines run at higher pressures and require a building regulator. For safety reasons, these are often installed outside, so they’re easy to spot. New York has put limits on when natural gas can be included in new construction, so these lines might become a thing of the past, joining other abandoned lines underground. Because of the cost and complexity of decommissioning utilities, it’s not uncommon to simply abandon them in place. In fact, there are plenty of lines underground that aren’t doing anything at all, including a once-sophisticated pneumatic tube mail delivery system, which spanned 27 miles and delivered mail throughout the city until it was shuttered in the 1950s when the maintenance costs started outweighing the benefits.

Also like other places, New York City needs telecommunications: telephone, cable, and fiber lines. And like all their other utilities, these are running below the streets, weaving their way around everything else near the surface. But unlike a lot of the utilities, telecommunications are somewhat standardized. That’s because, in 1891, the city granted a franchise to a company called Empire City Subway. That name uses a more generic meaning of subway, an underground path, rather than the much more famous one you’re thinking of (which we’ll get to soon). Empire City Subway’s whole job was and still is to build and maintain a massive network of underground utility ducts. Anyone wanting to run telephone, cable, or fiber lines through Manhattan or The Bronx has to rent space inside those conduits rather than digging their own paths.

But even with that standardization, you can see that things are starting to get pretty cramped. In many cases, especially older parts of the city, very little of this infrastructure was planned out comprehensively. As each service was installed, it had to find space among the others. And because much of it is pretty old, most streets are either imperfectly mapped, or not mapped at all. Making repairs is its own kind of treasure hunt. Utilities workers and engineers often just call it “the spaghetti” because of how big a mess it can be. To combat their proverbial pasta problems, the City is working on a comprehensive, 3D database of underground utilities to help with construction, repairs, and emergency response, but it’s an enormous challenge to combine records that, in some cases, can be more than 100 years old.

Vacuum excavation has made a big difference in being able to dig around utilities safely. Water or compressed air breaks up the soil and the truck can suck it up. This makes it a lot quicker to expose buried utilities without damaging them, especially compared to an excavator or backhoe bucket. Often you have to support utilities from above during repairs to keep them from sagging down and flexing, which could lead to breaks and service disruptions. Older materials used for pipes and ducts like cast iron and vitrified clay are brittle and relatively weak. Taking away that continuous underlying support to repair something below requires a lot of care. This is just a ton of work compared to cities where there’s less density in the buried utilities. But that’s still not everything under the street.

Of course you have sewers that carry wastewater away. They’re usually deeper than the other utilities, and for a few reasons: Sewers rely on gravity to maintain flow, so the slope carries them further below the ground. You also generally want to have them below water lines, just to be absolutely sure that a sewage leak doesn’t find its way through the soil toward fresh water. Of course, all American cities have sewer lines, but New York has some big ones. They’re not quite like the cartoon-style tunnels with handy ledges that act like walkways for crime fighting reptiles, but some of the hand-laid brick pipes are still in use. This is a big city that produces a lot of wastewater, and when all those lines start to converge and concentrate toward the treatment plants, it takes big pipes to carry it all. They may not look quite like the cartoons, but some actually are cavernous enough to walk through standing completely upright.

These big old sewers are made even more complicated by stormwater. Much of the sewer system in New York City was built before modern environmental rules. The goal back then was to get it out of town, not to treat it, so it didn’t really matter that the stormwater was diverted into sewage pipes. It was a good thing, actually, because it diluted the sewage that was just being discharged directly into waterways. The problem is that, now, it matters a lot. On rainy days, the treatment plants don’t have the capacity to clean up both the sanitary sewage (the stuff that comes from sinks, toilets, and showers) and all the stormwater runoff from streets and buildings. So they still have to discharge untreated sewage on occasion. The city has roughly 400 outfalls where these “combined sewer overflows” occur.

Of course, dumping raw sewage is highly restricted under modern laws. You can’t pollute natural waterbodies without consequence. But, it’s a practically impossible problem to fix, at least all at once, since these systems were built this way over decades and decades. Instead, the city operates under a consent decree with the state that basically says, “We won’t fine you for the overflows, but you have to implement a plan that puts a stop to this eventually.” Lots of major US cities are in the same boat. And New York City really has been working to fix it. From curbside rain gardens to massive underground retention tanks like the one at Newton Creek, New York is slowly chipping away at the number of sewer overflows. In newer parts of the city, the sanitary and storm sewers are entirely separate. Catch basins line the streets where stormwater is diverted to underground pipes. That’s just one more separate system of underground utilities to find space below these crowded streets. Go a little deeper and there’s more.

New York City is famous for its subway, one of the largest and busiest rapid transit systems in the world. They mostly run in cut-and-cover tunnels below the streets. This is a simple idea: dig a trench down from the surface of the road, construct the floors, walls, and roof of the tunnel, then backfill and replace the road on top. It’s disruptive to the street, but much simpler and more cost-effective compared to tunneling methods that don’t disturb the surface. There are many instances of multi-level tracks where one line crosses another that required elaborate steel framing to support everything during construction. Deeper sections of the subway, like those under the East River, required alternative methods. The most recent projects have used tunnel boring machines, which are much more expensive, but increasing the depth helps avoid existing utilities and minimizes disruption at the street level. And you also have the ventilation structures and the elevators and escalators and stairs that connect between subways and the surface. The stations themselves are mostly subterranean, since that’s where the trains are.

Of course, there are tunnels for cars and trucks below the surface in New York City too that pass under the East River and Hudson River. Sometimes it makes more sense than building a bridge, and other times it’s necessary for grade separation to keep traffic moving.

Go even deeper below the street, and we’re back to water. New York has three primary tunnels that bring the fresh water into the city from upstate. Tunnel 3 is one of the largest and most expensive capital construction projects in the City’s history. Started in 1970, it’s still under construction and will be for at least another decade. It’s really deep, roughly 650 feet or 200 meters below the surface. That’s nearly half the Empire State Building! That extreme depth is to avoid the underground traffic jam of all the other utilities we’ve talked about, but also because it keeps the tunnel in hard rock that’s better able to withstand the monstrous pressure inside.

The underground of New York is almost a city within a city. We kind of get used to having all these utilities and services that it’s easy to forget the physical space they all take up and the work that goes into installing and maintaining them. Besides what they provide us, there are little signs at the surface as reminders: manhole lids, puffs of steam, metal grates where you can barely catch a glimpse of the machinery below. And then sometimes the reminders are a little more in your face, like when an intersection shuts down for a careful and intricate replacement project, weaving new pipes between those that might be a century old. These kinds of disruptions are hard to love, but they do give you a chance to appreciate everything that’s underneath.

The Bizarre Flaw in the New Orleans Levees

[Note that this article is a transcript of the video embedded above.]

What happened in New Orleans in August 2005 wasn’t a natural disaster. Don’t get me wrong; Hurricane Katrina was a big storm. By many measures it was one of the strongest hurricanes to ever crash into a United States coastline. It’s in the top 10 for largest diameter, lowest central pressure, highest integrated kinetic energy and more. Storm surge along parts of the Gulf Coast crested at nearly 28 feet or 8.5 meters, the highest in US history.

There really are no words to describe the devastation Katrina caused, particularly in New Orleans and the surrounding areas. Roughly four-fifths of the city was flooded. In some areas the water rose above the rooflines of houses. The storm shredded the fabric of the city, and to this day, more than 20 years later, the population of New Orleans has not recovered to its pre-Katrina level.

If this were just a story about a major storm overwhelming the flood defenses, it wouldn’t undercut the tragedy, but it would be understandable. I think we generally recognize that these kinds of systems don’t offer absolute protection. The levees and floodwalls around New Orleans were a sprawling collection of structures built over many decades. It wouldn’t be out of question for a storm of the century to simply exceed what those protections were designed to handle. But that’s not what happened.

There were about 50 individual locations where levees or floodwalls were breached during Hurricane Katrina. Many of those were situations where the structure was overtopped by storm surge. But out of those fifty, only three of those breaches accounted for the majority of the flows that submerged the heart of New Orleans and led to nearly half of the disaster’s total fatalities and economic damage. All three of those breaches happened at surge levels below what the floodwalls were designed to manage. And I built a model in the garage to show you why.

This video is based on a chapter from my new Book, Disasters By Design: How Engineering Failures Shaped the Modern World, which you can pre-order right now at the links below. The book is all about engineering disasters through history and what we learned from them, and I really hope you’ll consider picking up a copy. I’ll tell you more about it at the end, but first, I want to show you this model. I’m Grady, and this is Practical Engineering.

The land we call New Orleans has a history stretching way beyond its 1718 founding or the birth of the United States. Originally inhabited by tribes like the Chitimacha, this area has always been a complex intersection of human settlement and flooding. It is strategically located on the Gulf Coast at the mouth of the Mississippi, the largest river in North America. But that prime location also comes with challenges, namely that the city’s exposure to tropical storms is extreme, and, being a river delta, much of its development happened on top of historically soft, marshy, and low-lying terrain.

In 1965 Hurricane Betsy rocked New Orleans, flooding more than 150,000 houses. At the time, it was one of the worst hurricanes in American history, earning the nickname “Billion Dollar Betsy,” because it was the first storm with damages in excess of a billion dollars. In response, the federal government, through the US Army Corps of Engineers, made a massive investment into the existing hurricane protection system around New Orleans. But they ran into an issue.

A levee is an earthen embankment meant to hold floodwaters back - kind of like a dam but parallel to coastline or a river instead of across it. We know that soil is naturally unstable on steep slopes. So levees have this trapezoidal shape in cross-section with gentle, stable slopes on the waterside and landside. Many of the levees in New Orleans were built decades before Betsy, and over that time, development encroached right up the toes of the structures. You can see the problem. If you want to raise the embankment to provide more protection, you need more space.

The solution they came up with is the “I-Wall” (not related to the eye wall of a hurricane). Steel sheet piles were driven into the top of the levee, creating a vertical floodwall on top. They got more height and more protection without the need to condemn property for a wider footprint, which was a major win.

All this flood protection infrastructure turned New Orleans into essentially a bowl. The levees keep the floodwaters out, but they also trap rainwater (or any other water) in. So in addition to the levees and floodwalls, the city is dotted with pump stations that move drainage within the city out of the bowl. The heart of this system is the trio of primary outfall canals that run south to north, carrying rainwater from New Orleans’ interior basins out to Lake Pontchartrain: the 17th Street Canal, the Orleans Avenue Canal, and the London Avenue Canal. Where most canals, rivers, and creeks in cities represent a topographic low point, these canals are somewhat of a high point in New Orleans. Runoff within the city naturally flows in street gutters and underground storm drains, eventually reaching one of the massive pumping stations associated with each canal. Those stations raise the water just a little higher than Lake Pontchartrain, and then gravity takes care of the rest, with the water flowing northward out of the city and into the lake.

The problem is that Lake Pontchartrain isn’t really a lake; it’s an estuary with a direct hydraulic connection to the Gulf of Mexico. It functions as an arm of the sea, which means it’s subject to the same tidal forces and catastrophic storm surges as the open ocean. When Hurricane Katrina pushed up a wall of water, the outfall canals stopped acting like drainage ditches and turned into conduits penetrating deep into the heart of New Orleans, inviting the weight and pressure of the sea miles inland. You can see why the breaches along these levees were particularly serious.

In the morning on August 29, just after landfall as storm surge began to peak, three failures along the outfall canals happened in quick succession. Maybe the most well-known was the breach on the 17th Street Canal levee and I-wall. That morning, water was several feet below the top of the wall. This structure should have been more than capable of holding back the surge, but a huge chunk of the levee shifted, leaving a 450-foot- (or 140-meter-wide) gap for a torrent of storm surge to flood the Lakeview neighborhood and surrounding areas.

It’s a little tricky to understand how something like this could happen, so I’ve built a model of a levee and I-wall in my garage to show you how this works. This is just a narrow tank that I built out of sheets of acrylic. On the left side, I have a pump that will move water into the system. This area represents the canal side where storm surge flowed in from Lake Pontchartrain. I have an overflow that will keep the water level constant. I don’t know how well this shows up but the soil that I’m using for the levee and its foundation is a blend of ground up plastic and sand. And then on the right I have a drain that kind of represents the pump stations that move surface water out of protected areas to keep them dry.

This took a few tries to get to work right. I’m sure anybody who’s built a garage storm surge simulation can relate. First the colored liquid dyes weren’t persistent enough. Then my I-wall sprang a leak. So I’ve got solid dye tablets in the soil and the leak hopefully fixed. And you can see that there’s not much happening when the water level in the canal is low. But watch what happens when I turn on the pump. The levee and I-wall hold the water from the canal back, keeping the homes and businesses on the land side protected. But keep an eye on the dye in the soil. I’ll speed this up a bit to make it clearer. Water is still making its way toward the landside of the levee, not above ground, but in the subsurface. It might seem obvious that this would happen, but it’s not a phenomenon we get to see with our eyes very often. And it makes one thing about flood protection systems abundantly clear: these levees aren’t just piles of dirt. They are engineered structures that have to consider some fairly unusual and unintuitive loading conditions. And the behavior of this subsurface flow is absolutely critical to their performance.

It’s easy to follow the flow in the model, but it’s harder to visualize the water pressure, so I coded this little companion toy simulator to go with my model. Thanks to Dan Shiffman at the Coding Train for teaching me javascript and P5. It takes a minute to converge on a solution, but when it does, it plots the pressure contours in the subsurface. Differences in pressure are what drive flow, so these contour lines are perpendicular to the flow lines you see in my model, and I gave the pressure a color gradient to make this more clear. You can see that, even though it’s not as high as the areas closer to the canal, there’s still water pressure in the subsurface on the protected side of the levee. And that’s a problem for a structure trying to hold back an immense wall of storm surge.

Soil grains work kind of like a pile of marbles. Squeeze them and they lock together, increasing their strength against shear. You can imagine if you were able to inject water between those marbles, it would push them away from each other. As a mass altogether, they would be a lot less strong, and it works the same way for soil. Engineers have to take this into account. The calculations when designing a levee aren’t just about how hard the water pushes against the structure, but also how hard it pushes inside the foundation. If the soil doesn’t have enough strength to resist the force of the water pushing sideways while the subsurface water pressure is also pushing the soil particles away from each other, it shears along a plane of weakness. That’s exactly what happened at the 17th Street Canal. The thing is, it happened at a lower water level than anybody expected, And here’s why:

It is no small task to assess the complex subsurface conditions along miles of levees and floodwalls. During the improvements made in the wake of Hurricane Betsy, geotechnical engineers drilled hundreds of boreholes, collected samples, and tested their strengths in labs. Here’s a look at that data plotted as a function of depth.

Now, when you’re designing miles and miles of levee, it’s not feasible to analyze every single inch. Instead, you pick representative sections, and you really want those sections to be the worst ones. That’s just efficient engineering. You know if the design works in the weakest spots, it will work everywhere. But for some reason, that’s not what was done. Knowing that kind of fundamental principle and looking at this data, what would you choose as your representative soil strength for design? If it were me, I might choose something like this - assuming a few of the test values are outliers that don’t represent the bulk properties, but for the most part, being conservative by assuming strengths at or below what was measured in the field. For some reason, the Corps of Engineers drew their line here, assuming the soil was stronger than a majority of the measurements. And what’s worse is that those measurements were taken along the centerline of the existing levees, where the weight of those structures had artificially consolidated and strengthened the soil. Outside of that centerline, the soil was actually much weaker. These design decisions led to the levee failing at water levels lower than the structure was designed to handle.

What made things even worse was a critical flaw of the “I-Wall” idea. It seemed elegant on paper: an efficient way to get more height out of a levee without widening its footprint. But when they were actually put to a true test during Katrina, those walls bowed out, opening a narrow gap deep into the levee’s foundation.

Let’s look back at the simulator to see the difference in pressure. You can see that with the so-called “water-filled gap,” you get more pressure in the foundation of the levee. This also shortens the length of soil shear needed to cause the levee to fail. All these factors resulted in the levee on the 17th Street Canal failing at a level below what it was designed to withstand. The same thing happened nearby on the north side of the London Avenue Canal. And actually, that failure was exclusively the result of the flaw in the I-Wall design. If the wall wouldn’t have deflected to open the gap, that failure wouldn’t have happened at all. But that wasn’t the only failure that happened on the London Avenue Canal, and the one on the south side was even more egregious.

This levee was largely the same as on the 17th street canal with the embankment and I-wall. And it had the same situation with water seeping through the foundation below. But there was one critical difference. The levee rested on a thin layer of marshy clay above a thick bed of permeable sand. Again, it’s hard to show pressure in the garage model, so let’s take a look at the simulation. You can see when I use a homogenous material for the soil of the levee and foundation, you get fairly nice, even spacing of the pressure contours. Let me make a change to add a layer of less permeable material like the London Avenue Canal south and run them side-by-side. With the marsh layer, you get a really different result. The pressure contours stack up right where the water comes out on the protected side. Geotechnical engineers would say there’s a steep exit gradient. And here’s why that matters.

Let’s just take a block of soil and look at it more closely. Because of the steep gradient, the pressure at the bottom of this block is different from the pressure at the top. That difference in pressure means there is a net force pushing it upward. If that net force exceeds the weight of the block itself, it’s going to move. And I can show it in my demonstration, even without the marsh layer.

I’ve reset the demo, but I added an extension to my overflow so the water level will be higher on the canal side now. Even though the pressure gradient through the subsurface is evenly spaced (since I’m using a homogenous material unlike the layers in New Orleans), it’s still steeper just because there’s a bigger difference between the water levels on either side. And because I’m using a lightweight plastic blended into the soil, the so-called “critical” exit gradient is much lower than it would be for denser sand. And it’s easy to see that we’ve exceeded that critical gradient. The particles are dislodging as the seepage comes out of the ground on the protected side. The erosion just keeps getting worse until the levee on the protected side fails. Eventually, the I-wall loses support, and you get a total failure.

This is exactly what happened at the London Avenue Canal South. As the storm surge rose, water seeped through the sand, building upward pressure on that marsh layer. Eventually the pressure got high enough that it literally blew a hole through it. With the sudden release of all that built up pressure, the subsurface water was able to erode a mountain of sand from the foundation. There are pictures of this - a neighborhood covered in what’s basically beach sand. All that erosion caused the I-wall to collapse, opening a wider hole in the levee and letting the full force of Lake Pontchartrain through.

This was not some unknown issue. The Army Corps of Engineers analyzed the seepage forces through the levee when they built the I-walls. But, inexplicably, they left that marsh layer out of the calculations. One of the investigation reports said it clearly: “if a more rigorous analysis had been performed at the time of design, the potential problem would have been predicted and corrective action taken.”

Those three breaches created a majority of the problems in the center of New Orleans, but there were a lot of levees and floodwalls that overtopped in the surrounding area during Hurricane Katrina. One of the worst cases was at the Industrial Canal, where the wall was lower than intended because of ground subsidence and confusion about elevation benchmarks. In that case, the water spilling over the top of the wall eroded the soil, causing the levee and wall to fail. And that gets to the heart of a very complicated problem with flood protection: you can’t build a 100-foot tall invincible wall around the city. You have to draw a line somewhere below absolute safety to a level that’s economically feasible. That means there are going to be potential storms that overwhelm the system, which makes designing levees even more complicated.

Let’s make a graph of depth versus damage. In an area with no flood protection, you might get a curve like this. Just a little damage at first, when depths are low, and those damages increase as flood depth grows. If you build a levee, of course you basically eliminate those damages below the protected level. And above the level, the curve stays the same. Once a levee overtops, even the protected areas flood, so there’s no change there. But over time, a funny thing happens. Before the levee was installed, the lowest places weren’t heavily developed because flooding happened regularly and people generally knew and understood the risk. But once those low areas are protected, and some time with no flooding at all passes, people build stuff in them: homes, business, et cetera. We put a lot of valuable stuff on that previously mostly empty land. So where the depth-damage curve with the levee started like this, now it looks like this. Just a tiny bit of overtopping means an enormous amount of damage because the former floodplains have all been built out. So if you look out toward the long-term, the damages you can expect from flooding over many decades is actually much more than what would have happened if the levees were never built. This concept is sometimes known as hazard creep, and it’s the thing that no one recognized in New Orleans. The shadow of the flood protection system had, in some ways, only made things worse.

In that way, not only did the engineering (particularly along the outfall canals) fail to protect the city (The Corps of Engineers estimated that roughly two-thirds of the deaths during Katrina would not have occurred had the levees and floodwalls not failed.), they generally created a situation where the damage and loss of life was much worse than it would have been in an unprotected area.

Of course, that is really just one tiny piece of the story of Hurricane Katrina. The investigations that followed filled volumes and volumes, tens of thousands of pages, just on the engineering issues alone. When you add in the challenges and controversies associated with the evacuation, rescue operations, and recovery efforts, you could spend your entire life trying to wrap your head around this one event. I know, because I spent a good chunk of my life researching it while I was writing my new book, Disasters By Design. I think there is a lot we can learn from past mistakes, and I really hope you’ll click on the link below or head to your local book store and pre-order a copy.

That’s the thing about engineering disasters: When they first happen, journalists scramble to tell the story with whatever information they can find; politicians and pundits offer premature reactions; and then, slowly, public interest wanes. By the time rigorous investigations are carried out and a consensus has formed around what really happened, most people have already moved on. So, in lots of cases, the documentation of an engineering disaster is this rigorous and complex forensic report produced months or years after the fact that a handful of experts read, and not much else.

Here’s the problem with that: For better or worse, a great deal of humanity’s most important lessons in engineering have been learned the hard way. Humans are capable of, and maybe even prone to, making stupendous errors in judgment, calculation, and prognostication. The only way we have reached this era of remarkable advancement and connectivity is through learning from those errors and striving to avoid repeating them. All the people who studied, investigated, and documented the engineering failures of history did so because they wanted, in some way, to help prevent us from making the same mistakes again. But they often wrote to an audience of other engineers, meaning many of humanity’s most important lessons are tough to wrap your head around unless you have the context, background, resources and education to parse and understand those sometimes impenetrable reports.

I wrote this book because I wanted to bridge that gap. I’ve spent a career working on infrastructure projects as a civil engineer and another career breaking down engineering concepts to people like you on YouTube. I wanted to combine those experiences and curate a collection of stories throughout history that I’ve found to be the most enlightening.

But these aren’t just stories about disasters and tragedy. They’re also stories about what we learned as a result, and how we transform tragedy into hope. Because some of the most important aspects of the built environment and the industry that surrounds it came as a result of something terrible. In that way, the stories of engineering failures are the stories of engineering itself and how it’s evolved over time.

New Orleans is a perfect example. The city’s flood protection system looks a lot different than before Katrina: There are huge gates on the canals to keep Pontchartrain from backflowing into the city. Levees have been raised and strengthened. The flawed I-walls were taken out. There’s a huge new storm surge barrier in Lake Borgne. But it wasn’t just steel and concrete structures that resulted from the storm. The Corps of Engineers completely rewrote the manual for levee engineering, incorporating the lessons learned. They did a nationwide review of existing I-wall projects. The federal government implemented the National Levee Safety Program to support owners and enhance public awareness. Lots of those lessons were incorporated into the International Levee Handbook that is used around the world. It changed so much in engineering, disaster response, and even government. It’s tricky to see how all this plays out, because it really just means quiet. There won’t be any news stories that some levee didn’t fail today. And that’s why I wrote the book.

To really get a sense of how the field of engineering has evolved over time, from antiquity to today, you have to look at the disasters that precipitated the big changes. And I really think having those major events laid out in order and explained so curious people can understand them, with lots of illustrations, is the right way to comprehend engineering history. So yeah, the book will be published in March of next year. I’m super excited to tell you about it, and really proud of how it came out. I know that’s a long time from now, but pre-orders are really important with books so they know how many to print. I know I ask you to click a link at the end of all my videos to support the channel via our sponsors, but this time, it means so much more. Grab a first edition for yourself. And if you have friends, professors, a significant other, or family members who are interested in engineering, it would make a great gift. Your support and encouragement has made this book possible, and I’m so grateful to have you along for the ride. Thank you for watching, and let me know what you think.

How to Demolish a Bridge

[Note that this article is a transcript of the video embedded above.]

In 2022, the world got a very cool new bridge. Two bridges, actually. The Iowa-Illinois Memorial Bridges carries Interstate Highway 74 over the Mississippi River between Moline, Illinois and Bettendorf, Iowa in the “Quad Cities” area. It’s a gorgeous pair of structures with the basket handle arches carrying each deck over the main span of the river. But even after they were finished, Iowa DOT had a problem. The two old bridges were still right there, also crossing the Mississippi River. And even though they were kind of cool looking, they just couldn’t stay. The bridges were already in poor condition, and without extensive ongoing maintenance, they would continue to deteriorate, posing a danger to the public, affecting the sensitive environment along the river, and even disrupting this critical shipping artery. The old bridges would have to come down.

Demolition, on its face, seems kind of easy. For a billion-dollar-bridge replacement project, the demo part feels almost like housekeeping. Smash the structure down or blow it up, then just pick up the pieces. No engineering needed. The truth is that it’s anything but. Demolition engineering, in many ways, is even more complicated than designing a new structure, and the I-74 bridge is the perfect case study in why. And don’t worry, there are explosions at the end. I’m Grady, and this is Practical Engineering.

The original I-74 bridges, with just two lanes each, were way overdue for an upgrade in capacity. This is actually a problem they faced and managed to overcome once before, many decades back. Despite looking like twins, the original pair of bridges were built a generation apart. The first span was completed in the mid 1930s, but car ownership and traffic exploded after World War II. Engineers decided that the best way to increase capacity was to build a nearly identical bridge right next to it. That new bridge was opened in 1959.

Neither bridge was ever intended to meet interstate standards; they predate the interstate system altogether. And yet, they found themselves carrying interstate-levels of traffic, way beyond what the designers in the 1950s, and especially the designers back in the 1930s, had considered. The lanes were narrow, there were no shoulders so they required a lower speed limit, which bottlenecked traffic on I-74.

Size isn’t everything, though; the bridges were also just physically wearing out. Like an old car, it eventually got to the point where the cost of replacing the bridges was outweighed by the constant maintenance and threat of disaster. In 2012, Transportation Secretary Ray LaHood toured this structure, reporting back that it was, quote, “one of the worst bridges I’ve seen in America.”

You would think that already being close to falling down would be to their advantage when it comes to demolition, but it’s quite a bit more complicated than that. These were big bridges with 3 types of structural designs. There’s these three span continuous truss units over the old non-navigable part of the river. There’s the deck trusses that kind of act as connectors. And then you have the big 3 span suspension section.

I’m sure you want to see the explosion, and I promise we’ll get there, but it’s basically the last step. Of course, there are lots of cases where it makes sense to just blast a structure down right away. You end up with a pile of rubble that you can manage with regular construction equipment. It can be much quicker, easier, and safer than dismantling a structure piece by piece, but it’s rarely true for bridges. Of course you’ve got the water that complicates things. Removing debris from below the water line is challenging. Long reach excavators can sometimes handle the smaller stuff, but you often need divers to rig the big stuff to be lifted out by cranes. That’s dangerous and difficult work. You also have shipping traffic to consider. This stretch of the Mississippi is a busy part of the inland waterway system, and closing it to clean debris out of the channel is a disruptive task. The other thing making it tricky, in this case, is the environment. There are endangered mussels living in the non-navigable channel below the continuous truss spans, so the demolition team couldn’t use blasting or even temporary supports in that part of the river. The only option was to dismantle the bridges more carefully and thoughtfully. Step one is to get the deck off.

The strategy here was to sawcut all the concrete into pieces small enough to move with construction equipment. An excavator with a slab crab attachment could lift each panel off the steel structure, swing around, and pass to a wheel loader to carry it off the end. Sounds simple, but it had to be done pretty carefully. Cutting the concrete into panels like this means that the reinforcement is cut too. And this is just a cool part of demolition engineering: using calculations and analysis not to design something new but to answer a tricky question like, “Can these concrete panels support the weight of a 35,000-pound excavator?” The answer was more complicated than just a simple “yes.” So the engineers imposed pretty strict positional requirements for the demolition equipment, in most cases making sure that the tracks of the excavator were always directly above the stringers instead of relying on the concrete deck panels to act like beams and transfer the weight.

Another challenge on the suspension span was asymmetric loading. It’s easiest to use the bridge to dismantle the bridge, systematically working your way toward either end. The problem is that if you take all the weight off one section of the bridge while it’s still remaining on other parts, the trusses are going to bow, the towers are going to deflect, and you could actually fail the bridge prematurely. It’s just like re-racking weights at the gym. If, instead of alternating, you take everything off one side of the barbell, you might have a bad time. So, on the suspension bridges alone, the deck removal was this multi-stage process.

Some slabs were removed with the excavator and loader. Others were popped up and left in place as counterweight to be removed by a crane later. It’s a lot more work with a crane (and slower), but it was the only way to get the deck off in a symmetrical way to avoid overstressing any part of the bridge.

Once the concrete deck was off, the contractor could start removing the steel trusses, beams, and stringers that make up the bridge structures. And this gets pretty tricky too. You can’t just go cutting up a bridge willy nilly. This is like jenga on hard mode with very high stakes, and that requires some structural engineering. On the continuous truss section, the demolition team wasn’t allowed to install temporary supports to avoid disturbing the mussels. So instead, they floated in the support on a barge. This allowed them to safely cut the trusses into pieces small enough for the crane to handle without causing a collapse.

The suspension spans were even more complicated. You can imagine how dangerous it is to cut a piece of steel that’s under significant stress. As soon as the member is severed, it could cause sudden movements and load redistributions within the bridge. Those cuts have to be carefully sequenced. When you’re actively weakening a structure, each step changes the stresses, shifting them around and altering their magnitudes and directions. You have to check each step before you do it to make sure it’s not going to endanger workers, ships below, or the environment. And there’s no way to know how much stress is in a member just by looking at it.

Instead, they had to create a structural computer model. But that’s not as simple as recreating the bridge in 3D. The order also matters. One cool example of this: the rivets in the connections for the top chords of each truss weren’t installed until after the concrete deck was poured. So, most of the load was being carried in the bottom chords in tension. When they removed the deck, the whole truss responded by going into what engineers call “negative bending.” The top chords were in tension and the bottom chords in compression, the exact opposite of what you would expect. That’s something that could have derailed the demolition plan without having gone through the exercise of modeling the bridge exactly how it was originally constructed, modified, and retrofitted over the years. It was almost half engineering, half a history exercise. The engineer even used old magazine articles to understand exactly where those stresses would be.

Here’s another tricky part. To lift the truss sections off the suspension bridges, they had to put a crane on a barge. Anyone who’s been on a boat knows that they’re not the most stable platforms for high-stakes, high-center-of-gravity work, especially when you add in huge loads and the need for precision. They did it for the original construction of these bridges, but that doesn’t mean it’s easy. Operating a crane from a barge involves dynamic loads from lifting and swinging. Barges often use these spud legs to help keep them in place, but there’s still a lot of engineering that goes into checking the stability of a barge for the variety of loading conditions. Those calculations help you pick the right crane, its configuration, limitations on pick weights and movements, et cetera, to make sure the work can be done safely. And just like the concrete deck, these truss segments had to be removed in a staggered manner to keep the towers from deflecting too much in one direction.

Interestingly, sometimes to demolish a structure, you have to add parts first. The original lateral load system had to be removed because of how the bridge would flex during demolition, but the engineers didn’t want people working on a bridge with no way to withstand wind loads. So they had to design and build these steel bumpers that could transfer lateral loads from the superstructure during the demolition process. In another case, they had to install bearing restraints on the trusses of the continuous spans, again to manage wind loads during the time those trusses were partially demolished. And in another example, they had to build an entire stiffening truss, made from pieces of the bridge that had already been removed, so the last deck truss could be lifted and removed as a single piece. All this steel was brought to a location on the shore where it could be cut down using this hydraulic shear and then sent off for recycling.

At this point, basically all that was left of each bridge was the suspension towers and cables. Since those cables are essentially one long structural member, there’s really no way to safely cut them loose. Imagine getting snapped by an enormous rubber band. There’s a lot of stored up energy there, and you don’t want any humans nearby when they come down. This is where the explosives come in. And like every other part of the process, this is tricker than you think.

Explosives used for demolition aren’t really like the ones you see in the movies. You’re not trying to use them to blow everything into tiny pieces. On buildings, you get a lot of breakup anyway because of the kinetic energy of falling. But really, the explosives are just strategically severing columns and beams quickly to start the falling process in a more controlled way. On a bridge, usually what you want is big pieces that can easily be removed from the water. So the explosives are more like small, very exciting saws that can cut quickly, simultaneously, and remotely.

Demolition contractors use shaped charges that sever structural members in a relatively controlled manner, and more importantly, a specific location. It matters a lot what these pieces look like after the blast. They have to be small enough to be lifted and transported out of the water. And you want them to fall in specific locations where they’re accessible without blocking the navigation channel. So, before the explosives are placed, there’s an entire process of precutting. The goal is to reduce each location where explosives will be placed down to flat plates or smaller sections so the blasts are sure to completely cut through. A worst case scenario is an incomplete explosive demolition that doesn’t fully bring the structure down. When that happens, the whole process becomes much more dangerous and difficult because you have to finish the job using workers on a structure where it’s not entirely clear where the stresses are or where it’s safe to cut.

On the I-74 bridges, workers cut the outer strands of the main cable, leaving only 7 of the 37 strands holding. This was done in four locations on each cable to break it up into manageable pieces. The towers were also cut in strategic locations to allow the shaped charges to sever completely and control which direction they would fall into the water. And, by the way, this leaves the bridge extremely vulnerable. You’re basically marching right up to the line of stability so that the explosives can kind of carry you over the finish line to bring the structure down. But it means the clock is ticking. You’re checking the weather. Working long shifts. You don’t want a storm bringing the bridge down before you get to push the button. But once you do…

I’ve mentioned a lot of pros and cons of explosive demolition, but there’s one thing they do better than anything else: the spectacle. You just can’t get around how cool it is to blow stuff up.

They got perfect shots for both bridges, bringing them down safely so that the barges could come out quickly to pick up the pieces. The road and navigation channel only had to be closed for a short period of time. Here’s a good look at how clean the cuts are from the shaped charges on the cables. And actually, they used explosives to demo the substructures later on. The piers in the more sensitive areas were taken out using conventional jackhammers as the final step. These sheet pile containment structures kept the debris from spreading out in the water. And of course they did a scan to make sure all the debris was picked up. It turns out one of those piers has become a great habitat for those endangered mussels, so you can see it still standing next to the new bridges. But other than that, there’s almost no sign that those old bridges were there at all.

I don’t just love bridges; I’m actually a licensed engineer, and part of that means every year I have to take a certain number of classes to stay current in the field. And that’s how I first heard of this job. Some of the engineers and contractors put the story together in a professional development class. So huge thanks to them for sharing so many interesting details of the project. And another thanks to Iowa DOT who answered our questions and shared photos and videos. I am so impressed with this bridge demolition. No shade at all to the new bridges, but honestly I think taking the old ones down was the coolest part of the entire replacement project. Bridge demolition is specialized, challenging work that takes a lot of engineering to get right. Hats off to the entire team on this project for getting it done so safely and efficiently.

Why Roads Get Washboards

[Note that this article is a transcript of the video embedded above.]

For as long as we’ve had clothes, we’ve done laundry. Long before detergent pods and spin cycles, people figured out the simple trick of rubbing textiles against rough surfaces, which squeezes dirty water out of the fibers and pulls cleaner water back in. Roughly around the 1800s, that idea got a dedicated tool that you still occasionally see today, even if only in jug bands. The washboard spread across the world as a faster, more practical way to scrub shirts and pants. Its use continued to grow until automated electric machines took over in the early 20th century. But we’re not here to talk about laundry. That era wasn’t just the heyday of the washboard. It also marked the start of the highway era (at least in the US), when road construction boomed, and traffic skyrocketed, including on the unpaved routes that connected farms, towns, and job sites.

If you’ve ever driven on a dirt road, you know one of their biggest problems. In some places, the surface organizes itself into a repeating pattern of ridges and valleys that can rattle your teeth, shake your suspension to pieces, and make the vehicle feel like it’s skating. It’s obnoxious at low speeds. At higher speeds, it can become legitimately dangerous because bouncing tires can cause you to lose control of steering and braking. In 1924, this phenomenon was known as “rhythmic corrugations.” By 1929, people were calling them “washboard” roads. And nearly a century later, they’re still a persistent problem all over the world.

What’s surprising is that this is still an active area of research. In the past decade or so, there’s been a burst of papers trying to explain exactly how washboarding forms and how to stop it. I was reading one of those studies when I saw the experimental setup they used. The moment I saw it, I knew I had to build one. Let’s talk about why this happens, and why there’s still more to learn about washboarding. I’m Grady, and this is Practical Engineering.

Roughly 35% of the countless thousands of miles of roadways in the United States are unpaved. That’s technically true, but a little misleading since most of those roads are in rural areas that carry only a tiny fraction of the traffic volume compared to the roads paved in asphalt or concrete. This is not a measured or reported statistic, but some back-of-the-envelope math shows that only about one percent of traffic happens on unpaved roadways in the US. That’s not true in other parts of the world, though.

We call that outer layer of a road the wearing surface (for obvious reasons). Of course, using a hard material like asphalt or concrete as the wearing surface has a lot of benefits: it cuts down on dust, it’s generally smoother and more comfortable to drive on, it’s easier and safer to drive at high speeds, and it’s more durable. But as good as it is, paving doesn’t make sense in every situation.

Almost everything in transportation engineering is an economic decision. Roads seem kind of mundane and simple, but you have to consider the scale. Think about a basic home improvement project like installing new flooring in a kitchen. Maybe it’ll cost you a few thousand dollars. Then consider that a mile (or 1.6 kilometers) of two-lane roadways has, very roughly, 600 kitchen’s worth of surface area. For a simple commute to work, we’re talking about thousands or even tens of thousands of kitchens now. This is silly, but the point is that roads are extremely expensive, not because they’re particularly sophisticated, but because they’re huge. You don’t get a monthly “road bill,” but if you add up gas taxes, tolls, and the slice of general taxes that fund roads, the average household in the US is paying roughly on the order of an electric bill each month. So engineering decisions about roadways are largely driven by costs. Here’s a simple example:

An unpaved road is pretty cheap to build. Before we consider the traffic, it sits pretty low on this graph. But as traffic increases, so do the maintenance costs associated with regrading, adding new gravel that’s lost over time, controlling dust through chemical application, etc. Its fixed cost is low, but its variable cost is steep because it’s not as durable. A paved road is expensive to build initially, but it requires less maintenance and that maintenance is less sensitive to traffic volume, so its variable cost is relatively flat. You can see there’s a breakeven point based on traffic volume, below which it makes more sense not to pave a road. And that’s just looking at traffic. When you factor in other complexities like the availability of materials, the abundance of skilled contractors who can do the work, and climate factors, there are plenty of situations where paving a road isn’t the best economic decision. In many countries, pavement is a luxury reserved for only the busiest highways. My point in all of this is to show you that washboarding isn’t just a problem of the past. It’s a real and present challenge, so it makes sense that we’re still trying to understand how and why it happens.

I mentioned there’s a lot of ongoing research into this question. Those papers can be pretty hard to wrap your head around. So I figured I’d build a demonstration so we can see it happen in real time.

Here’s how I set this up: In the center I have a stepper motor connected to a gearbox. I chose a stepper because the hardware is relatively cheap these days, and with the built-in encoder, I don’t need a separate system to keep track of the speed. The power supply and driver for the motor are over here to the side. I’m pretty new to this stepper stuff. This is not related to washboards, but I just think it’s cool. This controller sends pulses to the driver, which energizes the coils of the motor in a careful order. The trick is that it also knows whether the motor actually did what it said to do, so it can adjust or throw a fault if something goes wrong. For the 90s kids, it checks itself before it wrecks itself. It took me some time to wrap my head around it, but now I just have a button and knob that do exactly what I want.

This arm is attached to the gearbox and has a wheel assembly on one side and some bolts for a counterweight on the other. The wheel runs in a circular track of sand I made from polycarbonate sheets. I used polycarbonate rather than my typical acrylic because it can flex more without breaking. Turns out that not much will stick to it, though, so I ended up bolting the pieces together.

Now that this is set up, I can just push go and watch what happens. And what happens isn’t really too interesting at first. This does just about what you’d expect it to: the wheel digs a rut in the sand as it goes around and around the track. Right now, it’s moving at about one meter per second or roughly 2 miles per hour. I let this run for an hour with pretty much no change in the behavior. But watch what happens when I ramp up the speed just a bit. And actually listen too, because it’s easier to hear than to see at first. [Break for real-time footage.] After only a few laps around the track, you start to hear a rhythm.

It doesn’t take long at all before the wheel is literally galloping between each bump. I have to be honest - I didn’t think it was going to work this well. Let’s take a closer look to see what’s happening. I’m going to smooth out the sand and start this again. But, critically, there’s no way for me to make this surface mathematically smooth. There are always going to be some small, random bumps and dips. And it turns out that’s important here. My wheel has some freedom to move around this clevis pin, so it can go up and down bumps like a regular vehicle would. The wheel’s not rigidly being pushed through the material; it’s in a conversation with the sand. Because there is some irregularity in the surface, the force between the wheel and the sand is irregular too. When we hit a bump, the force is briefly higher, causing the sand to be plowed forward. And when the wheel encounters a dip, the force is less so less material is shifted.

That fluctuation in forces matters because the wheel starts changing the surface of the track unevenly. As the wheel rolls up off a bump, it unloads, but what goes up must come down. The wheel has inertia, so when it falls, the downward force on the surface is higher than when it’s just rolling along. If the speed is fast enough, that impact happens a little after the lowest point, so the sand is pushed and piled downstream of where the wheel hit, creating a new bump where there wasn’t one before. One bump becomes two, becomes three, and so on. Each pass makes them a little bigger and organizes them a little better until you get that rhythmic corrugation in the surface.

Eventually the bumps are big enough that the wheel follows a ballistic trajectory, literally jumping off the surface with each one, and slamming back down on the other side in just the right place to push more sand up the following bump. The only thing limiting the bumps from growing indefinitely is the natural repose of the sand. It can only stand up so steeply before it flows down into the next dip, filling it somewhat. That’s why these washboards can travel like sand dunes, slowly moving down the road.

Understanding all this, it makes sense why washboards often emerge in transitional areas of a road: curves, changes in slope, and at connections to paved roads. You need a fluctuation in tire force to get the process started, and these are places where, even without a bump, there’s a higher likelihood of a sudden change in force that could start to shift the material.

What’s interesting is how robust the instability is. You would think that the differences in vehicle weights, wheel bases, suspensions, and speeds would tend to smooth things out. It turns out, they don’t. In fact, you don’t even need a wheel to do this. Researchers have shown that even a simple angled plate that acts like a plow will form ripples above a threshold speed as long as it has the freedom to move with the bumps. It all feels a bit counterintuitive; it’d be easy to assume that an angled plate would smooth out the road like icing on a cake. A big part of the reason things don’t smooth out is that the washboards are self-perpetuating. They impose a periodic force on vehicle tires that, in turn, amplifies the forces creating the corrugations. It’s like the Tacoma Narrows Bridge where the wind amplified the torsion and the torsion amplified the wind's effect on the deck: a positive feedback loop of instability that only gets worse over time.

The washboarding of roads is just one example of what scientists call a pattern-forming instability, where structure emerges from a system that starts out more-or-less uniform. We see them everywhere in nature, from chemical reactions to the markings on animals. And washboarding has some natural analogs. In rivers, moving water transports sediments in a way that can form ripples on the bed. I caught a cool shot of this phenomenon in action for my video about reservoir sedimentation. At a larger scale, river meanders are often described in the same way. Even a straight channel will, under the right conditions, begin to drift sideways as curvature redirects the fastest flow toward the outer bank, strengthening erosion there and deposition on the inside. I have a couple of videos on that process, too. But the fluid doesn’t have to be water. In places without much vegetation, wind-driven sand doesn’t naturally smooth out. Instead, it forms dunes that can march across the landscape. In all these cases, the same basic feedback loop is at work: the driving force reshapes the surface, and the reshaped surface redirects the driving force. It’s the same thing with washboards. Small bumps change how a wheel loads and shears the gravel, and that shifted, uneven loading moves material in a way that makes the bumps grow into a repeating pattern.

Characterizing the factors that play a role in the formation of washboarding helps you make engineering decisions about how to avoid it in the first place. The obvious solution is to use a more durable wearing surface, like asphalt. But we’ve already covered why it’s not always the best option. Plus, it’s not like asphalt is immune to damage. Speed is the most critical factor for washboarding. Maybe slowing down is a useful suggestion on a personal driveway, but it’s not particularly viable as a widespread solution on public roads, especially the low-traffic roads where enforcement of speed limits is…less than strict. We want to get to where we’re going, and we want roads that can safely handle the speed. So most of the focus is actually on the materials used in roadway construction.

Typical roads use a base layer below the wearing course. It can get a lot more complicated than this, but in nearly every case, the base material is usually doing the heavy lifting. The material used for road base is broadly graded, meaning it has particles of lots of different sizes, from gravel size all the way down to microscopic clay-sized grains. Often it’s made from crushed rock, which naturally produces a good mix of particle sizes. All those different sizes, plus the angularity of the crushed stone, help the road base lock together into a dense, strong layer. It’s actually pretty remarkable. If you’ve ever walked on well-compacted road base, it feels almost as hard as concrete. But remember that a typical road base is meant to be confined by some kind of pavement. You don’t have to worry about particles being dislodged or shifted because they’re trapped underneath the asphalt. So, generally, the wearing surface for an unpaved road is going to perform best if it has more fines than a typical road base, locking those bigger aggregates in better and offering more resistance to shearing.

This is a pretty careful balancing act. Too many fines, and the road is excessively dusty when dry and slippery when wet. Too few fines, and the surface won’t lock together, leading to washboarding and raveling. It’s a pretty fine line to walk for an engineer, and the trouble is that it’s not always a simple matter to source the right material. Road base is widely available because we use so much of it, so quarries and suppliers tailor their equipment and processes to meet those specifications. It can be much harder to get an aggregate operation to produce or blend a custom batch of material that works well for unpaved roads. In fact, except for the biggest projects with a commensurate budget, the material chosen for an unpaved road is almost always a compromise between what works best and what the local quarry can give you.

Another option might be synthetic materials that can augment the properties of soil. These are used in all kinds of situations, including roadway construction, so it makes sense that they might help prolong the life of unpaved roadways and reduce the maintenance costs associated with washboards. I think it might be fun to test some of these out in the test track here, since I already have it built. Let me know if you’d like to see a video on that.

For something so common, washboarding is a reminder that we’re still actively learning how the built world behaves, especially when it isn’t perfectly rigid, smooth, or controlled. Researchers are still teasing apart the details of how speed, tire pressure, suspension, moisture, and grain size conspire to pick a spacing and then lock it in. Unpaved roads aren’t a relic of the past. They’re critical infrastructure for huge parts of the world, linking farms to markets, kids to schools, and communities to clinics. And if a road corrugates, it’s not just annoying. It slows everything down, shakes vehicles apart, can lead to crashes, and makes maintenance a constant uphill battle. The more we understand the physics of washboarding, the better we can engineer roads that stay safe, smooth, and reliable, even when the only tools you have are local materials, a grader, and a limited budget.

Fixing the Most Dangerous Dam in the World

[Note that this article is a transcript of the video embedded above.]

Mosul Dam rises 370 feet or 113 meters above the Tigris River in northern Iraq as one of the tallest dams in the Middle East. The dam was built in the 1980s, but, in a way, construction never really stopped. That’s because ever since the reservoir filled behind Mosul Dam, the ground has literally been dissolving, nonstop, below the structure. Almost immediately on filling, water started flowing through the foundation of the dam and back out on the downstream side. Just a year later, the volume of seepage was measured at 800 liters or about 200 gallons per second.

I usually hate to use the olympic-sized swimming pool equivalent, but in this case it makes sense because it was enough to fill one every hour of every day. And the issue is that, once a process like this gets started, it’s pretty hard to stop. So, for the past 40 years or so, the problem at Mosul Dam has been ongoing, scrutinized by some of the most preeminent engineers across the world and complicated by politics, bureaucracy, and, of course, armed conflict. Failure of a structure this large would be catastrophic; towns along the Tigris River would be fully wiped off the map, and some estimate that the breach wave would be so massive that even major parts of Baghdad, hundreds of miles downstream, would be submerged. In 2006, the US Army Corps of Engineers called it, unequivocally, “the most dangerous dam in the world.” That was 20 years ago, and Mosul Dam is still standing, in better shape than ever. And the story of how it got there is fascinating. I’m Grady, and this is Practical Engineering.

Mosul Dam is an earthen embankment dam not far from the City of Mosul in Iraq, built to generate hydropower and store water for irrigation and drinking. The hydro plant is on the west side of the dam with four turbine generators. You can see the massive surge tanks sticking up from the plant that absorb changes in pressure when the units are started and stopped. The dam has an outlet structure through the embankment here. It has a service spillway with radial gates here. And an auxiliary spillway with earthen fuse plugs here. Check out my videos on spillway gates and fuse plugs if you want to learn more about those types of structures after this.

The dam itself is impressive, but the rock that serves as its foundation is extremely complex, and in many ways, far from ideal. The geology of northern Iraq includes a lot of gypsum, a sedimentary rock that is widely used for things like fertilizer, plaster, and drywall. What it’s not widely used for is the foundations of dams. In fact, the consensus of experts involved on Mosul Dam throughout the years is that it was, all around, a terrible idea. One consulting group said that, quote, “the decision to locate such a major and important dam on the foundation rock mass which exists at the Mosul Dam site was fundamentally flawed.” That’s because of a critical property of gypsum, one that it doesn’t share with many other types of rock formations: it dissolves in water.

You might be familiar with limestone caves and karst geology, where water creates voids in the subsurface. Some of these can be quite dramatic like Carlsbad Caverns in New Mexico or Mammoth Cave in Kentucky. They’re formed because the limestone is just a tiny bit soluble in water, as long as it’s a bit acidic, which rainwater usually is. So over the course of millions of years, that water kind of carves away the earth from the inside. Gypsum, on the other hand, is roughly 200 times more soluble in water than limestone. It’s not quite like a spoonful of sugar or salt that dissolves almost instantly, but processes that usually take centuries in limestone are accelerated to human timescales in gypsum. And that’s especially true in the subsurface, because dissolution isn’t a linear process. More dissolving means more space for water which means more dissolving and so on. It’s a positive feedback loop.

Many dam failures have resulted from internal erosion, where water seeping through the soil or rock carries away particles, leaving voids. This process is what led to the demise of Teton Dam, which I covered in an earlier video. But where internal erosion can be combatted by designing filtration systems that catch waterborne particles before they escape the subsurface, you can’t easily filter dissolved gypsum out of seepage.

The designers of the dam knew the gypsum was going to be an issue, and they had a few ideas to address it. One was to install a blanket of bentonite clay lining the bottom of part of the reservoir. This would block seepage from flowing into the subsurface, at least in the dam’s immediate vicinity, lengthening the flow paths and thus reducing the total volume of the flow. However, the volume of material would be enormous, and the blanket layer would be fairly fragile to damage from boats or even strong currents. Another idea was to use a cutoff wall, basically a continuous subsurface diaphragm of some impervious material. The problem was that there were no machines that could trench deep enough to get below the worst of the gypsum. The idea they landed on was the same as at Teton Dam: a grout curtain.

Mosul Dam’s design included a continuous concrete tunnel running along the bottom of the structure. It had one purpose: to provide access to the dam’s foundation for drilling rigs and grout pumps. Political and schedule pressures pushed the government to finish the dam before the grouting was complete, but they knew they would have the access to the gallery tunnel to continue that process after the dam was in operation. Unfortunately, they underestimated how serious and complex a challenge they were setting themselves up to face.

As soon as the reservoir filled up, the problem became obvious. I mentioned the olympic swimming pools of seepage in the intro, but it wasn’t just that. Sinkholes opened up downstream of the dam as caverns formed in the geology below causing the surface to collapse. As time went on, those sinkholes started appearing closer to the dam, an aboveground hint at how the solution cavities were migrating in the subsurface. Essentially since its construction, operators have maintained a continuous grouting program, injecting a mixture of sand, cement, bentonite, and water into the rock below through drilled holes to try and plug up the voids. It’s basically a nonstop race between logistics and chemistry, because grout doesn’t fare well in flowing water and the foundation rock is constantly dissolving.

Recognizing the hazard they had created in the 1980s, the Iraqi government came up with a backup solution. Since it was clear that there really was no permanent fix for Mosul Dam, they would just build another dam downstream that would capture the flood if (and maybe when) Mosul Dam failed. Badush Dam started construction in the late 1980s. It would have a hydropower plant and store water for irrigation, but also include a huge empty storage pool to protect downstream cities from a breach of Mosul Dam. The project got about halfway finished before the geopolitical situation in Iraq ground it to a halt.

In 2003, a US-led coalition invaded Iraq as part of a larger war on terror in response to the September 11th attacks. As a major piece of infrastructure in the country, Mosul Dam had the coalition worried. Some early reports hinted that Iraqi forces might detonate the structure as an act of sabotage. But it didn’t take long to realize that the dam might fail on its own accord. They started coordinating with the US Army Corps of Engineers to assess the structure, whose report concluded that the risk was astronomical. That’s the source of the “most dangerous dam in the world” quote that has plagued the structure ever since. The truth is that the “danger” of a dam is a pretty complicated thing to characterize, and it’s not a statistic that’s widely tracked, especially at a global scale. But the fact that a government agency was willing to say it means a lot. And Iraq’s Ministry of Water Resources took the situation seriously and started working with a panel of experts to review the conditions of the dam. That panel largely came to the same conclusion: Mosul Dam needed serious help.

Coalition forces had bases and equipment along the Tigris River. The situation was concerning enough that they decided to move everything out of the potential inundation area if the dam were to breach. At the same time, a major part of the war effort was helping the new Iraqi government shore up the country’s infrastructure, including improving the grouting program at Mosul Dam. Even though it was really only considered a temporary solution, the consensus seemed to be that it was the only feasible way to address the foundation problems beyond the stalled Badush Dam project downstream.

Initial efforts by the US government to help at Mosul Dam turned into somewhat of a disaster. A few notable examples: The winning contractor for the grout plants submitted a concrete (not grout) mixing plant design, and somehow the review committee didn’t notice, despite it being printed on the front page of the submittal. By the time someone realized it, the concrete plants had already been delivered, and the US government had to pay the contractor to try and convert them into grout mixing plants. The material silos were poorly designed, with no ladders or braces. Some weren’t even bolted to the foundation. The loading ramp for the hoppers had no retaining walls, causing the slopes to slough off. Drills and pumping equipment couldn’t even fit into the grouting galleries below the dam. And the dam operations staff meant to run all this new high-tech equipment had only received a few weeks of training. The oversight report about the project was scathing. Millions of dollars had been spent on 21 contracts for almost no benefit to the dam.

Coalition forces continued efforts to improve the situation at Mosul Dam, but by 2010, the US was withdrawing troops from the country and handing off the reconstruction projects back to the Iraqi government. Unfortunately, that handoff was only temporary, as sectarian violence continued to plague the region. In mid-2014, the Islamic State (also known as ISIS, ISIL, and Daesh) took over several cities in Northern Iraq, disrupting the supplies of materials to Mosul Dam, which was still relying on nearly 24/7 grouting operations to keep the structure safe. That August, ISIS seized control of Mosul Dam, sparking new fears that the structure would collapse. For more than a week, the dam was out of the hands of the Iraqi government, and no one knew what the militants might do (or what they might not do). It was the same situation as before: Even short-term neglect presented a serious safety risk. Fortunately, the dam was recaptured by Kurdish and Iraqi forces, with the help of US air support, 8 days later. The dam was back in Iraqi hands, but the surrounding areas weren’t. With equipment looted during the brief seizure, the disruption of the workforce at the dam, and without regular shipments of cement, the grouting operation wasn’t being maintained. Equipment installed during the Iraq war wasn’t being used. Voids were going untreated, and concerns about the dam’s failure continued to grow.

Realizing that the Iraqi government was too fractured to manage the situation alone, the US decided to stay involved as Mosul Dam’s de facto engineer. In 2015, the Army Corps of Engineers led a task force to assess the condition of the dam, and the results were alarming. The US Embassy released a fact sheet based on their findings, saying that the dam had an “unprecedented risk of catastrophic failure” endangering between half-a-million and 1.5 million people along the Tigris River. A collapse would be a humanitarian crisis unlike almost anything in modern history. The situation was further complicated by the ongoing occupation by the Islamic State, making it difficult or impossible for residents to be able to evacuate to safer areas. Electrical blackouts, lack of government coordination, and poor communication would make things even worse in the event of failure.

The Iraqi government tried to downplay the alarm a bit. In an interview on TV, the Minister of Water Resources said, quote, “The looming danger to Mosul Dam is one in a thousand. This risk level is present in all the world’s dams.” I don’t know if he made that number up, or if it was actually supported by some kind of analysis, but anyone involved in risk management would find it hilarious if it weren’t such a serious situation. Assuming that’s an annual probability, which is what we normally use, and multiplying it by the consequences of failure estimated by the Corps of Engineers, you get an expected annual fatality rate of 500 to 1500 people. Nowhere in the world would anybody consider that acceptable. This is a graph often used to communicate tolerable risks on large dam projects. This green area generally means there’s not a lot of justification for making a structure safer. Yellow, you have to be more thoughtful. Red means unacceptable. Taking the minister’s estimate of probability, and the embassy's estimates of fatalities at face value, Mosul Dam would plot somewhere around here on the chart. That “most dangerous dam in the world” moniker doesn’t seem like hyperbole when you look at it like that. To quote Lieutenant-General Sean MacFarland, “If this dam were in the United States, we would have drained the lake behind it.”

The urgency finally spurred action in 2016. Iraq awarded a contract to an Italian company to rehabilitate the structure, including a massive operation to expand the foundation grouting program. It was one of the most unique civil engineering projects on the globe, with participation from the Iraqi government, the US (through the Corps of Engineers), the Italian military, and a number of international consultants. I actually talked with a few of the engineers involved on the project, and some of their stories are pretty wild. In the early days of the project, they were inserting engineers at night, by helicopter, to support the Iraqis who were operating the dam and install equipment that would let them monitor the situation remotely while ISIS was operating only a short distance away.

The entire project had to happen near the front lines as the conflict with the Islamic State continued to unfold in Iraq. Security forces were needed for the entire duration to protect the dam and supply routes for materials and equipment. That took some time to get set up, but eventually, the project team was able to establish a permanent camp at the dam. Over the next few years, all the grouting infrastructure, including batch plants, piping, electrical systems and drill rigs were replaced with modern equipment. Crews drilled more than 5,000 boreholes with a total length of drilling at more than 400 kilometers or 250 miles. 41,000 cubic meters (50,000 cubic yards) of grout were injected into the foundation along the entire length of the dam. Generally the way it works is this: you can inflate a rubber device called a packer using air or hydraulic pressure, creating a seal between the borehole and injection pipe. Or you just grout the injection pipe directly into the borehole. Then you can pump grout at very high pressure into the borehole, forcing it into voids, cracks, fissures. You just keep pumping until you reach a refusal criterion, a certain maximum pressure that you hold until the grout stops flowing. And you just keep doing it over and over and over.

All this work was done using a sophisticated computer system to keep track of pressure, depth, mix design, flow rate, and quantity of grout for every borehole, allowing the team to track progress, identify issues, and visualize the performance of the operation. From material delivery to batching to drilling and injection, every step of the process became a data point.

I love unique measurement units, and this project had a good one: As a quality control test, the contractor would try to inject water into the foundation rock after it was grouted up. A Lugeon is the loss of water of one liter per minute per meter of borehole length at an overpressure of 1 megapascal or about 145 psi. For all the permeability tests performed for the project, 98 percent had values below 3 Lugeons, a massive improvement over the conditions beforehand.

The project finished in 2019. It was a 3-year effort that cost more than half-a-billion dollars, but Mosul Dam lost its most dangerous dam title as a result. By all accounts, the dam is in a much less precarious position. The project won an award from the Deep Foundations Institute in 2022, highlighting the complexity and the danger of the work.

But this wasn’t like a typical construction project, because the work isn’t over. The goal was to get the Iraqi government set up to continue the process of maintenance grouting. The rock below Mosul Dam may have a lot more grout than it used to, but the gypsum is still soluble, and there’s still a massive reservoir constantly trying to push water through it. A major part of the rehabilitation project was training Iraqi staff to continue the fight. In that way, despite its magnitude, the project was sort of a half-a-billion-dollar bandaid. The grouting has never been considered a permanent solution, and even though this project resulted in an enormous improvement in the long-term prospects of the structure, it’s still a major, ongoing obligation.

Iraq is still planning for a more permanent fix. You can still see the half-finished Badush Dam on the map, downstream from Mosul, and finishing the job is still on the table if anyone can figure out how to come up with the billions of dollars it would take. Another option is that deep foundation cutoff wall considered during the original design. It would provide a continuous barrier for seepage passing through the porous rock below the dam. These are used on a lot of dams across the world, but it’s never been done on the scale and depth as would be required at Mosul. In 2018, the estimated cost for a cutoff was between 3 and 5 billion dollars, an almost unimaginable investment into a dam that already exists and functions today. Whether the electricity and water from Mosul Dam is even worth that scale of capital is something that will probably take a long time to decide. Until then, the government will keep pumping grout and Dinars into the rocks below in the nonstop race against a flawed foundation, but now with much more confidence that they can keep up the pace.

The Physics Behind the Thumb Trick

[Note that this article is a transcript of the video embedded above.]

Have you ever filled a bucket with water from the garden hose? It’s kind of a slow process. Or at least it feels slow while you’re standing there waiting. If you played with a garden hose at all, you know the trick of putting your thumb over the end to get a stronger jet. Obviously, the water is flowing faster with your thumb on than off. So if you do that - put your thumb over the end of the hose - to fill your bucket, do you think it’s going to fill faster, slower, or take the same amount of time?

Seems like kind of an elementary question, but I found this in the online notes for a college physics class. The only issue with the professor’s answer to the question is that it was wrong. Pipes seem simple, but there are a lot of misconceptions about pipes and how they work. The field we sometimes call Closed Conduit Hydraulics is a place where intuitions don’t always serve you well. And “closed conduits” matter. Lots of essential parts of our lives depend on fluids moving through pipes. So I put together a few demonstrations in my garage to try and correct some misconceptions. Let’s take a look at what really happens inside a garden hose, or really any pipe system to gain some intuition. I’m Grady and this is Practical Engineering.

The question I posed about filling up a bucket was from a lesson on continuity. The basic idea is that water isn’t very compressible. So in any closed system, there has to be the same amount coming in as there is going out. In mathematical terms, that looks like this. Velocity multiplied by a pipe’s cross-sectional area is the volumetric flow rate. So v-in, a-in is equal to v-out, a-out.

The professor’s answer was that the time to fill up the bucket will be the same, regardless of whether your thumb is over the end or not. The velocity out is higher, but the area is smaller. The volumetric flow rate should be the same in both cases. It sounds reasonable. Let’s test it out and see if that’s true.

I’m going to speed this up so you don’t have to suffer through the full duration. I used a big bucket to show the difference better. It’s not night and day or anything, but this makes it pretty clear that putting your thumb over the end of the hose actually slows down the flow rate. This is probably not earth-shattering news for you, but the reason for the difference is a little complicated.

Just to be clear, this demonstration doesn’t violate the principle of continuity. In engineering and physics, when we use conservation rules to solve problems or answer questions, we have to be explicit about the boundaries. Usually, that means applying a control volume, a defined region of space where we can easily describe inputs and outputs of flow, energy, momentum, and so on. In my demonstration, I can define a control volume here, and it’s easy to show that the flow rate through the hose is the same as that coming out of the end. Same thing with my thumb over it: the velocity in the hose is lower than the velocity leaving, but the area of the hose is larger than the nozzle I’ve formed with my thumb, so it equals out. But you can’t apply the principle of continuity across different control volumes. In other words, these are completely different situations. And if I change this demo up a little bit, it will be more obvious.

Now I have a mechanical thumb to constrict the end of the hose. In other words…a valve. Functionally, this does the exact same thing. When I turn the valve, it creates a varying obstruction across the pipe from wide open to fully closed. Let’s measure the flow rate for a full range of valve positions and see what happens. This is a chart of the data, and you can see there’s a pretty clear relationship. More restriction; less flow. This is the answer that the professor missed by assuming the flow rate IN was the same in both cases. Again, probably not earth-shattering news to anyone that when you close a valve the flow rate goes down. But you might not have ever considered, “Why?”

To answer that question, we have to look at a different conservation equation: energy. Basic physics separates energy into two forms: potential energy that is stored in some way, and kinetic energy: the energy of motion. Fluid in a pipe has both. Potential energy takes the form of pressure or elevation, kinetic energy in the form of velocity. The trick is that you can convert between types of energy, and of course, the total amount of energy in a closed system doesn’t change. And knowing this allows you to answer all kinds of questions. Let me show you an example.

This is a basic hydraulic system. A tank on the left and a pipe that constricts down, then expands back out. You know I love graphs, and there is a graph that makes solving closed conduit hydraulics problems a lot simpler. It’s called the hydraulic grade line, and it basically describes the potential energy in a fluid along its path. In the tank, there’s hardly any velocity, so all the energy in the fluid is potential energy. The hydraulic grade line is equal to the free surface. But once water enters the pipe, it picks up speed, so the hydraulic grade line drops down. The difference is the potential energy converted to kinetic. At any snapshot in time, we know that the volumetric flow through a pipe is constant. You really can’t have more water coming in than going out, just like we discussed with continuity. So the hydraulic grade line is constant as long as the velocity is constant.

The fluid has to accelerate as it goes into the narrower pipe. That converts more of the potential energy to kinetic energy, so the hydraulic grade line drops again. Same thing on the other side. The flow slows down as it expands into the larger pipe, so you get a conversion of kinetic energy back into pressure. If this seems complicated, just remember that the hydraulic grade line is basically the answer to the question: “If I tapped a vertical riser into this part of my pipe system, how high would the fluid go up it?”

I think this is intuitive for most people. It’s Bernoulli’s principle in action. But it’s missing something that makes it impossible to apply to our garden hose demo. Let’s hook up some pressure gauges, and you’ll see what I mean.

I put a pressure gauge at the beginning of the hose and one at the end. When I turn on the water, we see a pressure just under 70 psi or about 450 kilopascals at the upstream end. At the downstream end where the water’s coming out, it’s basically zero. That doesn’t jive with what we’ve learned so far about the conservation of energy. Let’s sketch out the hydraulic grade line to figure it out.

Here’s our hose. It’s close enough to level that we can neglect differences in elevation, so all the potential energy is in pressure. On the upstream end, it was 70 psi, and on the downstream end, essentially zero. That makes sense because the end of the pipe is exposed to the atmosphere. You can’t really have any pressure if you don’t have a pipe. That means our hydraulic grade line looks like this. We know in a pipe with a constant cross-section that the flow velocity isn’t changing, and yet, we’re still losing potential energy along the way. Where’s it going?

Well, we need to talk about losses. Of course we know energy can’t be created or destroyed, only converted from one form to another. We talked about pressure, elevation, and velocity already. But there’s also heat through friction in the system. No pipe is perfect. You’re always going to lose some energy along the way. Unlike pressure or velocity, frictional losses are unrecoverable. Once they’re lost, they’re lost. The garden hose example shows it perfectly.

Let’s assume the inlet pressure is always constant. It’s not really, since there are more pipes upstream of this point in my house’s plumbing. But assuming a constant inlet pressure, this hydraulic grade line is always going to look the same. You can make the pipe smoother, rougher, longer, or shorter, wider, or narrower. As long as the shape doesn’t change along its length, you’re always going to have the inlet pressure on the left, zero pressure on the right, and a straight line connecting the two. In other words, you’re always going to lose 100% of the potential energy in the water to friction from one side to the other. How’s that possible? It’s because the flow in the pipe will speed up or slow down until it’s true. Frictional losses are roughly a function of the fluid’s velocity squared, so higher speed means more losses. Again, assuming you can maintain a constant pressure on one side of the system, in effect, what controls how much flow you can get out of the other end of the pipe is how much friction happens along the way.

And, by the way, that’s kind of a tough question to answer. The friction is a function of pipe roughness and turbulence. Turbulence is a function of the flow rate, so you have to know the flow rate to calculate the friction to calculate the flow rate. So these computations usually require some iteration or at least some simplifying assumptions. I said that generally friction scales as a function of flow squared. I can show that in my demo with the pressure gauges. If this valve is closed, we get the full static pressure. There’s no movement, so there’s no frictional losses anywhere in the hose. I have the same amount of energy at the end of the hose as I do at the beginning. When I open the valve, the difference in pressure grows because the flow speeds up. And if we plot the difference in pressure as a function of flow rate, it looks something like this. Friction goes up a lot faster than velocity.

But friction in a pipe isn’t the only source of energy losses. Any transition in geometry is going to have losses, too. And now, we’re back to the thumb. We sometimes call pipe friction the “major” losses in a system and those at transitions “minor losses.” Researchers have measured all kinds of situations, making it possible to estimate how a pipe system will behave, no matter how complicated it is. And the results are pretty interesting.

For example, at a sharp-edged inlet into a pipe, the minor loss coefficient can be around 0.5. A higher number means more energy lost. If you round the inlet, you can get that coefficient down to 0.03. Huge difference. Same thing with expansions or contractions. If you have a sudden change, especially when the difference between sizes is larger, you get high loss coefficients. If you make the transition gradual, the coefficient goes down, since there’s less turbulence and gentler acceleration as the fluid changes speed. And every type of transition has an associated loss coefficient that can vary a lot depending on how smooth and consistent that transition is. In fact, valves take advantage of minor losses to give you some control over flow, and we already said that a valve is basically a mechanical thumb.

I have one more demonstration to show you. I’m going to fill this tank two more times. In one case, I put a cap over the hose with a hole drilled into it. In the other, I 3D printed this nozzle that has a smooth taper from the hose diameter down to the exact same diameter I drilled in the end cap. With an understanding of minor losses, it should be an easy guess which one can flow more water. And here’s the proof. Both hoses are discharging through the same-sized hole, but the one with a smoother transition but the one with a smoother transition lets a lot more water through. And if you compare the 3D printed nozzle with the fully open hose, it’s not quite the same flow rate, but it’s close, and it’s a lot closer than the sharp contraction created by the cap with the hole.

The point I’m trying to show with this is that a nozzle or any other type of obstruction you put in a pipe system doesn’t increase or decrease the flow from one side or the other. It just creates a loss in energy that slows down the whole system. Transitions and pipe roughness create friction, and the flow rate naturally adjusts itself until the available energy between two points is equal to that friction. And this is not necessarily intuitive.

For example, we often compare water in pipes to electricity in wires: pressure is like voltage, flow rate is like current, and a narrow or rough pipe is like a resistor. That analogy works pretty well for building intuition, but it breaks down once you care about the details. In a wire, resistance is usually close to constant for a given material and temperature, so current tends to scale more neatly with voltage. In a pipe, the “resistance” isn’t a fixed number. Friction losses grow faster than the flow and can change as the flow becomes more turbulent. But of course, you can build that intuition. Think about firefighters.

The operator’s job is to run the pump. They choose a throttle setting based on the pressure needed at the nozzle. How do they make that choice? Well, that depends on the diameter of the hose, the length of the hose, the elevation of the nozzle if you’re pumping up a hill or a ladder, and the characteristics of the nozzle itself. It’s important to get this right. Too little pressure at the nozzle, and you don’t get enough flow to quench the flames. Too much pressure and you can damage equipment or throw the nozzle operator around with excessive reaction forces. Firefighters learn the basics of hydraulics in training, but there are no desks with graph paper set up at a fireground to work through a bunch of engineering equations. Operators need good hydraulic instincts about how different configurations of hoses, apparatuses, and nozzles will affect the required pump settings.

Even the plumbing in your house follows these same simple hydraulic principles. If you have narrow pipes, or lots of bends, turns, and transitions, you’ll definitely notice if someone flushes the toilet while you’re taking a shower. The shared lines see higher total flow, meaning more friction, meaning less pressure. I mentioned earlier that we couldn’t really assume a constant inlet pressure at my hose bib. That’s because there are a lot of pipes and transitions from that point upstream. And it’s true from my house through my service line through the water mains all the way to the water towers and high service pumps at the treatment plant. The pressure and flow rate I can get out are almost entirely a function of how much friction the water encounters along the way, which is a function of both the flow rate and the geometry of the pipes. You may even notice that your water pressure drops in the mornings or evenings when everyone in your neighborhood is using more. It’s the same issue: more flow through the water mains creates more friction, converting kinetic energy into heat so you get less at the end of the line.

The garden hose is a backyard version of the same problem engineers and operators deal with every day: how much flow can you get through a real system, and what does it cost you in pressure? In a perfect world, you’d convert pressure to speed and back again with no penalty, but real pipes always take a cut. Sometimes that cut is spread out over a long run of pipe. Sometimes it’s concentrated in a single valve, elbow, or your thumb. Either way, the flow rate adjusts until the available pressure is fully “spent” on those losses. Once you see it as an energy budget, the weird stuff starts making sense.

The Wild Story of the Teton Dam Failure

[Note that this article is a transcript of the video embedded above.]

In 1975, after three long years of excavating, hauling, placing, and compacting soil across the Teton river, Teton Dam topped out at 305 feet or 93 meters high. Built by the Bureau of Reclamation, the dam was the flagship component of the Teton Basin project of southeastern Idaho, meant to provide flood control, power generation, recreation, and irrigation water supply for farmland in the Snake River Valley.

Not everyone was enthusiastic about the dam. The benefits were significant, but many felt that they didn’t make up for the environmental impacts, let alone the $100M price tag. Like most major water projects, it was controversial from the start. But politicians and dam advocates eventually won out. The lawsuits were resolved, and the earthen dam reached its final height at the end of 1975. Other parts of the project were still under construction, even as the dam wrapped up. In particular, the contractor was behind on a tunnel through the left abutment of the dam, called the river outlet works. This structure was needed to make controlled releases downstream. By the time the dam itself was done, the smaller auxiliary outlet tunnel on the opposite side was the only way to release water. Even so, the Bureau was eager to put the dam in service. Waiting for the completion of the river outlet works would mean losing out on a whole year of valuable spring runoff that could be put to use irrigating fields, floating boats, supporting fish, and generating electricity. So, the Bureau decided to start filling the reservoir anyway.

The first filling of any reservoir is a risky process. No matter how much you plan a dam, you never know how it’s going to respond to the immense pressure of actual water. Like many projects of its kind, Teton’s designers required that the initial fill be moderated, only allowing the reservoir to come up by a maximum of 1 foot (or 30 cm) per day. As it filled, engineers and technicians performed daily inspections and monitored the water levels in nearby wells. The goal was a measured process, loading the structure slowly so they could catch any problems and respond, even lowering the reservoir back down if anything seemed off.

Early in 1976, though, after a particularly snowy winter, it became clear that the spring melt was going to send a lot of water their way. Without the river outlet works, the dam’s ability to release water was limited. The auxiliary outlet tunnel just couldn’t handle the flow they needed. That meant the reservoir would have to fill faster than the foot-per-day limit. So, they relaxed the limit. The dam seemed to be holding fine anyway, and there really wasn’t another choice.

As the summer got closer, the reservoir was nearly full, just a few feet shy of the spillway. On June 3rd, engineers were doing their daily routines when they noticed water springing from the right abutment. The leaks were pretty far from the dam. It’s not completely out of the ordinary for little seeps to form in the vicinity as the local groundwater levels respond to a new reservoir. Two days later though, on the early morning of June 5, there were more leaks, and this time they were on the dam itself. Engineers quickly knew something was seriously wrong.

The water coming through the west side of the dam was first noticed around 7AM. By 10AM, the project’s construction engineer was peering into a tunnel carved through the embankment that was tall enough to stand in and extended into the dam as far as he could see. I’m Grady, and this is Practical Engineering.

Teton was a zoned embankment dam, meaning it was made of different kinds of compacted earth, each serving a different function. That soil came from borrow areas near the site of the project to minimize the cost of hauling it. Roughly 10 million cubic yards or 7-and-a-half million cubic meters of material were excavated from the abutment areas and even from the bottom of the river, hauled to the site, and placed in thin layers, then compacted into place. In the center of the embankment was Zone 1, built from wind-deposited silt (known as loess) that covered the upland areas at the top of the canyon around the dam. This zone of fine particles was the core of the dam, meant to provide the watertight barrier to prevent leakage. But even if the dam itself was watertight, there was another engineering challenge.

Below all the loess in the upland areas, the geology in southeastern Idaho gets a lot more complicated, thanks to an eruption of the Yellowstone Supervolcano about 2 million years ago. When a massive volcano erupts, it doesn't always flow like liquid basalt. Instead of oozing, it explodes into a gigantic cloud of hot, searing pulverized rock that’s somewhat misleadingly called volcanic ash. If that ash is hot enough when it lands, the weight of the layers above it squeezes the particles together into solid rock called welded tuff. But as that material cools, it also shrinks and cracks. Hot gases create voids. Soil and rocks picked up by the ash cloud create porous rubble zones. And seismic activity over millions of years breaks everything even further, leaving joints and fissures. As a result of all of those geologic processes, Teton’s foundation was essentially swiss cheese.

The Bureau of Reclamation carried out some early tests to try and seal up the porous rock with grout. Just a few holes to see how well it would work. They estimated it would take about 260,000 cubic feet of grout injected into the holes to seal them up. It ended up taking more than double that. Some of the holes simply didn’t stop taking grout, no matter how much they put in. The next year, they drilled holes to check if the grout worked out. The core samples they recovered were just as broken, fractured, and porous as before the pilot program. It just wasn’t going to work out. So the Bureau came up with a new plan: core trenches.

If they couldn’t seal up that worst top layer of foundation rock, they would just take the material out and replace it with something more watertight. So the designers of Teton incorporated key trenches below the dam, intended to cut off the flow of water underneath through the foundation. The trench was wider in the valley, but narrower up the abutments to help the dam “lock in” to the steep canyon walls. At the bottom of the trench, the Bureau went ahead with the grout curtain idea, but slightly modified from the pilot program. This time they drilled three rows of holes. The outer rows were grouted first to create an initial seal. Then grout was injected into the center row with the goal of completely filling the space without losing all the material upstream and downstream. Despite what they learned in the pilot program, the project required significantly more material than expected, ultimately injecting roughly 600,000 cubic feet (or 17,000 cubic meters) of grout into the dam’s foundation.

Once that was finished, crews backfilled the core trench along the bottom of the dam with the Zone 1 silt to create a watertight barrier. On the surface, it seems like a reasonable course of action. You have seemingly watertight materials from deep in the foundation all the way to the top of the dam, first with the grout, then the core trench, and finally the embankment on top. But some of the decisions made in that trench would prove fatal.

We’ll never know the exact single cause of the failure, since a lot of the evidence was washed away. But two teams of engineers investigated the wreckage, one from the Bureau of Reclamation and an independent panel. Both reached basically the same conclusion: the dam’s design was doomed to fail all along.

As the reservoir filled up, it reached the upper layers of rock that were the most fractured and started flowing into it. With no real barriers, the flow could move freely all the way to that silty core trench, and somehow, it could also get past it. It could have been that water flowed through “windows” in the imperfect grout curtain underneath. It could have travelled along an area that wasn’t well compacted during construction. Or the reservoir water pressure could have resulted in a phenomenon known as hydraulic fracturing, the same process used in oil and gas production, that you probably know as fracking. If water pressure exceeds the forces holding earthen material together (namely, the weight of the soil above), it can force open cracks, creating pathways for seepage. It’s possible all three contributed to the flow getting past the dam. All we know is that it found a path, and once it got past, it could flow freely in the cracks of the foundation downstream.

At 7:00 AM on June 3, 1976, two days before the collapse, that seepage was spotted on the right abutment. At the time, there was no way to know what path that water was taking. It was running clear, meaning it wasn't carrying away soil yet. But it was a hint of what the water was doing inside that fractured rock.

That Zone 1 material was well-compacted, but all it took was a small area of loose soil to break free. The issue is that, even when well-compacted, it’s not that hard for material to break away. Silt is a highly erodible material. Large particles like gravel and coarse sand mechanically lock together, plus they’re heavy, making it harder to pluck them free. Microscopic clay particles often stick together through intermolecular forces. But right in between, you have silt and fine sand. Too small to mechanically lock together, too large for intermolecular forces to dominate, and very lightweight. The loess soil from the plains along the river was practically the worst choice of material you could make for the core of an embankment dam.

Paradoxically, that erodibility was made worse by the silt’s strength. You’d think that a strong material would be a good thing in a dam, but in this case, you’d be wrong. Where other materials would naturally slump into voids as soon as they formed, “healing itself,” the silt was strong enough to maintain vertical walls and a roof. As it eroded away, it formed a tunnel, a process known as piping. Those pipes just created more space for water to flow and erode more of the dam away from the inside.

Another contributing factor was the narrow width and steep sides of the key trench. The soil distributed downward forces laterally into the rocky walls of the trench, kind of like an arch bridge converts loads into horizontal thrusts at the abutments. Had the full weight of the soil propagated straight down, it would have likely collapsed any tunnels forming through the core, but the arching action allowed them to stay open and grow.

Downstream, the rock was loose, and the fractures and voids were large enough to carry soil particles away. There was a "free exit" for the dam's internals. While Zone 2 of the embankment was intended to be a filter that would keep any erosion from carrying soil away, the actual seepage simply passed underneath it like it wasn’t even there.

By the morning of June 5, a channel had been hollowed out under the dam. Staff saw water exiting at the toe, and it was quickly getting worse. At 10:30 AM, a muddy geyser erupted from the face of the dam. Operators sent bulldozers to try and fill the hole, but the machines were swallowed by the eroding embankment, and the operators had to be rescued with ropes.

Only 30 minutes later, a sinkhole opened up on the upstream face, below the reservoir, giving the water a more direct path through the dam. Multiple witnesses saw a terrifying whirlpool form. The reservoir was draining like a bathtub directly through the center of the dam. More dozers tried in vain to push material on the upstream side to close the hole, but it was too late. The erosion worsened, allowing more flow, causing even more erosion in a runaway feedback loop until finally the embankment gave way at about noon, only 5 hours after the leaks were first noticed.

A wall of water thundered through the hole in the embankment. The breach wave moved quickly downstream in the Teton River, widening out once it left the canyon. The closest town in its path, Wilford, was basically wiped from the map, with nearly all the homes there swept away. Further downstream, Sugar City and Rexburg were similarly decimated, with the vast majority of the buildings rendered a complete loss. Then the smaller farming communities of Hibbard and Salem were inundated, homes and livelihoods washed away. The breach killed thousands of livestock, destroyed railroads, bridges, and farmland, and left thousands without a home. Ultimately, 11 people lost their lives. Had the failure happened at night when no one was watching the dam, it could have been hundreds or even thousands more.

Investigations concluded this was not a "freak accident." The challenges of erodible soil and fractured geology were well known long before the 1970s. The Bureau simply didn't spend enough money to address them. In the end, it was a simple case of frugality over safety. As one investigation said: “Defensive measures, such as rock surface sealing and adequate filters, were well within the state-of-the-art at the time Teton Dam was designed and should have been used.”

But even though it wasn’t a novel failure mode, the impact Teton had on the engineering community was enormous. To see a brand new structure, built by one of the most technically competent agencies in the US, fail in such dramatic fashion was catalyzing in many ways. The federal government responded by standardizing dam safety guidelines across all federal agencies involved in dam construction, rules that are still in place today. The event also spurred research into filter and drainage design for dams, along with a better understanding of the mechanisms of hydraulic fracturing. Finally, it showed engineers everywhere what was possible when geotechnical issues aren’t taken seriously enough.

The very first dam I worked on in my career was very much “Teton-esque.” It was designed by another firm, but during construction, they realized that the foundation was much worse than expected. There were pores, fractures, and even caves below the dam. They had to shut down the job and re-engineer it entirely. Our solution was a deep cutoff wall. It basically involved running an enormous vertical chainsaw along the length of the dam, then filling the trench with a watertight mix of rock cuttings, cement, and bentonite clay.

I vividly remember working through the complexities and challenges of dealing with that foundation material. It’s really easy to understand these failure modes in hindsight. But when you’re actually working on a real project, it’s hard to visualize how soil, rock, and pressure will actually behave underground. Things get pretty abstract, and you can lose sight of the stakes. It’s hard to explain what it meant to have that example of Teton at the top of mind for everyone on the design team. For all the computer models, testing data, and equations, we also had a real life example of what could happen if we got it wrong. It was grounding in a way that textbooks and excel spreadsheets are not. And I like to think that’s how we honor the victims of engineering disasters: to share the stories and keep the lessons from being forgotten to make sure it doesn’t happen again.

This Spillway Failed On Purpose

[Note that this article is a transcript of the video embedded above.]

Hurricane Helene was one of the more unusual tropical storms to hit the United States. In late September 2024, it made landfall on the gulf coast of Florida as a Category 4 hurricane. We’re used to seeing storm damage on the coast from hurricanes, but this time the worst damage was hundreds of miles inland. As Helene tracked northward across the Appalachian Mountains, it dropped a deluge of rainfall, swelling rivers, destroying buildings, washing away bridges, and ultimately causing more than 250 deaths in the US. Places normally immune to tropical storms faced flooding worse than anything in recorded history, with some areas receiving more than three feet (or 900 millimeters) of precipitation. The worst of the rain was in a narrow band centered roughly on Asheville, North Carolina.

Asheville’s primary source of water is the North Fork Reservoir northeast of the city. Built in the early 1950s, North Fork Dam impounds a relatively pristine portion of the Swananoa River. After some earlier major floods and six decades of service life, the dam was starting to show its age, so the City of Asheville embarked on a major rehabilitation project. The project included a new auxiliary spillway to help manage floods and make the dam safer under newer state regulations. It was finished in October 2021, and three years later, nearly to the day, Hurricane Helene hit the region. When it did, a part of that brand new spillway blew out, tumbling down the chute, unleashing a torrent of reservoir water downstream. In other words, it worked exactly like it was designed. I’m Grady, and this is Practical Engineering.

Nearly every dam has a spillway for a pretty simple reason: every once in a while, a big storm comes along. In most cases, it doesn’t make sense to build a dam tall enough to absorb a once-in-a-lifetime flood, and then keep that storage volume empty until one comes. It’s not a good use of resources. Even dams designed explicitly for flood control that intentionally keep some or all of the reservoir empty in anticipation of heavy rain usually aren’t intended to store the largest of floods entirely. Instead, we use spillways to discharge that water in a safe and controlled way so that it doesn’t overtop the dam or cause damage to the structure. I’ve done a bunch of videos about spillways if you want to learn more after this.

One of the most fundamental decisions when it comes to designing a spillway is whether to include gates. The vast majority of dams around the world use uncontrolled spillways, meaning there’s no way to make adjustments in real time. Usually, some kind of weir sets the elevation where the spillway engages and water naturally flows through. Depending on the configuration of the dam and the type of spillway, this might be the normal water level where the reservoir sits when it’s full. Other dams have auxiliary spillways that don’t engage until a higher level. In either case, once the water reaches the crest of the weir, it flows over. A chute controls and directs the flow down, and often a special pool or structure called a stilling basin helps dissipate the energy in the water, making it less erosive as it transitions into a natural channel downstream.

Most spillways are designed according to a simulated extreme storm called the design flood. In many cases, it’s the Probable Maximum Flood, essentially the most extreme inflow that we think is meteorologically possible. The flow through a spillway is proportional both to its width and the height of the water, called the “head” by engineers. So there are some tradeoffs here. For a given design flood, a smaller spillway means the reservoir is going to rise higher as the water builds up waiting to get out. This difference between the reservoir’s normal operating level and the maximum level during the design storm is called the flood surcharge storage. So, on top of the height you need to store the normal water in the reservoir, you also need extra height up to the top of the surcharge storage, plus usually some additional margin for waves. If you widen the spillway, you can get more water out quickly, decreasing the height of the surcharge pool and reducing the need for a taller dam. Smaller spillway, taller dam. Wider spillway, smaller dam. Both have costs, so it’s an engineering balancing act. But with an uncontrolled spillway, there’s no human intervention needed at all. The spillway discharges water when the reservoir reaches a certain elevation, and that’s it. There’s a fixed relationship between the reservoir level and the discharge rate, called the spillway’s rating curve. But sometimes you need more flexibility than that.

Adding gates doesn’t increase the width of a spillway, but it can change that second part of the equation: the head. And it really only makes sense for reservoirs designed to hold a permanent pool of water, usually for irrigation or water supply. Obviously, with an uncontrolled spillway, you have to choose a crest height above the level of that permanent pool or you would just lose all your water. You can only use the height above the crest to drive that water through the spillway. Not true if you have gates. Opening a gate instantly gets you a lot more head above the spillway crest, providing greater flow. That means, for a given design storm, a gated structure can be a lot narrower. You don’t need to rely on width to get the water out. Of course, the gates are an added expense, but there are situations where that cost is offset by the reduced width of the spillway. Plus you have a lot more flexibility. Discharge is no longer fixed to the level of the reservoir. You can adjust releases based on season or downstream conditions or even forecasted inflows, providing greater control.

Those gates don’t only add to a project’s overall cost; they also add to the complexity. You have moving parts, which means more wear and tear and more maintenance. Gates rely on hoists or hydraulics, seals, gearboxes, and other specialized equipment where knowledge and replacement parts aren’t always readily available. The other thing is: they need someone to open them when a storm comes.

There are plenty of spillways equipped with some level of automation, but in general you want a real human brain in the decision tree. Remember that spillways are a critical safety feature of a dam. The whole purpose is to protect the structure so it doesn’t breach during a flood, the consequences of which can be catastrophic. On the other side of that coin, opening floodgates can be dangerous to people and property downstream. Dams with gated spillways usually have elaborate systems to warn people when making releases, including lights, sirens, and sometimes even emergency alerts sent to cell phones. A gate opening when it shouldn’t can be almost as bad as one not opening when it should have. There are risks on both sides. That means nearly all gated spillways require someone to be on call 24/7/365 to make sure operations go to plan. It means checking the weather forecasts every day, testing gates regularly to make sure they stay operable, and having staff available mornings, nights, weekends, and holidays in case of a storm. It is a major obligation, especially when you consider the decades or centuries-long lifetimes of these structures. There cannot be a single day when someone isn’t available to handle a flood. For large organizations, like federal agencies or water districts, it’s definitely doable. Most of the largest dams in the world have gated spillways with whole teams of staff dedicated to their operation. But it’s still a challenge, especially for small owners like cities. So there is another option, kind of in between controlled and uncontrolled spillways, and its use is growing worldwide.

Behold, a fuse plug spillway. Let me put some water in this flume and show you how this works… You can see water builds up on the upstream side, but none is released yet. As soon as the water rises above the plug, things happen pretty quickly. The overtopping water erodes the fuse plug down, quickly washing it away and opening up a much larger area for the water to flow. It’s basically a floodgate made of dirt. Obviously, in my little demo, there’s not really a reservoir, so the water level drops pretty quickly back down. But you can imagine if there was a larger volume of water to release, the difference in flow rate before and after the fuse plug washed out would be pretty dramatic.

It’s funny because this is exactly what you don’t want to happen at an embankment dam. Overtopping is basically a worst-case scenario precisely because of how that erosion can cut through an embankment so quickly. But that erosive force can be used in a beneficial way on a spillway. It’s a little crude, but the advantages are obvious. You don’t need a person on site to operate gates, and there are no moving parts. Plus, maintenance for an earthen structure is a lot simpler than for mechanical and electrical components. Just like an electrical fuse is a small section of wire that fails before the main wiring fails, the fuse gate is like a mini-dam that fails before the big dam is at risk.

Of course, this takes some pretty careful engineering. The materials you use for a fuse plug have to be both sufficiently durable - able to consistently hold water back for non-overtopping reservoir levels - but also relatively erodible so that they will wash out in a predictable and controlled way when called upon to function. Usually, this means a zoned embankment, where part of the structure is pre-weakened using erodible materials like sands, silts, or fine gravel. Many fuse plugs include a pilot channel or notch to give the erosion a head start. So you tune both the materials and the geometry of the fuse plug so it performs as intended. And these are used in quite a few dams. One of the most famous examples is at Warragamba Dam in Australia that provides the primary source of water for Sydney. You can see that the service spillway in the center of the dam still uses gates to control more frequent, lower magnitude floods. But each bay of the auxiliary spillway is equipped with a fuse plug of earth and rock fill. The crests of each plug are staged so they don’t all wash away at the same time. As the reservoir gets closer and closer to the top of the dam, more of the bays will open up to increase the discharge capacity of the spillway. But these structures aren’t foolproof.

In 2003, the fuse plug spillway failed at Silver Lake Basin, a reservoir in a remote part of Michigan’s Upper Peninsula. No one was hurt, but the event prompted the evacuation of nearly 2000 residents. Bridges were washed out, and the failure inflicted millions of dollars of damage to the areas downstream. When the fuse plug overtopped, it eroded down as designed, but the erosion didn’t stop. The foundation soil was just as erodible, if not more, than the fuse plug, and water continued to cut downward until most of the lake had drained out. It’s a good case study in why engineers only use soil erosion as a failsafe measure in limited situations. It’s hard to predict and hard to control. So there’s a similar solution to this kind of fusible spillway that avoids it altogether.

I’ve removed the fuse plug in my demo and replaced it with something a little more elaborate. I mounted a sliding bracket on the side of the flume now. On one side is a float, and on the other is a little arm. And I have a crest gate mounted to the bottom. Let me get this set up and turn on the water. You can see just like the fuse plug, this holds back the water when the reservoir comes up. And actually, this gate can allow water over the top as it gets higher. But at a certain point, my mechanism slides up (pushed by the float), and the arm clears the top of the gate. When it does, the gate folds down, quickly opening up the spillway for a lot more flow.

This has a major benefit over fuse plugs in that it can release some water before it fully opens. The gate basically acts like an uncontrolled spillway until the reservoir reaches the literal tipping point. It’s not an all-or-nothing thing like the fuse plug. The other benefit here is control. I can adjust the float or the arm to change the exact point when this gate opens, unlike an erodible structure that has some inherent uncertainty around the amount and the duration of flow required to wash it out. But you might be thinking: “Grady, this is a mechanical system with a sliding bearing and moving parts.” And you’d be exactly right. You’re not likely to find a system exactly like this installed on a dam. It’s not even that reliable in my model, to be honest, so I wouldn’t trust it at full scale. I’m just using it to show the fundamental advantages because all of the fusible concrete spillways that I know of around the world use a proprietary system called Fusegates developed by the company, Hydroplus. I didn’t want to step on any of their patents by building a model in my garage, but the way they work is pretty clever.

Fusegates are concrete structures set on top of a platform with a chamber built into the bottom. An inlet connects the chamber to a prescribed elevation above the gate. When the reservoir reaches that target elevation, water flows into the chamber, pressurizing it just enough that the gate loses stability and tips downstream. The benefits are the same as my demo: namely, that you can discharge water before the gate washes out, and the precise control you get over when the gate tips. A lot of dams around the world have been equipped with Fusegates. In the US, a few high-profile projects include (of course) the North Fork Dam in Asheville, Canton Dam in Oklahoma, and Terminus Dam that holds back Lake Kaweah in California.

One important application of both fuse plugs and Fusegates is extending the life of an existing reservoir. I’ve talked about sedimentation in a previous video, where a reservoir gradually loses storage as it fills up with silt and sand transported from upstream. There are no easy fixes, and there are plenty of cases where dams have to be decommissioned or removed because they just don’t have enough storage anymore. For dams that use uncontrolled spillways, the volume typically reserved for flood surcharge above the spillway crest is kind of an untapped resource. So, there are projects where a fuse plug or similar-type spillway is retrofitted onto an existing dam to gain more storage without sacrificing spillway capacity. In some cases, this can save millions of dollars associated with decommissioning a dam and developing an alternative source of water. But there are some downsides too.

When a fuse plug or tipping spillway activates, it’s a major endeavor to put it back. Unlike a gate that you just close after the flood is over, replacing a fusible spillway is a construction project, which brings along all kinds of complications, like hiring an engineer, procuring a contractor, significant expenses, and a lot of time. The time is important because, until it’s replaced, you’ve lost a lot of storage in your reservoir. The other disadvantage to these systems is also what makes them useful in the first place: there’s no human control. It does make them safer; it also means that there may be little warning when they activate. In places with a lot of development downstream, that’s a big deal, because dramatic and sudden increases in water levels are dangerous. That’s why these systems usually break up the fusible structures into stages that give way at different reservoir levels, smoothing out the changes in flow as a flood passes through. But even then, it can still cause problems.

North Carolina came face-to-face with the issue when Hurricane Helene hit in 2024. A major impetus for the new auxiliary spillway at North Fork Dam was a previous storm, Hurricane Frances. Flows through the old spillways washed out key pipelines that carry water from the treatment plant at the dam into Asheville. In response, the city built a new bypass line to provide redundancy against failures. When Hurricane Helene hit and tipped one of the Fusegates at the auxiliary spillway, the surge eroded the channel downstream, taking out not just the original transmission lines but the bypass line too. So, somewhat ironically, the flood left major parts of the city without water for weeks.

The water crisis in Asheville was just one of the problems caused by Hurricane Helene along its path. But I think it’s important to recognize the tragedies that didn’t happen too, one of those being that North Fork Dam was never in any danger of breaching. Despite the incredible rainfall, and despite the fact that there was no way to control releases, the flood passed through exactly as designed. The fusible spillway tipped just when it was supposed to, allowing more discharge during an extreme event, and no one had to be there to push a button.

The Los Angeles Aqueduct is Wild

[Note that this article is a transcript of the video embedded above.]

On the northern edge of Los Angeles, fresh water spills down two stark concrete chutes perched on the foothills of the San Gabriel Mountains, a place simply called The Cascades. It’s a deceptively simple-looking finish line: the end of a roughly 300-mile (or 500 km) journey from the eastern slopes of the Sierra Nevada into the city.

On November 5, 1913, tens of thousands of people climbed these hills to watch the first water arrive. When the gates finally opened, water trickled through, but that trickle quickly became a torrent. The project’s chief engineer, William Mulholland, leaned over to the mayor and shouted the line that’s been repeated ever since: “There it is, Mr. Mayor. Take it!”

That moment was profound for a lot of reasons, depending on where you live and how you feel about water rights. LA didn’t become LA by living within the limits of its local resources. Its meteoric growth into the metropolis we know was enabled by an early and extraordinary decision to reach far beyond its own watershed and pull a whole new river into town. Today, roughly a third of LA’s water comes from the Eastern Sierra through the Los Angeles Aqueduct system. That share swings with snowpack, drought, and environmental constraints, but this one piece of infrastructure helped turn a water-limited town into a world city. It’s one of the most impressive and controversial engineering projects in American history.

But to really appreciate that water in the cascades, you have to look way upstream and see what it took to get it there. It’s gravity, geology, politics, and human ambition all in a part of the state that most people never see. Let’s take a little tour so you can see what I mean. I’m Grady and this is Practical Engineering.

When most people think about aqueducts, this is what they picture: a bridge carrying water over a valley or river. And, just to be clear, these are aqueducts. But engineers often use the term more broadly to describe any type of conveyance system that carries water over a long distance from a source to a distribution point. Could be a canal, a pipe, a tunnel, or even just a ditch. In the case of the LA aqueduct, it’s all of them, plus a lot of supporting infrastructure as well.

From the center of the city, it’s about a four hour drive to the Owens River Diversion Weir. It’s not accessible to the public, but it is the official start of the LA Aqueduct, at least when it was originally built. Here, all the snowmelt and rain from a huge drainage system between the Sierra Nevada and Inyo Mountains funnel down into the Owens River, where a large concrete diversion weir peels nearly all of it out of its natural course and into a canal. This point is roughly 2,500 feet (or 750 meters) higher in elevation than the bottom of the Cascades at the downstream end, which makes it obvious why LA chose it as a source. The entire aqueduct is a gravity machine. There are no pumps pushing the water toward the city. Half a mile of elevation change feels like a lot until you realize you have to spread it out over 300 miles. It’s all achieved through careful grading and managing elevations along the way to keep the flow moving.

That care is particularly important in this upper section of the aqueduct, where the water flows in an open canal. To do this efficiently, you need a relatively constant slope from start to finish. That’s a tough thing to achieve on the surface of a bumpy earth. Following a river valley makes this easier, but you can see the twists and turns necessary to keep the aqueduct on its gentle slope toward LA.

If it seems kind of wild that a city would buy up the land and water rights from somewhere so far away, it did to a lot of the people who lived in the Owens Valley, too. A lot of the acquisitions and politics of the original LA Aqueduct were carried out in bad faith, souring relationships with landowners, ranchers, farmers, and communities in the area. The saga is full of broken promises and shady dealings. Then when the diversion started, the area dried up, disrupting the ecology of the region, making agriculture more difficult and residents even more resentful. Many resorted to violence, not against people but against the infrastructure. They vandalized parts of the aqueduct, a conflict that later became known as the California Water Wars. In one case in 1924, ranchers used dynamite to blow up a part of the canal. Later that year, they seized the Alabama Gates.

About 20 miles or 35 kilometers downstream from the diversion weir, a set of gates sits on the eastern bank of the aqueduct canal. Because it runs beside the river valley, the aqueduct captures some of the water that flows down from the surrounding mountains in addition to what’s diverted out of the Owens River, particularly during strong storms. That means it’s actually possible for the canal to overfill. The Alabama Gates serve as a spillway, allowing operators to divert water back down to the river. This also helps drain the canal for maintenance or repairs when needed.

Those Owens Valley ranchers understood exactly what the Alabama Gates controlled. Open them, and the water would run back where it had always run, down the Owens River, instead of south to Los Angeles. The resistance simmered and flared for years, but it didn’t end in the dramatic showdown at the aqueduct. Instead, it ended at a bank counter. The Inyo County Bank was run by two brothers who were also key organizers and financiers of the resistance campaign. In August 1927, an audit revealed major shortfalls and ongoing embezzlement, and the bank quickly collapsed. Residents across the valley saw their savings wiped out or frozen overnight, shattering what was left of the community’s ability to keep fighting.

The Alabama Gates weren’t just a political flashpoint though. They also marked an important dividing line in the aqueduct’s design. LA knew that even if the ranchers didn’t release the water to the river in protests, a lot of it would end up there anyway through seepage. As the canal climbed away from the valley floor and crossed more porous soil, it would naturally lose its water through the ground. So, at the Alabama Gates, the aqueduct transitions from an unlined canal to a concrete-lined channel. It’s still open to the air, so there’s no protection against evaporation or contamination, but the losses to the ground are a lot less.

This design continues for about 35 miles (or 55 kilometers) through the valley. Along the way, the aqueduct passes the remains of Owens Lake. Once a large body of water, it quickly dried up with the diversion of the Owens River. Of course, there were impacts to wildlife from the loss of water, but the bigger problem came later: dust. All the fine sediment that settled on the lakebed over thousands of years was now exposed to the hot desert sun. When the wind picked up, it filled the air with fine particulates that are dangerous to breathe. Over the years, there have been times when Owens Lake is the single largest source of dust pollution in the entire country, and LA has spent more than a billion dollars just trying to fix this problem alone. The aqueduct passing along the hillside past the lake and its challenges is a reminder that the true cost of water is often a lot more than the infrastructure it takes to deliver it.

So far, it might be obvious that this aqueduct system is pretty fragile to be making up a major part of a city’s fresh water supply. Even beyond the vandalism and political resistance, there are a lot of things that could go wrong along the way, from bank collapses, earthquakes, diversion failures, and more. That’s why Haiwee Reservoir was originally built in a narrow saddle between two hills as a kind of buffer. With a dam on either side, it stored water up so the aqueduct could keep running even during a disruption upstream. It also slowed the water down, exposing it to the hot desert sun as a natural form of UV disinfection. In the 1960s, the reservoir was reconfigured into two basins to add some flexibility. That’s because, around that time, the LA aqueduct became two. While the open-topped canal section was large enough to meet demands, the underground conduit in the next section wasn’t. So, LA built a second one in 1970 to increase the flow. If you look at this map of the Haiwee Reservoirs, you can see that water has two paths: it can flow into the second aqueduct here from the north basin, or it can pass through the Merritt Cut to the south reservoir, through the intake there, and into the first aqueduct. This setup allows for some redundancy, along with regulation and balancing of the flows between the two aqueducts. Haiwee marks the start of the long desert run, with both systems no longer in open-topped lined canals, but running underground in concrete conduits.

There are a lot of advantages to running an aqueduct in a closed conduit underground, especially one this long through a desert landscape. There’s far less evaporation and less potential for contamination. It doesn’t divide the landscape at the surface level, so there’s no need for bridges, culverts, and wildlife crossings. Going underground also offers more flexibility when it comes to topography. You don’t have to follow the contours of the surface so carefully because if you come to a hill, you can just dig a little deeper to keep the constant slope.

Of course, those benefits come with a cost. An underground conduit is more expensive than a simple channel on the surface, and not all the problems with topography are solved. This is Jawbone Canyon, one of the biggest drops for the first aqueduct. Rather than taking a major detour around it, the aqueduct descends 850 feet (or 250 meters) and then ascends back up. This type of structure is often called an inverted siphon. I’ve done a video on how these work for sewer systems, and I’ve also done a video on flood tunnels that work in a similar way, if you want to learn more after this.

Unlike the concrete conduit, which really just acts like an underground canal with a roof, this is one of the places where the water in the aqueduct is pressurized. 850 feet of water column is about 370 psi, 26 bar, or two-and-a-half Megapascals. It’s a lot of pressure. These sections of pipe had to be specially manufactured on the East Coast, where the major steel facilities were, and transported by ship because of their size. They travelled all the way around Cape Horn, since the Panama Canal was still under construction. There are actually quite a few of these siphons crossing canyons in this section of the aqueduct, but Jawbone Canyon is the biggest one.

A little further downstream, the LA aqueduct crosses the California Aqueduct, part of the State Water Project. That system has a connection to LA as well, but this branch at the crossing actually heads to Silverwood Lake. However, there is a transfer facility, recently completed, that can pump water out of the California Aqueduct directly into the first LA aqueduct. This creates opportunities for LA to buy water that moves through the state system and offers some flexibility in where that water ends up. There’s also a turn-in that can move water from the LA aqueduct into the California aqueduct for situations where trades make sense. The second LA aqueduct passes underneath the state canal here. And this is a good example of the differences between the first project (built in the 1910s) and the second one, built in the 1960s. Over that time, the price of labor went up a lot more than the price of materials. Where the first one carefully followed the existing topography with bends and turns to minimize the need for expensive pressurized pipe, the second one could take a more direct path, reducing labor in return for the more specialized conduit materials.

After wandering more than a hundred miles (or 160 kilometers) apart, the two Los Angeles Aqueducts come back together at Fairmont Reservoir, in the northern foothills of the Sierra Pelona Mountains. This is the last major topographic barrier on the way to Los Angeles. There was no way to go up and over without pumps, so instead they went straight through. The largest project was the Elizabeth tunnel.

Here, the two aqueducts come together again into a single watercourse. About 5 miles or 8 kilometers of excavation through everything from hard rock to loose, wet ground became one of the most difficult parts of the entire project. The tunnel required continuous temporary supports along most of its length, followed by a permanent concrete lining. It was a monumental effort for its time and essential not only to cross the range. The Elizabeth Tunnel also delivers that water under pressure to the San Francisquito Power Plant Number 1.

This is the largest of the eight hydroelectric plants that run along the aqueduct, capturing some of the energy from the water as it flows downward toward LA. These plants are a major part of how the project paid for itself, and they continue to serve as an important source of electricity in the region today.

Continuing downstream, Bouquet Canyon reservoir adds another layer of operational flexibility. It helps regulate flow through the power plants and provides additional storage, a sort of insurance policy since this whole reach depends on a single major tunnel crossing the San Andreas Fault. In case of a major earthquake, it’d be best if Angelinos could avoid a simultaneous water shortage.

The aqueduct splits again just upstream of the San Francisquito Plant Number 2, which was famously destroyed by the St. Francis Dam failure. That reservoir project was designed to supplement the storage capacity along the aqueduct, but the dam failed catastrophically in 1928, just 2 years after it was completed, killing more than 400 people and destroying several parts of the aqueduct as well. The tragedy was one of the worst engineering disasters in American history. It put another stain on the aqueduct project, and it effectively ruined the reputation of William Mulholland, who was largely considered a hero in LA for all his work on the aqueduct and the city’s water system. The dam was never rebuilt, but workers restored the aqueduct to functioning service in only 12 days.

At Drinkwater Reservoir, the two aqueducts run roughly parallel through the Santa Clarita area, sometimes aboveground and sometimes below, before finally reaching the terminal structures that carry water into LA. Usually, the water stays in the conduits, which feed the two hydropower plants at the foot of the mountains. If the plants are out of service or there’s more flow than they can handle, you see excess water thundering through the cascade structures instead.

From here, the aqueduct drops out of the mountains and into the north end of the San Fernando Valley, where the water is treated and prepared for distribution. After filtration and disinfection, it’s stored in the Los Angeles Reservoir, the system’s terminal reservoir, so the city can smooth out day-to-day swings in demand even while the aqueduct’s inflow stays relatively steady.

For most of Los Angeles' history, that “finished water storage” was out in the open air. But in the 2000s, drinking-water rules pushed utilities to add stronger protection for treated water held in uncovered reservoirs. There’s a good chance you’ve seen their solution on the Veritasium channel or elsewhere: 96 million plastic shade balls that act like a floating cover, blocking sunlight to prevent water-chemistry problems and helping keep wildlife out. They’re the final protection for this water that traveled so long to reach the city. While the LA Reservoir is, in a sense, the end of the journey for this water, the original diversion way back at Owen’s River isn’t even technically the start anymore!

In 1940, LA extended the aqueduct system upstream northward by connecting the Mono basin and funneling its water through tunnels to the Owens River basin. Like Owens Lake downstream, Mono Lake began drying out as well. And also like Owens Lake, lawsuits, court orders, and environmental regulations have tempered the value of this water source, forcing LA to significantly reduce diversions and implement costly restoration projects.

That’s kind of the story of the LA aqueduct in a nutshell. The project seemed obvious from an engineering perspective. There was lots of snowmelt in the mountains; the city had the technical prowess, the funding, the elevation, and the political power to reach out and take it. The result was one of the most impressive works of infrastructure of the early 20th century. And continued efforts to expand and improve the system have made it even more efficient, flexible, and valuable to the many millions of people who live in one of the most populous cities in America, delivering not only water but also hundreds of megawatts of hydropower.

But it many ways, it was not only unscrupulous, but also short-sighted. Residents of the Owens Valley watched ranchland and farmland dry up as the water that had shaped their home was rerouted south. Native communities saw their homeland transformed with access to gathering areas disrupted, places made unrecognizable, and cultural ties strained by changes they didn’t choose. Wind picked up alkaline dust from dried lakebeds. Habitats were disrupted, and the birds that depended on these waters and wetlands lost part of what made this migration corridor work. It’s easy to see why the aqueduct remains controversial, and why what we sometimes dismiss as “red tape” around major infrastructure is often completely justified due diligence. As engineers, and really, as humans, we have to try and account for costs that don’t show up on a balance sheet, but can come back later as decades of lawsuits, mitigation, and restoration.

And even the aqueduct’s original thesis (that there’s reliable snowmelt up there, and a growing city down here) is starting to falter. In recent decades, the mountains have delivered less predictable runoff: more swings, more years when the timing is wrong, and more uncertainty about what “normal” even means anymore. California’s climate has always moved in long cycles, but the margin for error is thinner now, and no one can say with much confidence when or if the moisture the state depends on will return to its old pattern.

The hopeful part is that this is exactly where engineering makes a difference: at the messy intersection of geology, climate, culture, politics, and human need. The Los Angeles Aqueduct is a case study in what we can build when we’re ambitious, but also what happens when we treat a landscape like a machine with only one output. The next era of water engineers can learn a lot from it.

Earthquake-Proof Foundations

[Note that this article is a transcript of the video embedded above.]

In 1994, the Northridge earthquake struck the Greater Los Angeles area in California. The shaking was so intense in the west San Fernando Valley that buildings collapsed and bridges were ruined. Billions of dollars of damages were inflicted on parking garages, office towers, apartments, and warehouses, thousands were injured, and 57 people died. One factor that compounded the earthquake's devastation was the extent of the damage it caused to hospitals. Just when they were needed most, eleven of the area's hospitals were completely or partially closed, with patients evacuated to outside medical centers. But one hospital stayed open.

At USC University Hospital, now USC Keck, patients felt the rumble, but it was more of a gentle sway than the violent tremors felt in other parts of Los Angeles. The facility was one of the few in the area that was able to accept new patients after the earthquake. The reason it could stay open amid the chaos all comes down to the foundation. The hospital was built in 1991, only three years prior to the quake, and the design included what was a pretty innovative idea for its time (at least in the US). It is dead simple in theory: just physically decouple the superstructure from the ground. Of course, in practice, there are a lot of engineering details to get right. So I built a little model to show you how this works. I’m Grady, and this is Practical Engineering.

Earthquakes can put enormous demands on buildings. In seismically active areas for low- and mid-rise buildings, there really is no worst-case loading condition for structural engineers. It’s a big deal.

Earthquakes are a lot like floods: you can’t predict the timing, and their intensity varies tremendously. In most situations, you just can’t afford to design for the worst case. If you can imagine a destructive earthquake, there’s an even stronger one possible. Of course, they’re difficult to design against, and we’ll get to that, but there’s the additional challenge of just drawing a line in the sand that says: “This is the level of risk we’re willing to accept.” With floods, we often choose a somewhat arbitrary limit, like the 1-percent storm, to size drainage infrastructure and delineate the floodplain where development is restricted. Same idea with earthquakes. The code outlines site-specific seismic risks across the US and says, essentially, your building has to “survive” this amount of shaking. But that word “survive” might not mean what you expect.

Earthquake accelerations and the resulting deflections can be so strong that it usually isn’t feasible to design a building to completely resist the forces. Instead, we design buildings to absorb earthquake energy by deforming and yielding in a controlled and predictable manner. Seismic loads aren’t like other structural demands, like gravity and wind, where we add strength as needed to resist them entirely. If the “design earthquake” required by the code ever comes, the expectation is that ordinary buildings are going to be damaged, maybe even beyond repair. The intent is simply that they don’t collapse. Let’s go to the garage and I’ll show you what I mean.

I designed this little shake table so we can simulate an earthquake on the bench. It just uses a cordless drill for motion. I’m doing this in one dimension for simplicity, but most earthquake engineering has to consider shaking in any horizontal direction. Vertical motions usually aren’t as intense in earthquakes, and buildings are a lot stiffer in the vertical direction since they sustain a full G of vertical acceleration constantly, so it’s not as big a factor in design.

I built a little building using these magnetic pieces, and let’s give it a little shake. This is exaggerated for effect, but I wouldn’t want to be in there! It’s a little worse for wear after the quake, but as a whole, it’s still standing. This is what the code allows for many types of buildings. It’s all about life safety: protect the people inside. And it’s generally a reasonable tradeoff. For rare events like earthquakes, it’s often just not feasible to invest in a building that’s strong enough to come away from the shaking unscathed. But there are situations where it makes sense to have additional protection.

Hospitals are the perfect example. Even if the building doesn’t collapse, it might need to be evacuated and closed while the structure is repaired after an earthquake. Lives are still endangered by interruptions in procedures and lack of capacity. And think about all the expensive equipment and contents inside. Even if the building itself survives a major earthquake, a lot of that stuff is going to be damaged beyond repair. When you factor in the impact of ruined contents and the time it would take to get things back up and running, it tilts the calculus of savings versus safety when it comes to earthquake design. This concept is called resilience. For offices and homes, we might be willing to accept the risk of having to rebuild after an earthquake to avoid the cost of building an extremely strong structure. For hospitals, fire stations, emergency shelters, and other critical buildings, it’s just not acceptable. They need to be more resilient. But there’s a challenge there.

Let me stiffen up my building on the shake table, and I’ll show you what I mean. Now this structure is resilient against earthquakes. Give it a good shake and all the pieces are still connected just the way they were before. But watch what happens when I put some stuff inside. Almost all the accelerations from the ground are transmitted through the building, where contents feel the shaking. You can see how stiffness plays two parts: it keeps the structural members from bending and deflecting so much, but it also means that more of the accelerations are “felt” by the building and what’s inside. So sometimes we need an alternative. For that, we need to look at an earthquake.

This is an accelerogram of an earthquake. It’s a simple plot to understand: time on the x-axis; acceleration on the y-axis. It looks almost like a sound wave you might record through a microphone. By itself, it isn’t that useful at first glance, except, I guess, to let you know that earthquakes are “noisy.” The ground movements are broadband; they contain a lot of different frequencies. And frequency matters a lot in seismic engineering. Let’s go back to the garage to see why.

I have a few rods set up on the table now. These are very simple oscillators, and I can excite them with my drill. If I shake it slowly, the longest rod has an extreme response while the others are mostly unbothered. A little faster, and now it’s the medium rod responding. Faster still, and the smallest rod is responding while the other two are barely moving. The displacement is the same for all three oscillators, but the response is completely frequency-dependent. I think this is pretty intuitive. Instead of frequency, we usually talk in terms of period in seismic engineering. That’s the time it takes to complete one cycle of an oscillation. Every structure has a fundamental period at which it naturally sways. For short buildings it’s usually less than a second. For skyscrapers, it can be multiple seconds. And the closer an excitation gets to a building’s fundamental period, the more its response to that excitation grows.

Imagine if I could build a shake table with a lot of these oscillators, ranging in fundamental period, and then take our accelerogram from earlier and feed it into the demo. Some of the oscillators would shake a lot. Some barely at all. If we plot the response of the whole group, we get this graph called a response spectrum. This is the plot for a single event. For many earthquakes, acceleration tends to peak for periods less than a second. Plus, oscillators with a longer period naturally smooth out rapid ground motion. So it’s pretty typical that you have a stronger response at the lower periods. The building codes look at a wide range of earthquakes to create site-specific hazard curves that look a little like this: a high plateau at the shorter periods that decays as periods get longer. If you’re the designer of a low or mid-rise building, this curve is an issue for you. With a shorter fundamental period, your building is at the worst part of this curve; it’s the most susceptible to seismic ground movement.

This is why it’s common to see more damage in shorter buildings after an earthquake, when the skyscrapers survive unscathed. Lower buildings have shorter fundamental periods, so they have to be designed against much higher accelerations. But what if you could just adjust a building’s natural period so that it “feels” less power from the earthquake? That’s the idea with base isolation. Let me set up another example.

Now my building is on rollers, and I’m using rubber bands to hold it in place. I’ll put the two examples side-by-side so you can see the difference. Pretty remarkable. And here’s what happens to the stuff inside. What I love about this solution is how intuitive it is. You know, all that exposition about acceleration spectra and oscillators and ground motion response… you don’t need any of that to understand how this works. And because of that, it’s not really a new idea. The concept of using a loose connection between foundation and underlying soil has been used for centuries, and many historic monuments and buildings have probably benefitted from an isolated base, whether it was an intentional seismic protection or not. Lots of older patents on the idea make use of rollers or ball bearings, kind of like my demo. The general idea is the same in all cases: make the structure’s fundamental period longer to get the peak accelerations down. Modern designs typically use one of two solutions.

The first is rubber bearings. Instead of resting directly on the pile foundation, isolation devices separate the superstructure from the substructure. When an earthquake comes, the building acts more like a skyscraper, smoothing out the high-frequency ground motions so the floors feel less shaking. Early isolators were plain rubber, but it didn’t work that well. It would bulge out under the weight of the building, making the bearings less effective over time. Modern isolation bearings use a composite of steel plates with rubber in between them. This makes them highly elastic in the horizontal direction, but stiff in the axial direction so they can withstand a lot more load without bulging. And this system has a lot of advantages.

Of course you get the lengthening of the building’s fundamental period, which lowers its response to most earthquakes. That makes a huge difference in resilience for both the structure and the stuff inside it, shortening or entirely eliminating the time required to put it back into service after a seismic event. It also means that the structural members themselves don’t have to be as strong. I put the original, more flexible structure on the shake table with the isolation system, and it held up just fine. Even the stuff inside it was mostly unbothered by the shaking.

So you sometimes see an offset in the cost of the isolation system from the rest of the building, and there are situations where they pay for themselves entirely. They’re also also big for retrofits. Strengthening older buildings to meet modern seismic codes is often disruptive to architectural elements or even to the structure's function. New beams and braces require removal of decorative features or changing the floor plan. Instead, you can just temporarily support the building and insert isolators below. That’s exactly what they’re doing on the Salt Lake Temple in Utah, and the same idea has been used in a lot of seismic retrofit projects across the world.

That’s a lot of advantages, but we can do even better. Isolation lengthens the fundamental period, reducing the building’s response to an earthquake, but it doesn’t inherently absorb the energy. We don’t want our buildings to vibrate like a guitar string ringing out, so damping is also important. There are lots of ways to improve the damping of buildings. I’ve done a couple of videos on unique solutions in tall buildings. But base isolation systems make it easy. You can just use a special blend of rubber in the bearings that absorbs some of the energy instead of transmitting it into the superstructure (so called high damping rubber). Another option is to use a lead plug in the center of the bearing. As the plug plastically deforms, it absorbs the energy of the shaking. This reduces the accelerations of the building even further and helps kill the oscillations faster after an earthquake, so the building doesn’t ring like a bell. And not just buildings, but other structures too. Lots of bridges use similar isolation systems to manage earthquake loads.

Rubber bearings are a great solution and probably the most widely used base isolation system. But their properties can change based on temperature, they’re kind of bulky, they’re susceptible to damage from exposure to oils and ozone, and they can degrade with age. They’re not always the perfect fit for every application. An alternative is curved surface sliding bearings, often known by one of the most common trademark names, friction pendulum isolators. Instead of rubber, these use a sliding element on a curved surface to separate the superstructure and substructure. If an earthquake comes, the building rides on the bearings. Damping comes from the sliding friction, and the restoring force comes from the curved surface, keeping the building in place when the shaking is over. These get pretty creative with two or three separate elements that offer more control over the displacement and period under a wide range of accelerations. The amount of displacement a building experiences on the isolators is pretty important, because buildings don’t just float on their foundations. They’re connected to stuff!

People have to get in and out of buildings, of course. Usually, that’s not so complicated, since most structures don’t move relative to the ground, but it’s not true for base-isolated buildings. For the system to work, a building’s only connection to the ground has to be through the bearings. So buildings equipped with a seismic isolation system often have some kind of “moat” around them, providing a space to move. It’s usually pretty easy to spot an isolated building because it has a big gap all the way around, although often this joint is bridged with some kind of expandable cover, similar to the way expansion joints work on bridges. Utilities are also a challenge. You can’t have a solid water or sewer connection to a building that wiggles around. So, engineers have to design sometimes elaborate, flexible connections for water, sewer, gas, and electricity. Finally, you have to be thoughtful about long-period earthquakes. There are situations where a base isolation can actually amplify the shaking if it’s not carefully tuned.

Earthquakes are such a challenge in engineering: unpredictable both in time and intensity. There’s only so much you can do when the ground underneath your structure shakes uncontrollably. But I love this solution of base isolation because it’s so intuitive. It’s something my five-year-old would come up with: just put a suspension system on a building. But it works! It works really well, actually, to the point where it's been installed on thousands of buildings across the world. It’s not a catch-all, but I suspect that, as the technology improves and as we get more data about how these buildings perform in real-world situations, it’s only going to be more common to see little moats around important buildings in seismically active regions. Most people will probably never notice, but you and I will know what’s underneath.

So You Want to Build a Tunnel...

[Note that this article is a transcript of the video embedded above.]

It seems like homemade tunnels are kind of having a moment. Just about everywhere I look, it feels like someone is carving new spaces from the ground and documenting the process online. Colin Furze might be the quintessential example, with his wild tunnel project connecting his shop and house to an underground garage. You can watch the entire process in a series of videos on his YouTube channel, and he even started a second channel to share more details of the build. But he’s far from the only one.

TikTok creator Kala, lovingly nicknamed “Tunnel Girl,” has been sharing the almost entirely solo excavation of a tunnel system below her house, amassing more than a million followers in the process. Zach from the JerryRigEverything channel has an ongoing series about a massive underground bunker project. Not strictly a tunnel, but in the same spirit. In Wisconsin, Eric Sutterlin and a team of volunteers have built Sandland, which features a maze of sandstone tunnels in the hillside that can occasionally be seen on the Save It For Parts Channel. My friend, Brent, bought the abandoned mining town of Cerro Gordo and regularly explores the shafts and drifts on his channel, Ghost Town Living. And there are lots more. Wikipedia has a whole page about “Hobby Tunneling,” which it defines as “tunnel construction as a pastime.”

There’s something captivating about subterranean construction, delving into the deep, carving habitable space from the earth. In one case in Toronto, a tunnel was discovered in a public park, sparking headlines worldwide and fueling wild conspiracy theories about terrorist plots. Turns out, it was just a guy who liked digging. When he was interviewed by Macleans, he said (quote) “Honestly, I loved it so much. I don’t know why I loved it. It was just something so cool…”

What more can you say than that? Some of us just yearn for the mines. Plenty of people have front yards and back yards, but not everyone has an underyard. But the thing is: underground construction is pretty dangerous. And not only that; it also poses a lot of very unique engineering challenges that a hobbyist might not be prepared to solve. So I thought it might be fun to do a little exploration into modern tunnel construction methods used in public infrastructure and how those lessons can be applied to endeavors of the more homemade variety. Don’t take it as advice; I am a civil engineer, but I’m not your civil engineer. That said, maybe I can at least give you a sense of what’s involved in a project like this, and some things you might want to study further before you get out the pickaxe and helmet light. I’m Grady, and this is Practical Engineering.

I think one of the reasons that tunneling is so awesome is that the underground seems like a kind of no man’s land. It’s a different kind of wilderness - unexplored territory in a world where everything already feels explored. But it’s not really true. Land ownership is a tricky subject, but in most places, when you own land, you don’t just own the surface but everything below it as well. There are obvious practical limitations to that; some places separate mineral rights; and there’s plenty of legal nuance too. But in effect, it means that trespassing is still a thing below the ground. Land ownership is 3D. Major tunnel projects, whether for transportation or utilities, are preceded by the acquisition of rights, typically in the form of subsurface easements. In some cases, it can be a pretty nice deal for a landowner: getting paid just for a subway or highway, sewer pipe or fiber optic line that can run deep below your property without you even noticing.

So that’s the first rule of hobby tunneling: only do it where you’re allowed to. There’s an old internet legend about a plumber in Ireland who dug a tunnel from his house to the local pub. It started as a satirical news article that a lot of people believed. But I think the reason it spread is that it taps into a comforting fantasy: that if you go deep enough, the rules stop applying. Unfortunately, they don’t. Even a tunnel that never breaks the surface can still constitute trespassing, and “nobody noticed” isn’t a permit.

Speaking of permits, just like any other part of the built environment, there are often regulations around where and how you can construct a tunnel. I’m talking about building codes. They can feel frustrating to someone who just wants the freedom to build what they want on their own property. The thing is: codes really aren’t there to protect you from yourself. They’re to protect the safety and well-being of everyone else. They’re kind of society’s way of recognizing that the built world is more stable than the people who make use of it. A tunnel is likely to outlast the person who designed and built it, so authorities often want a say in how it's made.

This is a very broad statement, but I think it’s fair to say that we generally enjoy an expectation of safety when we interact with the built environment. The main reason for that is codes. They’re how we bake lessons from past tragedies into the next generation of construction. Codes are often written in blood, as the saying goes. So that’s lesson 2: get a permit. Even if you live in an area with no building codes, if you have a loan, and especially if you have an insurance policy, there’s a good chance that your lender or insurer is going to have something to say about a hobby tunnel. No one likes red tape, but digging deep means the stakes are high enough that some amount of prudence makes sense.

Of course, if you do dig in a place with building codes, those codes probably aren’t going to give you specific design criteria for a tunnel project. Instead, for unusual projects with high consequences of failure, they’ll tell you something pretty simple: you need to hire an engineer. There’s just too much that can go wrong for any building authority to trust an unqualified hobbyist to get the details right. And I want to show you just a few of the things that an engineer is going to consider in the process.

First and foremost is the ground itself. Not all tunnels are created equal because geology varies across the world. And nearly every part of a subsurface project is affected by geology. You don’t choose the design parameters; the ground does.

Take, for example, the excavation process itself. Look at the vast array of mining and tunneling equipment to get a sense of how conditions dictate the methods and tools. Maybe you have soft sandy soil that’s easy to dislodge with a shovel or pickaxe. Firm clays or softer rock might require power tools like a hydraulic arm or hammer drill. Hard and competent rock steps up the challenge, where larger rock milling or grinding equipment or even blasting with explosives is the only way to make progress. The tools you need to excavate depend entirely on what kind of rock or soil you have to work with, but that’s not all it affects.

In general, the ease of excavation is inversely correlated with stability. The more readily soil or rock particles come free from each other when you’re digging, the more likely they are to do it when you don’t want them to. I’ve talked about excavation safety in some previous videos. The gist of it is that soil and rock don’t love tension. Any time you cut a steep or vertical slope, it changes the forces between the particles, whether they’re tiny grains, cobbles, boulders, or slabs. Soil particles are strong against each other, but they don’t hold on so much if they’re pulled apart. Trench collapses are one of the most common causes of construction fatalities. Modern projects go to great lengths and expense to stabilize excavations like trenches and holes. Temporary shoring supports the walls of the excavations so they’re safe to work inside. And what is a trench if not a topless tunnel? But adding that top makes things more complicated.

For one, you have a roof of earth above you. Talk about tension in soil and rock. Trying to use them as a ceiling is basically the most unstable loading condition you can have. For two, we have to talk about the idea of earth pressure. Just like the water pressure goes up as you swim downward, the Earth does a similar thing. The deeper you go, the more stress the materials are under from the weight of the soil and rock above. Of course, earthen materials don’t act exactly like fluids. They can arch and shift load paths around an opening, but it’s extremely material-dependent.

For most types of soil, you basically need support as soon as you excavate, particularly for the roof. Usually, that means using a shield. This is a hollow box or tube that advances with the tunnel, providing temporary support for the walls and roof while leaving the face open for excavation. As you cut and remove the soil, the shield moves forward to support the newly excavated area. These have been used since the early days of tunneling, and even modern tunnel boring machines in soft soils use a shield for temporary support until a permanent lining is installed.

For rocky tunnels, engineers often use the concept of “stand-up time” for gauging safety and the urgency of getting supports in place. This is very empirical. It’s based less on the physics of the situation than on simple observations over a long period of time. The idea is that, if you can measure a few important properties of the rock mass you’re tunneling through (like strength and spacing of joints), you can get a rough idea of how long an unsupported excavation will remain stable. This chart shows that relationship between the roof span, the rock mass rating, and the stand–up time. You can see there’s a zone of immediate collapse where the span is too big or the rock mass is too unstable. And there’s a zone where no support is needed for small spans and competent rock. In between, there’s a time limit for how long you have to install supports, and that limit can vary from hours all the way to years.

Beyond temporary supports used during excavation, most modern tunnels rely on some kind of permanent support. This is important not only to protect the people and stuff inside from a collapse, but also for the stability of anything above. When a tunnel collapses, that movement can translate all the way to the surface, leading to settlements, sinkholes, and damage to buildings and infrastructure above, especially when the tunnels are shallow. I have a whole video on that topic you can check out after this. Major tunneling projects have extensive monitoring plans to check for movements and adjust construction accordingly. That might mean instruments like extensometers and inclinometers, high-precision survey equipment, and vibration sensors. For a hobby tunnel, especially if you’re building below a structure like your house, monitoring for movement is a smart move, even if it’s just a well-placed benchmark and a cheap laser level. Otherwise, your instruments become doors that won’t close and foundation cracks that weren’t there before.

Like temporary supports during construction, permanent tunnel supports vary a lot. For fairly competent rock with only some joints and fractures, where the instability is dominated by discrete blocks and wedges, support can be as simple as rock bolts. These anchors are used to stitch the rock together, and they work surprisingly well. I have a video on that topic, too, where I used model rock bolts to create a table of gravel. For tunnels where the risk of collapse is greater, many use concrete for permanent lining. The big projects that use tunnel boring machines often have an entire system that can take pre-cast concrete segments and assemble them, almost like Lego, on the backside of the machine. Then the annular space between the tunnel walls and lining segments is often pressure grouted so that you get a consistent transfer of ground pressure into the tunnel lining. You can use traditional cast-in-place concrete for lining, too. It’s easy to get good contact with vertical walls because the concrete can be cast right up against them. But, it’s a lot harder to place concrete for a roof section that makes good contact without leaving voids. Instead of that, many tunnels rely on pneumatically-placed concrete, sometimes known as shotcrete or gunite. That gives you the benefit of not needing forms, but much like using a tunnel boring machine, shotcrete does require specialized machinery and concrete mix designs that aren’t super accessible to a hobbyist.

Of course, even if the walls of your tunnel are supported after excavation, you still have the challenge of spoils. This is a little silly to say outloud, but this is one of the most difficult parts of tunneling. When you build something on the surface of the earth, the stuff that was there already (namely, the air) essentially gets out of the way on its own. With a tunnel, you have to do that work yourself.

Do a little mental math exercise with me. Multiply the length of the room you're in right now by its width and its height. Then multiply that number by the average unit weight of soil. If you don’t know it by heart, I’ll put it on-screen now in a few unit systems. Was your answer more than 50 tons? (Either metric or imperial - they’re close enough that it doesn’t matter here.) If you are in anything other than a small closet right now, it definitely was. And I don’t know the last time you moved 50 tons of something, but that is an enormous endeavor on its own, especially because most hobby tunnels don’t have the space or budget for heavy equipment that is normally used for earth-moving projects. And not only do you have to get it out, you also have to get rid of it somehow, unless you happen to have the land to keep a stockpile nearby. At least with mining, the muck often contains ore, which is a valuable resource. For hobby tunnels, and indeed nearly all tunneling projects, the spoils from excavation are essentially a waste product and represent one of the most difficult aspects of the entire process. In many ways, tunneling is a supply chain problem disguised as digging.

Even once you have support, you still have the challenge of water. It doesn’t just flow downhill; it also flows down into hills and any other permeable material it can find. Lots of homeowners with old basements understand this challenge. Providing structural support and keeping water out are two distinct jobs for a basement wall or tunnel lining system to do. It’s not feasible to make the walls 100 percent waterproof, even in underwater tunnels. Concrete cracks. Joints open up. It’s basically inevitable that water will get in, so a good design takes that into account. Modern tunnels are equipped with sophisticated drainage systems that collect water, whether it seeps in from the ground or gets in through the portals. Many tunnels even use a sloped profile so that water can drain out the ends through gravity. If that’s not an option, a collection sump and pump is the other way to manage the water. Just keep in mind that any materials that struggle in moist environments, like wood and unprotected steel, may not last long below the ground.

Speaking of humidity, air is another challenge in tunnel engineering, both during construction and afterwards. In any confined space, you can have higher concentrations of dust and gases that aren’t safe to breathe. Work in spaces like this comes with very specific safety rules, that include ventilation, gas monitoring, and a standby attendant to maintain communications and call for help if it’s needed. Ventilation is important after construction, too. Of course, vehicle tunnels have to deal with exhaust fumes, so their design can get pretty complicated. I have a few diagrams in my book, Engineering In Plain Sight, if you want to learn more. But even in a simple hobby tunnel, fresh airflow is critical. Depending on the layout, it can be pretty tricky to get fresh air IN and stale air OUT of the entire space. Ducting and fans are just one more of the complicated systems to juggle.

Another reason ventilation is so important in tunnels is the potential for fires. Engineers have to consider where smoke will go and how to keep tunnel occupants safe in the event of a fire inside. Hobby tunnels usually don’t have vehicles with combustion engines, but they still carry life safety risks like any habitable structure. So the layout should consider multiple routes for egress and fire suppression. And there are so many of those kinds of details that are, at the very least, worth consideration, even if not absolutely necessary for a small-scale personal project.

I love the idea of hobby tunnels. There’s an aspect of exploration and mystery that you can’t really get anywhere else than underground. I don’t have a tunneling project of my own, but I definitely live vicariously through the ones I see online. And I hope this video doesn’t feel like a wet blanket over any of that. Obviously, the risk profile for an individual hobbyist is going to be a lot different than for a public infrastructure project, so the design, construction methods, and feasibility all look different as well. This is not a how-to video (again, don’t take it as advice), but it’s also not me saying “You can’t do this.” I just think it’s interesting to consider the modern solutions to engineering challenges in large-scale tunneling and how those lessons might apply to intrepid hobbyists.

Do Retention Ponds Actually Work?

[Note that this article is a transcript of the video embedded above.]

This is the Historic Fourth Ward Park in Atlanta, Georgia. It’s got all the stuff you could want a park to have: landscaped walkways, benches, grassy fields, a playground, and even a splashpad and amphitheater. The focal point is the 5-acre or half-a-hectare pond running through the middle. But this pond isn’t just for looks. In fact, this park would never have been built at all except for the fact that it solves a serious flooding problem. For years, the Fourth Ward neighborhood struggled with drainage and flooding issues. In the 90s, the city came up with a plan: a massive underground tunnel to carry runoff away. Don’t get me wrong. I love flood tunnels. I have a whole video about them. But they’re not always the right call. One engineer in Atlanta had a better idea - a solution that would address the flooding issue for a lower cost, and significantly beautify the area, a rare opportunity to improve form and function.

You’ve almost certainly seen a stormwater pond, whether you realized that’s what it was or not. They kind of blend into the urban landscape to the point where they’re basically hiding in plain sight. Some have been turned into amenities in places like parks where the primary purpose is disguised. But I think it’s fair to say that no other single solution has been installed more extensively in modern cities or delivered greater cumulative protection against runoff than the humble stormwater pond. Let me show you how they work with a model I built in the garage and some of the ways these ponds are evolving in the 21st century. I’m Grady, and this is Practical Engineering.

The problem that stormwater ponds solve is pretty easy to understand. Storms bring water, and that stormwater has to go somewhere. Spray a garden hose on some grass and some concrete, and just watch what happens. How much of that water soaks in, and how much runs off the surface? Depending on the type of soil below the grass (and the duration of the experiment), the answers are pretty different for the two situations. Let’s do a little development to make this clearer.

Say we buy up this piece of land on the edge of the city. Add roads and sidewalks; some commercial parcels with parking lots; a park with a gazebo, tennis and basketball courts; apartments and homes with roofs, driveways, patios, and sheds. Before our project, this entire area was natural ground - soil that could absorb at least some amount of precipitation, allowing it to infiltrate into the earth, recharging aquifers. Now, it’s covered in all kinds of impervious surfaces. Let’s see what happens when it rains.

Essentially, two things can happen to rainfall when it hits the ground. It either soaks in or it runs off. How much of each happens depends on quite a few factors. For soil, it matters what kind. Sandy soils with large particles and interstitial spaces can absorb a lot. Clays, with microscopic particles and almost no voids, very little. It also matters how much water is already in the soil. If it’s wet before the storm, there’s less room for more water to flow in. And as soil absorbs water throughout a storm, its ability to infiltrate more decreases. Any water that can’t infiltrate the soil will run off into creeks or rivers nearby. But, for impervious surfaces like asphalt, concrete, and roofs, there aren’t really any variables. Essentially all the water that falls on them runs off. When it rains in our new development, all the runoff still flows to the same place: maybe into a channel that runs to a creek that eventually connects to a larger river. It’s just that now, there’s a lot more of it.

As I mentioned, depending on the type of soil and the size of the storm, the difference between pre- and post-development conditions can be pretty significant. But a single development usually only represents a small portion of the watershed for a creek or river. So, even with all these new impervious surfaces, the marginal increase in water levels during a storm downstream may be fairly insignificant. But zoom out to the scale of an entire city, and the problem becomes obvious. It’s basically all impervious. Development left unchecked can dramatically increase the frequency and severity of flooding because when it rains, a much greater proportion of that rain runs off into creeks and rivers instead of soaking into the ground. So, most cities don’t let development go unchecked, at least from a flooding standpoint.

Rules vary a lot among cities and across the world, but the most basic requirement you’ll see in most places is pretty simple: To get a building permit to develop a piece of property, you’re going to have to limit the peak runoff from the property to pre-development levels. That means that for a given storm, on a given site, you can’t have a higher flow rate after development than it would have been beforehand. Most development is going to involve adding impervious surfaces, whether they’re roads, buildings, sidewalks, or parking lots. And that means more runoff. You can’t just get rid of the water (in most cases), so somehow, you’re going to have to store it and release it gradually to keep the peak flow below pre-development levels. And the simplest way to do it is a pond.

This is my garage-built stormwater pond. It’s just an acrylic flume I use for some of my demos. But I’ve built this outlet structure that should slow down the water, backing it up into the pond.

I’m measuring flow with a meter on the inflow pipe. I also have a level sensor measuring the volume of discharge over time in this tank below. These are both feeding into an Arduino so we can look at the data.

I’m going to simulate a storm event using this valve. So, this is a hand-crafted, artisan inflow event. A typical storm has kind of a bell-shaped runoff curve. Starts slow, builds to a peak as more and more of the watershed contributes, and then tapers off as the storm moves away. And you can see that my stormwater pond captured some of that peak. Because of the outlet structure (that just has a small hole at the bottom right now), the discharge from the pond is much lower than the inflow. And, after a little post-processing, I can show you the data.

This is a plot of flow versus time. Inflow is the solid line. Outflow is the dashed line. The units are arbitrary since this is just for comparison, but I did calibrate the sensors so they match as closely as possible. You can see that the area under both curves is the same. Just as much water came out of the pond as into it. But the peak outflow rate was a lot lower. And that’s a big deal.

The peak of the flood is everything. That’s what determines how high the water rises downstream. It correlates closely with the total amount of damage that occurs. So most drainage rules in cities don’t really focus on total volume; they focus on the peak flow rate leaving the site. And you can see that the peak coming out of my pond is significantly lower than the peak going into it.

So great, the pond did its job. Problem solved right? But you know this wouldn’t be an engineering challenge if there wasn’t something to balance. You can imagine a pond with no outlet at all that just fills up with runoff. In that case, the peak discharge is zero. We’ve maximized the performance, right? Obviously not, since that storage is expensive, not only in the construction cost to build it but also in the valuable real estate it takes up on the site. So really, the optimal solution is the one that uses the least volume necessary, while still keeping the peak discharge below what it would have been without any development at all.

The problem is storms vary in intensity and duration. So most of the time, you’re going to have to show that your design works for several different storm events of varying magnitude. A little hole at the bottom of the outlet structure might work for a small one. However, for a larger storm, you can see that my pond fills up pretty quickly and eventually overflows. Sure, you could make the pond bigger to hold more volume, but we’re just trying to trim the peak off the flow rate to match pre-development conditions. We can release more water from the pond; we just have to be careful about how much.

When I remove this conspicuous piece of tape from my outlet, you can see that I’ve already built this in. I have a larger hole higher up on the structure, so it can release more during more intense storms. Let me simulate that now. You can see as the water reaches that level, the flows from the two holes combine, and we get more water released from the pond, so it doesn’t overflow. Here’s the graph of the small storm again. And here’s the graph for the big one. You can see that in both cases, we’re not completely eliminating the flow. The pond and outlet structure are just shaving off the peaks to reduce the impact of the impervious surfaces. But that can be a tricky thing to do when you have a lot of different storm magnitudes to consider.

Take a look at a stormwater ponds in the wild and you’ll start to notice the wide variety of outlet designs. Placing the various orifices or weirs is kind of an art as much as it is a science, because every site is different and every city has different rules. An engineer has to tune the structure to balance the amount of storage with the additional runoff from all the impervious surfaces. I added a third hole on top of my structure so it can handle a really big storm. The flow through all three holes in the outlet combines to create more flow out of the pond. Here’s the graph of that run. You can see the discharge is much higher, but it’s still below the peak inflow.

But, this gets quite a bit more complicated, because stormwater runoff doesn’t just create flooding. It also carries pollution. We think of rain as cleansing, but the stuff rain washes off the landscape has to go somewhere. That means everything from trash, oil, dog poop, sediment, road salt, and a whole lot more ends up in creeks and waterways. A lot of the contaminants in stormwater are either attached to sediment (or are sediments themselves). So stormwater ponds can serve double duty, reducing flooding and downstream contamination. You’re not going to get the water really clean like at a wastewater plant, but the treatment for suspended solids can be as simple as letting water sit still for a day or two so bits of stuff can settle to the bottom. You may have heard the terms detention pond or retention pond. We’ve been talking about detention ponds that simply slow down runoff, but they eventually empty out. Retention ponds are related, but they keep some of that water stored permanently, and it makes a big difference when it comes to treatment. Let me show you in the model.

I added a bunch more mica powder to the water so you can easily see how the water flows through the pond. Contamination is worst during the beginning of a storm, sometimes called the “first flush,” when streets and surfaces are dirtiest. You can see in my model, before the pond starts filling, everything suspended in the water is making it through to the outlet. The water is moving pretty quickly, and it’s relatively turbulent, there’s just not enough time for anything to settle out. But I can put a plug in the bottom outlet of the structure and prefill the pond so it acts like a retention facility. Now when I turn on the pump to the same flow rate, you can see a big difference. There’s a lot of turbulence where the water flows in, but things slow way down toward the outlet. It’s still just a scale model, so most of the mica powder is still suspended at the end, but you can imagine if we scaled this up so the water took several hours or more before reaching the outlet, most of the solids in the flow would have enough time to settle out.

And retention ponds have other benefits too. They help with groundwater recharge by giving water more time to soak in, and they often look nicer, since water features are an amenity, and these are often landscaped like any other pond you might intentionally install on a site. But, obviously, there’s a tradeoff here. You get cleaner water out, but you need a bigger pond, since some of the volume is already taken up before a storm arrives. However, there is a way to have your pond-cake and eat it, too.

Outlet structures don’t have to be passive like my demo here. Imagine if you could actively control how much water flows out of the pond based on sensors and weather forecasts. You could hold water in the pond for longer periods of time when there isn’t too much rain, improving the quality of the treatment, and then pre-drain the pond ahead of a storm, freeing up that space for the next runoff event. This is known as Continuous Monitoring and Adaptive Control - it’s basically “smart” stormwater management. It’s a pretty cool idea that’s only just starting to catch on in cities, but it has disadvantages too. One is disease vector control. Because there’s no stable pool, you can’t reliably stock fish to eat mosquito larvae, so there are limits on how long you can hold water before you have to drain the pond. It’s also quite a bit more technically sophisticated, so there’s a tradeoff there too. Usually, these types of systems are operated by specialized companies that install, manage, and maintain them. Some even sell the capacity on an open marketplace, allowing developers to buy credits in lieu of on-site ponds. This stuff gets pretty creative - addressing the lot-level needs of individual developments with larger, watershed-scale outcomes. And in fact, they’re often part of a larger idea called regional detention.

Even though on-site detention or retention is great in theory, it can be messy in practice: small lots don’t have room for meaningful storage, building dozens of tiny basins inevitably leads to uneven maintenance, the small pipes and outlets of minor ponds are more susceptible to clogging, and in some cases, they can actually make flooding worse. You could see on my graphs that detention lowers the peak at each site, but it also delays it. If many basins are designed with similar outlet controls, their attenuated peaks can arrive all at once at a confluence downstream, spiking the creek level worse than if there were no detention at all.

Water quality benefits are hit-or-miss, too, because performance depends on how each little system is built and maintained. So, there are cases where developers get together or a city or drainage district solves the problem at a regional scale, building a single, larger facility that can handle the runoff from multiple sites. By routing excess runoff to a shared basin (or a network of them), you gain real storage volume, coordinated release rates that match downstream capacity, and professional, centralized upkeep. It also lets you optimize water-quality treatment and pipe sizes across the area instead of overbuilding each parcel. Keep the small storms where they fall for infiltration and local benefits; send the larger pulses to regional detention so the watershed sees a calm, controlled hydrograph instead of a patchwork of ponds releasing a chorus of overlapping peaks.

I should make clear that detention and retention are far from the only stormwater management tools. Regional geology and hydrology often drive the design. I live near Austin, which has strict environmental rules because of the Edwards Aquifer. Where the limestone reaches the surface, contaminated runoff can easily enter the groundwater. So many sites in Austin require filtration ponds that actually pass water through a layer of sand before it’s discharged downstream, removing pollutants before they can reach the groundwater. I’ve talked about permeable pavement in a previous video, and there are a lot more solutions out there. Many civil engineers spend their entire careers solving urban stormwater puzzles, trying to balance the important watershed functions with the challenge of flooding and pollution. Detention and retention ponds are just one piece of it. Part park, part plumbing, mostly hiding in plain sight, they are often carefully tuned pieces of infrastructure that help keep the city’s head above water.

The Hidden Engineering of Runways

[Note that this article is a transcript of the video embedded above.]

September 2025 was an unusually bad month for runway overruns in the US. On the night of September 24th, an Embraer 145 with 53 people on board landed long at the Roanoke-Blacksburg Regional Airport in Virginia, overshooting the end of the runway. Just weeks earlier, on September 3rd, TWO similar incidents occurred on the SAME DAY, one a Gulfstream at Chicago Executive Airport and another a Bombardier at Boca Raton. In all three cases, the surface at the very end of the runway crushed under the weight of the planes’ tires. You look at the photos, and it looks like a mess, but these systems worked exactly as they were intended, preventing fatalities and serious injuries in all three cases.

We’ve all seen a runway before. At first glance, there’s not much to it: a strip of concrete or tarmac planted on the landscape with some extra markings and lights. It basically looks like a short section of highway. But if you look under the surface, there is a tremendous amount of engineering that makes these facilities entirely unique from anything else we build. I want to peel back the layers and show you what really goes into building a runway. I’m Grady, and this is Practical Engineering.

A fully loaded semi truck usually weighs on the order of 80,000 pounds (or 36 metric tonnes) and, depending on what state you’re in, legally maxes out at 60 to 80 miles per hour. Our highways are carefully engineered for vehicles in that weight and speed regime. Compare that to modern heavy jets that can weigh more than 500 tonnes or a million pounds, with takeoff and landing speeds around 180 miles per hour. Just like highways, the design decisions for runways - from length, to width, to shape, to materials and beyond - all have major implications on public safety. There is a long list of crashes and incidents that could have been avoided by better designs, and actually, a lot of the reasons we do things the way we do is because of lessons learned through previous tragedies. Maybe better than any other industry, the aviation world strives for continuous improvement through the understanding of past failures, and you can see evidence of that just about everywhere you look, including resources like SKYbrary.

The thing is, building a runway is an extremely costly endeavor. There’s practically no limit to the amount of money you can spend making one incrementally safer. So there’s always a balancing act between cost and capability. One of the most fundamental decisions that affects both sides is length. A longer runway can accommodate larger aircraft, but it can dramatically increase costs by requiring more land and more infrastructure. It can even affect the siting decisions, pushing an airport farther outside a city. It’s a pretty important choice. So important that FAA has a 40-page guidance document on length alone. Based on what you want to accomplish - whether it’s basic general aviation at a municipal field, air cargo, medevac, or serving as a backup to the Space Shuttle program - you first have to pick a critical aircraft: the one that requires the longest runway. But it’s more complicated than that, since takeoff and landing performance depends on a lot of factors. High temperatures and elevation reduce the density of the air, requiring more speed for the same amount of lift, which results in longer takeoff distances and landing rollouts. Slopes affect both takeoff and landing as well. Uphill takeoffs are harder because the engines have to fight gravity; downhill landings require stronger braking. The FAA says that for each percent of downhill slope, landing distance is increased by 10%. Manufacturers of aircraft can tell you the runway requirements for a specific make and model, or FAA has developed curves that can help you take these factors into account to decide a runway length.

When you’re driving on the highway, direction isn’t that important. Obviously, you have to get to where you’re going, but other than that, there aren’t many engineering requirements that change with the direction of the roadway. With runways, that’s not true. Whether taking off or landing, airplanes work best when facing directly into the wind. And in fact, they might not be able to land or take off at all under certain crosswind conditions. So the direction of a runway is a consequential decision. Prevailing winds vary a lot by location. In fact, one of my favorite types of diagrams, the wind rose, is specifically designed to show this at a glance. And if you look at enough wind roses, you’ll notice that, in some places, there’s not a prevailing wind direction at all. That’s why most large airports have perpendicular runways. Again, this is aircraft-dependent. Every airplane has its own crosswind limits. FAA generally expects runway orientation to provide about 95% wind coverage for the airport’s design aircraft, so in places without a strong prevailing wind direction, it takes a second runway to meet that target.

Length and direction are easy to notice, but there’s more to the geometry of a runway. In 2019, a Miami Air International Boeing 737 touched down in Jacksonville during heavy rain. The aircraft skidded off the runway and came to a stop in the St Johns River. 21 people were injured, but thankfully, nobody was killed. When the NTSB investigated the accident, one of the main contributors was that the runway was ungrooved. The water on the surface couldn’t squeeze out fast enough, instead building pressure in the contact patch between the tire and runway. It’s hydroplaning: the tires ride on the water instead of the ground, wiping out friction and directional control.

Just like in a car, planes need friction to stop. Larger jets have the benefit of aerobraking, using devices that reverse the thrust of the engines, but regular-old wheel brakes still do most of the work. And just like for cars, water makes that much more challenging, so there are a lot of engineering decisions that go into maintaining good friction on the runway surface. Like highways, most runways have a gentle crown at the centerline that drops off to the sides. This cross-slope helps shed rain and stops water from pooling on the surface. Larger airports install grooves in the runway surface that give water an escape path from beneath the tires, reducing the chance of hydroplaning in bad weather. And this isn’t just a one-time decision. Airports use friction-measurement equipment to monitor operational conditions. If the surface gets too polished from use or built-up rubber from the countless touchdowns, they have to clean the surface or even retexture with shot blasting to roughen it up.

Runways are a bit unusual because, when you think about it, they really have two very different jobs. Taking off and landing are pretty similar; one is essentially the reverse of the other. But in some ways, they’re entirely different. And so they drive the requirements for runway engineering in different ways. For example, it may feel like landing is the most dynamic moment in a flight, but it’s actually takeoff that usually governs runway length and strength. That’s mostly because of weight. A big part of the weight of a fully loaded airliner is fuel. An Airbus A380, the largest of commercial jets, has a max gross takeoff weight of over 550 metric tonnes. For a long-haul flight, nearly half of that weight can be in fuel. When an airplane touches down, even though the moment the wheels hit may feel impactful, the plane is much lighter. In fact, landings are so much less damaging to pavement than takeoffs that they usually don’t even count in load cycle tracking for the engineering design. It’s all about takeoffs, and to support those enormous loads, airport runways have some of the most heavily engineered pavement systems in the world.

This is something that you’ll almost never be able to see, but the amount of consideration and engineering below the surface is incredible. The FAA even has its own engineering software package, complete with a wonderful government acronym: the FAA Rigid and Flexible Iterative Elastic Layered Design or FAARFIELD. Just like highways, you basically have two choices for runway pavement materials. Rigid pavements generally use concrete. Flexible pavements use hot-mix asphalt. Their behavior and performance are pretty different, so the engineering is different too. Asphalt has a small but significant measure of give to it, which causes the effective width of aircraft tires to spread out in a cone underneath the surface into the deeper layers. This contrasts with rigid pavement, where a tire's effective width is its actual width.

Asphalt is a cheaper material, so it's used in the vast majority of paved airfields in the US. Concrete is stronger and stiffer, so most large-scale commercial airports use rigid surfaces. The tradeoff usually comes with volume. A rigid pavement has a longer design life, so the additional cost is offset by reduced maintenance and a longer interval before replacement. But in both cases, there’s a lot under the surface. It’s basically a layer cake of materials that all serve different functions.

Everything sits on the subgrade, which is the natural soil at the site. The quality of the subgrade really decides everything else. The soil strength, its potential for shrinkage and swelling, the depth of the frost line, and the depth of the water table will drive the design. If it’s really soft and mushy, the subgrade can be amended with sand, lime, cement, or geosynthetic materials.

Some pavements put a drainage layer on top of the subgrade. This is a permeable material, like gravel, that lets water get out of the system so it doesn’t soak the soils below, which might lead to softening and weakening over time. A runway is one place you don’t want a pothole.

Above that, many pavement systems (especially flexible ones) use a subbase. This is a layer of course material (sometimes even crushed up bits of an OLD runway). Practically, the sub-base adds thickness cheaply. Stress from wheel loads drops quickly with depth, so a layer of material that doesn’t have tight engineering specifications can accomplish the depth without driving up the cost too much. Plus, the subbase serves as a working platform so you’re not mucking up the subgrade with heavy equipment during construction.

Then comes the base course. This is the structural workhorse of a pavement system. It’s usually a mixture of high-quality crushed and uncrushed aggregates, specifically designed to lock together when compacted into a high-strength support. The goal is to distribute the point forces of wheel loads into the layers below. Lower stress mean less movement, which results in less cracking of the surface layer and a smoother ride over time.

On top of all that is the surface course that provides the friction and texture. Concrete pavements distribute forces, so they don’t require quite as much engineering underneath. For asphalt, friction is essentially its only purpose. The layers below do the heavy lifting. And if the surface course degrades, you can often mill it and overlay it with new material without having to rebuild the entire system below.

Separating the pavement into all these layers is about finding the right balance between performance, cost, and constructability. You could just build a 10-foot-thick layer of concrete and be done with it, but eventually those costs flow to the airline tickets, and no one would be happy to pay for that!

Since runways are essentially a connection to the sky, there are some quirks in their engineering to account for that too. One is the use of displaced thresholds. Sometimes, surrounding obstacles don’t allow for a gentle glide slope to the end of a runway. You don’t want airplanes diving steeply into a landing, so instead, we displace the touchdown point farther down the runway, while still allowing takeoffs to use the full length. Takeoff lengths are usually longer than landing lengths anyway, so this is a compromise worth making to take the best advantage of the surrounding airspace.

You can only displace a threshold so much, though. Sometimes design choices and sacrifices are made to accommodate unavoidable restrictions caused by nearby terrain or buildings. Airports have to exist in the broader context of developed areas. So, airport designers and managers have to ensure that imaginary zones called “obstruction surfaces” are free of buildings, trees, towers, and anything else you don’t want to get hit by a plane. These imaginary surfaces extend farther than you might think into the air space, providing safe approach and departure paths with comfortable margins of safety. Airports don’t usually have land-use authority, though, so keeping the airspace free from obstructions is a collaborative, and occasionally contentious, process between regulators, cities, landowners, and developers.

There are also areas of pavement at the ends of runways that aren’t intended to have planes on them at all. For example, larger runways include blast pads. This is one of my favorite elements of runway engineering. The powerful wakes produced by jet engines pick up grit and scour away the land behind them. If this is just loose soil or grass, the endless parade of planes will eventually dig a huge hole at the back of the runway! I’ve spent a lot of time working with concrete structures meant to curb erosion from flowing water, but there just aren’t that many pieces of infrastructure that are purpose-built to mitigate aerodynamic erosion. Blast pads can’t carry the weight of a jetliner, so they’re painted with yellow chevrons to tell pilots ‘stay off!’

Even when a runway is long enough to accommodate the air traffic it sees on a regular basis, accidents happen, and sometimes airplanes overshoot the end of the runway on takeoff or landing. Runways are required to have a certain amount of space beyond the pavement on all sides, called runway safety areas or RSAs. Like the clear zones along highways, RSAs provide an airplane with room to safely come to a stop without obstacles. There are some instances where space is tight, though. Urban infrastructure, a body of water, or other stuff can get in the way, making it less feasible to maintain so much open space around a runway. Luckily, there’s another option: Engineered Materials Arresting Systems, or EMAS.

These systems are manufactured from crushable material like lightweight concrete or foamed glass. In an emergency situation, they can dissipate a plane’s kinetic energy, quickly slowing it down so it doesn’t crash into whatever lies beyond. EMAS saved the day in all three major overrun incidents in September 2025. You can see just how effective it is in this footage from the September incident in Boca Raton. It’s like a much more sophisticated and carefully engineered runaway truck ramp for airplanes.

There’s so much more going on in the engineering and design of runways than I can possibly cover in one video. I’ve tried to focus on the hidden stuff: construction techniques and requirements that you don’t really notice when you’re a passenger looking through the window and may not even be familiar with as a pilot. I really love knowing how much goes into that stuff that most of us never have to think about. It makes me feel safer as a passenger. It’s a reminder that smooth and boring is usually the goal, and it takes a lot of work to keep it that way.