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

In 2015, an unusual incident happened on the construction site for a sewage lift station in British Columbia, Canada. WorksafeBC, the provincial health and safety agency, posted a summary of the event on YouTube. A steel caisson had been installed to hold back soil while the lift station could be constructed. One worker on the site was suddenly pulled into a sinkhole when the bottom of the caisson blew out. The cause of the incident was related to groundwater within the soils below the site. We don’t all have to live in fear of the ground opening up below our feet, but engineers who design subsurface structures do have to consider the impact that groundwater can have. The solutions to subsurface problems are almost always hidden from public view, so you might never even know they’re there. This video is intended to shed some light on those invisible solutions (including what could have been done to prevent that incident in BC). I’m Grady and this is Practical Engineering. In today's episode, we’re talking about how groundwater affects structures.

Groundwater has always been a little mysterious to humanity since it can’t easily be observed. It also behaves much differently than surface waters like rivers and oceans, sometimes defying expectations, as I’ve shown in a few of my previous videos. One of the most important places where groundwater shows up in civil engineering is at a dam. That’s because groundwater flows from high pressure to low pressure, and a dam, at its simplest, is just a structure that divides those two conditions. And what do you know, I’ve got an acrylic box in my garage full of sand to show these concepts in real life.

You can imagine this soil sits below the base of a dam, and I can adjust the water levels on either side of the structure to simulate how groundwater will flow. Blue dye placed in the sand helps show the direction and speed of water movement below the surface. A higher level on the upstream side creates pressure, driving water in the subsurface below the dam to the opposite end of the model. I’ll be the first to say it: this is not the most mind-blowing revelation. You probably could have predicted it without the fancy model. But to a civil engineer, this is not an inconsequential phenomenon, and for a couple of reasons.

First, water seeping below a dam can erode soil particles away, a phenomenon called piping. Obviously, you don’t want part of your structure’s foundation to be stolen from underneath it, and piping can create a positive feedback loop where failure progresses rapidly. I have a whole video on piping that you can check out after this one. The second negative effect of groundwater is less obvious. In fact, until around the 1920s, dam engineers didn’t even take it into account (leading to the demise of many early structures in history).

The engineering of a dam is largely an exercise in resisting hydrostatic pressure. Water in the reservoir applies an enormous force to the upstream face of a dam, and if not designed properly, that force can cause the dam to slide downstream or overturn. The hydrostatic force is actually pretty simple to approximate. Pressure in a fluid increases with depth, so you get a triangular distributed load. Once you know that load, you can design a structure to resist it, and there are a lot of ways to do that. One of the most common types of dam just uses its own weight for stability. Gravity dams are designed to be heavy enough that hydrostatic forces can’t slide them backwards or turn them over. But, to the dismay of those early engineers, pressure from the reservoir is not the only destabilizing force on a dam.

Take a look at this pipe I’ve included in the model that shows the water level between the two boundaries. If the base of a structure was below the water level shown here, the groundwater would be applying pressure to the bottom, counteracting its weight. We call this uplift pressure. Remember that the only reason gravity dams stay put is because of their weight, so you can see how having an unanticipated force effectively subtracting some of that weight would be a bad thing. Many concrete gravity dams have failed because this uplift force was neglected by engineers, including the St. Francis Dam in California that killed more than 400 people when it collapsed in 1928. Many consider this to be the worst American civil engineering disaster of the 20th century.

Unlike the hydrostatic force of a reservoir, uplift pressure from groundwater is a much more complicated force to characterize. It exists in the interface between the structure and its foundation, in the cracks and pores of the underlying soil, and even within the joints of the concrete structure itself. The flow of groundwater is affected by soil properties, the geometry of the dam, the water levels upstream and downstream, and even the subsurface features. How these factors affect the uplift pressure can be pretty challenging to predict. But engineers do have to predict it. After all, we can’t build a dam, measure the actual uplift force, and add weight if necessary. It’s gotta work the first time.

One way to characterize groundwater flow around structures is the flow net. This is a graphical tool used by engineers to estimate the volume and pressure of seepage in the subsurface. In simple terms, you divide the flow area into a curvilinear grid, where one axis represents pressure and the other represents flow. If this looks familiar, you might notice that a flow net is essentially a 2D solution to the Laplace equation, which also applies to other areas of physics including heat flow and magnetic fields. Developing flow nets is almost an art as much as a science, so it’s probably a good thing that groundwater problems are mostly solved using software these days. But, we can still use flow nets to demonstrate a few of the ways engineers combat this nefarious uplift force on gravity dams. And one common idea is a cutoff wall.

If water flowing below a dam causes so many problems, why not just create a vertical wall to cut it off? We do it all the time. But, how deep does it need to be? Some dams might have a convenient geological layer into which a cutoff can be terminated, creating an impenetrable envelope to keep seepage out. But, many don’t. Cutoff walls can still reduce the volume of flow and the pressure, even if seepage can still make its way underneath. Let’s take a look at the model to see why. I’ve added a vertical wall of acrylic below the upstream face of my dam, and we’ll see how it affects the flow. [Beat]. The groundwater flow lines adjust to go under the wall and back up to the other side of the model. If you look closely you’ll see a slight decrease in the uplift measurement pipe below the dam. The only thing I changed between this model and the last one was adding the cutoff wall. So why would the pressure decrease on the downstream side?

The flow of groundwater is described with a fairly simple formula known as Darcy’s law. Besides the permeability of the soil, the only other factor controlling the speed water flows is the hydraulic gradient, which consists of the difference in pressure over the length of a flow path. By adding a cutoff wall, I didn’t change the difference in pressure between one side of the model and the other, but I did increase the length of the flow path water had to take below the dam, reducing the hydraulic gradient. I can sketch a flow net over the model to make this clearer. The black lines are equipotentials; they connect areas of equal pressure. The blue lines show the directions of flow. Without a cutoff, the flow paths are shorter, and thus the equipotential lines are closer together. With the cutoff wall, the equipotential lines are spread out. That means both the volume of seepage and the uplift pressure at the base of the structure have been reduced.

Cutoff walls on dams have a long history of use, and nearly all large gravity dams have at least some kind of cutoff. It can be as simple as excavating a wide area of the dam’s foundation before starting on construction, and that’s a popular choice because it gives engineers a chance to observe the subsurface conditions and make sure there are no faults or problems before the dam gets built. Another option is to excavate a deep trench and fill it with grout, concrete, or a slurry of impermeable clay. For smaller or temporary structures, sheet piles can be driven into the subsurface to create a cutoff. One final option is to inject high pressure grout to create an impenetrable curtain below the dam.

The other way to deal with seepage and uplift pressure are drains. Drains installed below a dam do two important jobs. First, they filter seepage using sand and gravel so that soil particles can’t be piped out from the foundation. Second, they relieve uplift pressure by removing the water. Let’s see how this works in my model. Upstream of my uplift monitor, I’ve added a hole through the back of the model with a tube to drain seepage out. Instead of flowing all the way downstream, now some of the seepage flows up to and through the drain, and you can see this in the streamlines of dye flowing in the subsurface. Again, the effect is subtle, but the uplift pressure monitor is showing a slight decrease in pressure compared to the original configuration. There is less pressure on the base of the dam than there would be without the drain. Plotting a flow net over the model, you can see why it behaves this way. The drain relieves the uplift on the base by creating an area of low pressure below the dam. You can also note that the drain actually increases the hydraulic gradient by shortening the flow paths, so there’s actually more seepage happening than there would be without the drain. However, because the drains are installed with filters to reduce the chance of piping, that additional seepage is often worth the decrease in uplift pressure.

Many concrete dams include a row of vertical drains into the foundation, and some even use pumps to depress the groundwater level further, minimizing the uplift. I can simulate this by lowering the downstream level as if a pump was removing the water. Watch how the flow lines adjust when I make this change in the model. Like drains, these relief wells create more seepage below a dam because of the greater difference in pressure between the two sides, but they can significantly reduce the uplift pressure and thus increase a structure’s stability.

I’ve been using dams as the main example of managing groundwater flow, but lots of other structures have similar issues. Retaining walls and temporary shoring have to contend with groundwater challenges, including caissons, which are watertight chambers sunk into the earth to hold back soil during construction. Remember the worker I mentioned in the intro? He was on a site near a caisson. It’s typical to dewater a structure like this, meaning the water is pumped out, creating a dry area for construction crews to work. Let’s take a look at how this works in the model. I’m simulating the act of pumping water out of the caisson by draining out of the model at the bottom of the structure. When a caisson is dewatered, it is essentially working like a dam, separating an area of high pressure from low pressure within only a short distance between them. And, as you know, distance matters when it comes to groundwater, because the shorter the flow paths, the greater the hydraulic gradient, and thus the higher the volume and velocity of seepage.

If you look closely, you can see the sand boiling up as the seepage exits the soil into the bottom of the caisson. This elevated pressure in the subsurface and high velocity of flow means that the soil particles themselves aren’t being strongly held together. All it takes is a little agitation for the soil to liquefy and flow into the bottom of the caisson, creating a sinkhole that can easily swallow anything at the surface. One way of mitigating this hazard is dewatering the soil outside the caisson. Construction crews use well points, small evenly spaced wells and pumps, to draw water out of the soil so it can’t seep to areas of lower pressure. Caissons can also be driven deeper into the subsurface, creating a condition similar to a cutoff wall on a dam. They can even go deep enough to reach an impermeable layer, creating a better seal that prevents water from flowing in through the bottom.

Thankfully for the worker in BC, his colleagues were able to rescue him before he was consumed by the earth. Next time you see a dam, retaining wall, caisson, or any other subsurface construction, there’s a good chance that engineers have had to consider how groundwater will affect the stability. Even though you’d never know they’re there, some combination of drains and cutoffs were probably installed to keep the structure (and the people around it) safe and sound.