Ep 26 Folsom Dam Gate Failure

Engineering News – Robotic Exoskeleton Therapy for People with MS (2:15).


This week`s episode is about the Folsom Dam Gate Failure (4:50). On July 17th, 1995, while gate 3 was opening (7:54), it’s structure failed and the gate swung open, allowing a rush of water down the river. The main cause of the failure? Rust. (9:33)

Sources:

Engineering News

https://www.sciencedaily.com/releases/2021/05/210528133442.htm

Folsom Dam Gate Failure

https://www.researchgate.net/publication/333173706_LESSONS_LEARNED_FROM_THE_SPILLWAY_GATE_FAILURE_AT_FOLSOM_DAM

https://www.usbr.gov/ssle/damsafety/risk/BestPractices/Chapters/G1-FailureOfRadial-Tainter-GatesUnderNormalOperationalConditions.pdf

https://damfailures.org/case-study/folsom-dam-california-1995/

https://en.wikipedia.org/wiki/Folsom_Dam


Transcript

Hi and welcome to Failurology; a podcast about engineering failures. I’m your host, Nicole, and I’m from Calgary, Alberta. I attended a virtual podcast festival at the end of February this year. One of the sessions was on science communication. I didn’t quite know what that meant at the time, but I wanted to take in the entire festival and attended every session. Wikipedia says that science communication is the practice of informing, educating, raising awareness of science-related topics and increasing the sense of wonder about scientific discoveries and arguments. Turns out, I’ve been communicating science every day for over a decade, I just never had a name for it. In April I attended a prestigious two and a half day virtual science communication workshop hosted by the Best of Banff and Telus Spark Science Centre. And most recently I’ve been volunteering with the Science Writers and Communicators of Canada on the social media team for their June conference on the topic of Resilience. This has been a life changing experience. I’ve learned so much about communication plans, learned how to make session visuals for social media, I got a behind the scenes look at conference planning, and I even moderated a panel about resilient infrastructure. Which might be one of my coolest experiences of 2021, so far. And on top of that the Sci Comm community is amazing. Everyone comes from different backgrounds and specialties, but they all share a love of science.

All of that wonderfulness aside, between the conference, and the podcasting, and everything else going on in my life, I could use a nap or two. Luckily I have some much needed R&R scheduled for July which I’m really looking forward to. And I hear restrictions are going to be lightening up here in Alberta. Assuming its safe and our cases don’t skyrocket again, it will be nice to start working towards a new normal. Whatever that looks like.


This week’s engineering news is about exoskeleton therapy for people with multiple sclerosis. A group of MS experts at the Kessler Foundation in New Jersey did a randomized pilot study using robotic-exoskeleton assisted exercise rehabilitation (REAER). Say that 5x fast. They tested the effects of the robotic exoskeleton on mobility, cognition and brain connectivity as an effective intervention for people living with MS.

Multiple sclerosis causes deterioration of the insulating layer of the brain and spinal cord nerve cells, disrupting signals from the brain and causing physical and mental problems. We often take for granted that when our brain tells our leg to move, it moves; we don’t even really put much thought into it. But that is not the case for everyone. And unfortunately for people living with MS, there are not many therapies that currently exist to help manage those symptoms.

This four week study was small, including only 10 people; and all participants had significant MS symptoms. The participants wore a robotic exoskeletons that allowed them to walk further distances than they otherwise would. Distances that allowed them to receive the improved cognitive and mobility effects of exercise. From what I’ve gathered, for example, a person may need to walk 1km per day to see positive effects from exercise, but maybe they can only walk 10% of that. So they aren’t able to benefit from the effects of walking. But the robotic exoskeleton can help. It allows the user to walk longer distances and reap those benefits. Plus, you get to be part robot, even if temporarily, which is kind of like a superpower right?!? Pretty cool if you ask me.

Now, this is well outside my area of expertise. I like to joke that biology isn’t my strong science. Physics and chemistry are where I thrive. But I've heard about people with paralysis being able to repath the communication from their brain to their paralyzed limbs. Maybe not enough to walk again yet, but the ability to repath the signal is possible. I see the robotic exoskeleton as a way to repath the signals as well. I just think helping people walk again is really cool. Not all heroes wear capes.

If you want to read more on the robotic-exoskeleton assisted exercise rehabilitation (REAER) trial, check out the link in the show notes or head to failurology.ca.


Now on to this week’s engineering failure; the Folsom Dam spillway gate. I thought this was a really interesting failure. The majority of the water flows through the power generation section of the dam, but when it’s down from maintenance or the water levels are too high; the dam has several gates that can open or close to maintain the water level in the lake. And it operated just fine for 40 years. And then one day, while opening one of the gates, the gate buckled and just swung open like a door, allowing a rush of water down the river and draining almost 40% of the lake.

Like most failures, this didn’t just happen overnight and it could have been prevented. As much as this story is about the original design, it is also about the maintenance procedures that fell short. Even when engineers design the best and most robust infrastructure, if its not maintained properly or those maintaining it don’t understand how it works, it can still fail short of its expected life. Maintenance is a really important part of engineering. If a design is too expensive or complicated to maintain, guess what? It doesn't get maintained. If a design really complicated, it might not even get installed properly, but that's another story for another day.

Folsom dam is a concrete dam located at the SW corner of Folsom Lake in Northern California; 40km NE of Sacramento. The lake is fed from the north & south forks of the American River and it holds 1.2 billion cubic meters of water. The dam is 100m high and 430m long, flanked on both sides by earthen dams. Which sounds fancy, but are really just dirt hills acting as the banks of the lake. Construction started in 1951 and was completed in 1955, for a cost of $81.5 million USD, which is equivalent to about $812 million today.

Folsom dam was designed by the United States Army Corps of Engineers, the same folks who built the levees in New Orleans that I talked about in episode 2. Once Folsom Dam was complete, ownership and maintenance responsibility were transferred to the United States Bureau of Reclamation. The dam provides flood control, hydroelectricity, irrigation and municipal water supply to the region. In 2017 the dam underwent construction to increase flood protection in the area by adding an auxiliary spillway to prevent overtopping, which in most cases is extremely destructive to dams.

Folsom dam has 8 spillway gates, each 42ft wide and roughly 50ft high to release water from the reservoir either when the power generation equipment is down for service and/or when the water levels get too high. The five NW gates are used for normal operation and the three SE gates are used for emergencies. There's also a second hydroelectric dam, called Nimbus Dam, which is located 7km downstream. Remember, Nimbus, it's important later.

On July 17th, 1995 around 8am, the power plant underwent a scheduled shut down and water was diverted through the spillway gates to maintain lake levels and flow in the river downstream. Folsom lake was at full capacity at the time. While gate 3 was opening, a diagonal brace between the lowest and second lowest struts failed and the gate swung open to the left. This caused an uncontrolled release of about 40% of the lake causing the level to drop by about 11m. A 1,110 cubic meter per second flood of water was sent down the river; this equates to about 7000 bathtubs per second. Stoplogs were eventually used to stop the water flow so the gate could be repaired. Stoplogs are beams that are placed on top of each other in pre-made slots within the gate opening.

This was pretty bad, but there were a couple positives. One, the river downstream of the dam was rated for a capacity of 3250 m3/s, which is about three times more than the flow through failed gate 3, so there was minimal risk of downstream flooding. And the dam operator who was opening the gate that failed jumped in his truck and ran down to the Nimbus Dam downstream and opened its gates to prevent overtopping. One negative, well aside from the gate failure itself, was that the volume of freshwater reaching the San Francisco Bay was unusual that early in the season. This caused pacific salmon and striped bass to migrate months ahead of schedule. Poor fishies.

So why did the gate fail? The short answer is rust, but its a bit more complicated than that. To understand why the rust was such a significant factor, we have to talk about how the gates are built. The Folsom Dam gates, which are called radial gates, are made up of curved steel plates called skin plates which face the lake side. When the gate is closed, this surface sees the majority of the hydraulic loading. Supporting those plates, there are 4 horizontal struts which run from the skin plate to the pivot point of the gate. As the gate opens, it pivots on the trunnion pin, which is a fixed point, and the skin plate lifts and water flows underneath. The horizontal struts are numbered 1 through 4, with 1 being at the top and 4 being at the bottom. There are vertical braces in between each strut, these are numbered D, E, G and J starting from the pivot point and working back towards the skin plate. And there are also diagonal braces between the struts to provide additional support. The initial failure is believed to have occurred at connection point 3-E. With the second failure being at connection point 4-E. And the third failure being at connection point 4-G. I will put a diagram on failurology.ca so you can see how all of the structural members and connections are labelled. But for right now, just know that the failure occurred at connections on the bottom two struts, roughly mid way between the pivot pin and the skin plate.

The failures of the gate connections were the result of excessive friction at the pivot assembly which caused additional forces to be being placed on the structural members, exceeding the capacity of those members and connections until they failed. It’s like when you’re trying to loosen a bolt, a rusty one is much, much hard than a non-rusty one. A really important piece of this story to note is that gate 3 hadn’t been opened more than 600mm with a head no greater than 12m for decades. Head is the measurement of force applied to an element based on the height of water exerted on it. For example, if you had a 1m tall tube filled with water, the forces exerted on the bottom of the tube would be 1m of head. On the day of the failure, gate 3 was scheduled to be opened to 900mm with an applied head of 13m; more movement than the gate had seen in quite some time. When the gate was at 730mm open, the operator felt an “unusual vibration” and he stopped the motor. Within 5 seconds the gate failed and swung open like a door. This operator was the one who drove to Nimbus and opened their gates to prevent overtopping.

As I mentioned, the skin plate is on the lake side of the gate and sees the majority of the water contact. But the rest of the gate is still subject to rainwater, water spray and water vapour. Due to a reduced frequency of lubrication, lack of weather protection, and gaps in the pivot assembly, rust on the pivot assembly lead to increased friction and increased forces on the gate. Rust on the gate struts and other members and connections also weakened their capacity.

As was normal design standard at the time, the gate was not designed for any friction loads on the pivot assembly. They didn’t plan for this to rust. This has since changed. The United States Army Corps Engineers began accounting for rust and pivot assembly friction in 1966. Back when these gates were designed though, computer models were not available to calculate the loads placed on cross braces and smaller members; therefore the arrangement, size and connection details were based on the experience and judgement of the designer. The weight of the gate struts, vertical braces and diagonal members, essentially the gate structure, made up only 18.5% of the total weight of the gate. If the cross sectional area of all of these members had been increased by 15%, it would have only increased the weight of the gate by 2.7% and increased the cost by only 1%. And that is with a blanket oversize of 15% in area. If the designers were more selective about which members they increased the cross sectional area of, the overall impact could have been even less. Oversizing the gate structure would have greatly increased the safety factor with little impact to the project. They didn’t do this here, but it's interesting to think about. It’s important to weigh the impact of overdesign, with the increased safety factor. If there is very little impact to cost, design or schedules, and it could greatly improve the integrity of the design, it can be worth it to overdesign.

The pins in the pivot assembly used at Folsom dam were hollow pins made of carbon steel with an outside diameter of 800mm and an inside diameter of 600mm. While hollow pins themselves are common, it is important to note that hollow pins are typically larger in diameter and therefore have a greater surface area for rust. More rust means more friction and more risk of failure. As I mentioned earlier, the gates were not designed for rust or friction at the pivot assembly. Tests following the failure showed the coefficient of friction on the gate could have been anywhere from 0.22-0.28 when it failed. Friction coefficient is a value between 0 and 1; 0 being no friction at all and 1 being friction equal to normal force. An example of normal force would be the weight of the gate applied to the pin. As the friction forces were applied to the gate, the force on strut 3 was increased by 20% and the force on strut 4by 25%. This is significant. As members failed, the redistribution of forces was also significant.

The clearance between the pin and it’s housing was between 0.076 and 0.229mm when they were brand new. The rust was as much as 0.152mm thick and over time had filled this clearance space between the pin and its housing. Resulting in a higher coefficient of friction on the pin, housing and gate.

Since the gate had been opened to 600mm previously, it is a reasonable assumption, that had the gate been opened to no more than 600mm on July 17, 1995, the failure would not have occurred because regular use would have limited the amount of rust on that particular section of the pin. But since the gate opened beyond its normal use, the contact surfaces in the pin and assembly would have seen more rust and therefore more friction as it moved beyond regular use contact. All of that said, failure was pretty much inevitable because the operators weren’t aware of the risk of rust in the context of this failure. And so there was no maintenance plan in place to prevent a failure from occurring.

Carbon steel pins are still in use at some dams, but stainless steel pins and self lubricating bearings are now standard practice to mitigate rust and reduce the risk of friction in the future. Reports of existing carbon steel pins note vibrations and increased friction during operation as well as visible rust leaking from the pivot assembly. I think it’s also important to note that while reduced maintenance budgets played a factor, the lubricant they used did not conform to design specifications. It was a new, environmentally friendly lubricant that was not sufficiently waterproof and allowed water to enter the pin assembly and add to the rust. Forensic investigations and Finite Element Analysis of gate 3 offered a better understanding of radial gate behaviour which was used to revise design standards going forward. Folsom Dam gate failure was the only recorded gate failure in the US to occur while the gates were in operation.

Due to the quick thinking of the dam operator, the consequences were limited to damage to the gate and loss of water. The repairs cost $20 million dollars, including repair of gate 3 and upgrades to the remaining 7 gates. The Bureau of Reclamation formed a multi-disciplinary, multi-agency forensic team to investigate and determine the causes of the failure. They initiated a nationwide dam safety program to inspect and evaluate the structural integrity of radial gates across the US. Screening level evaluations of 78 radial gates found 15 structures that didn’t pass and were subject to further review. This also had a significant impact on the design practice of radial gate structures going forward.

So there you have it, the Folsom Dam gate failure. Design oversight, 40 years of rust and poor maintenance procedures resulted in the loss of 40% of the lake. Luckily, the river was able to handle the high flow and no significant consequences were felt downstream; except for the fish which unfortunately migrated much earlier in the season than normal.

For photos, sources and transcripts from this week’s episode head to Failurology.ca. If you’re enjoying what you’re hearing, please rate, review and subscribe to Failurology, so more people can find it. If you want to chat with me, my twitter handle is @failurology, you can email me at thefailurologypodcast@gmail.com, or you can connect with me on Linked In. Check out the show notes for links to all of these.

Thanks, everyone for listening. And tune in to the next episode of Failurology where I have a very special surprise for you. But more on that next time. Bye everyone, talk soon!