The day the lights go out in Oregon, S11E1

s11 e1


The long-feared 9.0 magnitude Cascadia subduction zone earthquake, which seismologists say is inevitable, will damage or destroy large swaths of Oregon’s electrical grid. How long will it take to get the juice flowing again? Weeks? Months? Professor Ted Brekken and his team are applying high-powered simulations to find out and to identify which parts of the system should be hardened against the quake at any cost.  

Season number
Season 11
Episode number

[MUSIC: “Blacksmith,” by Godmode, part of the YouTube Audio Library, licensed under Creative Commons.] 

STEVE FRANDZEL: In February, a big winter storm hit Western Oregon. Thousands of downed trees and broken limbs took out power lines all over the place. Almost 300,000 people lost power, some for more than a week. Some friends of mine in Salem had to bug out to a motel in Corvallis for four days, with their pets, to escape the near-freezing temperatures in their house. It was a mess. But that was a picnic compared to what’s in store after that big earthquake we keep hearing about.

The long-feared 9.0 magnitude Cascadia subduction zone earthquake, which seismologists say is inevitable, will be one of the worst natural disasters in American history. It could claim thousands of lives and force tens of thousands more from their homes indefinitely.  Buildings, roads, bridges, seaports, airports, electrical and telecommunications grids, natural gas and water systems — a lot of it will break. 

So on that happy note, welcome. I’m your host, Steve Frandzel. In this episode of Engineering Out Loud, we’ll zero in on the fate of the electrical grid and some ambitious research to figure out how much of it will fail, how long it will take to fix, and what can be done to get the juice flowing as soon as possible after the shaking stops.

[MUSIC: “The Ether Bunny,” by Eyes Closed Audio, licensed under CC by 3.0.]

FRANDZEL: From the College of Engineering at Oregon State University, this is Engineering Out Loud.

TED BREKKEN: We would probably be looking at being without electricity for several months, perhaps up to three months in the Willamette Valley, and along the coast you would look at being without electricity for perhaps up to six months. 

FRANDZEL: That’s Ted Brekken, a professor of electrical and computer engineering at Oregon State. His projections come from the Oregon Resilience Plan, a 2013 report that details what Oregonians can expect a major earthquake to do to the state’s infrastructure. It also lays out the priorities and the actions needed to survive and to bounce back. If I had to sum up the findings in one sentence, it’d be: We aren’t ready, but things are getting better. 

BREKKEN: Because we have these broad estimates from the Oregon Resilience Plan that’s based on expert opinion. And so, number one, I want to see how our numbers compare to that — not a qualitative analysis but this actual quantitative analysis that we’re doing.

FRANDZEL: To create that analysis, Ted is leading a team of three other researchers whose expertise covers power systems operations and analysis, geotechnics, earthquake engineering, mapping, and spatial analytics. Their main objective is to determine how much of the system will fail and how long will it take to restore. The foundation of their analysis is a computer model of the Western power grid, which spans 14 states and about two million square miles. Most of it won’t be touched. But if you’re in Western Oregon or Western Washington, Northern California or Southern British Columbia, well… 

[MUSIC: “Hypnosis,” by Godmode,  part of the YouTube Audio Library, licensed under Creative Commons.] 

BREKKEN: Yeah, so, the fault is large and it runs all the way from Northern California on up into off the coast of British Columbia. We expect probably the worst will be in Southern Oregon and then radiating from there. The severity of the shaking will be quite severe at the coast and will diminish to lighter levels of destruction by the time that you get to Bend for example. The magnitude of the shaking is actually highly dependent on geographical features, on the type of soil and how deep the bedrock is, and factors like this. And so we can make broad observations about the severity of shaking across hundreds of miles scale. But within that, there will be local variation as well. There’ll be some locations that might escape the most severe damage and other locations where things may be very severe. And that difference could be as much as just one city block from the next. 

FRANDZEL: Over that map of the vast Western grid, Ted and his team overlaid a digital representation of the grid’s assets and loads. We’re talking about up to 30,000 individual components from the team’s grid model. 

BREKKEN: So think all the transmission lines, all the distribution lines, all the transformers, all the circuit breakers, all the substations, all of those pieces of equipment.  

FRANDZEL: And also think power generators, like hydroelectric turbines. And think about loads, which is anything that pulls electricity from the grid. Lights, heat, refrigerators, computers, pretty much anything that plugs in.  

The electrical grid is just too big and too complex for any model to incorporate every single element, but enough of them are represented, or at least grouped together, to capture large-scale behavior. In Ted’s model, for example, a collection of substations that serves a town might be abstracted down to a single large substation. Or an entire town might be represented as one large load on the system. And the model must be geographically accurate. That seems obvious when you’re dealing with a specific area of Earth. But it’s often not necessary when you’re evaluating the mathematical underpinnings of a system, which in this case is the flow of electricity. 

BREKKEN: For this research, we do need a model that is geographically accurate, meaning the generators need to be where the generators actually physically are, and the loads need to be where the loads physically are, because the amount of shaking is tied to those locations. And I should also mention that I’m speaking specifically about shaking, because I think that’s what most people think of when you think of earthquake damage, and that is a significant factor. However, there are other modes as well.

FRANDZEL: Like landslides, rockslides, soil liquefaction, tsunamis, sinking ground, and even fires. Some of those are accounted for in the model’s computations. 

BREKKEN: And we then assign to that model certain seismic properties of all that equipment. And additionally, some estimates for how badly that piece of equipment is going to be shaken based on geographically where it is. And so that allows us then to run this probabilistic analysis where we can hit the system with sort of a virtual earthquake. We can kind of roll the dice on all these pieces of equipment and we can see which pieces of equipment fail, then we can see if we have a viable system after that. Do we have an electrical system that can still operate in any way? And we run this test then, this sort of probabilistic analysis, hundreds of thousands of times, and that gives us then a probabilistic view of how bad we expect the system is going to be hit.

[MUSIC: “Funky Toms,” by Jed Irvine, used by permission of the artist.]

FRANDZEL: On top of all that is a time element to predict how the system will fare in the hours and days after the earthquake. So the simulation is run shortly after the virtual earthquake, then again on the day after the quake to see how much of the equipment is still functional and what’s been fixed or replaced. Then again on day two, day three, day four, and so on.  

BREKKEN: Again, we have a certain system. Some assets have recovered, some haven’t. We again check whether that’s a viable system, we flag that result, and then we move forward another day. So let’s just take this specific example of a transformer. Let’s say that we roll the dice and a transformer has failed. We will virtually move the clock forward one day. And let’s say for this transformer, there’s a 10% chance on any given day that you’re able to get it replaced. S o we’ve moved the clock forward one day and we’d roll the dice on a 10% chance that we get that transformer back. Let’s say that we don’t, all right, well that transformer’s still out. And again, we do this for all of the assets in the system.

FRANDZEL: The estimates for how long it could take to fix or replace an asset were determined by talking with lots of experts. 

BREKKEN: If you had a failed transformer, right, what are the chances that you could have a replacement for it one day later, or a week later, or a month later? What is your organization looking at? How quickly could you get technicians on site? How quickly could you get the equipment in hand? How quickly could you get it installed and get it up and running again. That’s the number I’m really looking for. I want to see how long are we looking at recovering here.

But the other thing that we really hope to see out of the work is I think an understanding of where the weak points are. I think we hope to kind of get some idea for if there are certain bottlenecks, if there’s certain types of pieces of equipment or there’s certain geographic locations that tend to have an out-sized effect.

FRANDZEL: Armed with that kind of information — like which chunks of the grid are more important to preserve — Ted and his team can come up with recommendations for improvements. 

BREKKEN: We would hope to define those particularly key, very high-value either assets or locations. Say for example, this particular substation is really, really important. You have to make sure that the substation is still operational in an earthquake, because there’s so much other parts of a grid that depend on that being there. And we can make some kind of qualitative assessments of those things already, but it would be nice to see a more quantitative approach to that.

FRANDZEL: So maybe they recommend beefing up a very important substation by anchoring its components more firmly and giving it a better chance of functioning after the shaking stops. 

BREKKEN: You want them to be firmly connected to the ground, for example. If they’re not firmly connected to the ground, then the foundation can move underneath the piece of equipment and it can tip over, it can slide off of its foundation pad. You also have these pieces of equipment that are going to be moving relative to each other. So any wires or conductors that are connecting them are then going to be stretched, and those pieces of equipment are going to pull on each other and maybe amplify each other’s motion, and then you can have things break. So the standard is related to how those things are installed and how they’re connected to the ground and how they’re connected to each other in a way that gives them the best chance of surviving.

FRANDZEL: The standard he’s referring to comes from the Institute of Electrical and Electronics Engineers and covers the seismic resistance of electric substations. The Oregon Resilience Plan recommends applying the standard more aggressively. 

[MUSIC: “Two Moons,” by Bobby Richards, part of the YouTube Audio Library, licensed under Creative Commons.]

There are some important puzzle pieces left out of the model. That’s because earthquakes are equal opportunity destroyers. A big one will simultaneously disrupt or annihilate all the other lifelines I mentioned at the start: roads, bridges, seaports, airports, telecommunications, natural gas, water, sewers.

BREKKEN: So, when we look at the recovery of equipment, we’re actually not considering, say, the damage to a transportation system, to the roads and things like that. And obviously, that then greatly interferes with the ability to bring new equipment, for example, if we are looking at the electrical system. That hampers the ability of repair teams to get into areas and replace equipment or repair equipment. All of these lifelines, they’re all interconnected, and disruptions in one does hamper the ability to help with repairs in another. So that interrelatedness is absolutely a big issue, and it’s one that’s difficult to predict exactly how bad that will be, but we know it will happen. It can be challenging enough to understand what the impacts are going to be on one of those lifelines, but then when you add the interdependencies, that gets to be a pretty complex issue. You start as simple as you can, and then you add complexity as you move forward. And hopefully then in future years, this work will add up to even deeper understanding about how we’re connected to the other lifelines, for example.

FRANDZEL: So, the virtual earthquakes have been run. The thousands and thousands of recovery simulations have been logged. A full analysis of what it all means will take a while — maybe another year or two. Putting together a recovery plan to ease the burdens of that terrible day will also take some time. You can take a little consolation on from knowing that utilities and state agencies have not been idle when it comes to hardening our infrastructure. 

BREKKEN: The utilities in the area have been very aggressive since the publishing of the Oregon Resilience Plan, to implement those standards. And so we are currently in the midst actually of upgrading a lot of the electrical equipment, and there’s still a lot of work to do, but a lot has been done. And it’s encouraging to see that movement. 

[MUSIC: “Two Moons,” by Bobby Richards, part of the YouTube Audio Library, licensed under Creative Commons.]

FRANDZEL: This episode was produced and hosted by me, Steve Frandzel, with additional audio editing by Will Havnaer. Ted’s research colleagues at Oregon State are faculty members Eduardo Cotilla-Sanchez, Michael Olsen, and Armin Stuedlein, and research assistants David Glennon and Eve Lathrop. Our intro music is “The Ether Bunny” by Eyes Closed Audio on SoundCloud and used with permission of a Creative Commons attribution license. Other music and effects in this episode were also used with appropriate licenses. You can find the links on our website. For more episodes, visit, or subscribe by searching “Engineering Out Loud” on your favorite podcast app. Bye now.

Featured Researchers