Federal scientists find a way to reduce the cost of running nuclear power plants

The best listening experience is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s daily audio interviews at Apple Podcasts or PodcastOne.

If the nation is to have a so-called green future, many scientists and engineers believe that more nuclear power generation will be part of the equation. Nuclear weapons are expensive to operate. But researchers at Argonne National Laboratory think they’ve found a way to use artificial intelligence to cut costs. For more that Federal Drive with Tom Temin turned to Roberto Ponciroli, the chief nuclear engineer of the Argonne.

Interview protocol:

Tom Temin: There are many complications in the operation of nuclear power plants. Tell us what you see here in terms of artificial intelligence.

Roberto Ponciroli: The idea is that the nuclear industry hasn’t fared particularly well in terms of profitability in recent years. I mean a way to study 2018, more than 1/3 of the units are either unprofitable or scheduled to close. So the point is that the Department of Energy is trying to find solutions to improve profitability. To help nuclear energy, because as you said, nuclear energy better be part of the equation if we want a green future. And in particular, this applies to both currently liberated units and future units. So advanced nuclear reactors that could be used in the current timeline in the power grid. So the point is, there are a lot of avenues you can pursue to reduce the cost of capital and whatnot. But for now, our focus is on reducing operation and maintenance costs. An attempt is therefore made to limit the costs associated with the operation and maintenance intervention. The idea here is to try to use artificial intelligence, artificial AI algorithms, to try and introduce elements of autonomy and automation into the operation of a nuclear unit. So to try to delegate to algorithms error-prone, repetitive, lower-level tasks that are currently handled by human operators, and to try to increase the operators’ situational awareness, I’m trying to say that low-level things, so let’s say have algorithms that take care of it. And let human operators do more interesting and important things and play a more superficial role in running a unit.

Tom Temin: What operations could be applied here to free people from repetitive tasks etc.?

Roberto Ponciroli: As you said, if you have to run a nuclear unit, the nuclear unit is an incredibly complicated system. As I said, there are hundreds, maybe thousands, of light sensors everywhere. And so it really is based on teamwork. And it’s a very complicated job, you know, looking after the performance of components, for example, if a component is failing you have to be aware of that before you do anything, or if you have to do any kind of power transfer. Now there’s the penetration of renewable energy, right? So demand fluctuates, renewable energy fluctuates, and you have to try to manage those demand fluctuations. And you must try to adjust the output power of the device accordingly. And that’s not easy. And like I said, the operational nuclear unit is the result of teamwork, because there are operators, both technicians on site and people in the main control room, they have to talk to each other and this kind of communication protocol. So now it’s based on people talking to each other. But what if we could have algorithms that support these tests and favor the coordination of all these aspects? And when I say aspects, I mean diagnosis, control, and decision-making.

Tom Temin: We speak to Roberto Ponciroli, chief nuclear engineer at Argonne National Laboratory. The information you’ve posted on the subject, simply speaking of having artificial intelligence, gives an indication of when a sensor might be ready to fail. And that’s why you can replace it instead of having people test it and whatnot. That seems like a pretty mundane operation. But how much do you think something like this could save running an average plant in a year? Is that even a calculable number?

Roberto Ponciroli: I can give you a number, you know, usually because, as I said, sensors, like any other component, age. And that’s how the sensors are regularly maintained, right? From time to time you get up and check to make sure your senses aren’t biased or strayed from accuracy and all. And you know there’s a study that was really amazing to me when I realized what they said in the study that when they check the sensor they check all the sensors but about 5% of the sensors need servicing. The other 95% were fine. So it’s a very repetitive, time-consuming task. And although we can have an algorithm that tells us that this sensor is off, you have to do something about these sensors, but all the others are working fine. So don’t waste time with it. And the reason why our research focuses on sensors is because we thought that in this world we make decisions based on sensors. And if we have a very integrated system, okay, if you have sensors that tell you what’s going on and you make a decision and you take a control action, all these circuits are so integrated, if you make a mistake right at the beginning it spreads everywhere very quickly. So if your sensor gives you an incorrect reading, your diagnosis will fail, your decision will fail, and your control action may be genuinely ineffective, or worse, dangerous.

Tom Temin: Secure. And more broadly, knowing what level of operation a reactor should have to respond to the demands of the grid, it seems like the AI ​​could then go further beyond the scope of the facility and study weather patterns, wind and sun patterns, especially if you trying to fill in very unreliable sources like sun and wind, which is nevertheless a policy for so many grid operators that the AI ​​could look at many factors in the environment to say that it’s likely we’ll need more power in seven hours because a cloud gathers or whatever the case may be.

Roberto Ponciroli: That’s really true. Because right now there’s another project, I mean, right, we’re trying to optimize a base capacity of the units. So what I’m saying is that I know the pattern of the sun, the pattern of the wind, and the predictions for future demand. However, given that, we want to use as much as possible, wind and sun and solar energy. So what is the optimal installed capacity? So that applies if you need to build a new unit from scratch. But it’s also true as you said if you already have a unit and I’ll have to determine the best course of action over the next few hours over the next few days. Because as everyone knows, nuclear energy is kind of a slow dynamic, right? So when you switch power, you can’t go up and down like you’re on a roller coaster. Okay, you have to stay there for a while. So you have to think ahead of time what you want to do and what the optimal decision is and definitely all that data, all that pattern recognition, research, it’s just so helpful to tweak the operation of the unit over the next few hours or a couple of days.

Tom Temin: Well, Argonne National Laboratory has that knowledge, is there a way to encourage it, say, in the nuclear industry and even in the larger community of people who are thinking about what the future of the electric grid is going to be and the need for the growing electricity demand of the nation in the coming years and decades?

Roberto Ponciroli: Yes totally. I mean, our national labs are doing a lot of sensory research in our department right now. In our group we have been dealing with control, operation, diagnostics, optimization and so on for years. And definitely we have a lot of contracts with utilities and the Department of Energy. And that’s a very cool part because we’re working on real problems we’re asking the utilities, hey, what’s your problem? And we try to intervene and share our know-how to tackle this everyday problem that is becoming more and more important in recent days and from year to year.

Tom Temin: Sure we’re going to charge these Teslas somehow and we can’t do it on the grid we have if there are too many of them. And just one last question, by extension, let’s say the nation and politics and things are coming together and there’s a movement to actually install new reactors. Could be the process of location determination and proving the effectiveness of the sites etc. which sometimes takes decades. And that’s why they never happen. Could this be improved? You think with AI at some point?

Roberto Ponciroli: Well, in terms of politics, I don’t think that’s really my area of ​​expertise. So I’m not sure I can voice my opinion. What I can tell you is that this architecture that we’ve been working on, that we’re trying to develop, but basically it’s going to be very helpful for advanced reactor concepts. Especially in relation to my project I am talking about here. It’s about the advanced and advanced reactor coupled with memory, you know, the battery basically. And the more complex the system is, the more you need autonomous operation, because the sequence of tasks is, you know, complicated and extensive, and all I’m trying to say here is that this approach is going to be so beneficial for advanced reactors and will definitely help deploy this new technology.

Tom Temin: Roberto Ponciroli is the Principal Nuclear Engineer at Argonne National Laboratory.

#Federal #scientists #find #reduce #cost #running #nuclear #power #plants Source

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *