COVID-19 and Nuclear Power Risk Management
Good Day:
I had a lengthy phone chat with a colleague who has performed many, many risk assessments for nuclear power reactor risk issues. I contacted him to see if it was possible to use existing computer models to evaluate the risk implications of deferring inspections of components, relaxing the limits on work hours, etc.
As I suspected, he explained how easy it was to do. The challenge is in defining the performance and/or reliability reduction from a deferred task or relaxed limit. I posed several examples to him and asked how the potential risk implications could be modeled. For the 17 motor-operated valves on Limerick Unit 1 that Exelon sought to defer in-service inspections, the failure of the MOVs to perform their safety function (to open in some cases and to close in other cases) could be doubled. The model will then calculate the chance of core meltdown, which can be compared to the baseline (i.e., no impairments) case to see how the deferrals impacted safety margins.
For the case of relaxing work hour limits, the chances of worker mistakes in the model can be increased by multiples of 2, 5, and 10 (meaning the chance of an operator failing to start a safety system when needed is twice, five-times, or ten-times higher) to gauge the sensitivity of safety to human performance.
And the models can easily consider multiple impairments -- all 17 MOVs not being inspected this refueling, work hour limits being relaxed, torus coating remediation being deferred, etc.).
The colleague also apprised me of something I'd not realized (except in hindsight). The operating reactors vary in design and age. The risk value of a 2-inch valve at Plant X can be vastly different from that 2-inch valve's risk value at Plant Y. Similarly, some plants rely more on operators performing safety tasks than others; thus, the safety implications of relaxing work hour limits affect reactor safety differently. The NRC issued blanket relaxations of work hour limits despite knowing that the impact at some reactors would be quite higher than at other reactors.
But to the colleague's awareness, neither the NRC nor any plant owner has evaluated the risk implications from any of the assorted deferrals, reliefs, and relaxations. When I pushed back on this assertion, the colleague pointed out that if assessments had been performed, either the request or the approval would have cited the results.
The risk models are routinely used by plant owners for things like monitoring safety margins during refueling as the equipment is removed from service for maintenance and later returned to service. The NRC routinely uses the models to determine which crayon to pull from the set to color violations and kill time until lunch.
In any case, I prepared the attached slideshow to reflect my understanding of the risk assessment models and how one can account for deferrals, reliefs, and relaxations -- that it, if one were actually a risk-informed regulator and not a risk-deformed pretender.
Feel free to pass along the slideshow.
Thanks,
Dave
I had a lengthy phone chat with a colleague who has performed many, many risk assessments for nuclear power reactor risk issues. I contacted him to see if it was possible to use existing computer models to evaluate the risk implications of deferring inspections of components, relaxing the limits on work hours, etc.
As I suspected, he explained how easy it was to do. The challenge is in defining the performance and/or reliability reduction from a deferred task or relaxed limit. I posed several examples to him and asked how the potential risk implications could be modeled. For the 17 motor-operated valves on Limerick Unit 1 that Exelon sought to defer in-service inspections, the failure of the MOVs to perform their safety function (to open in some cases and to close in other cases) could be doubled. The model will then calculate the chance of core meltdown, which can be compared to the baseline (i.e., no impairments) case to see how the deferrals impacted safety margins.
For the case of relaxing work hour limits, the chances of worker mistakes in the model can be increased by multiples of 2, 5, and 10 (meaning the chance of an operator failing to start a safety system when needed is twice, five-times, or ten-times higher) to gauge the sensitivity of safety to human performance.
And the models can easily consider multiple impairments -- all 17 MOVs not being inspected this refueling, work hour limits being relaxed, torus coating remediation being deferred, etc.).
The colleague also apprised me of something I'd not realized (except in hindsight). The operating reactors vary in design and age. The risk value of a 2-inch valve at Plant X can be vastly different from that 2-inch valve's risk value at Plant Y. Similarly, some plants rely more on operators performing safety tasks than others; thus, the safety implications of relaxing work hour limits affect reactor safety differently. The NRC issued blanket relaxations of work hour limits despite knowing that the impact at some reactors would be quite higher than at other reactors.
But to the colleague's awareness, neither the NRC nor any plant owner has evaluated the risk implications from any of the assorted deferrals, reliefs, and relaxations. When I pushed back on this assertion, the colleague pointed out that if assessments had been performed, either the request or the approval would have cited the results.
The risk models are routinely used by plant owners for things like monitoring safety margins during refueling as the equipment is removed from service for maintenance and later returned to service. The NRC routinely uses the models to determine which crayon to pull from the set to color violations and kill time until lunch.
In any case, I prepared the attached slideshow to reflect my understanding of the risk assessment models and how one can account for deferrals, reliefs, and relaxations -- that it, if one were actually a risk-informed regulator and not a risk-deformed pretender.
Feel free to pass along the slideshow.
Thanks,
Dave
- Log in to post comments
