Modelers can be really smart, and their models can be really sophisticated. But there’s a danger that they, or those in positions of authority, actually believe the models. Recall that modeling low-probability events is fraught with difficulty. And they may be missing some simple, cost-effective solutions by over-relying on computer-generated scenarios.

A recent article points out both the seriousness of fires at nuclear power plants (remember the one at Brown’s Ferry, started by a candle), and issues surrounding fire modeling, such as

Instead of requiring repair of an identified problem, like faulty insulation on electrical cables, regulators will work with utilities to calculate the odds of something going wrong.

This all seems dicey, and as one person said:

Spending $20 million to move electrical panels around, he noted, might be better than spending $20 million “on analysis.” That way, at least the utility would have “really made a fundamental change in the physical plant itself” to enhance safety.

And what about suggestions we’ve seen for elevated water tanks that would simply dump cooling water on a reactor core, or on spent fuel rods, if temperatures get too high?

Interestingly, the Electric Power Research Institute has on sale for $50,000 a “”Fire Modeling Guide for Nuclear Power Plant Applications”, with the objective

To develop a guide for fire protection engineers to use fire modeling tools and methods to support day-to-day plant operation.

And high-priced modeling consultants probably cost way more than just the $50,000 for this guide. Maybe it would be better to spend the $50,000 on a detailed evaluation of equipment and procedures involved in fire prevention, by an independent consultant?