https://www.nrc.gov/reactors/new-reactors/design-cert/nuscal...
https://www.nrc.gov/reactors/new-reactors/design-cert/nuscal...
There's some hand-waving on safety issues in Table 1.9-2. NuScale passes the buck on flooding issues to the plant owner. Also, there's some hand-waving on software safety issues: "The NuScale design applies RG 1.152, Revision 3 and IEEE Std. 7-4.3.2-2003 that it endorses. For RG 1.168, the requirements of IEEE 1012-2004 are tailored to the NuScale I&C development lifecycle, which is different from the conceptual waterfall lifecycle in IEEE 1012-2004. The applicable tasks from IEEE 1012-2004 to the I&C development are mapped. Some administrative mandatory requirements in the standard conflict with established Engineering or QA documentation requirements. The requirements of IEEE 1028-2008 are tailored to the NuScale I&C development lifecycle." This may mean "We use Agile and your outdated rules don't apply".
The key concept is that, since these plants are passively cooled and don't require pumps for cooling, many of the concerns about keeping cooling water flowing to prevent a meltdown don't apply. "For the NuScale design, the safety analysis shows that core damage does not occur during any design basis event.", says the application. What could possibly go wrong?
A big concern is that NuScale proposes to put up to 12 reactors in one connected pool. I haven't found the part yet where they discuss what happens if one leaks and contaminates the whole pool. Or the pool springs a leak and drains into the ground.
These are not small plants. A 12-unit plant is 720MW of electrical power, so it's getting close to a typical gigawatt sized plant. More small units, instead of one big one.
The history of power reactor meltdown incidents is roughly as follows:
- Three Mile Island - meltdown, big containment vessel, fully contained, expensive but not hazardous.
- Chernobyl - fire, meltdown, no containment vessel, radioactive material dispersed over large area.
- Fukishima - meltdowns, undersized containment vessels breached, radioactive material dispersed over moderate sized area.
A certain amount of skepticism about designs that "don't need" a big containment vessel is perhaps justified.
Look at the IMSR.
You can find leaks in pools pretty easy by using a double lining and distributed sensors or in closed environments like this, by measuring net flows and looking out for any net change in volume.
A NuScale press release does say that. "The system is capable of providing 30 days of cooling without adding water to the pool."[2]
The web site hype goes even further: "Should it become necessary, NuScale’s SMR shuts itself down and self-cools for an indefinite period of time, with no operator action required, no additional water, and no AC or DC power needed."[3] But that's not what the NRC filing says.
Looks like they make bigger claims when they think they can get away with it.
[1] https://www.nrc.gov/docs/ML1924/ML19241A398.pdf [2] https://newsroom.nuscalepower.com/sites/nuscalepower.newshq.... [3] https://www.nuscalepower.com/
The 30 day claim was published in January 2012 in response to the Fukushima accident: https://www.tandfonline.com/doi/abs/10.13182/NT12-A13556
Also, in "NuScale Technical Report Long-Term Cooling Methodology, dated January 2017, TR-0916-51299" (https://www.nrc.gov/docs/ML1835/ML18355A973.pdf)
"5.3.4 Long Term Cooling Evaluation Period LTC analysis is limited to three days. This timeframe is considered acceptable because: (1) the most severe conditions will have been captured within the 72 hour window analyzed, and any conditions that could reasonably be expected to occur beyond this time period are thus bounded by the 72 hour calculation, and (2) after 72 hours, operator actions can be credited. When realistic initial operating levels, temperatures, and decay heat are considered, the scope of the LTC analyses bound the expected level decrease in the reactor pool ultimate heat sink over 30 days, for up to twelve modules transferring decay and residual heat to the reactor pool."
I read this as that they ran a number of scenarios with extreme initial conditions and at the end of three days all operating characteristics were effectively the same as with typical starting conditions and therefore they expected the 30 day simulations would give the same result no matter what the initial conditions were. I don't think they are hiding anything here.
The thing with nuclear reactors is that this sentence was thought to be true for most designs, including those that failed to varying degrees (RBMK, GE BWR, B&W PWR).