“In many cases” in this case means that safe parameters depend on the specific reactor model. Most reactors are made to safely scale output down to a certain degree, within certain timeframes. However, you can’t use reactors like you would use gas plants – powering them up from 0 to 100% output within a half hour. You also can’t use them like batteries that can switch between charging and discharging in sub-second increments. Rather, e.g. here in Germany, many reactors (now defunct) could, with some planning, scale between 60-100% within about a week. And e.g. the proposed SMR from Terrapower was supposed to just run full steam but be able to buffer energy as heat, so electricity output could still be modulated (Terrapower’s first SMR build was cancelled iirc, because of massive cost overruns). But in any case, that still means the Terrapower SMR would not provide “base load”; it would augment what’s needed (you know, if it had actually worked out).
Last I heard (seminar in Summer 2018) NuScale’s SMR was supposed to be able to do load following, but still needed some work analysis and design work to handle the effects of shadowing from control rods to prove they were safe for any power output history. I haven’t followed up since then, but I imagine that’s a bit of a complicated thing to simulate.
“In many cases” in this case means that safe parameters depend on the specific reactor model. Most reactors are made to safely scale output down to a certain degree, within certain timeframes. However, you can’t use reactors like you would use gas plants – powering them up from 0 to 100% output within a half hour. You also can’t use them like batteries that can switch between charging and discharging in sub-second increments. Rather, e.g. here in Germany, many reactors (now defunct) could, with some planning, scale between 60-100% within about a week. And e.g. the proposed SMR from Terrapower was supposed to just run full steam but be able to buffer energy as heat, so electricity output could still be modulated (Terrapower’s first SMR build was cancelled iirc, because of massive cost overruns). But in any case, that still means the Terrapower SMR would not provide “base load”; it would augment what’s needed (you know, if it had actually worked out).
Last I heard (seminar in Summer 2018) NuScale’s SMR was supposed to be able to do load following, but still needed some work analysis and design work to handle the effects of shadowing from control rods to prove they were safe for any power output history. I haven’t followed up since then, but I imagine that’s a bit of a complicated thing to simulate.