If this fault is judged as critical, and it seems to me it should be, then a little learning from ISO 9001 would seems appropriate. Laughing it off just does not make it (sorry whistle). So is the input specification to this issue is no explosions? If so then what are the processes and procedures to arrive at an output spec where this does not happen. Obvious, of course, most quality management is but first we set up IQ/OQ. If I sound like a burned out ex QC manager of an aseptic unit (I needed a job and it paid incredibly well) unit who suffered as a bought in consultant for a customer under a 483 violation (google that) from the FDA and what it took to fix that then I confess that is me. Injectibles most be sterile and coal stoves must not explode in the night - to me it is the same. So qualitative statements like "hysteresis must be reduced accordingly" would mean nothing to an FDA inspector and it means the same to me. Yanche muttering darkly about understanding anthratubes also is a non-sequitur.It seems that the entire thing is related to how much ash the coal makes. If your coal is not making much ash the SV and the hysteresis must be reduced accordingly to keep the fire high in the tube.
A graph correlating ash to hysteresis must be constructed and validated. That even assumes you believe at this point that is the only problem here and I suspect it is multivariate. Given the number of brilliant minds looking at this one a single source fix would have been discovered by now. There needs to be a explosion/no explosion area of the graph and the stove controls need to automatically prevent that from happening. My friends you are light years from that point - sorry these arguments are not convincing. It would be nice to fix this before someone blows themselves up. Never mind the human tragedy, an injury attorney will own AHS after the trial.