For the BLX performance it is very important that the entered standard devations of the samplers and online analyses are as correct and precise as possible.
If the Standard deviations are to high BLX will re-act slower because it assumes that the analysis from the XRF or online is unprecise.
Below are two figures illustrating the impact a too high standard deviation have on the perfomance.
The black arrow shows where the standard devations for the raw meal sampler (top graph) and the online sampler (lower graph) was changed to more "loose" values. There is a clear trend showing that the estimation becomes more unprecise with the higher standard devations
On Image #2 are shown how the standard devation influence the LSF, SIM and ALM values. The Red line is the actually measured. The blue is the estimated with the high standard deviations and the Green line is the new calculated based on low Standard deviation values. THere is a clear trend showing that with the lower standard devation the chemical modules are estiamted more correct.
If unsure on the standard devations from site - then ask for this relatively simple exersite to be done:
- The site sould take 12 spot samples of Raw Meal right after each other and from this material prepare 12 samples which should then be analysed.
- Then calculate the standard devation of these data.
- For
a PGNAA:
Do a trend in QCX of the medium standard deviation of the oxides covering 1 week. Then look for the lowest value for each oxide and this can then be used as a start noise value for the online. Be prepared to adjust these values as the tuning is ongoing.