paulspoon,
Hopefully you can retrieve data that was collected in the past on each run and determine what parameter(s) are *KEY* to the success of a batch. This is where statistical software like JMP by SAS (God bless you Jim Goodnight!) is just wonderful. Take *every* piece of information available, no matter how little you think it has an impact, and create a database. On each row for a batch, record the powder properties(product code, product batch number, particle size, density, acid number, pH when dissolved in water, any and every parameter you have), liquid properties (pH, composition, hardness, density, viscosity, acid number, again - any and every parameter you have), process procedure (initial liquid temperature, agitator initial RPM, temperature ramp, RPM ramp, final temperature, final RPM, etc.) and conditions (ambient temperature, who the operator was, who the lab person was, was vessel washed prior to run, time of day, etc.), and whether it was successful or not(yes or no).
Then, look at the data and see if you can determine what leads to success. If you can't, start collecting data that may be pertinent on every batch and evaluate the data as you go along. There *is* a parameter(s) that will be important, you just have to find it!
Good luck,
Latexman
