Abstract
Effective product quality control is only possible when it is known what effect changing a process parameter setting has on one or more product properties in quantitative terms. “Adding so much more bleaching earth reduces the colour by so much.” Accordingly, such quantitative relationships have to be determined experimentally in the plant. However, the raw materials used in the plant and the plant performance may be inconsistent, vary gradually over time. Diacylglycerols for instance, affect the rate of crystallisation and their concentration varies from one lot of raw material to another. Consequently, an investigation method is required that is not affected by these inconsistencies or eliminates their influence on the correlations being sought. Such an investigation method is described in the present paper. It makes use of changes that either occur inadvertently or that have been introduced deliberately and measures the effect these changes have on dependent parameters. “What effect has a change in hydrogenation catalyst level on hydrogenation time or trans content?” Since these changes are sudden, they are superimposed on the gradual changes and can thus be separated. By introducing changes on a systematic basis, the effect of these changes can be quantified to the degree of accuracy required for product quality control. Moreover, the effect can be used to optimise the process by a kind evolutionary operation.