Response Surface Optimization |

This module helps to find optimum values of technological or other (independent) variables, using observations of some output (dependent or response) variable obtained experimentally. Assumptions are that a) the optimum independent variables settings (e.g. temperature, pressure, drying time) correspond to minimum or maximum of the dependent variable mean, b) dependent variable values were observed for various settings of independent variables, c) minimum or maximum of the dependent variable mean exists and is not very far from experimental settings tried. The Response surface module fits a model through the experimental data – complete Taylor polynomial of second order and tries to find its extremum by looking for a stationary point (a point with zero first partial derivatives). When extremum (minimum or maximum) exists, its estimate and confidence interval are given in the protocol. When the optimized model has no extremum, stationary point corresponds to a saddle point and no optimum setting for independent variables can be found. Optimum independent variables setting can be located outside of experimental region, the estimate is less reliable in such case however.
Response Surface Optimization - Pdf manual
Example
Task: Find optimal process parameters
Data:
Protocol Text Output:
Graphical Output:
Reconstruction of the response "curve" for 1 parameter. Examples of 3D-quadratic surfaces for 2-parameter (factor) optimization |