utah,
While you and the others are waiting for the magical slides on transfer functions to come your way, consider that this concept and its understanding is not a new one. The development of a “transfer function” or least-square model of the process using Response Surface Methods have been with us for over 30 years. One should consider developing a process model beginning in the Analysis phase by identifying candidate inputs to evaluate. If you can use Design of Experiments with the process, then using a special set of DOEs will allow you to define a linear or quadratic model from the experimental data using conventional linear and/or non-linear regression. If your process is in the vicinity of optimal, then your data will reveal curvature in the model providing you with a quadratic model. If your process is further away from optimal, then your data will reveal little significant curvature in the model, and provide a linear model. You could search the internet or your reference for non-linear regression, Response Surface Methods, and Design of Experiments for additional detail.
Happy Hunting!
Ken
.
