A wide range of models are available to computer process simulators that require a large number of parameters which often are physical measurements, coefficients, calibration factors and functional forms. This makes it intuitively difficult in the majority of cases, to tune the parameters in order to simulate a circuit in agreement with experimental results. Unit calibrators operating on the principle of least squares minimisation of the difference between predicted and observed plant data, are relatively successful when the parameters are unconstrained and independent of each other. In the case of constrained parameters, a set of carefully chosen transforms is presented which map the unconstrained variables being minimised into the constrained parameters required for the model. This technique has been used successfully with the Powell minimisation algorithm and it may also be applied to other unconstrained minimisation methods. Necessary model characteristics are discussed to enable these transforms to be applied to interdependent model parameters such as points of a graph. Examples of the calibration of several unit models using these transforms are presented. KEYWORDS: Calibration, optimisation, transforms, process simulation.