System identification for a liquid flow process

Fluid process systems are usually difficult to model because they contain certain nonlinearities and have difficulties to measure several parameters. Therefore, experimental modeling, using measured data, is an appropriate modeling technique for such systems. In this paper, a mathematical model for a FESTO Process Workstation is developed using system identification procedures to formulate the transfer function between the pump voltage and the output flowrate. Input and output data sets are collected from the physical plant, using PC interface with FESTO FluidLAb PA software package. System Identification (SysID) Toolbox in MATLAB is used to formulate several structure models, with different orders, to capture the system's dynamic behavior. The models are tested, validated, and compared using several data sets. The best fit model is selected to represent the liquid process system.