Modelling and control of high temperature short time milk pasteurization process

Abstract

The shelf life of milk can be increased with pasteurization. The pasteurization is a multivariable interacting process. The temperature profile of milk should be maintained properly at each stage of the process to inactivate the bacteria as well as to retain the nutritional content of milk. The multivariable interaction between the manipulated variable and the control variable makes the control of pasteurization process as complex. The oscillations and overshoot in the milk temperature is undesirable. The control schemes should also be able to reject disturbances at the earliest. The various systems of HTST milk pasteurization plant in Thailand is simulated by eliminating the pre-cooling stage which results in the reduction of costs involved. The open-loop response of simulated model in MATLAB is analyzed for both the existing and modified model. The flow rate of hot water, fuel, and ice water are maintained at the same rate as that of the existing model. In the existing model, the milk temperature is controlled at 4° C at the exit of the cooling stage by keeping the compressor speed at 0.39 times the value of rated synchronous speed (Ns). However, In the proposed model, the speed of the compressor shall be kept at 0.594 times Ns. Also the cost is reduced by eliminating the pre-cooling stage and the cooling tower. This thesis is focused on the system identification techniques to select the best mathematical model that would be suitable to translate the physical behavior of an HTST pasteurization process by using parametric and non-parametric system identification methods. Among the transfer function model, Box-Jenkins (BJ) model and Empirical Transfer Function Estimate (ETFE) modeling structure, it seems reasonable to select the BJ model as a final choice. newline