Control and Simulation of FOPDT Food Processes with Constraints using PI Controller M.Y. Pua, M.C. Tan, L.W. Tan, N. Ab.Aziz and *F.S. Taip Department of Process and Food Engineering, Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysia Abstract—The most common type of controller being used in the industry is PI(D) controller which has been used since 1945 and is still being widely used due to its efficiency and simplicity. In most cases, the PI(D) controller was tuned without taking into consideration of the effect of actuator saturation. In real processes, the most common actuator which is valve will act as constraint and restrict the controller output. Since the controller is not designed to encounter saturation, the process may windup and consequently resulted in large oscillation or may become unstable. Usually, an antiwindup compensator is added to the feedback control loop to reduce the deterioration effect of integral windup. This research aims to specifically control processes with constraints. The proposed method was applied to two different types of food processes, which are blending and spray drying. Simulations were done using MATLAB and the performances of the proposed method were compared with other conventional methods. The proposed technique was able to control the processes and avoid saturation such that no anti windup compensator is needed. Keywords—constraints, food process control, first order plus dead time process, PI I. INTRODUCTION ID controller has become the most widely used controller since the last six decades, and is still being used nowadays due to its simplicity and flexibility. All real processes involve constraints that may appear in actuator saturation, for instance. More complex control strategies were usually applied for processes with saturation [1][2]. Saturation will cause windup which will cause performance deterioration such as large overshoot, large settling time and may even become unstable [2][3].In order to compensate this, anti windup compensator is usually added to the control feedback loop [3][4]. As an alternative, this research investigates the possibility to tune the PI controller when the system is under saturation without using anti-windup. II. MATERIALS AND METHODS This research emphasised on applying the methods of controlling a food processes with constraints. A simple method of tuning a PI controller that takes into consideration Pua Ming Yee is a postgraduate student in the Department of Process and Food Engineering, Universiti Putra Malaysia Tan Mei Ching is a postgraduate student in the Department of Process and Food Engineering, Universiti Putra Malaysia Tan Lee Woun is a postgraduate student in the Department of Process and Food Engineering, Universiti Putra Malaysia Norashikin Ab Aziz is a Senior Lecturer in the Department of Process and Food Engineering, Universiti Putra Malaysia Farah Saleena Taip is a Senior Lecturer in the Department of Process and Food Engineering, Universiti Putra Malaysia. (saleena@eng.upm.edu.my, Tel +603-89466357, Fax:+603-89464440) of the actuator saturation was developed. Focus was given on the first order plus dead time (FOPDT) model and the method was based on direct synthesis method. Typically, PI control is sufficient to apply in a large number of control processes, especially when the design requirements are not rigorous for the dominant first (second) order process dynamics [5]. The method was then applied to two different types of food processing and the performances were evaluated. Two different processes were modelled empirically and mathematically. They are the milk spray drying process and blending process. Both processes can be represented by first order plus dead time (FOPDT). The models are shown in Eq 1 and Eq 2 respectively where G P = Process transfer function, Process 1: Blending 1 10 01 . 0 4 + = − s e G s p (1) Process 2: Spray drying 1 8 . 7 2 5 . 3 + = − s e G s p (2) The method, M1 was based on direct synthesis method. In direct synthesis tuning method, there is a parameter λ that needs to be selected and is usually chosen to be half of the process time constant [6]. Simulations were done to find the relationship between lambda, λ and saturation level, U. This was done to ensure saturation can be avoided therefore no antiwindup is needed. Simulations were done to assess the performance of the proposed method, M1 with the existing tuning methods. Comparisons were made for processes with and without antiwindup compensator. III. RESULTS AND DISCUSSIONS Three existing different tuning methods were applied on PI controller; the Ziegler- Nichols method (ZN), Cohen- Coon methods (CC), Abbas method (AA) and the performances were compared to the newly developed method, M1. Simulations were done for processes without saturation, with saturation and with antiwindup. Figure 1 shows the control signals given by PI controller tuned using the 4 different methods for Process 1. PI controller tuned using the CC methods gave rise to the highest control signal, which indicates that it is easily saturated. On the contrary, the control signal obtained using M1 tuned controller was the lowest, which means that saturation is highly unlikely. P World Academy of Science, Engineering and Technology International Journal of Nutrition and Food Engineering Vol:4, No:6, 2010 522 International Scholarly and Scientific Research & Innovation 4(6) 2010 scholar.waset.org/1307-6892/11463 International Science Index, Nutrition and Food Engineering Vol:4, No:6, 2010 waset.org/Publication/11463