263 2011 International Conference on Recent Advancements in Electrical, Electronics and Control Engineering Simple tuned adaptive PI controller for conical tank process S.Anand SASTRA University Tanjore, India s.anand90@gmail.com Aswin. V SASTRA University Tanjore, India aswinvaradan@yahoo.co.in S. Rakesh Kumar SASTRA University Tanjore, India srakesh_84@yahoo.co.in Abstract−This paper proposes an idea for designing a continuously tuned adaptive PI controller for a non-linear process such as conical tank. In this paper, a simple tuning system is used to continuously tune the controller parameters in correspondence with the change in operating points. For each stable operating point, a FOPTD model was identified using process reaction curve method. The estimated model parameters are used to calculate the controller parameters for each operating points. Based on these calculated controller parameters and its operating points, a tuning system was created. The tuning system will able to interpolate and extrapolate the relation between control variable and the controller parameters over entire span of control variables. Finally, a detailed time-domain modeling of the conical tank was performed. Then the adaptive PI controller was implemented in Matlab and was simulated to verify its performance. Thus the adaptive controller was able to produce a consistent response regardless of parametric variations with minimum overshoots and minimum settling time. Keyword− FOPTD system, process reaction curve, conical tank I. INTRODUCTION Adaptive control has always been a successful methodology to control a system with parametric variations. A tuning system of an adaptive control will sense these parametric variations and tune the controller parameters in order to compensate for it. The parametric variation may be due to the disturbance or due to the inherent non-linearity of the system such as conical tank. In a conical tank the cross- section area varies as a function of level which in turn leads to parametric variations. The time constant and gain of the chosen process vary as a function of level. The most distinctive part of the adaptive controller is that it has a controller that is parameterized and an estimator [12]. In the direct approach, the parameterization of the controller occurs in such a way that the desired output is achieved in the closed looped system. In the indirect approach, there will be an element in the parameterized set for the system which characterizes the input-behavior of the system [14]. Hence in this there is calculation required to determine the controller parameters from the system parameters. Figure 1. Block Diagram Of Adaptive Control Conical tanks find wide applications in process industries. Conical tanks with gravity discharge flows are used widely as an inexpensive to feed slurries and liquids with solid particles to unit operations. Conical tank prevents the accumulation of solid particles at the bottom of the tank. Ziegler- Nichols [1] has developed a well known design method to provide a closed-loop response with a quarter- decay ratio. This technique has been used widely to tune the conventional PI controller [13]. A conventional PI controller has limitations in handling this nonlinear process [2] and there are various techniques which have been proposed to overcome these limitations [3]. Adaptive techniques such as gain scheduling, Neural-based control [4],[6], Fuzzy-based control[7], Model-based control [5] etc are employed in recent works to address for the control of non-linear processes. Besides these robust PID controller was also designed to address the model uncertainty by Ming Ge et al [11]. The adaptive system needs to identify the model parameters online by using estimation techniques [8]. This demands a need of complex and fast computational system. In the proposed system, the complexity of the tuning system was reduced. This tuning system contains only two polynomial of lower order to tune the propositional gain and integral gain as a function of level. This paper also propose a detailed modeling of the conical tank using mass-balance equations and the assumption such as time delay and initial value of level are made in order to 978-1-4577-2149-6/11/$26.00 © 2011 IEEE