EFFICIENT CONTROL OF LEVEL IN INTERACTING CONICAL TANKS USING REAL TIME CONCEPTS

Size: px

Start display at page:

Download "EFFICIENT CONTROL OF LEVEL IN INTERACTING CONICAL TANKS USING REAL TIME CONCEPTS"

Solomon Nicholson
5 years ago
Views:

1 EFFICIENT CONTROL OF LEVEL IN INTERACTING CONICAL TANKS USING REAL TIME CONCEPTS V. Karthikeyan Department of Electrical and Electronics Engineering, Dr. M.G.R. Educational and Research Institute, University, Chennai, India ABSTRACT In this research work an effort is made to design a real time control system for a non linear system and the performance is compared with existing methods. Level control in conical tank is a difficult task considering its non linear design. The proposed controller algorithm is implemented in LPC 1768 processor board along with µ cos II real time operating system for controlling level in the interacting conical tanks. The algorithm is written in keil C Platform. Process reaction curve method is used to design a mathematical model for first order process having dead time for each tank. The designed PID controller Algorithm is tested in real time with conical tanks system. On comparing the work done with existing ones it is found that interaction factor is very less in the proposed method. The results obtained shows negligible steady state error and very high stability compared to conventional methods. Keywords: conical tanks, PID control, LPC 1768, µ cos II, interaction factor. INTRODUCTION Flow and level are the main variables that need attention in process industry. The way of representing processes mathematically is called mathematical modeling. Mathematical modeling of conical tanks is very complex compared to linear tanks. The use of conical tanks plays pivotal role in various process industries like petro chemical, waste water treatment, food processing, concrete mixing industries etc. The main characteristic of conical tanks is its change in cross sectional area at each point of the tank. Conventional PID control algorithms produce best results in controlling the level of the linear tank, but it shows decline in performance due to the non linearity present in the Conical tanks. If the level of liquid rises in the tank, the corresponding area of the tank also increases. The existing decentralized controllers are not suited when there is a large interaction present between the Loops. In the proposed method a PID controller is designed and implemented in real time. The following table describes various process variables related to the work. Literature survey Feed forward controller is a good concept since the process variable is getting controlled before the disturbance has an effect on the process variable, but it has a drawback that all the disturbances cannot be identified and measured [2]. A model of PID controller for conical tanks was designed and comparison between the performance of internal model controller tuning with Particle swarm optimization tuning is done and found that PSO method is superior over heuristic PID tuning [6].Two centralized PI controller tuning methods davison method and Tanttu and Liestehto method are used for the control of interacting conical tanks and their performances are tested for both servo and regulator problems. Tanttu method gives better results [3]. A controller was designed based on skogestad tuning method.the designed controller performs effectively in the minimization of error [9]. Table-1. Process variable description. S. No. Process Variable Description 1 Pressure Force /Area 2 Absolute Gauge pressure + Pressure Atmospheric Pressure 3 Flow Square root of Differential Pressure 4 Flow Rate Amount of flow that flows in a given period of time M = ρ x Q 5 Level Length of liquid column or Pressure the column exerts over a Datum level A neural estimator for non linear Interacting process was designed and implemented it in python language and its performance is compared with fuzzy estimator [10]. The characteristics of the non linear spherical tanks is studied and performance comparison is made using four types of controllers and found FO PI controller reduces the error to good effect compared to other methods [1]. REAL TIME PLANT The Plant taken for study is shown in figure. It consist of three conical tanks, control valve, Differential Pressure transmitter, Reservoir, solenoid valves, hand valves, Current to Pressure Converter, LPC 1768 Board, Computer system loaded with Keil C. Generally an control valve is used for adjusting the input and output of process. Input signal 3 to 15 psi for control valve is coming from current to pressure transmitter. The tanks are fed with liquid from reservoir through control valves. The control signal from controller is given to I to P converter is 4 to 20 ma. Transmitter is device which is used for converting the sensor output to a signal in required form. Usually the 1747

2 transmitter output will be 4-20 ma. It measures the differential pressure from the bottom and top of the tank. Out of three tanks which may be connected in non interacting mode when tank 1 and 3 are sequentially connected with no connection to tank 2 or in interacting mode when tank 1 is connected sequentially with tank 2. Figure-1. Real time plant. Control loops interact when movement of actuator of one control loop affecting both its own process variable and also of other loops. The corrective action of one controller affecting the other process variable as well. It is called interaction between the loops Tank level can be measured by the hydrostatic pressure measurement of the tank P = Height of the liquid X Specific gravity Non linearity to linearity conversion Non linearity to linearity conversion is done at different operating regions of the conical tanks. Volume Calculation of Conical Tank is done as follows Let R - Radius of the tank at 100 % H - Height of the tank at 100 % r - Radius of the tank at x point h - Height of the tank at x point R/r = H/h r = R * h/h Volume at x point V = 1/3 * π * r2 * h Interacting control mode operation The various steps involved in the Interacting conical mode operation is as follows: a) Keep the hand valves MV 1, MV 4, MV 5, MV 7, MV 9, MV 10 AND SV 1 in open condition b) Valves MV 2 and MV 11 are in partially open condition c) Remaining hand valves and solenoid valves are in closed condition d) Power on the pump e) Run the controller algorithm with desired set point f) Based on the control valve movement reservoir tank starts filling tank 1 and tank 2 g) When process variable = set variable, levels in tank 1 and tank 2 are equal h) Introduce disturbance in tank2 by adjusting MV 11 i) Disturbance in tank 1 affects tank 2 ( interaction) Figure-2. Cross sectional view of conical tank. 1748

HARDWARE SETUP In this study the controller is implemented with the combination of LPC 1768 hardware alongwith µc/os-ii kernel are used to minimize the interaction present in the system.

3 HARDWARE SETUP In this study the controller is implemented with the combination of LPC 1768 hardware alongwith µc/os-ii kernel are used to minimize the interaction present in the system. The LandTiger V2.0 NXP LPC1768 ARM development board is a 32 bit Microprocessor used for real time applications. The Board has following features of 512KB on chip flash program memory, Color LCD display interface, USB 2.0 interface, 64KB SRAM for high performance CPU, Two CAN bus communication interfaces, RS 485 communication interface, SD/MMC card (SPI) interface, Two RS 232 serial interfaces, RJ45-10/100M Ethernet network interface, DAC o/p interface and ADC i/p interface, Standard JTAG test/debug interface. LPC 1768 is a high performance and Low power consumption Microprocessor. µc/os-ii is a real time multi tasking operating system kernel version 2. It is used for inter task communication and synchronization and it has superior features compared to other RTOS. The various features are: a) Portable, preemptive, multitasking kernel b) It can handle 64 tasks c) It supports processors up to 64 bit d) It has deterministic execution times e) µc/os-ii is simple to use and simple to implement KERNEL An real time operating system (µc/os-ii) fulfills the requirements of events that happens in real time. USB cable of the LPC 1768 Board is connected to USB port of the system and Serial port of the LPC 1768 Board is connected to Serial port of the system The PID Controller algorithm is programmed, compiled and downloaded to the LPC 1768 board by resetting the board, the control action for the conical tank starts. The experimental setup is shown in Figure-3. Figure-3. Experimental setup 1. USB Cable of LPC 1768(Connected to USB Port of PC), 2. Serial port of LPC 1768(Connected to PC Serial port) (Karthikeyan et al, 2014). CONTROLLER DESIGN The function of controller is based on measurement of process variables. Input/output data are measured from the conical tank plant. The main function of Controller is to produce control inputs which in turn depend on error value. The controller function is implemented in LPC Board and RTOS and tested in Real time non linear system. The efficiency of designed controller is tested using various time domain specifications like rise time, Peak overshoot and settling time. The mathematical model for level of tanks h 1 and h 2 are derived from the mass balance equation. The transfer function of the conical tank 1 and 2 can be approximated with transfer function of first order process with dead time RESULTS AND DISCUSSIONS The open loop response of tank 1 and tank 2 resembles that of the response of First order system having dead time. Both are sigmoidal in shape. Hence the tank response can be approximated with First order with dead time response. The open loop response of Level of Tank 1 and 2 is presented in Figure-4. Where t d = Dead time, K = static gain, = Time constant Mathematical model gives the mathematical relationship between input and output of the system which is an important factor in the design of controller. The process reaction curve method of controller tuning is used for tuning the PID controller for selecting best values for K P K I K D = Proportional gain, = Integral gain, = Derivative gain 1749

18 Process output (cm) Figure-4. Response of Tank 1 and 2.

4 18 Process output (cm) Figure-4. Response of Tank 1 and Process output (cm) The results are obtained in C Language which is worked in µ Cos -II operating system with LPC 1768 Hardware board. The screen captured image of error is shown in Figure-5, which clearly shows that error value is reduced at each instant Time Figure-6. Response of process output with time. 20 Error Error Figure-5. Error value at each instant Time Figure-7. Variation of error with time. The Process output (variable) controlled by PID Algorithm is shown in Figure-6. It is very clear that the response is stable in nature. The variation of error with respect to time is plotted in Figure-7. CONCLUSIONS A PID controller algorithm is implemented in LPC 1768 board with µ cos - II in keil C platform for the interacting conical tanks. The main objective of minimizing the interaction factor is achieved in this paper. Process reaction curve method is used for tuning the designed PID controller. On comparing the results the proposed method works efficiently in minimizing the error compared to conventional algorithms and the algorithm satisfies various time domain specifications like Rise time, Peak overshoot, settling time. The future scope of study will be designing the controller algorithm for three tank interacting system. 1750

5 REFERENCES [1] Priya Chandrasekar, Lakshmi Ponnusamy Comparative study of controllers for a variable Area MIMO Interacting Non linear system. International Journal of Engineering and Technology. 6(1): International Journal of Computer applications. 85(2): [11] Karthikeyan V, Ravi S, Anand M Robust Memory Management using Real Time Concepts. Journal of Computer Science. 10(9): [2] Vinay Kumar and Rekha Jha Liquid level Control of Multi tank system and their Performance analysis. International Journal of scientific research and education. 3(6): [3] Suresh Manic. K, Deva kumar. S, Vijayan. V and Rajinikanth. V Design of centralized PI Controller for interacting conical tank system. Indian Journal of science and Technology. 9(12): 1-4. [4] Lakshmanaprabhu S.K, Sabura banu U, Bharanitharan C Linear Quadratic Gaussian Control for two interacting conical tank process. International Journal of Computer applications. pp [5] Senthil Kumar K, Angeline Vijula D Implementation of two degree of freedom Controller using coefficient diagram method Techniques for Three tank Interacting system. International Research journal of Engineering and Technology. 2(9): [6] Mercy D, Giriraj Kumar S.M Design of PSO PID Controller for a non linear conical tank Process used in chemical Industries. ARPN Journal of Engineering and Applied sciences. 11(2): [7] Mohd Uvais, Prabhat Ranjan Sarkar, shaktidev Mukerjee Three tank interacting system level Control using Modern AI Techniques. International Journal of Research and Development in Applied Science and Engineering. 9(1). [8] Aravind. P, Valluvan.M, Ranganathan. S Modelling and simulation of Non linear tank. International Journal of Advanced Research in Electrical, Electronics & Instrumentation Engineering. 2(2): [9] Vadivazhagi. S, Jaya. N. Modelling and Simulation of Interacting conical tank systems. International Journal of Innovative Research in Science Engineering and Technology. 3(6): [10] Fareeza F, Rama Murthy B Design of Neural Estimator for non linear interacting process. 1751

Relay Feedback based PID Controller for Nonlinear Process

Relay Feedback based PID Controller for Nonlinear Process I.Thirunavukkarasu, Dr.V.I.George, * and R.Satheeshbabu Abstract This work is about designing a relay feedback based PID controller for a conical