Survey on PID Control System Design and Tuning Software Tools

Transcription

1 Survey on PID Control System Design and Tuning Software Tools Kiam Heong Ang and Yun Li Centre for Systems & Control, and Department of Electronics & Electrical Engineering, University of Glasgow, Glasgow, G2 8LT, U.K. Abstract- Tuning a proportional plus integral plus derivative (PID) controller seems conceptually intuitive by observing the behaviour of the controlled process in terms of overall response, steady-state error and transient, until the closed-loop system performs as desired. However, PID tuning by such trials-anderrors can be both an art and science, and can often be unreliable and tedious. The complication has led to the development of software tools that can guide designers through the tuning process with or without all the technical details understood. This paper presents a critical overview on available PID tuning packages and comments on a way forward. I. INTRODUCTION The proportional plus integral plus derivative (PID) control technique was invented in 9, largely owing to Sperry s ship autopilot. Such a controller provides control actions through a proportional (P) all-pass gain, eliminates steadystate tracking errors by enhancing low-frequency compensation using a simple integral (I) action, and improves transient and smoothness by enhancing highfrequency compensation using a simple derivative (D) action. These actions were first realised by electromechanical means and later by analogue electronics and now mainly by microprocessors. Since its invention, the PID control scheme has found wide applications in industry, owing to its simplicity and good performance provided by these three easily understood actions. The easy-to-use Ziegler-Nichols (Z-N) tuning method fertilised the popularity of PID control in almost all sectors of manufacturing industry [2][7]. The PID controller is now regarded as the bread and butter of control engineering. Today, about 90-95% of all industrial control problems are solved by this type of controllers in various forms [2]. The success and longevity of PID were characterised in a recent IFAC workshop [6], where over 90 papers on the state-of-the-art PID research were presented. The trend in tackling various tuning problems appears to be the integration of available methods in the form of software. However, there exists almost no publication on this trend. The aim of this paper is therefore to provide a survey on the PID tuning software currently available and to comment on future trends. A PID controller is generally represented by U G( GP GI GD E( p G( G i II. BACKGROUND s PD p d G s PI p Td s Ti s where the control action, u(t) = L - [U(], is taken upon the error signal, u(t) = L - [E(]. Equation () is the simplest and most common form, the parallel form or the standard form, as shown in Fig. for implementations. The controller in this form may possess complex poles. Fig.. PID Controller Structure Parallel Form If Ti > 4Td, only real poles exist and hence a series form for implementation yields, as given by: where G D ( G P ( G I ( I E ( s ) U ( s ) G D ( Td s Ti s 4T d / Ti 2 G P ( G I ( E ( s ) U ( s ) Fig. 2. PID Controller Structure Series Form () (2) (3)

2 In this series form, the PI factor may be implemented to counter actuator saturation without the need of anti-windup. Such a set-up is shown in Fig. 3. If there is no saturation, the feedforward-path transfer is unity and the overall transfer from UPD( to U( is the same as the last factor in Eq. (2). U PD ( Actuator model Actuator model s Fig. 3. Anti-windup PI part of a series form The widely practised Z-N method does not yield optimal performance and its application range is rather limited. Over the past half century, search goes on to find the next key technology or methodology for PID tuning [4]. An excellent summary on PID tuning methods can be found in [][3][5][6]. However, no tuning method so far can replace the Z-N method in terms of familiarity and ease of use in control engineering practice. A need for the development of PID tuning software has therefore arises []. The aim of PID tuning software is to allow a practitioner with minimal control knowledge to be able to tune a PID controller efficiently. Ultimately, it will increase the practising company s efficiency and profitability since it does not need to invest vast amount of manpower in tuning those PID controllers, but gain savings arising from improved quality. III. PID SOFTWARE FEATURES Table I shows some of the available PID tuning software. Most of them are generic except AdvaControl Loop Tuner (Advant OCS system), DeltaV Tuner (DeltaV workstation), Intelligent Tuner (Fisher-Rosemount PROVOX controller), OvationTune (Westinghouse DCS), Profit PID (Honeywell TPS/TDC system), PID Self-Tuner (Siemens SIMATIC S7/C7) and Tune-a-Fish (Fisher-Rosemount PROVOX controller). Note that Tune-a-Fish was discontinued on 2 nd April 2002 and ExperTune Inc. will handle any support or upgrade issues directly. IMCTune and CtrlLAB are meant for control design. Hence, it does not fit well into the category listed in Table I, but is suitable for learning and testing of controller designs. Nevertheless, they are still listed in Table I for information. Table II shows a summary of the known tuning rules. Note that RSTune TM and Tune-a-Fish are not shown in Table II because it is ExperTune that is embedded in their products. A. Model-Based Tuning Almost all the software mentioned in Table I employed model-based tuning method. First order process with time delay as shown in equation 4 is most commonly used to approximate the plant: T i U( s Ke G( (4) s where K is the process gain; τ is the process time constant; θ is the process dead time (or delay); and s is the Laplace domain variable. B. PID Controller Structures This poses most problems in PID tuning rules. As different vendors have their own design of controller structures, tuning rule for a specific controller structure does not perform well in another structure. One solution as seen in the software is to provide support for individual structures. C. Online Support Quite a few software packages in Table I do not support online operations, e.g. real-time sampling of data, on-line tuning etc. The common non-vendor specific interfaces supported are Microsoft Windows Dynamic Data Exchange (DDE) and currently the infamous OLE for Process Control (OPC ) [8]. OPC is an industry standard created with the collaboration of a number of leading worldwide automation and hardware/software suppliers working in cooperation with Microsoft. The standard defines method for exchanging real-time automation data among PC-based clients using Microsoft operating systems. Thus the aim of OPC is to realise possible interoperability between automation/control applications, field systems/devices and business/office applications. D. Operating Systems and Hardware/Software Dependent Microsoft Windows is currently the most popular platform supported based on Table I. is another popular software commonly used in control applications due to its ease of implementation. E. Tuning Methodology Setpoint following and load disturbances rejection are the two main concerns in PID tuning. Thus, most of the software mentioned in Table I have catered an option for user to select. Table II is listed according to the popularity of a tuning rule then followed by those with known tuning rules and lastly are those without any information on tuning rule used. It is apparent that IMC tuning rule is the most widelyaccepted tuning rule based on Table II. F. Other Features In addition to the traditional features provided by most PID tuning software, now the trend is to provide some additional features like diagnostic analysis, which is very helpful. An example is ExperTune which have a wide range of features for e.g. valve wear analysis, robustness analysis, automatic loop report generation, multi-variable loop analysis, power spectral density plot, auto and cross correlations plot, shrinkswell (inverse response) process optimisation etc.

3 TABLE I PID TUNING SOFTWARE Manufacturer Product Name (a) (b) (c) (d) (e) (f) ABB [9] AdvaControl Loop Microsoft Windows and Tuner Advant OCS system ACT [0] TOPAS.2 Microsoft Windows 2000 for single user AlgoSys Inc. [] SimAxiom TM (Off-line Microsoft Windows tuning) DynAxiom TM (On-line? tuning) Artcon Inc. [2] PITOPS TM Microsoft Windows BESTune [3] BESTune 4.4 Microsoft Windows and US$ 500 per copy Prof. C.B. Brosilow [4] IMCTune Microsoft Windows and Freeware Communications & Systems GRAPHIDOR Microsoft Windows COTEMS [5] Control Arts Inc. [6] Model ID & PID 3.5 Microsoft Windows US$ 699 for Tuning Software single user Control & Optimization Robust PID Tuning? Microsoft Windows Specialists [7] Control Soft Inc. [8] INTUNE TM 4.2 Microsoft Windows Dr. Doug Cooper [9] Control Station 3.0. Microsoft Windows US$ 895 per year for single user yearly maintenance license EMERSON TM Process Management [20] DeltaV Tuner [2] 5. DeltaV workstation and DeltaV controller running control software Intelligent Tuner [22] DEC OpenVMS VAX or OpenVMS AXP series and OpenVMS version 6. or later operating software; PROVOX 0-series, 20- series, 20-series SR90 controllers, or SRx controllers OvationTune [23] Westinghouse Process Control DCS EngineSoft [24] pidtune TM.0.5 Microsoft Windows and license ExperTune Inc. [25] ExperTune Microsoft Windows Honeywell International Inc. [26] Profit PID Honeywell TPS/TDC IPCOS Netherlands/Belgium [27] RaPID.2 Microsoft Windows and 3300 for single user for a year ISE, Inc. [28] Commander 4..4 Microsoft Windows Supervisory Software JC Systems [29] JC Systems Toolbox Microsoft Windows and US$ 495 per LabVIEW TM copy Lambda Controls [30] Control Loop Assistant.0c Microsoft Windows National Instruments TM [3] LabVIEW TM PID Microsoft Windows and Control Toolset for LabVIEW TM Windows Plant Automation Services, Inc. [32] TuneWizard TM Microsoft Windows

4 Rockwell Automation [33] RSTune TM Microsoft Windows and Allen-Bradley PLC-5, SLC 500 TM or ControlLogix PLCs Siemens [34] PID Self-Tuner 5.0 Microsoft Windows and S7-300/400 station; STEP 7 ( V3.2) and Standard PID Control V5 installed on programming device Controller Tuning Microsoft Windows US$ - Base price Straight-Line Control Co., Inc. [35] Techmation, Inc. [36] ProTuner TM Microsoft Windows TiPS, Inc. [37] Tune-a-Fish Microsoft Windows and Fisher-Rosemount PROVOX Controllers UNAC Automation Pty Ltd [38] EZYtune TM..02 Microsoft Windows US$ 99 per copy University of Glasgow [3][39] PIDeasy TM.0 Microsoft Windows Visual Solutions, Inc. [40] VisSim/OptimizePRO TM 4.0 Microsoft Windows and Professional VisSim 4.0 Xiera Technologies Inc. [4] GeneX 2.0 Microsoft Windows and Prof. Dingyu Xue [42] CtrlLAB 3.0 Microsoft Windows and Notes: (a) (b) (c) (d) (e) (f) Freeware Model-based tuning. Indicate software that matches the open/closed loop plant response data to a specific model. Support vendor specific PID controller structures. Indicate software that explicitly support vendor specific PID controller structures and not those just support some different generic PID controller structures. Support online operation. Indicate software that supports online operation like sampling of data, online tuning etc. Software version reviewed. Operating Systems and Hardware/Software Dependent. Prices. Please contact the manufacturer for updated prices on their products. Legend: Support Does not support? Probably support Information not available TABLE II PID TUNING RULES Tuning Rules Product Name Optimise for IMC IMCTune IMC Model ID & PID Tuning Software IMC pidtune TM Modified Ziegler DeltaV Tuner Performance setting ranging from no overshoot to very aggressive Nichols rules for PI Phase and Gain margin rules for PID Lambda tuning rules for PI Lambda-Averaging Level for PI Lambda-Smith Predictor IMC tuning TuneWizard TM Setpoint following or Load disturbances rejection IMC (Lambda) Surge Tank application

5 Robust PID Tuning Non-integral Process:-Modified IMC/Lambda tuning -Tuning using ratio of closed-loop to open-loop response time Integral Process: Tuning using closed-loop response time Advanced IMC based INTUNE TM algorithm Lambda tuning correlations Control Station Setpoint following or Load disturbances rejection ExperTune Setpoint following or Load disturbances rejection Quarter amplitude damping 0% overshoot Lambda (standard or level) Lambda Tuning Control Loop Dominante Pole Placement Method extended with Robustness Criteria (DPPM- RC) Assistant AdvaControl Loop Tuner Fast, normal or damped closed-loop performance GRAPHIDOR The software generate a 3-D plot with the range of P and I variables on the x and y axes and the error on the z axis. Objective is to find the deepest valley on the error surface i.e. the minimum error. Where the minimum error is, those are the P and I values that you want to use. Proprietary min-max algorithm Based on pole cancellation with gain and phase margin and closed loop damping factor specifications Based on closed loop time constant (beta) and 0%-90% rise time Profit PID ProTuner TM 32 Fast, Medium or Slow response to setpoint following or load disturbances rejection EZYtune TM Proprietary algorithm PIDeasy TM VisSim/OptimizePR O TM Use a generalised, reduced gradient algorithm named GRG2 to determine optimal values for the design variables subject to all the constraints CtrlLAB Optimise for ISE, ISTE, IST 2 E or Gain/Phase margins TOPAS Setpoint following or Load disturbances rejection Tight or Average level control SimAxiom TM Desired closed-loop response time DynAxiom TM PITOPS TM Setpoint following or Load disturbances rejection BESTune Controller tightness Intelligent Tuner OvationTune RaPID Setpoint following or Load disturbances rejection or Both Commander Supervisory Software JC Systems Toolbox LabVIEW TM PID Control Toolset for Windows PID Self-Tuner Controller Tuning 0 GeneX

6 IV. CONCLUSIONS PID, a structurally simple and generally applicable method, stems it success largely from the fact that it just works very well with a simple and easy to understand structure. Research during the past half century has overwhelmed the simplicity of the Z-N tuning rule. The present trend in tackling this problem appears to be the integration of various tuning methods in the form of software. Future trends include the inclusion of system identification techniques, such that the entire PID design process may be automated and be readily available for on-line adaptation. ACKNOWLEDGEMENT The first author would like to thank Universities UK and University of Glasgow for the sponsorship of his PhD. REFERENCES [] K.J. Åström and T. Hägglund, The Future of PID Control, in [6], pp. 9-30, [2] W.S. Levine (Ed, The Control Handbook, CRC Press/IEEE Press, 996. [3] Y. Li, W.Y. Feng, K.C. Tan, X.K. Zhu, X. Guan and K.H. Ang, PIDeasy TM and Automated Generation of Optimal PID Controllers, Proceedings of the 3 rd Asia-Pacific Conference on Control and Measurement, pp , Dunhuang, P.R. China, 998. [4] P. Marsh, Turn On, Tune In, New Electronics, 3(4), pp. 3-32, 998. [5] A. O Dwyer, PI and PID Controller Tuning Rules for Time Delay Processes: A Summary, Technical Report AOD-00-0 Edition, School of Control Systems and Electrical Engineering, Dublin Institute of Technology, Ireland, 5 May [6] J. Quevedo and T. Escobet (Ed, Digital Control: Past, Present and Future of PID Control, Proceedings of the IFAC Workshop, Terrassa, Spain, 5-7 April [7] J.G. Ziegler and N.B. Nichols, Optimum Settings for Automatic Controllers, Trans. ASME, vol. 64, pp , 942. [8] OPC Foundation, [9] [0] [] [2] [3] [4] [5] [6] [7] [8] [9] [20] [2] [22] [23] [24] [25] [26] [27] [28] [29] [30] [3] [32] [33] [34] [35] [36] [37] [38] [39] [40] [4] [42] d=8