OBJECT-ORIENTED MODELING OF VIRTUAL LABORATORIES
Contents
1. lox E lt x x x1 Masses Kg 10p__Liquid Temp C a Level Controller a 3 e SPiveselHeight 0 80 4 a0 PID Parameters B 2 3 eo ee 2 5 ao 1 a Heater CeCi OLATA ETE TU Turn tne neater off when liquid reachs the following temp C g time s xf time s x1 19 SourceMassFlow kg s SourceTemp C Chiller e A um the chiller on when one of the following conditions are met 3 5 G 3 Temp of the liquid mixture is C 3 4 3 50 a The conversion factor is V 5 00m 3 S2 9 urn the chiller off when liquid reachs the following temperature C Ohh PaCS ip hae A time s xf time s xP Pause Heater DialogPlots E3 C 15 THCC 120 j i E l epa jigy Normalized Consum a Conversion Parameters StartConversi MISt 1390 __RnW 1180 A A Heater Chiller oo Automatic Manual i Section r ee oy nae DA TA Laue time 3 x1 time 3 x1 Pump StateVars IsHeater IsChiller ReactionRate mol m 3 s LigFlow m 3is 0 00 1 0 2 MassA Kg 0001 MassP kg 725 e iS Fracta 0 1 o aA ea Masswater kg 3275 T C 045 Temp 020 Go Wh 2 ee lel a 2 4 BG l time s xf time s x1 a c Fig 4 a Ejs view of the virtual lab b Window menu to determine the policy of operation c Simulink model of the virtua2. uses the value of this output array i e O to refresh the control lab view Interactive parameters The interactive para meters are defined in physicalModel as con stant state variables i e with zero time derivative The interactive changes in the value of these parameters are implemented by re initializing their values The state re initialization is performed using when clauses and a built in Modelica operator reinit x expr It re initializes an state vari able x with the value obtained of evaluating an expression expr at an event instant Input variables The model symbolic manip ulations performed by Dymola to formulate the model according to a given state selection can require differentiating an input variable Martin et al 2004b In this case an error is produced Dymola cannot differentiate an input variable A valid approach consists in defining the input variables in physicalModel as constant state variables The changes in the value of these variables are implemented by re initializing their values In conclusion an analogous solution is ap plied to the input variables and the inter active parameters The zero time derivatives are included in the physicalModel model A new model is defined setParamVar It in herits physicalModel and contains the when clauses and reinit operators to change the value of the interactive parameters and input variables Four input variables are used to model the change
3. within Simulink s DymolaBlock blocks The views of the virtual labs have been programmed using Ejs A wa The modeling methodology proposed in Task 1 is discussed in this manuscript Its application to the implementation of the virtual labs is illus trated by means of the following two case studies 1 the control of an industrial boiler and 2 the control of a batch chemical reactor 2 MODELING FOR INTERACTIVE SIMULATION USING MODELICA A modeling methodology adequate for interac tive simulation using Ejs Matlab Simulink and Dymola is proposed It takes advantage of the modeling and simulation capabilities of Modelica and Dymola The common characteristics of the models intended for interactive simulation are dis cussed in this section Next the proposed mod eling methodology is briefly described Further details can be found in Martin et al 20046 2 1 Characteristics of the interactive models The proposed methodology states how a Modelica model can be formulated to suit interactive sim ulation using Ejs Matlab Simulink and Dymola The obtained Modelica model fulfills the following requirements Martin et al 20046 1 Computational causality of the interface In order to embed the Modelica model within a Simulink block the computational causality of the Modelica model interface needs to be explicitly set In other words interface variables needs to be classified into inputs and outputs Interactiv
4. OBJECT ORIENTED MODELING OF VIRTUAL LABORATORIES FOR CONTROL EDUCATION Carla Martin Alfonso Urquia Sebastian Dormido Dept Inform tica y Autom tica E T S de Ingenier a Inform tica U N E D Juan del Rosal 16 28040 Madrid Spain carla aurquia sdormido dia uned es Abstract The combined use of Ejs Matlab Simulink and Dymola with Modelica language has been successfully applied to set up virtual labs for control education The tasks completed to achieve this goal are discussed in this manuscript 1 the development of a novel modeling methodology adequate for interactive simulation 2 the object oriented design and programming of JARA a Modelica library of dynamic hybrid models of some fundamental physical chemical principles 3 the description of the JARA physical models in a way suitable for interactive simulation and finally 4 the implementation of the virtual labs This is illustrated by means of two case studies Copyright 2005 IFAC Keywords Software tools Education Automatic control Interactive approaches Object modelling techniques PID control Reactor control modeling Steam generators 1 INTRODUCTION Virtual laboratories supporting interactive simu lations are effective pedagogical resources for con trol education Dormido 2004 During the in teractive simulation run the students can change the value of the model inputs parameters and state variables perceiving instantly how these ch
5. anges affect to the model dynamic Interaction possibilities enhance the students understanding The students take an active role in the learning process and this promotes their motivation to study the subject Virtual laboratories can be implemented by com bining the use of three software tools Ejs Mat lab Simulink and Dymola with Modelica lan guage This approach takes advantage of the best features of each tool Martin et al 20046 Ejs capability for building interactive user interfaces composed of graphical elements whose properties are linked to the model variables Esquembre 2004 Matlab Simulink s capability for model ing automatic control systems and for model analysis Modelica s capability for physical mod eling http www modelica org and finally Dy mola s capability for simulating hybrid DAE mod els Dynasim 2002 Next the fundamentals of Ejs and its use together with Matlab Simulink and Dymola are briefly discussed 1 1 Fundamentals of Ejs Ejs is a open source Java based tool intended to program web based virtual labs It can be freely downloaded from http fem um es Ejs Ejs guides the user during the definition of the virtual lab and it automatically performs all the tasks required to generate the virtual lab code a Java application or an applet Ejs is based on a simplification of the model view control paradigm The virtual lab defin ition is structured in the followi
6. e changes on the model state As a result of the user interaction the interactive model needs to support instantaneous dis continuous changes in the value of the state variables In general different choices of the model state variables are possible Therefore several choices of the state variables need to be simultaneously supported by the interac tive model in order to provide alternative ways of describing the state changes N Ww F W o a dt 7 h F a J2gh V Ah F F kv Fig 1 Model of a process 3 Changes on the interactive parameters Time independent properties of the system are usually represented by model parameters However sometimes one of the interactive simulation goals is studying the dependence between the model dynamic behavior and the value of these properties In this case the property value can be instantaneously changed by the user s action remaining con stant between consecutive interactive changes These variables of the model are called in teractive parameters Changes in the value of the interactive parameters can have different effects depending on the state variable selec tion As a consequence interactive models need to support interactive changes in the value of the interactive parameters for dif ferent choices of the state variables Example 1 The model shown in Fig 1 will be used to illustrate these requirements Martin et al 2004b The voltage applied to the pum
7. ean 3 isState tank tank1 hIsState VIsState pipe pipe1 FIsState isState 1 isState 2 isState 3 end physicalModel The Boolean vector isState controls the state selection see Fig 2 a The output variable array O in Fig 2 contains the variables representing the actual value of the state in addition to the other variables linked OL oi StateSelection1 physicalModel x extends setParamVar isState Iparant isState Ol lvar CKparam CKvar Istate CKstate when change CKstate then a reinit xs1 Istate nl m1 end when SCLRALAMV AL moder StateSelectionN extends setParamVar Iparam Sarie ee ana j isState R aati l ee pe A Iparan l wal i lvar i CKparam CKparam Ae eee SE ek CKvar when change CKparam then CKvar reinit p Iparam Istate end when gt when change CKvar then CKstate when change CKstate reinit v Ivars reinit xsN Istate nN mN end when end when b c Virtual lab view Ejs interactiveModel Modelica StateSelection1 i Istate CKstate StateSelectionN extends Fig 2 Schematic description of the modeling methodology to the view elements Ejs
8. in et al 2004a The mathematical model of E mainFrame source process Heat flow source Volumes m3 q45_ OutputFlow mol s a Sa ee ae Ree vapor output flow water on SB o o 0 20 40 60 80 time s InputFlow liter s 20 40 60 80 time s log Heat J 20 10 0 1 21 Nk ow water input flow V 5 66m 3 H d 0 20 40 S 80 oO 20 40 oF 80 q time s time s g Pause Heater Dari Loara wv ieee A g KONER 55 Temperatures K 25 Pressure atm Liquid ot DET C Param Auto Man fu Pits i control Q Liquid StateVars o 20 cntrl solum i ad source Pump ee ee WaterT k 446 3 TI tassa ice Flowtliteris 0 4 vaporMol 1272 245 a a ul i i K E 1 Control module cntrl Control module Tem K 450 gasTemp ts 463 Band i A 7 5 Valve Pressure 35 0 Opening 1 0 a b outputPress atm 11 o 20 40 60 80 Oo 20 40 60 80 time s time s Fig 3 a Boiler model composed using JARA 2i b View of the virtual lab implemented using Ejs the boiler is found in Ramirez 1989 The input of liquid water is placed at the boiler bottom and the vapor output valve is placed at the top The water contained in the boiler is continually heated The diagram of the boiler model composed using JARA 2i is shown in Fig 3a Two control volumes are considered 1 a control vo
9. l lab mixture and the concentration of A and P the model parameters i e the reactor volume and section the area of the heat exchanger and the physicochemical data of the steam and cooling water and the input variables i e the inflow temperature and concentration can be changed interactively during the simulation run The sec ondary windows on the right side of Fig 4a con tain plots showing the evolution of some relevant process variables 5 CONCLUSIONS A novel modeling methodology adequate for in teractive simulation has been proposed It allows easy creation of the virtual labs by combining the use of Ejs Matlab Simulink and Dymola with Modelica language This approach has been suc cessfully applied to the implementation of the JARA 2i library The use of Modelica language has reduced considerably the modeling effort and it has permitted better reuse of the models The use of JARA 2i Ejs and Matlab Simulink to de velop virtual labs for control education has been illustrated by means of two case studies ACKNOWLEDGEMENTS Part of this work has been supported by the Span ish CICYT under DPI2001 1012 and DPI2004 01804 grants REFERENCES Dormido S 2004 Control learning Present and future Annual Reviews in Control 28 115 136 Dynasim 2002 Dymola User s Manual Version 5 0a Dynasim AB Lund Sweden Esquembre F 2004 Easy Java Simulations a software tool to create scientific simula
10. lume containing the liquid water stored in the boiler and 2 a gaseous control volume containing the vapor The vapor volume is equal to the difference between the boiler recipient inner volume and the water volume The boiling is a transport phenomena represented by a model connecting both control volumes The heat flow into de boiler the pres sure at the valve output and the water pump are modeled using JARA source models The Ejs view of the boiler virtual lab is shown in Fig 3b The plant has been modeled using JARA 2i Martin et al 2004a The user can interac tively choose between two control strategies man ual and decentralized PID The control system has been modeled using Modelica a PID is used to control the water level and another PID is used to control the vapor flow The manipulated variables are the pump water flow and the heater heat flow respectively The parameters of these PID con trollers can be changed interactively In addition the value of the model state variables mass and temperature of the water and the vapor parame ters inner volume of boiler and input variables temperature of the input water valve opening and output pressure can be changed interactively during the simulation run 4 CASE STUDY II CONTROL OF A BATCH CHEMICAL REACTOR The model of the batch reactor has been adapted from Froment and Bischoff 1979 In a batch reactor having a volume V an exothermic reac tion A P is ca
11. nerated by Dymola for the Modelica model Dynasim 2002 DymolaBlock block can be connected to other Simulink blocks and also to other DymolaBlock blocks 1 2 Contributions of this paper This software combination approach has been successfully used to program a set of virtual labs for chemical process control To achieve this goal the following four tasks have been completed 1 Proposal of a modeling methodology intended for interactive simulation Martin et al 2004b This methodology states how a Mo delica model can be formulated to suit inter active simulation Object oriented design and programming of JARA Urquia 2000 Urquia and Dormido 2003 JARA is a library of dynamic hy brid models of some fundamental physical chemical principles Its main application is the modeling of physical chemical processes in the context of automatic control Re formulation of JARA physical models in a way suitable for interactive simula N Near w we tion Martin et al 2004a The modeling methodology proposed in Task 1 has been applied The JARA library version that is written in Modelica language and intended for interactive simulation is JARA 2i Implementation of the virtual labs The phys ical models of the controlled chemical plants have been composed using JARA 2i The controllers have also been modelled using Modelica language Plant and controller mod els have been translated using Dymola and embedded
12. ng three parts 1 introduction html pages including educational content related with the virtual lab topic 2 model dynamic model whose interactive simula tion is the virtual lab basis and 3 view user to model interface The view is intended to provide a visual rep resentation of the model dynamic behavior and to facilitate the user s interactive actions on the model Ejs includes a set of ready to use visual elements that allows easy creation of the virtual lab view The graphical properties of the Ejs view elements can be linked to the model variables producing a bi directional flow of information be tween the view and the model Any change of a model variable value is automatically displayed by the view Reciprocally any user interaction with the view automatically modifies the value of the corresponding model variable Ejs virtual labs can run 1 stand alone 2 in conjunction with Matlab Simulink and 3 in conjunction with Matlab Simulink and with Dy mola In the first case Ejs gives the user a pro cedure to define the model and provides a set of built in ODE solvers to simulate it In the second case the virtual lab model can be partially or completely described using Matlab code and Si mulink block diagrams In the last case Modelica models can be embedded within a Simulink block the DymolaBlock block This block can be found in the Simulink s library browser It is an inter face to the C code ge
13. p v is an input variable The cross sections of the tank A and the outlet hole a the pump pa rameter k and the gravitational acceleration g are time independent properties of the physical system The liquid volume V and level h and the liquid flows F and Fin are time dependent properties Possible choices of the model state variables in clude e1 h e2 V e3 F where e represents one particular choice of the state vari ables If the user wants to change interactively the level value h the appropriate choice is e1 h Likewise if the user wants to change V then the right choice is e2 and if he wants to change F then e3 The model should support the following feature every time the user needs to change the state value the user decides to represent it in terms of a change in either the volume or the height or the flow Different choices are possible during a given interactive simulation run Changes in the value of interactive parameters can have different effects depending on the state variable choice For instance consider an instan taneous change in the tank cross section A If the state variable is the liquid volume V then the change in A produces an instantaneous change in the value of the liquid height h and flow F while the liquid volume remains constant On the contrary if the state variable is the height or the flow these magnitude values do not change as the result of an in
14. rried out in the liquid phase The reactor contains a heat exchanger and it can be operated with steam and with cooling water The Simulink model of the reactor and its control system is shown in Fig 4c The Modelica models of the plant composed using JARA 2i and the controllers are embedded within the SystemBlock and PIDBlock blocks respectively The virtual lab view is shown in Fig 4a The main window on the left side of Fig 4a contains the schematic diagram of the process above and the control buttons below Both of them allow the user to experiment with the model The user can interactively choose between manual and automatic control The automatic control is intended to perform the following operation policy see Fig 4b 1 Fill up the reactor with the reacting liquid The inflow is controlled by a PID 2 Preheat to certain temperature and let the reaction proceed adiabatically The heat ex changer is controlled by another PID Start cooling when either the maximum al lowable reaction temperature occurs or the desired conversion is reached and cool down to the desired temperature 4 Empty the reactor 3 The value of the PID controller parameters the temperatures defining the operation policy and the desired conversion can be changed interac tively Also the value of the model state variables i e the temperature and mass of the reaction
15. s in the value of the in teractive parameters and input variables see Fig 2 b two signal arrays Iparam Ivar containing the new values and two signals CKparam and CKvar for triggering the re initialization events State variable choices As many models state Selection1 stateSelection2 are defined as different state variable choices are needed e1 2 Each of these models inherits setParam Var isState array is set to the value adequate in each case and contains a when clause and a reinit operator to change the value of the corresponding state variable ar ray Two input variables are used to model the interactive changes in the state Istate and CKstate see Fig 2 c The array Istate contains the values used to reinitialize the model state The signal C Kstate is used to trigger the state re initialization event Interactive model The interactive model is defined interactiveModel model in Fig 2 d and embedded within a Simulink s Dymo laBlock block It is composed of all the mod els defined in the previous step i e state Selectionl stateSelectionN The value of the input array Enabled 1 N is set by Ejs see Fig 2 d and it selects which output is connected to the output signal O 3 CASE STUDY I CONTROL OF AN INDUSTRIAL BOILER The interactive simulation of an industrial boiler has been implemented by the combined use of Ejs Matlab Simulink and Modelica Dymola Mart
16. stantaneous change in A In this case the volume does change 2 2 Modeling methodology The proposed modeling methodology for interac tive simulation consists in the following steps 1 Physical modeling Object oriented modeling of the system using Modelica language For explanation purposes lets suppose that this model is called physicalModel 2 State selection control Modelica supports the user s control on the state variables selection via the stateSelect attribute of Real variables Otter and Olsson 2002 This attribute val ues include never the variable will never be selected as state variable and always the variable will always be used as a state This Modelica feature allows controlling the model state selection by means of a Boolean array The following example tries to illustrate it Example 2 State selection of the model in Fig 1 can be accomplished as shown below by means of the Boolean vector isState For instance if isState is set to the value false true false when instantiating the physical model then the volume V is selected as a state variable model tank Real h stateSelect if hIsState then StateSelect always else StateSelect never Real V stateSelect if VIsState then StateSelect always else StateSelect never end tank model pipe Real F stateSelect if FIsState then StateSelect always else StateSelect never end pipe partial model physicalModel parameter Bool
17. tions in Java Computer Physics Communications 156 199 204 Froment G F and K B Bischoff 1979 Chemical reactor analysis and design John Wiley amp Sons New York Martin C A Urquia and S Dormido 2004a JARA 2i A Modelica library for interactive simulation of physical chemical processes In Proc European Simulation and Modelling Conference pp 128 132 Martin C A Urquia J Sanchez S Dormido and F Esquembre 2004b Interactive sim ulation of object oriented hybrid models by combined use of Ejs Matlab Simulink and Modelica Dymola In Proc 18th European Simulation Multiconference pp 210 215 Otter M and H Olsson 2002 New features in Modelica 2 0 In Proc 2 International Modelica Conference pp 7 1 7 12 Ramirez W F 1989 Computational Methods for Process Simulation Butterworths Pub lishers Boston Urquia A 2000 Modelado Orientado a Obje tos y Simulaci n de Sistemas H bridos en el Ambito del Control de Procesos Quimicos PhD thesis Dept Informatica y Automatica UNED Madrid Spain Urquia A and S Dormido 2003 Object oriented design of reusable model libraries of hybrid dynamic systems Mathematical and Computer Modelling of Dynamical Systems 9 1 65 118
Download Pdf Manuals
Related Search