Object-oriented modelling of virtual-labs for - e-Spacio
Contents
1. In 11 this case the virtual lab has been implemented combining the use of Sysquake and Modelica Dymola The approach discussed in Section 3 has been applied The heat exchanger model composed using the JARA library see Fig 1a has been translated automatically by Dymola into an executable file In addition a Sysquake application has been programmed It implements the virtual lab view and controls the execution of the Dymola executable file The features of this Sysquake application that constitutes virtual lab core include 1 the application to the heat exchanger model of several identification techniques and 2 the design of control strategies using the linear models previously obtained by applying the identification techniques The challenge is to control the gas exit temperature by manipulating the water flow The virtual lab supports the automatic calculation of the plant linearized model This calculation is performed as follows see Fig 2a 1 the change in the value of the gas exit temperature in response to a step in the water flow is calculated simulating the heat exchanger model and 2 a transfer function abbreviated TF is fitted to this response In this identification procedure the virtual lab user is allowed to 1 change the parameter values and the input variable values of the heat exchanger model the simulation communication interval and the total simulated time 2 choose among different ident2. x1 time s x1e Play Pause Heater Chill pata THO 20 x1ify Normalized Consum m Conversion er nTSt 13 nw BE es oo fe 5 os ur the chiller on when one of the following conditions are met Heater Chiller K a Automate Manual Secton 33 SS aoa Temp of the liquid moture is C time 3 x1 time 3 x1 Pump StateVars IsHeater IsChiller ReactionRate mol m 3 s The conversion factor is LigFlow m 3 s 0 00 19 2 MassA Kg 0001 MassP Kg 725 osl Fracta 0 1 an 1 urn the chiller off when liquid reachs the following temperature C IMassWater ks 3275_T c 045 g Temp 020 oO vA 4 6 8 o 2 4 6 8 time 3 x1 time 3 x1 c Fig 6 Virtual lab with runtime interactivity of a chemical reactor a Simulink model of the virtual lab b view c window menu to determine the policy of oper ation 26 lel x Eg Sysquake 3 controlReactor sq Untitled FA Fie Edit Settings Plots Figure Layout View Window Help 18 x Run my ESE Experiment settings intial conc papery Liquid volume SP Signal offset Pume seinge Heater gt Liquid volume SP Signal amplitude Chiller is turned on when either gt Liquid volume SP Signal start time Chiller is tumed off when the liquid temp is PID parameters Heat Exch area m 2 3 3 I Mass of A Ka 0 0 p p n Se s ak SS Inlet liquid flow m 3 s 0 4 R E p E E p y
3. 00 400 600 800 time s Man StateVars WaterM Kg 1823 WaterT K 481 r water a fon vapo vaporMol 606 g i a g av Tem K 300 gasTemp k 481 g Boilin D ca Valve Pressure Opening 0 7 F outputPress atm 01 K3 w Q 0 200 400 600 800 time s 200 400 600 800 time s Fig 4 View of the industrial boiler virtual lab 24 b om Rn Control loop Figure Layout View Window Help ej nl walale 3 Sysquake 3 Stationary band Stel Reak Elongation band Process plant Bode Magnitude nnn Frstorder 1 cS 10 ay 1001 a o x F wis 120 v Parametric 9 9 Orders and Delays Number of Samples 140 A a 5000 164 2000 o 2000 a4 100 Steam Power Boiler Param Input Variables Bode Phase 150 V m 3 3 0 I Liquid flow m 3 s 1 82e 4 I mol vapor 700 l lj Valve opening 0 7 1 temp Vapor Q 450 TD Downstream press Pa 1 20e5 mass Water Kg 1600 DownStream temp K 300 EEEE temp Water K 420 T inlet water temp 49 300 I 100 04 100 Fie Edit Plots Figure Layout Yiew Window Help x Run Slaa Sample Time eager Liquid Vol Simulation time Liquid flow manipulated variable Bode Magnitude TF Liquic Volume Liquid F
4. 3 Email address carla dia uned es Carla Martin pole zero diagrams and frequency response diagrams i e Bode and Nyquist diagrams User s actions on the model can be performed by manipulating different ele ments of the view such as buttons sliders check boxes and certain graphic elements of the model schematic diagram The graphical properties of the view elements are linked to the model variables producing a bidirectional flow of information between the view and the model Any change of a model variable value is automatically displayed by the view Reciprocally any user inter action with the view automatically modifies the value of the corresponding model variable 1 1 Types of interactivity User s actions on the model are performed according to certain rules Two approaches are discussed next runtime interactivity and batch interactivity Runtime interactivity allows the user to perform actions on the model at any time during the simulation run He can change the value of the model inputs parameters and state variables perceiving instantly how these changes affect to the model dynamic An arbitrary number of actions can be made on the model during a given simulation run In the case of the batch interactivity the user s action triggers the start of the simulation which is run to completion During the simulation run the user is not allowed to interact with the model Once the simulation run is finished
5. Object oriented modelling of virtual labs for education in chemical process control Carla Martin Alfonso Urquia Sebastian Dormido Dept Informatica y Automatica UNED Juan del Rosal 16 28040 Madrid Spain Abstract Easy Java Simulations Ejs and Sysquake are two software tools specifically in tended for implementation of virtual labs They allow easy definition of the virtual lab view i e the model to user interface However the model definition capabilities and the numerical solvers provided by these tools are not the state of the art On the other hand the use of the object oriented modelling language Modelica reduces considerably the modelling effort and permits better reuse of the mod els Modelica is supported by the state of the art simulation environment Dymola Nevertheless Modelica does not provide the interactive capabilities required for virtual lab implementation The approach proposed in this manuscript is to combine the best features of each tool Ejs and Sysquake capability for building interactive user interfaces composed of graphical elements whose properties are linked to the model variables Modelica capability for physical modelling and Dymola capability for simulating DAE hybrid models This novel approach has been successfully applied to set up virtual labs for control education Key words control education virtual laboratory interactive simulation chemical Preprint submitted to Elsevier S
6. Step Response Process plant Bode Magnitude 400 if a6 80 350 o I X x 60 S00 Num 89 51 1790 15 Den 1 14 045 02 u0 0 y0 399 86 o 10 20 30 Aceptar 01 100 Heat Exchanger Param Bode Phase 150 Pipe length 1 0 nCO2 molar fraction 0 50f i D1 0 0189 i T Gas flow temp 400 00 D2 0 0222 I i Total molar flow 0 16 f ij D3 0 0381 I T Liquid flow temp 291 00 100 01 400 Bode Magnitude Robustness Margins Lead Compersator taB 150 xe Gain margin dB int Critical freq radis inf 100 Phase margin deg 100 19 Cross over freq radis 18 67 332 50 o 100 200 01 100 Control signal Bode Phase Nyquist 1 aiiis 4 270 i oj 2 4 260 ol J o 100 200 01 100 2 o 2 Fig 2 Virtual lab with batch interactivity of the double pipe heat exchanger a plant linearization and b controller synthesis 22 Boiling process becoceccuceu Control module cntrl Control module Fig 3 Modelica model of an industrial boiler composed using JARA 23 mainFrame aaa V 3 00m 3 Heater log Heat 5 7 g3 Auto Pump Flow liter s 0 2 plotsContr 3 0 Volumes m 3 OutputFlow mol s 0 gt N water iInput flow 200 400 600 800 ime s InputFlow liter s vapor output flow 200 400 600 800 time s log Heat J 200 400 600 800 time s Temperatures K O 2
7. al variables against time including 1 the setpoints and the actual values of the controlled variables 2 the manipulated variables 3 the water and the vapor temperatures and the boiling temperature corresponding to the actual vapor pressure and 4 the vapor pressure inside the boiler and the valve downstream pressure 14 5 2 Virtual lab with batch interactivity This virtual lab is intended to illustrate the synthesis of the boiler control sys tem This control system is composed of two decoupled control loops 1 the water level inside the boiler is controlled by manipulating the pump through put and 2 the output flow of vapor is controlled by manipulating the heater power The synthesis procedure is similar to the one discussed in Section 4 2 It is briefly described next The user is allowed to choose interactively the plant s operation point This is accomplished by setting the value of 1 the mass and temperature of the liquid and the vapor inside the boiler 2 the valve opening and its downstream pressure and 3 the flow and inlet temperature of the water Once the operation point has been set the user can launch the calculation of the two TF 1 a TF from the pump throughput input to the water level output and 2 a TF from the heater power input to the vapor flow output These TF are automatically fitted to simulated step responses by the virtual lab The user can choose a
8. arrative and the interactive simulation as an Java applet Then the user can run the virtual lab and or publish it on the Internet In addition Ejs provides an interface to Matlab Simulink This feature allows the combined use of both tools for virtual lab implementation the description of the model using Matlab Simulink and the description of the narrative and the view using Ejs The data exchange between the virtual lab view composed using Ejs and the model i e the Simulink block diagram is accomplished through the Matlab workspace The properties of the Ejs view elements are linked to variables of the Matlab workspace which can be written and read from the Simulink block diagram Further details can be found at the Ejs web site On the other hand a Dymola to Simulink interface can be found in the Simulink s library browser the DymolaBlock block Dynasim 2002 A Mod elica model can be automatically embedded inside the DymolaBlock block which can be connected in the Simulink s workspace window to other Simulink blocks and also to other DymolaBlock blocks Simulink synchronizes the nu merical solution of the complete model performing the numerical integration of the DymolaBlock blocks together with the other blocks In order to embed a Modelica model inside a DymolaBlock block the com putational causality of the Modelica model interface needs to be explicitly set Dynasim 2002 The input variables are calc
9. cience 27 January 2006 process control object oriented modelling Modelica 1 Introduction Virtual labs are effective educational tools for training of process engineers and plant operators They are distributed environments of simulation and visualization tools intended to illustrate some relevant properties of a system Virtual labs are composed of 1 the interactive computer simulation of the system s mathematical model and 2 a narrative that provides information about the system and the use of the virtual lab Interactive computer simulations provide a flexible and user friendly method to define the experiments to be performed on the mathematical model In teractive simulations allow the user to design and perform his own simulation experiments As a result the user becomes an active player in his own learning process which motivates him to learn Typically virtual lab programming includes the definition of the mathematical model and the virtual lab view The virtual lab view is the user to model in terface It is intended to provide a visual representation of the model dynamic behavior and to facilitate the user s interactive actions on the model The model behavior can be represented in different ways For instance plotting the model variables against each other and by means of animated schematic diagrams of the system In addition linear systems can be described using Corresponding author Tel 34 91398825
10. ea Den E naa aoe i SSS Temp of the mixture K 293 M Vase D S Concentration in vol of water 0 6 I Chiller temp K 288 ij Mass of A Water P Normalized consum Conversion 300r y r r 7 F r r r r Hester hiller 200 0s 500 100 o ot 1 A A A J OL A A A f A A A o 5000 0 5000 0 5000 Fig 7 Virtual lab with batch interactivity of a chemical reactor 2
11. ed using LME an interpreter for numerical computation which is mostly compatible with Matlab In order to allow the combined use of Sysquake and Modelica Dymola a set of LME functions have been programmed Martin et al 2005a These functions can be used from any Sysquake application They facilitate the bi directional communication between the Sysquake application and the executable file gen erated by Dymola for the Modelica model Further details can be found in Martin et al 2005a 4 Case study I control of a double pipe heat exchanger JARA library has been used to compose the interactive model of a double pipe heat exchanger JARA was originally written in Dymola language Urquia 2000 Urquia amp Dormido 2003 Later on it was translated into Modelica language The design methodology proposed in Martin et al 2004b Martin et al 2005b was applied in order to make JARA suitable for batch and runtime interactive simulation JARA contains models of some fundamental physical chemical principles in cluding 1 mass energy and momentum balances applied to ideal mixtures of semi perfect gases and homogeneous liquid mixtures 2 mass transport due to pressure and concentration gradients 3 heat transport by conduction and convection 4 chemical reactions and 5 liquid vapor phase change JARA s main application is the modelling of physical chemical processes in the context of automatic control Further info
12. ification methods including first order TF with delay second order TF with delay and non parametric identifica tion 3 analyze the obtained TF by means of Bode and zero pole diagrams and robustness margins 4 start the simulation run and 5 export the cal culated TF to another Sysquake application In addition the virtual lab automates the controller synthesis and analysis 12 The virtual lab supports the following user s operations see Fig 2b 1 to import the TF previously identified 2 to analyze the TF characteristics using Nyquist Nichols and Bode diagrams 3 to choose the controller type possible options are PID lead and lag compensators 4 to synthesize the controller i e to set the value of the PID s parameters and to specify the error and the phase margin of the system controlled by the lead or lag compensators 5 to simulate the closed loop linear and non linear models 5 Case study II control of an industrial boiler JARA library has been used to compose the interactive model of an indus trial boiler It is based on the mathematical model of the process provided in Ramirez 1989 The model diagram is shown in Fig 3 it has been rep resented using Dymola The input of liquid water is located at the boiler bottom and the vapor output valve is placed at the boiler top The water contained inside the boiler is continually heated The model is composed of two control vo
13. ke supports built in functions to include in the view different types of interactive plots and interactive graphic elements i e radio buttons sliders dialog boxes etc However the model definition capabilities and the numerical solvers provided by these tools are not the state of the art They support the block diagram modelling This modelling paradigm requires of explicit state models i e ordinary differential equations and the computational causality of the model must be explicitly set i e the blocks have a unidirectional data flow from inputs to outputs These restrictions do not facilitate the model reuse and they strongly condition the modelling task which requires a considerable effort For instance dummy dynamics need to be introduced in the model to avoid the establishment of systems of simultaneous equations The model programmer has to manipu late the model to transform its equations to the form of ordinary differential equations ODE As a consequence the modelling and simulation capabilities supported by these tools are not the best possible ones for describing the large models used in the physical chemical field 1 3 Contributions of this paper The physical modelling paradigm supported by the object oriented modelling languages is an attractive alternative to the block diagram modelling Astr6m et al 1998 Object oriented modelling languages support a declarative de scription of the model ba
14. low Bode Magnitude TF Vapor flow Heat SP Liquid Volume m 3 gt o0 18 dB dB SP Vapor flow moljs gt jl fi l i PID Liquid volume Controller id Volume PID paramete Vapor flow Controller r F v Lead Compensator 1w q vapor flow Error Constant ud vapor flow Phase Margin Lag Compensator sor flow Margin 50 50 18 ie o o 200 o 200 400 01 100 01 100 Vapor flow controlled variable Heat manipulated variable Bode Phase TF Liquid Volume KLiquid Flow Bode Phase TF Vapor flow Heat o 0 ol 8l 1e7 ii 50 d 100 R D OE 6 100 o 20 so o 200 400 01 10 01 10 Fig 5 Virtual lab to illustrate the control of an industrial boiler 25 IparamC4 Iparam_signal P Isi D lvar_sig _signal 0 ig Istate_s SetPointC m CKpara CKvar CKparamC1 _signal 12 CKstate_signal CKtsC1 SystemBlock a P CKvar F Ckstate n Pause Switch1 Clock Time to Workspace Pause Simulink 8 jo a aiai E x x18 Masses Kg 120_ Liquid Temp C 3 100 4 amp aol 2 2 60l eat ae xi o 7 ao eee Oa ae Level Controller time s x1 time 3 x1 g SourceMassFlow kg s SourceTemp C SPiveselHeight 0 80 l PID Parameters Za Z 5100 g 6 T E 2 Ee T Heater V 5 00m 3 IE 2 ott um the heater off when liquid reachs the following temp C o ws 4 6 8 o 2 4 6 8 T time 3
15. lumes in which the mass and energy balances are formulated 1 a control volume containing the liquid water stored in the boiler and 2 a control volume containing the generated vapor The model of the boiling process connects both control volumes The heat flow from the heater to the water the pressure at the valve output and the water pump are modeled using JARA source models 13 5 1 Virtual lab with runtime interactivity A virtual lab with runtime interactivity has been implemented applying the methodology discussed in Section 2 The objective of this virtual lab is to illustrate the boiler closed loop dynamic for two different control strategies manual control and decentralized PID The water level inside the boiler and the output flow of vapor are the controlled variables The pump throughput and the heater power are the manipulated variables The view of the virtual lab is shown in Fig 4 At any time during the simula tion run the virtual lab supports interactive change of 1 the control mode i e manual or PID 2 the parameters of the PID controllers 3 the mass and temperature of the water inside the boiler 4 the mass and temperature of the vapor inside the boiler 5 the total inner volume of the boiler 6 the inlet temperature of the water and its input flow rate and 7 the vapor valve opening and the pressure downstream The virtual lab view contains an animated diagram of the boiler and plots of sever
16. mong the following identification methods see Fig 5a first order TF with delay second order TF with delay and non parametric identification The virtual lab supports a set of graphical methods to analyze the fitted TF including Bode and pole zero diagrams and it automatically computes the robustness margin In addition the virtual lab allows to export the TF to any other Sysquake application Finally the virtual lab facilitates the design and analysis of the two controllers 15 see Fig 5b The water level inside the boiler is controlled using a PID The gas flow can be controlled using a PID a lead or a lag compensator The user can change the controller parameters and the error and phase margin specifications of the compensation networks 6 Case study III control of a batch chemical reactor The model of a batch chemical reactor has been composed using JARA library An exothermic reaction A P is carried out in the liquid phase The reactor contains a heat exchanger which can be operated with steam and with cooling water The reactor s operation policy is the following Froment amp Bischoff 1979 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 adia batically the heat exchanger is controlled by another PID 3 start cooling when either the maximum allowable reaction temperature occurs o
17. o Workspace blocks transmit the value of the output variables from the Simulink block diagram window to the Matlab workspace Ejs reads the value of these output variables from the Matlab workspace and writes the value of the input variables in the Matlab workspace The view of the virtual lab programmed using Ejs is shown in Fig lc The main window of the virtual lab view on the left side of Fig 1c contains 1 an animated diagram of the heat exchanger which displays the tempera tures of the gas the liquid and the wall by means of a color code 2 buttons to pause reset and play the simulation run 3 sliders to modify the liquid and the gas flow rates and their inlet temperatures 4 a text field to set the molar fraction of carbon dioxide in the gas mixture and 5 checkboxes to show and hide the following three secondary windows Geometry Parameters Modify State and Characteristics The Geometry Parameters window contains text fields that can be used to modify the pipe length and diameters The controls placed in the Modify State window allow changing the temperature of the medium inside each control volume Finally Characteristics window displays several plots of the model variables 4 2 Virtual lab with batch interactivity A second virtual lab has been developed Its objective is to illustrate the appli cation to the heat exchanger of some linearization and control techniques
18. of virtual labs intended for control education References Astr m K J Elmqvist H amp Mattsson S E 1998 Evolution of continuous time modelling and simulation Proceedings of the 12 European Simulation Multiconference 9 Calerga Sarl 2004 Sysquake 3 User s manual Calerga Sarl Lausanne Switzerland Cutlip M B amp Shacham M 1999 Problem solving in chemical engineering with numerical methods Prentice Hall 18 Dynasim AB 2002 Dymola User s manual Version 5 0a Dynasim AB Lund Sweden Esquembre F 2004 Easy Java Simulations a software tool to create scien tific simulations in Java Computer Physics Communications 156 199 Froment G F amp Bischoff K B 1979 Chemical reactor analysis and design New York John Wiley amp Sons Johansson M Gafvert M amp Astr m K J 1998 Interactive tools for ed ucation in automatic control IEEE Control Systems Magazine 18 3 33 Martin C Urquia A amp Dormido S 2004a JARA 2i A Modelica li brary for interactive simulation of physical chemical processes Proceedings of European Simulation and Modelling Conference 128 Martin C Urquia A Sanchez J Dormido S Esquembre F Guzman J L amp Berenguel M 2004b Interactive simulation of object oriented hybrid models by combined use of Ejs Matlab Simulink and Model ica Dymola Proceedings of 18 European Simulation Multiconference 210 Martin C Ur
19. quia A amp Dormido S 2005a Object oriented modelling of interactive virtual laboratories with Modelica Proceedings of 4 Interna tional Modelica Conference 159 Martin C Urquia A amp Dormido S 2005b Object oriented modeling of virtual laboratories for control education Proceedings of 16 IFAC World Congress paper code Th A22 TO 2 Ramirez W F 1989 Computational methods for process simulation Boston Butterworths Publishers Urquia A 2000 Modelado orientado a objetos y simulaci n de sistemas h bridos en el mbito del control de procesos qu micos Ph D Thesis UNED Urquia A amp Dormido S 2003 Object oriented design of reusable model libraries of hybrid dynamic systems Mathematical and Computer Modelling 19 of Dynamical Systems 9 1 65 Wittenmark B Haglund H amp Johansson M 1998 Dynamic pictures and interactive learning IEEE Control Systems Magazine 18 3 26 20 i Er as mer a j m co os ooo ak ak 4 Ai ap jheatexchmodel Bele E c Fie Edt View Simulation Format Tool
20. r the de sired conversion is reached and cool down to the desired temperature and 4 empty the reactor 6 1 Virtual lab with runtime interactivity The methodology discussed in Section 2 has been applied The reactor and the controllers have been modelled using Modelica language and they have been embedded within Simulinks DymolaBlock blocks SystemBlock and PIDBlock respectively see Fig 6a 16 The virtual lab view is shown in Fig 6b The main window on the left side 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 corresponds with the operation policy previously described The value of the PID controller parameters the temperatures defining the operation policy and the desired conversion can be changed interactively in the virtual lab s window shown in Fig 6c At any time during the simulation run the user is allowed to change inter actively the value of 1 the model state variables i e the temperature and total mass of the reaction mixture and the concentration of A and P 2 the model parameters i e the reactor volume and section the area of the heat exchanger and the physical chemical data of the steam and cooling water and 3 the input variables i e the inlet temperature and concentration The secondar
21. rmation about JARA library can be found in Urquia amp Dormido 2003 Martin et al 2004a The JARA model of the heat exchanger is based on the physical model de scribed in Cutlip amp Shacham 1999 A gaseous mixture of carbon dioxide and sulfur dioxide in the tube is cooled by water in the shell The JARA model allows two modes of operation cocurrent and countercurrent flow The temperature dependence with the spatial coordinate has been modeled by di viding in control volumes the flow paths of the water and the gas and the pipe wall This approach allows for local variations in physical properties and heat transfer coefficients The diagram of the JARA model is shown in Fig la it has been represented using Dymola 4 1 Virtual lab with runtime interactivity The objective of this virtual lab is to illustrate the dynamic behavior of the open loop plant It has been implemented by combining the use of Model ica Dymola Matlab Simulink and Ejs The procedure described in Section 2 has been applied The Simulink model of the heat exchanger is shown in Fig 1b The Modelica model of the plant whose diagram is shown in Fig la has been embedded within the DymolaBlock block see Fig 1b The blocks connected to the DymolaBlock inputs MATLAB Fen blocks transmit the value of the input 10 variables from the Matlab workspace to the Simulink block diagram window The blocks connected to the DymolaBlock outputs T
22. rol systems and for model analysis Modelica capability for physical modelling and finally Dymola capability for simulating hybrid DAE models This software combi nation approach is discussed for implementation of virtual labs with runtime interactivity Section 2 and with batch interactivity Section 3 This novel approach has been applied to the implementation of a set of virtual labs Their topic is the open loop dynamic and the control of three process units widely used in the chemical industry a double pipe heat exchanger Sec tion 4 an industrial boiler Section 5 and a batch chemical reactor Section 6 The objective of these virtual labs is to help the students to 1 understand the behavior of the plant non linear models 2 apply some linearization tech niques 3 analyze the plant linearized models using zero pole Bode and Nyquist diagrams and 4 design the PID lead and lag compensators re quired to control the processes 2 Implementation of virtual labs with runtime interactivity Ejs is an open source Java based software tool intended to implement virtual labs with runtime interactivity Esquembre 2004 It can be freely downloaded from the Ejs web site http fem um es Ejs Ejs guides the user in the process of creating the narrative the model and the view of the virtual lab It generates automatically 1 the interactive simulation as a Java application and 2 HTML pages containing the n
23. s T SSN g ols tsa sre 2ec r fan H Jwh Jay ayh Jwh dan Han H 1 Iparam tel tel tel tel Hel fe tel te ie ua 2 frre param sional E E EE E aE E oh vo CKparam ma E AE AE O IE AE AE r on Ckvar CK ssignal a CKstate_signal Bi de eet ee ee ee dol aaa D ClockTime to Woispace 5 x Ready 100 i i TE A a b mainframe OO E E Characteristics xl 60 Liquid temperatures Gas temperatures mu 50 Input flow 0 17 40 1 E kaa g KAUHEAA AA EEA AEN 30 50 J 20 Pipe length 1 te 2 timo 0 50 100 150 0 50 100 150 Outer diam pipet lo0o22 time s time s Inner diam pipe2 0 038 Pumps temperatures Wali temperatures i L 100 100 c 0 aD 150 0 50 150 time s time s 10 Erguin Faw 03 pas plow os oal J a aas 3 J 3 X oal 202 J P 02 oi __Pause Geometry Para _ Modify state F 00 00 Play 1 l J i i 0 50 100 0 50 100 150 Vi Show diagrams Molar fraction CO2 0 600 ESS J com time s time s Fig 1 Double pipe heat exchanger a Diagram of the Modelica model composed using JARA library b Diagram of the Simulink model c Virtual lab view 21 b z s Plots E a aaa gt le ofa identificationHEBoton sq Untitled alale BS Sysquake 3 controlHE3Boton sq Untitled PID parameters
24. sed on equations instead of assignment statements The computational causality is not included in the model Thus a model can adapt to more than one data flow context The modelling knowledge is rep resented as differential algebraic and discrete equations that may change by being triggered by events i e hybrid DAE models Some examples of object oriented modelling languages are EcosimPro gPROMS and Modelica The Modelica language http www modelica org has been designed by the developers of the object oriented modelling languages Allan Dymola NMF ObjectMath Omola SIDOPS Smile and a number of mod elling practitioners in different fields It is intended to serve as a standard format for the external model representation so that models arising in differ ent engineering fields can be exchanged between tools and users Modelica lan guage is supported by the Dymola modelling environment http www dynasim se The use of the Modelica language reduces considerably the modelling effort and permits better reuse of the models However Modelica does not provide the interactive capabilities required for virtual labs implementation The approach proposed in this manuscript is to combine the best features of each tool Ejs and Sysquake capability for building interactive user interfaces composed of graphical elements whose properties are linked to the model vari ables Matlab Simulink capability for modelling of automatic cont
25. the results are displayed and a new user s action on the model is allowed 1 2 Virtual lab implementation software Several virtual lab packages conceived to illustrate some selected topics in automatic control have been implemented For instance ICTools and CCS DEMO Johansson et al 1998 Wittenmark et al 1998 are two packages developed at the Dept of Automatic Control Lund Institute of Technol ogy In both cases the virtual labs are implemented using Matlab Simulink http www mathworks com a general purpose simulation environment sup porting the graphical block diagram modelling As Matlab Simulink is not spe cially suitable for interactive simulation the development of new modules re quires a considerable effort In addition there are software tools specifically intended for implementation of virtual labs These tools 1 provide their own procedures to define the narrative the model and the view of the virtual lab 2 guide the virtual lab programmer in these tasks and 3 automatically generate the virtual lab ex ecutable code Easy Java Simulations http fem um es Ejs and Sysquake http www calerga com are two of these tools A strong point of these tools is that they allow easy definition of the virtual lab view Easy Java Simulations hereafter cited as Ejs provides a complete set of interactive graphic elements which are ready to be used in a simple drag and drop way to compose the view Sysqua
26. ulated from other Simulink blocks while the output variables are calculated from the Modelica model In addition the Modelica model needs to support the discontinuous changes in the value of its state variables parameters and input variables which are the result of the user s interaction In some cases several choices of the state variables need to be supported simultaneously in the model in order to provide the user with alternative ways of describing the state changes A design methodology has been developed in order to implement Modelica models fulfilling all these requirements In particular this methodology pro vides a systematic procedure to adapt any existing Modelica library to a for mulation suitable for runtime interactive simulation All the details of this design methodology can be found in Martin et al 2004b Martin et al 2005b 3 Implementation of virtual labs with batch interactivity Sysquake is a commercial tool intended to develop virtual labs with batch in teractivity Calerga 2004 Typically a Sysquake application contains several interactive graphics which are displayed simultaneously These graphics con tain elements that can be manipulated using the mouse While one of these elements is being manipulated the other graphics are automatically updated to reflect this change The content represented by each graphic and its depen dence with respect to the content of the other graphics is programm
27. y windows located on the right side of the view see Fig 6b contain plots displaying the time evolution of the relevant process variables 6 2 Virtual lab with batch interactivity The virtual lab view is shown in Fig 7 It contains sliders to change the model parameters the initial value of the state variables and the input variables The Settings menu allows the user to see Fig 7 1 change the parameters of the control policy 2 set the communication interval and the total simulated time and 3 launch a simulation run The view contains plots displaying the time evolution of some process variables including the mass of A P and 17 water the mixture temperature and the pump throughput 7 Conclusions The feasibility of combining Modelica Dymola with Ejs and Sysquake for im plementing virtual labs with runtime and batch interactivity respectively has been demonstrated Ejs and Sysquake are software tools specifically intended to develop virtual labs Their strong point is the programming of the virtual lab view The use of Modelica language reduces considerably the modelling effort and facilitates the model reuse In order to implement this software combination approach 1 a novel mod elling methodology adequate for interactive simulation of Modelica models has been proposed and 2 a Sysquake to Dymosim interface has been pro grammed This approach has been successfully applied to the implementation
Download Pdf Manuals
Related Search