User`s Manual – Hybrid Controller - nees@berkeley
Contents
1. Figure 7 Main panel for MTS software Structural Test System STS nees berkeley User s Manual Page 20 In addition to the MTS controller the Simulink model PCSimulation mdl should be opened Figure 8 shows a modification of this model renamed PCSimulation hybrid mdl which has been modified to conduct a hybrid simulation The root model contains two blocks the controller block which is an S function that connects to the STS software in simulation mode and the model block which contains models for the actuators and hydraulic supply Similar to a real setup the controller sends valve commands to the actuator models and receives feedbacks representing the response of the actuators The user should become familiar with the model subsystem particularly the actuator and specimen models Figure 9 shows the model subsystem and the implementation of a hybrid simulation algorithms using the force feedback from the actuator models and generating the displacement commands for the servo hydraulic controller 7 PCSimulation hybrid E Oj xl File Edit View Simulation Format Tools Help mfr Nomi FI scones function 1 controller valvecmd fesdback UC B PC Simulation Note Execute initislize m before running simulation Ready 100 FixedStepDiscrete Figure 8 Root box diagram of Simulink model PCSimulation_hybrid mdl nees berkeley User s Manual Page 21 PCSimulati2. nees O berkeley User s Manual Page ii LIST OF TABLES Table 1 Training specimen specifications Error Bookmark not defined Table 2 Test matrix of the grout test sess Error Bookmark not defined Table 3 Estimation of the coefficients of friction from the grout tests Error Bookmark not defined nees O berkeley User s Manual Page iii EXECUTIVE SUMMARY This user s manual of nees berkeley is To be written when manual is complete nees Oberkelev User s Manual Page iv Chapter 1 INTRODUCTION 1 1 Mission and Definition of nees berkeley The mission of nees berkeley is to provide a leading equipment site specializing in earthquake response simulation of large scale structural and non structural systems through real time integration of computer models and physical test specimens in a reaction wall facility The nees berkeley facility is an integral part of George E Brown Jr Network for Earthquake Engineering Simulation NEES established by the US National Science Foundation The role of nees berkeley is to ensure the availability of the state of the art technology in large scale earthquake hybrid simulations involving computer modeling and physical testing and innovative application to the field of earthquake engineering The nees berkeley equipment site supports large scale simulation by e Leading the development of the hybrid simulation methods to enable o High
3. MANUAL PREPARED BV GILBERTO MOSQUEDA POST DOCTORAL RESEARCHER STRUCTURAL ENGINEERING MECHANICS AND MATERIALS DEPARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING UNIVER UNIVERSITY OF CALIFORNIA BERKELEY September 23 2004 TABLE OF CONTENTS LIST OF ILLUSTRATIONS 1 5 sssovovsorsavens M II DIST OF TABLES FL rase eo ep et Pa ELT a S ENSEM soeesasveusesnessesdos NNUS rs RUE BEER USUS S RUD FREE RE VER UNSERES parata Ba sa i TRAE m EXECUT VE SUMMARY oss Es Skosat SeSe IV CHAPTER 1 INTRODUCTION 6 ssssicsssssesscsioseoieasssnsacesesesivessensacsioseassasaesiseesesessscaessscscdesiebioossesesese sine iesebsed aievieasaessaeessese e 1 1 1 MISSION AND DEFINITION OF NEES BERKELEY eee e e emen enne ness eset ese ese ese e ese ese etes sees sees eee 1 19 GIAI UOIUAUTT aa 1 1 3 INFRASTRUCTURE OF NEES 9 BERKELEY cccccccccececececececeeeeeeecececeeeseeececeeeseceaeeeaususuaeaeaeeuaeauauauauauauauauauauanananaea 2 1 4 NON NEES INFRASTRUCTURE AVAILABLE FOR NEES 9 BERKELEY ccccccccccccccccsccesccccesecescscscsesssssescessssseseseeees 3 1 5 LOCATION AND CONTACT INFORMATION ccccccccccscecececececeeeeececececeeeeeceeeeececeeucucesusueeseseeeaeauauauauauauauauauauauaeanaes 4 1 6 MANUAL GAY OUT B Abu dec ee de Blues eee he B B derde Alba seats kJ 4 CHAPTER
4. generation of the actuator commands The separation of these two tasks into different processors provides an expandable distributed architecture for simultaneous testing of multiple substructures as show in Figure 3 Moreover increased processing time can be dedicated to the integrator task for applications with large numerical structural models In a local testing configuration the network is replaced by a shared memory bus Systrans 14 to maintain fast communication rates for real time continuous algorithms In the case of geographically distributed testing Ethernet replaces the network link As will be demonstrated in the discussion of the experimental results network communication time is random and therefore not suitable for real time algorithms A solution based on a finite state event driven controller design is discussed next nees berkeley User s Manual Page 8 2 3 Event Driven Simulation In cases where task execution times are random a clock based control scheme could fail if the required processes are not completed within the allotted time As an improved alternative to the clock based scheme used for real time applications an event driven reactive svstem based on the concept of finite state machines Harel 15 is proposed that responds to events based the state of the hvbrid simulation svstem The event driven svstem can be programmed to account for the complexitv and randomness of real svstems and thus take acti
5. Host and Hybrid Controller Host are both connected to their real time counterparts via Ethernet The data acquisition host is connected through GPIB utilizing a fiber optic cable that provides increased throughput The three real time computers are connected through scramnet forming a closed circular loop between the three nodes as indicated in Figure 5 Both the MTS servo controller and the PI data acquisition system consist of specialized hardware design to execute a particular task However both of these systems contain user adjustable settings and tunable parameters that can be optimized for a specific applications via nees berkeley User s Manual Page 13 the host The 5 servo controller s main task is to provide the closed loop force or displacement control of the actuators using an enhanced PID tvpe algorithm Basicallv the servo controller accepts a displacement or force command and generates the proper command signal for the servo valve that attempts to move the actuator to the commanded position The PI data acquisition hardware provides the signal conditioning and calibration capabilities to obtain measured data from analog sensors which can be digitized into engineering units and saved to a file The onlv user programmable component in this svstem is the xPC real time target Of particular importance numerical algorithms for conducting hybrid simulations can be programmed to run in the xPC environment in real time xPC has
6. STS with setting uNEES set setting file If there is a problem with running the STS properly it is advised to reset the MTS Hybrid Controller RST button in the instrumentation room 2 set an oscilloscope to show a Scramnet command and displacement force feedback at particular actuator or Actuator 1 in this example 3 in the main STS panel the program source should be selected as Scramnet and simulation mode should be enabled Figure 10 nees berkeley User s Manual Page 23 Structural Test System Settings uNEES set E File Calibration Configuration Operation View Service HPSOn Q interlocks Enabled Reset Master gain 100 HSM Nel L gt Program cramet 2 Source m Desired span 100 0 EE CC V Enable simulati 3 Current span 0 0 Allow Scramnet modeswitch _HSM Lo Bypass controllers Stop Run Pause KWOTI HSM Hi Q Podready B Scramnet OK Figure 10 Main panel for STS in RTSimulation mode On xPC Host machine 1 open RTSimulation mdl Simulink model Figure 11 compile it and connect it User can use play and pause buttons in the Simulink model s menu or enter tg start and tg stop from the Matlab prompt 2 Push plav button or type tg start in the Matlab prompt 3 Make sure that the flags for displacement control are assigned to zeros they should be equal to one for a force control mode 4 Change value in Displ bl
7. command for the actuators at a rate of 1024Hz nees berkeley User s Manual Page 26 the rate of the MTS controller using polvnomial approximations The continuous test is scaled in time by selecting the duration of the integration step time step updated in initialize xpc m The actuator target displacement commands are updated and the step number is incremented in the xPC in each step XPC reacts bv generating the command signal for the MTS controller and returns the measured values at the target command Please examine the Simulink model to verifv the channels used in the MTS controller Steps to run the simulation In matlab switch to the directory D MTSmodel ucbSTS Simulink event_model Run initialize xpc m This defines some variables for the Simulink model Build and download the Simulink model FastHybrid_eventmdl mdl to xPC Start actuators make sure disp scram cmd is set to 0 Start the real time program for example tg start Pe e c pu m Start the numerical analysis by running localsim 2dofshear You should see the real time response computation in the host computer showing the global displacements and the story shear vs story drift for both stories The xPC plots the generated displacement command signals and the state of the controller 0 extrapolate 1 interpolate 2 slow 3 hold Especially if you do other things on the host PC you should note that the slow and hold states are activated b
8. memory on one computer is copied to the other nodes on the network In order to avoid conflicts between the computers using shared memory a memory map has been defined to coordinate the reading and writing of data from xPC MTS and PI Table 1 lists the reserved partitions in the scramnet memory The MTS controller reads and writes to the first memory partition of size 1024 words or 4KB A word is defined as 4 bytes or 32 bits This includes data written by the MTS controller and data written by xPC for use by nees berkeley User s Manual Page 15 the MTS controller The following 1024 words of memorv are reserved for the PI heartbeat internal clock or counter and current sensor readings The next 1024 words are also reserved for PI for the storage of calibration constants for the sensor channels This calibration data is only updated on scramnet when requested by the user after a calibration parameter has been changed The remainder of the scramnet memory is available to the user for example to transfer data between other computers that may be added to the scramnet ring Table 1 Reserved memory partitions in shared memory resource Partition Resource 1 1024 words MTS Data 1025 2048 words PI heartbeat and real time sensor readings 2049 3072 words PI calibration constants User defined Remaining memory The memory map for MTS consisting of the first 1024 words has been coded into the controller and cannot be
9. size 544 tvpe int16 PI reserves word length 32 bits memory slot for each channel but since the A D converters are 16 bit the first 16 bits are empty followed by the 16 bits of actual data Further PI reserves 8 memory locations per card this is the maximum number of channels per card in the system There are 32 cards 2 Racks with 16 cards each with 4 channels 128 Channels and there are two cards with 8 channels 16 channels of analog input only Therefore 544 intl6 variables are read by xPC where every other integer is blank data and for the 4 channel cards the last four channels are blank For example to access the counts for channel Pacific 65 1 0 0 which is located in RACK 1 CARD 0 CHANNEL 0 we need to skip RACK 0 which holds 16 cards That is 16 Cards 8 Channels card 2 integers Channel 256 Integers Therefore integers 257 and 258 are reserved for Pacific 65 but the 16 bit integer data is stored in 258 257 is blank The Simulink model Pacific scram mdl together with InitialzePI m demonstrates how to read a value written by PI from scramnet and covert to engineering units nees berkeley User s Manual Page 18 Chapter 4 EXAMPLE APPLICATIONS Five examples applications of the hybrid simulation controller are presented to demonstrate the use of the testing system First the contents of the directory containing the main software is reviewed followed by the demonstrations The first example s
10. speed testing o Advanced and robust computational tools o Hybrid substructures i e computational and physical substructures o Geographically distributed hybrid simulation e Implementing data metadata capabilities of NEES grid e Simulating tests in a virtual mode prior to physical testing e Allowing versatile capabilities for specimen and setup configurations e Integrating a NEES Equipment Site to the number one rated Civil Engineering education program The accessibility of nees berkeley and its proximity to several major research universities national research laboratories and leading earthquake engineering professional community provide an excellent environment for efficient collaboration with other researchers Moreover nees berkeley provides excellent ancillary research infrastructure including office space computational facilities and the 100 000 volume Earthquake Engineering Library 1 2 Objective This user s manual is aimed to familiarize the interested user with the hybrid simulation controller provided by nees berkeley It is assumed that an external user of the nees berkeley facility will only program hybrid simulation algorithms using the Matlab xPC environment and operating their program during the test Thus this manual focuses on using Simulink and xPC nees berkeley User s Manual Page 1 along with the MTS controller and Pacific Data Acquisition for the implementation of hybrid simulation algorithms I
11. to deal with random delavs since the hold period can be arbitrarilv long However the ramp hold procedure introduces a number of other errors In order to maintain the benefits of continuous testing an event driven procedure is proposed for conducting continuous tests over a network that minimizes if not eliminates the hold phase at each integration step The distributed hardware architecture utilizing event driven controllers is also presented 2 1 Hybrid Simulation Test Method The equipment used for quasi static testing in most structural testing facilities can also be utilized to conduct hybrid tests The basic components of a pseudodynamic test setup and their interconnections are illustrated in block diagram form in Figure 1 The required tools are 1 a servo hydraulic system consisting of a controller servo valve actuator and pressurized hydraulic oil supply 2 a test specimen with the actuators attached at the point where the displacement degrees of freedom are to be imposed 3 instrumentation to measure the response of the test specimen and 4 an on line computer capable of computing a command signal based on feedback from the transducers The primary task of the on line computer is to integrate the equation of motion utilizing the restoring force vector r which is composed of forces from experimental and numerical substructures A time stepping integration procedure is used to solve the discretized equation of motion for disp
12. two modes 1 it can communicate with the MTS VME console that runs in real time using TCP IP or 2 it can communicate with a windows model of the controller running on the same computer The settings file set determines the mode of operating If there is a settings set file in the same directory as STS exe then by default this file is used to start STS Otherwise the user is prompted for a settings file Table 3 below shows the first few lines of a set file which determines if STS runs with hardware or runs in windows simulation mode In addition the Enable Simulation option must be checked in the STS main panel for simulation of the actuators as shown in Figure 7 Table 3 STS settings for use with hardware or PCSimulation mode File set for using STS with hardware File set for using STS with PCSimulation MySystem MySystem podDriver pod dil podDriver coopsDriver Super dll coopsDriver SuperSim dll userPassword userPassword Structural Test System Settings gil set File Calibration Configuration Operation View Service HPS On interlocks Enabled Reset D Master gain 100 HSM Nel Program Function Generator ri Function Generator el Desir d span 100 0 Enable simulation 3 Current span 0 0 Allow Scramnet modeswitch HBM Qu Bypass controllers Stop Pun Paice Peer HSM Hi Podready B Scramnet OK
13. 2 TEN e OD PE dede 5 2 1 HYBRID SIMULATION TEST METHOD ns Aineen n n n n n nnn nnn nnn nnn nnn n nnn nnn nea n nu np pp nara nanna nanna 6 2 121 CONTINUOUS TESTING i e eret esee ertet ka eden 7 2 2 DISTRIBUTED HARDWARE 8 2 3 BVENT DRIVEN SIMULATION eee ie dee ere red pe eene 9 CHAPTER 3 DESCRIPTION OF HARDWARE eeee ee ee ee ee ene eetn oeste tnos eee tone eese ene e eaae sese to Pese eene etta aee e toes era 12 3 1 HARDWARE COMPONENTS dore eee E ee o ee ee e A Ro he aient eoe e y dere e p Neue eee 12 3 2 sHARDWARE ARCHITECTURE si eere eee eo exeo eroe eu ave tee eee aee eese eve eee ee ene Ve ge e eee e e E ege uve ew derden 13 3 32 SCRAMNET MEMOR Y MAP issa roce denderen id o ec eso eco ee A ec oto o deense A e eto ee o 15 REMAINING MEMORY 16 3 4 SCRAMENT ACCESS FROM XPC sese erster eie 16 CHAPTER 4 EXAMPLE APPLICATIONS scctessssvcstesssussssgetesssesces sbesessvsesescsvsestestveteessevstecsssvents seuvecesssduedsesseucdanbedesodesebedses 19 4 CONTENESCOF FOLDER UCBSES 2 i B e a decsendes 19 42 IPESIMULATION MODE i nete bt tee vete tee de vo ether i eie 20 4 3 FAST HYBRID SIMULATION EET 25 4 4 SLOW CONTINUOUS HYBRID SIMULATION nne
14. access to data from both the MTS controller and the PI data acquisition system through scramnet Further xPC can generate actuator commands for the MTS controller The scramnet memory map and access to data on scramnet from xPC for use in numerical algorithms is discussed in more detail in the sections that follow To improve the performance of the hybrid testing system xPC and the hybrid controller can be synchronized to insure that xPC updates the command in time for each control cycle This is accomplished by having the Simulink model interrupt source set to the MTS clock rather than its own clock The MTS models for fast hybrid testing presented later are set trigger from the MTS clock Due to this synchronization the Simulink model must runs at a base rate of 1024 Hz nees berkeley User s Manual Page 14 Control room xPC Host PC Ethernet Hybrid Controller Host PC xPC Real time target Data Acquisition Host PC am ee ii KI Fiber MTS Pacific Instruments Hybrid Controller Data Aquisition Instrumentation room To Actuators To sensors HSM pod Figure 5 Illustration of computers and networks that form the hybrid simulation testing system 3 3 Scramnet Memory Map Scramnet provides 2MB of shared memory between computers in its network Each computer contains a 2MB memory module that is mirrored to all other computers on the network Data written to local
15. acquisition host PC with GPIB Dell Precison 630 Workstation with Windows XP runs the graphical user interface to manage the data acquisition hardware including the logging of data Two software programs are used for this purpose PI660 is the main software for typical operation and Panel60 serves more as a debugging tool The active sensor channels are defined and calibrated through the PI660 interface Data saved to disk using PI660 is stored in the local hard drive Data acquisition hardware with scramnet Pacific Instruments rack mount hardware contains signal conditioners a scramnet card and a DSP dedicated to data acquisition Logged data is collected and time stamped by the local DSP before being sent to the host PC for writing to disk Sensor data is written to scramnet memory locations in units of counts The calibration information to convert counts to engineering units in only known by the host although the calibration factors can be written to a separate memory location in scramnet after calibration is completed It should be noted that the Pacific Instruments hardware can write to the scramnet memory but cannot read from it 3 2 Hardware Architecture The physical location of the hardware listed above and the interconnections between these machines is illustrated in Figure 5 The overall setup consists of three real time computers with scramnet cards and three hosts that serve to interface with each of the real time computers The xPC
16. actuator with 351 kN 300 mm 79 kip 12 in nees berkeley User s Manual Page 3 36 One actuator with 338 kN 300 mm 76 kip 12 in 37 Two actuators with 285 kN 300 mm 64 kip 12 in 38 One actuator with 222 kN 300 mm 50 kip 12 in 39 Two actuators with 156 kN 2540 mm 35 kip 100 in 40 Four actuators with 129 kN 100 mm 29 kip 4 in 41 Two actuators with 89 kN 300 mm 20 kip 12 in 42 Two actuators with 53 kN 300 mm 12 kip 12 in 43 Reciprocating dvnamic shaker capable of developing 22 kN 5 kip of inertia force up to 10 Hz lt Check the above list and see if there is opposition to include the non nees equipment gt 1 5 Location and Contact Information Nees berkeley is located in Building 484 at the Richmond Field Station of the University of California Berkeley Address Richmond Field Station Building 484 Mailing address 1301 S 46th Street RFS 451 Richmond CA 94804 Phone 510 231 9527 Fax 510 231 9471 Web site http nees berkeley edu 1 6 Manual Layout A short background describing the hardware architecture for hybrid testing including capabilities for geographically distributed substructures are provided in the background section in Chapter 2 Chapter 3 describes the individual hardware components that compose the hybrid controller and their role during a simulation After presenting the equipment Chapter 4 presents several example applications in detail
17. ast Hybrid SC150 init typeConv11x uint32 typeConv12 dig inps SC150 unpac u a Simulink model FastHybrid mdl b Blocks to acces input from scramnet Figure 6 Simulink model template for accessing scramnet memory to MTS controller The sensor readings collected by PI are also written to the scramnet memory and can be read by the xPC Target Use of this data requires programming for two reasons 1 due to the volume of data available from PI 128 16 Channels xPC should be set to read only the nees berkeley User s Manual Page 17 necessarv channels to reduce overhead and 2 the PI data written to scramnet is in units of counts from the A D converters and needs to be converted to engineering units The code below demonstrates how to read the PI heartbeat and 128416 data channels into partitions 25 and 26 of the scrament memory map respectively following 23 partitions defined for MTS SSTART PACIFIC INSTRUMENTS PARTITIONS sblank space partition 24 Type partition 24 Size uint32 866 Pacific Instruments heartbeat Memory Address 0X1000 partition 25 Type int32 partition 25 Size Muss Wo gd sPacific Instruments 128 channels 16 partition 26 Type intle partition 26 Size 1544 First a blank partition 24 is defined to complete the 1024 words reserved for MTS Starting at word 1025 OX1000 a single partition of variable tvpe int32 is defined for the heartbeat followed by a partition of vector
18. ations nees Oberkelev User s Manual Page 28 nees Oberkelev User s Manual Page 29 Chapter 5 CLOSING REMARKS In addition to the hybrid controller FlexTest GT is also available at nees berkeley for quasi static and dynamic load controlled tests or for using a standard pre programmed hybrid simulation algorithm Please refer to the FlexTest GT manual for further information on use of this system nees berkeley User s Manual Page 30 Chapter 6 REFERENCES 10 11 Takanashi K Udagawa K Seki M Okada T and Tanaka H Non linear earthquake response analysis of structures by a computer actuator on line system details of the system English translation of paper in Transactions of the Architectural Institute of Japan March 1975 229 77 83 Takanashi K and Nakashima M Japanese activities on on line testing Journal of Engineering Mechanics 1987 113 7 1014 1032 Mahin S A Shing P B Thewalt C R and Hanson R D Pseudodynamic test method Current status and future direction Journal of Structural Engineering 1989 115 8 2113 2128 Shing P B Nakashima M and Bursi O S Application of pseudodynamic test method to structural research Earthquake Spectra 1996 12 1 29 54 Magonette G E and Negro P Verification of the pseudodynamic test method European Earthquake Engineering 1998 XII 1 40 50 Magonette G Development and application of large sca
19. e input from scrament block Figure 6b demonstrates how a nees berkeley User s Manual Page 16 continuous partition is input to the Simulink program then upacked into the separate partitions The partitions are defined in the Matlab workspace bv running initialize m to create the memory partition structure node Each partition is defined by specifying the variable type and vector size for example master span baseAddress 0 partition 1 Address Ox dec2hex baseAddress 4 partition 1 single partition 1 Size X control modes partition 2 Type uinb32 partition 2 Size num2str nAct The base address is defined for the first partition the following partitions fill the memory immediately following the previous partition More detailed information on defining scrament memory partitions for Simulink can be found in scramnet doc including the creation of the node data structure after the partitions have been defined input from scramnet output to seamnet R a3 tel master span O double ctl modes ENE typeConv displ cmds force cmds REN displ fbks ELI mwa force fbks e cour eee ofS valve cmds EM NE typeConv5 terms unpack m user ducs V 3 Lc SC150 read typeConv7 term10 user encs zi dig mee PT E typeConv typeConvS mil double UC B F
20. ecause of delays in Matlab 4 6 Fast MOST In order to run a simulation using NEESgrid software including NTCP special software needs to be installed and the operating system configured To verify settings all computers involved both for local and remote simulation must be enabled to run the MOST simulation see http www neesgrid org software neesgrid2 2 doc php This verifies that the NTCP plug in is correctly installed in Matlab There are two options to run Fast MOST 1 using the original Matlab based simulation coordinator and computational model or 2 using a Java based simulation coordinator and computational model The demonstration below makes use of the Java version because it runs faster by multi threading the tasks with the experimental sites The Matlab version can be ran by opening two more Matlab sessions below and running SimCoordinatorFast m and NCSA Comp Site m one in each of the additional Matlab windows The configurations for the simulation are set in most config m for Matlab based programs and in C neesgrid 2 2 matlab JavaSimulationCoordinator input config xml Make sure nees Oberkelev User s Manual Page 27 the IP numbers for each site are set to the proper NTCP clients used in this simulation The instructions below are specific to using the nees berkeley NEESPOP and running one experimental site using the Hybrid Controller hardware in the laboratory The numerical simulation and the remainder of the e
21. eedback will remain constant 8 User should try changing the controller gain Operation gt controllers to see how the actuators response changes to the command signal In order to run the model using the hybrid simulation the program source needs to be changed to Scramnet In this setting the controller will look for the actuator commands from the Simulink model In addition the Run button must be pressed to set the span to 100 Upon playing PCSimulation_hybrid mdl the command for actuator 1 should begin to change Users wishing to run a PC simulation on their own computer can do so by simply copying the USBSTS folder to their computer and running the PC simulation as indicated above Apart from Matlab Simulink no additional software or hardware is necessary 4 3 RTSimulation Mode RTSimulation mode provides an option to conduct a real time simulation with xPC xPC Host and MTS Host involved From xPC Host a displacement command or a force command can be sent through Scramnet to the MTS Host computer A Simulink model is provided to perform real time reading and writing in and from Scramnet A windows based application STS simulates the behavior of the actuators and the controller on MTS Host The user can check communication between the xPC xPC Host and MTS Host Two programs are used to run RT Simulation STS software configures as noted below and the Simulink RTSimulation mdl or a modification of this model On MTS Host machine 1 run
22. for video recording of testing and tele presence In addition to the physical facility nees berkeley utilizes staff with extensive experience in earthquake engineering experimental research including previous experience in conducting research for off site researchers 1 4 Non NEES Infrastructure Available for nees berkeley 20 21 22 23 24 25 26 2 28 29 30 31 32 33 34 35 In addition to the nees berkeley equipment mentioned above the following equipment can also be made available to researchers using nees berkeley Two 300 kHz 15 bit A D converters with 144 channels of signal conditioning Three 100 kHz 12 bit A D converters with 48 channels of signal conditioning Five sets of 16 channels each 12 bit resolution ISA BUS data acquisition cards without signal conditioning 424 channels 16 bit resolution GPIB BUS with signal conditioning A total of 35 static actuators with the following specifications Two actuators with 6672 610 mm 1500 kip 24 in Two actuators with 4190 kN 250 mm 942 kip 10 in Two actuators with 2046 kN 510 mm 460 kip 20 in Two actuators with 2046 kN 250 mm 460 kip 10 in One actuator with 1334 kN 300 mm 300 kip 12 in Four actuators with 667 KN 910 mm 150 kip 36 in Four actuators with 556 910 mm 125 kip 36 in Two actuators with 556 KN 610 mm 125 kip 24 in One actuator with 543 kN 510 mm 122 kip 20 in One
23. highlighting the steps involved in conducting a hybrid simulation nees berkeley User s Manual Page 4 Chapter 2 BACKGROUND Hybrid simulation is a method intended to evaluate the seismic performance of structures The principles of the hybrid simulation test method are rooted in the pseudodynamic testing method developed over the past 30 years Takanashi et al 1 Takanashi and Nakashima 2 Mahin et al 3 Shing et al 4 Magonette and Negro 5 In a hybrid simulation the dynamic equation of motion is applied to a hybrid model which includes both numerical and experimental substructures Typically the experimental substructures are portions of the structure that are difficult to model numerically thus their response is measured in a laboratory Numerical substructures represent structural components with predictable behavior they are modeled using a computer Hybrid simulation procedures have advanced considerably since the method was first developed Early tests utilized a ramp hold loading procedure on the experimental elements Recently developed techniques along with advancements in computers and testing hardware have improved this test method through continuous tests at slow Magonette 6 and fast rates Nakashima 7 The potential of the hybrid simulation test method has been further extended by proposing to geographically distribute experimental substructures within a network of laboratories then link them thro
24. hows the use of the controller in PC Simulation Mode which is a useful standalone tool for training and pre test simulations PC Simulation uses a Simulink model to represent the behavior of the actuators and specimen and a Windows based controller model thus no hardware is necessary The three following examples demonstrate the use of the controller in fast hybrid mode which is used for a real test There is a third option for simulation labeled real time mode in which the controller hardware is exercised but actuator and specimen models are used instead of the physical actuators The real time mode is not described here since its use is similar to the two models presented 4 1 Contents of Folder ucbSTS MTS related software is found within the ucbSTS folder Both the Hybrid controller host C ucbSTS and the xPC Host D MTSmodel ucbSTS contain copies of this folder The files referred to in this manual should be available in the xPC Host Folders and files that should be of interest to the user are listed in Table 2 Table 2 List of files in ucbSTS folder File or Folder name Description Api Matlab based api that can be used to change parameters settings and other STS commands See apiExample m Documents Contains files that describe the use and setup of the hybrid controller including the scramnet memory map Matlab Matlab m files for importing MTS binary data files bin into the Matlab workspace Simuli
25. lacement d velocity v and acceleration a at time intervals t iAt for i21 to N t Cv r f The subscript i denotes the time dependant variables at time t At is the integration time step and N is the number of integration steps The mass matrix M damping matrix C and applied loading f are typically modeled as part of the numerical simulation Numerical methods used to solve the equation of motion are discussed in Mahin and Shing 12 The same methods are extended to hybrid simulation nees Oberkelev User s Manual Page 6 d actual imposed displacement hydraulic supply p d command displacement d E L d m measured displacement i D A ae ae 55 servo valve rm measured restoring force pe SIENA B actuator Controller on line computer ix zt servo hydraulic system da AID dm specimen transducers A D 1 experimental substructure Figure 1 Test setup for hybrid simulation 2 1 1 Continuous Testing Applying a continuous load history rather than a ramp hold load history improves the measured behavior of the experimental substructure Magonette 6 The improvements are largely based on the elimination of the hold phase and the associated force relaxation in the experimental specimens Continuous testing methods require a real time platform to ensure the commands f
26. le continuous pseudo dynamic testing techniques Philosophical Transactions of the Royal Society Mathematical Physical and Engineering Sciences 2001 359 1771 1799 Nakashima M Development potential and limitations of real time online pseudo dynamic testing Philosophical Transactions of the Royal Society Mathematical Physical and Engineering Sciences 2001 359 1851 1867 Campbell S and Stojadinovic B A system for simultaneous pseudodynamic testing of multiple substructures Proceedings Sixth U S National Conference on Earthquake Engineering June 1998 Watanabe E Kitada T Kunitomo S and Nagata K Parallel pseudo dynamic seismic loading test on elevated bridge system through the Internet The Eight East Asia Pacific Conference on Structural Engineering and Construction Singapore December 2001 Tsai K C Yeh C C Yang Y S Wang K J Wang S J and Chen P C Seismic Hazard Mitigation Internet based hybrid testing framework and examples International Colloquium on Natural Hazard Mitigation Methods and Applications France May 2003 MOST Multi site On line Simulation Test NEESgrid 2003 http www neesgrid org most nees berkeley User s Manual Page 31 12 Mahin S A and Shing P B Pseudodynamic method for seismic testing Journal of Structural Engineering 1985 111 7 1482 1503 13 Nakashima M and Masaoka N Real time on line test for MDOF systems Ea
27. ly Two dynamic actuators with force capacity 667 kN 150 kip static 556 kN 125 kip 9510 mm s 20 in s stroke capacity 510 mm 20 in Two dynamic actuators with force capacity 979 kN 220 kip static 623 kN 140 kip 9510 mm s 20 in s stroke capacity 1020 mm 40 in Three static actuators with force capacity 1460 kN 328 kip compression 960 kN 216 kip tension stroke capacity 1830 mm 72 in A high performance MTS digital control system capable of operating up to 8 dynamic actuators simultaneously A fiber optic network linking the MTS controller and a number of local personal computers enabling high speed hybrid simulations A new MTS FlexTest system and a xPC digital control system both capable of carrying out hybrid simulations nees Oberkelev User s Manual Page 2 13 14 15 16 17 18 19 A new 128 24 channel data acquisition system A wide variety of transducers Capabilities for telepresence such as high resolution still photography NSTC and DV video cameras and teleconferencing A personal robot avatar for off site users to traverse the laboratory to view progress and talk to laboratory staff students and others on site Capabilities to computationally simulate the seismic response of test specimens and develop computational models of substructures for hybrid simulations using the Open System for Earthquake Engineering Simulation http opensees berkeley edu Modern video equipment
28. modified by the user The corresponding memory map used by xPC has been specified by MTS as a Matlab M file initialize m and should not be modified to insure compatibility with the controller The format in which the PI data is written to scramnet is also pre defined However the location in memory to which the PI data is written can be specified through PI660 software Specific details on the MTS and PI memory maps are included in Appendix A 3 4 Scrament access from xPC Nees berkeley users need to access data from xPC or other PC s on the scramnet network for use in their algorithms Computers can be added to the scramnet network for additional tasks such as OpenSEES simulations or data streaming to the NEESPOP These computers also need to be programmed to read from the specified locations in the shared memory However only access from the xPC is discussed here To tools to access scramnet memory associated with the MTS controller from the xPC are well defined For a hybrid simulation the Simulink model FastHybrid mdl provides the inputs and outputs to the controller through scrament which form the inputs and outputs necessary to interact with an experimental specimen in the Simulink structural simulation The Simulink template FastHybrid mdl shown in Figure 6 shows the Simulink blocks that access data from scramnet in real time Figure 6a shows the complete model listing all the inputs and output and Figure 6b shows the details of th
29. n ennen en nennen en tetett tett etete rte neven ennen eneen enen ennen 26 AS IRAST MOST AEAEE EE E TIOTA TENE EEEE EE EN ETE EEE DE ETETE EEE OTETA 27 CHAPTER 5 APPENDIX A SCRAMNET MEMORY MAP ERROR BOOKMARK NOT DEFINED CHAPTER 6 APPENDIX B CONTROLLER SIMULATION MODES nae eee eo eee en toes e etta see seen seen 34 nees Oberkelev User s Manual Pagei LIST OF ILLUSTRATIONS Figure 1 Test setup for hybrid simulation eenn enne 7 Figure 2 Ramp hold and continuous load histories enn anne ev onnvenneeeenneenneerenseenseerenseenseerenseenseerenseenseeeenveenseeee 7 Figure 3 Distributed hardware architecture for geographically distributed testing eenen ereen 9 Figure 4 Event driven scheme using a polynomial predictor corrector to continuously generate actuator commands n B A 10 Figure 5 Illustration of computers and networks that form the hybrid simulation testing system 15 Figure 6 Simulink model template for accessing scramnet memory to MTS controller naeve 17 Figure 7 Main panel for MTS software Structural Test System STS emmen nnennennnnzznennnznnnznnzznnnznnzznnzna 20 Figure 8 Root box diagram of Simulink model PCSimulation hybrid mdl 21 Figure 9 Block model of PCSimulation hybrid mdl essent 22
30. nk Contains templates for Simulink models STS exe Main STS control software set Setting file for STS software contains stored gains and calibrations for a setup File is automatically generated by STS by saving settings Simulink PCSimulation mdl Simulink model template for running a simulation without hardware Simulink RTSimulation mdl Simulink model template for running simulation with main controller in the loop runs in real time but models for the actuators and test setup Simulink FastHybrid mdl Simulink model template for running a hybrid simulation with real controller and actuator Simulink Initialize m Contains model input parameters for above Simulink models including actuators specimens and scramnet memory map nees Oberkelev User s Manual Page 19 4 2 PCSimulation Mode PCSimulation mode provides an option to simulate a hvbrid test using a single computer with no specialized hardware Simulink models are provided to simulate the behavior of the actuators and a windows based application simulates the controller The user needs to incorporate specimen models and hybrid simulation algorithms into Simulink and combine with the actuator models to run a customized simulation Two programs are used to run PC Simulation STS software configures as noted below and the Simulink PCSimulation mdl or a modification of this model The STS interface to the MTS hybrid controller can operate in
31. o the MTS servo controller The software STS can be used to calibrate and tune instrumentation servo valves and actuators prior to a test During a test the Hydraulic Service Manifolds HSM are switched on and off from this machine and the program source for the actuator commands is selected The controller Program Source is set to Scrament mode in order to enable actuator commands from the xPC Otherwise the command signal can be produced locally using the function generator or other available sources Hybrid controller hardware with scramnet MTS Rack mount hardware consists of MTS VME console including digital actuator controllers signal conditioners and interlock mechanisms The controller is preset to run at a frequency of 1024 Hz which is the update rate nees berkeley User s Manual Page 12 for the servo valve commands and the fastest rate at which data can be sampled All servo hydraulic components on the lab floor connect to this hardware including servo valves and feedback instrumentation for control force displacement deltaP The MTS hardware includes 24 signal conditioners used primarily for feedback sensors analog I O digital I O and encoders All measured data from the sensors and internal control commands are written to the scramnet memory also at a rate of 1024 Hz If the controller program source is set to scramnet the actuator commands are obtained from a predetermined scramnet memory location Data
32. o the slow state In the slow state extrapolation continues at a reduced velocity to keep the actuator in continuous motion while allowing more time to receive an update Upon receiving the next target displacement the D_target TimeOut TimeOut TimeOut Legend State mm State Transition Path Event causing State Transition EventffunctionCall Figure 4 Event driven scheme using a polynomial predictor corrector to continuously generate actuator commands nees Oberkelev User s Manual Page 10 interpolate state is activated If the update is not received within a set amount of time the slow state needs to TimeOut as well to place the actuator on hold until the target displacement is received Longer delays possibly due to the integrator crashing or a network failure could indefinitely delay the controller receiving an updated displacement For this rare event the hold state can also time out and force the system into free_vibration or any other desirable state to dissipate the energy in the physical specimens and end the test The free_vibration state is intended to fully unload the physical substructure based on locally stored mass and damping ratio for the test specimen nees berkeley User s Manual Page 11 Chapter 3 DESCRIPTION OF HARDWARE The hybrid simulation control system is composed of various computers Each individual computer is first described in Section 3 1 then the interconnecti
33. ock for Actuator 1 to some value and press connect button in the Simulink model The displacement value will be passed through the Scramnet to the MTS Host so the displacement command will take this value in the STS window of the MTS Host computer Once the model is playing tg object is started any change in Displ box will result in a change in the STS oscilloscope presentation on the MTS Host machine If the run button is pushed the change in the Scramnet command will cause a change in the feedback command nees berkeley User s Manual Page 24 CI RTSimulation Ex File Edit View Simulation Format Tools Help Dcus BIT hk Eten SHAS RAE Constart2 Constant Constant input from scramnelt output to seramnet master span 14 Constant tarmi Target Scope cil modes 16 4 master span displ om er Scope PC em toroe 3 Force displ sme Forces force ftis pl Rate Limiter thes Rate Limitert force omdz valve om s Constanti user dues ActuatoiMdiz user anes ig ines temm12 init 8C150 init UC B RT Simulation Ready 100 FixedStepDiserete Figure 11 Simulink RTSimulation mdl for real time simulation 4 4 Fast Hybrid Simulation This example demonstrates the use of the hybrid controller to run a real time simulation The integration algori
34. on sweep model E lol xl File Edit View Simulation Format Tools Help DISS See 22 ah function Ready output to udp input from udp ActustorPsylosdMdl Figure 9 Block model of PCSimulation hybrid mdl Before presenting a simulation using the actual hybrid simulation algorithm a simpler model is presented to demonstrate the use of STS In this example the STS software is used to excite the actuator directly without use of the hybrid simulation algorithm The followings steps should be followed 1 Start ucbSTS STS exe and select a set file 2 Start Matlab and switch to directory ucbSTS Simulink 3 In Matlab run initialize m and open Simulink model PCSimulation mdl or PCSimulation hybrid mdl both will work 4 Run Simulink model clicking on the plav button 5 In STS set program source to Function Generator and open operation gt function generator Select Act 1 Displ and create a command signal for example a sine wave nees berkeley User s Manual Page 22 6 Select view gt oscilloscope and set CH A and CH B to Act 1 disp cmd and Act 1 displ fbk then hit auto to start viewing data 7 In the main panel hit run to start the function generator You should see the generated displacement command in the oscilloscope The displacement feedback should follow the displacement command Note that if the Simulink model is not playing the displacement f
35. on to minimize the random effects on experimental substructures The programming procedure is based on defining a number of states in which the controller can exist in and the transitions between these states that take place as specified events occur Nakashima and Masaoka s 13 algorithm reacts to events in the sense that the algorithm switches from extrapolation to interpolation after the integration task is completed However the variance in task completion times for their application was minimal They used an explicit integration method and the DSP running these tasks had a dedicated and reliable connection to the servo hydraulic controller This algorithm will not function effectively for distributed hybrid simulations involving the Internet since random delays are likely to occur The state transition 1 1 1 1 DSP PID Signal Servo hydraulic i Generation control NETWORK Actuator Load cell Remote Substructure A Analysis Site DSP PID Signal Servo hydraulic Generation control Load cell Remote Substructure B Figure 3 Distributed hardware architecture for geographically distributed testing nees berkeley User s Manual Page 9 diagram in Figure 4 shows the implementation of event driven version of a polvnomial predictor corrector command generation method This algorithm continuouslv updates the actuator commands using the same approach under normal ope
36. ons that form the hybrid simulation controller are described in Section 3 2 Special attention is given to the shared memory link in Section 3 3 particularly on how to access data from this memory for use in hybrid simulation algorithms Details on how to access the shared memory resource from xPC are discussed in Section 3 4 3 1 Hardware Components The following computer equipment forms the hybrid simulation testing system Simulink Host PC Dell Precison 630 Workstation with Windows XP is used to program Simulink models with data signals mapped to scramnet memory Once the Simulink model is downloaded to the xPC control of the model including start stop and parameter tuning can be maintained from this computer although the program actually runs on the xPC Target xPC Target PC with scramnet Dell Precison 630 Workstation contains two operating systems 1 xPC boot from floppy and 2 normal Windows XP boot from hard disk In xPC mode this computer functions as a real time processor and runs compiled Simulink models in real time The programs running on the xPC are downloaded and controlled from the Simulink Host There is little user interaction with the xPC model In Windows mode this computer is able to access scramnet although not necessarily in real time and can be used to view data in Windows applications Hybrid controller host PC Dell Precison 630 Workstation with Windows XP runs the graphical user interface t
37. or solving the equation of motion The signal generation task on the other hand computes the displacement path of the actuator using polvnomial approximation procedures Nakashima and Masaoka showed that third order polvnomial interpolation and extrapolation of known displacement values from previous steps provide accurate displacement and velocitv predictions in the current step The to this polvnomial approximation procedure is that the computation time is small and actuator commands can be continuously generated at small constant time intervals For each integration step the actuator is kept in motion after achieving the target displacement by predicting a command signal based on polynomial extrapolation of the previous target displacement values Meanwhile the integrator task is carrying out computations for the next target displacement Once the integration task has been completed and the target displacement is known the controller switches to interpolate towards the correct target value An advantageous feature of this algorithm is that the communication between the integration task and the signal generation task is minimized 2 2 Distributed Hardware Architecture The typical architecture of a hybrid simulation controller consists of the integration loop commanding the inner servo hydraulic controller loop as previously shown in Figure 1 A single processor is used to compute both the integration of the equation of motion and the signal
38. or the servo hydraulic controller are updated at deterministic rates Constant update rates allow for the control of the actuator velocity thus allowing for a continuous load history non zero velocity on the experimental elements The difference between the ramp hold and a continuous load history is shown for one simulation step in Figure 2 Note that the continuous procedure utilizing a predictor corrector approach reduces the velocity demands for the same time interval An example of a predictor corrector technique for continuous loading is summarized below computation computation application ke E di g continuous D predictor hold 5 oo ramp ho 2d Bot hold o ko AT simulation time step actual time Figure 2 Ramp hold and continuous load histories In their algorithm for real time testing Nakashima and Masaoka 13 separate the computations in the on line computer into two tasks running at different sampling rates 1 the response analysis task which carries out the integration of the equation of motion and 2 a signal nees berkeley User s Manual Page 7 generation task which provides displacement commands to the servo hvdraulic actuator at a rate faster than that of the integration time step These two tasks run on a Digital Signal Processor DSP in real time using a multi rate approach The response analvsis task deals with the tvpical numerical algorithms f
39. ration conditions and takes action for excessive delavs This diagram consists of five states extrapolate interpolate slow hold and free vibration The default state is extrapolate during which the controller commands are predicted based on previously computed displacements while the integrator computes the next target displacement The state changes from extrapolate to interpolate after the controller receives the next target displacement and generates the event D_update The event D_target is generated once the physical substructure has realized this target displacement The model then subsequently transitions back to the extrapolate state and sends updated measurements to the integrator The smooth execution of this procedure is dependent on having a reliable network connection and selecting the run time of each integration step sufficiently large for all of the required tasks to finish Small variations in completion times for these tasks will only affect the total number of extrapolation steps versus interpolation steps The advantage of the event driven approach is that logic can be included to handle excessive delays For example if the system is in the extrapolate state longer than a specified time the actuator can deviate from the intended trajectory or even exceed its target hence limits need to be placed on the number of allowable extrapolation steps A simple solution is to generate the event TimeOut which will transition the controller t
40. rthquake Engineering and Structural Dynamics 1999 28 4 393 420 14 Systran SCRAMNet Network Systran Corporation 2003 15 Harel D Statecharts A visual formalism for complex systems Science of Computer Programming 1987 8 231 274 nees berkeley User s Manual Page 32 Chapter 7 APPENDIX A SCRAMNET MEMORY nees Oberkelev User s Manual Page 33 Chapter 8 APPENDIX B CONTROLLER SIMULATION MODES Structural Controller Simulation Modes B K THOEN MTS 17 Dec 02 nees berkeley User s Manual Page 34
41. t is expected that experienced on site personnel will operate the main servo controller or Structural Testing System STS controlling the actuators Documentation on the use of the STS software package is available for interested users MTS 2003 Separate documentation on use of the Pacific Instruments data acquisition software is also available The user should refer to related users manual for setting up test specimen in the strong floor calibrating instrumentation and tuning actuators 1 3 Infrastructure of nees berkeley 1 2 Oy 10 11 12 The nees berkeley laboratory consists of the following A 6 m x 18 m 20 ft x 60 ft structural tie down floor with 60 mm 2 5 in tie down holes located in an array at 910 mm 36 in on center to accommodate service load of 445 kN 100 kip acting either up or down for each tie down hole A 17 8 MN 4 000 kip capacity Southwark Emery Universal Testing Machine An overhead 107 kN 12 US ton capacity bridge crane 15 2 m x 30 5 m 50 ft x 100 ft paved construction area Twenty four 3050 mm x 2740 mm x 760 mm 120 in x 108 in x 30 in hollow modular reinforced concrete wall units that may be post tensioned to the floor using 12 threaded rods with a prestressing force of 445 kN 100 kip per rod to build one or more reaction walls up to 13 m 42 5 ft in height with the maximum shear force and bending moment of 1780 kN 400 kip and 5420 m kN 4000 ft kip respective
42. thm is programmed into Simulink using an S function and runs in real time In addition a polynomial approximation procedure is incorporated to generate commands to the actuator at 1024 HZ regardless of the rate of the integrator The numerical integration algorithm can be scaled in time to run the simulation at slower rates The procedure to run a fast hybrid simulation is very similar to the procedure for PC Simulation mode 1 Once an actuator has been tuned and attached to a specimen the STS software controlling the real hardware should be set to scramnet program source HSM nees berkeley User s Manual Page 25 turned on and the run button pressed to set span to 100 The controller is now ready to accept commands from scramnet Before hitting run it is advisable to check that the signal Act scram cmd is 0 If not this should be set to 0 by trying a quick tg start tg stop from Matlab in the xPC Host assuming a model is loaded 2 On the xPC Host open the Simulink model FastHybrid sdf mdl Run SDF_input m to set parameters for the structural model 3 Build and download the model to xPC If STS is ready type tg start in Matlab to start simulation Note that the simulation starts as soon as tg start is entered It is important that the actuators be ready to go at this point 4 Data can be saved to a file using STS limited to control signals in MTS controller or xPC Target 4 5 Slow Continuous Hybrid Sim
43. ugh numerical simulations using the internet Campbell and Stojadinovic 8 The infrastructure of the George E Brown Jr Network for Earthquake Engineering Simulation NEES provides the experimental equipment the analytical modeling tools and the network interface to research complex analytical models with the simultaneous testing of multiple large scale experimental substructures using the distributed hybrid simulation approach Geographically distributed hybrid simulation has already been carried out jointly between Japan and Korea Watanabe et al 9 in Taiwan Tsai et al 10 and in the U S as part of the NEES efforts MOST 11 However these applications of distributed hybrid simulation have used the ramp hold procedure to load the experimental substructures As such they are not benefiting from the advanced continuous methods that can improve the measured behavior of the experimental substructures and the reliability of the test results The difficulty in applying real time based continuous algorithms to distributed applications stems from their lack of suitability with tasks that involve random completion times Random completion times in network communication numerical integration and other such tasks could compromise the stability of real time algorithms because they may not complete in the time required by the real time test clock The ramp hold loading procedure can be readily nees berkeley User s Manual Page 5 applied
44. ulation In the previous example the complete simulation algorithm including the integrator and signal generation tasks were programmed within Simulink Particularly for the integration algorithm the programming for large structural models can become difficult in Simulink This example demonstrates the use of multiple processors and easier programming using Matlab The complete structural model and the integration algorithm are programmed in Matlab thus can be done for complex structures The xPC Target runs only the signal generation task which is based on an event driven control strategy to deal with possible delays in the non real time Matlab environment The signal generation algorithms run in real time using polynomial approximations to generate actuator commands at 1024 HZ The necessary files to run a hybrid simulation locally using matlab and xPC can be found in ucbSTS Simulink eventmdl The numerical model consists of a 2DOF shear building model where the story resisting forces are obtain from two experimental substructures The numerical analysis in localsim 2dofshear m sends two target displacements and receives measured response values from the Simulink model FastHvbrid eventmdl mdl running in real time on the xPC The Simulink model includes a Stateflow block eventcommand Note that this requires Stateflow and Stateflow to be able to compile and run in real time This block generates a continuous displacement
45. xperimental sites are simulated on a separate computer referred to as the remote computer The remote computer assumes the Matab software is installed here NEESGRID 2 2 release is used 1 Start NTCP a Log in to NEESPOP using SSH neespopd berkelev edu See System Administrator for account b Make sure no NTCP processes are currently running ntcpd stop c Start 6 NTCP processes ntcpd c 6 start 2 Prepare Remote Simulation a Open A Command console to directory C neesgrid 2 2 matlab JavaSimulationCoordinator gt b Open 4 instances of Matlab to directory C neesgrid 2 2 matlab FastMOST 3 Prepare Local Simulation a Open Matlab to directory D MTSmodel ucbSTS Simulink event_model b Open Simulink model fast hybrid eventmdl mdl and load model to xPC NOTE the model should be downloaded from this directory because files are automatically generated in the current directory cd to C neesgrid 2 2 matlab FastMOST d Start actuators 4 Run Simulation a In remote command prompt type ant run b Start the local simulation in Matlab by typing Cal Exp Site c Atthe remote site start the four other experimental sites one in each of the Matlab windowns CU Exp Site UIUC Exp Site UBUF Exp Site Lehi Exp Site For best results start in this order 5 After the test has completed view simulation results in C neesgrid 2 2 matlabJavaSimulationCoordinator outputsimulationCoordinator dat or in the current workspace in the Matlab simul
Download Pdf Manuals
Related Search