model based design approach for hvdc applications€¦ · steven‘s software emt simulation needs...

Intern © Siemens AG 2014 Alle Rechte vorbehalten. Answers for energy.

Model Based Design approach for HVDC applicationsMATLAB-EXPO, Munich 2014-07-09

Markus Engel, Siemens AG, E T TS PLM C&PCo-Authors: Sebastian Behnisch, Dr. Christian Siegl, Christina Zenk, Marcos Pereira, Kevin Römer

Intern © Siemens AG 2014 Alle Rechte vorbehalten.

17.06.2014Seite 2 Engel / E T TS PLM

Agenda

Current situation

Challenges

Goals & Constraints

Solution

Conclusion

Outlook



Current situationTurnkey solution business & customers with high buying power We have to deal with manifold contributions – see hawker‘s tray of C&P

Control & Protection System

PAS

SICAM RTU

SIPROTEC 5

TFR Acquisition-PCs

1 2 3 4 65

TDCPLUS

Control

VBE

CCTV

S7 / PCS7 Komponents

Air ConditioningDiesel aggregatsRaw water

Chiller

Waste water

Fire detection

Door access

Weather data



Challenges: System & Software engineers work on parallel paths …

Susie‘s Simulation

Steven‘s Software

HVDC1: Blackhillock (1+2)

HVDC-Hub

HVDC5: Future B (DC Source)

HVDC4: Future A (DC Source)

HVDC2: Spittal (DC source)

VA

HVDC3: Kergord (DC Source)

VA

exBLK_4

exBLK_5VA

VA

DS

C_3

DS

C_3

DS

C_4

DS

C_4

DS

C_5

DS

C_5

DSC_2

DSC_2 DSC_1

DSC_1

wP_2

wP_4

wP_5

Udp_Hub

Udn_HubVA

VA

VA

VA

exBLK_11

exBLK_12

VA

VA

wP_12

wP_11

VA

VA

exBLK_2

HVDC2src

+-

HVDC12src

+-

Smpl_I Smpl_I

Smpl_I

HVDC11src

+-

wP_3

exBLK_3

Smpl_I

HVDC3src

+-

HVDC4src

+-

Smpl_I

HVDC5src

+-

Smpl_I

C1Cable2S1

C2S2

Cable2C

C1Cable2S1

C2S2

0.1 [ohm]

0.1 [ohm]0.1 [ohm]

0.1 [ohm]

+/- 320 kV @ 800MW37 km

C1Cable3S1

C2S2

Cable3C

C1Cable3S1

C2S2

0.1 [ohm]

0.1 [ohm]0.1 [ohm]

0.1 [ohm]

+/- 320 kV @ 600 MW255 km

Cable4C

+/- 320 kV @ 600 MW100 km

C1Cable4S1

C2S2

C1Cable4S1

C2S2

0.1 [ohm]

0.1 [ohm]0.1 [ohm]

0.1 [ohm]

C1Cable1S1

C2S2

Cable1C

C1Cable1S1

C2S2

0.1 [ohm]

0.1 [ohm]0.1 [ohm]

0.1 [ohm]

+/- 320 kV @ 1200 MW124 km

Cable5aC

C1Cable5aS1

C2S2

C1Cable5aS1

C2S2

0.1 [ohm]

0.1 [ohm]0.1 [ohm]

0.1 [ohm]

+/- 320 kV @ 800 MW200 km

VA

VA

VAVA

VA

C-programs(VHDL)

Electro-Magnetic Transients (EMT) simulation for development & optimizationof- control parameters & algorithms- steady state & dynamic performance- protection coordination

Software development in Step7/CFC (graphical & block-oriented language) & C/C++ / FPGA applications for- closed loop control- open loop control- protection and measurement

Sebastian‘s Software



Algorithms for- open-loop

- closed-loop- Protection

Electricalnetwork

models & control

algorithms

Closed loopalgorithms

Challenges: … with the same objects developed in different tools and they aredependent from each other!

Susie‘s Simulation

Steven‘s Software

EMT simulation needs- Control & protection algorithms- Communication delays- ….

Software development needs- Parameters / structure for control algorithms- ….

Sebastian‘s Software



Goal: One platform independent Model Based Design tool for all targets

Design Tool

Generate C-Code for DSP‘s

Generatefunction blocksfor Simatic TDC

(PLC)

Enable PC-basedtesting

Allow multiuserengineering

Generatefunction blocks

for EMT-simulations

Constraints:

Different automation platforms to besupported

EMT simulation program PSCAD isstate-of-the-art in HVDC business(often PSCAD models have to bedelivered to customers)

System & software engineers workin different departments



Which integration platform fits best?MATLAB/Simulink seems to be a feasible approach for HVDC

Criteria PSCAD CFC/TDC MATLAB/Simulink

integration in software development suite - - ++

rate of coverage of development lines -- - ++

test/commissioning support o ++ o

independency from automation platform o -- ++

working in distributed teams o + ++

know-how protection o o +

fulfillment of single-source principle - - ++

life cycle costs o o +

▲▲



Model Based Design approach using MATLAB/Simulink & code generationwith Embedded Coder

MATLAB/Simulink

OS1

TDC

OS2

DSP

OS3

PSCAD

OS1

…

OSx



MATLAB/Simulink & PSCADApproach is feasible, but some issues to be investigated!

USERFORTRAN

INTERFACE SUBROUTINE

P27AC_IF3.c

SIMULINKC CODE

MAIN PROGRAM

SIMULINKC CODE

PSCAD

P27ACinterface

test model

U_M_I1_AC_Volt_LL_2_or_3_PhaseU_M_I2_AC_Volt_LG_1_Phase_or_L

U_M_I3_AC_UVolt_LG_1_or_MU_M_Block_27AC_UVolt_or_3phl

U_M_Block_at_meas_FaultU_M_T1_Activ_Time_Stage1U_M_T2_Activ_Time_Stage2U_M_T2_Activ_Time_Stage3

U_M_TPW_Prewarning_Time

Y_M_OT1Y_M_OT2Y_M_PRWY_M_OT3

DWORK2

1

0.3

0.10

PDE counters (6)- Only here for test

0.130.16

110

USERC

INTERFACE FUNCTION

P27AC_IF3.f

Main approach: Unchanged use of C-code produced

automatically by Simulink Generation of lib files to be used by PSCAD

Optimal workflow to be developed with PSCAD users: Accessibility of parameters Avoid jumps between Simulink and PSCAD

as far as possible

Fortran wrapper for interfacing with PSCAD necessary Benchmark: Comparison of simulation results of Simulink-based block and original PSCAD block

without deviations



MATLAB/Simulink & Simatic TDC:Referenced models enable variant management, distributed engineering andefficient version control

One main model link to multiple and hierarchical referenced models

Implementation of different variants (e.g. different closed loop control structures) ofone referenced model possible (same external interfaces)

Integration with version control tools (e.g. SVN or GIT) possible

Referenced models enable distributedengineering

Clearness due to hierarchical structure



MATLAB/Simulink & Simatic TDC:Reuse of well-proven TDC library functions by C-macros and use of Legacy Code Tool

Integration of existing macros in Simulink:

C-Source file: hvdc_ccp_lib.cH-Header file: hvdc_ccp_lib.h

Use of Legacy Code Tool:

- Generates the following files:*.mex file for simulation*.tlc file for Simulink Coder

- Generation of library for Simulink

Exisiting function blockswhich should be re-used



MATLAB/Simulink & Simatic TDC:Automatic C-code generation for use in TDC function block generatorConventions defined and workflow tested

referenced model „CRPC“ MATLAB-Script Source files

TDC function block generatorgenerates library

Conventions:- Block names Simulink =

TDC connector names- Signalnames in Simulink=

comments in TDC- Single task within one model- Same datatypes

Additional connectors compared to original:- Version of Simulink model- Enable/disable input- Failure output- Internal variables output



MATLAB/Simulink & Simatic TDC:Runtime of Simulink generated function blocks is at least 50% less thanoriginal TDC function blocks by applying feasible granularity of packaging

performance advantage of Simulink-generated function blocks can bereached by packaging of suitablenumber of function blocks within oneSimulink model

Stateflow suitable for modelling ofplant sequences

1:1 conversion of standard functionblocks from Simatic TDC to Simulinknot feasible

Run

itme

in p

erce

nt

100

200

300

SimulinkSimatic TDC

Comparison of runtime on target platform between original Simatic TDC standard function blocks andidentical function within one function block generated from Simulink



MATLAB/Simulink & Simatic TDC:Simulink concept of constant values and signal busses is useful forparameterization of functions

All parameters are handled within separate referenced models, input e.g. via .csv file Distribution to „consumers“ via signal bus Existing procedures / tools for parameterizing remain as they are

PAM: Parameter ManagerPageTDC: Parameter Generation Tool for TDCHW Config Tool: Definition of hardware configurationFBGEN: generates function blocks from C-CodeStep 7: Engineering tool for Simatic TDC



MATLAB/Simulink & DSPBenchmark against interface-identical software of customer project positive

Critical issues:

• Code efficiency (due to very fast control cycle within µs-area)

• Proof of accurate dynamic performance

Benchmark executed in the following steps:

1. Implementation and test of developed algorithms in MATLAB/Simulink

2. Benchmark between automatic generated Simulink-based software and

(plain C) software of customer project

a) in offline simulation environment (PSCAD)

b) in realtime test environment

Benchmark Results: Control accuracy & dynamic performance nearly identical Runtime performance of Simulink-based control 14% better than original C-code



Conclusion

Overall feasibility of approach is proven andshows significant advantages! platform independency performance savings in project engineering

Closed loop control algorithms migratedinto Simulink models and tested

„One click solution“ for generation of TDC-and PSCAD-lib‘s



Outlook

Migration and test of open-loop andprotection algorithms Use of stateflow Simulink based (automatic) component test setup Integration test in realtime simulation environment Benchmark of runtime behaviour against original

software

Integration of different components‘ models Enable PC based integration tests

Evaluation of Simulink‘s EMT simulationcapabilities

model based design approach for hvdc applications€¦ · steven‘s software emt simulation needs...

Documents