Emtp Mahseredjian R&D

1,997 views
1,854 views

Published on

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,997
On SlideShare
0
From Embeds
0
Number of Embeds
11
Actions
Shares
0
Downloads
49
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Emtp Mahseredjian R&D

  1. 1. EMTP‐RV Research and development p Jean Mahseredjian Professor jeanm@polymtl.ca École Polytechnique de Montréal École Polytechnique de Montréal Thursday, April 29
  2. 2. History: R&D project History: R&D project • R Research and development organization: Development  h dd l t i ti D l t Coordination Group (DCG‐EMTP) g g p • EMTP: Electromagnetic Transients Program, developed since the  70s, major versions in 90 and 96 • Completely new software and technology: EMTP‐RV • Large and complex project: total duration 5 years Large and complex project: total duration 5 years • First commercial release version 1 in 2003 g • Large scale software with more than 1 million lines of code • New computational engine and New graphical user interface  (GUI) • C Commercialized: www.emtp.com i li d t • DCG Members: Hydro‐Québec, Électricité de France, CRIEPI  ( p ), (Japan), Entergy, American Electric Power, Western Area Power  gy, , Administration, US Bureau of Reclamation, Hydro‐One, CEATI 2
  3. 3. Old EMTP software and technology New Computation Methods New EMTP RV  New EMTP‐RV (Restructured Version)  3
  4. 4. Support and development Support and development • Level 1: Neil MacKenzie, Capilano Computing e e e ac e e, Cap a o Co put g • Level 2: Awa‐Marie Ndiaye, CEATI • Level 3: Jean Mahseredjian, École Polytechnique • Development: Jean Mahseredjian, Chris Dewhurst (Capilano) – Team at École Polytechnique • Luis Daniel Bellomo research associate Luis Daniel Bellomo, research associate • Many Ph.D. students • Many M. A. Sc. students • Special developments: Special developments: – Several funded projects with Hydro‐Québec – Several funded projects with EDF p j • Major contributors: – Hydro‐Québec – EDF – Developments, funding, funding of research
  5. 5. Courses on EMTP‐RV Courses on EMTP RV • C Courses in 2008 i 2008 – Australia (May) – Saudi Arabia (June) Saudi Arabia (June) – Madison (University of Wisconsin) – Montréal (September) Montréal (September) – Paris (Supélec, September) – Orléans (Vergnet, éoliennes, September) (Vergnet, éoliennes, September) • Courses in 2009 – Special course for Hydro‐Québec, March Special course for Hydro Québec, March – Croatia, April – New Orleans, US, November , ,
  6. 6. Other courses Other courses • Courses on transients (not software) – Seoul, South Korea, Sungkyunkwan University,  , , g y y, April 2009 – Special long course every year École Special long course, every year, École  Polytechnique de Montréal (web page) – Seoul South Korea Sungkyunkwan University Seoul, South Korea, Sungkyunkwan University,  August 2009
  7. 7. New version 2.2 New version 2.2 • What is new in 2.2 – Full compatibility with Vista – New documentation system with new navigation features – Various improvements and additions to models. The data handling  features for several models are now simplified to allow easier loading  when separately calculated data. h l l l dd – New capability to store complete circuits in libraries. A circuit  appearing in a library folder now becomes listed in the library Parts  Palette and can be dragged and dropped into a design just like  Palette and can be dragged and dropped into a design just like standard parts. This is a very powerful feature that provides easy  access to user circuits and allows maintaining more complex models  g through libraries. – Subcircuits are now given the Model or Physical attribute in the  Subcircuit Info menu. A model subcircuit is primarily intended to  define the operation of the device represented by its parent symbol. A  physical subcircuit i i h i l b i it is primarily used to contain some of the system. The  il dt t i f th t Th devices inside the subcircuit represent actual physical elements of the  system. The physical subcircuit may contain Model subcircuits. This  distinction allows propagating computed data into Physical subcircuits distinction allows propagating computed data into Physical subcircuits for visualization purposes.
  8. 8. New version 2.2 New version 2 2 –SSeveral new scripting methods, including: dynamic  l i i h d i l di d i modification of device symbol using a separately  stored symbol drawing. stored symbol drawing – Several improvements • New ScopeView New ScopeView – Vista compatible – Several improvements • A A new HVDC model benchmark (for 50 Hz and 60 Hz networks)  HVDC d lb h k (f 50 H d 60 H k) originally developed by professor Vijay Sood (University of  Ontario Institute of Technology) is now available upon request.  This work resulted from a collaboration with Sébastien  Dennetière (Électricité de France) and École Polytechnique de  Montréal.
  9. 9. Scenario attribute Scenario attribute • Allows changing scenarios in one easy step • Each device is given a Scenario attribute and a  Scenario.Script attribute p – Built‐in • Simple user‐defined scenarios Simple user defined scenarios dev=defaultObject() Scenario=dev.getAttribute('Scenario'); switch (Scenario){ switch (Scenario){ case '1' : dev.setAttribute('Exclude','Ex') break; case '2' : dev.setAttribute( Exclude ) dev setAttribute('Exclude','') break; }
  10. 10. Recently completed R&D projects Recently completed R&D projects • 0 0‐Hz startup of Synchronous machine fS h hi – Project EDF R&D, Clamart – Allows using the synchronous machine model without  60 Hz or 50 Hz initialisation –SStarts from 0 Hz. f 0H – Allows studying the machine startup and  synchronization onto the network synchronization onto the network – For pumped storage studies – For black start st dies For black‐start studies • Improved wind generator models
  11. 11. + Network PLL L Grid Grid + + frequency a angle • IPST 2009 Ir re Curre ent - Contr roller + Ire + U K Current Limiter PLL + + Control J M h Speed & Torque Inver rter Contr roller + Rotor Position P SM - + SM frequency dji S D Act tivated ed Activate Frequency Excitation Vol ltage control when System when f > 47 Hz Controller H f > 47 Hz - SM + voltage tiè - -+ + IPST‐2009 paper, U. Karaagac, J. Mahseredjian, S. Dennetière Grid Grid d Activated a Pumped Storage Power Plant Unit voltage frequency when Angle z Δf < 1Hz Controller SM angle Δθ < 45° Modeling and Simulation of the Startup of  Grid angle -+
  12. 12. • Measured and simulated frequencies Measured and simulated frequencies 51 50 Hz) quency (H 49 Freq 48 47 105 110 115 120 125 Time (s)
  13. 13. 1250 A) urrent (A 1200 Machine field cu 1150 M 1100 95 100 105 110 115 Time (s) Machine field currents
  14. 14. 4 x 10 V) ltage (V 1.8 1.6 16 -line vol rms line-to- 1.4 s 1.2 80 90 100 110 120 Time (s) i () Machine terminal rms li t li M hi t i l line‐to‐line voltage lt
  15. 15. 5 0 W) ower (MW -5 ctive Po -10 Ac -15 -20 80 90 100 110 120 Time (s) Active power delivered by the machine
  16. 16. Improved Wind generator models Improved Wind generator models • Generic models – Detailed – Mean‐value models • Matching of PSS/E results for slow transients M t hi f PSS/E lt f l t i t • Initialization scripts p • Flicker meters • Work completed by L. D. Bellomo and J.  W k l d b L D B ll dJ Mahseredjian (École Polytechnique)
  17. 17. SW1 + WINDLV1 WTG1 1.00/_6.6 10 generators 34.5/0.69 2 ? 1 + ZnO O1 ZnO 1.00/_6.3 WINDLV2 + YD_1 BUS12 34.5/0.69 Network 1 2 MAIN_SW 1 2 + + + + SW2 + 230kVRMSLL /_0 230/34.5 nO3 Zn WTG2 ZnO + ZnO2 10 generators 5Ohm ZZ ZnO + .25 34.5/0.69 2 1 0.99/_5.9 WINDLV3 SW3 + WTG3 10 generators
  18. 18. 80 60 40 20 V) (kV 0 -20 -40 -60 60 -80 0 0.5 1 1.5 2 time (s) 3.5 3 Voltage 2.5 25 2 (pu) 1.5 1 0.5 Obvervoltage trip signal Crowbar signal 0 0 0.5 1 1.5 2 2.5 time (s)
  19. 19. Improvements to the load‐flow  module (next versions) • P Presentation and location of worst mismatch locations i dl i f i hl i • Presentation and location of reactive power violations • Presentation of PQ power on transmission lines (on the  design symbols) • Automatic calculation of tap positions l l f – Automatic initialization for tap control signals • Automatic calculation of asynchronous machine slip  l l f h h l from mechanical power or electrical power • Th The area control notion t l ti • Attribute scripting for device data based on LF solution
  20. 20. Toolboxes • CRINOLINE l t CRINOLINE: electromagnetic compatibility ti tibilit • EGERIE – Short‐circuit analysis package Short circuit analysis package – Automates short‐circuit studies • Harmonic analysis – Harmonic source models – Analysis tools – Compensator models Compensator models • Parametric studies – Advanced functions, high level scripting – Scenario studies • LIPS: Lightning impact on power systems – Automation level for lightning analysis Automation level for lightning analysis
  21. 21. Other works Other works • C Conversion of remaining device scripts to the object‐ i f i i d i i t t th bj t oriented version • Scripts for automatic layout of signals automatic Scripts for automatic layout of signals, automatic  connections for building entire networks • Simplified SVC model: controlled inductance (currently  p ( y available) • Switching to the Intel compiler – Compatibility of DLLs bl f • New C/C++ DLL (prebuilt) for direct interfacing through  DLL (IREQ) DLL (IREQ) • New DLL specific to control systems, based on  p perturbation theoryy
  22. 22. Modeling of transmission lines and  cables • C Current limitations li i i – The Wideband model may encounter numerical problems • Can be fixed by user manipulations of the fitting function not Can be fixed by user manipulations of the fitting function, not  simple • Complex research problem in the literature, many papers • Prominent problem for short cable • Development of a new fitting method: WVF • Contribution of an error control technique in time‐ b f l h domain –M More robust, stable model b t t bl d l • Results presented in IEEE papers
  23. 23. H = exp − YZl ( ) H=e ( T Λ T −1 ) = T e Λ T −1 cn N M ⎡ N m cij mn ⎤ ( − s ⋅τ ) H mode ≅e ∑ − sτ H ij ( s ) ≅ ∑ ⎢ ∑ ⎥e m m =1 ⎢ n =1 s + pmn ⎥ n =1 s + pn ⎣ ⎦ 4 10 Magnitude of modes in H(1,1) WB Calculated 2 4,5,3,2 wb , , , 10 0 gnitude 1 10 1 wb Mag -2 10 6 wb -4 10 7 wb 7 6,3,2,4,5 -6 10 0 2 4 6 10 10 (Hz) 10 10
  24. 24. CEWB FDQ WB Δt 1 microsec 2 CEWB_V10 CEWB V10 FDQ_V10 WB_V10 1.5 Voltage 1 0.5 0 0 0.01 0 01 0.02 0 02 0.03 0 03 0.04 0 04 0.05 0 05 0.06 0 06 0.07 0 07 0.08 0 08 t (ms) WB CWB FDQ Δt 0.1 μs 2 CWB_V10 FDQ V10 Q_ WB_V10 1.5 age Volta 1 0.5 0 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 t (ms)
  25. 25. Other R&D based on EMTP‐RV Other R&D based on EMTP RV • New hysteretic reactor model, completed M. A. Sc. project – Better fitting method • Other hysteretic reactor models: – Preisach based model (University of Toronto), completed – Programming of the old EMTP type 96, started  • Vacuum breaker model, currently available • Fast to superfast computations Fast to superfast computations – The dynamic phasor approach for slow transients (stability analysis  needs) – Relaxation techniques q – Automatic adjustment of synchronous machine solutions for slower  transients – Parallel computations p • Using the Multi‐Core processors • One simulation to many simulations – New solution methods for control systems New solution methods for control systems – FPGA programming of a sparse‐matrix based solver solver
  26. 26. New solution methods for control  systems (research) • I Improvement of speed t f d • Reduction of Jacobian matrix size (demonstration  prototype) • Elimination of the matrix based solver • E i Estimated gains in speed: 5 to 10 times d i i d 5 10 i • Research on a single system of equations: power  and control‐diagram based models d t l di b d d l
  27. 27. Control system equations N6 L1 L2 L3 L4 L5 N7 Iterative solver ⎡1 ⎤ ⎡ xN7 ⎤ ⎡ fN7 ( xL4 )⎤ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ 1 ⎥⎢ xN6 ⎥ ⎢ fN6 ( xL5 )⎥ ⎢ 1 −kL5 ⎥ ⎢x ⎥ ⎢0 ⎥ Computation of Jacobian matrix  Computation of Jacobian matrix ⎢ ⎥ ⎢ L5 ⎥ ⎢ ⎥ by perturbation ⎢ −kL4 ⎥ ⎢ xL4 ⎥ ⎢ 1 1 ⎥ ⎢x ⎥ =⎢ ⎢0 ⎥ ⎥ 1 -1 ⎢ ⎥ ⎢ L3 ⎥ ⎢ 0 ⎥ ⎢ 1 −kL2 ⎥ ⎢ xL2 ⎥ ⎢0 ⎥ ⎥⎢ Jx = b ⎢ ⎥ ⎢0 ⎥ ⎢1 1 -1⎥ ⎢ xL1 ⎥ ⎢ ⎥ ⎢ ⎥⎢ 1 ⎦ ⎢u ⎥ ⎥ ⎢s ⎣ ⎥ ⎦ ⎣ ⎣ ⎦ 27
  28. 28. Other R&D based on EMTP‐RV Other R&D based on EMTP RV • Database! • Development of portable data modeling Development of portable data modeling  methods – P t bilit t d d CIM V il VHDL? Portability standards: CIM, Verilog‐VHDL? – Data – Portable modeling between applications • New IEEE Task Force New IEEE Task Force
  29. 29. G1 + + Equivalent G2 + 144 + 144 BUS1 + + B ZnO+ SEND + + 24 193 + BUS2 P Load Details Q REC ZnO+ SVC V I NemiscauCLC + + + Nemiscau b780 Nemiscau_b780 CXC82_ZnO CXC63_Z ZnO Very large networks Very large networks + ZnO 330 MX 330 MX O + ZnO 330 MX + + + L7082 L7063 + + + + CXC82 CXC C63 CXC81_ZnO + + CXC62_Z ZnO Radisson_b720 O + ZnO + ZnO 330 MX 330 MX + + L7081 L7062 + + + + CXC81 CXC C62 + + CXC80_ZnO CXC61_ _ZnO + ZnO 330 MX 330 MX O + ZnO + + L7080 L L7061 + + + + CXC80 CXC C61 Eastmainlasarcelle B19 B18
  30. 30. Hydro‐Québec Network Hydro Québec Network • IPST‐2009 paper, L. Gérin‐Lajoie, J. Mahseredjian • Complete network (L) p ( ) – The complete Hydro‐Québec network is organized  using a multilevel hierarchical design structured on 6  using a multilevel hierarchical design structured on 6 pages in the GUI. There are a total of 30000 physical  devices and 28000 signals. The list of physical devices  g p y includes 19000 control devices and coupled 3, 6 or 9‐ phase devices are counted once. The signal count  adds 8000 power nodes to 20000 control system  signals.
  31. 31. • Complete network (L) – Th The top level listing (subnetwork contents are not counted) of main  l l li i ( b k d) f i devices is: – 1100 transmission lines representing the existing 1560 lines and  derivations – 296 three‐phase transformers representing the existing 1500 three‐ phase units connected in Ynyn, DD, Dyn, Ynd, Ynynd, Yndd and ZigZag grounding banks grounding banks – 532 load models representing a total of 36000 MW. All medium and  high voltage shunt capacitors and inductors were modeled separately.  Some loads were modeled with the transformer and shunt capacitor p at the lower voltage level. – 7 SVC (Static Var Compensator) models of 300 Mvars and 600 Mvars.  The SVCs have been combined on some buses by creating 600 Mvar models. d l – 32 series capacitor MOVs and 303 nonlinear inductances used for high  voltage power transformer saturation representation. – 99 99 synchronous machines (SM) with associated controls representing  h hi (SM) ith i t d t l ti more than 49 power stations and four synchronous compensators. All  synchronous machine devices are matched to corresponding load‐flow  type devices for specifying the PV constraints used for initializing  type devices for specifying the PV constraints used for initializing machine phasors at load‐flow solution convergence. All machines are  given a single‐mass model except one nuclear power plant generator  modeled using 10 masses.
  32. 32. • Reduced network Reduced network – The reduced network has a total of 24000 physical  devices and around 24000 signals. There are 4000  power devices and 2500 power nodes. The listing of  top level devices is: – 170 lines with 75 lines at the 735 kV level 53 at 170 lines, with 75 lines at the 735 kV level, 53 at  315 kV, 23 at 230 kV and 19 at 120 kV – 90 three‐phase transformers p – 27 load models, 7 at 315 kV, 6 at 230 kV, 4 at 161 kV, 6  at 120 kV and 4 at 13.8 kV for a total of 33800 MW – 7 SVC models d l – 39 synchronous machines with AVRs for representing  31 power stations and 3 synchronous compensators  31 power stations and 3 synchronous compensators for a total of 35600 MW of generation.
  33. 33. Substation no.1 Substation no.4 400 400 300 300 200 200 100 100 0 0 0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 Substation no.2 Substation no.5 400 60 300 40 200 100 20 0 0 0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 Substation no.3 Substation no.6 400 100 300 200 50 100 0 0 0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 Frequency (Hz) Frequency (Hz) Frequency response (positive sequence impedance) plots for the  complete (blue) and reduced (green) networks. Left column plots show  three 735 kV b t ti th 735 kV substations and right column plots show three 315 kV d i ht l l t h th 315 kV  substations.
  34. 34. Substation no.1 - Bus voltage (pu) Substation no.2 - Bus voltage (pu) 1.04 1 1.03 1.02 1.01 1 0.99 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 Line no.1 - Transmitted Power (MW) Line no.2 - Transmitted Power (MW) 2280 2460 2260 2450 2240 2440 2220 2430 2200 2420 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 Power plant no.1 - Power flow (MW) Power plant no.2 - Power flow (MW) 2620 5660 2600 5640 2580 5620 2560 5600 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 Time (s) Time (s) Network initialization test without SVCs, L‐Network (blue),  R‐Network (green) and PSS/E (red)
  35. 35. Substation no.1 - Bus voltage (pu) Substation no.2 - Bus voltage (pu) 1.04 1 1.02 0.995 1 0.99 0 99 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 Line no. 1 - Transmitted Power (MW) Line no. 2 - Transmitted Power (MW) 2300 2440 2250 2420 2200 2400 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 Power plant no.1 - Power flow (MW) Power plant no.2 - Power flow (MW) 5700 2600 2580 5650 2560 5600 2540 2520 5550 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 Time (s) Time (s) Network initialization test with SVCs, L‐Network (blue),  ( ) R‐Network (green) and PSS/E (red)
  36. 36. Substation no.1 - Bus voltage (pu) Substation no.2 - Bus voltage (pu) 1.08 1.04 1.06 1.02 1 1.04 0.98 1.02 0.96 1 0.94 0.98 0 98 0.92 0 92 0 5 10 0 2 4 6 8 10 Line no. 1 - Transmitted Power (MW) Line no. 2 - Transmitted Power (MW) 2400 2600 2300 2500 2200 2400 2100 2300 2000 2200 0 5 10 0 2 4 6 8 10 Power plant no.1 - Power flow (MW) Power plant no.2 - Power flow (MW) 2800 6000 5800 2600 5600 2400 5400 2200 5200 0 5 10 0 2 4 6 8 10 Time (s) Time (s) Simulation of a 3‐phase fault and loss of a 735 kV transmission line,  Simulation of a 3 phase fault and loss of a 735 kV transmission line L‐Network (blue), R‐Network (green) and PSS/E (red)
  37. 37. a) Generator frequencies at James Bay Complex 66 64 Hz 62 60 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2 b) Prospective TOV at LVD7 1 pu 0 -1 -2 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 c) TOV at LVD7 with LVD7-Montreal tripping 2 1 pu 0 -1 -2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 time (s) James Bay system voltage oscillations due to an  extreme disturbance
  38. 38. 30000 devices 28000 signals
  39. 39. 0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000 Solved time‐domain sparse matrix for the L‐Network,  50269 non‐zeros
  40. 40. CPU ti i timings (s) for a 10 s simulation interval ( )f i l ti i t l CPU Timers L-Network R-Network GUI File (design) load 9 4 Data generation 10 3 Load-flow solution 181 (6 iterations) 21 (7 iterations) Steady-state solution 0.48 0.12 Time-step 100 µs 200 µs 100 µs 200 µs Time-domain network equations 4710 2548 538 276 Time-domain control equations 846 435 715 389 Time-domain updating 409 210 75 36 Time-domain solution total 5965 3103 1328 701 99 min 52 min 22 min 12 min

×