Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
Emtp Mahseredjian R&D
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. 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
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. 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. 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. 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. 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. 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. 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. +
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. • 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. 1250
A)
urrent (A
1200
Machine field cu
1150
M
1100
95 100 105 110 115
Time (s)
Machine field currents
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. 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. 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)
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. 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. 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. 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. 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. 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
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. 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
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
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. • 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. • 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. 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. 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. 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. 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. 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
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