2. 2
Outline
2
- Presentation of real-time simulation
- Large power system simulation technique
- Micro-grid challenges for real-time simulation
- Distributed simulation approach
3. 3
Presentation of real-time simulation
How many type of real-time simulation exist?
Pure Simulation
Controller Plant
IOs
4. 4
Presentation of real-time simulation
How many type of real-time simulation exist?
Controller Plant
IOs
Real-time simulation
5. 5
Presentation of real-time simulation
How many type of real-time simulation exist?
Rapid Controller
Prototyping
(RCP)
IOs
Controller
Plant
IOs
Controller Plant
IOs
Real-time simulation
6. 6
Presentation of real-time simulation
6
How many type of real-time simulation exist?
IOs
Controller
Plant
IOs
Hardware-In-the-Loop
(HIL)
7. 7
Presentation of real-time simulation
7
How many type of real-time simulation exist?
IOs
Controller
Plant
IOs
Hardware-In-the-Loop
(HIL)
Controller
Plant
Real-plant
Power amplifier
Power signals
IOs
IOs
Power-Hardware-In-the-Loop
(PHIL)
17. 17
Outline
17
- Presentation of real-time simulation
- Large power system simulation technique
- Micro-grid challenges for real-time simulation
- Distributed simulation approach
18. 18
Large power system simulation technique
18
ā¢ Specialized software
ā¢ ARTEMiS
ā¢ SSN
ā¢ Traditional method
ā¢ Distributed parameter line
ā¢ Stubline
ā¢ Voltage/Current source
27. 27
Outline
27
- Presentation of real-time simulation
- Large power system simulation technique
- Micro-grid challenges for real-time simulation
- Distributed simulation approach
28. 28
Micro-grid challenges for real-time simulation
28
ā¢ Distributed parameter line (DPL)
1 km
šš = 50 Ć 10ā6 šš = 500 Ć 10ā9
29. 2929
ā¢ Stubline
š¶ =
šš 2
šæ
Micro-grid challenges for real-time simulation
SI PU
Nominal power 100 kVA 1
Nominal voltage 600 V 1
Nominal frequency 50 Hz 1
Line impedance 1.1 mH 0.1
Capacitor
conductance (50 Hz)
2.18 ĀµF 0.0025
Capacitor
conductance (5 kHz)
2.18 ĀµF 0.25
šš = 50 Ć 10ā6
30. 3030
ā¢ Stubline
š¶ =
šš 2
šæ
Micro-grid challenges for real-time simulation
SI PU
Nominal power 100 kVA 1
Nominal voltage 600 V 1
Nominal frequency 50 Hz 1
Line impedance 1.1 mH 0.1
Capacitor
conductance (50 Hz)
2.18 ĀµF 0.0025
Capacitor
conductance (5 kHz)
2.18 ĀµF 0.25
SI PU
Nominal power 100 kVA 1
Nominal voltage 600 V 1
Nominal frequency 50 Hz 1
Line impedance 1.1 mH 0.1
Capacitor
conductance (50 Hz)
87.27 nF 0.00001
Capacitor
conductance (5 kHz)
87.27 nF 0.01
šš = 10 Ć 10ā6
32. 32
Outline
32
- Presentation of real-time simulation
- Large power system simulation technique
- Micro-grid challenges for real-time simulation
- Distributed simulation approach
In this presentation Iāll identify difference between microgrid and large network. Also the different application of RTS in microgrid application.
1- Using RTS alone with both controller and plant simulated is pure simulation and not RT simulation.
2- If IO are used to communicate between controller and plant then it is RT simulation
3- When real-plant is available, RTS can be used to iterate different command law (very expensive DSP)
1- Once youāve bought a small DSP and implemented your controller, plant can be simulated for various test (see what happen for a 40kVdc phase-phase fault not safe for lab test)
2- HIL can also be used to test some hardware using power amplifier.
1- Once youāve bought a small DSP and implemented your controller, plant can be simulated for various test (see what happen for a 40kVdc phase-phase fault not safe for lab test)
2- HIL can also be used to test some hardware using power amplifier.
First you find equations to be solved and then you discretize them. Using fixed-step solver
First you find equations to be solved and then you discretize them. Using fixed-step solver
Fixed-step solvers are used since the model need to be synchronized with real world.
1- read input
2- solve model
3- output results
If the model to solve is too big, it might not be possible to solve within 1 time-step
One solution is to increase the time-step.
Doing so a stable model can become unstable
If the model to solve is too big, it might not be possible to solve within 1 time-step
One solution is to increase the time-step.
Doing so a stable model can become unstable
If the model to solve is too big, it might not be possible to solve within 1 time-step
One solution is to increase the time-step.
Doing so a stable model can become unstable
If the model to solve is too big, it might not be possible to solve within 1 time-step
One solution is to increase the time-step.
Doing so a stable model can become unstable
Using decoupling method, equations can be decoupled and solved in parallel. Allowing to keep a smaller time-step.
We need to go from 1st matrix to the 2nd
Using decoupling method, equations can be decoupled and solved in parallel. Allowing to keep a smaller time-step.
We need to go from 1st matrix to the 2nd
Specialized software, with proprietary methods can be used
Literature also refers to more open technique
Taking into account propagation delay in a line, two system can be solved in parallel
Rule of thumb 100km take at least 50Āµs
Smaller time-step can still be used
Dynamic of the system is rather slow, therefore 50Āµs is ok.
Taking into account propagation delay in a line, two system can be solved in parallel
Rule of thumb 100km take at least 50Āµs
Smaller time-step can still be used
Dynamic of the system is rather slow, therefore 50Āµs is ok.
Taking into account propagation delay in a line, two system can be solved in parallel
Rule of thumb 100km take at least 50Āµs
Smaller time-step can still be used
Dynamic of the system is rather slow, therefore 50Āµs is ok.
Propagation delay again but parasitic capacitors are added to achieve same parameters as DPL with exactly 1 step delay
Propagation delay again but parasitic capacitors are added to achieve same parameters as DPL with exactly 1 step delay
Propagation delay again but parasitic capacitors are added to achieve same parameters as DPL with exactly 1 step delay
When a large state is available, like a DC capacitor. If DC voltage is constant over 1 time-step, it can be decoupled.
Each converter has a controlled voltage source, and the value is obtained by injecting each dc current in a capacitor
When a large state is available, like a DC capacitor. If DC voltage is constant over 1 time-step, it can be decoupled.
Each converter has a controlled voltage source, and the value is obtained by injecting each dc current in a capacitor
Now instead of 50Āµs, Ts has to be around 500ns.
Also generation is much small machine, therefore the inertia is much smaller too.
Basically dynamics of smart-grid are much faster, requiring smaller time-step.
In the case of STUBLINE, at 50Āµs, for a 0.1pu of impedance, gives a capacitor of 0.0025pu. Less than 1% losses so it is negligible at 50Hz.
At 5kHz, it becomes 0.25pu which greatly impact results
This can be all solved by reducing the time-step.
In the case of STUBLINE, at 50Āµs, for a 0.1pu of impedance, gives a capacitor of 0.0025pu. Less than 1% losses so it is negligible at 50Hz.
At 5kHz, it becomes 0.25pu which greatly impact results
This can be all solved by reducing the time-step.
1- Many RTS use different technology for IO and for computation
2- During input/ouput, conditionning can be done. Filtering, pulse generation
3- To some extend, the whole model could be done on FPGA reducing time step and therefore losses and latency
4- CPU can be used for slower computation, like a controller at 50Āµs while FPGA is running much faster
Taking the different structure of microgrid:
Ring
Radial
Mesh
Regrouping some part of the model, it could then be distributed over different RTS.
This can be achieved if
High speed link are available
Small time-step of simulation
When flexible RTS are used, they can be used for different part of the project.
1- can be used for pure simulation and RT-simulation
2- Removing the CTRL from RTS, can be interfaced with external controller
When flexible RTS are used, they can be used for different part of the project.
1- can be used for pure simulation and RT-simulation
2- Removing the CTRL from RTS, can be interfaced with external controller
RCP simulation can be achieved if the microgrid is available in the lab.
Even a mix between RPC, PHIL, etcā¦
Some part of the circuit can be simulated, some can be real, interfaced with power amplifier