Using HIL and P-HIL for cost effective and reproducible testing of power electronics with OPAL-RT Germany GmbH

Show notes

Real-time simulation helps accelerate development and derisk early implementations in the race to transform to e-mobility and renewables. (Power) hardware-in-the-loop plays an important role. Among notable current needs is to simulate the new scenarios presented by grids with converter-driven energy sources, not least in testing for resonances, or black start conditions. For real-time simulation - particularly integrating power hardware - this presents new challenges to balance stability, accuracy, speed, and scale.

Prof. Marco Jung talks to Ravinder Venugopal, Managing Director at OPAL-RT Germany, about the issues and approaches (from parallel processing to solvers and FPGAs) in episode 3 of the"Sound On. Power On." podcast.

They also discuss the future, with digital twins harnessing live data to improve real-time simulations, with the support of AI.

Get your PCIM Europe ticket here: https://pcim.mesago.com/events/en.html#conversionteaser

More about OPAL-RT Germany GmbH: https://www.opal-rt.com

Show transcript

Recording:

Recording: Sound on. Power on. Your power electronics podcast, powered by PCIM Europe.

Prof. Marco Jung:

Prof. Marco Jung: Hello everybody, and welcome to the third episode of Sound on. Power on. Your power electronics podcast, powered by PCIM Europe. My name is Marco Jung. I'm a Professor for E-mobility and Electrical Infrastructure at the Bonn-Rhein-Sieg University of Applied Science at Sankt Augustin. As well as Head of Department, Converter and Electrical Drives at the Fraunhofer Institute for Energy Economics and Energy System Technology at Kassel. Both are located in Germany.

Prof. Marco Jung:

Prof. Marco Jung: Additionally, I'm the chairman of the IEEE Joint IES/IAS/PELS German Chapter. And today our technical theme is using hardware-in-the-loop and power hardware-in-the-loop for cost effective and reproducible testing of power electronics. And this I will discuss with Ravi Venugopal from OPAL-RT. Hi, Ravi. How are you?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Hi, Marco. I'm doing well, thanks. And I'm very glad to be on this podcast with you, on Sound On. Power On. to talk about PHiL and PHiL.

Prof. Marco Jung:

Prof. Marco Jung: Great Ravi. All that sounds really nice. And I see you're really great. So, where are still you located now?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Well, I'm in Nuremberg, right where the PCIM Europe conference will be held in just under a month. It's where we have our office, the OPAL-RT office for the Dach region. And I'm proud to be leading a great team that's a little part of the power electronics revolution that's underway.

Prof. Marco Jung:

Prof. Marco Jung: Great. You're still next to me, but I think we will see us then, on the PCIM located in Nuremberg. So, maybe some information for our listeners. Ravi and I working since, let's say two years together in realtime simulation and power hardware-in-the-loop for teaching and education. So, it's really nice stuff. And colleagues of mine at Fraunhofer IEE have a longer history together with OPAL-RT and with Ravi.

Prof. Marco Jung:

Prof. Marco Jung: They have done several research projects on realtime simulation and power hardware-in-the-loop. And I think so Ravi, I'm looking to your face and I see your whole business life, you have worked let's say, in realtime simulation. Please give us... Yeah, let's say a briefly overview of your experience and maybe of your company, OPAL-RT.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Yes indeed, Marco. Yeah. So, I've been involved with realtime simulation from the time I was a master's student about 30 years ago. Over my career, I've seen it grow in from its origins in the aerospace industry to becoming dominant in the automotive industry. And now it's playing a key role in the energy industry, to enable the transformation to renewables and with the focus on de-carbonization. So this is really... I think it's a tremendously interesting topic that's evolved over the years and found new applications. And there are so many examples of catastrophic software failures. Such as for example, the 2003 Northeast blackout in the US. There was the loss of the Mars orbiter in 1995, there was an Ariane 5 rocket failure in 1996. And you could argue that many of these failures could have been avoided with comprehensive hardware-in-the-loop testing.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And maybe I should say a little bit about in the loop testing when I mention it here, for our listeners who are not aware of the technology. So, the basic idea is that you connect your real equipment, your controllers, your automation, your protection to a realtime simulation of the system that they're controlling or protecting. And this device doesn't know it's not in the real system. And you can do extensive testing based on realtime simulation of the plant that it's controlling. And the nice thing is it's a simulation, the risks are low. And you can do a whole lot of tests that might not be feasible in the real world. And so, you can test in a very comprehensive way going beyond what you could do in a lab or a field test. And this really de-risks the implementation of your control in automation.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Now coming to OPAL-RT. It's been a pioneer in using realtime simulation for hardware-in-the-loop testing for power systems. Its origins are in Hydro-Québec, which had a really strong real simulation group or lab, back in the 1970s. And this was driven by need. In Canada, you have a lot of hydro power generation that's remote from the load centers and the cities. And you have equipment deployed in rather difficult to access places. And this was a problem because in the middle of the Canadian winter, driving 600 kilometers to fix something, you didn't find that many volunteers to do that. And so Hydro-Québec really started doing a lot of hardware-in-the-loop testing before they deployed the equipment.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And this was done, before the nineties, using first analog simulators and then supercomputers because of the computational requirements. But then as we got to the nineties computing power started increasing rapidly and our founders, Jean Bélanger and Lise Laforce, who were working at the Hydro-Québec realtime simulation lab decided that they would start OPAL-RT to use commercial off the shelf processors. You know, the more powerful processors that were coming out of Intel and so on. And well, each one was not powerful enough. If you hook them up to compute in parallel, you could start addressing some of the computational requirements for this. And this was how the company started in the nineties. And yeah, from there, we've moved, we've grown in the energy sector, of course. But also in the aerospace and automotive sectors.

Prof. Marco Jung:

Prof. Marco Jung: Great. I think we can learn a lot of what's happening in Canada as a best practice, because Germany will implement in future a lot of LNG and hydrogen power in the infrastructure. And I think in Canada has done it in the past, maybe we can have a look to Canada from Germany's side.

Prof. Marco Jung:

Prof. Marco Jung: Great. Before we start now to go a little bit deeper in retail simulation and power hardware-in-the-loop and so on, please, Ravi, I want to know from you what technology innovation at OPAL-RT or technology step made the most impact on your life?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: So, that's a really interesting question, Marco. So from my side, I truly believe that OPAL-RT's contribution to the simulation of power electronics, especially using FPGAs to handle the fast dynamics, has been revolutionary. Beyond the ability to just simulate these systems, which is a challenge in itself because of the fast dynamics and so on and so forth. It's also built a software environment where an electrical engineer can test out power electronics controls without knowing anything about FPGAs. They're not FPGA programmers. They're not experts in this field. So, what we've been able to do is create a software environment where a power electronics engineer can just create the schematic of the converter or inverter that they want to test the controls for. And just in effect, with a couple of clicks, have the simulation running in realtime on the FPGA.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: So, I'd say this technology has had a big impact and will have a big impact with E-mobility with the de-carbonization, power electronics are just... They're going to continue to grow. And rapidly, because of everything that's happening across the world right now. And just to give you a couple of examples, I'd say a good one is SNCF, the rail operator in France. And recently, they've looking at having their trains run at 350 kilometers an hour, instead of 320. And they were upgrading their power electronics and the controls for this purpose. And they used our systems and EHS, our power electronics simulation suite, to do the testing of their controllers. And we've got other companies. For example, G, that's used our simulation platforms for a range of applications, from energy systems to marine applications. So these are a few examples, but I would really think that coming to impact, it's been the contribution to power electronic simulation.

Prof. Marco Jung:

Prof. Marco Jung: Your last sentence now leads us in our topic for today, let's say, for power electronics. And in the last episodes of Sound on. Power on. I discussed with Frede and with Peter, let's say, the future trends in power electronics. As well as using wide band gap devices in power converters. Now, it's for me really important, and I think for our listeners too. Can you give us a look inside for what we can use realtime simulation in this context, with power electronics? And what are the advantages of hardware-in-the-loop?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: So, looking at the way the power electronics industry is evolving, as you said wide band gap devices are going to make a big difference. Everyone's looking to have better power to volume, power to weight ratios. And they're going to higher switching frequencies, using wide band gap devices. And this in itself produces some challenges. You do need to have controls that are operating at faster switching frequencies. And the applications range from E-mobility to energy systems, to even commercial consumer devices.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And so, that's one side of it. That your controls are going to have to evolve to be able to implement these wide band gap devices. On the other side, for certain architectures like dual active bridges for resonant converters, the controls have to have very specific characteristics. And testing all these controls for wide band gap device based converters will require a lot of testing, a lot of development. And here's where hardware-in-the-loop can really play an important role.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: First, it allows you to do extensive testing, de-risking the implementation on the real hardware. The second is also that you could start your development before you have the actual, physical hardware. So you can get started. You can start doing a lot of development work with the realtime simulation. And then as you get closer to deploying your controls, you can do extensive fault testing, fault conditions. You can really cover the range of scenarios that the controls would be working in, before implementing them. And I think this de-risks and accelerates the process. And it's a competitive industry. You need to get out there as fast as possible.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: That's one side. And then you can also do work with power hardware-in-the-loop. And I know Marco, that's one of your interests also, to test the complete inverter and converter assemblies, including the controls. And this opens up another range of testing. So not just the controls, but the complete converter system.

Prof. Marco Jung:

Prof. Marco Jung: Ravi, as you said, there exist several application possibilities for realtime simulation. Let me add one topic. Big power grid simulation, which needs a high computing power. Phase simulation are in the area of 100 milliseconds up to one to three seconds. EMT needs smaller step size than let's say, 50 microseconds. But for switched model like power converters, using wide band gap semiconductors like for example, silicon carbide or gallium nitride. So step size needs to be done in the area of nanoseconds. I know OPAL-RT uses parallel computing, or better, a multi-core simulation strategy. Can you describe your approach in more detail and what are the limits?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Yeah. Great question, Marco. So, I'll handle the first question first, where you were talking about grid simulation. And this is definitely one area as you start having more converter driven, distributed energy resources in the grid. You do want to see what the impacts are. You want to see how they interact. You want to see if there are any resonances. So, from design of filters to just seeing how the network behaves. And also when you have to perhaps study things like black start conditions with converter driven systems. So there's a whole lot of grid simulation that needs to be done. And yeah, with large grids, of course there's a lot of computation that needs to be carried out.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And as I mentioned before, we've been working on this for many years. The structure of OPAL-RT has been built around parallel computing, the structure of our platforms I should say. We cover both phaser simulation in the order of milliseconds and EMT, as you said, in 50 microseconds or below. For this, we use multi core CPUs. Again, it's not always trivial to do this, because there are latencies in data transfer, which can compromise the accuracy of the simulation. So you have to have specific numerical method. You have to have ways of decoupling the system in such a way that it makes sense. And this is something that we spend a lot of time on. But there is a limit. We can get down to maybe five microseconds, 10 microseconds, something like that. But there are many applications that have used these CPU configurations. ABB has done a lot of work with us for their medium voltage drives. But now, as you said, as we going to wide band gap devices, as we're going to really high switching frequencies, we really need to be in the sub microsecond, nanosecond range.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And the only way that we found to do this effectively is to go to FPGAs, where we can assimilate relatively large networks or numbers of switches on FPGAs. So, we can do right now with our most powerful platform which uses a Xilinx Virtex-7 FPGA, we can get up to about 288 switches in under a microsecond. But really, to do effective realtime simulation, you need to be running at about a hundred times the switching frequency. And so there are challenges. I mean, we are hitting the limits as we get into the hundreds of kilohertz type switching frequencies. We need to find ways and it involves everything from using multiple FPGAs, parallel computing, to finding better solvers, to trying to identify which effects are really important when you do testing.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: So it's a big challenge. There's a lot of work to be done. And this is where we also look to work with partners and universities and collaborating to see how we can develop this technology.

Prof. Marco Jung:

Prof. Marco Jung: Great. As you said it now, what do you think about to approach or using an improved solver versus small hardware? What do you think about what is a better solution for the future?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Yeah. So, this is a really good point and I briefly mentioned it earlier. And so, we have to work on both sides really, to address the challenges that are coming up. With respect to solvers, especially for switching systems, we do have to find good ways to model switches. Because again, roughly speaking with each switch, you can have two sets of equations. If you have N switches, you have two to the power of N sets of equations. And this can grow very rapidly. So, you need to find good ways of modeling your switches.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: We have particular solver that... What we call a fixed admittance matrix solver for our eHS generation four solver, that's the earlier generation. And it enables us to simulate large power electronic systems, but it also has a limit in terms of accuracy because we're making approximations. And so there's always this trade off between accuracy and speed of simulation. And then, when you look at more hardware that does provide you the ability to go bigger but then there are issues with latencies, with data transfers. Because once you're talking nanoseconds this sort of parallelization is not just a matter of adding more hardware power. That doesn't work. You have to be smart about the way you split the system. So my answer to that question is, it's really we have to work on both sides to meet the challenges that are coming up on the future.

Prof. Marco Jung:

Prof. Marco Jung: So, change our focus. I think one big advantage for hardware-in-the-loop is power hardware-in-the-loop used for reproducible testing. But what is about the latency between the realtime simulator and the amplifier? I have some experience from the past. Do we need new interface requirements? I think that will be a hot research topic in the future. Maybe together, for us.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Oh, absolutely. I mean, it is a hot research topic. And power hardware-in-the-loop is always challenging. Because at the end of the day, you're simulating something, you're adding in power amplifiers, you're creating latencies that will not be there in the real world. Latencies automatically imply stability issues.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: But there are a couple things to keep in mind. The first is that when you're testing an entire inverter system, usually what you're doing is you're simulating the grid or some other device on the realtime simulator. And the dynamics there are in the tens of microseconds, so it buys you a little bit of time in terms of handling the latencies with the power amplifiers and so on, are typically in the range of 10 microseconds or less. They still make a difference. And you still have to be careful about the way you close the loop when you're doing grid simulation with a real inverter or converter. But it depends on the use case.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And we are spending a lot of effort in figuring out how we can design interface algorithms to enable closed loop stability of a PHiL system without losing too much accuracy. But then, there are times when we have to do just power electronics PHiL. And now your model's running on the FPGA. And now we're talking dynamics, which we're simulating with time steps in the nanoseconds. And this is especially relevant when you're doing, for example, motor emulation. You have your inverter, and then you're emulating the motor on the FPGA, and you have the power amplifier in between. That's creating latencies of a few microseconds. And in this case, it is very tricky. And it is a research topic that we are looking at in detail.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Or another use case might be you have a converter driven system for the grid with multiple converters. You have one real converter, but then you have multiple simulated converters on the FPGA. And then again, the latency for the PHiL is an issue. So to be honest, we don't have all the answers right now. And it is a research topic. And definitely, we'd be very happy to work with you and your group on this.

Prof. Marco Jung:

Prof. Marco Jung: Yeah. Great. And I think for this theme latency, you have a topic of a paper for the PCIM conference. I think it's dealing still with PV testing, right? Can you give us a short overview of this?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Yes. Quite right. So, the paper talks about designing an interface algorithm for testing PV inverters in a power hardware-in-the-loop setup. And depending on the filter configuration in the inverter, whether it's an LC or an LCL, there can be stability issues. And the way we typically interface a converter and inverter in a PHiL setup, is you add a low pass filter to help with the stability. You basically try to filter out the high frequency current feedback, so that it doesn't trigger any instabilities.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: But when you do that, you are obviously compromising your accuracy. And you need to have a trade off between stability and accuracy. And this paper deals with the particular method of addressing that stability, accuracy compromise, and talks about a method in which we can do this. So, it's both analytical and it also shows experimental results in an actual power hardware-in-the-loop setup.

Prof. Marco Jung:

Prof. Marco Jung: Great. So that we will see from your side at the conference. And I think they have a booth at the fair. Can you explain us a little bit more, what we can see there from OPAL-RT?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Oh, yes. It would be my pleasure to do so. Yeah, of course we would have a booth at PCIM in Nuremberg. So this is where we're located for the DACH region, as I mentioned earlier. And it just an amazing show with so many exhibitors. It's so big, both in terms of the trade exhibition and all the papers and the community that comes there. So yeah, we'd be very happy to welcome all our listeners and you to come by the booth and visit us.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: There's going to be a lot of interesting things. We have our experts, of course, that will be able to talk to you about hardware-in-the-loop simulation and the technologies we are developing. If you have questions based on what you've heard in this podcast, feel very welcome to come by and talk to us. So, first of all, there'll be people who have spent a good part of their professional careers studying and Hil and PHiL for power electronics.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Secondly, I think you'll be able to see some demos of our products in the space. Especially our eHS gen five power electronic simulation system. Which enables you to test systems with high switching frequencies. Tens, hundreds of kilohertz. And also topologies such as the dual active bridge and some of the newer types of converters that are being deployed in the field. So, I think that anyone who visits our booth will be able to get a really good idea of HiL. I don't think we'll be having a power hardware-in-the-loop setup because that's a little more difficult to set up in a booth. But we definitely will be able to talk to you about what we're doing in the space of power hardware-in-the-loop.

Prof. Marco Jung:

Prof. Marco Jung: Thanks. So, let us have a look to the future. I think using hardware-in-the-loop for digital twins, as well as using AI for parameter generation are really next steps. What we will see in the future?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Yeah. No, it's really an exciting time. We have huge challenges. Revolutionary technology. Digital twins are really a critical part of our technological roadmap. We see them as being increasingly important for both development and operations. And we are looking to leverage our CPU FPGA realtime simulation platforms, for them to be the basis to create online digital twins. Which would basically have realtime simulations integrated with live data coming from the field. Now, as we do this, obtaining good data to build these models for digital twins is definitely going to be a challenge. And also, validating these models as they get bigger and bigger and bigger is also going to be a challenge. And I think AI could provide a solution for this, both in terms of being able to optimize parameters or be able to choose good parameters. Because at any point you're not going to have a hundred percent complete information and you're going to have to make some trade offs. And I think AI definitely has the potential to help with parameterizing these systems.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: On the other side, I think AI could also be used to generate test cases. As we test and we find issues, it's hard to manually go and redefine the tests. And maybe we would be much faster if we used AI to use results in a feedback loop, to generate new test cases to isolate where the problem might be. Or figure out the best way to correct the fault. So I think this is going to be a big development. Really figuring out how we can use AI in the field of realtime simulation. And also, as we have big systems we're going to generate an enormous amount of data. We're going to have to find a way to analyze them and make them interpretable by humans. And I think AI is going to play a role in that.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: And so, we're actively in engaging in R&D and in advanced technology programs with both academic and industrial partners. And yeah, there's a lot that can be done. We'll have increasingly powerful tools. And yeah, our CEO likes to say, perhaps our imaginations will be our limitations with all the things that are happening.

Prof. Marco Jung:

Prof. Marco Jung: Yeah. I think it's really hot stuff. There exists a lot of algorithm let's say, for solving. But to extract the parameters, to get the correct parameters. Let's say for example, power hardware-in-the-loop, using for testing in reliability or something like this, I think it's really hot stuff. Yeah. And I think we need a lot of research in the future in this fields. Yeah, to use AI in the good way.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: The use of AI in the field of realtime simulation is at its infancy right now. We are still trying to figure out what it can do. And there's a lot of potential. So, you are absolutely right that this is a topic for a lot of research and development. And I think it'll continue over the years to come.

Prof. Marco Jung:

Prof. Marco Jung: One theme is important and a matter close to my heart. Education and training. I believe the key to the success of realtime simulation and power hardware-in-the-loop is education. My university offers various courses on the subject. OPAL-RT also are represented in some courses. Can you say something more about this, and what do you think about the future needs?

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Oh, we are definitely on the same page on this, Marco. realtime simulation, HiL, PHiL are all relatively new subjects in the wider world of power electronics. And one of the most common problems that we hear from our customers is that they simply don't have the trained resources to get into HiL or PHiL. So it's extremely important for universities to build curricula on these subjects. Just like your university Hochschule Bonn-Rhein-Sieg is doing.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: There are a few universities that are taking the lead like you are. I believe KIT in Karlsruhe is doing a little bit of this. We've been working with a couple of universities in the US, in Lebanon, in a few countries around the world. In developing programs to teach hardware-in-the-loop, power hardware-in-the-loop for power electronics and power systems. So, we are collaborating with universities such as yours to build coursework. And we do feel that it's most important that the professors and the lecturers design the courses. But we're trying to work with you, to support you with the tools to demonstrate the concepts and to get students hands on experience.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Yeah, we also see this as something that's going to keep us busy in the years going ahead, in building up education and teaching infrastructure to enable the widespread use of hardware-in-the-loop and power hardware-in-the-loop technologies.

Prof. Marco Jung:

Prof. Marco Jung: Great. Okay Ravi. Thanks for your statements and I hope we will see us at the PCIM in Nuremberg and discuss future possibilities. Thanks, Ravi.

Dr. Ravinder Venugopal:

Dr. Ravinder Venugopal: Thank you very much Marco. It's been a true pleasure being on Sound on. Power on. And I'm looking forward to seeing you and some of our listeners at PCIM in Nuremberg in May.

Prof. Marco Jung:

Prof. Marco Jung: To all the listeners, wherever you might be, thank you very much for listening. We hope you have enjoyed today's episode and have gained some valuable insights. Make sure to subscribe to our podcast, which is available on all major podcast platforms.

Prof. Marco Jung:

Prof. Marco Jung: If you would like to share your feedback with us, please do so via email to podcast-pcim@mesago.com. We'd be delighted to meet many of you live and in person at the PCIM Europe in Nuremberg from 10th to 12th May. Tickets for the trade show and conference, as well as further information are available at pcim-europe.com, or via links in the show notes.

Recording:

Recording: We hope you enjoyed this edition of Sound on. Power on. Powered by PCIM Europe. Do subscribe and share.

New comment

Your name or nickname, will be shown publicly
At least 10 characters long
By submitting your comment you agree that the content of the field "Name or nickname" will be stored and shown publicly next to your comment. Using your real name is optional.