Digital Twin for IED Testing: Comparing Physical and Virtual Device Performance

Paulo Sérgio Pereira Júnior is the author of the work comparing performance between digital and physical devices through closed-loop testing. Paulo, the floor is yours. Good morning everyone, my name is Paulo Júnior, I’m a director at Comprove.

So we’re going to present this work today regarding the Digital Twin and how the performance and test results are with this Digital Twin compared to the same physical IEDs.

This work will show the results we achieved by physically testing the IED and testing it through the cloud of its digital replica, comparing these results. Initially, we’ll talk about what a digital twin is. The digital twin is a digital replica of a physical device or system. That’s the definition of the digital twin.

So we have our system here, and we have it here in its digital replica. What benefits does this bring us in terms of IED design? What benefits will this IED design bring us? It’s possible to validate protection adjustments in advance. We can reduce our TAF and Tac time. It can also be used for team training, in addition to performing closed-loop tests. So we have the condition of virtual IEDs that are already a reality. We’re seeing its use and implementation. It has hardware independence. You have an unlimited number of signals. So you’re not susceptible to the number of physical inputs and outputs that device would have.

And for the testing tool, we have the PSM, which is the Power System Simulator. It was created by Comprove over 15 years ago, and it allows the realization of transients in the device, so physical device testing. It simulates real situations. It has tests with virtual IEDs through COTEC Digital Twin, and it allows the realization of open and closed-loop tests through the step-by-step method.

So how does our closed-loop work in relation to a real-time simulator? For example, to do a test of a voltage and speed regulator, it’s necessary to work with a real-time simulator, and we’ll understand why. Initially, in our modeled system, we have an interaction as follows: we have here, for example, terminal voltage, current, speed that reaches our voltage and speed regulator, and this voltage and speed regulator returns to our modeled system, for example, mechanical power and field voltage. These variables are analog variables that change at every instant in time. This doesn’t happen, for example, for our protection. When we’re going to test a protection IED, it’s possible to do this simulation in steps because the behavior and interaction are different.

So our modeled system where we inject and pass voltages and currents to this protection IED either in analog form or in the M50 environment through SLE values, and the return of this protection, the communication or interaction occurs in the form of Trip. That is, the protection IED can interact with our modeled system in two ways: either it will send to open the breaker or it will send to close the breaker. In this way, our step-by-step simulation in closed-loop can give the same results as a real-time simulator.

So we’ll understand exactly how it works. Initially, the system simulates with a condition here. There’s a normal condition and then a fault closure. This condition is simulated. Then this condition is generated. This signal is injected, and the IED response is captured. Here we have the IED response. With this response, in an intelligent way, the system will then simulate again, taking into account the capture of this breaker actuation and simulating with the time already modeled for the breaker opening. And this. In the end, we have this new waveform that is injected again and the final IED response. So this last generation is exactly the same generation that the real-time simulator has injected into the IED. We’ve even done tests with real-time simulators from UNIFEI. We did a work on this that was disclosed in another STPC in the past. Here our system compares, and this is only possible due to the repeatability of the protection IED.

So let’s understand a little more about this testing tool and compare until we reach this new ecosystem of our Digital Twin. So our testing tool will model the entire system: equivalents, breakers, TPs, TCs, including TCs entering with their saturation curve, whether flux by current or voltage by current. And it can even be done by surveying in the field through our CS7122 equipment, which captures the saturation characteristic of the TC and exports the file that can be imported here from the PSM so that it can be simulated as the real condition of what was surveyed.

And how does the interaction occur here with our IEDs under test? So the testing tool exports analog signals for each of the IEDs. It also exports digital signals that are the positions of our breaker, and it captures the Trip, the opening or closing command of the breaker from the IED. In a real physical condition, we have the physical devices here, IED1 and IED2. Now in a virtual device, a digital twin, we no longer have this physical condition, and this communication happens where the IED is located in the cloud, and this exchange of information occurs through the internet, where inside the Server, the digital replica is running. The information is passed to this server so that it processes it, and then this server returns the information to the testing software with the results, and this is done in steps, several times, until we reach our conclusion.

So this is the proposal, and the idea is to compare these results afterwards. So in terms of methodology, let’s go. We have the PSIM, which is the transient simulation software connected to a Comprove CMS 710 panel. This panel is injecting the currents to do the test here of our line differential 87 R86 current for both terminals. The actuation contacts are received here, sent by the panel. This is in the physical methodology of the physical IED. And we have here in the virtual part an API of the Digital Twin where the software will talk to this API, and the digital twin that runs in the cloud on a server will respond in this condition.

So we replaced a physical condition where I would have a physical test panel and physical IEDs with a virtual condition where the relay is running in the cloud, and I don’t need a test panel there. So basically what we have here in both cases is our Power System Simulator software.

The question is more: is the virtual IED a faithful copy of the real IED? It has the same behavior, has the same algorithms, the same functionalities, the responses are identical, and I only see one way to answer these questions: what are we going to test? So comparing the result of the same condition in physical IEDs and virtual IEDs, and let’s see how we had this condition.

So on top of this, we placed here a conventional environment comparing with Digital Twin in a 500 kV situation, an 87 STL line differential protection, and 260 cases were run, divided into 10 groups, and these cases are the usual, traditional cases used to do, for example, a protection study, a real-time simulator. Here we have our system, the line that was modeled, breakers, TC, including the TC model here with real condition with its saturation curve, a burden to vary this TC burden to force it to saturate. Here we have the fault conditions and the equivalents of both sides.

So what were these cases? These 260 cases were separated into 10 groups. Initially, there were 50 cases for internal faults: a three-pole reclosure with and without success, varying the fault location throughout the line in 25% steps. Then another 50 cases of evolutionary internal faults with an angle of incidence of 0 to 90 degrees, with bipolar reclosure with and without success, and the evolution occurring after the start of the fault cycle, one cycle later. Then internal faults on the line were done with variation of the fault resistance from 5 ohms to 200 ohms. Then another 10 cases with external faults to verify that the 87 should not operate. Then 10 external faults followed by internal faults, so the fault started external and after six cycles became an internal fault, and verifying this behavior in this condition with reclosure with and without success. Also, 40 cases of external faults with burden resistance varying, aiming to cause both light and heavy saturations. Also, 40 cases of faults with external faults followed by internal faults, also varying the burden resistance. Then 12 cases here verifying the terminal, SE1 substation 1 and substation 2. And 10 more cases here varying the frequency of our system from 57 Hz to 72 Hz in the middle of the line, verifying the behavior of our protection for frequency variation. And here 8 more faults without communication, that is, the line differential would not operate to verify the actuation of the other backup protections. So there were many cases there, we went through each one.

And let’s make a comparison here, take a case for us to compare: a phase-to-ground fault at 0% with an angle of incidence of 0 degrees, a reclosure without success, and we repeated this case, for example, 20 times so that we could do a statistical analysis of the IED behavior. So here is our fault case. We have here the currents of substation one on top, the currents of substation two, and here our signals coming from the IEDs. So the first four signals are from the IED substation one, then from the IED substation two. Here we can see a zoom here in the fault waveform. Notice that it was a condition with a reclosure without success. When it reclosed, the fault continued, and we have here below in the digital signals: first trip of substation one and actuation of differential 87, then trip of substation two and differential actuation, then in pink we have the reclosure command of both substations. Then we have again the Trip 87 and already the actuation of the STIF of both sides in both substations. With this character of repeatability, we set up a table to do a statistical analysis of these results of that case in detail.

So let’s focus here on the condition we have in yellow: the value of the analog signal, that is, here we have the physical relay and on this side in green the virtual relay. And let’s compare here, for example, before the reclosure, the average trip time for the physical relay was 14 ms. Already for the digital relay, we had a response of 10 ms. So we have a delta of 3.71 ms. The digital relay was faster than the physical relay. Then in substation two, we have here 14.5 ms, the average time of the physical relay, that is, injecting analog, and we have here in green 10.32 ms in the virtual relay, a difference of 4.22 ms and 3.71 ms as a result there of our tests.

So our test setup that we had in the physical ones here are the two IEDs here, the physical test equipment that we need. So both the physical test equipment and the IEDs to perform the test in a conventional system. And here in a Digital Twin, we have the software, which is the testing software, and here the Digital Twin software. You work entirely in a virtualized way.

What are the conclusions? So from this work, we have here 520 cases were tested, that is, 260 cases on the physical relay and repeated these 260 cases on the digital twin model. Then we had the condition of variation of the Trip between the methodologies. This variation was around 4 ms. This was the difference in the closing of the physical contact, that is, why did that virtual relay become about 4 ms faster? Because it doesn’t have the time of the physical relay to close. This was the condition that was found in relation to our tests. It was demonstrated the testing tool showing this condition of closed-loop interaction in steps and hardware independence. So the testing equipment that you have there, you’re not limited to the number of current outputs that that testing equipment has, voltage outputs, binary outputs, etc. The equipment is also virtual equipment. It’s equipment that depends only on software to be tested. So you don’t have the physical equipment. You have a solution that is software to test that virtual IED, that our Digital Twin, being then a more economical solution. You don’t have to invest in physical devices, and that’s what I would like to present to you. Thank you very much for your attention, and we remain at your disposal. [Applause]