Hello, my name is Steven. Welcome to Hyfindr Tech Talks, where we aim to understand the technology that makes the hydrogen economy work. Today we are going to talk about leak testing bipolar plates. This is a very specialized topic, but I'm very happy that for this we
were able to get an expert who works at MACEAS. They have been specializing themselves in this over several years. He himself has piled up, I think, an engineering degree, he has a an apprenticeship in this area and business degrees on top of that. He's leading all the efforts that Worthmann and MACEAS are driving in this field. I'm very pleased to welcome Daniel Schönbohm. Hi Steve, hi everybody. Nice to meet you, nice to be here. Hello Daniel, thank you for coming all the way to talk to us about this and just to kick things off, what is leak testing for bipolar plates? Leak testing is part of the quality assurance and you want to test if your bipolar plate - after the wedding process - is leak tight. If it's tight, it's good. Okay so if it
doesn't have anything, well that's pretty obvious. Who uses this, who does this? Usually, the fuel cell manufacturer or companies who build bipolar plates for the fuel cells and for electrolyzers. These are the main customers in this case. Okay so we are essentially very close to - let me say - what can form any heart of the hydrogen industry, electrolyzers, fuel cells. Both sides: Using and producing. Both necessary parts of the production process. Can you please take us a
little bit deeper into the process. How does this work? I mean it sounds so high up there, but take us to the top. Generally speaking it's very simple, but the devil is in the detail as you know. I have some pictures with me. So this is our, I would call it, standard machine, so we are machine builder in general. And we are
focusing on leak testing, so this is our machine which we developed especially for leak testing bipolar plates. It's a double station machine and it can test four bipolar plates at the same time. Double station means one station here one station there? Correct. So it is a human being that will be standing there? In this case yes, a human loading the bipolar plates and unloading the bipolar plates, but the process itself is running automatically. If we go to the next slid now, we can see it more in
detail. So we have in each station, which are independent - you can run both sides fully independently. You can test the plates here, you bring the plates in here, in this and this. You have always two plates in the machine. Then it's
closing and the process starts, then you can't see much anymore. So that stamp goes down and then that works? Right, then the leak testing itself starts. So of course, we are curious to see what's happening inside there. Yeah that's true, but unfortunately from the outside it's not possible to see it, so we can just - this is a view from inside, you see the other side on the top here, but in general it looks like that, more or less. So this is essentially what we saw in the photo, like the top view of one of those? Yes, for example this side here. Now we're looking at the top here.
So at the outside we have the sealing of the chamber, we call it the vacuum chamber. The gray one? Yes. We are working with helium, you are using a tracer gas like helium to detect leaks, so we have a mass spectrometer installed which is detecting the helium atoms coming out. Therefore we have a vacuum chamber, because we want to detect very small leaks. That is only possible with this tracer gas like helium
and in a vacuum chamber. So that's how it works. Just to explain it a little bit more in detail: We put in the parts. Either a human or a robot. Both is possible, we already built both. They put them in, then the machine is closing. Then we have the sealings on both sides of the tool, so we seal the three circuits. We have the sealing of the outside, we have the sealing for the port of the cooling and we seal the complete area for the anode and cathode side. Oxygen and hydrogen side. So you have a sealing on the outside here, you have a sealing for this channel and this channel? And then also around. To have the flow field closed. So then we create everything. We
put - you can see - 0 bar everywhere, we need a very deep high vacuum, in order to find this small leaks. So just to understand again why we are seeing two plates here, because in the machine there are always two plates being tested at the same time? That's a question of cycle time. So you double your cycle time? Correct. You can also test one. For the explanation of the process, only one would also be fine, but here we have a double? Okay, so now we've understood, now we have zero bar, really zero zero? Nah, nearly.
It's very low, maybe 0.001 or something like that, but very, very low very low vacuum. It's not not 100% zero, because it's not possible, but it's a very low vacuum. Then you have a low or zero vacuum in there, in this channel, in this channel and in this? Yes, everywhere. Then the next step, that's a very important step - and the longest step in the whole process - we have to check what is the helium background. So we create a vacuum everywhere, then we have
to check with the mass spectrometer how the helium background is, how much helium is still inside. Because it needs to be lower than what we want to detect afterwards. That's a very important step. So you kind of like you know how much there is, so you have a reference point. Yes! Then we put it on zero an then we put helium in and we can detect what is coming out. So this is a very important step for the whole process, so then the next one is filling, for example, in the first step it is always filling the cooling circle with a mixture of helium and air. So we have 90% air 10% helium. With a certain pressure, usually for the fuel cell bipolar plate it's up to two bar. In most of the cases it was around 1.7 bar. With the graphite plates it's a little bit less - 0.7, 0.8 bar, something like that. So we put
the helium inside the cooling circuit first. So we fill it completely with helium - or with the helium air mixture in this case. Then we have our mass spectrometer which is then opened and connected to the outside of the chamber to the oxygen circuit and to the hydrogen circuit - so anode and cathode side. So now the mass spectrometer is looking into the chamber, into the circuits. If there's helium coming from the cooling circuit. There's always something coming, there's always a leak, but it depends on how big or how small. That's why we have our limit leak rate. Which in this case is - we call it -
one times ten minus five millibar times liter per second. So 0.00001mbar*l/s. Thanks for writing that out! So mbar*l/s. This is our standard picture/figure and usually the specifications are one by 10 minus four, but we have 10% helium that's why we go down to 10 minus five. And that is really the limit. Due to helium background, due to the speed of this process. To go lower than that you'll need much more time and it will be really difficult. But this is what we found out - the limit of this process. And now the mass spectrometer is detecting the
helium atoms coming out, it's counting. It's counting how much helium there is? Correct, and then we know exactly the leak rate for the outside, for the chamber, what is going out of the plate. And also to the other two circuits. Then we will repeat it for the other circuits. I haven't shown it yet, but it will be the same we get everything out. We evacuate everything, we do the helium background check again and then we fill the oxygen side with helium. So coming back to the design. Essentially what you did here, you put the helium in there and then check what's on the others and you then take the helium out of this? Yes, everything to zero. So you go
back to zero? At the beginning. Like we saw in the first picture? Yes, exactly. Then you do the background check again. Very important because we need it as another new zero. Then we fill not the cooling circuit we fill the
oxygen circuit or the hydrogen - doesn't matter in the end. Then we check to the other side. Then, after this is done, you evacuate everything and fill the hydrogen side. So then we have three tests in one. And then we have seen all the weldings, all the glued areas, if it's a graphite bipolar plate. Then we have checked the complete bipolar plate within 30s. Within 30s you do all that? The testing itself. Of course the moving and the
handling, but the test itself takes around 30/31/32s. So you detect the leak by knowing when the mass spectrometer finds more helium than you want? Correct. If it's lower, 10 minus six for example, everything is fine nothing happens. If we are higher, 10 minus four for example, then it's a
Not Okay case. I think you brought a picture of your machine of how it does that? Yes, so this is our HMI - the screen of the machine - so you can see here this is the first part we did, the background check, so we had to have a certain helium background here, which is okay to test. Then you can see the helium which is in. The 12% roundabout. Then we have the value one, it means the cooling, oxygen and hydrogen circuit. So for every circuit you get a leak rate and as long as they are lower or smaller than you have defined, then specified, then everything is fine and you get a green light, if not there's a red light. Okay, so here we already have several greens here. Then this is not done yet. Okay, so this is how that looks. Yes, it's working at the moment. Wow that's super interesting I have
a couple of questions around that. So first of all I know you say you choose helium because it is a super light gas. Yes, a noble gas, just doing nothing and very small. So maybe a silly question, but why didn't you take hydrogen, to test what is going to be used for hydrogen later anyway? It would be a good idea, because you run it with hydrogen in the end. The problem is, if you use hydrogen, to use 100% hydrogen is very difficult, because of exploding areas and so on. Then you can use e.g., the, as we call it, forming gas. This is 95% nitrogen and 5% hydrogen.
But then you have to take a much lower leak rate, so we have 10% helium, then it's only five, so you have to go lower and lower. And to use hydrogen in the vacuum chamber is not possible today in this certain way, with this leak rate. In a certain way it's possible, because you don't get this hydrogen background. It's always higher than you want to test. So this is what we are working on at the moment, but at 10
minus four, there is maybe the limit. You can do it by sniffing on the atmosphere level, but in the vacuum chamber it's really difficult at the moment. Especially in this field of leak rates. Okay, so if you want the reliability and everything that we need, then hydrogen is not the best thing to use to test it? Not today, hopefully the developments are getting forward so that we get the possibility to use it, because helium is expensive, it's not endless.
So hydrogen would be a good solution, but as I said, 95% nitrogen 5% percent hydrogen mixture, leak rates, vacuum. So I think you've said one word that I also wanted to ask you about, in terms of expense. I know that helium is an expensive gas and you use it for testing. So you create the zero vacuum, then you blow it into one of the channels essentially and then you take it out again. What happens? Do you then blow that same helium into another one or not? No, it's gone unfortunately. That's why we use only 10%, so we have this mixture, we have a mixing unit and there we mix it to 90% air 10% helium, because this 10 are gone. They are going out to the exhaust and that's it. You can also use a recovery unit.
But that's also only 90%, so in the end it's the same result. Okay. So essentially that gets ejected? Ten percent are gone, yes. Just out of curiosity that goes into the air? Yes, we have to bring it out of the factory hall where we have the machine because you don't want to have helium everywhere to influence your measurements, that's why you have to bring it outside. Okay, so that means essentially in this process you are using a bit of helium as you go along? Yes. Okay. Everyone talks about the stacks being also an important cost element in the whole thing. So I believe, if you are doing this as a quality assurance this will also influence the cost. Can you tell us a little bit more on that side, the cost side and how
things are going? So unfortunately the machine is not the cheapest machine on the earth, it's true. We have a lot of equipment inside, like vacuum pumps, four in total we have the mass spectrometer which is high in cost the machinery around, mixing unit and everything. So they are the cost for the machine at one time and then of course wear parts like sealings and everything and of course the helium consumption of course. If you have a high capacity, a lot of parts coming out then you need the helium and the worker on front, that's why, the next picture will show it, you can use also a robot. Fully automated. Can be fully automated? Yes, then you can of course reduce the cost for the worker. And that's why we've also
develop this machine. We built this machine, it's running more than one year now in Sweden at a big bipolar plate manufacturer. We built this machine, fully automated and the plates are coming in here with the box, we have a vision system here, detecting or seeing the position of the plates, the robot is grabbing the plate and bringing it into the machine. If they're okay they get out to another conveyor here, that can bring it to the next process step.
If not we collect them here. Okay, so that is essentially - also in terms of an outlook - that you use obviously robotics to help. One thing I forgot to ask, when you find the leak, what happens then? If you find a leak of course you get an alarm, a notice on the screen, that you have a Not Okay case. A red blinking one. You know the one of the two which one it is? Not yet, that is the problem. We know the station yes, left or right station. Maybe you know also the channel cooling or oxygen. But you don't know if it's the left or the right
plate in this vacuum chamber. That's the problem if you test more than one plate at the same time. Then you have to do a retest. In the same tool you have to put in the dummy. We call it dummy it's just a metal plate or whatever. To have it closed this side. Then we retests the two plates again. Then you know this one is okay, this one is not okay. Because you can have everything. You can have both are okay, because it's always the sum of leaks. In
the helium vacuum technologies, it's the sum of leaks. Both can be okay, both can be not okay or only one can be not okay, the other one can be okay. So every case is possible. So the first step is to find out which one it is. When you do find it out, do you know where the leak is? Unfortunately not. Okay so you don't know, so now what happens to that bipolar plate? Because we are in this vacuum chamber, so you cannot see where it is. As I said it's just the sum of leaks which you measure so they have to find out which one of the two it is, but then they have to recycle it or I don't know what they are doing. It gets marked. Yes, they get it marked as
Not Okay, we put it in the red box for example. And then the manufacturer has to do something with it, whatever it is. I don't think they will throw it away, because it's very expensive. But you will not find where the leak is. You can sniff it for example, then you can go over the the welding seams, but very complex. That is probably a bit of a manual effort. Plus you need to then repair and that's not worth. Well we have on this channel actually some videos on
the manufacturing of bipolar plates also very interesting and a hell of a complex process, but I think it's really good to know now how we can leak test these. Well it's part of the process. That's obviously part of the quality assurance and making sure that these things work for a very sometime, isn't it? Yeah, that's true, and hopefully for the future it will grow, so we have now sold eight machines in total. One of them is running fully automated,
other customers who already bought the second one now. So they want to grow, to have the fully automated line for, we talk about a cycle time of three or four seconds now. So maybe in a few years it's one second because they always talk about it. Three or four seconds? Even though it takes 30s to do the testing? So this machine which you can see here with the robot. This has a cycle time of 12s in total. In total? Yes, coming in and coming out. And then, of course, if you want to achieve six seconds, four seconds then you need a second one, third one, because the process cannot be faster. Either you have lower
leak rates to detect or you would only test the cooling circuit for example or you only test 10% of all the bipolar plates whatever. Maybe it's for the future, it could be a way not to test 100% of the plates, but then your quality must be better. Wow, but thank you Daniel for giving such an outlook already. It was a pleasure.
Got me all excited now. A 12s cycle time. That's an announcement on time and also an announcement at the time, that we have run out of time. Unfortunately we have, but I want to thank you very much for coming and explaining this topic to us, I have learned a lot. I hope you too have watched and learned something as well. If you like it please leave a comment. You can follow us on this channel, we appreciate that.
Also give us a message on LinkedIn or subscribe. And also I need to tell you, that you can find people like Daniel on Hyfindr.com, with all the systems that you need, all around us. For the manufacturing and also later for the development of actual fuel cell systems or electrolytes or whatever it is that you want to do in the hydrogen economy. We wish you well with it and want to thank you for taking the time to watch and thank you Daniel again. Thank you. Have a lovely time thank you bye.
2023-09-11