Below is the text version for the "Hydrogen Fuel Cell Generator Demonstration at Microsoft Data Center" H2IQ Hour webinar held on February 23, 2024.
>>Kyle Hlavacek: Hello, and welcome to this month's H2IQ Hour webinar. Today we have an overview of Caterpillar's demonstration of hydrogen fuel cell generators supplying backup power at a Microsoft data center in Cheyenne, Wyoming. My name is Kyle Hlavacek with the Department of Energy's Hydrogen and Fuel Cell Technologies Office, supporting stakeholder engagement and other outreach activities.
Please be aware this Webex webinar is being recorded and will be published online in our H2IQ Hour webinar archives. If you experience technical issues today, please check your audio settings on the audio tab. If you continue experiencing issues, please send me a direct message. I'll now hand it directly over to Paul Wang with Caterpillar for today's presentation. Paul? It's all yours.
>>Paul Wang: Alright. Can you hear me?
>>Kyle Hlavacek: Yep. You're good. Go ahead and share.
>>Paul Wang: Let me know if the screen's coming up.
>>Kyle Hlavacek: It's starting to come up. There you go. You're good.
>>Paul Wang: Alright, thanks Kyle for the introduction. My name is Paul Wang with Caterpillar, and I'm here to share a successful demonstration that we had with a hydrogen data center project sponsored by the DOE, the award here ending in 9252.
About four years ago we asked a question, can hydrogen power a data center for 48 hours? And where the 48 hours comes from is from agreements that data centers might have with their customers that they would have enough fuel on-site, should there be a loss to grid power. Today, that function is being performed by diesel gen sets, which are a very reliable way of producing backup power in the case of loss of grid power. And beyond videos of cats and dogs that you might have on Instagram or other platforms, to our personal lives, and to society as a whole, data centers are critical pieces of infrastructure for—whether it's financial transactions or even national security.
And so, it's key for data centers to retain their uptime and availability. And so, we asked a question whether hydrogen can provide this source of backup power. And the answer to this question is yes.
On the right you see a fuel cell, a Ballard fuel cell, 1.5-MW fuel cell that's doing an extended run. To our knowledge, no other manufacturer has had the experience of running in excess of four plus hours at a live data center, and we've done this many, multiple times at this Cheyenne, Wyoming, site that Microsoft has, in all sorts of harsh conditions. And we've run this during freezing conditions when the temperature is below 32°F and had no issues at all.
And so, I really like this picture on the right here because it shows the products that are coming out of the fuel cell stack just being water, and all you see in that picture in this nighttime run is that of water vapor.
And what we had proposed in that solution is a 1.5-MW fuel cell plus a battery microgrid, and so this is more than just a standby-type solution. We can run this microgrid islanded or we can be connected to the grid and provide power that way as well. And so, this this solution is definitely more than just a standby-type solution that can be applied at this data center for standby power as well as peak shaving and even continuous running if you have the hydrogen source for that. And what this solution brings is increased resiliency, lower carbon intensity, and an integrated system.
We have another companion presentation in the 2024 AMR slides. If you go to that slide deck, there's going to be a lot more information on this integrated system and how the fuel cell pairs to the batteries. And Patrick for Ballard is also going to talk some more about the pairing of fuel cells with batteries for this type of application and beyond.
So going on to the next slide, there are really three pieces color-coded here. We've got the blue piece, which speaks to design, safety, and analysis. In this portion we're looking at the greenhouse gas impact, how you size this power plant, what the hydrogen infrastructure would look like on a data center—and the fuel cell does not take liquid, cold liquid hydrogen, but it takes gaseous hydrogen, and we'll have a picture to show how that's implemented at the data center to provide gaseous hydrogen at the right pressure and temperature to the fuel cell.
As we've mentioned, this is a first-of-a-kind demonstration ever at a live data center with a megawatt-scale fuel cell, and we've demonstrated a 48-hour runtime as well as multiple extended run hours exceeding four hours each time. All in all, we generated in excess of 160 megawatt-hours at Microsoft's data center.
And, the green piece here is where Caterpillar comes in with regards to controls, the systems architecture, and being able to flow power from multiple directions, whether it's from a hydrogen source through the fuel cell, or the batteries, or the grid, and to send power in multiple directions, and so you can charge your batteries from the grid or with hydrogen, although that might be an expensive proposition. We want to acknowledge the financial assistance received from the government. And on to the next slide.
In terms of the DOE potential impact and relevance. The picture on the right is one that you might have seen multiple times in many different presentations. I'll just speak to it in a bit here. Today, primarily, if you look at the right side of this picture, hydrogen is mainly used for oil refining and producing fertilizer. But we're looking at hydrogen as an energy source that's produced from excess energy, whether from renewables, nuclear, or fossil with carbon capture, and to redeploy that energy in power generation as needed.
We talked about the 48 hours of fuel cell as well as the fuel cell being capable of reaching a full load rating of 1.5 MW. The relevant DOE goals here are systems development and integration, looking at how commercially ready this solution is, what are the greenhouse gas impacts, and technology validation.
And this is the picture of the setup, the microgrid setup at the data center. We're connected to the— Microsoft's electrical spine here in the background, and if we move to the foreground and go to the bottom right, we'll start at the fuel source. We've got liquid hydrogen that's converted into a vapor through the vaporizers, and it comes along the gas train on the concrete pad to the fuel cell, about 75 feet away.
And around the 1.5-MW fuel cell are the balance of plant, and you see the cooling package in the background there as well as the inverter to convert the DC power to AC power, a transformer, and a switchgear that connects the two battery sources to the fuel cell and connects everything to the electrical spine. And the microgrid operation in an islanded mode, we are sending power to a load bank that you see at the top left of your screen.
Safety is really important to Microsoft, Caterpillar, Ballard, and the other people that are performing in this project. And we had many people that worked on-site from day one when we signed the agreement for this, the hydrogen safety plan has been in our minds, and we've worked it from day one basically. And we had a lot of work and participation from the different team members that I'll speak to in the next slide, just coming together and sharing expertise and best practices.
And this was not done in isolation. We had a lot of feedback as well as live participation from the Hydrogen Safety Panel—when we did a one-week process hazard analysis, we had the Hydrogen Safety Panel, or three members from the Hydrogen Safety Panel that were with the engineering team the entire way just to look at different potential scenarios and ways to mitigate those risks.
We had a pre-startup safety review that we did on-site. And as part of the preparation for being on-site and working on-site, everyone has to perform training either in person or online and certify that they understand the risks that comes with liquid hydrogen and batteries.
And again, want to give special credit to the Hydrogen Safety Panel. We have training provided to the national guard, the first responders in the sheriff's department at Cheyenne. And this site has two unique types of fires, one being hydrogen fire and the other one's battery fires, and so we provided the hydrogen fire training through the Hydrogen Safety Panel with Nick Barilo, and Caterpillar also provided fire—a battery fire training for our batteries.
And we created a lot of documents including safe job procedures and lessons learned to maintain a safe work site, and we're proud to say that during the course of construction and demonstration and demobilization, we've not had any recordable safety incidents and we're really proud of that record.
In terms of the participants, Caterpillar was the lead for this overall project, systems integrator, controls development, and responsible for installation and demonstration. Microsoft was the host of the demonstration site and providing data center requirements. Ballard's the supplier of the 1.5-MW fuel cell and provided support for installation of the fuel cell at site, commissioning, and test support. And Patrick's gonna speak to more of the fuel cell side and the various applications you might have at data centers.
The National Renewable Energy Lab is a big player in the space of hydrogen safety, providing some techno-economic analysis and greenhouse gas impacts. And the next two slides that I will speak to are with regards to the TEA as well as the greenhouse gas impacts. Linde provided the liquid hydrogen equipment as well as the fuel for demonstration. McKinstry was a EPC firm that provided design layout, site prep, install, and were on-site during testing and decommissioning.
This is the first slide from NREL. This is on the TEA. We looked at seven different scenarios here, looking at different combinations of providing backup power at the data center. And the peak load again, or the max rating here that we're looking at is 1.5 MW. We've considered two variants of fuel cells. One's a bespoke tailor-made stationary fuel cell for stationary power and the other one is taking something from a heavy-duty application and trying to [inaudible] in a stationary application with some cost differences. And the other variation is looking at liquid versus gaseous form of storage. Of course, we have to consider the baseline case, which is the diesel gen set. And the other case that you see that sticks up is the primary batteries, and this comes to be the most costly form of energy for providing 48 hours of backup power because if you go beyond a certain amount of energy that's required, then you quickly have to shift to hydrogen, and that that crossover mark could be between say 8 and 12 hours of energy that you need. And so, at that point in time a fuel cell with hydrogen storage is the way to go.
And based off the NREL analysis here, what we're showing directionally is that there's benefits from a cost standpoint by going to liquid because of the energy density that you get from that. And there's potential for heavy-duty fuel cells that are adapted for stationary applications to come to cost parity in the future with the diesel gen set. And a lot of this is based on projections of fuel cell adoption and volume that you might get from manufacturing.
Next slide from NREL, it's titled, here, greenhouse gas emissions analysis, but to be more specific, it really should be life cycle analysis. And, typically, this has been done on a wells-to-wheels basis. In our case, this is [inaudible], and we've got a diesel gen set again as the baseline case, and this is using a fossil fuel. If you use a lower carbon intense fuel such as biodiesel, you can have life cycle greenhouse gas emissions reductions by switching to a different fuel source that's less carbon intensive.
The other pieces that we look at here are using a natural gas gen set. We looked at producing hydrogen from a site that's not close to the point of use and being transported with a truck that's powered by hydrogen, whether that's a gaseous form of hydrogen or liquid hydrogen.
And then we've also looked at off-site SMR and transportation of that, of that fuel to on-site. This is with—on-site wind as the lowest, or clear winner, if you wanna pick a winner in this case.
And we did look at some other applications using the grid. As mentioned earlier, this was done in Wyoming and this might be a snapshot in time, the—just comparing the grid power here, we've got three variants, we've got Wyoming, Washington State, and Virginia where a lot of the data centers are being built and have been there. The three different mixes are—Wyoming in the past, or a snapshot in time, was a lot of coal in the mix, and Virginia has got a lot of natural gas, Washington State's got a lot of hydro. And so those are some of the differences that you see with the carbon intensity. And so, the takeaway here is that potentially with on-site production, using on-site wind, there's a path to lower carbon intensity for a hydrogen fuel cell.
In terms of gaps and barriers to adoption, I think we've mentioned just the hydrogen cost and availability is a challenge, and a lot of the gaps and barriers to adoption here are known things, and the DOE is working on them through different goals and programs. And so today, we don't have the hydrogen where we need it to be. For the project that we've demonstrated here we had the hydrogen—liquid hydrogen transported from Ontario, California. And so, yeah, this is a known challenge today as we implement the hydrogen hubs and get more hydrogen online that's going to help with the cost and availability question.
The power density of fuel cells, that's another thing that—so if you look at across the street at the data center, 3-MW diesel gen sets in a 40-foot container, and it's over about half the power here with this 1.5 MW installation. First cost for fuel cell and systems, again with broader adoption and manufacturing that would come down. The last one is space requirements. The picture that we showed had a lot of space because of some of the setback distances, and, you know, a first-of-a-kind demonstration, there's possibility to have better integration and design for a future data center whether if you choose to have it on-site like we've demonstrated here, or you have that hydrogen facility off the fence of the data center. And so, these are just some gaps and barriers in adoption but nothing that's not already known and not something that's insurmountable or you know, a showstopper, if you will.
In terms of the conclusions, we had a successful demonstration and zero safety incidents. We're really proud of that. The fuel cell plus battery microgrid solution provides applicability beyond standby and the solution provides increased resiliency for the grid, a lower carbon intensity solution, and an integrated system. With that I will end my presentation. Thank you for your attention.
>>Patrick Savoie: Kyle, should I just start?
>>Kyle Hlavacek: Yep, go ahead, share your slides, and it's all yours.
>>Patrick Savoie: OK, alright, so let me know if you can see my screen.
>>Kyle Hlavacek: Yep. We can see it. You're good.
>>Patrick Savoie: OK, alright, thanks, Paul. That was really insightful for any, you know, I would say novice or trying to pencil out what is required in these larger fuel cell projects. It was great to work with you and the whole Caterpillar team.
So, for this part, we simply just wanna give you an overview of, you know, fuel cell technologies and which space it makes sense today to look at large-scale stationary fuel cells. with of course the fundamental reason behind all of this is that they are truly zero emissions. I think Paul pointed it out that the only byproducts that you have with a PEM fuel cell is water, heat, and of course electricity.
So, my name is Patrick Savoie. I'm a key account manager at Ballard Power Systems, also leading the business development side of the business for North America in the stationary space. So pleased to be giving you just a high-level view of what fuel cells can do and hopefully you'll come away with just a little bit more knowledge and perhaps some questions.
So, in a nutshell, just a quick, you know, elevator introduction on Ballard Power. We are now in our 45th year—I should probably change this—of operations with about 1,200 employees globally. A lot of the work, I would say, upstream of where we are today was really on IP and developing the secret recipe for fuel cells. Ballard being, you know, I think one of the first companies to really commercialize fuel cells, the market didn't really have any products to bear whether it was on the MEA plates or the membrane technology or also the catalysts.
So, a lot of the in-house development, well, of course brought us to where we are today with nearly 1,400 patents. We are publicly traded on Nasdaq and also TSX in Toronto, Canada. We strive to carbon neutral—neutrality by 2030. So, in the middle of the screen, basic market segments where Ballard is playing today is in the bus market, which is the most mature today, with over 1,400 buses running with a Ballard engine. The truck business is, I would say, next to the bus with 200—sorry, 2,300 trucks running. The majority of those are through our Weichai joint venture partner in the Chinese quadrant of the world.
We have been working with several train companies, Stadler and Siemens to name a few, to develop, I would say, regional commuter hydrogen trains. We've developed specific technology and fuel cells for the marine segment. And last but not least, the stationary application, which has been sort of the slowest market up to date simply because the size and the sheer magnitude of these fuel cells that are required, as Paul pointed out, this 1.5 MW fuel cell is the largest to our knowledge in the world for PEM-based fuel cells.
So, all in all, I think maybe the last takeaway is the middle bottom line where we've got over 175 million kilometers, about 150 million miles, that have been accomplished by vehicles running with our fuel cell engines. So, what is the Ballard promise? Of course, we've proven the technology. I think we're, this is one key takeaway, is that we're no longer at a stage where we need to prove that a PEM fuel cell works.
PEM is the acronym for proton exchange membrane, and I'll explain that a little bit in the next slide. So, we're past the—you know, the technology works, the technology is mature. Really now it's all about trying to figure out where does it make sense? What are the applications where it is a cost-effective solution? Of course, with the underlying benefit that it is a zero emissions solution.
So, the reliability, efficiency, and the promise when it comes to Ballard, we support our customers right to the end, knowing that fuel cell is a bit of a novel technology for a lot of people. And I think Paul and his team can be a testimony to that, that we really supported Cat right up until the end of this project.
So, this is just to give you a snapshot of when we say fuel cells, I mean there are multiple different types of fuel cells, but the main two families of fuel cells that we hear about are the PEM, as I indicated, proton exchange membrane. That's what we make. Solid oxide fuel cells are a different technology. They operate at a much higher temperature and typically if they are using alternative fuels such as methanol or ammonia, we'll still have some residual SOx and NOx. So, it's not a truly 100% zero emission. Much better than diesel, no doubt, but not necessarily 100% zero emissions.
The other main differentiator between a PEM and a solid oxide is the fact that PEM has the ability to load follow, so just like a diesel gen set where it can ramp up and down. Contrary to a solid oxide, which should be used more as a base-load-type fuel cell, which operates at a steady state power output.
And finally, today the PEM I would say dollar per kilowatt is lower than a solid oxide fuel cell, and we are striving towards being at parity with a diesel engine by 2030. So that is the goal that Ballard has set. So, we continue to improve product durability, power density, and also cost per kilowatt.
This is a funny peanut butter sandwich, but it's just to give you a sense of why do we talk about batteries when we're also talking about fuel cells.
So, in every application, whether it's onboard vehicles or on stationary, there will be some sort of battery pack to provide various different, I would say, system integration needs. So, on the one hand, the batteries are kind of like the jam in the sandwich. It's short sprints of power. Typically, we talk about C ratings if you're familiar with battery C ratings, so that's the ratio between power output versus the energy content, and also has a real knack for these instantaneous transient response. So, think of a UPS system for a data center, of course, just like you at home, you don't want your data to be lost or the screen to turn off, so that instantaneous backup power is possible with batteries. Whereas with fuel cells, it takes a little bit longer to ramp up the fuel cells.
So, the hydrogen fuel cell is more like the energy content of your sandwich, which is for a much longer steady stream of energy to this overall solution. So, batteries are there really to do a couple of things. Obviously for the transients. Number two, to start the fuel cell because, just like a car engine, you need to sort of get the pumps and the radiators and the compressors going. So, the battery is also part of the solution for that.
And finally, you also can look at a larger battery to really increase the total output capabilities if you've got a lot of inductive loads, essentially loads where there's a surge. And that basically is what we're referring to as these transient response.
So, in a nutshell, what we currently offer is a 200-kW building block for large integrated fuel cells. So, you can see the one unit to the left-hand side of your screen. Then you've got a sort of semi-integrated system with fuel cells and DC/DC converters. And then what it could look like, or essentially what we did with the Caterpillar project, is a containerized solution. In the case of what we're showing here is the cooling is on top of the skid, on top of the container, whereas with the Microsoft DOE project, the cooling was on the ground. It's just to give you a sort of footprint because Paul talked about that.
And so where do we see fuel cells making sense, for particular markets? So, on the EV chargers, we see a lot of demand coming because the grid is constrained and there's a lot of necessity for more current or more grid connections for large vehicle fleets. So, either to compensate the wait time, because you can wait between four or five years sometimes to get grid connection in certain parts of the country, or there is simply just not enough grid power at that location or at that transformer station.
Number two, shore power, so a lot of the ports when the large ships come in, they still need to be connected or have power to hotel loads like lighting and heating or cooling. And so, in certain parts of the world, yes, you've got an umbilical cord to connect to the grid, but in other parts they actually run tugboat generators and connect these ships to that. So, to remove obviously the diesel emissions, you could think of a standby or stationary power system at ports.
Data center, well we talked about it. And microgrids, well again, sort of remote island areas where diesel gen sets are still, you know, the main power output, and you can think of combining that with wind and solar. And so really where these diesel gen sets are running, you could replace those with a fuel cell.
So, just to give you a couple of extra bullet points on, you know, these individual systems or market segments. So, I guess that's basically what I just explained, avoid costly and timely grid delays using—the grid is not ready to support the growth of the EV vehicles, and so on. So, again, just on the shore power, reducing emissions, eliminating diesel, bringing power to ships when they are at berth.
And on the, sorry, backup power centers for data centers. Well, of course, in other areas perhaps of the world where you would have unstable grid reliability, and you could also think of, of course, much longer duration.
So, this is one point that I'd like to bring to Paul's comment while—when you're comparing a fuel cell to a diesel gen set, the power density is definitely much bigger with the fuel cell, point taken. However, I think that the comparison should also be with batteries. If we had been running this 48-hour system with batteries, it would have required at least 10 times more footprint or more space to have that amount of energy in lithium-ion batteries.
So, the comment is not that I'm against batteries. Of course, I think batteries completely make sense when you're looking at a 2- to maybe 6-hour storage and backup time. But anything beyond that is just too large of a footprint for batteries and really, I don't think that that would be the right approach, and neither from a capex point of view as well. And so, here's just a, again, a synopsis of microgrids.
So, I think that basically brings us to the project. I think you've got a lot more details in Paul's presentation, but just to give you a quick snapshot of what was the scope for Ballard, because Caterpillar of course talked about McKinstry, they talked about Linde, so all these other technologies bring to make this project successful.
From our point of view, what was Ballard's scope? So essentially the fuel cell, and the cooling part, so what we're seeing here are the cooling just behind the container, and we also had some heat management there as well.
The bottom part of the picture, you can see this is just another version, or before we had painted it with the Ballard name, and you can see the stacks there with the red, the red piping. So, these are all the stack assemblies that you can also service from an exterior rollout tray. So, pretty cool design there. So overall, I hope that gives at least a little bit of an overview of what fuel cells are all about.
I'd be more than happy to, you know, answer questions. And thank you for this opportunity. And again, you know, two thumbs up for working this DOE project with Caterpillar. It was just an amazing, an amazing project to work on. So, thank you.
>>Kyle Hlavacek: Thanks, Patrick. Appreciate that presentation. And that concludes our H2IQ Hour for today. Once again, I'd like to thank Paul and Patrick for today's presentation. The slides and a link to the recording of this webinar will be available within the coming weeks in the H2IQ Hour archives. Be sure to subscribe to HFTO news to stay up to date. Thank you for attending, and we look forward to seeing you at our next H2IQ Hour webinar.
[End of audio]