Toward Zero-Emission Shipping with Fuel Cells and Model-Based Design

[AUDIO LOGO]

KJERSTI WERGELAND KRAKHELLA: Hi, everyone and welcome to this section. My name is Kjersti, and I'm a researcher in the two companies Clara and Alma. Today, I will talk about how we use models for fuel cells and how we use this towards zero-emission shipping. First in the presentation, I will go through a bit of our projects, because I think we have rather cool projects before I dive into some slides about fuel cells and how we model them.

So first, our projects. We have a lot about fuel cells in our companies. But we also have rather cool projects like the torch to light the Olympic fire. We had it withstand a lot of rain and wind, and need to withstand a lot of time. So we had these guys running around in rather a small town in Norway, making sure it could last different weathers.

You also had this one here, where we grew tomatoes in space because astronauts need fresh food. And I think we got the tag the world's most advanced flowerpot. But later, on it was more fuel cells-- for example, this one, where we made fuel cells for the moon, because on the moon, you have really cold nights, and the nights last for 14 days, I think. And fuel cells produce a lot of heat.

Later on, we had more on the maritime sector for shipping that we made these container-based fuel cells, and the company Alma was started. And fuel cells for shipping are rather important because the shipping industry stands for quite a lot of the global emissions of CO2. 30% of the emissions in the EU in the transport sector is from the shipping industry. This is why it's really important that we actually find a way to get zero emission.

The next part is about fuel cells. So fuel cells, you have hydrogen and oxygen. And these two, when they react, they produce only water. If the reaction's happening, it's a spontaneous reaction. But if you find a way of separating it, like for electrolyte or a membrane here, and you get the electrons to pass in an outer circuit, you have electric current, which is why this is important.

The special case about solid oxide fuel cells is that they have a material that could withstand much higher temperatures-- up to 800, and some up to 1,000 degrees. At this temperature, you can have cracking and reforming, meaning that you can use fuels like ammonia, methane, and methanol, for example. Operating on higher temperatures also means that you have available heat at higher temperature, which is way more valuable. In addition, the efficiency increases.

Now, for the observant attendant here, you can see there's a C in this one, and in this one meaning that you reduce CO2. For the even more observant attendants, you know that ammonia is produced in steam methane reforming mostly producing CO2. The reason why this is zero emission is that for these two, the methane and the methanol passing through here, the CO2 coming out here is not mixed with air.

So it's rather concentrated. Ironically, this is positive because a higher concentration of CO2 means that it's simpler and cheaper to remove it. For the second part, the ammonia, when you produce ammonia at a certain site, it's way easier to do a carbon capture than having it on the boat, where you have to have the capturing of CO2 following the boat. This is why this is a simple way of getting towards zero emission shipping compared to motors.

Summarizing this, the reason why fuel cells are easier to use than motors is that the efficiency is increased. In our lab, we measure, like, 62% to 63% efficiency. And that's why I don't think there's an issue reaching 70% or higher of the efficiency here. Internal combustion engines have 40% to 50%.

The second part is the fuel. In the solid oxide fuel cells, the one we are using, you can use almost the same fuels as you use in an internal combustion engine, while the PEM fuel cell, which is one you normally have in cars, you could only use hydrogen. In addition to this, you have the benefits that it's low maintenance. The reason for this is that you have no rotating parts, no vibration, which also make [INAUDIBLE] sounds.

If you remember from the first slide this container-based fuel cell, it's easy to maintain modularity, meaning that if you need 1 megawatt, you can build a container for 1 megawatt. But if another customer asks for 10, you just use 10 of the same container. The last one I talked about was carbon capture, which is simpler for fuel cells.

Now, the really cool part about our company, and which makes it easier to model, which is the next part of my presentation, is that in our labs, we actually can do material characterization, meaning that into my model, I get electric conductancy of different materials. I also go down in the factory where I can have my fantastic design of this place here. It's a heat exchanger.

This guy told me this was a really bad idea. So I went up and decided again. But the thing is, I get all this data from the heat exchanger, which I can put into my model, which makes it way easier to make models. And last, we have a lab where we actually can test the fuel cells, meaning that I get a lot of data for my model.

So when my boss came to me and asked me to make a model of this, the first thing we had to consider was the thermodynamics. Our system has different components like condenser, heat exchanger-- of course a fuel cell, and an afterburner. And I found this wonderful webpage where you actually can find all you need there-- entropy where you actually can find the energy that is transferred-- for example, the entropy which is used for the electric power and how the energy is related to the increase in temperature.

Another guy was employed and told me, this is simply the way I'm doing this. We actually implement Python into MATLAB. And in this formula, you add the temperature, the pressure.

And you don't even need to know the formula for water. You can just write water, which was really annoying, because this was really simple. We later discovered that using it with Simulink, the runtime went a bit up. So we ended up using a combination of these two. But the fact that we can do that is pretty awesome.

So the next part is the fuel cell. Inside a fuel cell, you have different reactions. For the ammonia, it reacts to hydrogen and nitrogen.

For the methane, you have other reactions. And the last one here, you might have recognized from the last one, where you have hydrogen and oxygen reacting to water. Each of these either decreases the temperature or increases the temperature, meaning that whenever you have these reactions, I need to make sure the fuel goes up in temperature or down in temperature regarding what reactions are happening.

In addition to this, when you have reactions happening, you need to make sure that you always know what components are there. This is a graph made for us by MathWorks, where you can follow different components. For example here, you can see the CO going down and the methane going down. And you can see you have an increase in hydrogen and CO2.

The next part is the electrochemistry. In a fuel cell, you have different reactions happening. So depending on your electric power or your current, there is a corresponding voltage. You have different equations for this.

You don't need to know all these equations. The voltage is dependent on things you can measure, like the temperature and the pressure, or constants like the ideal gas constant or Faraday's constant. You also have things that are more difficult to measure, like the limiting current or this [INAUDIBLE] slope or the area of specific resistance.

And here comes the advanced part. Being in a company where you actually can measure these things makes your model way easier. I actually can get all these graphs from the lab, where you have the relationship between the stack potential and the current density, meaning that if I request the current density or a power, I can read off the potential. And this is important because the power, which everyone seems to be interested in, is a current times the potential. Everything else from your actions that doesn't go into your electric power go into heat, meaning that I can know how much heat is either produced, heating up your fuel, or is reduced, cooling down your fuel.

The way we've done this is that we use Simulink. In Simulink, we use something called buses. And they are handled the same way as you handle structs, meaning that if you go into a bus selector, you can get out whatever I request-- temperature, the pressure components. These components are not logical in the order they are, but some of you might have noticed that.

But the thing is that you can request one of them throughout your system, regardless of where you are, like you put in a temperature sensor or a pressure sensor or gas tomography to find your components. When they pass into a MATLAB function, like these ones, they can be handled almost the same as you handle a struct, which makes it very easy to know how to handle temperature and pressure. You just read them out instead of using vector, which also works. It's just difficult to remember the order of what you inserted into your vector.

The last part of the presentation, before I sum it up, is that you use MATLAB. You use the model in different ways, depending on where you are in projects. Don't know if your bosses do this, but they do this to me all the time where they ask, if you look into the glass ball, what will this cost?

And the thing is, the model can be used for this, because I know approximately how much high heat I need to have on the fuel cell to make it work. And I know what power they are requesting because that's usually the only thing that they know at the beginning of a project. And I can tell them how much of the fuel they need per hour to make this work.

Usually lately, we've gotten a lot of projects. Seems like people want to go into a greener industry. And we go to the next stage, which is the shipping industry. The shipping industry is interested in the weight and the volume because they're going to place it on the ship.

And often, they have a cooling loop or a heating loop on the ship, and they need to know how much heat is needed or could they get back from our system. And because I know the power that needs to be delivered, I can go into my model and figure out we need this amount of fuel cells for this to work. Because we have experience from the lab, we also know approximately the size and volume of this. In addition to this, because of the model, we have the temperature and how much heat is available for the shipping industry.

For the next stage, it's also an important part because when we are buying different components, like a heat exchanger, they always ask what is the highest flow you will have. And I can go into my model and see, actually, we can't go any higher on this system. So I know the flow. I know the temperature at different places. And actually, more importantly, I know the composition, meaning that if there is hydrogen, we need to make sure there is no material that can have hydrogen embrittlement. If there's oxygen, we need to make sure there is no corrosion, or ammonia, which is dangerous in other ways meaning that it's way easier to both buy components, and it's also easier for making components, designing them for worst case or design case of our systems.

The last stage is where we go through, for example, the safety, where I need to figure out do we need an extra valve? How is the pressure around the system? How can we regulate this? Have we thought about everything going up and down in load and power to see how the system changes when we are operating on realistic systems?

I will give one example of how I have modeled one component in the beginning and at the end of the system. A heat exchanger, you have one fluid going through it one way, going up in temperature, and the other one passing, going down in temperature. In the beginning, I know that a fuel cell needs to have 700 degrees to be able to start cracking or reforming.

So I set this one because we don't know anything about the heat exchanger. We just know what it needs to do, meaning that I can figure out how much energy needs to be transferred to this fluid or gas. For the shipping industry, they usually want to know how much heat is then transferred to the cold fluid or the heating fluid, meaning that I can also figure out the output temperature of the other fluid. Using this, I don't need to know anything about the component. I just need to say something about how much energy is transferred.

Later on, when we figure out heat exchange here, that hopefully will do the job. We use different sets of equations. We don't need to know these equations either. But this little guy is a heat transfer, which is dependent on area and material of the heat exchanger, meaning that I can test how good is the heat exchanger with different flows. And if we have cases where we don't have the temperatures we wanted, what will happen then? In this way, I can actually test the entire system before we build it, but still with the one component we bought, which is rather neat.

In addition to this-- because later on, it wasn't only me modeling, we had to make some collaborations between us modeling in our company. We ended up using Simscape because it's easier. We can drag and drop components. But we had a rather advanced system where we had at least eight species, I think. And if I'm correct, I think Simscape had, like, four.

So we needed some help, but it seemed like this wasn't too difficult to fix. You can make a system adapted to your actual physical system. In addition to this, you need a thermodynamics input and some for the reaction rates. But then, same Simscape is really easy to use.

The second part was we needed a way to collaborate, because when you're modeling-- particularly our system, where you have probably not a person being expert on modeling, and an expert also being expert in fuel cells and an expert in heat exchangers and piping and everything, you need collaboration. And we found the best way to do this was using Azure DevOps, where we actually could collaborate. It was awesome using it together with MATLAB. Probably a lot of other ways, but this worked for us.

The last one was that we got a bit worried when new programmers came in, and we had made code that was may be difficult to jump into because it's a large system. We found a way of using a custom library. You actually can make your own components, add into a library, and can drag and drop. You still need to read the code how to use it, but it's way simpler getting into the code.

And that was the presentation. I will try to sum it up shortly in one slide. We found benefits with using MATLAB and Simulink in the prediction, so that investors and new customers can get good estimates of future systems, like efficiency, fuel utilization, and available heat.

The techno-economic insights-- because when you're buying new components, it's easier when you have a model telling them temperature, pressure, composition, or if you're making your own components. Simplified component manufacturing by setting mass flow, pressure, and temperature early on in the project. And the last one-- when you can implement what you find in the lab into your model. Thank you so much for your attention.

[AUDIO LOGO]