Change the World with Autodesk Fusion 360: Become a CFD Engineer with Generative Fluids

MACIEJ JASKIERNIK-DETKA: Hello, everyone. My name is Maciej Jaskiernik-Detka. I am a senior software quality assurance engineer at Autodesk. And today, I'll show you my presentation, which I named "Change the World with Autodesk Fusion 360-- Become a CFD Engineer with Generative Fluids."

Today, I will shortly present you some fluid path features, such as pressure drop reduction, flow balance, obstacle omission. Later on, I will show you applications in manufacturing. Especially, I will focus on how to create manufacturable parts based on your design. And in the end, I will show you how to find fluid path in Fusion 360.

I need to make one statement. Fluid path and generative fluid is the name of same sub-model in Fusion 360. So I will use both names during the presentation.

Let's start with pressure drop reduction. The slides I'm going to show you were prepared for the students, which were visiting our office in Krakow about one year ago. And I was asked to show them what is [INAUDIBLE] the fluid in general and what does a CFD engineer do for a living.

So the easiest problem for CFD engineer to solve is connecting two pipes into one. The goal is to achieve the lowest possible pressure drop in the connector. And in a minute, I'm going to tell you why.

In big companies where I used to work, first, this problem is handled by design engineer. So design engineer is always focused on producing cheap and easy-to-manufacture solutions, like the one in the left picture. So the idea is that somebody is going to cut the pipes, weld them again, and create the connector, as in the picture.

Later on, design engineer exports this idea as a CAD data and provide it to CFD engineer, like I used to be. Now it's time for CFD simulation, based on which I will calculate the pressure drop.

And now imagine that this connector is a part of the bigger system-- for example, internal combustion engine cooling system. So if the pressure drop in such system is high, it means we need lots of power to get it running. Therefore, we increase the fuel consumption in the internal combustion engine. And this is definitely not good either for the user nor for the environment. And usually, it comes out from the simulation that the pressure drop is too high.

So we have two possibilities-- either to increase the water pump, and also the fuel consumption, or to optimize this pump, maybe change the manufacturing solution. But typically, option number two is chosen.

So now I need to prepare some tips because I'm not able to prepare a new design on my own because I do not have almost no design skills. Therefore, I'm only preparing some tips, as you can see in the picture on the right side. I'm drawing some lines in PowerPoint. And I'm providing it to the design engineer.

But later on, he or she comes to my desk and asks, but what's the radius of this line? And I'm saying, I have no idea. I just draw it in PowerPoint. You need to figure it out. And that's problematic because this process that takes lots of time. And therefore, it's quite expensive. But fortunately, we have generative fluids.

In generative fluids, only need-- only thing that you need to do is prepare a simple shape as a starting one for the optimization. So you need a separate body for inlet or outlet and, as I said, a starting shape.

The aim of this optimization is to minimize the pressure drop. So first, a CFD simulation is run, based on which a sensitivity field is calculated. Sensitivity field indicates which part of this model should be enlarged or where it should be cut.

So as you can see in this slide, iteration by iteration, you get smoother and smoother shape. And therefore, you decrease the pressure drop. The result is shape as presented on the screen right now.

And during the presentation, which I showed to the students, I told them that for sure, this one can be manufactured, either used as a core in the casting or printed using 3D printers. The truth is I haven't-- I wasn't sure about that. Luckily, a couple of weeks later, I took part in a 3D printing training. And as we have a 3D printer in our office, I decided that I need to manufacture it to prove that I was right.

So I started with preparing the outer shape because as you imagine, generative fluid is optimizing the flow path-- so the inner shape. Therefore, you need to create the outer shell that you can manufacture later on. So I started doing this. But later I thought, if this is a connector, you need an interface, for example, to clamp the hose on it. Therefore, I extended inlet and outlet a little bit.

Next, I wanted to get rid of the support in the printing process. So support is the scaffolding that needs to be introduced during the printing process to support the structure. And I introduced a leather shape in the middle to get rid of this support, and also to stiffen it a little bit because I had no idea if 2 millimeters of the wall thickness, which I assumed, is enough.

In the end, I thought, but what about sealing? When you clamp a hose on the connector, you want the interface to be perfectly sealed. Therefore, I introduced the ribs on the inlet and outlet interface, as presented in right picture. And I was ready to print it. And I was really proud of this process because it was the first thing that was manufactured by me and designed by generative fluids.

And at this point I thought, I need to go to the AU to show you this. I thought that if manufacturing is so easy, I will have lots of ideas. But I need to think of something more complex. But I was wrong at that point.

Nevertheless, I started thinking about the flow balance. So flow balance is a generative fluid feature which adapts the geometry to distribute the flow evenly between all outlets. So if we have a structure with one inlet and two outlets, the aim of optimization will be not only to minimize the pressure drop, but also to provide the same amount of the fluid to each outlet.

For demonstration purpose, I prepared a model with one inlet and six outlet. And honestly speaking, it was spring. And I was cleaning my garden. And I thought, a perfect design would be a garden sprayer which distributes the flow, water flow, from one garden hose to six outlets or evenly around the sprayer to my plants. And I started preparing it.

But as I told you, my manufacturing experience was almost zero. I cannot say it was zero because I created connector previously. But during the first run, I haven't thought about manufacturing restrictions at all. And it was a big mistake.

So this process of preparing the final design took me two and a half months. And I will share with you with my thoughts that I had during this process in the second part of the presentation.

Now, the important thing is that I was able to print my final design in one piece. And what's more, I run a detailed CFD simulation using Autodesk CFD. And that simulation showed that the flow imbalance is less than 5%. So for me, it was perfect.

Final design was a mix of manufacturing restrictions, flow rate uniformity, and pressure drop reduction. As you can see in the picture, I was able to print it and test it in my garden. And I was really happy about the fact that I made it. And I think it is really useful design, at least for me. But the question is, can you think of engineering application of this attribute?

So as my background is automotive, I can-- I prepared some examples for you. Let's imagine that we have a battery pack for the electric vehicle. So in this battery pack, you have cells, which are represented in the picture with red cylinders, and cooling channel, which is represented by yellow body.

So every time you charge the battery or discharge it, the cells warm up. So you need to cool them down. And typically, in such a system, you will have one coolant pump. Therefore, you need to distribute the flow from one inlet to all the cells. So it means that you need a distribution manifold.

I prepared such a manifold. We have one inlet and six outlet. So green bodies are keeping the boundary conditions. And they will be not modified during the simulation. Yellow body is a starting shape, which I prepared using automated modeling. And if you don't know this feature, I really recommend you to get familiar with it because it speeds up the design process a lot.

The outcome of this optimization was the geometry that you can now see in the screen. So please take a look at the outlets. The flow velocity close to the outlet is almost 2 meters per second in all areas. Therefore, we might think or assume that the flow balance is properly distributed or the flow is properly distributed.

Nevertheless, to prove this information, I, again, run a detailed CFD simulation using the distribution manifold and cooling channel. And again, this simulation showed that on the outlets which are now on the left side of the geometry, the flow imbalance is around 5%. Of course, the flow in the cooling channel can be still improved. But I think it's really good for the first shot.

Another attribute of generative fluid is obstacle for omission. To visualize this possibility, I prepared the sprayer with one inlet and one outlet. Honestly speaking, I was not sure if the structure can handle the water pressure inside. So I prepared a stiffening structure between-- which will be later used to connect the outer shell of the top-- the bottom part. It is represented at the-- with the right shape. This body is defined as the obstacle. So optimization algorithm will omit it during the optimization process.

And as you can see, this obstacle is introduced in a way that takes some of my flow path. Therefore, it's decreasing the cross-section area and, for sure, increasing the pressure drop. Fortunately, generative fluid can provide a perfect shape which perfectly omits the obstacle. It's very smooth. So we can expect that pressure drop was reduced. But we need to prove it. Therefore, I, again, run a CFD simulation and printed this structure.

So simulation showed that pressure drop assuming 1 meter per second before the optimization was 14.5 kilopascal. But after optimization, it was reduced to 6.35 kilopascal, whereas a test in the field showed that spray distance has increased from 140 to 210 centimeters.

The reason why the pressure drop was reduced over 2 times and water spray distance has not increased by 2 times is the fact that the nozzle inclination is 30 degrees. Therefore, it means that the water goes higher than further. And I focus only on the distance from the sprayer.

Again, we need to think, is it really useful? So back in the days when I was responsible of preparing cooling systems in the internal combustion engines, I always had a problem with something called thermal management component.

So this is the model that distributes the flow, for example, from engine water jacket, charger air cooler, or engine air cooler to the radiator, to the bypass of the radiator, because you know that, for example, during winter, when you start up your engine, you don't-- and the coolant is almost freezing, you don't want to cool it down. You want to warm it up as soon as possible. That's why you need this bypass.

And of course, you also want to warm up air in the cabin. Therefore, you need another airflow to the passenger compartment heater. But if you ever try to change a bulb of-- in your car, you know that the packaging under the hood is really tight. So most probably, we'll need to protect some space for other engine systems. And this protected space is represented by red cylinder in the picture, whereas the shaded area is our initial or starting shape, again, prepared by the automated modeling.

Later on, we need to think of how to connect this model to the engine. Most probably, it will be bolted. So we need to protect some space for the bolts, but also for the tools in order to later on access the bolts because you need to screw and unscrew these bolts. And this space is represented by blue cylinders in the picture.

Based on these assumptions, I prepared a generative fluid model. So we have green bodies, preserves, which are, again, keeping monitoring conditions and red bodies, obstacles. Yellow body is a starting shape. And of course, I could redesign it right now to omit the bolts. But I wanted to show you that generative fluid can handle a complex problem.

So flow velocity on the inlets is assumed or-- to be 1 meter per second, whereas I needed to play a little bit with the pressures on the outlet to avoid throttling. The result of the optimization process is geometry that avoids all obstacles. And what's more, if you take a look, we cannot see any rough edges. That means that pressure drop is quite low.

Of course, to create this component, lots of work needs to be done. We need to prepare interfaces for the bolts, design the outer shell, lots of things. But designer who is not familiar with the CFD can use this shape as the internal flow structure. And if further optimization is needed, he can-- he or she can always use preserves and obstacles for detailing and target volume to change the general dimension. I will elaborate about the target volume a little bit more in a couple of minutes.

Let's sum up. Fluid path can smooth your design and reduce the pressure drop. It can provide you even flow distribution and omit all obstacles that-- occurring on your system. So I think it is really useful thing.

Now let's focus on applications in manufacturing. As I told you, my story or my journey with a content sprayer was-- took about two and a half months. And one of the most important things I learned that-- is that starting shape is really important because [INAUDIBLE] the fluid can only optimize your idea. It cannot provide a new design.

In the bottom on the right, you can see geometries that were-- the alternative geometries that were prepared by automated modeling. But I decided to use the one on the left side. And if that was a good decision we'll see in the next slides.

Another important part is defining a target volume. You need to provide it as a percentage of the starting shape. So optimization algorithm needs to know whether to increase your starting shape or decrease it, and how much.

It's quite obvious that in the beginning of the process, you have no idea what should be the final volume. Therefore, I recommend to run multiple simulations and choose the one that fits best to your requirements. In my case, I chose the one with 30% of the target volume.

When you choose your final design and you want to generate a new model, click on the "Design from Outcome," wait a while, and when it's ready, click Open Design. This operation will provide you a new model you can start working on. And as my sprayer was too complex to print it in one part, I decided to split it to the base and the top part. And I prepared a-- slightly different workflows because a base is very simple shape. Therefore, all you need to do is start from creating the surface with the offset.

Think of the offset as the final part wall thickness you want to achieve. So in my case, I wanted to have a 3-millimeter wall thickness. Therefore, I defined the distance offset as 3 millimeters.

Next, you need to thicken your surface to the inside to perfectly represent the shape that you use during the optimization process. Therefore, I used thickness value of minus 3 millimeters.

Now it's time for detailing. As I wanted to use two parts and connect them later on, I needed the interface for sealing. I also introduced some positioning pins. And, of course, I needed a connector between my garden hose and the sprayer.

Now I can start working on the top part, which is a bit more complex. So I was not able to create the surface offset with distance of 3 millimeters because such surface would have self-intersections. Therefore, you need to start with creating the surface with offset of 0 millimeters, just representing exactly the shape you obtained.

Next, you need to mesh it. Use Tessellate tool. And at this point, you can use default meshing options because it's not really important. But as you can see in the picture, there were several surface meshes created for each surface. So to simplify the workflow, you need to combine them into one using Combine Face Groups tool.

When it's done, you can start repairing the surface. And if you use Repair tool and choose the repair type to Rebuild and rebuild type to Accurate, you will be able to provide the offset and create a volume mesh. Again, think of this offset as the final part's wall thickness. I've used 3-millimeter distance.

Right now, you need to convert mesh into a solid body. This operation is really time-consuming. One thing you need to consider is reducing number of elements to speed up. Of course, if you reduce the number of elements, you will lose some quality of the surface representation. Therefore, you need to think on your own about doing this step. Nevertheless, if your-- when your solid body is ready, the last thing you need to do is cut the inner flow path inside it. So using the optimized geometry, we can prepare shape for the fluid flow.

Remember that offset is providing some volume not only to the side, but also to the top and bottom. And in my case, I needed to cut the bottom 3 millimeters to, again, open the connection between top and bottom. And nozzles were prepared in the separate steps using the simplified workflow because these are really simple shapes.

Finally, I obtained two geometries which were ready to print. Base was not a problem. But top structure contains quite a lot of overhangs. Therefore, lots of scaffolding was needed during the printing process. And as part of this, supports were not able or impossible to be removed.

Later on, mechanically, I needed to use a PVA, which is water-dissolvable material. So you can print it, put it into water, and get rid of the support. Perfect. But unfortunately, this material is quite unstable. And as you can see in the picture, the printout quality was quite poor. Therefore, I needed to simplify my idea and start printing process again.

If you consider 3D printing, you need to imagine that one layer is printed over the previous one. So it's-- the easiest shapes to create are pyramids or towers because it's always easier to put a smaller object on the bigger one than the opposite. And that was also my idea.

So my initial shape during the second run were six towers connected with a circular connector in the bottom. And the outcome of this optimization process is visible in the middle. Using the workflow I showed, I prepared the outer shape. And I was ready to print.

As you can see in the left top pictures, the printout quality is good. But I was really unsecure about how to connect these two parts. And my colleague advised me to use zip laces because they are indestructible. But unfortunately, they just don't look good. Nevertheless, I used them. And as you can see in the movie, I was able to test my garden sprayer for the first time.

But I had a couple of problems. First, I didn't like the way it looked. The zip laces are indestructible. But I didn't like this idea.

Second, there was a sealing issue between these two parts. And the last problem is spraying distance. As you might see in the movie, the distance, which was provided with water, is-- the circle is about 30, maybe 40 centimeters of the radius. So it's quite small. Therefore, I needed to prepare another design.

I was thinking about pyramids and towers. And I came up with an idea to change the circular connector to be a bit more similar to the pyramid. So it has a triangular cross-section. And it's easy to print. But we need to optimize it. And also, I used a rectangular inlet. And this was very strange idea for me because I had never seen rectangular pipes. So I thought, maybe I'll be first in the world to use such pipes. We'll see.

But as you may expect, there is another issue that the connector, the silver one with the cross-section on-- of the triangle, will be not utilized by the flow. Therefore, generative fluids during the optimization process will try to remove it.

What's the solution? During the optimization process, you are provided with several designs, starting from the initial one. And in the end, you get the final, most-- or the fully optimized one. But there are several in the middle.

So the solution is to find one which still contains manufacturing features and is properly balanced regarding the fluid flow and is smoothened to reduce the pressure drop. And in my example. I choose the outcome number 10, which is now visible in the middle of the screen.

So I'm ready to prepare the outer shape. Of course, I need a custom connector between the hose and the garden sprayer. But it was not a big problem, as I was really happy because I fulfilled all my requirements. And I was ready to print.

I was very nervous during this process because it was the most difficult or most complex part that I ever manufactured. And I couldn't look to the inside to check the print quality. Nevertheless, the printing process was done with no issues. I have no sealing problem because it's one part. And I obtained even flow distribution, which I knew from previously run CFD simulation.

As you can see in the pictures, I really love this idea. And I use it in my garden. So it's quite useful for me. But let's focus on the lessons I learned.

So at first, think of the manufacturing restrictions while you are preparing your initial shape. This is very important. As you might see in my process, it defines either you are able to manufacture your part in the end or not. And that's why we need to focus on the second thing, which is that-- the fact that it's better to choose a manufacturable design than fully optimized one.

So it is true that done is better than perfect. What's more? Try getting inspiration from the forms around you. And usually-- not always, but usually-- the simplest ones are the best. And the most important thing-- if you fail, don't give up. But improve your design and try again.

So where is fluid path in Fusion 360? Whenever you open up Fusion 360, you start in a design workspace. So as I told you in the very beginning, you need a separate body for inlet or the outlet and a single shape for the starting shape. When it's ready, click on the Design button to change the workspace, and choose Generative Design Workspace. When the pop-up appears, check the Fluid Path and click on Create Study. You are now ready to set up your model.

At least one inlet and one outlet is needed to perform a simulation. Starting shape is not a must. But I really recommend to use it. Keep in mind that by default, water is your fluid medium. But you can change it to the [INAUDIBLE] or select Custom Fluid if you want.

So now you know where to find fluid path. You know that it can provide you a smooth design, reduce the pressure drop, and the amount of the energy you need to use to put the flow through your parts. It can provide you a proper flow distribution if multiple outflows are in your system. And it can avoid all the obstacles that you define.

I think that you are also ready to prepare manufacturable parts based on your ideas. Therefore, I think you are ready to change the world. But if you consider that my examples are too easy or not realistic, the best solution is try it on your own now. And please share your feedback using Fusion 360 Feedback Hub.

Thank you very much. I hope you enjoyed, and have a good day.