ToffeeX
Ten years ago, a question was asked.
It was asked by Marco Pietropaoli [MP] – then a PhD candidate at Imperial College London, and today CEO of ToffeeX – who had identified room for improvement when it came to the design of, in particular, industrial cooling systems. He wanted to find the most efficient way of controlling airflow, he wanted to land on the best shape for a coolant system, and he wanted to be sure of both of those aspects before manufacturing commenced.
ToffeeX is the answer to that question. Since spinning out of Imperial College London, ToffeeX has commercialised a physics-driven generative design software platform, further developed its capabilities, and secured customers in a host of industries.
At TCT 3Sixty earlier this month, Pietropaoli sat down with TCT Magazine to discuss the company’s beginnings, the key capabilities of its flagship product, and the maturity of generative design software.
Marco, can you tell us about ToffeeX and the journey the company has been on to where you are today?
MP: It was about 2014, 2015, when we really realised that additive manufacturing, especially for industrial application, was really increasing the value of many applications. You had 3D printing increasing the reliability on possibilities, and unlocking the potential for very efficient designs, more efficient cooling systems for aircraft, more efficient cooling systems for batteries, for example. So, really giving clarity of the possibility of making the step forward towards net zero. But the question was, the gap in the market was, what is the best design to control the flow? What is the best shape for a coolant system? The human mind cannot really think clearly because it's biassed by experiences and limited in the 3D abstraction, so we say, ‘okay, let the physics design it.’ So, let's create a software that is using the information coming from the physics simulation to build the optimal design. And this is really the question that was answered during my PhD, developing something called topology optimization for fluid dynamics, which is still today the core technology software.
Very soon, we gained some traction and then we spun out. And after further developing the technology for a couple of years, we build our technology, our software, in an online platform. And now we have customers that use this powerful tool for these smart designs on their own. The technology is now way more advanced, it takes into account manufacturing constraints, for example, so the customers can pick if the design is going to be suitable for 3D printing with or without support or machining, with all the parameters, the material, the flow. So, it's really up to the customers to pick what are the main parameters for the manufacturing and for the physics and then navigating all these trade-offs and picking what is the best design. So, it's really shifting the role of the designer from [taking a lot of] time to find, possibly, a suboptimal design to navigating many optimal designs and picking the best one for a more system level optimisation.
So, what challenges would say your software product is solving?
MP: I would say the biggest value we're giving to the customers is massively shortening the design and the R&D iteration. So, historically, you have design engineers working on CAD software, you have then analysts working on CAE software, for example, for the structural analysis software, etc. Usually, you have a cycle here. This group of engineers are talking and iterating on continuously improving the design based on the analysis, sometimes based on the experimental topic and be plugged into these design workflow, and massively speed up, even cutting all these design cycle and giving in a matter of hours, rather than months and years, what is the optimal design, which is not only found in a couple of hours, but it's also more intricate, more efficient, more tailored to the specific working condition of the customer. Some of our customers use the final design as it is. Some other customers use the design as a very good first guess what an optimal design is and they continue. So, in this context, it's very important to emphasise the fact that Toffee is not replacing engineers, it's not replacing existing software, it's augmenting with existing teams and the existing design workflow in terms of software.
Get your FREE print subscription to TCT Magazine.
Exhibit at the UK's definitive and most influential 3D printing and additive manufacturing event, TCT 3Sixty.
Tell us more about the physics-driven simulation capability of the ToffeeX platform.
MP: It's very important. What I believe and what is my message, always, is in a bigger space of artificial intelligence and data driven application, let's not find more solutions than problems. So, in the space of generative design, we are focusing more on the simulation driven, physics driven design, generative design, because this gives a big generality, general application, we simulate the physics, the core physics is the same, regardless of the working condition for an aircraft or for appliances or for cars. So, the core of the software is based on physics, on heat exchange, on mixing, on chemical reaction. So, it's simulation driven, we use the data instead, to capture some phenomena that are not captured by the physics. One example is improving the turbulence model. In that case, it is specific to the working condition. And we have some machine learning, some neural networks for aerospace applications, some orders for heat exchangers, for batteries. And so, we believe the software is very fast, you can find designs in a couple of hours. So, there's no need to go faster there, because we're really shrinking from months to hours, but using instant data, to create a hybrid solution to improve accuracy and to correct some design.
You cater for both the additive manufacturing market and the more traditional manufacturing markets – what are the opportunities for ToffeeX in both markets?
MP: When I was talking about our story, I didn't really mention, we started to fill the gap, there was a market opportunity on how to fully exploit the potential of additive manufacturing. And this was really the core, the beginning of our journey. But very soon, we realised that still, the majority of the market, the majority of the industry, was not and is still not ready, probably, to adopt additive manufacturing. And so we adapted our software to contain manufacturing constraints inside the optimisation process, inside the design process. So, now you can click from our platform your manufacturing technique and the software will create the best possible design within the limitations of your manufacturing technique, it's very easy. When you see a picture from our design, if it's for 3D printing, you see these three-dimensional intricate geometries. If you click, for example, for stamping or for machining or for chemical etching, you see geometries that are more 2D, some extrusion, which are clearly suitable for traditional manufacturing. In terms of the different industries and the different applications, we see the biggest application on additive manufacturing is on aerospace, and also on energy. And then down to automotive, for example, I would say mass production, automotive and appliance, obviously, they are still much closer to the traditional manufacturing technique, but this has to be mass production.
How would you assess the maturity and industry acceptance of generative design?
MP: I would say I saw the journey starting more or less five or ten years ago; it's improving. So, there is much more awareness of the adoption of generative design. I would like to mention that there's still some education to be done in the market. But the positive is there are not many companies doing the different aspects of generative design from structure, lattice, as we do simulation driven. And it's really the shifting towards the digitalisation of industry and winning the biases that these tools are not replacing engineers but they are empowering engineers to go much faster and much farther. So, I saw this literally in the last five years [begin] to be massively understood by the engineers. This was probably first in aerospace and energy, but now is arriving in automotive and mass production as well. So, I'm very confident that in another five years having a generative design tool within the software portfolio will be the norm.
How do you imagine the ToffeeX product will evolve? What capabilities are your users working for?
MP: Yeah, so I mentioned that we're mostly working on fluid dynamics. There's an optimisation for the behaviour of 3D components and the majority of our customers are using it for heat exchange and for flow uniform, for flow control. But there is much more, for example, we just developed a tool to optimise also for carbon capture. So, our development of the technology will mostly follow two ways. One is including more physics, chemical reaction, changing phase, etc. Those are just examples. And the other one is including more and more manufacturing constraints. We work with additive manufacturing, of the 3D printers you can pick the material and the resolution, the overhang, etc. We work with milling, tuning size, etc. stamping, but there's so much more. There is casting. There is injection moulding. So, by doing this, we empower more engineers of different industries to use the potential of generative design.
