TOffeeAM: Topology Optimization For Fluids Engineering & Additive Manufacturing
What happens if we let the fluid design the part for us.
The role of software in Design for Additive Manufacturing has shifted in recent years from blocker to enabler. CAD systems that have been developed for decades hand in hand with other manufacturing technologies were understandably not optimized for creating, or simulating the geometry capable of being manufactured additively.
We now see a new generation of software entering the market from both existing CAD vendors and startups that not only enables the creation of geometries suitable for additive manufacturing, but also sometimes guide the creation of engineering solutions using simulation and/or computational design.
TOffeeAM is a spin-out of research at the Imperial College London focusing on simulation driven optimization of thermal-fluidic applications.
In the following interview with Antonio Di Caterino we learn about their software, how they got started and where they are seeing traction with the design of thermo-fluidic applications for both additive and subtractive manufacturing processes.
Can you introduce us to TOffeeAM, your software and your team?
To explain what TOffeeAM is, it is probably best to start by answering a question we are often asked: "what does your name mean?"
TOffeeAM is an acronym for Topology Optimisation For Fluids EnginEEring and Additive Manufacturing.
Today TOffee is a cloud-based, multi-physics software that automates the design process to maximise the performance of thermo-fluid components based on multi-objective optimization.
The core of TOffee is based on a multi-physics solver which solves the thermal and fluid dynamics equations within the specified design space. The software iteratively uses the data output by these simulations to create a design that is optimised to match users' desired output.
Imagine you need to re-design – or even better to design from scratch – a heat exchanger. With TOffee, you only need to input a few parameters and the software will design it for you.
As inputs, the software only requires a design domain, which is the space in which the final part can be built, i.e. your spatial constraints, the property of the fluids and materials involved, and at least one target - or more - to optimize (for example minimizing the pressure loss across a manifold).
The software is very fast and can produce designs in minutes and hours, not weeks and months. Moreover, being a simulation-driven optimisation, you don’t need to design and then validate your design in a different software – the output of the optimization is already based on high fidelity simulations. Everything runs under the same hood.
In terms of our history, the project traces its roots back to 2015 with the work of the three co-founders: Francesco Montomoli, professor at Imperial College London and current CEO of the company, Marco Pietropaoli and Audrey Gaymann, then PhD students and now, respectively, CTO and CAIO of TOffeeAM. Following the completion of their PhDs, TOffeeAM was founded in 2019.
In 2015, Additive Manufacturing was increasingly becoming a concrete and vibrant reality, but one question still remained: how can we exploit this incredible geometric and constructive freedom to design new components that are more efficient than what could be done before?
Although topology optimisation and generative design for the structural optimisation of components was something already well known and widely used, the scenario was different when it came to optimising components to maximise thermo-fluid-dynamic performance.
Because of the nonlinear and unintuitive nature of the equations that describe the physics of fluids, to think that you can rely solely on human intellect and ingenuity to come up with an optimal design is almost impossible.
The idea was simple: What if we let the fluid design the part for us?
What was needed was to rely on the equations themselves, and let them determine the best shape to give the final component.
Just as the course of a river is traced to minimize the resistance of the water flowing through it or the veins of leaves are routed to spread nutrients as efficiently as possible, Toffee solves the equations that describe the phenomena involved, creating a design that maximizes efficiency.
If that sounds difficult… It is because it is!
The team currently consists of 13 people, including 6 PhDs, with a background mainly in the aeronautical industry and experience from companies such as GE, Rolls-Royce and ESA.
Your software was originally developed to solve thermal management problems in gas turbine engines, how did your team approach that engineering problem?
That's right, the first project we worked on and the one that also gave birth to the code is related to the thermal management of turbine blades. It was a natural choice given the environment of the Department of Aeronautics at Imperial College London and the experience of our CEO Francesco.
For gas turbines, turbine blades are often the limiting component.
The highest temperature in the cycle occurs at the end of the combustion process, and it is limited by the maximum temperature that the turbine blades can withstand.
To put it into numbers, these fluids can reach temperatures of 2000K, a value that well exceeds the melting point of the blade material (around 1400K). Turbine blades therefore need to be actively cooled.
To cool a turbine blade, different methods of cooling, such as internal air channels can and have been used.
It’s important to understand that even small improvements in heat transfer or reductions in maximum temperatures can have significant effects on the lifespan of various components.
It has been shown (Bunker, R.S., 2009. The effects of manufacturing tolerances on gas turbine cooling) that a 1% increase in the nominal operating temperature of a turbine blade can lead to a 33% reduction in the lifespan of the blade.
In a field like aeronautics, where people understandably try to be conservative and preserve what works, attempting only incremental improvements, we wanted to adopt a completely different approach, one that would allow us to create a new design that was not conditioned by existing solutions.
We see that human beings are inclined to take inspiration from their surroundings and from what has been done before, and this inevitably leads to a conservative approach when it comes to innovation.
With TOffee, we had a tool in our hands that freed us from any preconceptions about what the final design of the cooling channels should be, and in the end it was a winning choice.
The most revolutionary idea, if you want to call it that, was to emulate the natural processes that lead to evolution and result in structures that are perfect for the function they perform, without excesses or deficiencies.
These principles, translated into mathematical models and powered by physical models, have led TOffee to be a tool capable of automatically designing optimised cooling channels, valves, heat exchangers and so on, which not surprisingly recall complex natural structures, surpassing (as nature does) human intuition.
With this in mind, using TOffee made it possible to generate very efficient cooling designs that can minimize the temperature the blade has to withstand, with an increase in efficiency of up to 40% compared to traditional designs.
To give some context to the cost savings this improvement leads to, a single turbine blade can cost more than $10,000 to manufacture, and a single gas turbine engine will contain hundreds of blades.
A 33% longer lifespan can result in savings of millions of dollars per engine across the lifespan of the engine.
Again in automotive applications, thermal management seems to be where you gained your initial traction, can you explain what are the similarities and differences between designing for aero and auto applications?
When it comes to Automotive there’s a plethora of possibilities related to thermal management optimisation.
Nowadays, the shift to electric vehicles has further increased the need for good thermal management.
Not only the engine, but also all electronics need to be cooled in order to work efficiently, increase the lifespan of the battery, or avoid problems that could be catastrophic, like thermal runaway.
Given the expertise of the team on thermal optimisation, it was more as a natural consequence to approach the automotive industry from the same angle.
While it is true that the transition to electric vehicles has had a positive environmental impact compared to traditional internal combustion engine cars, it is also true that the production and disposal cycle of a battery can often produce an equivalent amount of emissions over the years.
Being able to increase the lifetime of batteries has the dual advantage of reducing costs and reducing the environmental impact even more.
This is a common trait of both aerospace and automotive, both sectors where increased performance translates into cost reductions of various kinds, from production costs to environmental costs.
Speaking of differences between the two sectors, however, the automotive sector, especially the commercial one, is more related to mass production.
It is therefore important to have a tool that is not only able to optimise and improve efficiency, but also to do so with limits on the final design to make it compatible with rapid production, whereas Additive Manufacturing is mainly used in 'premium' sectors such as F1, luxury brands, or indeed aerospace.
How do you incorporate lessons learned from both applications into your software?
This is a very interesting question because we actually learnt lessons with each new application that allowed us to improve the software.
Being able to touch the real needs in the R&D world of large industrial realities allowed us to gain insight into current and future issues, and inspired us to change our vision as well.
These experiences have allowed us to understand how each of these fields (but also segments within the fields themselves) have different needs, and we have adapted our software to suit different environments.
Probably the most important lesson was the realization that, although the initial idea of the software was to develop a tool that exploited the full potential of AM, we realized that this is not necessarily the best way forward for everyone – other manufacturing methods are still very common and will remain widely used for mass production.
Therefore, TOffee now optimises for 'best design within your specific manufacturing constraints and within your costs of manufacturing'.
How do your customers incorporate TOffeeAM into their existing workflow with other software and manufacturing processes?
TOffeeAM is a so-called SaaS, software-as-a-service, which can be accessed by subscription via the cloud, a bit like Netflix.
Currently, all pre-processing is done within the platform, from the assignment of boundary conditions to meshing, and then the analysis and optimisation.
Since the software is simulation-driven, at the end of the process it is possible to either obtain the design created by the software and download it as an STL file or to download all the parameters such as temperatures, pressure values and so on.
All that remains for the user to do is to generate a geometry to initially feed to the software as a design domain and then send the file obtained through TOffeeAM to a slicer.
The ultimate goal of TOffeeAM is to provide engineers and designers with a tool that simplifies and improves their work, enabling them to exploit the full potential of additive manufacturing.
In addition to continually improving performance, TOffee aims to be a complete 360-degree software tool that allows the user to go from idea to printed design in the shortest possible time, from design space creation to optimisation and slicing to create the file for printing, as well as create APIs to integrate our optimiser with other software existing today.
At what point in the design process do you think engineers should consider TOffee and what about an engineering problem would indicate TOffee could enable them to design a solution.
I think the best way to use the software is to give it as much freedom as possible.
We have often been asked to use TOffee to improve this or that component to achieve a higher efficiency within an assembly that is now well frozen.
While it is certainly possible - and has been done - to use the tool as a means of redesigning and improving a component, being able to have a design that is free of any bias allows the software to reach its full potential.
As mentioned, in fact, we are often constrained as designers to adapt to solutions that come from common use or experience, but the change of direction we want to propose with TOffee is towards design beyond what the human mind can think of.
However, it remains open also to the possibility of using the tool simply to have a guide on the best direction to follow, and then simplify it to adapt it to the internal production logic of each company for example.
There is only experimentation to be done, but the advice is to use it as much as possible in the early stages of a project.
How can someone get started with your software?
Getting started with the software is quite simple. It is possible to access a 14-day trial that has some limitations compared to the standard licence.
We generally prefer to have an initial meeting with a potential user to understand how the software can be useful to them and to help them set the best parameters for optimisation.
If you are interested in trying it out, just send an email to firstname.lastname@example.org or book a slot at the link https://calendly.com/toffeeam/30mins-toffeeam
DfAM is a reader-supported publication. To receive occasional emails and support the research, consider becoming a free or paid subscriber.