About Us

Why we started

When we founded ThermoPhys in Trondheim, Norway, in 2024, one of our main motivations was to close a gap we had been watching from the sidelines for years: advances in thermodynamics and process modeling took decades to reach industry. The most advanced and accurate models were often described as too slow or unstable to be useful in process simulation. One consequence is expensive overdesign and slow innovation in emerging technologies such as CCS and hydrogen.

Another motivation was the lack of modern and collaborative features in established process simulators. Many of them felt to us like software from the 1990s, and much of the workflow they encouraged was inefficient, and not well suited to ensuring safety or encouraging innovation either. We didn’t feel that this combination justified the extreme prices we had to pay.

But AI changed everything

As we started building our cloud-based software in 2024, however, AI suddenly took off. That opened up a completely new possibility, and we started building around it. The architecture and design of our software have been developed so that the underlying models, calculations and workflows can be easily accessed by AI, and can scale as AI gets better. And it gets better very, very fast.

This has already changed what we can do with the software. We have gained capabilities that we did not originally plan for, simply because we built the platform around AI rather than trying to add AI to an existing process simulator afterwards. Some engineering workflows are now 1000x faster than before.

Why we are unique

There are very few people in the world with deep scientific expertise in thermophysical properties and calculations. We happen to have brought several of them together at ThermoPhys.

Why does this matter when building the next generation process simulation platform? Because when a difficult process simulation fails, the problem is nearly always in the thermodynamic routines. We know that from hands-on experience. And if those routines were developed by somebody else, which is usually the case, who do you call when the simulation does not work?

It is also a matter of computational speed. Since we started in 2024, we have speed-up some of our most accurate thermodynamic models by more than 1000 times, while making them substantially more robust and predictable.

Our bottom-up and careful approach where accuracy and robustness are essential, means that not only can you trust the answer, but also the AI that builds the process for you, or checks that its safe.