More than simulation.

Trained on millions of simulations, Flowsheet Copilot gives you recommendations in each stage of new process design

Build expert systems from no prior knowledge

The engineering of new production facilities heavily relies on previous processes, patents or bibliography. Flowsheet copilot uses reinforcement learning to explore all possible solutions, from no previous knowledge, and learns from the experience he adquires interacting with process simulation. AI learns which combination of equipment leads to a good quality product, low CAPEX and OPEX while minimizing waste and GHG emissions. After that, all the knowledge is in your hands.

Flowsheet generation

Generate flowsheets from scratch. All the process streams and utilities

Cloud computing

Train and execute all your models in our cloud platform

No data needed

Data is generated automatically through simulations, while AI learns from them

From lab to full scale

How it works?

Key benefits

Reach engineering decisions
faster than ever before

Intemic allows chemical engineers to make design decisions quickly from early stages,
combining conceptual and detailed engineering into a robust automated sequential
decision-making process.

With Flowsheet Copilot, engineers can explore more options, with less time, and take the product to market faster.

Define your raw materials, your products, the kinetic models of your reactions, if any, and the exploration requirements that best suit to your application.

Let yourself be guided by Flowsheet Copilot, while you are the pilot. Get recommendations at every step for an economically and environmentally optimized process design, simulating digital twins of existing equipment or technology available in the market. Recommendations are flexible and data-driven: As you modify the compositions of your feed, the recommendations will optimize in real time adapting to your raw materials.

Currently, engineers have lenghty iterative processes for each new project, which makes it last up to months.

After a market research, a new product of interest is visualized and its manufacturing seems to be sustainable in both economic and environmental issues. The objective is to evaluate the viability of its manufacturing on an industrial scale, so the company decides start the project with a group of engineers.

Engineers look for previous similar flowsheets or patents, to avoid having to search from scratch, which lead to suboptimal results if an innovative process wants to be evaluated. Selecting the different technologies and unit operations it's a difficult task, because it's based on engineers intuition. These early decisions contribute greatly to the final costs and consumptions of the facility and its environmental impact, and are key to making the project profitable after doing all the detailed engineering. The CAPEX and OPEX costs are estimated with an error of 30%, which means uncertainty and risk for the project.

Each unit operation have to be manually introduced and connected in the flowsheet, based on the conceptual design. A lot of assumptions based on previous patents or engineering bibliography are made to be able to build and run the first simulations, whether its on a process simulator or in excel file.

Process simulators are like videogames with complex rules that chemical engineers play, guided by their intuition and experience.
Simulators give errors, calculations can be long, and for every modification done in the flowsheet (recycles, modifying parameters, ...) the mass and energy balances change, some unit ops remain at suboptimal conditions, and a lot of procedures need to be done again.
It feels like solving a rubik cube.

Once the flowsheet structure is defined, economic and environmental metrics can be improved by optimizing a set of parameters with methods like MINLP (Mixed integer nonlinear programming). Apart from the fact that the optimization is based on already defined unit operations and its conexions (which does not lead to global optimum), it takes a lot of time to define all the parameters, its boundaries, and perform the MINLP calculations.

Once the equipment characteristics are defined, engineers spend months searching for the right equipment, having to adapt to what is on the market, because building custom equipment is very expensive. After asking for prices and availability, if the purchase is finally made, the equipment will probably not behave exactly as in your simulations.

Call To Action

Want to see it in action?