Who is the client?
One of the world’s largest aerospace manufacturers.
What was the problem?
Today’s aircraft require the cabin to be pressurised and the temperature to be controlled by onboard environment control systems. Aerospace manufacturers use simulations to test these systems, but the emergence of new propulsion technologies (such as electric and hydrogen) have made these simulations increasingly complex and time-intensive to run.
That was certainly the case for our client.
To gain the necessary insights for each aircraft, their simulations needed to analyse between 10–15 physical parameters. They wanted a new model that relied on significantly fewer parameters while delivering the same predictive accuracy.
What was the Expleo solution?
We began with a data study of our client’s existing simulation. From there, we combined our data with a machine learning algorithm which allowed us to identify a smaller set of parameters that could be used for testing.
This gave us the foundation to build a new physical surrogate model. This model only needs to analyse three physical parameters — rather than the previous 10–15.
How did that help?
Our new physical surrogate model has replaced the previous simulation. And the results have been dramatic. Our client can gain the same level of insights on cabin pressure and temperature, but faster.
What were the results ?
- Our client is benefiting from a huge simulation timesaving.
- The new physical surrogate model delivers 95% accuracy — matching the testing precision of the previous simulation.
- This same methodology can be re-used to conduct simulations on several other aeronautical related systems.
Could it work for me?
Yes, our approach can help improve simulation testing for any industrial manufacturer. By analysing your current digital environment, we can develop new models that reduce testing time and complexity.
In the aerospace sector, manufacturers often have to run complex simulations that place a huge strain on time and resources. Using machine learning algorithms, we can develop more efficient models that deliver the same insights and predictive accuracy while drastically cutting testing time.
Anthony Laffond, Data Technical Leader, Expleo
Speak to an Expert