Task
The gap between the panes of an insulating glass unit is filled with a gas mixture. The background is the energy requirements for buildings under the German Building Energy Act (GEG).
The goal was a way to predict the thermal conductivity of a given gas mixture.
Approach
dAIve GmbH trained a neural network in dAIve Predict that predicts the thermal conductivity of gas mixtures. The training data come from a range of sources in the technical literature.
The case was created as a demonstration, not as a client project.
Result
The trained network predicts the thermal conductivity of a gas mixture.
The model could be transferred into databases or a control system. This was not done in the demonstration case.
Source: Application example from the former daive.de website (section “Material”)