Project Description

Thermal pre-dimensioning of electronics

close-up of electronic circuit board with processor

As it as been presented at our user conference ScilabTEC, Valeo is using Scilab for early stages pre-dimensionning of electronic systems, basing its approach on mathematical methodologies (modeling with thermal impedance).


Several tools / Several Teams


Mix Between several Tools:

  • CFD Simulations or Tests for transfer functions extraction
  • Mathematical (Scilab) and/or SPICE models for temperature prediction

Collaboration by several teams:

  • Thermal CFD Simulation expert for transfer function extraction
  • Electronic Designer to run various configuration or complex profiles
  • Improve communication between the two teams for equipment optimization


Transfer Function Extraction

Depending on the project status, different approaches are relevant:

  • For Project in design phase
    • CFD Simulation model is build
    • Number of run is equal to number of power injection point (m)
    • These are transient runs with step injected power (relatively simple)
  • For Project with existing hardware
    • CFD simulation or
    • Measurements using thermal sensor, IR camera, …

Temperature environment and Injected power during CFD simulation or test shall be close to real configuration (linearization at operating conditions).

“0D” Model Construction

The Model chosen depends on the method used for temperature profiles calculation. Two options are possible:

  • Temperature predicted using mathematical software (Scilab)
    Pure mathematical model needed
  • Temperature predicted using SPICE simulation software
    Assembly of RC network circuit needed

Use of SPICE simulation needs to add a calculation step to define the equivalent RC networks with potential additional errors

Two models are always generated to identify amount or errors produced at each step (due to system linearization then to RC model generation).

Here is the expression of the mathematical “0D” model:



In the end the mathematical approach allows much faster modeling and processing, giving accurate results (see graph attached), appropriate for early stages decision strategies.