Search

Introduction

Quick Overview

Getting started

Release Notes

Measurement Acquisition

Measurement Acquisition

Test Bench Experiment Setup

Bench testing is critical for calibrating the mathematical models behind virtual thermal sensors. This guide outlines best practices for recording high-quality measurements that ensure your virtual sensors accurately reflect real-world thermal conditions.

Introduction

Virtual thermal sensors are designed to be deployed in real-time embedded hardware, providing accurate temperature estimations. Calibration against experimental measurements ensures these virtual sensors reliably reflect real-world thermal conditions.

The quality of your test bench measurements directly influences the fidelity and reliability of your virtual sensors.

General Guidelines

Sensor Placement

  • Position external thermocouples to observe temperatures in otherwise inaccessible parts
  • Create detailed diagrams of sensor placement positions
  • Document the types of sensors used (thermocouples, NTCs, etc.)
  • Install sensors in multiple radial and longitudinal positions for comprehensive coverage

Data Recording Format

  • Use standardized formats:
    • CSV files with timestamps and signal labels
    • MAT files containing data tables
  • Twin Fabrica can import these formats directly (some preprocessing may be required)

Essential Data to Record

  • Temperature profiles across multiple components
  • Power model inputs (current, speed, frequency)
  • Cooling system parameters (inlet/outlet temperatures, flow rates)
  • Ambient temperature conditions

Test Conditions Coverage

  • Capture warm-up, steady-state, and cool-down phases
  • Test multiple power and cooling setpoints
  • For electrical motors, evaluate different rotational speeds
  • For cooling systems, test range of inlet temperatures and flowrates

Special Test Recommendation

For liquid-cooled components, conduct an "isolated cooling test":

  • Keep system in still condition (zero power)
  • Change cooling temperature and flowrate
  • This evaluates thermal material response without power loss uncertainties

Acquisition Frequency

  • Set temperature acquisition frequency to capture the fastest thermal responses expected in your application
  • More data points are better than too few

Managing Initial Conditions

Initial condition mismatches can compromise the accuracy of virtual sensor calibration. Follow these guidelines to ensure proper initial condition management:

Best Practices

  • Start tests from isothermal initial conditions whenever possible
  • For non-isothermal conditions, continuously record temperatures during power-off periods
  • Avoid using test segments without known initial conditions
  • Conduct tests with zero power while varying boundary condition

Best Practice Summary

  • Accurate Sensor Placement: Position sensors to capture temperatures in all critical areas
  • Complete Test Phases: Include warm-up, steady-state, and cool-down phases
  • Record Cooling Parameters: For liquid-cooled systems, document inlet/outlet temperatures and flow rates
  • Diverse Operating Conditions: Test various load levels and temperature conditions
  • Redundant Measurements: Prefer data redundancy over gaps
  • Steady-State Conditions: Maintain steady cooling during measurements
  • Ambient Recording: Document ambient temperature close to the tested component
  • Start from Isothermal: Begin tests from isothermal conditions when possible
  • Continuous Recording: For non-isothermal tests, record temperatures during power-off periods