Digital Environmental Data

Statistical methods play a central role during the design and assessment of vehicle components with respect to durability and reliability.

The entire process starts with the description and modeling of the usage variability, which arises due to the combination of different mission profiles or driver behaviour and the varying environmental conditions in different regions of the world. For instance, the mixture of applications for a tractor of a certain class such as plowing or transport varies from customer to customer as well as the regional conditions such as topography, size of fields etc. Accordingly, the loads of the machines or vehicles vary considerably and a proper statistical modeling is required including factor models, ANOVA-type methods, handling high variances etc.

Define the Process for Durability

First, we have to provide answers to following questions:

  • How to model customer loads?
  • How to evaluate data from customer uses?
  • How to plan measurement campaigns to determine load data?
  • How to translate the results into test tracks or test programs?
  • Which failure probabilities to prove and how?
Durability
© Fraunhofer ITWM
Distributions of Loads and Strength

The next step is to derive reference loads, which are used within the development process and are fundamental for release tests. Measured data is analysed and extrapolated to a given design life based on the classification of factor models (which type of load cases are relevant) and the application of usage models (how often do these load cases occur in a certain customer group). High quantiles of the corresponding distributions are used to define the reference loads.

Volkswagen Commercial Vehicles Benefits from the Methods

That process was, among others, executed in a common project with Volkswagen Nutzfahrzeuge and reported during the VDI conference Commercial Vehicles 2015 in Eindhoven. Some of the applied methods, e.g. extending the evaluation of measured data using geo-referenced information and the statistical analysis based on factor models, we have been presented in more detail during the DVM-conference of the AK Betriebsfestigkeit 2016 in Steyr.

 

Prove Reliability

For demonstrating the required reliability of a component based on release tests, safety factors are introduced to cope with the remaining uncertainty due to lack of knowledge in the extreme tails of the load and strength distributions. In addition, statistical reasoning is needed for estimating how many specimens have to be tested in order to prove a certain reliability with a certain confidence.

Finally, we evaluate the reliability tests using statistically validated methods. This applies to comparatively simple component tests as well as to very expensive and long-running complete vehicle tests. The optimal planning of such tests is a decisive point.

  • Are many short tests or very few longer tests better suited?
  • How can you get the maximum amount and quality of information and benefit from a few tests?

We develop and apply methods for supporting the sketched process with focus on the vehicle industry including construction and agricultural machinery. However, other applications such as wind turbines or railway industry are also covered. Most of the methods are applicable as well within the assessment of energy efficiency or fuel consumption.

References

  • Weyh, T., Speckert, M., Opalinski, A., Wagner, M.: Planung einer Messkampagne durch Osteuropa mittels der Fraunhofer Software VMC („Virtual Measurement Campaign“), VDI-Bericht: Nutzfahrzeuge 2015 Commercial Vehicles 2015 Truck, Bus, Van, Trailer, VDI-Berichte/VDI-Tagungsbände 2247, Vol. Bericht 2247, 2015.
  • Speckert, M.; K., D.; Lübke, M.; Halfmann, T.: Automatisierte und um GEO-Daten angereicherte Auswertung von Messdaten zur Herleitung von Beanspruchungsverteilungen, DVM Bericht 143, 2016.

 

Application Fields

 

Project News / 15.12.2020

Autonomous Driving in the Commercial Vehicle Sector for Trucks

In the IdenT research project, we are contributing with our mathematical expertise together with our partners to actively drive the transformation of autonomous transport in the area of trucks.

Geo-referenced Analysis and Virtual Measurement Campaign (VMC®)

Since several years, the department MDF at Fraunhofer ITWM is investigating the analysis of geo-referenced data to support and improve the derivation of design loads as well as the assessment of energy efficiency by statistical reasoning. The approach is mainly motivated by the desire to understand usage variability and the resulting scatter in loads or fuel consumption and to provide methods for making use of that information within vehicle development.

 

Virtual Environments for Developing, Testing, and Validating Autonomous Vehicle Functions

 

REDAR

Our 3D laser scanner vehicle REDAR (Road and Environment Data Acquisition Rover) is used for high-precision georeferenced measurement of roads and environments.

 

EU Project »ECOTR AVID«

The European LIFE funded project »ECOTRAVID« (»Emission and Consumption Optimized Transport Missions Using Virtual Drives«) aims at demonstrating the efficiency of a virtual driving simulator based on the software »Virtual Measurement Campaign« (VMC) developed at our institute to reduce the fuel consumption of heavy duty transport and the associated CO2 emissions.

Mathematics for Agriculture

 

Cognitive Agriculture

In the Fraunhofer lighthouse project COGNAC we work on the optimization of agricultural processes through modern methods of data analysis or Machine Learning.