arrow_back_ios

Main Menu

See All Pages
arrow_back_ios

Main Menu

See All Pages
arrow_back_ios

Main Menu

See All Pages

Valeo basing vibration tests on realistic specifications helps Valeo deliver reliability and performance at right cost

The key to success for automotive suppliers is to design components that meet reliability targets, keep warranty costs low, while at the same time achieve low production costs. Traditionally, automotive components are tested using generic specifications based on historical damage profiles with little connection to the vehicle that the component will be used on. The result is that the component may be overdesigned for some types of vibrations while remaining vulnerable to others. 


Valeo has pioneered a new approach that transforms road load data acquired from the vehicle on which the components will be used into a test signal that realistically represents the fatigue loads seen by the component during its life on the actual vehicle. Valeo worked with HBM Prenscia to implement this approach in nCode GlyphWorks test data analysis software and uses this software to quickly generate synthesized test signals to drive fatigue test systems. The result is that the component is designed to meet the specific requirements of the vehicle, resulting in a reduction both in manufacturing and warranty costs.

Valeo is one of Europe’s largest car parts suppliers, providing components for nearly all global car manufacturers. Valeo is organized into four business groups: Comfort and Driving Assistance Systems group (interior controls, security systems); Powertrain Systems (engine and electrical systems, transmissions); Thermal Systems (climate control, compressors, engine cooling); and Visibility Systems (lighting systems, wiper systems). Valeo Service also distributes products to the automotive aftermarket. Valeo has about 120 production sites in 27 countries and more than two-thirds of its sales come from Europe.

Solution requirements

  • Quickly generate synthesized test signals to drive fatigue test systems.
  • Convert time history data acquired into realistic vibration specifications that provide a much closer match to the loads experienced in real vehicles.

Reliability at the right cost

Designing products that deliver the right performance and reliability at the right cost is critical to success in the automotive components business.
“But in reality different vehicles place very different loads on their components,” Delaux continued. “For generic specifications to cover many different types of vehicles requires that they incorporate large safety factors but even then they do not necessarily cover the complete range of damage seen in particular vehicles. Historically we have seen failures and failure modes in the actual components on the road that were not seen in testing with generic specifications. We’ve also seen failures in testing that were not seen in the real vehicles. Furthermore, designing to generic specifications makes it necessary to overdesign components which adds considerably to their cost.”

Spectrum analysis for dynamic characteristics
Waterfall analysis rotating machinery analysis

Figure 1: Spiked data before (top) and after (bottom) correction

Converting time histories to realistic specifications

Valeo has been working for about a decade to develop methods to convert time history data acquired into realistic vibration specifications that provide a much closer match to the loads experienced in real vehicles. These time histories are generally captured by customers on a test track using accelerometers, strain gauges, pressure sensors, etc., and in some cases acquired by Valeo itself. The French military with its GAM-EG-13 standard, the American military with its MIL-STD-817 and the French scientific society, the Association pour le développement des Sciences et Techniques de l’Environnement (ASTE), have all developed fairly consistent methods of converting time history signals into vibration specifications for driving fatigue test equipment. Delaux said: “I would like to warmly thank Christian Lalane, Henri Grzeskowiak, Bruno Colin, Bernard Colomies, Jacques Vanuxeem, Lambert Pierrat and Armand Delanghe, who have built these strong methodologies.” The challenge for the automotive supplier that attempts to pursue this approach is accepting large volumes of time history signals from test tracks around the world and developing an efficient and accurate method of converting them to fatigue test specifications.
Figure 2: GlyphWorks is used to detect and correct peaks, deviation, loss of signal, saturations, etc. Close-up of corrected data - blue is original, red is correctedThe first step in this process is to input the time history data into GlyphWorks using the time series input glyph. The anomaly detection algorithms within GlyphWorks are used to detect outliers. A measurement specialist reviews the anomalies and corrects spikes while keeping an eye out for other defects, such as saturation, that might require the data to be discarded.

Figure 3: Stationary test values including signal, probability density, running mean and running standard deviation
Figure 4: Sources of noise in different frequency ranges

The first step in this process is to input the time history data into GlyphWorks using the time series input glyph. The anomaly detection algorithms within GlyphWorks are used to detect outliers. A measurement specialist reviews the anomalies and corrects spikes while keeping an eye out for other defects, such as saturation, that might require the data to be discarded. 

Global statistics are then computed on each channel such as max, min, mean, standard deviation, RMS, skewness and kurtosis. These stationary test values are used to quickly assess whether the signal is centered around its mean and its peakiness. Statistical analysis is performed to determine whether the mean and standard deviation values remain constant over the sampling period. These analysis help the measurement expert determine whether to use a deterministic or probabilistic approach to calculate the fatigue damage spectrum.
Figure 5: Fatigue damage spectrum calculation with GlyphWorks
Figure 5 shows an example of a fatigue damage spectrum calculation using GlyphWorks. The deterministic approach was used in this example because of the non-stationary nature of the signals. The fatigue parameters used are typical for metals: Basquin slope b=8 and dynamic amplification factor: Q=10. The same parameters are used for derivation of the synthesized spectra. The objective is to find a power spectral density or swept sine whose fatigue damage spectrum is equivalent to the fatigue damage spectrum generated from the time history. This is accomplished in GlyphWorks using an iterative approach that involves adapting the power spectral density until its fatigue damage spectrum overlays with the target fatigue damage spectrum within a certain margin. In the example, the synthesized equivalent power spectral densities are calculated to have an equivalent damage in an exposure duration of 30 hours per axis.

A statistical safety factor is added to account for variations in applied loading and the fatigue strength of the component. An additional safety factor is used to account for the limited number of durability tests that will be performed and for the reliability targets of the carmakers. The total safety factor is a product of these two numbers.

The realistic specification that was created in the example above was compared to generic specification provided by the automotive manufacturer as shown in Figure 6. The overlay shows that the synthesized specification exceeds the generic specification at frequencies under 200 Hz, typically where the engine orders are important contributors to vibrations. On the other hand, the synthesized specification has a lower magnitude than the existing profile in the high frequencies. In this example use of the generic specification may lead to overdesign at high frequencies and underdesign in the low frequency range.
Valeo has used these methods in the design and validation of over 400 engine cooling products. The company typically begins by using fatigue analysis software to evaluate the PSD specification against the current design. If the simulation shows the current design cannot stand up to the realistic spectra then the design is iterated until engineers are confident that it will pass the test.

The realistic specifications produced by the new method have proven to be considerably more accurate than generic specifications. For example, in one case an automotive OEM asked Valeo to use a generic swept sine test on a plastic component. Valeo measured the time history data and also computed the damage based on a realistic profile. Designs were created to meet both specifications. The design based on the realistic specification saved 100,000 Euros in tooling cost, 0.60 Euros per part in manufacturing costs and also provided a weight savings of 10%. “Much higher but more difficult to measure savings are achieved through our ability to more accurately design products to meet real world conditions using reliable specifications,” Delaux said. “Parts designed with generic specifications sometimes are unable to stand up to loads generated by specific vehicle configurations. Designing parts to reliable specifications can generate huge reductions in warranty costs by ensuring that the part is able to meet the requirements of the application.”
Figure 6: Realistic test specification (red) overlaid on generic specification (blue & green)

Test data processing and durability analysis

nCode GlyphWorks is a data processing system that contains a powerful set of pre-defined tools for performing durability analysis and other insightful tasks such as digital signal processing. Designed to handle huge amounts of data, GlyphWorks provides a graphical, process-oriented environment that contains leading analysis capabilities for saving both time and money in environmental qualification and product validation.

Ready to achieve success through failure prediction?