The ArbeitsKreis KUndenKOllektive (KUKO) was founded ten years ago by AUDI, BMW, DaimlerChrysler, Porsche, VW and LMS. The KUKO group had set as its goal to analyze the real customer usage in a statistical sense. Since accurate load data is a critical factor in durability engineering, the KUKO participants joined their forces to find more reliable, better manageable and less expensive ways to gain load data that can be trusted completely. This intense and yearlong focus on load data analysis and synthesis has generated innovative insights that have led to a new methodology, which is now offered by the Engineering Services division of LMS. The KUKO engineering services package includes the equipment needed to gather and process the test data, support in selecting representative groups of drivers and dedicated data processing software that tunes the prediction models and generates the load data you can count on.
Accurate Load Data Required
The challenge of durability engineering is designing parts with the right durability performance. Failing durability engineering generates overdesigned parts that are too heavy and expensive, or underdesigned parts that are unreliable and potentially even life-threatening. Despite the availability of top-class durability simulation solutions, the durability engineering process is still somewhat at risk, due to the fact that the fatigue-life predictions involved are highly sensitive to load data. To be trustable, load data must accurately represent the damage that is accumulated in car parts as a result of the realistic usage of real customers.
Today, different methodologies are used to generate load data. A mixture of different road surfaces consisting of specially designed test tracks, public roads and synthetic events are quite commonly used by professional test drivers. Test cars are instrumented with high-count measurement channels to acquire large sets of forces in real-time. The extensive testing equipment and special wheel force transducers required make these setups expensive. They also generate large data amounts that are only manageable with high effort, especially when data collection campaigns are executed in different regions and by different test drivers. The fact that these traditional approaches are based on the long time experience of the car manufacturer, and do not take into account real statistic customer parameters is probably their biggest disadvantage.
Real Customers – Real Roads
The KUKO approach introduces a new concept that enables automotive OEMs to perform efficient, reliable and representative measurements within cars that are driven by real customers during their regular everyday life. The key benefits are that the gained loading results are statistically representative and that the data amounts and costs involved are limited. KUKO’s modeling capabilities make it possible to reduce the channel count, to use accelerometers instead of expensive wheel force transducers and to focus exclusively on predefined events of high interest.
Predicting accumulated damage
During the first phase, a limited number of cars are fully instrumented with the purpose of making the KUKO model ready for use in the specific group of car types that will be part of the focused test program. Measurements are performed during many different driving situations on all kinds of road profiles. Measurements of a limited number of easy-to-measure parameters of customer behavior are combined with loading measurements based on wheel force transducers. LMS TecWare based Rainflow-counting techniques are used to calculate the accumulated damage from the measured loadings. KUKO prediction models are tuned to correlate between the reduced set of measurement channels and the accumulated damage of wheel forces. Sophisticated statistical modeling techniques are used for this purpose. An advantage arising from KUKO modeling is that most of the required information during the real test program is acquirable through just a few measurement channels. In addition, these channels are easy-to-measure, since regular accelerometers can be used instead of more expensive transducers. Other information that KUKO takes into account is gained through the CAN bus.
Large-scale Customer Data Collection
The second phase represents the real test program and takes between six months up and one year and allows sufficient number of kilometers to be accumulated, usually more than 10.000 km per customer. A lot of test cars participate in this test program and are driven by real customers. The KUKO approach is very suitable for use in large-scale test programs, as the measurement hardware required is very limited and affordable. No wheel force transducers are needed. Each individual test car is instrumented with a small box with dedicated hardware that connects to input channels that describe the customers behavior, e.g. three accelerometers at the car’s center of gravity, and one on1 a front wheel (vertical acceleration). The test box is automatically started with turning the car key, and also accesses the CAN bus to acquire information that relates to velocity, gears, brakes, ABS, etc.. During the test program, the predefined events of interest are automatically recognized. These events typically represent driving situations that are most relevant to the fatigue life of car parts, such as significant car accelerations and decelerations, high speed cornering, crossing pot holes, shifting gears, or driving over Belgian blocks. During the tests, the data is acquired and the recorded event series are processed. Along the way, special data reduction techniques are applied to the time series to further downscale the amount of data. All data are consolidated in event-specific statistical distribution functions.
Statistically Relevant Data
The third phase represents a statistical inquiry through questionnaires. Its purpose is to gain the road type and the driver type distributions that correspond to the specific usage. The event-specific statistical distribution functions obtained in phase two are weighted with the road type and driver type distributions obtained from the questionnaires. This way, the durability target is designed to correspond to a specific geographic region and corresponding driver style.
KUKO is the right approach when it comes to designing reliable durability targets. This approach provides good and statistically relevant durability targets in terms of pseudo-damage values, which can be used to validate actual test tracks and scenarios. With the obtained durability targets, virtual simulations and rig tests can occur with greater accuracy, leading to car parts with the right durability performance. The approach is being applied by car manufacturers participating in KUKO and is also commercially leveraged at other automotive OEMs. Next to durability-specific target setting, the KUKO method can also provide insights that relate to fuel consumption, engine torque and other requirements.
