As new materials and production methods are discovered to help achieve this greater power density, testing becomes ever more important as an approval tool—only standing up to the tests will prove what is destined to get a new generation of vehicle moving.
These are the challenges that drive the engineers at Porsche Engineering to excel. Testing requires them to define loads that are representative of day-to-day use and to figure out scenarios for applying these loads to the separate components and assemblies. All relevant loads and functions need to live up to requirements and testing aims to verify this within the shortest-possible time frame. To do so, the engineers employ all manner of instruments, from simulation to engine bench to test drive.
Simulations use physical and behavior-based models of driving environment, driver and vehicle. Modern computer models, such as Porsche Engineering uses, possess a depth of data accumulated over decades. They allow reliable depiction of highly complex OEM requirements, for example the capacity to use a powertrain in a variety of vehicle models. To identify in sufficient detail the load profiles for powertrain components across a variety of product applications, simulating what are referred to as load collectives is indispensable. The load profiles invariably produce conflicting interests in terms of strength, weight and costs. These conflicts need to be resolved before the second step, in which design collectives are drawn from the various load profiles. The design collectives are themselves in turn used to dimension the units that will form the module system.
This process is highly dynamic because its application itself modifies and refines a vehicle’s driving and comfort qualities. These changes to driving characteristics can result in considerably increased component strain and exceed the design’s failure limits. These findings are then fed back to simulations of component load using new application data to calculate load collectives. Without ever using any actual components, an iterative process continually compares the load collectives with the design collectives to identify necessary modifications.
The very first prototype components can already be used for engine bench trials. Usually, components are subjected to short-duration trials that examine various factors at a highly accelerated rate: oil supply, bearing load, tooth-flank load, tooth-root load are tested in regard to the design’s failure limits in defined stages up to component failure. This kind of testing is extremely component-specific and can only convey a limited level of information. What it can tell you, however, is whether a certain design actually works at all.
Once development progresses further, full units become available for trials. What is referred to as endurance cycle testing has proven particularly suitable. It tests loads at a comparatively accelerated rate while at the same time supplying highly reliable data on a unit’s sub-components. During testing, the subcomponents are put together as a unit consisting of combustion engine, traction motor, transmission, control units, drive shaft and half-shafts. Tire contact with the road surface is simulated using consumer units.
By employing actual control units for engine and transmission, the engineers can examine application data sets in direct conjunction with the powertrain. Here, too, the goal is to identify the failure limit under a simulated powertrain load. Intelligent online control of the differential coupling ensures correct distribution of drive torque to the front and rear axles. With the test run progressing further, online adjustments to gradient resistance allow compensation for variables like changes in drive power or system efficiency due to tribological variations or wear of components and lubricant fluids. As a consequence, the powertrain reaches a high level of maturity even before its first use in a prototype vehicle.
The most accurate depiction of all powertrain components’ wear is achieved by performing real-time tests on a vehicle test stand, although this method is by far the most time-consuming. Such testing supplies the most reliable data on all components. Real-time testing can be based on driver models that incorporate vehicle parameters or be based on real driving data measured in a trial vehicle.
Real-time testing does not progress at an accelerated rate but does, however, take into account the entirety of driver variables and the ambient conditions of the acceptance trial. The test uses vehicle speed and output speeds as setting variables. These values can be specified either directly as velocity or artificially through a driving profile. The aim is to minimize the potential for error at the least-possible deviation from the specified targets. Though no online adjustment of driving resistance is available for compensation, intelligent driver models facilitate dynamic real-time tests. The model’s driver logic decides autonomously whether to emulate the speed profile or to drive freely within the specifications of an allowed speed range.
While the powertrain is under development, overall vehicle design continues, too. Consequently, real-time testing on the test stands is supplemented with measurement data from test vehicles. Road tests are held in actual road traffic and on dedicated test sites, for example in Nardò. When performing vehicle measurements, Porsche Engineering aims to obtain data of sufficient statistical reliability on what the characteristic powertrain loads will be in everyday vehicle operation over driving distances of up to 15,000 kilometers. The test drivers drive only on dry roads during the trials as dry conditions entail the greatest material stresses. Measurements take into account all essential variables, including route composition, driving style and vehicle cargo. The engineers subsequently evaluate the measured values and derive load collectives for precisely defined customer usage profiles.
The measuring equipment for the road tests is designed to be impervious not only to usual day-to-day driving but also to the extreme loads produced by abusive handling, for example like what engineers call a “performance start.” To obtain measurement data from vehicle operation over a period of three weeks and a distance of 15,000 kilometers, the measuring equipment used is particularly reliable in terms of system startup, redundancy and storage capacity. The primary focus of load-collective measurements are wheel torque values. Together with the gear ratio, wheel torque allows it to determine precisely what loads the powertrain is being subjected to and what damage this is causing on the powertrain’s various components. What makes these measurements so difficult are the rotating parts (which can only be measured using sliding contacts or telemetry), the powertrain moving in all directions at once, the temperature effects (heat radiating from the exhaust system, transmission or engine), electromagnetic interference (EMC), and the effects of the weather.
The measuring instrument used here is what engineers call a measuring flange. Because of their high component rigidity, measuring flanges are especially resistant to abusive loads (performance start, μ-split).
• Acceleration sensors that record vehicle acceleration along the x-, y- and z-axes
• GPS receivers that determine position using longitude and latitude as well as elevation, gradient, and curve angle and radius
• Thermometers that measure the ambient temperature, component temperatures, and the temperatures of the operating fluid
Digital measurement values primarily describe the communication signals exchanged among the control units via the vehicle’s bus system. While only a few years ago a low number of CAN systems sufficed to ensure data communication, more recent vehicles additionally utilize the FlexRay system. FlexRay’s architecture allows it to transfer far larger amounts of data. The measurement sequences’ signal scope usually lies between 100 and 600 signals measured at a rate of between 500 hertz (torque) and one hertz (temperature). To record the data, Porsche Engineering employs powerful data loggers designed for large volumes of data and high transfer rates. Transmission development requires additional transfer protocols (for example XCP, CCP).
The system is started by means of a central switch that controls a central power supply system capable of serving all amplifiers and sensor blocks. This is intended to prevent any incorrect measurements being made due to the system components inadvertently remaining deactivated. The same also applies once measuring has been completed: any components inadvertently remaining switched on will discharge the vehicle battery before the next measurement run begins.
To reduce data loss during measuring, there are specific saving points along each route that the drivers are required to observe. This way, should the measuring equipment encounter problems between two waypoints, only the incorrectly measured route section will need to be repeated rather than the entire route. However, this results in multiple files for each measurement run, which need to be merged again in a subsequent step. Measurement data is correlated to route type by means of a route signal, for example generated from GPS coordinates when on a mapped route. Another method is to generate a route signal using additional measured values (speed, steering angle, acceleration and others) that are then computed in a complex process. This produces a highly dynamic signal capable of processing specifically even standstill times in a traffic jam or when driving in urban areas. If blocked roads force a driver off the planned route, different measurement records will no longer correlate. In such cases, this method allows automated allocation of a route type (for example city, country road, freeway).
Over the course of measurement, a number of measuring errors can occur resulting from the nature of the systems used. These need to be rectified during a subsequent follow-up and of course before any measurement data is handed over to customers.
Such measurement errors result first and foremost from flaws in the measuring flanges and shafts used to record torque and include torque offsets, calibration skips, spikes, drifts or “no-values.” Because—due to the sheer volume of raw data—manual rectification of these measurement errors would take months, Porsche Engineering has automated detection of these kinds of errors. Unusable data is removed by search engines, eliminating the need for manual validation of the data sets. All this aims at one thing: to supply the customers as quickly and as efficiently as possible with the reliable data required for designing and configuring an essential vehicle assembly. Porsche Engineering is fully aware of the great responsibility this entails and takes it on with confidence in every project.
Porsche Engineering pours its time and resources into keeping its testing scenarios at least one step ahead of tomorrow’s automobile technology. Specifics include the programming of simulations that depict driving loads in even greater and more varied detail. Automated driving is set to greatly affect the composition and severity of powertrain loads in future, opening up potential to, for example, utilize load reductions for tailored lightweight design. Our cutting-edge testing incorporates these considerations, just as it incorporates considerations on how future human behavior is likely to shift test parameters.
Text first published in the Porsche Engineering Magazine, issue 01/2018
Panamera Turbo S E-Hybrid: Fuel consumption combined 2.9 l/100 km; CO2 emissions 66 g/km; electricity consumption (combined) 16.2 kWh/100 km