A trend toward the integration of analytical investigations and testing hardware prototypes has emerged as a result of increasing pressures related to shortened development cycles and the desire to save costs. The combination of physical and virtual testing can accelerate the design process by identifying and subsequently eliminating physical and virtual prototype deficiencies, such as model parameter inaccuracies.
In the early design stage, FEA and multibody dynamic analysis have been widely used for vehicle development. One of the most important issues for a good FEA model is to establish accurate loads and boundary conditions, but there is often no clear path for the analyst to establish these. One of the methods is to model the full vehicle together with the digital roads and obtain the load and boundary conditions by using multibody dynamic analysis—e.g. via software such as MSC.Software’s ADAMS. This approach is called Virtual Proving Ground. The advantage of this method is that little or no experimental data acquisition is required. The disadvantage is that an accurate full-vehicle model is needed. But accurate vehicle models are difficult to create because some of the components, such as tires, are especially difficult to model.
For many years, automotive companies have used laboratory simulation testing to improve the quality, durability, and reliability of vehicle components. Laboratory simulation testing provides accurate durability and performance information, and it reduces product development cycle time. To further accelerate development, it would be valuable to replicate the laboratory test in the analytical world and conduct virtual testing.
To conduct virtual or laboratory-based testing, some kind of road load data is usually required. However, when the prototype is not available, it is impossible to collect road load data. Therefore, a generic method to calculate the desired road response from existing information must be used. The RPC Pro Software iteration process can then be used to reproduce the road data using a virtual test rig and a vehicle model. Consequently, the load time history and fatigue life of each component can be predicted long before a prototype is available. After the initial prototype is available, real road load data can be collected and lab simulation testing can be conducted. The road load data from proving ground and from generic road approach can then be correlated. The virtual and physical test results can also be compared.
One of the virtual testing examples is to simulate spindle coupled full vehicle durability tests for the purpose of completing virtual durability evaluations on components and full vehicles before a prototype is available. The reproduction of measured spindle loads is achieved on a virtual model of a passenger car coupled to a six-degrees-of-freedom spindle coupled test system.
In this virtual testing activity, inaccuracies in the vehicle model and/or road load data can cause inaccurate model response if only the road loads are applied to the vehicle model. For this reason, a non-square RPC iteration method can be used to try and reproduce not only spindle loads but also expected spindle motion. During RPC iterations, weighting values can be used so that the best compatible solution for load and motion will be reached. As a result, both load and motion can be adequately reproduced using the virtual model. The ability to find a compromised solution between load and motion and to produce meaningful results in the presence of model and road data inaccuracies is another important advantage of virtual testing.
One of the methods aimed at overcoming the limitations of virtual testing is real time hybrid simulation. The idea of this approach is to test the hard-to-model vehicle components/subsystems and model the rest of the well understood components/subsystems. The test system and the analysis model form a hybrid system that is solved in “real time.” This approach is named mechanical hardware in the loop (mHIL).
One example of a mHIL system is the Four-Corner Damper system developed by MTS Systems. In this system, the four physical dampers are tested using four damper test machines. In the vehicle model, the dampers are replaced with single component forces. The inputs of the test system are displacements calculated by the vehicle model. The measured damper forces are sent to the single component forces representing the dampers in the vehicle model.
By conducting mHIL tests, Nissan was able to find an unwanted irregular behavior in vehicle motion of a vehicle design. The cause of the irregular behavior was traced to a specific component, and a countermeasure was successfully adopted prior to the vehicle prototype test. The Four-Corner Damper system was considered by Nissan Motor to be “beneficial to secure high performance and quality of the first vehicle prototype.”
Another mHIL example is the Quarter Car System developed by MTS Systems. In this system, the vehicle model calculates tire vertical motion, shock tower vertical motion, and tire lateral force. These signals are used as input commands of the Quarter Car test system to excite the physical specimen that includes a tire and a suspension with an active air spring that is difficult to model. In this hybrid system, the active air spring is controlled in part by the vehicle motion that is calculated by the virtual model. The vertical shock tower force that is measured from the Quarter Car test system is applied to the vehicle model so that realistic virtual vehicle motions will be predicted.
To address the limitations of real-time techniques, MTS Systems has developed an alternative approach to hybrid simulation for durability applications. The Hybrid System Response Convergence (HSRC) technique enables physical and virtual components with multi-degree-of-freedom dynamic behavior to be simulated as if they were a coupled, hybrid system. The various physical and virtual components of the system are operated sequentially as open-loop systems, rather than simultaneously, thus avoiding the requirement to run in real time. The HSRC method uses an iterative approach that modifies the inputs to the physical system until the physical and virtual system responses achieve a compatible dynamic solution at the hybrid interface.
When the property of the physical specimen is not critically rate-dependent and the requirement of real-time simulation can be relaxed, more hybrid simulation methods are available. The University of California, Berkeley has developed a soft real-time hybrid simulation program, OpenFresco, that can connect a test system with any FEA software that allows a user-programmed element to be defined. The user-programmed element, called an “experimental element,” in the analysis model acts as a representation of the component/subsystem that is tested in the lab. The experimental element communicates with the test system through OpenFresco to send command and receive response. Since nearly all commercial FEA and multibody dynamics analysis software has user-programmed element function, they can all be linked to OpenFresco and be used to conduct hybrid simulation. LS DYNA, Abaqus, and other software packages have been used in this approach.
Virtual testing and hybrid simulation, including mHIL, HSRC, and soft real hybrid simulation, are effective tools to study vehicle performance and durability properties. These tools bring the expertise and knowledge of the vehicle testing and modeling activities together to enable more efficient and accurate design evaluation. With proper usage, these tools can provide vehicle component loading, motion, and durability information at an early development stage, well before full vehicle prototypes are available.
This article is based on SAE Technical Paper 2011-01-0029 by Shawn S. You and Dave Fricke of MTS Systems Corp.