Verification of hardware-in-the-loop as a valid testing method for suspension development

Abstract

A need for a cost effective, versatile and easy to use suspension component testing method has arisen, following the development of a four-state hydro-pneumatic semi-active spring-damper system. A method known as hardware-in-the-loop (HiL) was investigated, in particular its use and compatibility with tests involving physical systems – previously HiL was used predominantly for Electronic Control Unit (ECU) testing. The suitability of HiL in the development of advanced suspension systems and their control systems, during which various vehicle models can be used, was determined. A first step in vehicle suspension design is estimating a desired spring and damper characteristic, and verifying that characteristic using software simulation. The models used during this step are usually low-order, simple models, which hampers quick development progress. To predict vehicle response before vehicle prototype completion, many researchers have attempted to use complex and advanced damper models to simulate the vehicle’s dynamics, but these models all suffer from some drawback – it is either based on empirical data, giving no indication of the physical parameters of the design sought; it may be overly complex, having many parameters and thus rendering software impractical; or it may be quick but based on the premise that there is no hysteresis in the damping character. It can be seen that an obvious answer exists – use a physical commercially available or prototype damper in the software simulation instead of the mathematical model. In this way the suspension deflection, i.e. the true motion of the damper is used as excitation, and the true damper force is measured using a hydraulic actuator and load cell. The vehicle mass motions are simulated in a software environment. This is basically what HiL simulation does. The HiL method was verified by comparing HiL simulations and tests to globally accepted testing methods, employing widely-used vehicle models: linear single-degree-of-freedom (SDOF) and two-degrees-of-freedom (2DOF) or quarter-car models were used. The HiL method was also compared to a non-linear physical system to verify that the method holds for real vehicle suspension geometries. This meant that HiL had to perform adequately at both ends of the suspension-testing spectrum – base software and real system simulation. The comparison of the HiL and software/real system simulation was done using the “Error Coefficient of Variance” (ECOV) between the compared signals; this quantitative measure proved very sensitive and performed dubiously in the presence of signal offsets, phase lags and scaling errors, but remains a tangible, measurable parameter with which to compare signals. Visual confirmation was also obtained to back the ECOV values. It was found that even using a relatively low-force actuator, the HiL simulation results followed the software/real system responses well. Phase lags and DC offsets in the HiL simulation’s measured signals (as well as the real systems responses) has an adverse effect on the performance of the HiL simulation. Special attention must thus be paid to the zeroing of equipment and the amount/type of filters in the system, as these affect the HiL results dramatically. In all, HiL was proven to be a versatile and easy to use alternative to conventional mass-based suspension testing.