Integrated Corner Technology - a Key Element of Zero-Prototype Engineering

Integrated corner technology - a key element of zero-prototype engineering

by John Whitehead, Technical Manager, Vehicle Dynamics, MIRA

John Whitehead BA, moved to MIRA in 1985 from Michelin's Vehicle Dynamics Department, which he joined in 1968. He has considerable experience in tyre-adhesion testing and in vehicle suspension analysis and development using subjective and objective methods including laboratory testing. John was instrumental in MIRA's decision to invest in a suspension kinematics and compliance facility. He has been involved in projects for many vehicle manufacturers worldwide and represents the UK on the ISO's sub-committee concerned with vehicle handling.

One of the hottest issues in new vehicle development is the move towards zero prototype engineering (because prototypes are extremely expensive and take time to build). Do without them and you save time and lots of money. In this, and the following four articles, MIRA staff describe the advanced technologies - many of them ground-breaking - that the company has developed, integrating its varied skills and expertise to achieve zero-prototype capability in vehicle corner engineering.

A popular misconception of how a new car gets from drawing board or, more accurately today, from computer screen to production is that the shape, style and appearance - the most visible features of a new car - are the driving forces.

However, a car is a complicated machine, and design is not simple. For example, a bodyshell must not only look good and provide the required packaging, it must be aerodynamically efficient, structurally sound (to support the rest of the car), collapse correctly in a crash and have the appropriate stiffness and vibration characteristics (to support the ride, handling and refinement goals). The suspension must not only have the correct characteristics to achieve the ride, handling and refinement goals, it must also be durable.

Design to production has traditionally been a long step. But over the years, with improving capability of design, the trial-and-error development process has reduced in duration. Nevertheless, a new model development programme may last two to three years. Often 'mule' vehicles are used to prove certain aspects, such as the engine or the suspension. Following this are prototype phases, typically two, where development is conducted on hand-built, or 'soft tool' vehicles. Final minor development is done on pilot-build, or 'hard tool' prototypes built off production tooling. Only then can Job One mark the start of production.

This process is not only time consuming, it is also expensive, particularly in the building of 'soft tool' prototypes. Since cars were invented the automotive industry has sought to reduce both time-to-market and development costs. In the USA the emphasis has been on reducing time-to-market, while in Europe cutting development costs has been the high priority. So it is not surprising that zero- or virtual-prototype engineering, which meets the demands on both sides of the Atlantic, has become such a hot issue.

A definition of zero-prototype engineering is the deletion of 'soft tool' prototypes from a development programme. To make zero-prototype engineering a reality requires robust initial design (getting the physics right) and improved fidelity of computer simulation. To achieve the latter requires improved representation of some components within the computer simulation.

For a suspension system, computer representation of the metal parts generally gives reasonably accurate results. This is because a metal part that is working within its design envelope behaves elastically and linearly. A spring is a good example; when subjected to a force a spring will deflect. When the force is removed the spring will return to its original position (elastic behaviour). If it is subjected to twice the force, it will experience twice the deflection (linear behaviour). Unfortunately, a suspension system contains parts that behave non-linearly and with friction or hysteresis. These parts are the rubber bushes, the dampers and tyres. The rubber components suffer hysteresis; although in essence they behave elastically, when a force is removed they do not recover fully to their original positions or shapes. Dampers suffer from both friction and, due to the internal valves, some hysteresis.

Computer representations fall broadly into two groups: analytical and empirical. Analytical representations are used for objects whose behaviour and characteristics can be accurately described by mathematical equations. Usually this means for objects that behave elastically and linearly, such as the metal spring. An analytical representation is an emulation of the real thing, and can be used to predict, for example, the behaviour of any spring, given the correct dimensions and material properties. An empirical representation is a simulation that depends upon testing of the real thing to formulate a system (which may be a mathematical expression) that returns the correct output for the required input. Such a simulation is valid only for the object it was designed to represent, for example a tyre of particular make, model and size.

A computer simulation of a complete system will generally comprise analytical and empirical representations. Thus, a simulation of a suspension system will include analytical representations of the metal parts and empirical representations of the bushes, dampers and tyres.

Hitherto, empirical representation of components exhibiting friction, hysteresis and/or non-linear behaviour has been difficult and has led to errors of prediction. This is the case in dampers where, for example, the representation could replicate the amplitude non-linearity but not the hysteresis or frequency dependency. However, a recent MIRA research programme, concentrating on improving the representation of the non-linear components within suspension systems, has produced significantly improved results.

MIRA's research programme focused on three areas - handling, ride and durability. Handling simulation used a different tyre representation from that used for ride and durability. For handling, the tyre representation used was the well known and tested Magic Formula which essentially returns cornering forces for given inputs of steering demand and vehicle speed. For ride and durability, a new Rigid Ring representation was tried. This simulates tyre stiffness and vibrational characteristics, essentially returning vertical and longitudinal forces transmitted to the car for given tyre inputs and vehicle speeds. For bushes and dampers, the representations used incorporated neural networks. The advantage of such representations is that they can simulate the hystereses and frequency dependencies of components as well as their amplitude non-linearities.

The car and its suspension system were represented from design data using ADAMS. Using MIRA's kinematics and compliance rig, which measures the geometry changes of a suspension system due to the application of displacements and forces, the suspension of the real car, a production model, was adjusted to give characteristics close to design intent.

The real car was fully instrumented to measure whole body dynamics parameters for handling, and vibration parameters for ride and durability. In both cases, wheel force transducers were fitted. Standard objective handling tests were performed on both the computer representation and the real car. Excellent correlation was obtained for wheel forces and for vehicle response to steering inputs between the simulation and the real vehicle. For ride and durability, a contributory factor in achieving correlation between the computer output and test results from the real vehicle was more accurate reproduction of the wheel inputs through creation of the 'virtual proving ground'. This was done by detailed representation of the surfaces of MIRA's proving ground used for testing the real vehicle.

Although empirical representations of components have been based upon test results from real components, these representations could be based on data constructed from desired behaviours of components. Thus the representations could be incorporated into the computer simulation of the total vehicle at the design stage. This process would help formulate the target behaviours of the components required to achieve target behaviour of the vehicle.

MIRA is now able to support its design services with the type of engineering analysis described above, and thus bring zero-prototype engineering a step closer to reality.