Simulation Replaces Prototyping

Dec. 9, 2004
As simulation gets more realistic, prototyping becomes an expensive last resort.

Simulation replaces prototyping

Dick Kading
CAE Market Manager
LMS North America
Troy, Mich.

The spindle from a front suspension, designed by Audi, shows fatigue damage predicted by simulation.

Audi engineers model the front end of a new car as part of the suspension, then run it through simulated test driving to gather load data.

To ensure doors and latches withstand repeated slamming, BMW modeled the doorclosing process, starting with door-frame components.

Engineers reused a door model from a previous engineering effort.

A model of the door latch was inserted into the door model, and both were then placed in the door frame.

To redesign a mining truck, engineers modeled it then put it through simulated work cycles of loading, traveling, unloading, and then traveling to reload.

In recent years, the automotive and off-road industries have increasingly used modeling to simulate the operation of vehicles and subsystems prior to prototyping them. This saves time and money. But it isn't as easy as it sounds. Certain types of analysis, for example, need specific types of data, much of which is unavailable until prototypes are built. And how do you test designs that aren't even finished?

At Audi, designers combined testing and simulation to improve predictions on durability (or life, in terms of cycles) of key suspension components. This lets them redesign components to eliminate weak spots before building prototypes.

In the past, automakers used simulation software to estimate component durability. But such software needs loading data for each component, which requires prototypes, often instrumented with more than 100 strain gages and other sensors. But such data isn't usually known until late in the design process, a point where major changes are expensive and affect many other components and subassemblies. Furthermore, any major changes would call for new measurements on each component.

For these reasons, previous attempts at predicting suspension loads prior to prototyping have generally focused on road testing an earlier version of the vehicle, measuring loads at the spindles. These are the shafts on the front and rear suspensions where the wheel and bearing mount. The redesign is then based on these loads. But loads are highly dependent on vehicle mass, spring and damper characteristics, and other factors. So the accuracy of estimated loads from previous designs is questionable, even when factors are introduced to account for differences in the weight between old and yet-to-be developed vehicles.

Measured road profiles can be used as a starting point, but in practice they are not always available. Test engineers prefer to use actual measure response data and adapt it to new designs. To sidestep these limitations, Audi wanted to convert loads on spindles into spindle displacements, then use them to evaluate new suspensions in a multibody simulation (MBS) that calculates loads on individual components for various road and driving conditions. Loads and material properties are then fed into the FE models in LMS Virtual.Lab Durability to predict each component's fatigue life.

Audi runs MBS on a previous-generation vehicle to create estimates of spindle displacements. The accuracy of these initial estimates is improved by using LMS Hybrid Road software. It uses mathematical methods to iteratively adjust calculated displacements, comparing measure and computed loads. Once iteration is complete, the displacements can drive the MBS of the new design.

Resulting displacements are used to put an MBS of the new vehicle through its paces, generating load histories which are fed into the fatigue-life solver, with material properties, geometry information, and unit-load stress results from FEA on individual components. Local-strain and critical-plane techniques calculate life predictions for body-in-white and suspension components, the goal of the project. The local-strain approach is based on linear elastic-stress analysis and includes an elastoplastic correction factor that accounts for plastic deformation in local areas. The critical-plane approach examines several potential crack-initiation planes and determines the one with the biggest damage.

As vehicle-crash regulations tighten and comfort standards rise, car companies are spending more time on vehicle body and door design. For example, adding safety structures and softer rubber elements as seals to reduce noise increases loading on components when doors are closed. And changes in trim packages for increased passenger comfort, such as new acoustic material, can change the durability of doors and body components.

To avoid problems and maintain its reputation for quality, BMW tests physical prototypes to guarantee the durability of door and body components. The test delivers reliable results, but it has its drawbacks. It requires expensive prototypes, and evaluating hundreds of doorslam events takes considerable time. And if testing uncovers a problem, designs must be changed, prototypes modified, and tests rerun, adding more time and cost to vehicle development.

So BMW wanted to augment prototype testing with predictor simulation for evaluating the door and body components' ability to standup to being slammed. In preliminary work, they defined three phases in a door slam. First, the door hits and deforms a rubber grommet on the body. Energy absorbed by the rubber is converted to heat and elastic energy. Then the door hits the closing hook and is stopped. In the last phase, stored elastic energy tries to push the door back, but the hook prevents that. Capturing time-dependent contact between components is essential in simulating load transfer between door and frame.

The FE model used to simulate door slams consistsof the door, body frame, grommet, and lock. The FE mesh of the frame was taken from a model that had been used for stiffness calculations. Engineers used just the area around the front-left door to reduce the number of elements and refined the mesh density around the lock hook, the area they thought would be critical. They also reused a door model from earlier side-impact simulations. They connected the frame and door with joint elements, creating a model of about 60,000 elements and 65,000 nodes.

The rubber grommet was modeled as a single row of solid elements with a skin of contact shells. To account for rubber's nonlinear behavior, it was modeled out of low-density foam. Material parameters of the rubber were calibrated by comparing calculated and experimental data from a simple test setup. The plastic buffers, coating, and bump stops were also modeled as low-density foam. The lock was made of solid and shell elements. Surfaces on sliding contact were modeled in greater detail to avoid unrealistic peak loads when contacting components slide over the mesh edges. Prestressed spring elements were included, as well as a damper element, to account for friction in the lock.

Rigid spiders, a mathematical constraint relationship, connect beams to sheet elements rather than simple nodeto-node connections. This simplified beam placement, making it easier to connect dissimilar meshes.

Engineers validated the model by comparing simulation and experimental results, which showed an acceptable level of correlation. It became clear that the force acting on the closing hook is significant. Validation also showed a high sensitivity to lateral positioning of the hook, local stiffness around the lock, and the rubber's properties.

Transient simulations for three different closing velocities and two different designs were performed using Ansys LS-Dyna FE. In these simulations, the modal superposition technique accounts for rigid-body door motion, extensive deformation of the rubber seal, and the door's modal oscillations. Global as well as Rayleigh damping were also used. Stress outputs from the FEA became inputs for LMS durabilitysimulation software, which converted localstress histories into fatigue-life predictions.

"Both prototype and simulation results showed that the modified design improved the total life by a factor exceeding 20," according to Mr. G. Tokar of Structural Analysis at BMW. "And results predicted by the strain-life approach accorded better with physical test results than results produced by Rupp's approach, which turned out to be too conservative. Using rigid spiders instead of simple node-tonode connections caused the spot-weld beams to transfer higher moments than those that really occur. We adjusted this by establishing a new back-calculated W^hler curve for the modified spot-weld modeling."

The project demonstrated that combining nonlinear FEA and fatigue predictions is a good way to go. Once refined, simulation could predict fatigue life for body and door components without waiting to build and test more prototypes.

Liebherr Mining Equipment in Newport News, Va., builds huge, ultraclass mining trucks. The ratio of payloadto-vehicle empty weight is critical to productivity and operating costs. The truck's overall gross weight, however, is limited by tire capacity. This means that every pound removed from the vehicle adds to its payload.

Conventional ultraclass trucks use a single-axle design and the frame carries payload and dump-body weight, so it experiences substantial vertical bending loads. This design has been optimized to where the newest trucks offer payloadtoempty vehicle weight ratios of 1.4 to 1.6. Liebherr engineers designed tier TI 272 truck so that it weighs just over 165 short tons empty and carries a 320-ton load, almost twice its own weight.

Liebherr used two independent rear axles, each oscillating around its own suspension pivot. This means road undulations and obstacles are primarily absorbed by just one wheel pair without affecting the other pair. And even the involved pair continues carrying about the same load. And in curves, the four rear tires travel at different speeds, which improves traction in wet conditions.

The new design has wider dumpbody pivot suspensions. At 55% of the total vehicle width, they are almost three times as wide as pivots on conventional trucks. This wider stance delivers more stable load distribution and decreases wallowing on less-than-perfect roads. The weight of the payload is also spread further apart, reducing bending stress on the frame.

Unlike conventional trucks, the frame does not support the dump-body and payload. Instead, they are carried in front by two hoist cylinders mounted behind the cab. In back, they are carried by two dump-body pivots that, in turn, transfer weight through suspension struts and wheels to the ground. This lets Liebherr use a lighter frame, the primary reason the TI 272 hauls more payload than other trucks with the same weight.

The new design is a major departure from existing trucks, so it is no surprise that prototypes suffered fatigue-life problems, particularly in the frame. Liebherr decided to redesign the frame to address fatigue and increase the payload. The first step was to test prototypes under working conditions with strain gauges. Results confirmed that frame loading was substantially different than on conventional trucks but they were unable to measure forces acting on components, which is needed for FEA.

Liebherr modeled the truck using LMS multibody simulation software, a multibody dynamics program based on a differential algebraic solver. Results correlated well with strain-gage readings and provided more information than physical testing alone, including forces and load-case analyses on all components.Engineers refined the model to contain over 30 rigid bodies representing all major moving and some welded or bolted components and assemblies. The frame, for example, is treated as a flexible body and is represented by modes from an FE-beam model.

The dump-body is modeled as a rigid body connected to other components by bushings with a stiffness determined by FEA to simulate flexibility. Components are connected with appropriate joint elements modeling spherical, revolute, cylindrical, or translational joints. Bushing elements model rubber pads, and transnational spring-damper-actuators model nonlinear stiffness and damping of struts and hydraulic cylinders. Expression-force elements replicate torque for brakes and electric motors, forces for steering and hoist cylinders, as well contact between hoist cylinders and guides. Contact elements model axle-rotation stops that limit axle pivoting. And curve elements specify characteristics of electric motors, hydraulic pumps, oleopneumatic struts, and tires.

Acceleration, braking, steering, backing, dumping, and towing are initiated by proportional-integral-derivative controllers separately or in combination with one another. Engineers stitched these situations together to simulate complete hauling cycles.

Time-domain multibody simulation from the LMS program calculated forces used as load cases for FEA. Liebherr's engineers used insights gained from simulation to strengthen the frame and other structural components, without putting back too much weight. Other useful results from the simulation include each body's global position and orientation, velocity, and acceleration. Position data is used to create realistic animations of the truck in operation which have proven valuable in understanding complex dynamic behavior, and in concisely conveying sophisticated information. Most importantly, simulation delivered the predictive engineering information needed to improve the design without relying on costly prototypes. The TI 272 is now in production and there are plans to carry modeling one step further by using it to generate load histories for durability analysis.

Sponsored Recommendations

How BASF turns data into savings

May 7, 2024
BASF continuously monitors the health of 63 substation assets — with Schneider’s Service Bureau and EcoStruxure™ Asset Advisor. ►Learn More: https://www.schn...

Agile design thinking: A key to operation-level digital transformation acceleration

May 7, 2024
Digital transformation, aided by agile design thinking, can reduce obstacles to change. Learn about 3 steps that can guide success.

Can new digital medium voltage circuit breakers help facilities reduce their carbon footprint?

May 7, 2024
Find out how facility managers can easily monitor energy usage to create a sustainable, decarbonized environment using digital MV circuit breakers.

The Digital Thread: End-to-End Data-Driven Manufacturing

May 1, 2024
Creating a Digital Thread by harnessing end-to-end manufacturing data is providing unprecedented opportunities to create efficiencies in the world of manufacturing.

Voice your opinion!

To join the conversation, and become an exclusive member of Machine Design, create an account today!