Designers traditionally have valued the use of both physical testing and computer analysis in developing components and systems for a wide range of products, from automobiles and aircraft to medical equipment and industrial machinery. For the most part, however, testing and analysis are performed as separate activities, generally by two distinct groups that typically have little contact with one another or with designers.

A preliminary finite-element analysis (FEA) or dynamic simulation may be performed in the early stages of an automotive transmission design to determine stresses and loading, for example, with results guiding the engineer in sizing components and estimating fatigue life. Later in product development, physical testing of a prototype might show premature failure, in which case further computer analysis is done to pinpoint and correct the problem.

Several of these analyze-build-testfix cycles might be required before the product performs satisfactorily, adding significant time and cost to the development cycle as engineering change orders are prepared, drawings modified, new designs approved, tooling reconfigured, and hardware re-built. This often results in schedule delays and cost overruns as designs are analyzed and tested over and over. Worse yet, quick fixes just to get the product out the door often cause further problems in other areas.

This traditional approach has two major flaws. First, analysts often must make simplifying assumptions and speculate how complex variables will play out in the actual product, so their “virtual prototype” computer model may not accurately reflect the operation and behavior of real world parts. Secondly, critical data from physical testing generally is not available until late in product development, when changes are the most difficult, and expensive.

Combining real-world with simulation

A new approach called hybrid simulation overcomes these obstacles by combining real-world test data of existing parts with dynamic analysis of the entire mechanical system early in product development. The tools are coupled through embedded data management capabilities that correlate the two types of data in a single virtual prototype model. In this method, known component behavior is used to calibrate computer analysis, thus eliminating unnecessary testing, simulation, guesswork, and the risk of overlooking important performance parameters.

In the design of a new vehicle transmission, for example, test data from standard parts such as bearings are readily combined with CAD models of new structural components to represent the entire subassembly. The technology lets the product design be systematically refined at each stage of product development, with systemlevel target product specifications cascading down to subassembly and components level requirements.

Hybrid simulation extends the reach of traditional virtual prototyping by arriving at more accurate results sooner, giving users the capability to design right the first time. Engineers quickly and cost-effectively compare various alternatives, pinpoint and correct problems early in development, and tune the design so the manufactured product delivers exactly the right performance.

Such product optimization leads to fewer but better hardware prototypes by ensuring that quality is designed into the product, thus shifting the balance of engineering activity from troubleshooting during prototype testing to analysis early in the conceptual stages. The aim is to provide innovative solutions that address the entire engineering process, not just point solution tools for performing mechanical simulation and data analysis tasks faster.

Benefits throughout product development

Hybrid simulation methods may be applied throughout the product development process to deliver significant benefits at each phase of design.

Early in the design cycle of a product, test results on similar components and systems from previous projects are extremely valuable in tuning and updating hybrid simulation models for improved accuracy and reliability. Maintaining this bank of test data is especially useful in modeling the effects of various connections such as bolts, washers, and gaskets. This validated database of components can be modified and reused for future programs and is extremely valuable in determining overall system characteristics of a company’s own products as well as those of the competition.

Using such a database of test results early in design makes hybrid simulation an especially effective tool in comparing alternative configurations and evaluating what-if scenarios. The approach also provides load data, including internal loading such as the vibration excitation an alternator bracket would be expected to withstand. Load data serves as experimental boundary conditions for detailed component design, with detailed models used to represent selected parts while the complex overall structure is “approximated” by test boundary conditions.

Later in the design, hardware components that will actually be used in the product gradually become available and can thus be tested, with this data used to refine hybrid simulation models. This improves simulation quality by more closely capturing the complex effects that determine damping, for example, and influence total response levels. Also, simulation models based on this actual test data are typically smaller and thus can be optimized and run much faster. In the final phases of product development, validated virtual models can complement experimental troubleshooting by evaluating different corrective measures. Of course, the whole purpose of hybrid simulation is to use early prediction to avoid these major issues in the final phases of product development. But, often the local phenomena which cannot be predicted early may be solved after more detailed analysis. In these cases, hybrid simulation is valuable in guiding the troubleshooting tests and directing engineers toward workable solutions.

Tools and technologies

Hybrid simulation is performed through the use of a variety of integrated tools and technologies, each developed to handle a particular set of tasks.

FEA is one of the most widely used methods for analyzing mechanical structures. The method determines characteristics such as stress or deformation in components. The structure is divided into manageable chunks (elements), each of which can be represented with relatively simple equations that the computer readily solves and combines to determine the behavior of the entire structure. The program then converts the analysis results into a graphical form for easier interpretation, usually showing problem areas in red.

Kinematics determines the action of moving mechanical parts, mostly those resembling linkages as well as rotating or sliding components. Software animates the action to ensure parts go through the required motion without bumping into each other. Moreover, properties such as displacement, force, velocity, and acceleration at various points are determined.

Multibody dynamic simulation analyzes displacements and forces in complex mechanical systems such as vehicles and industrial machines. Programs generate and solve the equations of motion for the entire system and are thus able to predict the overall behavior of the product. Using this type of program, engineers can thus determine how aircraft landing gear will perform when the tires impact the runway, or if a valvetrain’s motion will remain stable throughout a speed range.

Acoustics systems have capabilities for monitoring as well as predicting airborne sound levels, structuralborne noise, sound pressure levels, and fluid-structure coupling. Such tools are useful in predicting the sound level inside a cavity, estimating the sound field around the structure, or even calculating the structural response to an acoustic load. While products are still in development, engineers can predict, for example, the sound inside a passenger car or noise radiating from an industrial machine on a shop floor.

Vibration studies are closely associated with acoustics, with the two often referred to as vibro-acoustics and using FEA techniques in the studies. Vibration is typically analyzed not only as the source of sound and noise within and around the structure but also as a source of fatigue loading on components throughout the structure as they move and deflect. Transfer path analysis is often used to work back from the symptoms of noise and vibration problems to their sources.

Durability analysis predicts the fatigue life of systems based on test data as well as FEA results and loads from kinematic solutions. Software detects potential crack initiation locations and determines through stress and strain analysis how many loading cycles will transpire before parts are likely to fail. Loading histories usually can be imported from a variety of test data formats.