Authored by
Rendell Hughes
Vice President - Virtual Product Development
International TechneGroup Inc.
Milford, Ohio
transcendata.com

Edited by Leslie Gordon
Leslie.Gordon@penton.com

Related articles:
Neutral file formats fare differently in collaborative design
http://tinyurl.com/b3sma2

Different CAD systems don't alway talk well with each other. What lies under the hoo of data-transfer problems?
http://tinyurl.com/bsmg3s

It is tough to come up with a bulletproof engineering design. It can take a lot of time and many iterations to get all the details right. These iterations might seem small, but one minor change can unleash a cascade of minor changes in a product cycle. In other words, one event sets off a chain of consequences in a sort of “domino effect.”

For instance, consider the design of a new automobile that must get 40 mpg. First, an industrial designer creates the initial vehicle shape. A CFD analysis determines forces caused by air resistance. Drivetrain engineers then estimate the power needed to propel the automobile based on airflow resistance and estimated weight.

The last step shows a need for more room under the hood, so designers change the hood shape which, in turn, forces the driver seat’s height to rise for visibility. Now the cabin must get higher to provide clearance between the driver’s head and the ceiling. The new shape gets sent back to CFD which finds significantly more wind resistance, so the power-train team boosts engine power — and so on.

Some software developers have addressed this problem with applications such as Isight from Simulia, Siemens Knowledge Fusion, and Parametric Technology’s Behavioral Design suite. Leading product-development companies have also come up with their own methods. For instance, General Electric Aircraft Engines (GEAE) uses what are called “linked master models” involving about 300 components (out of the roughly 30,000 components making up a jet engine) considered critical because they drive cost, reliability, and performance.

GEAE followed rigorous procedures in selecting the components, identifying critical attributes, and defining how these attributes related to the CAD, CAM, and CAE models. These models varied widely in complexity, from simple spreadsheets and internally developed analysis programs, to sophisticated finite-element methods for CFD, stress, buckling, fatigue, and thermal studies. With the relationships established, GEAE users changing the value of an attribute in one model would cause the value of the same or related attributes across all models to also change.

Our company combined procedures like those created by GEAE and CAD, CAM, and CAE technologies to build the Linked Intelligent Master Model (LIMM) Web-based system. It provides a way for manufacturers to manage the domino effect with a database that links key product parameters to engineering models. The database also stores mathematical relationships between parameters and acceptable tolerance ranges for each parameter. Also included are adapters for various CAD, CAM, and CAE applications, homegrown analysis systems, and data sources.

The process of deploying the LIMM system helps companies identify critical components, determine primary engineering attributes, and establish relationships between products and models. This information helps engineers quickly evaluate design alternatives by changing parameter values and seeing the impact of their changes in the context of multiple analysis and design models.

In general, designing in the LIMM system involves changing values in models, updating models, running simulations, and evaluating simulation results. The design-to-spec of a crane for unloading cargo ships helps illustrate the system’s use.

First, users type the values from the customer specification into the LIMM system. The customer spec provides overall load capacities, crane dimensions, and operating conditions. A thorough structural analysis requires the evaluation of the crane in several different operating conditions, so users must analyze the crane model with eight different wind directions and three different wind speeds. In all, it’s necessary to analyze just over 300 load cases and crane configurations to ensure the structural design meets industry standards and customer requirements. This step also requires that engineers specify the values of critical attributes not included in the customer specification.

At this point, the data from the customer specification and the major component parameters are in the LIMM database. Users need only press the update button to generate the geometry for meshing.

Here, the LIMM system provides a template to adjust parameters at the next level of detail. For example: parameters for the boom can consist of material, plate thickness, and the like. Engineers can use the settings for these parameters from the baseline or another design, thereby minimizing or eliminating the need to type-in the values.

These actions ready the LIMM system for the deflection analysis of the crane. Users update the section and material-properties data for each beam in the analysis model and specify the load cases to run. The system generates geometry based on load case and crane configurations, generates mesh, applies loads and boundary conditions, executes the solver, and generates specific reports.

Engineers repeat the structural analysis as many times as necessary to attain the deflection targets while minimizing the use of material. (Raw material accounts for 70 to 80% of the product cost.) The LIMM system lets users input a value, press the update button, and then have the entire set of models regenerate and the simulation execute.

Based on these results, users adjust parameters at a lower level of detail in the product geometry model. For example, users can change the boom’s plate width, cross-section shape, and the like. Engineers can use the settings for these parameters that were previously typed in, thereby minimizing or eliminating the need to reinput them. Once detail parameters are available, users press the update button to generate the detail CAD models.

Next comes the selection of major subsystem components. Examples of subsystems include the trolley, control, braking, and lifting systems. To select the motor for the lifting system, for instance, users first type the productconfiguration data (which may have already come from the customer specification) into the motor-selection template. The template runs the motor performance analysis program to determine the motor requirements.

This action results in queries to a motors database to find all the motors that meet or exceed requirements. Should users find no match, they can contact vendors to find an appropriate motor and type its attributes into the database for future use and iterations. Once users pick a motor, the overall crane CAD model updates, with the motor selected to confirm fit and ensure the absence of interference issues. The LIMM system also updates, storing the motor’s mass and inertia properties for use in the vibration analysis.

In the previous step, users selected a lifting motor. Next comes determining the dollar amount for the motor to include in the overall crane cost. For this project, there were several potential sources for cost data: the ERP software for components previously purchased, the pricing database for vendor-quoted components, quotations from fax, mail, or e-mail, and the price the sales engineer believes the vendor will quote. The sales engineer can specify which value to use or input a newly negotiated value.

Finally, it’s necessary to sum costs and profits to determine the sales price for the crane. The LIMM system tracks material usage and component costs as well as waste, machining, assembly, and welding factors. It also tracks the cost for insurance, commissioning, and transporting the crane to the site for presentation to internal management as well as to support customer meetings and proposals.

As described, the entire procedure might sound like a typical development process. However, the LIMM system lets product-development managers review up to 20 times as many iterations as performed historically, as well as cut engineering time. For example, GEAE’s linked model environment halved the time needed to create a fan-disk design and slashed 90% from the time it took to design high-pressure compressor stages.