Machine Design

Does model-based engineering make sense?

The concept of “model-based engineering” (MBE) has generated a lot of buzz lately and perhaps rightly so. As you probably know, this approach tackles product development using a kind of digital “master model” (not necessarily CAD) from which all downstream activities can be derived to build the product. The approach replaces ambiguous documents and can eliminate the need for physical prototypes before a particular design has been chosen. Engineers can simulate and iterate as much as necessary to refine the model while meeting requirements and adhering to design constraints.

In this regards, it is helpful to distinguish between “design” and “engineering.” Design is merely creating the geometry. Any CAD jockey can do “design.” Engineering, on the other hand, uses physics-based rules to develop functional machines and mechanisms.

A few firsthand examples of MBE came from one of the many workshops conducted at the recent Congress of the Future of Engineering Software (COFES), held annually in Scottsdale, Ariz. Speakers there each gave an example of how they used the approach to build two entirely different systems. It seems that the master model can be quite different in scope and setup, depending on the nature of the problem. But it must usually allow for a certain fluidity. Why? Because product development itself is fluid. There is also the underlying assumption that it is necessary to logically model and simulate the entire system operating in its setting to understand the system’s behavior.

In one example, David Thomas, Sr. Project Leader, of The Aerospace Corp., Los Angeles, says model-based engineering only works when design models at the appropriate level of fidelity are integrated across engineering discipline boundaries. His example: A small, interdisciplinary team of engineers had higherfidelity models for mechanical CAD, structures, thermal, and optics. They combined these with lower fidelity, Excel spreadsheet models for a spacecraft bus and its associated components. The result was an initial design for an infrared telescope created in less than 400 hr. In this case, engineers generated initial optical, CAD, and structures models for the telescope in Comet Solutions MBE software. First-order constraints for the telescope were determined during an earlier predesign phase that described the orbit and the telescope.

Integration of the CAD and structures model in the MBE environment let the designer optimize for launch vibration loads in only 4 hr. The resulting CAD design for the telescope then became an input to a spreadsheet-based conceptual design for the mission. This design produced additional information (orbit details, solar array size, overall payload geometry) needed to complete a thermal design for the telescope. Designers completed their initial design of the infrared telescope payload in the MBE environment by integrating models for thermal (Thermal Desktop), structures (MSC Nastran), and optics (Code V). The resulting MBE let them evaluate changes to the telescope image quality during orbit.

Integration of engineering-design models within an MBE workspace helps diagnose design-performance problems as they arise. For example, initial analysis revealed bending of the primary mirror (optics) was degrading the telescope image. The structures model showed the culprit to be excessive clamping force at the mirror hub mount, due in part to low temperature. With this root cause determined, designers could change designs and repeat the integrated analysis much more quickly (by factors of 2× to 3×) compared to standard practice.

In another case, Matthew Loew, a new-product development chief engineer at Joy Global in Milwaukee, is using model-based engineering to develop mining equipment that weighs millions of pounds and stands several stories high. Again, the master model consists of multiple domains (structures, performance, reliability, cost, as well as product geometry) and mixed fidelity models.

For example, engineers initially investigate the structures models with closed-form calculations in a spreadsheet. They ultimately develop a finite-element model with fine shells and solids meshes, coarse meshes, and beams. Loew loosely coupled the models (unlike the telescope example) because, as he put it, “The design process in this case lacked a formal order.” The models range from those for nonlinear static FE, multibody dynamics, deterministic and stochastic reliability calculations, 1D performance simulations, and engineering-content CAD. Typically, models need more fidelity as systems mature. However, it is possible to actually degrade the fidelity in some areas to concentrate on areas of continued interest in development.

What do you think about the model-based design approach? Write us and we might print your answers here.

© 2012 Penton Media, Inc.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.