Vous êtes sur la page 1sur 5

education & training

How Modeling and Simulati "Business" to Their Skills

odeling and simulation (M&S) have been replacing rules of thumb for 20 years or more. M&S precision allows both tolerances and processes to be tightened and cycle times for all sorts of things to be shortened by eliminating trial-and-error / build-itand-break-it methods such as physical prototypes. For defense contractors like Lockheed Martin Co., modeling and simulation have always played a huge role in engineering modern weapons systems. Almost since the dawn of the Information Age, simulations helped determine the best ways to design and produce. New challenges will move M&S out of the closed world of engineering, where they came of age, and into the realm of "business" where the adding machine still trumps third-order equations. Thanks to the obvious benefits of M&S, engineering will be more deeply involved in so-called business decisions, and earlier. Large companies now refuse to undertake any significant capital investment without simulating the proposed facility, its processes and products. In the world of defense contracting, this means the project criteria generated by the systems engineers will be fed into simulations to highlight potential problems, solutions, to determine where to optimize or where to take risks. Remedies for these can be factored into responses to requests for proposals (RFPs) and bid packages.

This presents a great opportunity to M&Sand great challenges in justification and quantification. It is no longer enough to account for time spent analyzing things by saying, "We tried 16 different alternatives and after eight months came up with the best one". Instead, the justification should reflect the impact of M&S on key performance indicators such as cost, reliability (warranty costs) and time-to-market.

Management is likely to retort:


"Why did this take eight months? We needed the information in 60 days!" "Why did you need to look at 16? Back-ofthe-envelope or hand calculations might have eliminated most of them!" "It would have been great to know that information three months ago when we bid on the proposal!" In other words, the engineers bias toward thoroughness is being overshadowed by the high value business people place on timeliness. This is not to say that engineering is being rushed or dumbed down. But it does recognize the fact that the final 20% or 30% (or even higher) of effort in many engineering analyses"refining the data"often doesnt change the outcome. Observing this trend for many years, M&S software vendors have made their systems easier and faster to use even as they became more powerful. This makes the tools accessible to more engineersand nonengineers. Product designers and manufacturing engineers now routinely use tools that not long ago required a Ph. D. in engineering or computer science or both. M&S software vendors have also developed powerful tools such as morphing meshers and parametric preprocessors. Among the results: In animated CAD-like simulationsclearanceand-collision analyses, often called Digital Manufacturingthe time needed to import objects and build dimensionally accurate scenarios has plummeted 90% or more, according to most vendors and users whom coauthor Thornton has interviewed. In finite element modeling and analysis (FEM/FEA), there has been a similar drop in the time needed to generate the elements that represent physical properties of objects and

Maritime Systems and Sensors of Eagan, MN using Star-CD

"...the engineers' bias toward thoroughness is being overshadowed by the high value business people place on timeliness."
Under todays fixed-price contracts, with the demise of the Soviet Union, and budget deficits, the Good Old Days of cost-plus contracts are mostly gone. In a complete reversal, letting any cost miss their budgets can be painful in terms of workforce incentives missed and management bonuses foregone, not to mention profit-margin squeezes and declining share prices and follow-on contracts/work.

July 2005

education & training

ion Engineers Can Add


Mark Westphal, P .E., Lockheed Martin Co. Engineering Process Improvement Center, Cherry Hill, N.J., USA Jack Thornton, Principal, The ParadigmTilt Organization, Santa Fe, N.M., USA
Using Tecplot 10 for the figures and DAC 97 for the DSMC simulations (from MAGallis-SandiaLabs)

meshes that replicate how objects behave under stress or strain (to include nonlinear analysis). And for both M&S the amount of detail that can be accommodated in models has been multiplied by the surge in computer power and speed. In short, the tools and the computer resources are in hand or readily available and inexpensive. Its time now to change the way management looks at M&S, which means it is time for the people in M&S to change the way they see themselves. Analysts have to seeand projectthemselves as part of the overall business rather than just a part of engineering.

functionality for the most sophisticated users at the expense of tools needed by the broad middle of the engineering profession. The education of most mechanical engineers. All too often, modeling and simulation are not required coursework so developing those skills is left to employers and vendors. The ad-hoc, late-in-the-game approach of many companies. They wait too long to start analyses and simulations. In short, a more pragmatic and comprehensive "businesslike"approach is required from all concerned. Modeling and simulation allow companies to make much smarter bids and find ways to cut costs in major ways: By finding better methods sooner. By pinpointing discoveries that can be reasonably anticipated. By avoiding unrealistic promises in performance or delivery time. By eliminating opportunities for error. By understanding that actual, physical testing is the most expensive way to verify requirements of the contract. A good example is in engineered composites, where product and process are much more tightly intertwined than in older manufacturing processes. In traditional cutting and forming, where processes are mature and well understood, only the worst practices really harm the product and the best new practices make only marginal improvements.

"It's time now to change the way management looks at Modeling & Simulation"
M&S needs to be a part of the design and development process from the beginning, at the RFP stage, even before bidding. To the extent this is done, the early analyses can greatly improve the terms and scope of a bid and the business environment of the contract, and the potential payoffs are huge. Marketing and business units understand this, of course, but it has not yet fully penetrated into the analysis and design "trenches" of many engineering organizations. There are other challenges: The organization of most M&S software companies. They still prefer to develop

July 2005

education & training

In composites, however, processing (fiber, resin and lay-up) remains less than fully developed and anything short of best or fully understood practices can significantly reduce physical properties and ultimate performance. Small changes in process can have huge effects on the properties of the final composite product. And since composites processes are evolving rapidly, they merit continual analysis. At the same time, the uses of composites are booming. They are in everything from radar-eluding Stealth aircraft to body armor to automobiles and other commercial products.

Analysts understand the criteria by which M&S results are evaluated. Effort is placed where the biggest time and cost savings are. That proper training is available and that experience required for analysis and simulation is applied. M&S should be integral to each engineering teams push to make better products in better ways. As an example, using a minimum of "make-and-break" prototypes to address all key design and analysis issues at some time in a new product cycle.

"Analysts must accept that calculations of their costs vs. cost avoided will determine the trajectory of their careers just as much as the accuracy and effectiveness of their analyses."
The authors believe that the real opportunity in M&S is using them as early and as logically as possible. But that can be achieved only if some major cultural changes take place among users, corporate managers, educators and developers

Cultural Change No. 2:


Managements must, in turn, accept and value the strengths of the various M&S approaches, and understand the remaining shortcomings. M&S cannot predict or see everything. Analyses and simulations are only as good as the power of the software, the skills of the users, and the realism of the assumptions from the systems engineers or the marketing people. We feel that analysts should be regarded as any other corporate resource, neither more nor less important. This means M&S should plug away at the same problems right alongside the designers and manufacturing engineers. M&S should be encouraged to identify potential problems and not be the last minute check-off or last minute fix solution. As part of this, management and M&S together should consider ways to ensure that the contributions of M&S are as visible as the contributions of every other sector of engineering. This can work only if managements insist that analysts be fully aware of all the requirements developed by systems engineering. Analysts will then know what is expected of them and will have the right elements, meshes, solvers and simulation engines. In certain ways, management itself must become more businesslike in its approach to M&S. M&S needs more thoroughly defined criteria as part of an integrated engineering process (IEP), perhaps in terms of minimal margins of error or when stresses and strains seem real but unquantifiable. This is where M&S criteria are refined and developed, and where trend lines can be studied to predict where M&S will be needed in the future. This should encompass the level of modeling effort, from penciland-paper calculations to probabilistic design with its dozens or hundreds of values for one or more variables. The mindset that "the computer is always right" must be challengedin the ranks of the users as well as management. The computer is right only when the

Cultural Change No. 1:


The authors feel that engineers who do simulation and finite-element analyses must become more businesslike in their dealings with management. Conceptually, their services are like any other product: it has costs, customers, values and benefits. These factors must be clearly understood. Analysts must stop advancing their cause with techno gee-whiz; M&S technologies/concepts are as much as 30 years old. Instead analysts must drill down to the real value of their services and not be diverted by what is "cool." This means analysts should quantify their contributions in readily understandable terms as gains in quality, speed, safety, durability, cost etc. They should also standardize their reporting methods/formats and the methods by which their models are correlated. Analysts also must develop an appreciation for the dollars-and-cents issues of M&S. They must learn to speak and think in the language of the "suits," i.e. how to quantify benefits for a given level of effort and time expended. Analysts must accept that calculations of their costs vs. cost avoided will determine the trajectory of their careers just as much as the accuracy and effectiveness of their analyses. On the other hand, M&S should never be a speed bump or a check-off on a to-do list. Managers can help by making sure that:

July 2005

education & training

Using Tecplot 10 for the figures and DAC 97 for the DSMC simulations (from MAGallis-SandiaLabs)

problem / model fully represents reality in the assumptions made in the model. This means management must make sure analysts know what they are doing. Engineers may know the functionality of M&S software quite well but may not understand the underlying theories (sometimes referred to as "buttonology"). Without real-world testing experience, they may lack the "horse sense" to put the proper and realistic boundaries around a problem. After three decades of demonstrations, managements should be willing to accept that virtual testing of digital designs and prototyping is superior to make-and-break. Breaking digits is vastly cheaper and faster than breaking metal or tearing apart composites. Of course, some make-and-break will always be needed to validate computer models. There is another and equally important value to digital make-or-break. Physical make-and-break can never completely reveal what happened just before a part failed. And that is what designers most need to know. One can, of course, get a really good idea from make-and-break and all failures leaves telltale signs. But an FEA model can do that very clearly, if the part breaks as the analysis said it should, without a lot of interpretation. It can be slowed way down or even run in reverse. However, when a part does not break, it may or may not validate the analysis model. Then there is a high level of interpretation of what has been really correlated.

This is especially important in FEA with its elements, meshes and solvers. This should include mastery of the basics behind the software geometry, boundary conditions, material properties, load paths, valid/proper assumptions, etc. Engineering school faculties must also grasp the implications (positive and negative) of FEA and simulation migrating "downward" and "outward" in engineering organizations, and not just at aerospace companies either. FEA can no longer be regarded (or dismissed) as some sort of computer-graphical subset of statistics that only "big and high tech" companies use. They are very common at all levels of product technology.

"After three decades of demonstrations, managements should be willing to accept that virtual testing of digital designs and prototyping is superior to make-and-break."
Analytical tools built into the designers solid modeling packages, initially simplistic at best, get better with every software release. All of the big mechanical computer-aided design (MCAD) software providers own or have tight relationships with one or more FEA software vendors. And all the FEA and simulation packages can and do use MCAD models for pre- and post-processing. Both are huge steps in shrinking time-of-use and ease-of-use. Since many designers may know very little about how FEA or simulations actually work, the "master analyst" concept is very important. Working together, engineering schools and user corporations might take on the training of these power users. They could help generate the parameters within which designers could do their own basic analysis and track their efforts. This could go a long away in reducing wait time in design and analysis.*

Cultural Change No. 3:


Given that M&S are going to loom ever larger in engineering and business decisions, we feel that engineering schools must do a better job of teaching tomorrows engineers. Both user companies and software vendors must demand M&S additions to engineering curriculums-as a requirement for graduation and not just the rare elective. They should start with a solid understanding of the underlying theories.

July 2005

education & training

Similarly, team oriented design/analysis projects should be strongly encouraged. That is how nearly all engineering graduates will apply their skills in the working world. Design projects must include analysis. Most engineering faculties must do a better job of preparing students in the subjects of Geometric Dimensional & Tolerancing (GD&T); the ASME Y14.5 standards; the implications of 3D model centric design; producibility analysis / design for manufacturing (DFM); machine shop/CNC experience; cost analysis; knowledge of basic Microsoft Corp. Office products; and writing skills with proper grammar. This material may seem boring or not "real engineering" but it is essential if good designs are to be created, explained, and then produced on time and within budget.

suddenly, in retrospect, seems to have arrived overnight. And just as suddenly the old rules of thumb seem so inadequate. At one time, a long time ago, designers, manufacturing engineers and analysts worked in groups over drawing boards, arguing and discussing the design and each others intent. Each knew what was important to the other person and for the product. As CAD and analytical software developed along separate paths, this close cooperation was lost. The invention of functional groups has had an impact, too. Today, with much more complicated designs, the same type of interactions, backed up with massive amounts of data, will take place. Superior and faster design cycles will result. The once close cooperation that had been lost to technology can be restored by M&S. We feel that industry must plan for these developments or they will occur haphazardly, as the various software vendors push their own interests and directions. If thats allowed to happen, the majority of users probably will benefit only in haphazard ways. The MBAs and accountants long ago established primacy over the engineers. Engineers may be loath to admit it, but their work has become more careful and more comprehensive as a result. Inspiration, creativity and insight are still prized. But they are checked out before any checks are written. They must make hard financial sense. Great business plans are now just as highly prized as great aircraft, amphibious and assault ships. After all, without great business plans, and with the days of cost-plus gone, the funding for these weapons systems might never be appropriated. And really great business plans require M&S. BM

Cultural Change No. 4:


The authors feel that software vendors could do a better job of supporting users with business metrics in their software and tools for presentations to the upper echelons of customer organizations. The problem with this is no vendor or user will ever provide what could be seen as a "bad" metric. Only the most altruistic of organizations will disclose something that failed or doesnt work as well as expected or as the competition, no matter how valuable the lesson learned. An impartial organization like NAFEMS might be able to address this despite data sensitivity and political challenges. It would also help if analysis data were available in easily shared, traditional neutral formats such as STEP or IGES files used in the MCAD world. Some work on this has been done in analysis neutral data files like STEP (ISO 10303) AP 209:2001.

Conclusion
Computer modeling constitutes a revolution in design engineering as well as in production engineering. But its biggest benefits probably wont be limited to engineering, and certainly shouldnt be. Instead they lie near the realm of Microsoft PowerPointat the interfaces between engineering and the rest of the enterprise including such nonengineering functions as marketing and business development. In addition to all the engineering benefits spelled out above, both are tremendous aids to comprehension. Revolutions are not completed, however, without accompanying cultural changes. These run the gamut from simple acceptance of new opportunities to complete upheavals in methods. New technologies dictate new ways of looking at the world, at daily tasks, and at the way work is done. Real change comes slowly but then

Contact
Jack Thornton, The ParadigmTilt Organization jackt@paradigmtilt.com

Notes
* SEE "The Future of Mechanical Analysis" by Mark Westphal, BenchMark magazine, NAFEMS, April 2004.

10

July 2005

Vous aimerez peut-être aussi