Industry Outlook
Systems Engineering - Tomorrow's CAD?
By Raymond Kurland, President, TechniCom, Inc.
Introduction
Let's face it; CAD is used more for documenting and detailing the design than developing it. Today's CAD systems are quite facile at making geometric changes driven by parameters. This inherent "bottoms up" approach can lead many large projects to become mired in details, causing CAD vendors to focus most of their resources on making the detailed design process more efficient.
Instead we propose that both users and vendors consider changing their emphasis to the more top down approach of Systems Engineering. That is, approaching the entire design by considering all product aspects, from beginning to end of the product lifecycle. In the last year we have seen small changes in the industry leading in this direction. We believe that it is a natural evolution of CAD systems.
What is Systems Engineering and how does it differ from CAD? Unfortunately, like PLM it is a broad topic and defies a neat definition. Systems Engineering starts by "working with" overall system requirements. For example, in the case of a hair dryer, the price target, the quantity to be produced, performance characteristics (such as CFM, ambient temperature, output temperature, etc.), exterior shape design, input voltage, and so on. Now a top level design takes place, breaking the design into subsystems that can be assigned as sub-projects. Critical to the subsystem parsing is an understanding of how these subsystems interrelate. Even in this simplistic case we have complex inter-relationships. Consider that there are mechanical space reservations, electrical wire routing and an electrical control system, mounting information, and thermal and CFD relationships between the housing and other components. Ideally, we would like to mathematically describe these inter-relationships, making them more readily available for system simulation.
Now the design engineering can take place, taking into account the requirements, existing designs and available parts. During the design process, and as the implementation evolves, systems engineering techniques "oversees" the process. As the design becomes more and more detailed, the systems engineering plan should be continually compared to actual system performance. Mixed simulations should continue to occur, with projected product performance substituting for the actual results, as needed. Such early involvement comparing actual versus projected performance will truly allow early error detection in the design process or show where designs are going astray from their original intent. Naturally there will need to be continuing comparisons of actual results versus the original requirements.
Image an even bigger step. Functional requirements are coupled with designs; changing a functional requirement changes the design.
Unfortunately no such integrated design system exists today. Instead we have bits and pieces of software which can be pieced together.
Today's technology limits
The mechanical design systems of today excel at the early stages of automating design changes. While small changes are correctly evaluated (and by small I mean changes that do not dramatically effect the topology), large changes often cannot be solved by pure geometric systems. In fact, large changes often require an understanding of the process in order to select the correct design.
Let's consider how even a simple assembly using today's technology resists automated changes. Assemblies typically connect using only geometric mating constraints. If these mating constraints change, how is the corresponding sub-system to react? Suppose two co-linear shafts connect via a coupling. If one shaft increases in diameter, how is the coupling or shaft supposed to change? Since no knowledge of the behavior of the coupling is known, a human designer is required to intervene. When should shaft sizes change because of greater torque? What should they change to? Should the coupling method change? To what - a clevis pin or a spline connection? And so on.
Solving interdependencies by embedding each connection with knowledge rules could solve the problem. BUT such rules only work for predetermined changes! And a rule would have to be written for every connection. Clearly an impossibility. On the other hand, knowing the characteristics of the coupling might enable the system to compute such a change. Yet mechanical systems have no way to describe what happens at each interface. Several problems exist in making such a concept operable.
1. No language exists to clearly define what subsystems do. Even in the simple case of a bolt, we know it has thickness, thread pitch and depth, and material type. But what does it do? How does its performance change as its parameters and characteristics change?
2. Beside geometry, changes include operational characteristics, geometric sizing and tolerances, external stresses and reactions, motion behavior, and on and on. These changes are quite complex to describe, and in fact, often impossible to describe with a regular language. Many systems can only be described using non-linear charts, such as RPM vs. power output.
Needs
What is needed is a system that offers an integrated solution starting from the beginning of the design process and working toward a capability that offers unique tools to help customers solve these problems. Systems Engineering concepts will work more viably when tied directly into the CAD tool. Requirements capture and checking for product compliance starts by making specific product requirements visible and offers a way to interpret these requirements so that they can directly drive geometry. Extending, then, directly into top down design by controlling key assembly constraints allows building and investigating overall designs early.
If detailed designs allow their parts and assemblies to be "knowledgeable" then such top down designs can be moved rapidly to detailed design.
Product simulation allows the operational characteristics of the system not only to be checked against specified performance, but should allow careful evaluation of design alternatives, without the tedious process of detailed product redesign. Thus, this closed loop design should prove if the design meets the requirements!
Customer benefits
We were able to find a few customers that have begun to implement some portions of systems engineering, notably by driving parts and assemblies from knowledge based rules. Such customers have only begun to realize the results of systems engineered design. Some of their benefits included: faster development processes, improved handling of design changes, shorter lead times for custom orders, more accurate bids, more design iterations within the allotted development time resulting in better product consistency, and lower manufacturing costs.
Summary
Systems Engineering addresses the effective design of reliable systems within cost and time constraints. Companies that employ Systems Engineering techniques are the ones that will best be able to understand the effect of and reaction to changes anywhere in the product lifecycle.
The next challenge facing manufacturing companies is to progress from cost cutting to increasing revenue and margin through innovative product development. Often such innovative designs require rethinking the product delivery processes. Simulating the essence of these processes allows product innovations to be proven before deployment.
In the past we have been unable to wed the tools necessary to perform such a system approach. Only today have we begun to see the ability to use this approach for mechanical engineering. Today these techniques are widely used for software. Unfortunately we are in the infancy of such systems for mechanical engineering.
We expect that CAD/CAM/CAE vendors will be in the forefront of delivering such integrated software. The technology being delivered piecemeal today by these vendors include requirements capture and management, front end system simulation to validate and optimize design decisions, design centric system based modeling, knowledge driven capture and automation, integrated digital simulation, and the ability to manage and control not only critical product data but the variables that drive such a system.
Customers need to make the decision to change their business practices to accommodate a systems engineering, product lifecycle planning approach. The time is now for such decisions!
Keep in mind that systems engineering is likely happening within your company for many of your current products. However, without a specific systems engineering organization, it is likely an informal process with mixed or unrecorded results. We recommend that management consider how design trade-offs are made and by whom. A formal Systems Engineering process should greatly benefit the visibility of how key product decisions are made. Any decision to evaluate these processes can greatly benefit from independent advice.
When you are ready to proceed on an implementation, carefully consider a stepped approach that delivers value during each phase.
About the Author
Raymond Kurland is President of TechniCom, Inc., a market research and analysis organization that specializes in understanding, consulting, and writing about Mechanical Engineering product development software. TechniCom produces a monthly newsletter, specialized reports, and offers a continuing research program for software vendors. Ray also speaks at conferences on the subject and frequently consults with users considering embarking on re-evaluating their product development systems. He can be reached via email at rayk@technicom.com.
|
