The future of mechanical CAD
High-end systems will continue to exceed the capabilities of mid-range modelers, and new elements of CAD technology will make design engineers more visible (and respected) within their organizations
John R. Baker, P.E., Product Manager, Unigraphics Solutions Inc.
When it comes to addressing the problems facing designers and engineers, CAD vendors are shooting at a moving target. For a long time, the problem was simply to define the shape of a part digitally. CAD developers have achieved that goal, to the point where today's mid-range CAD programs can model all but the most complex parts. In fact, geometric authoring tools are approaching the level of a commodity product.
The current challenge that CAD must address is to change the nature of product design from "reactive" to "predictive." It is reactive now because a designer typically has only a fraction of the information he needs when he starts a job. For example, he may know that certain components of a product will be milled, but he probably doesn't know which machine will be used or what the travel limits of the tool are. These factors might affect the design. But the designer doesn't find this out until the design has been sent on to manufacturing. They alert him to a problem regarding tool travel and then he reacts, redesigning the part as necessary.
The CAD system of the future must allow designers to incorporate manufacturing requirements and evaluate product performance early in the design process, when they can make those changes inexpensively. This will shift the design engineer's role from one of reacting to that of predicting, or meeting in advance all of the requirements for the product. This is not something a commodity geometry-centric tool can do. A predictive engineering environment requires a high-end CAD system supported by a wide range of integrated, advanced knowledge-capture technologies. In fact, this is how high-end CAD will continue to differentiate itself from mid-range tools. High-end systems will address the broader challenges facing product designers with the ability to capture an evaluatable product description, complete with associated processes, that accommodates influences from all aspects of the extended enterprise.
The importance of knowledge. Knowledge-enabled CAD, the foundation of the predictive engineering environment, represents the next big step in the evolution of mechanical CAD technology. To put it in perspective, let's consider two other shifts of this magnitude. One was the change from drawing-based technology to feature-based digital models.
A more recent shift, taking place in the last five years, has been the emphasis on process-based solutions. This came about as the CAD vendors developed a better understanding of their customers' needs. Unigraphics Solutions, for example, created an industry-specific approach for CAD/CAM/CAE implementations called "process threads." A process thread supplies all the tools necessary to move an idea from concept to finished product with no data translation at any step of the way. We picked the term process threads because this approach "threads" the required functionality through the organization, from start to finish, much the way a strand of thread is woven through a piece of cloth.
The next major shift in high-end mechanical CAD technology will be characterized by capabilities related to knowledge-based engineering, for which we have identified three primary enablers:
Process wizards--applications that combine best practices in engineering with proven technologies to produce applications that are directly applicable to specific process threads. One of the main benefits of process wizards is that they make advanced technologies (such as WAVE, Master Model, etc.) available to the user in such a way that he does not need to have expertise in these technologies. Process wizards make high-end CAD easier to deploy so that a high degree of productivity gains comes quickly.
Optimization--critical to front-loading the design process with product performance information. By incorporating computer-aided engineering results as design criteria, better decisions can be made regarding the finished digital product.
Rule-based modeling is key to incorporating a broader set of design criteria into the product design cycle. For example, abstract criteria (such as government regulations or cost criteria) can be incorporated right alongside traditional numerical constraints. Today's knowledge-based engineering tools function in this manner. They interpret rules to produce numerical data for the design process. In a predictive engineering environment, these rules will exist as an integral part of the digital model.
What additional capabilities must a CAD system encompass to support a predictive engineering environment? One, called language encoding, refers to the ability to impose a broad set of design criteria on the digital product. This is critical for allowing rule-based modeling. Another necessary capability is process integration, or the ability to interact with external sources of information, such as spreadsheets, existing knowledge-based applications, and customer-developed programs.
A third requirement is the ability to perform automated, iterative analysis. We are not advocating the elimination of specialized analysts with all simulation being done at the designer's desktop. Rather, we see the use of intelligent "design advisors," applications that incorporate the knowledge of specialists in such a way that it guides design engineers as they create products. When a designer modifies a wall thickness, for example, a design advisor may inform him that this will affect the structural integrity of the part and that he should verify that this will not compromise the quality or performance of the product. The design advisor might offer to perform an immediate structural analysis of the part. Finally, a visual development environment is needed to serve as an interface to these capabilities and make them accessible to the user in a graphical fashion. The knowledge-based technologies mentioned above are available today, but are only usable by between 3 and 5% of the mechanical CAD com- munity. By providing access to this technology in a graphical environment, we expect that at least 80% of our customers will be able to take advantage of this technology.
The importance of design engineers. By broadening the set of design criteria to include non-geometric data (government regulations, costing information, rules of thumb), engineers can extend the value of digital products beyond the design engineering department to the entire enterprise. Currently, only 5 to 10% of people in an organization contribute to the creation of CAD models or use them as input for their jobs. If the other 90 to 95% could both contribute to and benefit from CAD data, think how much more valuable it would be. All departments would have the ability to provide early input to product design, setting the stage for true collaborative product development.
And all departments would benefit directly from the work of the design engineer. That is nothing new, of course. An organization always benefits from the work of the design engineers. Without them, how would new products get developed? But many people in the typical company today aren't really sure what design engineers do. Ironically, that has been caused in part by the replacement of drawings with digital design data. Drawings at least provided a visible end product of the design engineer's work.
In a predictive engineering environment, design engineers have the greatest understanding of all aspects of product develop- ment, from structural integrity to market appeal. And engineering models not only drive design and manufacturing, but are also of value to other fundamental departments, including sales, marketing, and technical support. When everyone is contributing to design projects and using the fruits of the design engineers' labors--i.e. a true collaborative product development environment the design engineers' contribution to the organization will be more visible. As a result, the design engineers--especially the ones who best understand the entire process of bringing a product to market--become more valuable to the enterprise. While this may not be the primary focus of our efforts as developers of tomorrow's mechanical CAD solutions, perhaps it may be a pleasant by-product of those efforts.
The future of CAE
First, there was 2D, then 3D. But simulation, including finite element analysis and other tools,is 4D
Ken Blakely, Sr. Vice President, MSC.Software Corp.
Simulation is the act of prediction. In mechanical engineering, simulation comprises FEA (finite element analysis), CFD (computational fluid dynamics), kinematics, and manufacturing process simulation (forging, mold flow, etc.). In the broader area of product development, simulation is performed on a broader scale to determine how design and process tradeoffs affect development cost, time to market, and customer acceptance.
Simulation is really the next step in the design and manufacturing of products. We have seen the software tools move from 2D to 3D; simulation is 4D, adding the time dimension. Most simulation today occurs after the design is set, and is used to verify that the design will perform as intended without violating design constraints. This verification process is done using MCAE (mechanical computer-aided engineering) and testing, and often occurs too late in the design-manufacturing process to have any real impact.
Today, a product's form (i.e geometry) often dictates the overall design. However, geometry is only one of three high-level parameters available to the mechanical designer. How the 3D geometrical model looks, what the material consists of, and how the manufacturing process is assembled are the "global design variables" the engineer needs.
Tomorrow's design process will be different. In that new process, a product's function will dictate its form, materials, and manufacturing processes. First, the design attributes--weight, stress, deflection, reliability, safety, noise, comfort, manufacturability, affordability, time to market, profit, and customer acceptance--will be specified. Then the design's form, materials, and manufacturing processes will be synthesized. Simulation will play a key role in creating this new paradigm, and will need to be done after a product is conceptualized but before it is detailed. In fact, the detailing--done as a traditional CAD function--will be an output of the simulation, turning concept into reality.
In order to achieve this, the following changes will need to take place in simulation technology, user environments, and com- puting infrastructure:
Wider range of simulation. For function to drive form, simulation needs to be much broader than it is today. The designer needs to take into account stress and vibration, kinematics, fluids, assembly modeling, reliability analysis, human factors, manufacturing, and costing. The entire life cycle of the product, from manufacturing to shipping to operation to disposal plays a vital part in this process. Engineers must be able to simulate all functions, with the resulting design synthesized form the results.
Robust design. Today's simulation assumes that both design and operational environment can be defined in terms of known parameters. However, in reality this is not the case: The "as built" product has variations and differs from the "as designed" product. Tolerances and variations exist in geometry, shape, and manufacturing processes. Engineers don't typically know the operating environment. In today's simulation, we assume that all inputs are known and we compute the response, optimizing our design to maximize the desirable characteristics at one design point. Robust design assumes that we don't fully know the inputs, then finds the best design by taking the uncertainty and variation into account. This leads to a design that can operate over a wide range, making the design more reliable and safe.
Process and task automation. In a broad sense, simulation today can take into account many of the attributes listed in item number one, and does so via separate applications. This means that multiple people are involved, each being an expert in a particular application. This methodology will not suffice in the future, where the trend is away from specialists and towards generalists. Tomorrow, we will see an increased use of user environments that handle all of the design attributes at once, automating the entire conceptual design/simulation/detailed design process. This will most likely be accomplished by creating custom systems that are specific to a company's way of doing design, leading to tremendous productivity gains over today's methods. Applying automation on a smaller scale--to tasks, as opposed to the whole process--will lead in the future to many similar benefits. Process and task automation can be product, industry, or company specific.
Re-use of previous models. In addition to process and task automation, another way to be more productive is to make use of existing simulation models. Many designs are made today by modifying an existing design so why not perform simulation in the same manner? Engineers can do this by storing existing models and reusing them in the context of a new or modified design. Models can be finite element models, geometry, spreadsheet-type conceptual models, or any type of model used in the simulation. Full use of parametrics--including geometry, materials, manufacturing processes, and the operating environment--will facilitate context-based reuse.
Interoperability with enterprise systems. Much simulation today is most often applied in order to verify a design and is separate from the overall product development process. To be better integrated into the process, simulation should come earlier to drive the design, and it should integrate with the company's enterprise systems, such as ERP (Enterprise Resource Planning). Today, simulation is somewhat integrated with other engineering software such as CAD, PDM, and test, although the level of integration can certainly improve. We see increased acceptance of data exchange standards to facilitate this, such as the STEP AP209 exchange standard, which is for finite element data.
The Web. The Web is transforming the way we access information, share data, and make transactions, and will also transform the way we perform simulation. An increasing amount of simulation software will be demonstrated, delivered, and supported via the web. "Pay for use" will also become more prevalent than it is today. It is also likely that an ASP (Application Service Provider) model will result, with large number-crunching software on servers and user environments on the client side, changing the way we utilize simulation software. An example of this paradigm today may be seen on the Engineering-e.com website. The DataMart uses MSC.Mvision technology to provide OnDemand access to high quality materials data. This framework that illustrates how we believe engineers will utilize the web for their tools of simulation.
With these changes, simulation will become the key component in mechanical design, turning concept into reality.