Abstract

Developers and manufacturers of complex systems are finding themselves in more and more competitive situations which force faster and lower cost product life cycles. More powerful computers and advanced software are helping to significantly shorten time to market with improved products. The key to achieving these improvements is a methodology called Total Product Optimization. Using state-of-the-art software and computers, major industrial organizations can automate their design process, optimize their integrated designs through rapid product modeling, performing "what-if" studies, simulating and diagnosing failures and, very importantly, capturing a knowledge base (often called a "Virtual Engineering Model") for the product that can be used throughout the manufacturing, testing, operations and field maintenance parts of the life cycle. This paper describes why this is important to technical businesses today and how it can be achieved with a low investment.


DEFINITION OF TPO

Many companies today have a segregated approach to designing new products or making major upgrades to existing products. Technical disciplines tend be grouped together into "silos". They perform their portions of the product design with relatively little interfacing with other technologies. There are plenty of meetings and phone calls but the actual design work is done with processes and tools that become unique to the discipline. Some companies are beginning to take the first major step toward integrating disciplines into "Integrated Product Teams" or IPT's. These teams work together in a common work area, report to a common boss, charge to a common budget and have one common focus. NASA, along with other advanced technical organizations, have been leaders in encouraging this type of culture change. Figure 1 indicates how this is done.

After the IPT concept has been implemented in some form, the next logical step is to produce a complete model of the product such that the design of the complete product can be developed and optimized as a total entity. This Total Product Optimization (TPO) concept can yield huge advantages, especially when cost, time-to-market and other competitive considerations are important.


HOW TO INITIATE A TPO ENVIRONMENT

The effort to create an integrated model of a product, especially when it is a large system like an automobile, airplane, missle, assembly line or refinery, can be truly awesome. It is seldom practical to just drop existing discipline models and start over with a new complete model.

An incremental phase-in of a TPO environment is usually necessary. The early selection of a capable total simulation tool and a computer system is required. The tool must have all the key features which will allow it to ultimately be the central simulation tool. Because of the need for the incremental phase-in, it must interface, either directly or indirectly, with common discipline-specific models which already exist. Some of these models may later be incorporated into the central tool and others, due to their uniqueness, may always remain separate but integrated from control and data passing points of view. Other very important central tool characteristics are;


· a friendly and flexible user interface, point and click capability, natural language dialog

· the ability to generate multidiscipline models; mechanical, electrical, thermal, chemical, etc.

· predefined or easily adapted interfaces to existing models

· the ability to perform trade studies and what-if's and to quickly show product performance under varying conditions

· the ability to insert simulated failures and to show system performance under these conditions
· the ability to accumulate "design knowledge" about the product such that it can be applied to testing, product operation, field maintenance or later improvement

· capability to become a real-time operation and diagnosis tool

· a highly developed and proven hardware interface

Once a tool is selected and users have become accustomed to it, organizations can begin to reap the benefits of optimized products and shortened development cycles.


THE INFLUENCE OF COST ON TPO

It is rare in industry today that the cost of producing and maintaining a product is considered early in the design process. It is even more rare that the consumer's cost-of-ownership is considered. Airlines are particularly sensitive to ownership costs because having to replace expensive parts or performing constant servicing is costly and keeps the aircraft out of service. When design choices are made for various parts, those with low maintenance characteristics can be traded against initial cost of the part. When design decisions are made with these costs being part of the optimization process, much greater customer satisfaction and market share can result.

Oftentimes, particularly in Government acquisitions, complex products must be designed to cost no more than a certain amount. The functionality of the product therefore is limited by what the buyer can afford. Cost modeling becomes very important in this case because individual functional capabilities can become part of the cost decision.


Some companies today have separate cost models which can be integrated with a central tool to help with TPO. If not, development and addition of a cost model, either as a stand-alone model or as part of the central tool, can be well worth the expense. Cost can become an integral part of a company's knowledge base.



HIGH VALUE OF KNOWLEDGE BASES

Finding efficient ways to capture knowledge automatically in a computer has eluded engineers and computer scientists for many years. It has been clear that if technical knowledge could be stored in a computer in such a way that it could be used for product improvement, testing, maintenance, fault diagnosis, failure detection, product operation and in many other ways, a great deal of money could be saved by eliminating the re-development of information bases by each organization that a product passes through. It also would reduce the reliance upon "special" people who always seem to be the only ones who have all the design knowledge of a product. Figure 2 shows a typical product life cycle over which a single knowledge base can be very effective.


USE OF KNOWLEDGE BASES IN THE FIELD

Next to the total system design optimization, use of a common product knowledge base is an area of high leverage cost savings. In many industries, maintenance contracts with customers are common. These generally are fixed price so the speed with which diagnoses are made, parts replaced (the correct part the first time!) and the product returned to service translates into higher profits and more satisfied customers. The automotive industry is on the leading edge of capturing design knowledge only once for a product and transferring it to dealer shops for direct use in trouble diagnosis and repair.


SELECTION OF THE CENTRAL TOOL

A relatively new design, optimization and diagnosis tool is available on the market today. It is called RODON, (the Greek word for Rose). It is already demonstrating its amazing capabilities in many complex technical and medical applications in Europe and the United States. It is produced by an engineering and research company in Germany named R.O.S.E. Informatik. This product takes advantage of state-of-the-art modeling techniques, friendly user interfaces and fast computer workstations.


RODON, THE TOOL OF CHOICE

The functional analysis tool "RODON" is a model based simulation, monitoring, analysis and diagnosis tool that integrates engineering methods with Artificial Intelligence technology. All these functions can be performed using the same knowledge base or Virtual Engineering Model. The modeling approach is fashioned after a powerful method many engineers apply to diagnose a system they are developing where initially they only know its nominal functionality and the topology, i.e., the components it consists of and their connections. In other words, they know the configuration and the intended behavior, but do not have any experience with failures and their ultimate effects. At first, they build a mental model of the system. Then they identify how the hypothesized failure symptoms match with the nominal and off-nominal behavior of components in the functional chain or net i.e. they propagate the symptoms backward through the system until they find a functional unit which seems to explain the symptoms. They then identify the failure mode of the component after which they anticipate (i.e. simulate) how this matches with the full set of observations in order to either confirm or invalidate the hypothesis. Insertion of failure modes of components into rodon is very important because they are decisive in the performance of the diagnosis and are critical parts of the knowledge base.

The essentials of this approach are as follows: A technical system consists of a number of functional units or components which are connected with each other to form an assembly or system. The generic behavior of the individual components is known and can in general be described in mathematical form including tabulated sets of measured values. The behavior of the system as such is then defined by the behavior of the components and their connections. The mental reasoning process described above encompasses diagnosis (backward reasoning) as well as simulation (forward reasoning). The basis of the mental model is the nominal or regular behavior of the functional units and not the symptoms of the failed system. Faults are detected and isolated on the basis of the deviation of the faulty behavior from the nominal behavior.

The capabilities of RODON as stated above can generate significant benefits by allowing simulation of multiple systems incorporating different technical disciplines. Nominal operation can be observed across all of the systems meaning that all interfaces are understood and that no mismatches have occurred. It is important to note that this is being done at design time rather than after systems have been built and physical integration started when redesign becomes very expensive. Further, system operation under various failure conditions can be observed. The manner in which fault symptoms propagate from one system to another can be simulated and captured into the knowledge base for instant use later when the total product is in the field and real failures occur. Diagnosis by repairmen or maintenance shops is significantly aided by prior knowledge of failure signatures.

RODON thus is a tool which fits the TPO concept and, through its knowledge base, follows the product from concept to deployment and maintenance. The operating modes of rodon are consistent with the design and implementation life cycle phases of products, whether they are new designs or improvements of previous designs.


LOW INVESTMENT

RODON has a unique capability to integrate into existing design and simulation environments so that companies do not have to discard old design methodologies quickly. Over time, RODON can phase into and upgrade current processes. A tool like RODON can help overcome the "legacy factor" which tends to impede beneficial change in established companies. RODON creates an environment for systematic improvement of employee proficiency. Investment in process improvement occurs continuously in progressive companies and RODON can become the centerpiece of such improvement. RODON is part of the improvement investment and can be the impetus for continued progress in the important area of product optimization.


CONCLUSION

The principles of focusing engineering teams on specific products and the use of modern design optimization tools like RODON are extremely important to process improvement and overall design optimization. Creation of a knowledge base from the start of a product's life cycle and carrying it through to its end produces a new level of product engineering and operation efficiency. Risk is significantly reduced by focusing engineering manpower into dedicated teams and using a central modeling tool like RODON to integrate multiple engineering technologies to achieve a technically optimized product.


REFERENCE

Werner Seibold, A Model Based Diagnosis and Simulation System in Practical Use - The Concept of RODON. In Third International Workshop on Principles of Diagnosis, October 1994

July 1995
Bill Lokken

Return to White Paper Page