Automotive manufacturing companies increasingly rely on collaborative design and development to shorten product cycles. This means that people responsible for different aspects of a product throughout its life cycle now need access to accurate and up-to-date design information. Good quality design data are more important than ever in helping organizations work together. At the same time, design data quality is somewhat of an uncharted land for most manufacturers. This is even true for companies that have six-sigma goals on the factory floor. When they audit their engineering data for quality, many find that the simplest of design inaccuracies cause the most waste and iteration in development time—misnaming CAD models, for example, or building geometry that can't be manufactured.
For years, companies used manual checking methods to correct inaccurate digital models and to reinforce design standards. Engineers would spend days poring over and manually fixing digital models before they could release them to the organization. Now, software for automating design quality from companies such as Prescient Technologies, Inc. cuts down on this administrative time and allows manufacturers to practice design-quality assurance concurrently with design.
In 1999, Prescient gathered data from manufacturing companies all over the world to determine the extent of the design-quality problem. The initial results from an engineering quality audit program show a pervasive problem with data quality in the process of engineering design, with a consistent level of error in engineering data tested across aerospace, automotive, consumer products, and electronics industries. Quality audits were performed in companies with as few as 100 and as many as 160,000 employees. Over 3000 separate product models were analyzed during the study, of which only 225 passed the appropriate set of each company's defined standards, while 70% of the models failed standards that companies categorized as "critical."
Prescient's quality audit process is customized to look at the engineering data standards of most concern to each specific company. These can be common industry design standards and best practices, or a set of quality issues unique to the organization. The standards can be as straightforward as naming conventions or as complex as manufacturing requirements and guidelines for electronically building the geometry of a part.
"Data quality has been one of the most significant issues in the product development process for a long time," noted John MacKrell, an industry analyst with CIMdata, Inc. "To achieve the full benefits of advanced engineering and data-management solutions, companies must have information of the highest possible quality."
"The audit has only scratched the surface of the issue of data quality in engineering," said Gavin Finn, President and CEO of Prescient. "Although there are more analyses to do, the overwhelming rate at which data failed defined standards was consistent across all the audits. The high percentage of errors is even more noticeable because the results are not limited to company size or market. This problem spans the entire manufacturing industry. The problem is a consequence of the increased pressure to utilize digital data throughout the automated product-development process."
The issue of quality in engineering data is growing more crucial as digital-modeling software becomes integral with automated product development. Inaccurate or incomplete design data affect the product-development process in a number of ways. Models are used increasingly by other product development functions such as manufacturing, procurement, and documentation, so errors in design data add rework time and cost downstream. Design errors also reduce the capacity for data to be exchanged between different software systems. The result is that models are often recreated from scratch, or substantially reworked, and this limits a company's ability to utilize legacy models in new designs.
"Companies have already acknowledged that inaccurate design data cause additional costs and problems for people downstream of engineering who need to use the data," stated John Racine, Vice President of customer service and implementation at Prescient Technologies. "But until recently the technology has not been available to quantify the size of the problem. We never expected the data failure rate to be this high. Neither did the companies that participated in the audit program."
"Companies have historically accepted these costs as a side effect of business in the digital age," stated Finn. "However, given the constant pressure to keep product development costs down, it is very timely that technology can now help organizations put some quantitative numbers not only to understanding the problem, but to solving it."
Product-modeling software systems are inherently flexible, offering engineers a number of ways to create, assemble, and annotate a digital model. However, without a set of defined guidelines for model structure and design, the danger is that each designer will create models according to his/her own individual methodology, rather than in accordance with company standards. As Prescient's AuditQA program discovered, this causes problems for the organization, because even designers on the same team may not be able to make simple changes to one another's designs.
The audit programs showed that some companies have developed a set of design standards and best practices for engineers to follow, but reinforcing those standards has been a challenge. Software tools that can measure models against standards have only recently become available.
"It comes down to questions of training and the consistent use of best practices," added Finn. "Engineers want to do the best job possible, and they need tools to guide them toward the best way to use the system for the specific job at hand or for a particular customer."
For additional information on the quality audit process, visit www.prescienttech.com or circle 75.
SAE Off-Highway Engineering June 2000
