Manufacturers with six sigma goals are beginning to see
that quality improvement starts in engineering design
by Gavin Finn, Ph.D. The evolution in mechanical-design software during the past decade has
permanently changed the manufacturing industry. Three-dimensional computer-aided design, manufacturing and engineering (CAD/CAM/CAE) software programs are close to fulfilling their promise to deliver fully
integrated design and manufacturing. CAD design software saves time and money in product development by reducing or eliminating the need for physical mock-ups, allowing for early detection of interferences
between components and enabling quick design iterations that result in product optimization. To further this integration, design engineering is taking new quality assurance lessons from
manufacturing. On factory floors all over the world, attention to quality in product-manufacturing operations now determines market success. Attention to quality in the operations used to produce engineering
designs--in other words, attention to the data produced by digital modeling activities--is proving to be equally fruitful. Practitioners of design quality assurance are reaping great benefits by pushing
quality awareness earlier into the engineering process. The quality challenge As a rule, design engineering
has lagged behind the shop floor in awareness of product and process quality. Unfortunately, very real costs are associated with inattention to design quality. If errors or omissions in design data are not
addressed early, more costly changes are required later in the product development process (see Figure 1). Figure 1: Cost of Changes
For example, a major aerospace company recently discovered that it was
losing millions of dollars as a result of inaccurate digital design data. Product quality on the manufacturing floor was high, and customers were receiving
excellent finished products. However, the company didn't apply the same rigorous attention to assessing and controlling the quality of the digital modeling
process in design engineering. Errors and inconsistencies in the 3-D CAD models regularly caused hours of rework for engineers and significant problems
for others in the organization. Engineering managers were stunned to learn of the product development costs they could have avoided by applying quality assessment and control to design.
On each digital model generated in the design of the company's products, engineers were spending at least one additional hour to examine and prepare the
electronic data for manufacturing. Multiplied by the 10,000 digital models generated per month, the time allotted for data cleanup was substantial. A
typical design package containing 80 parts required 80 hours in cleanup, equivalent to two weeks of work for one person. This didn't include the
additional time required when models had to be returned to designers for corrections, nor did it count the cost of errors that slipped through and caused headaches later on in tooling or production. The wasted time in engineering was only part of the problem. The more troublesome costs, and the most pain for manufacturing, came from resulting delays in aircraft assembly schedules. Manufacturers in other industries, such as automotive, electronics and consumer products, have had similar insights into the importance of design data
quality. Some of the traditional inattention to quality issues in the creation of digital models may be attributed to a perception that the design process, unlike
the manufacturing process, is not a series of repeatable actions that can be controlled. But the regular, repeatable errors that design data contain prove this perception false. Repeatable processes The manufacturing engineers at the aforementioned aerospace company saw a
$150,000 design data error return repeatedly in separate digital part models. The problem was a mismatch between the part dimensions represented in the
model itself and the measurement specified by designers in text on the dimension line. Because of the design error, the first articles produced were consistently
wrong. Each individual digital model was always fixed. However, because the design process that produced the error wasn't fixed, the error continued to
occur--and continually had to be fixed--at a cost of $150,000 each time. If this isn't a set of repeatable actions, what is? The example shows that the
company had a method in place to identify instances of manufacturing error, but it had not yet developed a method to identify and correct error in the design process.
This story is widely representative of circumstances in many companies around the world. The beauty of integrated design and manufacturing is that it cuts
product cycle time, but successful integration hinges on the quality of the design data passed to manufacturing. Engineering design practices that are subjected to
process improvements will aid this collaboration instead of hindering it. Enterprise-ready engineering products
There's another reason for the lack of attention traditionally paid to design data quality: Engineers don't tend to think of a design as a product that can be
measured and compared with standards. However, a digital design is a recognizable set of electronic data. It is a product produced by an engineering designer.
There are many steps in the product realization process, from conceptual design to service and retirement. At each step, a new "product" is delivered to
people in the organization who rely on the information to continue development. For example, manufacturing engineers generate assembly diagrams, which
technicians use to put assemblies together. The assembly diagram is a product that engineers supply to the technicians. Likewise, the "engineering product"
generated by design engineers is the digital product model. This includes the 3-D CAD model of the design and the associated nongeometric data, such as ma- terials requirements.
The engineering product is used throughout the organization by manufacturing engineering, procurement, documentation, customer support, suppliers and
vendors, and manufacturing. All of these downstream users are engineering's customers. How well the engineering product meets the requirements of these
customers is a measure of the engineering process's quality. If an assembly diagram is deficient in some way, manufacturing time is lost in correcting it--or
worse, in rebuilding incorrectly assembled items. Unless quality control begins earlier in the product development process, the design organization may never be made aware of these problems. Data interoperability Today, many manufacturers acknowledge that the engineering process is
flawed because digital design data cannot easily be shared. Problems with data interoperability occur as a consequence of inattention to design data quality, and
they hobble entire industries. For manufacturing organizations to collaborate productively, both internally and with suppliers and vendors, they must be able
to produce interoperable data (data that can be exchanged easily between one software package and another). In March 1999, the National Institute of Standards and Technology issued an
extensive report on the consequences of poor interoperability in digital product data. The study, called "Interoperability Cost Analysis of the U.S. Automotive
Supply Chain," stated, "Imperfect interoperability [of digital data] imposes at least $1 billion per year on the members of the U.S. automotive supply chain.
By far the greatest component of these costs is the resources devoted to repairing or reentering data files that are not usable for downstream applications." Design data for different users Let's look at just one instance of design data produced for downstream use.
Digital product models are regularly used as the basis for creating toolpaths for part manufacturing. The catchphrase "art to part" describes the goal of seamless
design-through-manufacturing, but digital design data usually must be corrected before toolpaths can be generated. The reason the process is, in reality, anything
but "seamless" stems from differences between product design software and tooling software. For example, in a shaded solid model, it may appear that two curved surfaces
are touching. However, when the model is imported into the software program used for tooling, it becomes apparent that the curves do not actually connect.
The problem occurs when the two software systems have different mathematical tolerance requirements for the construction of curves. To the designer viewing
the model with the design software, the digital model looks perfect. But when the manufacturing engineer tries to make a toolpath using numerically controlled
software with tighter tolerances, a gap between the surfaces appears, preventing generation of the toolpath. In this case, some accounting for tooling
requirements up front in the design process would help the company create an engineering product that is more enterprise ready and that minimizes rework.
Most manufacturers currently have no way of telling how suitable their engineering product is for enterprise needs. In order to do so, design engineers
must submit the engineering product to quality testing procedures. This is where design can learn from manufacturing quality assurance.
Manufacturing quality assurance lessons Like the physical product produced on the manufacturing floor, the digital
product model has many characteristics that can be identified and measured. The difference is in the way measurements are taken. Quality assurance
professionals use metrics, or standards, such as the distance between two holes or the thickness of a wall, as a basis for deciding if the physical parts produced
on the shop floor have been made according to the design intent. They use the appropriate tools (such as calipers, lasers, coordinate measuring machines and
so on) to take measurements. They then determine if the physical measurements conform to the expected values. All of these procedures are planned ahead of time as part of the company's
overall quality control process. Defects or variances discovered in parts can be traced back to specific errors in manufacturing operations. When enough of the
same errors continue to occur, quality professionals locate the problem in the manufacturing process and correct it.
Quality planning for digital design data works by answering the same three questions:  What do we need to measure?
How should we measure it?
How can we fix the errors we find and improve the overall design process so that we avoid the errors altogether?
Metrics for design quality The operations used to make the engineering product--in this case, the company's digital modeling practices--provide opportunities for inconsistencies
and errors, just as in manufacturing. Developing standards and methods of measuring adherence to standards involves identifying which of these operations
are critical. Critical operations will vary by manufacturer and may include geometrical accuracy or format of the data.
For example, let's say that a particular company's automated material resource planning (MRP) system regularly checks every digital product model for
electrical components to generate bills of material. The MRP system is set to look for electrical components in the same place in each design: in a particular
data layer. If electrical components are inadvertently left out of the specified layer, the MRP system will never know it. It will generate an incorrect bill of
materials, and manufacturing will discover the error--much after the fact--as inventoried components come up short.
A quality assurance approach to avoiding this problem would begin with the identification of data format as a critical metric. Design quality software would
then measure and analyze the design data for correct format. In fact, the software would determine whether each electrical component was in the correct
layer and would flag any that weren't. The software would automatically correct the error or alert designers to fix it before releasing the data to the MRP system. Standards to develop Companies must develop four categories of metric standards to judge the quality of the engineering product:
Design standards reinforce the use of company-defined best practices for 3-D modeling. These standards might include details such as the correct method
for digitally representing a hole so that machining software can interpret the data. Geometry standards encourage consistent and accurate handling of geometric data and help avoid problems with data exchange, analysis and
manufacturing. As was explained previously, geometric gaps that appear in toolpath programming can be invisible to design engineers. Certification standards ensure that only quality-tested product information is
released to downstream users. A reliable system for releasing data ensures that assembly diagrams, for instance, are complete and correct. Management standards provide guidance for strategic use of design quality
data. The organization's other systems, such as MRP and enterprise resource planning, are undermined when they receive poor quality design data. Substantial savings The costs and manufacturing rework that can be saved by instituting quality
assurance in the design process are substantial. Organizations willing to audit their design data for quality errors usually discover immediate rewards.
A remarkable example comes from the experience of a major appliance manufacturer. The sheet metal enclosure for one of its appliances consists of
seven parts, each of which requires a different sheet metal die. Each of the seven dies costs $1 million to produce. Before manufacturing the dies for the
latest model of the appliance, the company analyzed the design data using DesignQA, a design-quality software package from Boston-based Prescient
Technologies. It found design errors that would have rendered each one of the dies useless. Simply by analyzing the design data, the company was able to prevent a $7 million manufacturing mistake. Improving design quality Quality assurance programs such as six sigma have proven that companies gain
both a competitive advantage and an economic advantage when they pay attention to the quality of the manufactured product. There are three elements of
design quality: the adequacy of a design relative to the functional requirements, the accuracy of the geometrical data and the suitability of the design data to
meet downstream product data requirements. As we have seen, an organization is achieving design quality when the engineering product it creates can be effectively used by downstream organizations. Many practices developed for quality in manufacturing can be adapted for use in design quality assurance. Similar steps apply: (1) Detect errors or deviations;
(2) assess the errors detected in order to decide what to fix; and (3) correct the 3-D model so it conforms to requirements.
How software tools can help The most effective way for a manufacturing company to address engineering design quality is to implement a complete program of automated quality
assurance. The aerospace company previously discussed used PrescientQA software to determine the exact nature of its data quality problems and the cost
burden of ignoring them. The direct cost of not preventing the errors turned out to be more than $50,000 for one month for just one design team. The quality
software allowed the organization to quantify improvements and set a program for the future. Engineering design offers companies a whole new arena within which to apply
quality practices. The first step is often a quality audit. When the aerospace company performed a quality audit of its design process, it discovered that 50
percent of the errors found in the design data would have led to critical fabrication errors. These errors would have rippled through manufacturing. The
moral of the story is that design quality is tied inextricably to finished product quality and, therefore, to a company's ultimate success.
About the author Gavin Finn, Ph.D., is president and CEO of Prescient Technologies Inc. and is a leading industry spokesperson on the issue of design quality and
its effect on the engineering process. Finn has served as a consultant for such activities as quality improvement in the engineering process, design
and engineering automation, and the integration of design and manufacturing. E-mail him at gfinn@qualitydigest.com . |
