The role of CAI in competitive manufacturing
Inspect: from the Latin inspicere – to look deeply, carefully, within
by Peter Marks
Editor’s note: This is the second of a two-part article. The first part looked at the evolution of inspection and CAD within the context of manufacturing trends. This part examines tools that are beginning to revolutionize the middle ground between design intent and real-world production. |
Just as CAD has revolutionized geometric design, a new generation of geometric scanning and inspection tools is beginning to revolutionize the transformation from design intent (increasingly 3D CAD augmented with tolerance information) to real-world production.
We can now quickly capture thousands of measurements from almost any complex part and compare actually produced geometries to their idealized CAD counterparts. While this technology is still somewhat new, participants in a technology forum earlier this year were clear that benefits are compelling, even at this stage of development when equipment costs are still fairly high.
Accuracy was less an issue among forum participants than expected, although there is still major work to be done among vendor and standards bodies to allow meaningful comparisons between equipment.
The group identified at least 12 “middle-ground” functions between nominal design and full production where 3D scanning technology frequently offers a compelling advantage:
 |
1) 3D capture.
This is the creation of a nominal digital model in those cases where there is either no design reference (e.g. some legacy designs), limited design data (e.g. manual or digital 2D drawings), or incorrect data (e.g. errors or ambiguities in the 3D data). The term “reverse engineering” is often used to describe this process. A better term might be “forward engineering,” since the goal is to have the many benefits of a 3D model going forward into manufacturing and product support. In addition, the usual notion of reverse engineering is to capture one instance of an existing component and recreate it to the exact dimension. The forward-looking approach is to consider such issues as wear in the existing part, the other parts with which it must mesh, and suitable tolerances for manufacturing. This is the difference between a cheap clone and an engineered replacement.
2) Process selection.
There are often multiple combinations of materials and processes (casting, welding, molding, machining, etc.), with different cost and performance tradeoffs. Selecting the ideal combination is a non-trivial process. In the past, companies tended to target processes which they already employed. Some construction equipment companies, for example, were comfortable with large castings as a starting point and others employed fabricated (welded or bolted) structures.
Today, picking the best processes for both cost and performance can be the difference between dominating a market and being number two or three. Something else to consider: The best process a decade ago may not be the best process in light of new manufacturing options, improved materials, or changing customer requirements.
Companies face at least three hurdles in changing processes, even when it is a matter of survival and success. First, they must recognize the need for change. Just as there is an “innovator’s dilemma” in product design, so there is a dilemma in confronting new manufacturing processes. Second, they must address the many subtle details in changing from one process to another. The part may be basically the same, but entirely new issues such as draft angles, shrinkage factors, wear patterns, locating points, and the like emerge. Third, there are the usual issues of human resistance to change. People are always reluctant to risk new methods, especially if the old methods worked well for so many years.
In all cases of process selection, 3D scanning technology is often the best way to characterize both prototype tools and parts.
3) Tolerance design.
 |
The point is that engineering is not complete until real-world variation is accounted for. For reasons of cost, we accommodate as much variation as possible. For reasons of performance, quality and safety, we must clearly specify the limits of acceptability.
Tolerance design is one of the most misunderstand disciplines in modern manufacturing. Designers are typically given the responsibility for tolerance selection, but they often lack the motivation, time, training and support to do the job. In many cases, the task of reconciling these issues (tolerances too tight or too loose) falls into a formal or informal process of neglect or negotiation between engineering, tooling, inspection and manufacturing. Significant differences in cost and performance often hang in the balance.
One participant noted that getting tolerance design right early in the process can save tens of thousands in tooling costs and also avoid the potential for catastrophic failures. A combination of analytical tools (tolerance stackup analysis) and experimental methods (prototype production and testing) is commonly employed.
4) Manufacturing process optimization.
Materials and processes are selected based on rules of thumb or comparatively simple models of cost and capability. Depending upon the process, specific details such as casting or molding shrinkage factors must be determined. While there is usually experience and (sometimes) analytical tools to support this, prototype tooling and experimental methods are often required. As any manufacturing engineer knows, the desired part geometry is not the same as the mold or die geometry. Metal bent to exactly 90 degrees will spring back to something less. Machined parts cut to 0.0001 accuracy may distort to something else as they cool or are unclamped. Process design aims to characterize a process so precisely that the actual parts produced (and scanned) fall within design tolerances.
Retooling is a special case of process optimization. A company may have gone through years of process design, perhaps trial and error, to arrive at a set of working jigs, fixtures, molds, dies and other tools. That experience may be lost or undocumented, but the company still needs to retool the process. Perhaps the old tools are worn. Perhaps production is expanding. Perhaps production is moving to another location or different equipment. Characterizing the existing tooling (3D scanning) is typically the starting point for capturing past knowledge and procuring new tooling.
5) Tooling inspection.
The production of tooling is commonly outsourced. Even if tooling is built in-house, there is still the issue of conformance. Assume that a company has selected the ideal process, carefully determined and documented tolerances, and accounted for every bit of shrinkage, spring back and manufacturing variation in its tool designs. Now, the company must assure that the tools as built match their designs. This is comparatively simple for things like one-dimensional gauges and fairly complicated for complex shapes with tight requirements for fit (e.g. mating surfaces), finish (e.g. no surface defects), aesthetics (e.g. perfect visual blending) and performance (e.g. airflow). Tooling inspection, increasingly through scanning methods, is typically employed. In those cases where precise control of tolerances is critical, companies employ two tooling stages. First, prototype tooling is used to verify and refine the process. Second, production tooling is used to begin quantity production.
 |
6) First-article inspection.
If everything has gone according to plan, prototype tooling produces nearly perfect parts and production tooling offers that ideal combination of fast and affordable production of parts and assemblies that meet every user requirement. One of the benefits of many net-shape and numerically controlled processes is that if the first part is good, then the sources of subsequent variation such as mold or die wear or tool breakage are easier to track than in less-controlled processes with more room for operator error.
Modern manufacturing increasingly relies upon first-article inspection of controlled processes. Especially for net-shape processes (molding, casting, forming, stamping, etc.) and multi-axis machining (complex surfaces), scanning or a combination of scanning and precise coordinate measurement of critical features is the best available technology.
One point made during the technology forum was that the part designer (or company management, for that matter) rarely understands that the cost of first-article inspection and verification may be on the order of $50,000, because this cost is built into the tooling cost.
Functions one through six, above, typically report to engineering, manufacturing engineering, pre-production, tooling, and other groups outside of day-to-day manufacturing and operations management. At this point, responsibility for dimensional control typically passes to a production-oriented quality control. In some companies, these people report to operations. In other companies (especially if manufacturing isn’t trusted or the company wants to show customers its commitment to quality) QC reports higher up, perhaps to the CEO.
Production advances
While the first six functions are among the areas with greatest near-term return on investment for scanning technology, we are also seeing significant advances in production applications:
7) In-process inspection.
The first ideal is to select a robust manufacturing process – one where sources of manufacturing variation are rare and comparatively easy to track. The second ideal is to track those sources of variation in real time.
In automotive machining lines, for example, cutting tool wear or breakage is a likely source of variation. Rather than produce several out-of-spec parts (engine blocks, valve bodies, etc.), in-process inspection methods such as contract probes have been commonly employed. In electronics fabrication (e.g. circuit board stuffing and soldering), vision systems are common.
While 3D scanning methods offer great promise for real-time inspection, a variety of issues such as scanning equipment costs, speed, accuracy, environmental conditions (coolant, swarf), retooling costs, data processing speed, and the like limit their use to specialized high-value applications. In the near future, 3D scanning is expected to become increasingly competitive with other contact and non-contact in-process inspection methods.
8) End-of-line or end-of-cell inspection – hard vs. soft gauges.
A sampling of parts is commonly selected to assure that a manufacturing process has not drifted out of control. For simple parts with a few critical dimensions to be checked, hard gauges are commonly employed. But, the cost of hard gauges skyrockets as gaging requirements become more complex and as different products (requiring new gauges) are introduced.
Industry is seeing increased demand for “soft gauges,” where a single piece of equipment (such as an articulated arm with a scanner) can be quickly programmed to gauge different parts. The payback can be very quick – and is sometimes justified with the replacement of a single hard gauge – with reuse beyond that being almost pure profit after the reprogramming cost. Faro and Perceptron were cited as key scanning hardware vendors in this area.
9) Detailed coordinate verification.
For complex parts and assemblies with many critical dimensions, samples are commonly taken into a quality control lab for detailed verification. Decades ago this work was laboriously completed with height gauges on granite surface plates, optical comparators, and the like – and it is still done this way in many facilities.
More recently, coordinate measuring machines (CMMs) have displaced purely manual measurement methods. Scanning methods are now poised to take this a step further. While the best technology depends upon the application, a current generalization is that scanners are ideal for quick characterization of an entire complex geometry. A hybrid approach, where scanning is used as a first step and a CMM is used either for problem areas revealed by scanning or critical dimensions required by contract is often best. In response to this, CMM vendors are producing machines with interchangeable touch probes (traditional CMM) and optional non-contact scanners. Metris, Laser Design, Inc., Kreon and Nextec were some of the suppliers identified by forum participants.
10) Tool wear monitoring, prediction and rework.
As noted earlier, net-shape processes such as casting, molding and stamping tend to eliminate some sources of variability. If the first article is right, the next few 1,000 articles are more likely to be right – at least compared to hand crafting. Tool wear is, however, one of the common sources of process capability degradation. Given that some net-shape tools (e.g. auto body stamping dies) cost hundreds of thousands of dollars, there is significant work going on to predict tool wear, increase tool life, and monitor tools for rework or replacement. Scanning technology is the ideal adjunct to any effort to understand tool wear, experiment to achieve longer life, monitor wear, and verify rework.
When dimensional problems arise in a once-capable process, the investigation usually goes first to quality control. If the problem isn’t obvious or the technology required to solve the problem is fairly sophisticated, the investigation is often thrown back to whatever people in the company handle the middle-ground issues. Which bring us to an 11th step.
11) Dimensional troubleshooting.
Subtle changes in suppliers, material composition, tooling wear, and the like can throw a process out of tolerance. If scanning technology has been a benefit in previous steps, it’s a near certainty that it, along with sophisticated point-cloud analysis software, will be an even greater aid to problem identification, analysis and solution.
Two additional applications for the new scanning technologies are similar to numbers 10 and 11, above, but performed to audit or streamline specific supplier-contractor relationships.
12) Outgoing and incoming inspection.
Many customers require either a sampling or 100-percent inspection of outgoing parts. Scanning is often the most productive method, the best to assure conformance with a 3D digital model, and (one might predict) increasingly preferred as a near fool-proof verification method.
Our group did suggest that suppliers might not always prefer scanned inspection, since it commonly reveals “problems” (in quotes because they may or may not be functional issues) that had not been revealed in older methods of inspection. Over time, we can expect suppliers and their customers to work out the problems, much as they continue to work on problems with communicating 3D design data.
Incoming inspection has basically the same objective as outgoing inspection – further assurance of quality – but is performed by the buyer or user. Either sampled or 100 percent, the objective is to assure that nothing gets into the system that can cause problems for the ultimate customer. In some cases, both outgoing and incoming inspection, perhaps on a sampled basis, are required.
The decision of whether to require outgoing or incoming inspection, and at what level, speaks to issues of trust and competence in supplier relationships. Some industries are pushing inspection requirements out to their suppliers. Others are pushing most requirements out, but still sampling incoming items. Still others are ignoring the issues, hoping to save costs, and ending up being disappointed. The dramatic rise in outsourcing to low-cost suppliers is now putting all this in flux.
So, to summarize, there are at least a dozen application areas where non-contact inspection (especially laser or white-light scanning) is already making, or is poised to make, dramatic advances.
High-payback applications
The experts participating in the forum are bullish on the new technology. They noted that the bad experiences of just a few years ago (shiny parts that wouldn’t scan, accuracy less than needed, millions of points that choked the software) had increasingly turned into success stories with the latest scanners and software. That said, they are also mindful of remaining limitations and cost barriers. There are still relatively few economies of scale, for example, for hardware vendors, especially at the high end. The following seven criteria distinguish applications with high paybacks – a “yes” to three or more of these would suggest taking a closer look:
 |
1) Complex geometry or surfaces, where a few simple measurements are not enough to characterize the component. Turbine blades and circuit breaker housings are examples.
2) Non-rigid materials, where touching the part will alter the measurement.
3) Assessing differences between “as designed” and “as shaped” geometry due to the manufacturing process. Accounting for shrinkage, distortion, spring back and other conditions requires at least two (prototype and production) tooling cycles. A combination of #1 (complex shapes) and #3 (net-shape processes gone slightly astray) is a big win for scanning applications.
4) Out-of-tolerance parts or products that pose significant risks, ranging from customer acceptance through efficiency to safety, if they get into the field.
5) Complex interactions between geometry and product performance, with performance being anything from the physics of fluid flow to the aesthetics of car body design.
6) Separation of design and manufacturing functions. It’s already a moat rather than a wall, and outsourcing further separates know-how and responsibility. Scanning is the simplest and most-complete method of getting a valid comparison between “as designed” and “as tooled” or “as built.”
7) A need to better characterize the middle ground between nominal design and full production, to survive and thrive in a global economy.
Competitive implications
In some respects, the new CAI technologies are a two-edged sword.
Cutting one way, they make it easier to capture the geometry of a product and create a clone or derivative design. Products that haven’t changed much over time, whose look, feel or intellectual property is up for grabs, become vulnerable to low-cost manufacturers. Milwaukee Electric Tools, the former U.S. maker of power tools, was a frequent target of low-cost clones selling at about one-fifth the price. There wasn’t enough “worth more” in the bearings, motors, switches and manufacturing equation for some customers. So, it shouldn’t have been a surprise when Milwaukee (along with AEG) was acquired by TTI, a Hong-Kong-based supplier.
Cutting the other way, CAI gives companies a powerful tool to get the combination of performance, price and manufacturing process just right. An example is circuit breakers that meet the highest standards of reliability for just a little more than a bet-your-home-and-family clone. Another example is turbine blades whose value propositions range from “too complicated for anyone to copy” to “so much more fuel efficient and reliable you’d be crazy to choose second best.”
Manufacturing companies that want to survive and thrive need to be great at design and manufacturing. Increasingly, that means paying attention to the role of CAI in competitive manufacturing.
Peter Marks is president of Design Insight, a consultancy for new product development. The firm’s methods of customer research, process improvement, and technology selection have helped clients add several billion dollars of incremental product and service revenue.