It's not just data

Manufacturing Engineering, Jan 1999 by Erickson, Mark A

To detect problem trends, pinpoint causes, and form solutions, users need relevant data in understandable chunks

I need more data before I can make an intelligent decision." How many times have you heard that statement? Actually, the typical manufacturing professional swims in data-great waves of data that increase every day, every shift, every hour, and threaten to drown a person in a flood of seemingly unrelated numbers.

No, it's not simply more data that manufacturing professionals need, but coherent data presented in an understandable context. That's the key to intelligent decisionmaking. Nowhere is this need more evident than in automotive assembly, where features like points or patches, holes, slots, corners, and edges must be checked over time-over the life of the process. Many measurement systems now in place in auto plants aim to do just that. Most, however, fail to live up to their full potential for ensuring not only functional integrity but also the excellent fit and finish crucial to success in today's globally competitive vehicle market.

Why? Two reasons: first, many existing systems simply aren't powerful enough to do the job; second, and more important, they don't give a sufficiently broad range of users the ability to easily group the data in understandable chunks. Sure, a QC expert sitting in his or her quality command post can usually analyze the inspection data coming from measurement systems on the factory floor and arrive at intelligent conclusions regarding the health of the manufacturing process.

Unfortunately, such conclusions may emerge after the fact, after scrap has been made or-equally important in today's environment-after less than optimum product has been built. Today users need measurement systems that let front-line decisionmakers quickly access relevant data in understandable chunks, detect problem trends, pinpoint causes, and formulate plausible solutions.

Next comes plant-wide data management. A plant-wide system should collect and store dimensional data in a single database open to multiple users, both local and remote. That means it must be compatible with a plant's existing networking system, not a QC orphan or outcast, so that people can use it to compare measurement data from different devices and transfer data to other database systems in the plant via local area networks.

Finally, users need a suite of fast, easy-to-use analysis tools that allow both QC experts and front-line decision makers to zero in on the root causes of problems and potential problems. People must be able to comprehend and report the inherent or potential build and position inconsistencies of the various parts that form the final assembly. This capability makes it possible to ensure the final assembly's best fit, given the limits of each part's inherent errors. Users also need a system that can communicate between the database and the gages that supply the data, so that the gaging process can be fine-tuned at a speed comparable to the manufacturing process.

When 100% in-line measurement of hundreds of points generates oceans of data, and the process variation waves start getting bigger and bigger, threatening to swamp your quality, your competitive advantage, and your career, you need a robust system guided by a comprehensive and processcontrol-oriented methodology. Systems like these enable their users to zero in on the largest wave, ride it to its source, locate and address the problem, then move on to the next wave, and on down the line.

Such data surfing produces better process control, resulting in superior fit, easier assembly, fewer rejects, and fewer warranty claims. Strategically, these benefits spell higher quality and lower cost. All of the hardware and software components for systems like these are currently available. All that's needed is the wisdom, the will, and the plant-floor savvy to integrate and use them.

Mark A. Erickson

Training Manager

Perceptron

Plymouth, MI