Engineers spend hours perfecting analytical models without fully understanding the environment in which the final products must survive. Design criteria often comes from guesses made early in development about loads, the number of time they are applied, and for how long. At best, uninformed decisions like these result in overdesigned, costly components. At worst, they drop to the bottom line in the form of increased warranty claims, poor performance, and unhappy customers.
Customer Usage Profiling (CUP) is one way out of the dilemma. It is a data-acquisition and analysis method for recording product performance after a product or prototype is in customer hands. Profiling activities involving direct measurements and statistical analysis can take the guesswork out of product development. While this type of process has been used in the automotive industry for some time, others are beginning to take advantage of the information gathered from well thought-out programs.
CUPs capture unbiased, real-world data while the product is being used, giving a true picture of product performance outside the confines of a traditional laboratory. Collecting and understanding this information decreases the likelihood of under or overdesigning in the early development stages. This translates into reduced warranty and manufacturing costs. Usage profiling can benefit manufacturers of everything from mining equipment to miniature cell phones.
CUP systems exploit the computer and electronic advances which have brought about a proliferation of digital data-acquisition systems. Equipment that once filled a small truck now fits into a box the size of a deck of cards, and smaller custom acquisition systems can be manufactured.
Before taking the first measurement, a suitable strategy must be developed for gathering the required data. Engineers must be weary of over complicating the project by trying to measure every nuance of a product's operating environment. Maintaining more data channels than necessary adds cost and complexity. Conversely, care must also be taken to capture all information pertinent to performance. History and warranty records often provide insight into which measurements are important.
In addition, the level of customer interaction must be defined before instrumented products are put in the field. Based on project requirements, one of three levels of interaction is usually adequate:
- Users actively participate in the acquisition. Collecting data in a factory environment is one example. Temperature information for a recent project was collected on a large piece of food-service equipment as it operated. The machine's operator actively monitored the tests.
- Users are aware of acquisitions but do not actively participate. A detailed example follows later.
- Users have no knowledge of the data-gathering activity. This type of acquisition is used when customer knowledge could bias the data. Keeping the collection process completely autonomous and hidden make this the most difficult.
A recent project illustrating this procedure involved a fleet of recreational vehicles. Standard production vehicles were instrumented with channels that measured acceleration, displacement, temperature, and event occurrences. The data-acquisition equipment and sensors were hidden. In addition, tests ran completely autonomously. No interaction whatsoever was required for several weeks at a time while the systems recorded performance data.
In addition, most CUP programs fall into at least three phases: product instrumentation, field-data collection, and data analysis. During product instrumentation, data-acquisition equipment is installed and its operation verified. Standard transducers and required signal conditioning are often not available, so special sensors and electronic equipment may have to be designed for the project.
After instrumenting a product, it can be given to customers to begin collecting data. Depending on the required level of interaction, this may simply entail handing it over to a group of users. Typical data-acquisition systems log several megabytes of information. Based on the number of channels and type of data recorded, this may be enough storage for weeks of uninterrupted operation. Data are typically uploaded periodically as different people use the product. This allows recording performance as a function of different types of users. In addition, studying field data helps determine whether or not the system is operating correctly.
Data from a CUP program and appropriate analysis can answer questions regarding product performance. Depending on a program's goal, CUP data can be used to do everything from determining maximum load levels to quantifying why a specific product is viewed as "more comfortable." It can answer a wide range of questions such as, how many times will this door be opened (or this latch pushed or switch activated) in the normal service life of the product? Or, what is the 95th percentile (or the 90th, or the 50th, or the maximum) load that users are putting on a product's armrest (or faucet handle, or trigger switch)? How do users in Ohio, Florida, or Arizona use the product compared with those in Taiwan, Germany, or Mexico?
A recent CUP project with furniture manufacturer Herman Miller Inc. shows the capability of the process. Manta helped institute a program that was based on customer profiling of a standard office seating line. Data was used to generate laboratory tests simulating the most destructive users and understand an office chair's daily environment. Based on a preliminary study of several seating lines, the Ambi model chair was chosen for the customer usage field test. The chair was a good candidate for extrapolating results into other chair lines because of its particular construction and well-defined geometry.
Twelve chairs were instrumented and given to customers at six different locations over a four-month period. Over 100 users were eventually studied. Data was harvested at one-week intervals.
Six measurements of load and position were made on each chair. Three channels were dedicated to measuring loads and moments at the seat post. In addition, moments were measured on the seat back and armrest. To evaluate the chair's configuration during various loading conditions, the seat-back position was also measured.
Commercially available computer systems collected field data and 12-V gel cells provided power for each system. If the chair was not disturbed for 2 hr, the power was shut down to conserve resources. An accelerometer would "wake up" the acquisition equipment when the chair was moved or jostled.
After completing the data-collection phase, Manta engineers analyzed results to define the severity of each user based on a fatigue-life model. They came up with a spreadsheet of the user's physical information (sex, height, weight, and occupation) and an indicator of the fatigue sustained by the chair in a week. This let the manufacturer quantify durability requirements for office chairs based on customer use.
In the second portion of the data-reduction phase, Manta and Herman Miller engineers developed an accelerated laboratory bench test. It simulates use by applying varying loads to different areas of the chair. Dynamic loads are applied and repeated until the required damage has accumulated. The magnitude, direction, number of repetitions, and order of the loading events correlate to data from the CUP program. Using this test, Herman Miller engineers now simulate years of use in a matter of weeks. This helps verify new designs, modifications, and manufacturing processes so their chairs perform to company durability standards.