A Practical Approach To Automating Data Collection
When they first got their automatic data collection system online, this Indiana moldmaker thought they'd have to charge customers to document parts measurement. Instead, the system has saved them time and money.
By CHRIS KOEPFER, Associate Editor, Modern Machine Shop
Merkler Machine Works, Inc.(Fort Wayne, Indiana) is a 35-person job shop that specializes in making molds for extruded rubber products, primarily used for automotive weather stripping and hermetic seals on appliances. It also designs and manufactures special machinery for material handling applications. Although they've grown the special machine businesses, the shop's bread and butter comes from mold making.
Until about two years ago, the company made and shipped molds to its tier-one suppliers with little more than reputation to assure the molds were right. They might include a shop-stained sheet of paper on which dimensional readings were recorded but that was about the extent of documentation. And it was enough for their customers. That is until one of Merkler's customers, responding to their own customer’s demand, issued what amounted to an ultimatum: "Find a way to document your measurement of our parts or lose our business."
 That'll get any job shop's attention. It motivated Merkler to begin a process that has already paid dividends. Working with their customer to define specifically what was needed, Merkler began shopping for a data collection system that would satisfy their customer's current need for documentation — and provide enough flexibility to allow Merkler to grow the system as needs change. That led them to L.S. Starrett Company (Athol, Massachusetts).
Long known for its gages and instruments, Starrett has a product line that combines computer hardware and software with gages, to help shops starting from scratch get into automated data collection. It's a building block system designed to allow a shop to grow into quality automation at its own pace. Merkler found it fit their needs.
Customer Driven Supplier Chain
Almost everyone works for someone. For its mold making business, Merkler serves about 30 plants representing four major customers. "When one of our major customers gave us an ultimatum in effect to provide documentation in the form of reports verifying that specified mold dimensions were checked, or we'll pull our business, we took action," says Loren Bergren, plant manager.
With quality paramount in virtually every industry—automotive, aerospace and appliances for example—the drive to document process procedures is pervasive. As this documentation drive spills over from one supplier tier to the next, some businesses must take action or lose work. Merkler chose to take action.
Why Automate?
On the surface, it might sound like automatic data collection is designed for shops doing sophisticated quality programs. Certainly it can dovetail into SPC or other analysis systems, but that's not a prerequisite for automating data collection.
Actually, any shop that measures dimensions on workpieces can probably benefit from automating data collection. Even if the data go no further than a file in the shop office, taking variability out of the collection process insures that the data are correct.
Many jobs shops, Merkler included, have shied away from SPC because their lot sizes are too small to require sampling. Not having to measure all of the parts in a run is one big advantage to doing SPC. It allows statistical conclusions to be drawn about how a process is working without the need to sample the entire run.
 Merkler and many shops like them run small batches (under ten pieces) and can easily measure 100 percent of the parts. They chose to automate data collection, and measure all of the parts in a run, rather than apply SPC to their small quantities.
Peace Of Mind
Merkler's mold making operation deals with small quantity runs, usually under ten pieces, they can readily check all of the parts being made. A given mold may have six or seven critical dimensions called out for measurement by the customer. These dimensions are highlighted on the part print and each of these must be inspected. Further, since the molds are multi-piece, Merkler also specifies some dimensions be measured even though the customer doesn't require it.
"The customer is usually concerned about the dimensional accuracy of the mold cavity and focuses its measurement requirements there. But because each rubber mold is built from pieces, sometimes as many as seven, each mold part must fit precisely with the next. Without close monitoring, error buildup can ruin an entire mold set. We've learned to also take measurement of other mold dimensions outside the cavity that can impact assembly of the mold pieces, and automation of data collection has sped up that process,'' says Mr. Bergren.
One of the biggest gains for Merkler since implementing the automated data collection unit has been peace of mind. The system is very close to fool-proof. As part of their quality control procedure, Merkler's customers will randomly check the specified mold dimensions on a shipment.
Now, if a problem arises with a mold that has shipped, 99 percent of the time it turns out the customer is doing something wrong at their end. The documentation generated by the automatic data collector confirms the measurement procedure followed and the gaging used. "Usually the customer has not duplicated one of these items in their test," says Mr. Bergren.
To Err Is Human
That confidence in the accuracy of measurements wasn't always the case. With hand-recorded dimensions, mistakes happen. "Nobody makes mistakes on purpose,'' says Mr. Bergren, "they just happen. Our data collection system has come very close to total elimination of recording errors.''
There are several potential problems when an operator is responsible for recording recorded dimensions. First of all is the operator's handwriting. That may sound simplistic, but poor readability can promote errors. Operators may write the wrong numbers or transpose numbers. And because the recording is done at the machine, shop contamination can spoil the paper - "Is that a coffee stain or a decimal point?''
When the recording sheet comes back to the office with the completed job packet, the data may be transferred to ship with the parts. With each handling of the data, numbers can get transposed, zeroes may get dropped, and wrong guesses made about what the operator actually wrote. With several transcriptions of each dimension, there is potential for error.
How Does It Work?
Starrett's system eliminates the potential for mistakes by taking the human element out of the data stream. In operation, the operator still gages each specified dimension on each part but rather than write each reading on a separate piece of paper, the gage is wired into a data collection unit.
The reading from the gage is recorded directly on a portable memory card that replaces the paper record. When a job is sent to the shop a standard 64K memory card goes with the job packet. On it are measurement procedures for the job being produced. The tolerance dimensions are called out on an LCD screen on the hand held data collection unit. The operator is even instructed as to which gage is recommended for best access to the prescribed part feature.
 The operator gages the parts and each dimensional reading is recorded directly from the gage. When a job is finished, the packet, complete with memory card, is sent to the office. The data collected from the gage is loaded through a card reader into a standard PC. The PC is running; a software program designed to capture and chart the data.
At this time, Merkler does not use the SPC function built into the software but plans to begin charting some high volume jobs coming in the near future. "There's no point in charting the data from our short run molds using SPC," says Mr. Bergren. "But it's there when we need it. Because our shop has become so measurement conscious already, it's a very short step from what we're doing now to charting SPC data with the help of this software.''
For Merkler's immediate needs, simply getting direct data collection has more than paid for the system. With the software program, reports can be generated that fulfill the customer's need for certifiable data. The reports not only indicate that a specified dimension was checked, they give exactly where each reading fell relative to the upper and lower limits of the tolerance band. "Our customers are delighted,'' says Mr. Bergren.
As word of the automated system has spread among potential customers Merkler has landed several jobs based on the their ability to supply documentation of their data collection. The pressure to create pass-through documentation is pervading many industries, especially automotive. "For us," says Mr. Bergren, "it's been a competitive advantage.''
Value Added
"Not knowing how the system would impact our operation, we planned at first to charge customers for this service," says Mr. Bergren. "It quickly became clear," he continued, "that the savings in paper and handling of the data was saving the shop money. We were able to actually lower quotes on jobs that used the data collection units." Now all mold measurement is done using the data collection system.
Starting Small
Like many job shops, Merkler has a lot of capability to handle the variety of work that comes through the door. Over 80 machine tools dot the shop floor. These machines are divided into seven departments to simplify routing of work through the shop.
"The departments that finish the molds were the initial focus of our data collection efforts," says Mr. Bergren. "In our shop that's the CNC and EDM departments."
With the success of automated data collection in the first two departments, Merkler plans to implement it throughout the shop. "Our plan," says Mr. Bergren, "is to place a data collection unit and appropriate gages in each department. And when we do that we'll link all of the units together to directly input the data into the front office computer—bypassing the memory cards."
With the Starrett system, Merkler can upgrade the amount of automation it needs. Data collection can be confined to remote collectors, like Merkler's phase one. Or the remote collectors can be wired into central data collectors that can then route the measurement information from the shop floor to the office PC.
Because the software on the PC is set up f-`or statistical, the shop can automatically run charts and provide SPC data to customers who want it. "That's the beauty of this system," says Mr. Bergren, "the components are independent of each other and allow us to use as much or as little automation as we need."
The Unexpected
There were also some unexpected benefits for Merkler derived from automating data collection. One of the ways is setup reduction on the CNC department's machining center. Prior to introduction of the data collection unit, a test block was used to verify the part program. The cavity was cut into the test block, then measured, and the machine comped as needed to meet any difference.
Now, rather than cut a test block, the operator roughs the first mold cavity on the machining center and measures the cut. The gage is set up to allow for the difference between the rough cut and the finish cut. Again, the machine is comped to make up any difference. Savings comes from elimination of the need for a test block.
Small Steps Add Up
Merkler has taken the necessity of measuring critical workpiece dimensions, and with a nominal investment turned parts checking into an asset for the company. Data collection units like Starrett's start at approximately $2,000.
Automating data collection can be a short-cut to streamlined data collection efforts. And if done in small steps, like Merkler did, will cause a minimum of disruption while providing a springboard for expanding quality analysis and documentation.
Condensed version reprinted from the April 1995 Modern Machine Shop. copyright 1995 by Gardner Publications, Inc., 6600 Clough Pike, Cincinnati, OH 45244-4090.
|