The drive for precision has reached a high point among the six automatic molding machines of TKA-Waupaca’s Plant 1 in Waupaca, WI. From this small industrial town west of Green Bay and north of Madison, come some of the most dimensionally precise gray iron sand castings produced anywhere.
Plant 1 produces brake rotors and brake drums, transmission housings, hydraulic pump housings, flywheels, intake and exhaust manifolds, cylinder heads, crankshafts, pulleys, brackets, and covers for electric motors. These castings go into cars, SUVs, trucks and buses, as well as construction machinery and farm equipment.
With respect to inspection measurements precision is held to 0.002 or 0.003 in., a span smaller than the grains of sand fed into its molding machines. There are four primary reasons why Waupaca’s inspection teams work to these levels:
- Customers have automated the machining of castings they get from TKA-Waupaca. In those operations, there are no human machine tool operators to adjust tool offsets or tweak NC programs. Without those standards, very little part-to-part variability can be accommodated in machining.
- Many parts are used by customers as-cast. That is, as received from the foundry and with little or no machining.
- Tight tolerances reduce the time castings spend in grinding operations. This speeds deliveries and saves labor.
- All dimensional inspections of molds are measured from known reference points on the molding machines. This ensures that every mold’s ram and swing halves fit together exactly. It also ensures that every mold is properly positioned and aligned on the conveyor that carries it to the pouring area. Alignment for pouring controls what was traditionally a cause of variability among castings in a job lot.
Tracing the Root Causes of Variability
It is not enough, in other words, to ensure that the molds are dimensionally correct in and of themselves (i.e., measured only in reference to the individual mold’s coordinates). Instead, the precision demanded by TKA-Waupaca’s customers requires that molds be checked against the tooling that produces them.
If the molds were measured just by themselves, some of these tolerance errors might cancel each other out. The dimensional errors would still exist, but they may be harder to find. More importantly, the errors would be much harder to trace back to root causes in the molding and coresetting machines, such as ram or swing arm maladjustments or bushing wear.
This can only be done by verifying that the pattern is correctly pressed into the sand as each mold is produced and that the core setting machines are running correctly. Once dimensional relationships are established among patterns, cores, and key components of the molding machines, molds can be checked by reference to known points on the molding machines. This adds to the need for precision because in linear measurements such as these, tolerances stack up.
If molds and cores are removed from the production line and carried to a conventional coordinate measuring machine (CMM), there is no verifiable way to address problems that occur within the molding machines or core setters themselves. Symptoms can be identified, but that’s all. Once a mold or core’s dimensional relationships are severed from the molding or core-setting machinery that produced it, there is no straightforward way to translate dimensional discrepancies into actions to be taken by the tooling maintenance people.
Taken together, these factors are why TKA-Waupaca Plant 1 engineers, tooling experts, and machinery maintenance people now do most of their dimensional measurement with a portable coordinate measuring machine in one of the hottest, noisiest and most confined areas of the foundry. That’s between the automatic molding machines and the ladles that pour metal into the molds.
Plant 1 is a high-production, and thus fast-paced, operation. Each of its six automated molding machines makes about 300 molds per hour. That’s five or six molds per minute, which works out to a new mold every 10 or 12 seconds per machine. From one to six parts are produced per mold, and job lots are always more than 100 pieces.
Maintaining Volume and Precision
It’s a complex task to maintain a rate of casting production while casting to high standards of uniformity and low tolerances. To get the job done, Plant 1’s layout department uses PowerINSPECT dimensional inspection software from DELCAM Inc., Windsor, Ontario, Canada, and a portable CMM from ROMER Inc./CimCore, Farmington Hills, MI.
“Customer demands pushed us to use the portable CMM and software we chose,” reported Steve Buck, Plant 1 layout engineer and the primary user of the Romer-PowerINSPECT system. “Customers insist we maintain tight control over the way the patterns are pressed into the sand in the molding machines. When we look for root causes of variance, we go all the way back to the pattern that makes the impression the mold’s sand and start measuring from there. We also have to make sure that the cores are set correctly, that no core shifts occur, and that there is no shifting between ram and swing halves of the mold,” he added.
“That could never be done with a conventional fixed CMM because the inspections always have to be referenced back to the machines. We have to work to common zero points on the machines,” he explained, “and not just to the mold.”
Fulfilling the demanding standards of the foundry’s customers is what drives the selection of the measurement systems in use. “PowerINSPECT allows us to generate complete, step-by-step reports to the customers for verification of inspections on production, sample, and first-time castings,” Buck added. “Once the point data has been electronically downloaded into a template, all the statistical process control and trending functions are available to us.” Templates are made with Excel spreadsheets from Microsoft Corp.
Waupaca is also using DELCAM’s automated report generator to create initial sample inspection reports (ISIRs) for Caterpillar Inc., one of their major customers. “We designed the report to look just like the forms that Cat requires,” Buck said. “This saves us a lot of time; before, we had to enter the data manually. This also eliminates a potential source of error since the inspection report data is downloaded right into each ISIR.”
“Our in-house troubleshooting measurements are very helpful for preventive maintenance,” Buck noted. “This includes identifying things like wear in bearings and bushings before they begin to cause ‘slop’ in the equipment and before we find inaccuracies in pattern setting and out-of-tolerance molds.”
Among the system’s many tasks are assisting the maintenance department with setting up automatic molders, ensuring that molds are square, that they do not shift, and that they remain aligned when they come out of the molding machine chamber. TKA-Waupaca often measures a mold’s parallelism and perpendicularity as it comes out of the molder, right before the molten iron is poured in. “Sometimes we even have to measure inside the molders for the pattern positioning and orientation relative to the machine’s ram and swing arm,” Buck said.
The Romer arm— a 9 ft. Model 3000i — is in constant use. To get the best measurements in and around the molding machines, they clamp the arm as close as possible to the machine’s swing arm and its actuating cylinders — sometimes right to the bed plate of the machine.
One major consequence of the CMM’s portability is that the foundry’s measurement and dimensional verification has moved almost completely to the production floor.
Steve Buck has also been pleased with the portable CMM’s durability. “The arm is often used within 10 or 15 ft. of the bull ladle full of molten iron, which is over 2,600° F.,” he noted.
One reason for the arm’s durability and performance stability in harsh conditions is its carbon-fiber composite arm. “We know it will be dimensionally stable no matter how close we get to those ladles,” Buck pointed out. “The system is impervious to foundry dust, too, and is unaffected by the slight vibration from the molding machines.”
Pattern Shop Use
The measurement equipment is also used in the pattern shop for checking cores, core boxes, and patterns. This is much faster than the previous practice of bringing them into the layout department to the conventional CMM. One reason this has speeded the procedure is that core boxes and patterns are heavy, awkward, and sandy/dusty. Engineering changes are measured both with and without CAD.
The new measurement system was “an easy choice” for other reasons, too. Buck cited:
- The ability to change probes on the fly without having to recalibrate the arm.
- Better measuring tolerances.
- The arm’s light weight and counter-balanced design.
An additional advantage of the equipment is that conventional CMMs require that every measured point be programmed, a tedious and time-consuming task. PowerINSPECT accepts data from manually selected and clicked points. (Both types of CMM compare points’ actual locations in 3D space against the nominal locations from downloaded CAD files, then plot out the differences.) Layout work on the production floor was done with Verniers, a type of caliper used to measure fractions of larger dimensions.
Customer demands for the best possible as-cast surfaces drove TKA-Waupaca to try the highly sensitive PowerPROBE, also from ROMER/CimCore. This probe tip requires almost no pressure to register a point, whereas the pressure needed to register a point with conventional probes can leave a detectable dimple in the surface of the mold. In the as-cast part, the dimple produces a detectable bump — tiny but still unacceptable in some applications. Probes currently used include a 15 mm ball, an 80 mm ball with a 6 in. extension, and a 20 mm ball with a 4 in. extension.
Ease of Use
Ease of use is a big issue for Waupaca. “None of us using the system now had any prior experience with CAD layout,” Buck said. “At first, we found that laying out jobs without a knowledge of CAD was very intimidating. Yet the arm and the software were easy to understand and very easy to learn use effectively. ”
Keeping the layout process simple is important for another reason, too. Blueprints and plots from CAD files are no longer routinely supplied with the jobs to foundries. Now the foundry is expected to work from downloaded CAD data derived directly from the customer’s math model. Issues of data traceability aside, consistently holding 0.002- or 0.003-in. tolerances working from plots and blueprints has never been easy.
Buck has worked 18 years in foundries, starting in maintenance and reaching the layout room only after long stints in production and grinding. “All the people doing the layouts and inspections here have worked their way up through foundry operations,” Buck noted.
Ease of use has a physical benefit as well. Measurements made directly from the molders often require working in cramped spaces. To ease the burden, Waupaca purchased a stand for the measurement arm. “The stand provides a lot of additional stability and it has a shelf for the laptop, too, which is very convenient in tight spots,” Buck noted.
The “acid test” of dimensional measurement is repeatability. Buck explains: “One day we were asked to measure a bolster plate for one of the automatic molding machines. The machine operator was a real ‘show-me’ kind of guy. Bolsters are used when we have to run small patterns in larger machines. Bolster measurements are complicated. First, we have to level and align. Then, we measure through two bushings, measure a preset dimension to a round bushing, and then measure a series of planes on two different surfaces.
“Two days later, we were asked to come back to measure a large number of additional bolsters. The operator slipped the bolster we had already measured into the new lot. As he hoped, we didn’t notice it, so we measured it again — but the second time in different ways. Both sets of recorded dimensions came within 0.001 in. repeatability. This is amazing — especially considering the arm had been trucked in and out twice, and set up, torn down, and packed away twice.”
Edited from information supplied by Romer Inc./CimCore, Farmington Hills, MI