Temperature Compensation

Temperature variation is one of the most significant sources of gaging error. As manufacturing tolerances get tighter and the margin for gaging error gets smaller, it becomes an issue that must be addressed.

Most materials expand as they heat up. For every inch of steel, a 1oF increase causes expansion of approximately 6 microinches. For brass and copper, the figure is 9 microinches, and for aluminum, 13 microinches. If the objective of the inspection process is to determine a part's true size, its temperature must be known. Based on the ISO's very first standard (ISO 1 issued in 1931), that temperature is automatically assumed to be 68oF (20oC).

But few inspection processes monitor, much less attempt to control, workpiece temperature. Many quality managers assume that any thermally induced variation in the part will be matched by like variation in the gage and the master: everything will expand and contract at the same rate, and everything will work out just fine.

This is far from true. Gage, master, and part—the three hardware "components" of a gaging system—may be of different materials, so the effects of thermal expansion will differ even if they're all at the same temperature. And the components won't necessarily be the same temperature. Parts that have recently come off a dry machining process may be several degrees warmer, and may remain so for hours. Parts machined under coolant may be cooler. The gage or the master might be sitting on a bench in direct sunlight, or under a heating or cooling vent. Temperature stratification within a room may create temperature differences between components placed near the floor, and components placed on a high shelf. The relative masses of the components may make a difference. For example, an engine block may take longer to reach equilibrium with ambient temperature than a bore gage. And in some instances, thermal variation may work in opposite directions for the gage and the workpiece, compounding rather than canceling the error. For example, high temperatures will cause bore gage contacts to grow longer, which will naturally result in ID measurements that are smaller than actual. On the other hand, the ID of a thin-walled part, like a bearing shell, will grow larger with higher temperatures.

These errors can be significant. As an example, let's use an aluminum part with a critical dimension of 4.0000 inches, and a steel master to zero the gage. The shop is hot today, but both the part and the master are at equilibrium at 80oF (or 12oF above "standard"). Master and workpiece have expanded as shown:

Steel Master: 6µ by 4 by 12 = 0.000288 inch

Aluminum 13µ by 4 by 12 =

Workpiece: 0.000624 inch

Error caused solely by the different coefficients of thermal expansion for the different materials is 0.000624 inch - 0.000288 inch = 0.000336 inch.

Now assume instead that shop temperature is a perfect 68oF and that master and gage are both at ambient, but the workpiece just came off the machine, and it's 80oF. The entire 0.000624 inch variation in the part will show up as measurement error.

Some companies attempt to control this problem by trying to control the environment. Essentially, you would have to turn the shop into a controlled lab environment—and that would be really expensive.

A better approach is to measure the temperature of the part, master, and workpiece, and compensate for thermal variation based on the known coefficients of expansion. This is now practical on a production basis using special devices like those from Albion Devices, Inc., (Solana Beach, California), which interface with electronic gaging systems. Typically, two small, industrial-hardened sensors are installed on the gage: one to measure the temperature of the gage itself, and one to measure the workpiece or master when either is staged. The system can be programmed for different coefficients of expansion of the various components, and the results are fed into an algorithm which generates a temperature-compensated measurement result on the gage readout. (Additional compensation factors may be built into the algorithm, to correct for unusual elements in gage geometry, differences between a workpiece's surface and interior temperatures, and similar variables.) Such a system will typically reduce thermally induced errors by 90-95 percent. In our second example with the aluminum workpiece, that would bring the error down to 62 microinches or less—a figure most shops can live with.