KILN CONNECTION: Understanding Variability
April 1, 2011
"Variability is the enemy of quality." This has become the battle cry for manufacturing excellence in many facilities that I visit. After engineers measure the range of process parameters, they analyze the required operational changes necessary to narrow the range and standard deviation of specific values. Control limits are established, and considerable effort is expended to minimize the variations.
But like the many programs for quality, consistency, and yield that have come before it, the process of measuring variations and subsequent discussion is often confused for progress. A surprising number of plants spend significant funds in an effort to improve processes through adjustment but, after a few years, the resulting progress is limited; yield improvements are modest, at best.
Variation TypesQuality experts separate variations into two types. The first type is common cause; these are variables inherent within the process. Hundreds of such variables can exist, and they generally cannot be eliminated without changing the process in some significant manner.
The other types of variables are defined as special cause, and this class of variable is far more discernable: something happened within the process, often abruptly, that caused properties to deviate widely. These kinds of variations normally provoke instant action and firefighting in order to determine "what went wrong." Frequently, a ceramic factory will spend much of its time trying to eliminate the special-cause variables through the establishment of control charts, standard procedures, etc., while the common-cause issues are noted-but rarely addressed.
Fixing special-cause problems is often easy; better monitoring, adjustments and procedures usually allow things to resume 'normal operation.' Sometimes, however, it is the common-cause variation that must be addressed before lasting and continuous improvements can be made.
Fixing Common-Cause VariationsAs an example, take a look at the Datapaq curve shown in Figure 1, which displays the temperatures in the preheating zone of a tunnel kiln in several locations on a single kiln car. In the case of this kiln, the differences in temperature and the relative heating rates are completely location-dependent. The top levels of the kiln car are hot, and the bottom levels and car interior are cold.
For this particular kiln, no amount of adjustment will fix the problem. Occasionally, losses will occur due to cracking or insufficient oxidation; these problems are usually caused by heating that is either too fast or too slow in certain locations. Such high losses can and do result in futile kiln adjustments until the yield problem goes away. Instead, the focus should be shifted to the common-cause problem root-poor temperature uniformity-and the kiln's design should be modified to eliminate the problem.
This is a management function (not an operator's responsibility), yet few companies address these inherent problems. Fixing common-cause problems will often provide the largest window for successful firing, even when upstream variations are a little out of kilter. The goal in kiln operations should be to provide the best chance of success, even when the greenware coming to the kiln might be imperfect.
Variability truly is the enemy of quality. Recognizing the difference between inherent common-cause problems and special problems-and addressing both-can make the difference between continuous improvements of quality/yield, and the standard firefighting of special problems alone that does nothing but fix the occasional bad yield issue.