Information Equals Dramatically Improved Results
To Fix the Problem, First Get the DataIn many kiln troubleshooting missions, an inordinate amount of time is spent trying to figure out the details of specific problems. For example, when troubleshooting product defects, it is important to know where and when the piece was fired, its setting location, etc. Pinpointing how the kiln relates to the problem is always a challenge, but the job is easier when direct comparisons can be made regarding different firing circumstances. To use an elementary example, when a product defect arises in a single kiln, one obviously concentrates on that particular kiln. Alternatively, if a problem suddenly occurs in all kilns, we look at the common processing traits from raw material properties through firing. But what happens when the defect is associated with only a few kiln cars, or kiln cars of products on a sporadic basis? Or suppose the problems are concentrated in one position of the car setting area?
While this detective work is interesting, it takes time away from correcting the root cause of the problem. In most facilities, very little statistical data is available to pinpoint the defect quantity, and only vague references exist to the frequency of occurrence and location of the problem.
Life in a ceramic factory consists of coping with variables—defined and ill defined—but we often discover what causes defects by comparing different operations. For example, the defects in sanitaryware production often vary from kiln to kiln. Knowing what makes a defect occur is important, and being able to relate it to a particular firing condition is crucial. Multiple, well-monitored kilns can be used to discover the origination of firing defects scientifically, by analyzing the differences in atmosphere conditions, circulation and temperature. Without multiple kilns, divining these subtleties would be nearly impossible.
Textbook Quality and Process ControlExperts tell us that timely information is the best process management tool. When solving problems, the trail gets cold easily; problems are a lot harder to solve “after the fact.” Also, when you know about problems as they are occurring, you can correct them more rapidly, with much less ware lost during the correction process. Finally, people are always more motivated to solve current problems so they can have the satisfaction of seeing the results improve as quickly as possible. It is never as rewarding to fix missed defects found in the warehouse as it is to work on the process and see result improve instantaneously. Despite these principles, very few ceramic manufacturers know how to set up the proper process control system.
The Edesa Facility ApproachThe moment—yes moment!—the ware emerges from any of the several kilns at Edesa, several things happen immediately:
As kiln cars exit the kilns, operations people already know whether a defect is occurring and the frequency of occurrence, and examples are available to everyone in the factory for inspection. There is no “lag time” for discovery. Through cross-referencing, product output in shape, color and quantity is known minute by minute for inventory data. Furthermore, the kiln supervisory control and data acquisition (SCADA) system gathers the temperature profile so that the precise firing condition for each piece can be reviewed as necessary. This early warning system pays dividends in fast response to defects.
Rapid ClassificationPieces unloaded from the kiln car are placed on a conveyor for transport to the sorting and inspection area. This first in/first out system results in complete inspection and full process classification within 10 hours (maximum) of the kiln exit.
In the sorting and inspection area, employees complete the final inspection, tabulating all data for manufacturing—including casting dates, casting operative, drying data, etc.—and checking product aesthetics in detail. Defects from the standard are noted, and keypad entry of all parameters is made. This information is also sent to the network, and cross-references are automatically made to define the piece history.
Kiln Process Improvements Using the Data SystemAdditional efforts are made to maintain appropriate control points in each of the kilns. Aside from the normal temperature limits applied to each temperature and pressure monitoring point, kiln oxygen and pressure curves are evaluated frequently. Traveling thermocouple curves are generated and analyzed in detail.
Despite these efforts, defect aberrations occasionally occur and often arise out of variations in raw materials. Kiln control requires the largest possible margin for firing products that occasionally have marginal properties, and this sometimes means shifting kiln control points in special situations.
For example, Edesa has had some difficulties with cooling cracks (dunting) of certain shapes. From the data system described, we already knew the frequency of occurrence and the principal setting locations where the defect occurred, and we were able to eliminate other production variables. Analysis of all kiln curve parameters showed them to be in control and normal. Accordingly, cooling rate changes were necessary to eliminate this problem. But adjusting a large sanitaryware kiln cooling system is often risky. It is quite easy to go from a small dunting problem to a very large one in a matter of hours. Furthermore, detecting this type of crack is difficult at times—a skilled inspector may have to die test the pieces to determine the true loss percentage. Further downstream impact—in product refire, for example—might not show up for weeks.
Unfortunately, we only had a short time to make this improvement, but the rapid data feedback made the solution relatively painless. After analyzing the raw material through DTA (differential thermal analysis) testing, a new cooling curve was designed, and gradual, continuous adjustments were made over a period of only three days to reach our target curve. Production results were monitored essentially hour by hour, and the influence of our progression of adjustments was readily seen, essentially in real time. Under these circumstances, a series of adjustments normally requiring a period of weeks was made in a very short time with relative safety to the ware.
Fortunately, our kiln manipulations were quite successful and resulted in the elimination of the problem. But if we had made any missteps, the information system would have told us almost immediately, and a recovery could have been made easily with very little negative yield impact.
Commissioning a New KilnIn 1999, a new Swindell Dressler kiln was commissioned, along with a full automation system for car movement and management. The benchmark for accepting the new firing system was that the performance must be equal to—or better than—the existing tunnel kilns in the plant. Here, again, the data monitoring system provided the appropriate criterion for acceptance. The information was also beneficial to the Swindell commissioning team, as crucial adjustments could promptly be made and tracked. The kiln was placed into operation very rapidly post-lightup. Having the correct information in a timely manner clearly shortened the required timetable.
Continual FeedbackProduct improvement and defect elimination is hard work. However, by putting certain tools in place, a manufacturer can enhance the speed of defect elimination, and even more importantly, provide the necessary continual feedback so that yields can be maintained at a very high level with no “surprises.”
A further benefit of the system described at Edesa has to do with the human factor. People like to see the results of their work as promptly as possible. Seeing the results of hard work in near real time provides special incentive and extra satisfaction to employees. On the flip side, because defective ware is displayed prominently in front of the kiln, and the data can be accessed as developed, there is nowhere to hide. “Out of sight, out of mind” does not occur at this facility. The results include higher continuous yield, consistently lower costs and enhanced employee satisfaction.