In my two previous columns I discussed specification limits and process control limits, which led to thoughts about product disposition limits.
Along with process control limits, product disposition limits are arguably just as important to manufacturing operations. The role of disposition limits, however, is significantly different, but many organizations confuse them.
Fundamentally, disposition limits are focused on product, not process, control. The decisions they drive are focused on what to do with product that has already been processed through a specific process step or set of steps. The basic decision involved is whether a specific group of product should be allowed to move on for further processing and eventually become finished product worthy to be shipped.
It might sound sophomoric but product following within suitably determined disposition limits would result in the lot moving forward for additional processing. However, product falling outside disposition limits would result in the product delegated for disposition which include scrap, rework or derate for another purpose.
Specifically disposition limits differ from process control limits in three areas.
Disposition limits are applied to a finite group of product that has already been manufactured. Control limits, on the other hand, are applied to the manufacture of current and future operations of a process for variable amounts of time and processed product.
Disposition limits are focused on product control to minimize overall producer and customer costs. Control limits are focused on process control and are ideally determined by appropriately balancing false signal rates with required levels of sensitivity. As Dr. Walter Shewhart, the inventor of the control chart, stressed managing with process control limits requires resolving special cause variation to maintain control.
Disposition limits and process control limits differ in the amount of risk they impose on a manufacturing operation. It sounds strange but the risk associated with determining the fate of a finite lot of product outside the appropriate limits is often perceived as much less than the risk of determining the fate of the associated process. Something has to be done with the product that has already been produced outside the appropriate limits but that decision is only applied to that finite lot. However, adjusting a process will potentially impact all future product through the affected process step.
Because of the inherent differences, and confusion, between product disposition and process control, it can be desirable to separate these functions as much as possible. Many quality managers prefer to separate process control engineers and product quality engineers, at least in part, for this reason. Some organizations fail to do this which can create situations in which the statistics used to drive both functions are not only the same value, but frequently have identical limits.
Where organizations do not separate these functions some make it work but it can lead to one function dominating the other. Typically in those situations it is the process control function that suffers because the product control decision is necessary and carries less risk, or perceived risk.
These two functions can be decoupled by using different limits or, more desirably, using different statistics to drive the decision-making for each function.
One consideration is when a sequential control statistic (for example, a run rule, an exponentially weighted moving average or a cumulative sum) can be used for process control, while the periodic sample average can be used for product disposition purposes. These are related statistics, not the same, but they would have different sets of limits.
What would Drs. Deming and Juran suggest? To my knowledge they did not address this issue so it is anyone’s guess if the two greatest quality giants of the 20th century would support this approach. Juran would likely suggest doing what made sense and that would deliver effective results. Deming would likely be more inclined to support reliance on simple process control approaches.
From my experience, most manufacturing managers are reluctant to discard the perceived safety net that disposition limits provide. Disposition limits, even misunderstood, are likely to be part of any manufacturing operation.
In summary, for manufacturing perspective specification limits are primarily for interactions with customer and management. Process control and disposition limits are more directly related to daily decisions made in manufacturing operations than are product specification limits. Control limits drive decisions on process operations (to adjust or not adjust). Disposition limits drive decisions on product management (scrap, rework or derate).
Do you—or more specifically does your organization—know your limits? Many organizations continue to operate in relative uncertainty related to their routinely used limits and thus endure daily arguments, frustrations and inefficiencies.