Statistical Process Control (SPC) is generally a Six Sigma tool. A few people and organizations will use SPC as a stand-alone methodology. Generally, it is not a tool found in the Lean methodology’s toolbox. That is because the process of conducting Statistical Process Control is considered non-value-added.
However, there is a common, problematic phenomenon that undermines the success of the Lean method for many of us. Our effort to eliminate non-value-added (NVA) process steps encourages us to develop a trustworthy process that does not need monitoring so we can stop the NVA effort to monitor it. Unfortunately, change happens.
Here is an example of how the phenomenon bites many of us. It should sound familiar. A business gets manufactured components for its finished product from a small list of tried-and-true, trusted suppliers. The suppliers’ quality processes and systems have been evaluated with careful scrutiny and the delivered components have been measured and tested until the business declared these suppliers worthy of the “preferred” or “partner” supplier list and on-going measurements and tests are no longer required.
One day, however, one of these suppliers, in order to compete with the others, or in order to increase capacity to keep up with demand or growth, or because one of its contracted services goes out of business, changes who it uses for the finish treatment of the manufactured parts. There may or may not be a mechanism in place that ensures our business of interest is informed.
The process that was qualified by the business is no longer the same process. What if the new process is not as reliable? What if the new process doesn’t meet the business’ needs at all? How would that business know it is assembling and shipping and selling defective goods?
Most businesses that adopt a this-supplier-doesn’t-need-monitoring practice still schedule occasional audits of supplier processes and goods in order to address this issue, but how effective are those schedules? How thorough are the audits and investigations? Do those audits include complete measurement and testing of delivered goods to ensure continued satisfaction of qualified specifications?
My observation is that those audits are generally more focused on providing an appearance of continued checks and communication of expectations than on thoroughly re-evaluating how goods are made and what performance is actually delivered. The audits have been “Leaned” to a minimum effort as well. Also, if the audit is an annual event, that gets postponed or delayed because there are too few auditors for all of the contractors to our very lean business, how many escapes might there be before the change is detected by an audit?
The problem of not knowing, of not being sensitive to changes in processes, is a problem that SPC is specifically designed to address. Its sensitivity to change is why it is a common Six Sigma tool. Let me give a very brief explanation so we can talk about how it can solve the problem already described.
The method of SPC uses regularly collected data to draw what is essentially a line graph of measurable output performance over time. The statistics play a role by comparing new data points with past data points to mathematically assess the probability that new data reflects the behavior of previous data. If it does not fit existing data statistically, then we investigate to determine what happened in the process that caused a mathematically noticeable change in process behavior or performance.
It can warn us of changes and trends before the changes turn into quality or performance problems where expected performance is no longer satisfied. Also, over the years, a huge variety of data collection methods and types have been incorporated so that virtually any type of output data can be monitored.
In essence, SPC can warn us of a pending problem, can predict a problem, before it occurs. Also, the statistical elements help keep us from over reacting to every little fluctuation in performance and making costly, unnecessary adjustments. The statistics involved are very basic, the formulas simple, and so we do not need to be mathematical or statistical experts to effectively use the SPC method.
So, let’s take another look at the example business described above. Suppose that the components it purchases from a short list of three suppliers must meet three critical criteria.
- Critical dimensions must be precise
- The material must be of particular strength and hardness
- The finish must be of minimum durability
In the scenario described, how would the business know that Supplier A is using a new material, that Supplier B has recently replaced some machinery and the process for cutting or producing the critical dimensions is different, or that Supplier C has added a new finish treatment subcontractor? Even a periodic audit by a process quality representative might not reveal or identify these process changes, but the output might easily be affected in each case. In fact, the output is guaranteed to be different, though it may not be problematic.
What if, however, the new and different outputs do cause a problem? How would the business know? The suppliers are all listed as “don’t inspect incoming material” suppliers because they have earned that rank and the quality certificates from the suppliers pass muster to guarantee that they have built according to print and contract specifications. The quality function will “rubber stamp” the assembled product and ship it, no questions asked, entirely unaware that a problem has been introduced.
However, if the business incorporated a practice of measuring and inspecting one piece of every supplier’s incoming shipment and keeping that record on a run chart according to the SPC method, the business would very likely know immediately when a supplier changed a process, machine, or subcontractor, and would quickly know if that change was a potential problem. Unfortunately, doing so means accepting a NVA cost and process step.
There are ways of practically minimizing that cost and process burden, and most businesses with a strong SPC practice have developed many. Back to our example, businesses that deal with critical dimension usually have a metrology laboratory and use it as necessary. It is not difficult to schedule and manage regular measurements of in-house or delivered components.
Likewise, that same metrology lab probably also contains the equipment for measuring Rockwell hardness and can collect that data at the same time as the critical dimension measurement. A check of the material certificate from the supplier could be made at the same time. That leaves just the finish durability to address.
Unfortunately, finish durability is not as quick and easy to assess. It requires a destructive test that is not performed in a few minutes. In fact, something like a salt-fog chamber test can take weeks. Now we are into the kind of non-value-added quality checks we are trying to eliminate within the Lean methodology.
Here’s how I like to advise addressing this kind of scenario. Pretend that the quality check is value-added. It’s not hard, actually, to rationalize.
Change is going to happen. Being prepared for it and actively planning for and preventing problems that are brought on by change is value-added because it prevents defects. Therefore, proactively monitoring processes for change is value added. Therefore, the salt-fog test is value added.
I can hear the arguments and anticipate some comments. It’s a rationalization, but it’s also, I argue, not impractical and not entirely outside the intent of the Lean methodology. Consider the business trade-off. An undetected problem might cost the business a great deal more than the monitoring process will cost the business. Yes, there is a business decision that must be made between “might cost” and “will cost.” That’s business.
I reiterate; change is going to happen. The cost of monitoring very well might be more value-added than non-value-added. The cost of monitoring is certain to be less than the cost of a defect that goes undetected for a long time.
The trick is to apply the Lean methodology to the process of monitoring the critical performance parameters with an eye of removing waste from the monitoring process rather than removing the monitoring process. It’s a very important distinction.
In the example, the metrology equipment and skills can be easily scheduled or otherwise “processed” in a very Lean way to make it as efficient as possible. Likewise, an eye for Lean process can quickly and effectively organize equipment and activity for the salt-fog chambers around a demand that balances on-going quality monitoring as well as new process and product development demands.
The paradox is this. Ongoing monitoring and quality checks of process outputs are generally, non-value-added, so as Lean practitioners we try to eliminate them. Unfortunately, things change. Changes increase the risk of defects, and without those non-value-added quality checks, we won’t catch them.
So, to address the paradox, we must decide which waste is more acceptable, potential defects, or guaranteed NVA quality checks. We can argue that preventing defects is value-added, not NVA, albeit that is only true if our quality checks do indeed prevent defects, and so the argument goes in circles.
We can effectively reduce the NVA cost of quality checks by applying the Lean methodology and methods and tools to the quality check processes, which makes it easier to swallow. It’s a matter of accepting the necessity of the processes instead of accepting their elimination.
If we decide to do regular quality monitoring, we don’t need to use the statistical methods of SPC. However, I argue that as long as we are collecting regular data points, why wouldn’t we get the most out of our efforts? SPC can identify changes we aren’t aware of, can sometimes predict the production of failures before they happen, and aid us in increasing the efficiency and effectiveness of our processes.
If the balance of value-added and non-value-added work is the root of the issue, then SPC is the most value-added way to conduct what is generally considered non-value-added quality inspection. It’s also not that hard to do once the habits are ingrained.
Look at your Lean program’s processes for monitoring on-going process performance. If your organization is running in a danger zone because apparently NVA monitoring processes have been eliminated, consider the risks associated with that behavior. Work with your peers and leaders to question the balance of risk and cost. Consider if some on-going monitoring isn’t potentially worthwhile.
If you do on-going monitoring, or decide to add some back in, give serious consideration to the SPC method of process monitoring. Because of its ability to predict problems before hand, and its discipline for preventing unnecessary process tampering, it is a very effective tool for remaining Lean in our efforts to monitor outputs and control processes.
Stay wise, friends.
If you like what you just read, find more of Alan’s thoughts at www.bizwizwithin.com
Filed Under: Rapid prototyping