Q: When we have a parameter like flatness in a part, the lower control limit is zero. When we create a control chart, upper control limit and lower control limit both must appear in the chart. How does one make control limits for such one-sided parameters?
Krishnan P M
A: Your question of how to interpret and construct the lower control limit in the case of the flatness can be answered in two ways. But to do so, we need to answer two questions. First, which control chart should be used in this case? Secondly, what meaning can we assign to the lower control limit?
The flatness measure is a measurement of the deviation from a perfect surface. This measurement is usually represented by the standard deviation of the data gathered. So, the distribution of the flatness measure is not usually a normal distribution. A perfect flat surface should have a flatness of zero, as you mentioned in the question.
Since flatness is a standard-deviation measure, a control limit customized to the actual flatness distribution should be used. If the actual flatness distribution is used, an upper control limit of 1%, 5% or even 10% can be identified.
The interpretation of a part's being outside either control limit is similar to the regular interpretation in statistical process control (SPC). There should be no lower control limit problem, unless you are trying to find out which batch is an unusually good batch.
If an X-bar chart is used in a group of flatness measures, the out-of-control point means the flatness measure either is too large or too small compared with the flatness average measure—that is, it might have a shifted mean for the new group of data. A part that measures out of control on the low end can be interpreted to mean you have unusually good surfaces compared with what you had before.
If you use the regular X-bar control chart, a part that is deemed out of control is interpreted the same way as it was taught in any SPC class. In this case, out of control on the low end is good, and you should determine the cause. If you can find the cause, you should improve your process based on your findings, and the control limits can be revised downward.
Shin Ta Liu
Principal consultant, Lynx Systems
Q: What is the difference between corrective and preventive action?
A: The difference in corrective and preventive actions can be subtle and is a frequent sticking point among people new to the quality field. But there is a significant difference between them.
If a situation is causing a failure or any other issue in the process, then something needs to be done immediately and therefore is a corrective action because you are correcting something that is not working properly.
For example, if you receive complaints from customers that the relief valve in a process is not rated for the right pressure or a capacitor for a circuit is not rated at the right operating voltage, corrective actions need to be put into motion right away. It's very likely that a temporary solution will be applied, and a more in-depth analysis will be performed to determine the root cause.
There are cases, however, when you don't have an issue at all, or when a solution can be implemented later. The process is not as efficient as it could be in that case, but it's still functional, and there is no immediate risk.
It might have been operating that way since day one, but a more efficient, cost-effective or safe mode of operation exists. Some of these actions start as customer suggestions or the recognition of an opportunity for improvement by someone inside the company.
For a side-by-side comparison of the significant differences between corrective and preventive action, see Table 1.
David Bonyuet, CQA, PMP
Q: We are in the business of providing printing solutions to individual end users, small and medium businesses, and large accounts. We develop, manufacture and service laser and inkjet printers. How would I go about creating a best practice value stream map that will address and enhance end user/customer experience?
A: Addressing and enhancing the end user/customer experience is a broad goal that calls into play everything your company does. It will likely require you to invoke more tools than just a value stream map. But you probably realize that, so I'll focus solely on developing a best practices map.
You asked about approach. You will need to address all phases (departments) of your business. Since an end user's or customer's experience is a function of their expectations, start with mapping how you gather and document those expectations.
As you do so, ask yourself if you are really interacting with all the needed people and capturing all you need to know about the products and services desired. What product and service functions are the most critical? What metrics will be used to assess performance?
Once you capture the customer's expectations, make sure you have a process for proactively verifying them as needs change. Also be prepared in the event the customer fails to communicate those changes properly to you because, as the supplier, you might be held accountable for any dissatisfaction. Not fair? Maybe, but you asked about best practices, and communication is a key consideration in being perceived as a superior supplier or partner.
Map your design and manufacturing processes, including how you work with your suppliers. Are you embracing design excellence? Do you use tools such as the house of quality (which is part of quality function deployment)? Are you applying a range of lean manufacturing principles and not just the value stream mapping tool to drive out all wastes and associated costs so you can give the customer the most competitively priced product or service possible?
Map your pricing process. How do you establish initial contract pricing proposals, and how do you ensure you negotiate consistently from customer to customer?
Map your service processes, including telephone and field support. How do you take those processes from average to exceptional in the eyes of the customer? Do you have preventive maintenance calls built in, or are you focused on repairing breakdowns? Preventive calls might seem expensive, but losing customers is even more expensive. Preventive calls might be one way of verifying customer expectations periodically.
That should be enough to get started!
Peter E. Pylipow
Sr. design excellence engineer, Vistakon
Orange Park, FL
- Lean Lessons columns, Quality Progress, February 2006-December 2007, www.qualityprogress.com.
- Dave Gleditsch, "The Lean Spectrum of Work," Northwest Lean Networks, www.nwlean.net/article0604.htm.
- Trent Wall and Dave Levine, "Atkins for the Office: A Lean Office Could Beef Up Your Ability to Surpass the Competition," Northwest Lean Networks, www.nwlean.net/article0904.htm.