I was teaching the Statistical Process Control & Capability module during a Lean six sigma Green Belt class and one of the students asked, and I'm going to paraphrase, "When is enough, enough?" I think this would be a normal question to ask while learning how to Make Value Flow. The question came from one of my Quality friends, and I'm sure he was thinking about all the different processes he supported that needed improvement and certainly not when he could put his feet in the air and his "you-know-what" in the chair.
If you have been measuring your performance, then you should have some amount of data collected around the problem you are experiencing. If you are having quality problems, you may have Defects Per Unit (DPU), or Defects Per Million Opportunities (DPMO). If you are having delivery problems, you may have Cycle Time or Number of Days Early/Late. If you are having cost problems, you may have Cost Per Unit or Weekly Printing Costs. Either way, you need some data that represents your process.
We can see the performance is in control and the stable, but is the process meeting the customer's needs? This is always the right question!
When we use SPC we are looking for indicators that signify when to start asking questions about the process. There are four that are the easiest to use and remember; 1) Any points outside of the control limits, 2) Six points consecutively going up or down, 3) Nine point consecutively above or below the center, and 4) 14 points consecutively going up and down. Having a process that performs within these four parameters means that our process is in control and stable, which indicates the flow is predictable.
Here is your SPC warning! Do not let any of you Quality Engineers catch you putting specification lines on your SPC chart.
What it does not tell us is are we meeting customer expectations. This is where we take our Voice of the Customer specifications and lay them on top of a histogram of the data. From here we can see our performance compared to customer specs of no more than 21 days (USL, LSL).
 |
| Current State |
And as we can see from the fine specimen above, when our cycle times are 22-25 days we are not meeting customer expectations. Our process is in control, but not performing to the wants and desires of the customer. In fact it may be possible that our favorite customer is looking for a new supplier of their information needs as we speak and I think that immediate action is required.
At this point you should be constructing a charter, assigning one of your belts to lead the project, and kick-off the team. Work your best as a Project Sponsor, Smart Person or Supplier/Customer and document the results in an A3. When you have enough data with the new process you can see how the center and spread fit within the specifications.
If you have identified the root causes and adequately implemented a new process that has improved flow, then you are done with that process for now. The biggest indicator is the edge of the distribution is some amount away from the specification. For you math nerds, below is the formula and you want to get as close to 1.5 as you can.
 |
| Improved Process #1 |
Based on our results we can see that are delivering everything on time, and if we finish early we can hold the deliverable until the customer is ready. Be wary of producing too much too fast. That would make our work-in-process increase, which means the window between "doing work" and "getting paid" increases and this is detrimental to cash flow.
At the end of the day we want our products to flow without stopping though our value-added operations to meet customer demand. Deliver too late and the customer must wait (Wait Time), but if we deliver too early the customer has to over-handle (Excess Processing) the product until they are ready to use it and this is "Inventory" they have to hold until they are paid. If you have paid any attention over the last 5 years, this over-capitalization is one of the causes of the downfall of the American automobile industry. Leadership is the decision-maker, but the following histogram represents one possible best fit.
 |
| Improved Process #2 |
How do you use your data for improving flow?
No comments:
Post a Comment