|
Design of Experiments (DOE) Lego ExampleNow that you know a little more about the basics of design of experiments, let's look at an example. Suppose that your company designs and manufactures Lego's, and wants to make sure it is making the "perfect" lego. Now, we all know that Lego's are trademarked and manufactured by only one company, however, lets assume that for this example that your company has hundreds of competitors all striving to design the perfect Lego themselves, and therfore, put you out of business. Let's assume that your Lego's are being injection molded in a single cavity tool, one part at a time. The response, or output that we are interested in is the outside length of the Lego. We want our "perfect" Lego to have an outside length of 57.3 mm. If we can make our Legos at exactly this size specification, they will fit together perfectly for our customer and as we all know, customer satisfaction is the reason we continue to make a paycheck. Our process knowledge tells us that two factors affect the length; Fill Speed and Hold Pressure. We want to know what happens to the length as we vary Fill Speed from 2 inches per second to 5 inches per second and Hold Pressure from 4600psi to 6500psi. The 2 and 5 inches per second for Fill Speed are just different set points, often referred to as levels. Obviously, your company would prefer your fill speed to be as fast as possible in order to have a shorter cycle time, and in return, be able to create Lego's faster than your competition. To set up our designed experiment, we are going to do 4 test runs on your machine, and create 5 parts per test run. A table set up using DOE Wisdom software is below:
As you can see from the table above, four unique tests, or runs, were conducted. For run number 1, we set Fill Speed at 2 inches per second and Hold Pressure at 4600psi. Once our process had reached a steady state, five shots were completed. The outside length was measured on all five parts and recorded in the table below. (Side note: Far too many companies will not take into account the accuracy and precision of their measurement system, however, when conducting a designed experiment, it is extremely important that your measurement system is both accurate and precise. An un-reliable measurement system will destroy a project and render it useless. If you would like more information on measurement systems, just click here).
For run #2, the Fill Speed was held at 2 inches per second but we changed the Hold Pressure to 6500psi. No other changes were made. Again, samples were taken after a steady state for the process was reached. The outside length was measured on all five parts and recorded below:
Runs 3 and 4 were completed in a similar manner. The final results are shown below:
Now that we have collected all the data we need, we can move onto graphical analysis. In many cases, much of what is to be learned from your experiment can be gleaned from a few simple graphs. Let's learn about one of my favorites, a main effects plot (below)
In this graph, we learn that Hold Pressure has a much larger impact on Outside Length than Fill Speed does (since the slope of the line is much greater). We also learn that the effect of each factor is proportional. (i.e. if we increase our setting on our factor, the response (length) will increase as well.) Now lets get back to our ultimate goal of creating the "perfect" Lego. Remember, to make a better product than our competition, we need to have an Outside Length of 57.3mm. Let's now look at a contour plot (again, from DOE Wisdom software.)
In our contour plot, Hold Pressure is plotted on the Y-axis and Fill Speed is plotted on the X-axis. Outside Length (our response) is displayed as contour lines in the graph. From reading the contour plot, we can assume that any combination of Hold Pressure and Fill Speed that falls on the line labeled "57.3" will produce our "perfect" Lego. So, to produce the "perfect" product at as fast a rate as we can (so we can successfully produce faster than our competitors), we should set Fill Speed to 5 (the maximum our machine will allow), and Hold Pressure to 5570. Pretty simple, huh? That is the basics of Designed Experiments. In producing just 20 units (5 runs at 4 setting each), we can figure out a way to produce a product to perfect specifications at a speed our competitors can't match. Now, imagine how long that could have taken by doing what quite a few companies actually do, trial and error. While they are busy doing a few hundred costly runs, mixing and matching settings, we have produced the "perfect" Lego, and given us a leg up on our competition. Launsby Consulting has been teaching design of experiments since 1991. To learn more about how we can help your company become more competitive, feel free to email me personally at Matt@Launsby.com, or give us a call at 1-800-788-4DOE (1-800-788-4363) or 719-282-1143. I hope you have found this example informative, and feel free to pass it on to anyone you like. However, if you are going to show this in a class or conference, please credit Launsby Consulting. Thank you for your time, and I look forward to hearing from you. Launsby Consulting offers a variety of tools on design of experiments.
Launsby Consulting has been an industry leader in teaching designed experiments since 1991. To view our partial client list, just click here.
|
||||||
