The Third Law of Thermodynamics and Murphy’s Law both require that things break down. When do you get around to fixing them? Maintenance culture tends to fall into one of two patterns. The first is to fix it when it has to be fixed, i.e., the machine will no longer run saleable product. (Sadly, these cultures often have a DOFER mentality; it will do for now.) The second is to fix it when it should be fixed. Here we mean repairing/replacing a component at the most economic moment in its life cycle. This is the practice we will expand on briefly in this post.
However, before we begin we need to introduce an industrial engineering measure of component life called MTBS or Mean Time Between Service. We could apply the easy metric to just about anything that wears out regularly in our plant such as rubber covers or slitter blades. In fact, many maintenance departments already record cover life data and some high-end slitter systems automatically record blade life.
The cost of maintenance is easy to quantify. For the rubber cover it might include the cover grinding costs, the occasional recovering costs (every nth regrind), labor, machine downtime, etc. Unfortunately, the costs of NOT grinding are less quantifiable and are often ignored until the very last minute when it becomes painfully obvious something needs to be done. (Recall the fix it when its broke culture discussed above.) This practice is guaranteed NOT to be economically optimum, but is somewhat inevitable because management tends to undervalue what they can not easily put a dollar sign too.
Let us change that by putting a dollar sign on NOT grinding. You can do that for many situations where defects result from a deteriorating component. Let us take for an illustrative example, that a worn cover will cause a coating streak, corrugation or some other lane-like defect and that you have a defect code for that defect. All you have to do is plot the defect cost versus time. You will probably see something approximating a sawtooth. That is where defect rates slowly rise with time and then suddenly drop when a key component is changed out; in this case we used a worn cover as an example. The cost of not grinding is the area under the curve and to get the per annum cost we simply multiply by the maintenance interval. Note that if we decrease the maintenance interval, as seen in the dotted curve, the costs of not grinding decrease. We can get this new curve for shorter lives even without running the experiment because the slope remains the same. It should come as no surprise that (relevant) defect costs are reduced when going from the 50 day regrind interval (orange) to the 25 day regrind curve (yellow).
One other thing to note. That is there is likely a base rate of defects even when the roller is new. This is either due to the brand new roller not being good enough (yes, it is possible) or due to other causes not related to the roller cover, such as poor web profile as one of many examples.
Stay tuned to the next post when we show how to easily determine when the roller should be changed out to save your company the most money by economically optimizing maintenance intervals.