How astrology can improve your production scheduling

Is the sequence of the orders on your bottleneck machine a Leo or a Virgo? Taurus or Gemini? This post will show you how astrology can improve your production scheduling results, and how visualization can provide you with insights far beyond the reach of any KPI framework.

Creating a sequence of orders on a machine is the essence of production scheduling. This task would not be so complicated if machines were not engineered to be versatile and to be able to produce a large variety of products and process a wide range of materials. Differences are typically in product dimensions, input materials (whether it’s steel grade, aluminum alloy, chemical composition of liquids or plastics, etc.), after treatment or any other kind of property that influences the required means of production. But all this production flexibility comes at a cost: instead of simply sorting your orders based on their urgency, all kinds of rules and constraints about good or allowed sequences need to be taken into account. Before you know it, planners are fighting full time against the number one killer of productivity at many facilities: change over time.

The question then arises: how good is my schedule in terms of productivity? The best way to find the answer is to define a KPI on the schedule and inspect its value. Your manager will undoubtedly be interested in that value (applauding you for an OEE of 95%), and you can favourably compare that value to yesterday’s value and even that of the days or weeks before to assess how the schedule quality is developing over time.

But what does this analysis really tell you about today’s situation? Let’s say you obtain, with the same product mix, an OEE that is significantly less than yesterday’s value, let’s say 70% – instead of the 95% you were so proud of? Then something is really wrong.

However, if you know that today’s product mix is significantly different from yesterday’s, how can you assess if 70% is actually a very good achievement given the current situation? It’s like getting a low grade for your test and being expected to do better next time without being explained the errors that you actually made. Yet many scheduling systems are trying to do exactly that… sophisticated planning and optimization systems, however, are able to provide keen analysis and enable KPI-oriented planning – which is impossible without proper visualization aids.

One of the nicest examples of this I’ve seen was in scheduling continuous resources in a metals factory recently. The world of production scheduling is not black and white, and working with machines is not always a question of ‘can do’ or ‘can’t do’, that it’s not always a question of ‘two hours change over time’ or ‘4 hours change over time’. In fact, there was no sequence-dependent change-over rules at all in this factory. The mills, picklers, and levellers and all typical resources you find in a metal factory were simply running around the clock, literally.

That means that each material is actually welded to the next material, thereby pulling it into the machine. After production, the materials are separated again, but from the machine’s point of view, a virtually endless coil is created that runs through it. The resource is never stopped during this process. Of course, there are restrictions as to which coils can be welded together. A 0.5 mm thin coil of 400 mm wide, weighing about 5 tons, is not able to pull in a 4 mm thick coil of 1200 mm wide weighing 20 tons – that would be like a Volkswagen beetle trying to tow a huge truck over the Alps.

It’s funny though, when asking what the actual maximum allowed difference was, nobody in the factory was able to answer conclusively. “We can go extreme sometimes, but that means we have to run the machine slower. Otherwise, the smaller coil will break”. Hmmm… Varying production speeds based on the actual sequence, without having clear boundaries of what is allowed and what is not: how on earth to support a planner in creating a good sequence for such a machine?

As it turned out, a visualization that later became known as the “Stars in the Sky” was the best planning aid conceivable. It was a simple, black, 2-dimensional plot, with different attributes (in this case, width and thickness) on the axes. Every order that was scheduled on the machine is a dot in this plot (or a “Star in the Sky”), and by connecting the Stars as they were in the sequence on the machine, we obtained our final picture.

Is the picture of your production schedule one big galactic disaster, with lines crossing each other all over the place, or is it a nicely drawn constellation? Well-balanced sequences that safeguard high machine speeds result in clean pictures. With a sophisticated planning and scheduling system, a planner is able to spot where the sequence can still be improved, and get that 70% OEE back up to the 95% OEE or confirm that 70% is actually an excellent score because there is simply no better constellation possible given the current situation.