To simulate or not to simulate?

Citation

Loughlin, C. (2001), "To simulate or not to simulate?", Assembly Automation, Vol. 21 No. 3. https://doi.org/10.1108/aa.2001.03321caa.001

To simulate or not to simulate?

To simulate or not to simulate?

Our theme for this issue is "Productivity improvements with simulation". This is intended to cover both the benefits to the system integrator that can arise from simulating a system as part of the design process, and also the benefits of improved productivity for the manufacturer.

The trick with simulation is doing the right amount of it for the task in hand.

For example, if you need to check that two components can be placed at locations A and B without interfering with each other, then you will need to accurately model their 3D external shape and then use the simulator to position them where you want and check that the two do not collide or overlap. If the two components are cubes, then it is just the work of a few minutes to define their dimensions and move them about in the virtual world of the simulator. However, most real world components are not cubes and the process of specifying them can take hours, days or even weeks.

Of course, if they were simple cubes you probably would not bother with the simulation at all, so the conclusion must be that, if simulation is worth doing, it is normally pretty difficult, and it will take quite a few man hours to do anything meaningful. The trick is in deciding just how good your simulation needs to be. Certainly if you are only interested in exterior shape then there is little point in also modelling the interior. Also it may only be necessary to model the surfaces that are likely to come into contact, leaving the rest as a simple cube. By the same token, any surfaces that may come into contact must be modelled as accurately as possible.

A poor simulation may be worse than no simulation at all, as it may give false confidence.

As an example of this I recently had to move a steel boat 11m long and weighing 2.5 tons out of a barn where there was just 40mm clearance between the widest point on the boat and the doorway. The boat had to be on its side and fixed into a cradle, on which it was then dragged through the doorway. Boat hulls are very complex 3D shapes with varying curves in all directions and so I elected to make do with a number of 2D views which showed the boat mounted in the cradle and the doorway. From the comfort of my own desktop I was able to see that the whole thing would fit with about 20mm clearance to spare.

On the day in question all went according to plan until we came to the door, whereupon the boat jammed solid against the top of the doorway. After several rather anxious hours pondering various alternatives we determined that the boat was getting stuck because it was too high in the cradle, and the reason for this was two diagonal struts that had been included to strengthen the cradle. These were supporting the hull and stopping it sitting low enough in the cradle. The solution was to move the boat forward in the cradle by about 200mm to a position where it could sit a bit lower. We greased the top of the boat, anchored the cradle to the ground, crossed our fingers and pulled the boat forward the 200mm with a tractor. It ended up giving us 10mm clearance and the boat then departed the barn with both barn and boat intact.

Fortunately, this story had a happy ending but it could easily have been otherwise. A full 3D model would have taken ages to program but would almost certainly have highlighted the problem.

In this issue we have a feature, written by Jack Thornton, describing how a simulation of the maintenance of a military aircraft was essential both for optimising the layout of components to ease the maintenance task and, more fundamentally, for ensuring that maintenance was actually possible. The investment of time and money in producing such a simulation is very considerable indeed, and perhaps only justified for such high value contracts.

One problem with simulation is that the same amount of effort is required to create the simulation regardless of the quantity of the product that is eventually to be manufactured. Complex systems will take a lot of effort to simulate, but by the same token they are just the sort of systems that most need simulation in the first place. Simpler systems will take less effort, but the need is less as well.

There is, however, one particular area where simulation has got to be worthwhile, and this is for modular automation systems. The whole purpose of modular automation is that the various component parts can be reused in different sequences and combinations in order to create a wide variety of very different automatic assembly systems. Effort put into the simulation of a given module will repay itself manifoldly each time that module is used. The simulation will have to be very good, but it will be well worth the effort.

The decision "to simulate or not to simulate" comes down to whether the cost of simulating is less than the likely cost of not simulating. Do not underestimate the effort that good simulation will require, or the benefits that are likely to arise.

Clive Loughlin