Risk management within a production factory context is a complicated topic even when viewed in broad terms. The range of distinct potential problem areas is wide and varied, and much of the threats facing a factory are outside of direct control, being dependent on third party suppliers or contractors. However, for the purpose of this discussion, a very small segment of factory risk management is going to be looked at – the mechanical process line. The mechanical process line encompasses, for this discussion, any moving part of the line. This includes conveyors, motors, bearings, actuators, pistons, pumps, applicators etc.

Risk in a production line typically hinges around those associated with, or resulting in reduced product output. Most production managers will have in place an acceptable percentage of rejects (damaged, faulty, out of specification) as well as an acceptable percentage reduction in overall output of the line (no rejects, but slower running line).

Any moving part in a production line, when it begins to malfunction, or operate outside of desired specification, will have an impact on the production line, either in terms of affected goods, or reduced output. This impact can be periodic, with intermittent issues and product failures, or it can be continuous, ultimately bringing the production line to a complete stop. On intermittent issues, a quick evaluation can be made by managers – is the effect of such importance and lost production value, that it is worth stopping the production line, or attempting troubleshooting while the line is running, or is the effect minimal, or within financial tolerance, that the line continues and the problem is fixed during routine maintenance. With continuous problems, the usual response is to stop the line, often simply due to the fact that the line needs to be cleared. Then, while the line is being cleared, an evaluation of potential problems can be made. Sometimes problems are highly visible and self-evident – for example an actuator running out of sequence. Then it is a simple fix. However, many production lines these days are running at high speed, which results in an area of concern being visible, but the actual cause cannot be determined as it is too fast to see.

Experienced maintenance staff are invaluable to a factory, and by themselves, can offer a significant reduction in risk. Their range of experience and intuitive understanding of machinery and process control can often result in a very quick diagnosis of what is wrong, and the remedial steps to correct this. However experienced staff are increasingly rare and invaluable (risk of poaching of staff), and as a result, a comprehensive risk management plan needs to take into consideration the loss or unavailability of experienced maintenance engineers.

Focussing then on mechanical issues that are typically too fast to see what is happening, regardless of whether the impact is negligible or major – there are several options a line manager can take. Firstly, start slowing down the line, and see if that reduces or removes the problem. This is a valid, although usually temporary option, which results in a reduced output, but there is still ongoing production. Secondly, to continue with the production speed as is, with periodic stops as necessary, to clear the line. Thirdly and usually the last resort, a complete stop of the line, when production is so badly affected and waste levels so high that production cannot continue. With option one and two, typically maintenance staff are immediately dispatched to start trying to trouble shoot the line – through intuitive understanding, as well as trial and error tweaks and adjustments to suspected components. During these as well, managers will likely assign increased numbers of production staff to the line, in order to facilitate the clearing and restarting of the line as required. To save time, these staff need to be in close proximity to the line. With option three, the line would be run periodically, either with live product or dry runs if possible with maintenance engineers in attendance. Regardless of which option taken, now identification of the problem begins. By reducing the line speed, it may become evident which particular component is the problem. This component can then be serviced, altered or replaced as required, and then the line run at full speed to verify that the problem is sorted out. However, if reduced speed does not identify what the problem is, then similar to option 2, as well as option 3, essentially trial and error based adjustments need to be undertaken, as the exact problem is unknown. This can be a quick, or a very long process.

However, there is a very useful tool, which can be implemented immediately upon noticing that there is a problem. High-speed cameras, which can record from several hundred to many thousand pictures per second, have developed from film based and unwieldy systems, to digital, compact and very versatile systems. The human eye is typically able to resolve “speeds” of around 25 pictures, or frames per second (fps). Thus typically, old film projectors, video camera recorders etc have recorded at speeds of ±25fps. While these could be slowed down, and even viewed frame by frame, there are insufficient frames to provide necessary data. However, if a camera, with 1280×800 pixel resolution is recording at 1000fps, with a 1000mm view of the production line, and a tub of cream 55mm wide is moving at 2 meters per second, the following is noted:

1. The tub of cream passes from one end of that 1-meter window, to the other in 0.5 seconds.
2. 500 distinct pictures of the tub are taken.
3. A picture is taken for every 2 mm that the tub travels.
4. The tub of cream occupies 43 pixels out of 1280 pixels.

From this, a very clear understanding of what is happening to the tub of cream as it passes through the 1-meter area is achieved. With 500 distinct pictures, which can be viewed one at a time, as well as played back in video mode and selectable speeds will show any effect of components on the tub. Movement and orientation of the tub, as well as any processes associated with it – filling, capping, labelling, will be clearly seen. Thus, should there be any problems, with filling, capping or labelling for example, the sequence of pictures will show exactly what and where in the process the error occurs. If the problem is being caused by a bearing or conveyor wheel jiggling, or an actuator functioning at wrong time for example, this would also be visible. At times, a much smaller area could be re-filmed to give a higher resolution of the problem area to get greater detail.

Time taken to produce the evidence? Assuming that access to electricity and all safety concerns are dealt with (example opening machine cover), from arrival at the factory floor with camera, tripod, cables, laptop, lens and lights, it would take 10-15 minutes to produce the first footage. Acquiring a closer view for higher resolution may require moving the camera closer, or using a zoom lens. Next footage available in 1-2 minutes. Now the exact evidence is available of the problem components. Immediate service/fix/replacement can ensue and the line restarted and re-filmed to confirm the problem has been dealt with.

Offered below are two examples that we have been involved in.

Firstly, before we were involved in high-speed cameras, a production factory called us to order a new stepper motor. This was duly sold and delivered. A week later they called again, now they wanted a new planetary gearhead to fit the previously purchased motor. This was ordered and two weeks later delivered. A few days after that the now irate production manager called to say that the stepper motor and gearhead were pieces of …. and we as a company were a bunch of …. and ….. Needless to say, I was at the factory as soon as possible, in order to try and get the full story. As so often happens, we had received none of the actual story. We had simply had requests for components, without detail of why they were needed. In this case, it was the maintenance contractor who had told the manager, “You need to order this”. Which he duly did. What had been happening is that a filling machine had been having problems. The filling machine filled 10 tins at a time, row by row for 10 rows, before these were then processed out to the lidding machine. Periodically while a row of tins was being sent out, the next row was being brought forward, leading to 20 tins being crushed and mixed. The machine stopped, was cleared and restarted. The maintenance contractor believed it was the stepper motor which was malfunctioning, and then it was the gearhead at fault. However, with both replaced the problem still occurred. Viewing the machine in operation, together with a fault, intuition said it was not the motor itself, as it appeared that there was movement in the rows before this motor started. However, it was happening just too quickly to know for sure. After watching three faults, intuition said it was either a proximity switch at fault, or that the PLC was sending incorrect pulses to various motors. Both the maintenance and manager disagreed. They then went out and purchased someone else’s stepper motor and gearhead and the problems continued. Finally, after 3 more weeks, a new maintenance company contacted us asking for a loaner stepper drive. This we loaned to them, and they quickly deduced the drives were functioning as per spec. Then the PLC was tested and it was found a module in the PLC was giving intermittent problems, resulting in wrong step and direction pulses being sent at times. Module replaced, machine worked perfectly. Time taken to fix problem – 7-8 weeks, cost of lost production unknown. Maintenance contractor lost contract for entire factory. With a high-speed camera, it would have been proven within minutes that while the one stepper motor was still running, the other one started, causing the problem. Without it, intuition and trial and error, with 7 weeks of frustration.

More recently, a call came in for a rental of a high-speed camera as a food production machine had been giving problems for 6 weeks. During this time, the machine could run at ±60% of full speed, with intermittent issues, but as soon as they tried to increase speed, the number of errors increased. During this 6 weeks, maintenance crews had worked tirelessly trying by trial and error to solve the problem. However they had no idea where in the machine the root of the problem was. Within minutes footage showed that on a small feeder line every now and then one packet was twisting slightly as it was laid down. This caused the edge to brush against the guide rails, slowing it down fractionally. This caused the next packet to run into it, which caused a bigger slow down, until the machine tripped. Now it was immediately clear where the problem was. Simple adjustments to the pneumatic system, and new footage taken to prove problem resolved. Immediately the machine went back to 100% and no further problems were seen. During the time there, up to 10 maintenance staff were on the machine, 4 extra production staff were on standby to clear the machine, and further down the line, 3 staff were standing around each time the machine tripped. In this example it is very easy to calculate the costs of this:

1. Lost production, running at 60% for 6 weeks;
2. 20-30 rejected packets every time the machine tripped;
3. Up to 10 maintenance staff at their daily/hourly cost;
4. Increased production staff as well as staff unused during down times;
5. Inspection, reworking and manual packing of rejected packets.

If we assume the following (understanding that this is guess work on rates and costs but gives an indication): profit per packet R0.05 at full production of 1000 pieces per minute, maintenance staff daily rate R750, with average of 3 staff per day, production staff at R440 per day, with average total of 4 affected (including inspection and reworking), with the machine running conservatively 10 hours a day, 5 days a week, for 6 weeks.  Over that 6 week period, almost R500 000 in increased costs and lost production. This was on one machine. This, versus the cost of a high-speed camera system (decent entry level units, with lens and lights R250 000), or even using rental services.

As a final example for risk management, there was a client who contracted rental services in order to provide a database of baseline footage of machines, while they were functioning optimally. This footage was to be used as reference as soon as there were problems.

Here it can be easily seen from these examples that high-speed cameras can play a very important role in fault finding, especially on fast-moving production lines where the process is too fast for human eyes to resolve. Although expensive, investing in the high-speed camera asset, or knowing available resources, therefore can be a very valid risk mitigation plan, as it provides necessary data very quickly, as well as removing dependency on trial-and-error fixing, which can be time consuming. In a very competitive market place, lost production and increased costs are becoming greater risks to the viability of production lines and factories.

[David Horne]