Organisations in the industrial sector and in particular manufacturing are being transformed and disrupted by the irresistible forces of cognitive computing, cloud and the Internet of Things. This phenomenon, characterized as the fourth industrial revolution after robotics and automation, is a hot topic for manufacturers around the world as they seek to understand how they become the disrupters rather than the disrupted. The Internet of Things in particular will provide the cornerstone for this revolution, as it provides the means to continually monitor production.
How do we become the disrupters in a revolution?
The big question though, is how and where to start?
The best answer is with a combination of business-led direction (“top down”) and technology exploration (“bottom up”). The top down view provides the focus – for example, prioritizing a recurring failure because it regularly incurs the biggest losses through halts on the production line. The bottom up provides the means to identify root causes through better instrumentation. For many, simply measuring equipment use and understanding the true efficiency of their operation is a quantum leap forward.
Whilst a lucky few may have the luxury of a new factory to build, the reality is that many manufacturers have already invested significantly in equipment and skills to operate it over many years. In such ‘brownfield’ scenarios, therefore, the manufacturer is unlikely to be able to justify a rip-and-replace strategy to take advantage of new technology. Furthermore, the installed equipment will either be of an age where digital instrumentation is not inbuilt, or the instrumentation that is available is deliberately locked away in highly proprietary vendor systems serving a point purpose.
Optimising existing frameworks with supplementary instrumentation
In both cases, the best coping strategy is to add supplementary instrumentation, either in the form of add-on external sensors (e.g. vibration, sound, temperature) or if possible, by taking a feed from any proprietary on-board sensors so the data already being captured can be harvested in a broader context. Which sensors are required may require some exploration. Anecdotal experience from operators may provide the key – for example, if there is a rule of thumb that a particular failure occurs on hot days – plus some good old fashioned common sense. If a production line is often halted without warning because pressure has been exceeded in the assembly of an engine, we might reasonably assume that by measuring pressure on a continuous basis, a trend might emerge that we can capitalize on.
The good news is that this process can begin small, and grow with experience.
In the case where the manufacturer simply requires better insight into the operation of their factories, the exercise might simply be one of gathering sufficient data to determine more accurate KPIs, for example the overall equipment effectiveness (OEE) which compares expected output with actual. On the other hand, where the organization is looking to reduce the impact of machine failures, we typically see a two-step process.
- Add instrumentation to gather more context associated with a failure, and provide a means of analysis to develop and test hypotheses about relevant trends.
- Having determined a hypothesis, develop automated rules and analytics to take corrective action either to reduce the impact of an impending failure, or avoid failure completely.
In both cases, best results occur when a limited scope pilot is attempted (perhaps a single cell or piece of equipment) to validate the approach, before embarking on a bigger roll out.
In summary there is a great opportunity, the technology is ready now and you can get going quickly – why not start your revolution today?
If you’d like to start using the data in your manufacturing environment quickly and make real changes fast, why not take a more detailed look at the IBM Watson IoT Platform. To find out how IBM is helping automotive companies apply cognitive IoT technologies to their products and manufacturing processes you can see some examples in this IoT in Automotive website.
Share this post: