If we take the `buzz’ around Industry 4.0, and the resurgence of interest in Artificial Intelligence (AI), add a sprinkling of imagination, and we end up with some fantastic possibilities for manufacturing systems of the future.
Industry leaders have known for a long time of a need to combine competitive advantage with minimal costs. If we could somehow develop the ‘best’ product whilst controlling the creation and retention of intellectual property, and have someone else make the goods. Of course, this is a difficult balancing act that demands a finer-grained, more detailed understanding of manufacturing processes, environmental economics and human behaviours than most manufacturing companies possess.
A great deal of manufacturing control exists as tacit knowledge, gained by machine operators and their supervisors over a period of time.
The introduction of computers, and lately, the widespread availability of very cheap computation through embedded and IoT devices, has enabled considerable strides to be made in pursuit of resource optimisation.
But, as any economist knows, there is considerable value to be had if manufacturing organisations can collaborate at the process level, rather than through board-level business agreements. If we can use technology to realise and re-engineer the concept of the ’supply chain’, then there will be opportunities for waste minimisation, enhanced quality, increased efficiency and ultimately profit maximisation.
So, what are the challenges that so far have prevented the manufacturing domain from realising this hitherto hidden value?
Manufacturing systems have three inherent characteristics that need to be dealt with. First, there is the distribution of activity, that in many cases is parallel, and this is difficult to comprehend at the macro level. Second, manufacturing systems must be able to adapt to changes in economic demand, which means that there is a large degree of flexibility that needs to be accommodated. Third, such systems have also to accomodate uncertainty in many guises; economic downturns and booms; material shortages and abundance; employee skills currency, etc.
Taken together, a manufacturing system is a highly complex entity and is thus an interesting and potentially rewarding area to study.
If we return to the shop floor of a manufacturing company though, how do the staff cope with uncertainty in material flows, erratic sales forecasts, and staff shortages, to name just a few of the daily issues?
Such is the demand now for organisations to offer and deliver enhanced customer service levels, that originated in swifter lead-times and are now being compounded by mass-customisation to accommodate tailored customer needs at a massive scale, that it is becoming impossible for staff to comprehend the necessary variables and make sensible business decisions in the time available. They need assistance, whether it be visualisation to aid comprehension of complex data, simulation to understand the potential impact of different strategies, or guided-automation to reduce the number of options from thousands to maybe three or four choices.
A conversation with an experienced production supervisor or production manager will quickly reveal the experience that is called upon daily to deal with systemic problems with product design, manufacturing processes, organisation, scheduling, etc., that are all additional factors that need to be considered when managing manufacturing operations.
Many manufacturing companies are working with computer systems that at one level enable the transactions to be completed, but which also hinder the organisation’s ability to be flexible and adaptive to uncertainties. In effect, there are restrictive characteristics that are embedded with in the manufacturing systems that staff learn to cope with.
Thus, if we start to consider the collaboration possibilities of Industry 4.0, by way of accessible, cheap computational devices, there is also the potential for chaos if a collaborator cannot contribute positively to the shared desires. In fact, there are likely to be casualties for those companies who cannot adapt quickly enough.
Since the computational hardware that enables data processing and sharing via networking is increasingly becoming embedded in manufacturing plant, it is inevitable that the solution to enhanced collaboration must require software at least in part; we must ignore the human factors of such changes at our peril, but the underlying connectivity of processes, data, knowledge exchange and visualisation for comprehension are all facilitated by tools that can abstract us away from the detail, while ensuring that the detail informs the decision appropriately.
In effect, we need to go back to the basics of software engineering; our systems need to be fit for purpose, meaning that they deliver what we need from them in a predictable way. They also need to be able to tolerate changes and be flexible. And they should be documented in a way that enables future developments to be understood in the context of the existing system capabilities. As we move towards adaptive systems, that perhaps use statistical learning or genetic algorithms to generate new functionality through system use, the requirement for self-describing systems becomes much more relevant.
If a system design is documented, it can be tested, even before it has been written as program code. If the functionality is tested against a known set of cases, it can be wrapped-up as an abstraction to simplify the design of other systems that can utilise that functionality.
Formal methods are approaches that can support the development of systems that do operate in complex environments, that have consequences if uncertainties are not catered for. Such methods use mathematics to precisely define concepts and system states, whilst also enabling logical reasoning to be utilised to test a specification to ensure that it meets what is required of the eventual system. Perhaps losing a network connection between two items of manufacturing plant is something that a factory can tolerate and overcome. But what would the impact be if a number of machines, across several sites (or maybe cross-cutting more than one industry), upon the optimisation of a system that ensures a mission critical order is delivered complete?
It is the informal that we deal with on a daily basis, and this is something that needs to be addressed as part of system design, as sell as system integration. Cyber Physical Systems rely on communication and collaboration to manifest enhanced performance. As such, to trust these systems needs more formality in their make-up.
Manufacturing systems are complex, but the next significant development opportunity of Industry 4.0 will only increase the complexity of design, validation and implementation of such systems. We need to replace informality with formality in our response.