Digitalisation framework for smart factories

Introduction

Globally, there is a growing chorus of voice among the manufacturing communities to gravitate towards “Smart factories”. Smart factories are production environment which enables facilities, machines and logistics chains to be manged without human intervention. They are highly digitized shop floor that continuously collect and share data to proactively address issues, improve manufacturing process and respond to new demand. Additionally, these factories focuses on building hyperflexible, self-adapting and  intelligent manufacturing capability. Furthermore, such factories are a holistic transformation of people, process and technologies to achieve the intended performance or business goals. While notions of smart factories have existed for several decades what has brought it within grasp of reality is the maturation of the Industry 4.0 technologies.

One such technology is IIoT or more specifically sensor technology. Manufacturers are installing sensors on machines and building IIoT platforms to introduce a new layer of intelligence and to track critical KPIs such as temperature, vibration, moisture, uptime, wear and tear, and more. But installing sensors merely is not enough. If the focus still remains to pull data manually, run reports, analyse the results, and only then take actions, it is huge loss of efficiency and also a limitation of the ability to meet customer expectations. So the critical part is not generating swamp of machine specific data but gleaning insight from that data.

Ergo, it is essential to enrich the IIoT data in the factory with business context, and leverage it throughout the value chain. This requires modern cloud based applications that can integrate ERP driven business information with IIoT driven manufacturing operations data. Such applications can tell if a machine is running hot and heading for a failure. They can also tell the impact on the business – if a manufacturing run is at risk, which customers are affected, and the cost associated with any downtime.

Developing a digitalisation framework

The path to a smart factory can be built by thinking in terms of a digitalisation framework tailored specifically for manufacturing. Such a framework should aim to bring the physical and digital world closer by transforming activities related to product designing, engineering, manufacturing and servicing. It should also help in creating an unified data view, stringing set of related data as it flows through various business processes and functions. It is useful to understand in greater detail what could be the various aspects of this digitalisation framework.

The first aspect is installing a methodology (platform, process and people) to collect, analyse and distribute product development and manufacturing data, throughout the network, to support insight and decision making. Here emphasis should be given to collect data seamlessly, real time, accurately and with relatively less effort. While IIoT sensors are best choice for purpose of data collection, manufacturers can also consider mobile application based data collection as a viable alternative. Such mobile applications should collect data as part of an all-encompassing end to end manufacturing workflow.

The second aspect is tracking quality of product or production system. To do this it is essential to digitally capture the crux of product design – recipes, CAD drawings or product specification data (defining how the product should behave and be used). Once captured digitally this information can be leveraged to create simulation and emulation of real behaviour for the manufacturing environment both at line and machine level.

The third aspect is providing clear instruction for the workers to achieve highest throughput and quality standards. This requires training of the workforce and identifying opportunities of optimisation – defect reduction, machine effectiveness, safety and sustainability. It also requires capturing data related to the way production is executed such as process information, machine parameters and laboratory tests data.

The fourth aspect is digitalising product research and development to make better products that run efficiently and improve speed to market. At the same time receiving feedback from the clients on how the products are being used and feeding that as an input into research, development and design. The fifth aspect is digitalising the transformation process of a product concept to manufacturing operations and circling back manufacturing data into the product development process.

All the above aspects of the digitalisation framework are going to generate a massive amount of data. All these data need to be integrated from a data management perspective and also from the perspective of creating a virtual shop floor experience. Without an integrated data model it is not possible to optimise the value chain. Moreover, to be effective the digitalisation framework will require manufacturers, customers, and supply chain partners to have established well run automated manufacturing (physical) and business (process) systems.

Alternative to fully autonomous smart factories

One of the primary objective of a smart and autonomous factory environment is to build partners and an ecosystem who can offer their factory as a service. In the absence of a fully autonomous and fully integrated system, an alternative approach is to  achieve this is by establishing manufacturing command centres. These centres can use three levels of information (of manufacturing operations) visibility– the machine/operator level, plant line/order level and, finally, globally (across all locations, processes, and production assets) – to improve decision making and harmonise global processes.

It is worthwhile to delve deeper into the approach of building a manufacturing command centre. To begin with it is necessary to carefully identify a potential area that can be carved out independently where a command centre can be established. Next step is to build a digital workflow for this process, automating data collection wherever possible, and integrating data from different parts of the process. Third step is planning to make all the process related information readily available in a flexible environment like the tablet or mobile phone of an expert. Also, the information should be available from a machine perspective, a plant line perspective and from an audit perspective, and this should be globally available across all the sites to be focussed on in the future. 

Once a command centre is established it can run as an open loop where there are still people responsible on the site, but the command centre is now supervising and providing remote support in this environment and being able to guide operators and give them additional information. This allows the command centre  to have visibility into any problems or bottlenecks, with information on how they can be optimised using closed-loop control algorithms. All this will contribute eventually to building of the smart and autonomous factories.

Conclusion

It is important to recognise that while many smart factories contain highly automated systems, automation is not a prerequisite for realising the goals of smart factories. It is data and not automation which is the key foundation underpinning all smart factory use cases. Technology that enables the acquisition, orchestration, and analysis of relevant data will empower the humans running the factory to make faster, more informed, and ultimately better decisions.

Looking forward, it is apparent that the future of manufacturing is a sort of symbiotic relationship between humans and machines, with data-driven tools bridging the gap between physical and digital world, enhancing the production capabilities of employees, and extending manufacturing beyond the traditional factory set-up. Technologies such as robotic process automation (RPA), IoT, augmented intelligence, machine learning, blockchain and responsible artificial intelligence will contribute significantly to creation of smart factories of future. However, all of these technologies are currently at various stages of evolution and will be mass available in the time frame between 2 to 10 years.