The Rise of Networked Production and Manufacturing-as-a-Service
Published on : Monday 20-01-2020
Half-hearted application of technology ‘band-aid’ on the old, legacy manufacturing strategy and structure is not a solution to upcoming threats and challenges, cautions Somil Gupta.
Manufacturing industry is geared up for another revolution. It started with Industry 4.0 to connect machines to cloud to collect data and real-time information to drive efficiency and transparency in processes. And now we face a new threat and also a new opportunity for a major transformation in production structure and business model – welcome to the era of Networked Production and Manufacturing-as-a-Service.
The global changes in consumer behaviours and needs are creating disruptions:
• Shorter product lifecycles
• Demand for personalised products and services
• Demand for connected and managed products, and
• Demand for ‘Same Day Delivery’ and 24x7 access and support.
Consumer buying journey has become complex and unpredictable limiting the reliability of demand predictions and forecasts. Market dynamics have also become faster, more complex and unpredictable than what traditional manufacturing systems can handle. ‘Batch of One’ is no longer a far-fetched concept; it is now a reality where consumers want and are willing to pay for hyper-personalised products and services. Adidas is setting up two ‘Speedfactories’ at its suppliers in Asia to get more flexibility and variation in
in Asia
product design, better utilisation of production capacity and faster response to consumer needs.
The classic manufacturing methodologies like Toyota Production System and Lean manufacturing are excellent in maximising efficiency and throughput but fall short in handing variability and uncertainties. In lean setups, uncertainties are expensive. A part of uncertainty can be avoided using better forecasting models but most of the current uncertainty is a result of inherent complexities in the supply chain and consumer behaviour. And these complexities cannot be ‘predicted’, they must be ‘managed’.
A leading indicator of an impending crisis is evident from China’s manufacturing sector that is experiencing severe slow-down and margin erosion this year due to demand contraction in both international and domestic markets. China’s manufacturing industry is built for high volumes and when volume shrinks, profitability is severely impacted. And this not limited to China. Globally, sustainability and environmental concerns are driving investments in sharing and circular economy and re-manufacturing. These trends disrupt the volume game and further add to demand uncertainty. Therefore, dependence on sustained high volume and inability to handle variability in demand and product mix is no longer sustainable and threatens the survival of manufacturing companies relying on old structures and strategies.
A lion’s share of investment until now has gone to ‘avoiding’ uncertainties instead of managing it. This includes integrating machines and information systems horizontally and vertically with suppliers, partners and customers. While these integrations create information super-ways, they also create dependencies and domino effects where a disruption in one part of the value stream reverberates throughout the chain. The second strategy for managing uncertainties was ‘levelling’. Simply put, buffers were created to absorb fluctuations in demand. These buffers included safety stocks, capacity buffers, maintenance buffers, etc. As supply chains are getting leaner, these safety buffers are increasingly seen as ‘inefficiencies’ to be minimised/eliminated. But when you start taking out the cushions you start getting hit by the bumps.
In order to manage uncertainties and variabilities, manufacturing needs to become highly flexible and agile – which is not the case today. We have a ‘Line and Process Mindset’ where all manufacturing decisions are taken considering the line or process as the fundamental unit. It works when volumes are high, variability is low, and the line/process is stable except for some minor adjustments in intermediate process steps. But as we get into a higher number of variants and fast product changeovers, line and process thinking doesn’t work. There are too many parameters that impact the efficiencies and throughput. Optimisation becomes an ongoing process where the line must be balanced at every changeover – new tooling, process parameters and instructions. In a linear line/process approach, such changeovers result in sub-optimal production. Whereas line managers prefer to run their lines as smoothly and efficiently as possible because demand fluctuations put a dent on prices and further chip away margins for manufacturers. Is there a better way?
The networked production offers a viable strategy for managing both variability and volume. It is a paradigm shift where lines are not seen as indivisible unit rather as logical sequence of process steps – process chains for manufacturing different product variants. The machines or workstations are the ‘nodes’ in networked production that are robust, flexible and can self-organise based on the demand patterns. Each node ‘knows’ exactly which component is needed at which moment and what instructions and tooling is needed to execute a process step.
This new paradigm takes inspiration from successful methods and practices from other industries and domains, e.g., Agile principles of software development with quick turn-around times, faster data-driven decision cycles and lower planning overheads. And from ‘Uber-ification’ or network economy where complex fulfilment strategies can be implemented using flexible assets and advanced analytics. Networked production utilises modular, multi-functional machines as nodes and the production strategy is derived by creating the material flow or a logical ‘line’ from the intersection of prioritised order pipeline and available machine capacity.
There is a continuous stream of incoming order and continuous fulfilment by 24x7 operating machines with failure redundancies. When a machine is about to fail (predictive analytics), the entire setup – configurations, tools, programs and instructions is cloned onto another machine that can quickly take over the load while the original machine is repaired. For every product in the backlog, the required throughput is constantly matched with the available throughput at different machines for sequencing and making decisions like schedule, material flow, switchovers and batch sizes. This also includes adjusting for changes and disruptions in demand patterns, order priorities, material availability and other production or supply chain disruptions. The resultant operations in not a marching band but an orchestra of production units participating in implementing order streams deployed as ‘production loads’ on the manufacturing infrastructure.
Consequently, this level of complexity cannot be managed by humans and requires a supervisory intelligence – An AI-based Production System that can continuously optimise the order backlog, material flow and other parameters to get the maximum throughput at maximum OEE. Organisations have already started laying the foundations for such intelligence. There is a drive in many large manufacturing organisations to create master data repository combining data throughout the enterprise like engineering data, manufacturing data, logistics and service data and customer data from disparate systems like machines, shop floor, PLM, ERP, CRM, DMS and many other enterprise systems into a single data warehouse or lake. Over time, these data repository will generate analytical insights and data assets like reports, dashboards and models to provide an accurate picture of ongoing production.
But analytics can only create so much value. The next generation of agility, flexibility and responsiveness also requires creating new roles and changing the decision structures to enable frontline employees make faster decisions with data and autonomy. Organisations must transform from an approval-seeking culture to an entrepreneurial culture where employees are empowered to take decisions and supported in decision analysis and guided for better decision-making in future.
This cultural shift towards data-drive ways of working and use of AI in production will have a great impact on the jobs and quality of work. Against common perceptions, this new paradigm will create many new high-value jobs. Let’s take Formula 1 for example – before 1970, there was no telemetry in the F1 cars and most of the driving and pit stop decisions were made by the driver. Fast forward 50 years, cars transmit more than 3 TB of data for every race from 200+ sensors which is analysed by 30 trackside analysts and anywhere from 30 to 200 analysts back at racing headquarters. Similarly, when a supervisory intelligence takes over production coordination, it sets a new and much higher productivity baseline than what’s possible today. Manufacturing facilities will require large mission control teams to maintain the infrastructure and quickly resolve problems to maintain such high productivity levels. This mission control team will consist of analysts and experts in machines operations, supply chain, logistics, maintenance, production and many other functions to monitor the production in real-time and shape the production strategy. This is already a reality in the mining industry where
equipment OEMs like Sandvik and Epiroc are working with global mining companies to collect data and optimise various mining processes using automation and AI.
The growth of network economy also creates new opportunities for the manufacturing industry and pave the path for Manufacturing-as-a-Service. Today, customers don’t pay for the products, they pay for convenience and express deliveries. As manufacturing infrastructure becomes more flexible and networked, facilities will be able to manufacture a broad range of products and easily shuffle order pipelines. It enables manufacturers offer ‘dynamic pricing’ to command higher price for express orders and a lower price for standard orders produced in spare capacity for better utilisation. Signify (formerly Philips Lighting) offers a similar service to its consumers for ordering custom-designed luminaires 3D-printed at scale at competitive price. And many more are to follow.
In conclusion, a half-hearted application of technology ‘band-aid’ on the old, legacy manufacturing strategy and structure is not a solution to upcoming threats and challenges. The industry is in dire need of a new mindset and strategy. Networked production offers a solution to managing uncertainties and challenges of variability and volume. It also offers an opportunity to tap into the burgeoning network economy through manufacturing-as-a-Service. If we possess the wisdom to reflect on the recent past and apply the lessons to manufacturing, the future is clearly networked, agile and servitised and waiting to generously reward the early adopters.
Somil Gupta, Digital Technologist, Speaker and Entrepreneur, is an AI Strategy Advisor for Nordic manufacturing, automotive and retail companies in IoT, AI and Industry 4.0 topics. Earlier, he led the business development for digital topics for Bosch in the Nordics. He consulted C-Suite executives in understanding the developing and monetising digital solutions, e.g., I4.0, IoT, AI and Blockchain. Somil consults clients in bridging the gap between their digital vision and current capabilities to realise that vision.
