Leading Digital Lean in the Smart Factory

 
 
Engineer.jpg

The performance of a smart factory is often measured by its ability in collecting, analysing and aggregating data into meaningful information. This data is then used to predict future events, derive decisions and influence the future state of a process. A smart factory gets strongly influenced by the manufactured product itself, including its design, functionality, unique information, integration, software and human knowledge and experience. Each of these areas, must seamlessly fit within a lean digital roadmap which paves the way to the digital future by driving speed, flexibility, quality and productivity to the limits in a secure environment. We call it lean digital roadmap due to the fact that lean principles must be applied before digitalisation comes into the picture. When we think about a product, presently it is conceptualised as digital with a 3D-model as the base. In addition, it must be enhanced to a full functional description of the product in multi dimensions. This is required to derive and test manufacturing and service strategies. All of this data and information needs to be stored in a data back bone to guarantee that every involved party has access to the same single source of truth.

The journey to become a smart factory should always begin with becoming a lean factory first and can be characterised by several lean rules stated in 1999 by Spear and Bowen. One rule is known as, that all work and process steps, shall be highly specified as to content, sequence, timing and outcome. Processes must be described without any waste, and artificial intelligence and machine learning can be used to make them even more efficient. On the other hand, waste should not be automated or digitised either. As an example, a full automation line can be analysed using the same methodology as analysing material flow by value stream mapping. Simulation tools can help to analyse existing data to evaluate the process. While counting the inventory line to get transparency in regards of work in progress and the relation between through put time and cycle time one can evaluate how much money has been spent in a full automation line to automate non-value add contribution. In our studies we found out that between 50-70% of automation costs are related to handling, transportation and manipulation of the product and this is in general a non-value add. The objective must be to streamline the production process steps even before automating them and measure the effectiveness by the relation of costs spent for value added and non-value added automation. In the same way, business processes can be analysed in regards of the information handled in the work stream for engineering or administrative tasks. Only streamlined processes can be automated and digitalised efficiently. With digital methods, the lean vision could really become true due to virtual implementation of efficient and proved concepts of machines, lines and processes that enable a fast ramp-up and line utilisation in the real world.

Another lean rule states, that every customer-supplier connection must be direct and that there must be an unambiguous yes or no to send requests and receive responses. This can be realised very efficiently with defined digital supplier interfaces with standard communication protocols for tracking and tracing. So called “digital knowledge management” supports the preventive and operative supplier management with sustainable benefits as a contribution of the business success. The full digital integration of suppliers and the early involvement in engineering and series production generates data transparency over the entire lifecycle, opens new business models for service and improves data analytic potentials due to new data sources. In this context, thinking “lean digital” also means that the pathway for every product and service must be "simple and direct.” This aims for shorter lead times during product development and production introduction by utilisation of the digital twins' product, production and performance. New services via internet and cloud computing allow direct access to valuable insights about utilisation of the product and services. From a plant development point of view, a very important lean rule states, that any improvement must be made in accordance with the scientific method, under the guidance of a teacher, at the lowest possible level in the organisation. Personal experience shows that well established lean performance systems which are describing the principles, methods and tools are able to improve operations by about 3-5% annually. The nature of lean principles is to drive continuous improvement programs on the lowest possible hierarchical level. Today, software and information technology is used in a limited way or using proprietary systems with no interconnection to other function, departments or even plants. In this area, a re-thinking is necessary to benefit from future digital solutions. While productivity benefits via continuous improvement programs will melt down by better designed and virtually tested products, virtually commissioned on optimised production lines, the integration of digital twin solutions will drive further productivity while reducing IT and software investment risks. A common reason why some companies struggle in their smart factory initiative is that their vision of a smart factory is more technology than business and process oriented. Those companies also often underestimate the effort and investment required to achieve their smart factory vision. Some of the reasons why artificial intelligence and reinforced learning is so hard to put into production are that it requires an accurate simulation of the environment, or a full digital twin of product, production and performance, which is far harder for robotics and automation than for other areas. Typically, such algorithms require profound domain knowledge and a massive number of training iterations before they converge into meaningful information making the outcome uncertain, expensive and in many cases difficult to reproduce. It looks like, that many production and automation challenges can be solved with artificial intelligence and the investment risk can be lowered by following principal lean implementation guidelines while rethinking the basic continuous improvement approach. Digitalisation enables customised information on the right level of employees. This can lead to powerful solutions with immediate impact while closing the gap between ideation and production with digital twin solutions.

About the Author

Dr. Gunter Beitinger is Vice President Manufacturing at Siemens

 
Siemens.png
 
 
Richard Smith