Smart manufacturing and the smart factory enable all information about the manufacturing process to be available when and where it is needed across the entire manufacturing supply chains and product lifecycles. The phrase ‘smart manufacturing’ is now widely used to represent a new wave in manufacturing innovation. There are several descriptions of smart manufacturing in the published literature, which refer to the use of large collection of digital data and their manipulation in the manufacturing space. Recently, an increased use of sensors in the manufacturing floor for measurements to collect data has been noted. Their addition to data pool is being pursued under the umbrella of IoT.
The ‘black box’
Processes using CNC, robotics, automation, etc intrinsically combine data manipulation, together with physical processes. It is the repetitive use of all processes that distinguish manufacturing from research, design, product development, etc. The information processes are ubiquitous and involve many activities that occur inside and outside the manufacturing plant, which are also described as front-end and back-end operations. Repetitive engagement of these processes invariably creates large sets of data/information. They are also generic and hence, mostly independent of the physical process involved. The data pertaining to information processes are readily available to access, aggregate, curate and analyse. Much of the smart manufacturing attempts to exploit this information-driven data and their manipulation to gain efficiency and better economics.
Physical processes are unique. Conception and execution of the physical processes require a unique set of data. But such data is generally hidden from plain sight. Hence, they remain in the black box known and are accessible to only a few closely associated with the process. This dichotomy of information sources is illustrated in Fig. 1. The limitations of treating the physical processes as a black box can be overcome when we recognise that every physical process is an input/ transformation/output system. The details of physical processes can be captured using the STIMS Diagram (Fig 2). The system view of physical processes used for manufacturing is further illustrated in a generic fashion in Fig 3. Such framework enables capturing a set of vital data unique to the given physical process. This unique data set pertinent to grinding processes is illustrated in Fig 4.
When a new machine is installed in the shop floor for manufacturing, it is not just the machine, but the entire process solution that is installed. This implies that every one of the parameters described in Fig 4 has a specific value (data) for the newly installed process solution. The process cannot function in the component manufacturer’s shop floor as envisioned unless all these parameters are repeated identically and as documented during process validation, usually in the shop floor of the machine tool suppliers. In other words, when a new machine tool is installed, the entire system should be specified in quantitative terms, not just the parameters pertaining to the machine tool. The collection of such unique data pertaining to the grinding process can start with comprehensive documentation of every part of the STIMS Diagram pertaining to the given process. We can consider this as the “birth certificate” for the new process installed in the shop floor.
After the unique data pertinent to the system – the System Document–is captured, then any changes to the system should be tracked in a single database. Invariably, every change in the system will be reflected in the in-process signal. Such signals are generated every time the grinding wheel touches the part., acquisition, analysis and interpretation of in-process signals based on certain rules are the skillset of process engineers. Such well-trained engineers proficient in process science and system view of physical processes are very few. IMTMA, IIT–Madras and Micromatic Grinding Technologies (MGT) have been collaborating with STIMS Institute in this education and development effort over the past few years.
In the world of smart manufacturing, the initial data as well as every change to the system necessitated on the shop floor, along with their corresponding in-process signal, will be available in a single searchable digital database. With such a database, rule-based process management can be automated. Tools of Machine Learning and Data Analytics can increase the efficiency of even the best process engineers. With this database in place for the manufacturing process, reports can be generated tailored to the needs of various levels of manufacturing personnel. Also, this vast array of data can be integrated with the front and back-end data sets for further use in analytics based on data science.
Data and manufacturing process
This ambitious program to aggregate data from physical processes and automation of such practices is now in progress in a collaborative research effort between STIMS Institute, Advanced Manufacturing Technology Development Center (AMTDC) at IIT–Madras and Micromatic Grinding Technologies. The project funded by DHI–GOI is titled, ‘Automation of Grinding Process Intelligence (AGI)’. This project was initially focused on grinding processes. The longterm goal is to develop a digital platform applicable for all physical processes used in manufacturing.
The collaborative project between researchers of IIT — Madras, MGT and STIMS Institute is specifically being guided by Prof Ramesh Babu, IIT–Madras and N K Dhand, CEO, MGT. As part of the project, IoT-enabled grinding machine has been developed for automation of in-process signal acquisition, analysis and remote access. The work is in progress to integrate this together with a system-level database for process management, including generation of health check reports. The researchers, their work and the industry-academia collaboration forms the backbone of this project.
Relevant data exists to deal with the manufacturing process as a system, but such data is spread across various resources, both inside the component manufacturing company and with their many suppliers and the component end-users. Some of the information required for system output is not often documented at the unit process level. It should not come as a surprise to anyone that most of the physical processes are treated as ‘black box’ and without any relevant information pertaining to the in-process signals and analysis. These basic ailments must be addressed immediately in order to move towards new directions for meaningful smart manufacturing.