How edge computing improves cost efficiency for manufacturing companies

1. Solve problems around the essence of manufacturing

For manufacturing companies, the operating level can be evaluated by profitability, which can be further broken down into the production of qualified products with high cost efficiency, that is, stable product quality and continuously decreasing costs, and fast delivery capabilities. In other words, quality, cost and delivery are the core topics of the manufacturing industry, while for intelligent manufacturing, it is about solving the optimization of materials, time, energy consumption, etc. in the process of "global coordination", thereby further improving the cost efficiency space.

In terms of product processing quality and speed, they have been maximized under cost constraints. Therefore, new optimization space comes from "integration" and "collaboration". The other dimension lies in the challenges brought by personalization. In order to solve the cost optimization of collaboration and the cost efficiency problems brought by personalization to production, the core issue of edge computing is to aggregate and globally optimize information flows.

Figure 1- Returning to the essence of manufacturing to solve operational level problems (Uya Information Technology)

As shown in Figure 1, from Ms. Fu Yuan of Youye Big Data, data includes basic production and operation improvement, operation, health management of machinery and equipment, and then global optimization, and finally autonomy, with learning analysis and decision-making capabilities. In this process, various data-based tasks will be encountered. These tasks are what edge computing focuses on. In addition to basic control, various levels of problems including data connection, aggregation, storage, presentation, and learning need to be completed by edge computing. Therefore, edge computing is not a computing problem, but a series of architecture and specification interface problems around tasks.

2. Data connection: the key to connecting IT and OT

It has to be said that the first obstacle encountered in promoting edge computing is the "connectivity problem". It is not as easy as imagined for the IT world to access the OT world, even after 30 years of development of bus technology. Figure 2 is a 2014 World Economic Forum survey and assessment of the difficulties in promoting IoT in North America, Europe, and the whole world. Although it is not a new result, it is not outdated in reality.

Figure 2- World Economic Forum 2014 research report on the difficulties in the development of IoT

The lack of interoperability standards is the key factor affecting "smooth" access to data. On a single machine, data from various fieldbuses or based on real-time Ethernet can be accessed smoothly. However, when "collaboration" is required between machines, the data is divided into isolated islands by different buses. Breaking down isolated islands seems to have been mentioned for many years, but it is still impossible to achieve in real factories.

Figure 3 - Development of fieldbus development

For IT vendors, accessing OT devices encounters bus obstacles, because you have to develop interface conversion hardware for each bus and write drivers. Even if real-time Ethernet remains physically consistent, you still need to write variable address table mapping and reading programs. All of this makes access between IT and OT "uneconomical" - this is a very critical point. The key to whether technology can be promoted lies in this. If there is a lack of economy, neither technology providers nor users can benefit from it, so it cannot be promoted.

OPC UA and TSN have been entrusted with this historical responsibility. As the IEC62451 standard, OPC UA has been formulated as a semantic interoperability specification by RAMI4.0, IIRA, etc. It solves the problem that various entities in heterogeneous networks can communicate with each other through the same semantics.

TSN solves the real-time data transmission mechanism required for OT periodic control tasks in the network, as well as the transmission problem of large-capacity non-periodic data of IT. Through VLAN switches, two different types of data can be transmitted in the TSN network, while data can be transmitted through standard Internet methods when leaving the TSN network. The biggest benefit is that IT can have transparent access to OT.

3. Vertical Layers of Edge Computing

Figure 4 shows an edge node (Edge Node) of a field acquisition station (based on OPC UA TSN) to a control (Embedded Edge Controller), which can achieve secure data security and aggregation to the edge computing server (Edge Computing) and then to the vertical application process in the cloud through OPC UA/MQTT.

Figure 4-Vertical architecture and technology applications of edge computing

In addition to the physically visible implementation architecture, how data is applied is also the key to the whole problem. For factories, edge computing is mainly used to implement some local applications, which may include the following applications:

(1) Application of data collection and collaboration

For the most typical factory applications such as OEE calculation, it is necessary to calculate the quality, machine operation time related parameters, utilization rate, etc. and display them to the factory management level in real time so as to observe the status globally.

(2) Planning and analysis issues

For many factory applications, it is necessary to consider online planning issues, such as glass cutting, how to conduct online inspection, classify grades, and plan them according to the system orders in the CRM, so that different grades of glass can be cut and processed as needed and diverted to different packaging lines. For printing factories, how to reorganize different orders to minimize material waste are all issues that need to be solved by edge applications.