Since the birth of the concept, in the industrial field, “digital twin” has brought unattainable and difficult stereotypes to employees because of the universality of application scenarios and fields, the diversity of technologies involved and the huge differences in implementation methods.
1.What is digital twin? What are the common misunderstandings?
2.How to make better use of the value of digital twin technology?
The content of this paper is compiled from the speech delivered by Dr. Lin Shiwan, CO chairman of American industrial Internet Alliance Technology Working Group, architecture task force and digital twin interoperability task force, at the 2021 (10th) global automation and manufacturing summit.
From the perspective of production and operation scenarios, Dr. Lin Shiwan peeled the cocoon layer by layer to help deeply understand all the knowledge points of “digital twin” from concept to technology application.
Three misunderstandings of digital twins
The high complexity of industrial system, the diversification and variability of production environment and working conditions, and the fragmentation of industrial knowledge accumulation make it even more difficult to carry out the application of various new technologies. In order to avoid detours, enterprises must avoid these common misunderstandings before stepping into the digital twin “track”:
Myth 1: 3D simulation data display ≠ digital twin!
When people refer to digital twins, they often think directly of three-dimensional simulation display, especially some data displayed in the background of three-dimensional simulation. This understanding is one-sided.
The three-dimensional simulation display is only the mapping of three-dimensional space. The result is to make the display of data, status or events and the browsing of the system more intuitive. It can give visitors a cool impression by displaying it on a large screen. However, this is only an expression of man-machine interface. Without the support of digital twin data and algorithms, these displays are of little significance.
Of course, if the digital twin has rich data and powerful algorithms, and the corresponding industrial app has powerful functions, most of the production, operation and maintenance operations can be solved automatically without manual intervention, the necessity of three-dimensional simulation display will also be weakened. And if there are algorithms that can effectively monitor, alarm, record and process the equipment status, the existing “virtual Patrol” will have little effect.
Of course, in addition to meeting the previous process regulations, 3D simulation, as a kind of digital twin model, establishes a visual model based on the actual spatial parameters of physical entities and the topological relationship of space, especially combined with AR, will continue to play a unique role in design, equipment disassembly and maintenance operation guidance, and the replay of operation events of moving equipment.
Myth 2: traditional analog simulation model ≠ digital twin!
Simulation is to establish a model in the computer, reproduce the essential process in the physical system, and carry out experimental calculation by adjusting the model input and control parameters. It is used to study and evaluate the existing or design system characteristics and behavior, and find a feasible or optimal design. It is widely used in the manufacturing industry.
“But generally speaking, the cost of actual system construction is high, and some tests take a long time or are dangerous. It is obviously a means to get twice the result with half the effort by using computer simulation.”
And the actual system is generally very complex and limited by the current technology. When establishing the simulation model, most of them need to be simplified. Only focusing on key factors, ignoring secondary factors, or simulating only some aspects of the system can meet the requirements of verifying whether the design results meet certain design requirements (such as safety production) in the design process, but the calculation accuracy is not easy to meet the requirements of supervision and Optimization in the production process.
In addition, most of the simulation model software used in the design process is based on the simulated data, which is used in batches and is not connected with the real-time data on the production site. It is not easy to support continuous flow computing in terms of computing performance, and it is difficult to support the control and optimization of the production and operation process in the digital twin.
Generally speaking, analog simulation is an important supporting technology of digital twinning. The simulation model in the design process is only an important part of the digital twin algorithm model, which is not equal to the digital twin itself.
Myth 3: Digital twins ≠ “alchemy”!
“Digital twin is not a golden touch, but an opportunity for continuous improvement.”
The effectiveness of digital twinning depends on the complexity of equipment or production process, the difficulty of mode judgment, state prediction or optimization strategy calculation, and the completeness and accuracy of sensor data collected from the field for calculation. Just as people’s cognition of the real world has a continuous learning process, the digital twin’s grasp of the physical world also has a process from rough to fine.
As a methodology, a framework technology system, and even a middleware for the digital architecture of production and operation environment, digital twin is not a magic that can be achieved overnight.
Enterprises should not take digital twins as a golden touch technology, and unrealistically expect digital twins to solve difficult problems in one step. The core of digital technology is software. In essence, it is a technology of continuous improvement, while software without continuous iterative improvement has no vitality.
“Digital twin is an idea and methodology of industrial digitization, a technical system and technical ability.”
1.As a basic concept, the core of digital twin is to use sensing data and computing to realize in-depth cognition and intelligent decision-making of physical entities, effectively control and manage these physical entities and better serve mankind;
2.As a technical concept, digital twin is to realize the mapping of physical entities in the computer, which belongs to the problem of computer engineering and the category of software engineering. To map a physical entity, it is necessary to calculate its state, such as mode judgment, root cause analysis and state prediction. There can be a variety of implementation methods.
Computing is the core of “digital twin”
In the early stage of computer development, we only had assembly language, which was almost programmed with the machine language of the computer processor. It was a long string of instructions without any structure. It was brain-consuming, time-consuming, error prone, difficult to check, and there was no reuse.
Later, high-level languages, such as FORTRAN and C, introduced logic and data structure, and could organize the code in the way of function according to the task. Since the function can be called in multiple places as a code unit, the reuse of the code is realized.
However, these programming languages belong to the process programming paradigm. For large-scale program systems, there are many functions that need to be decomposed, the call relationship is complex, the difficulty of design, development and maintenance will become very high, reuse and reuse will be difficult, and the cost of software development and maintenance will become very high.
This way makes the software design more in line with people’s way of thinking in real life. It not only makes the design more structured, but also enhances the reusability and maintainability of the software, improves the software quality and reduces the software development cost.
As a set of methodology
As a methodology, object-oriented programming paradigm (OOP) can be used for reference in digital twin design. The design of digital twin can extend the object-oriented programming paradigm to the entities of the physical world, represent the physical entities in the way of objects in the software, and establish the corresponding software objects for each physical entity.
In digital twin, the attributes and states of physical entities are represented by data, and their behavior is simulated by algorithm model. In addition, the object-oriented design method can support the use of unit objects to build more and more complex systems in the way of building blocks. Starting from the component number twin, it can build the digital twin of equipment, units, production lines, workshops and even the whole plant, becoming the digital representation of the whole plant.
The digital twin is constructed in this way, which structurally reflects the attributes, states and behaviors of physical entities, shields the complexity of the site, and simplifies the construction of digital industrial applications.
As a technical framework system
Digital twinning maps entities in the real world and calculates their state, such as mode judgment, root cause analysis and state prediction — for this purpose, digital twinning should not be used as a terminal application for users, but as a supporting technology for the application of digital chemical industry, It can even be built as middleware in these application architectures.
In order to map the state and behavior of equipment on the production site, digital twin needs to establish corresponding image objects in the software. If we learn from OOP, we need to set the parameters corresponding to the equipment in the digital twin, including attribute, status, instruction and other types.
On this basis, these parameters are connected with the data collected from the equipment one by one. If a certain operating parameter of the equipment changes, it will be reflected on the corresponding digital twin almost in real time. This is a process of sorting out equipment data, and it is also a process of precipitation of industrial knowledge into software.
After that, the algorithm model is used to analyze and calculate these data and map the behavior of physical entities. For example, determine or predict the operation mode of the equipment, whether it belongs to the operation abnormality, whether it meets the process requirements, whether it meets the energy efficiency requirements, etc. if the abnormality is determined, analyze the root cause of the abnormality and the solution strategies, which will use the mechanism model and data algorithm model.
However, the final solution must take the operation characteristics and behavior of the equipment mapped by the digital twin as the input, make appropriate decisions and implement them in combination with the business logic and production rules of production and operation management, especially the principles and methodology of lean management. These logics can generally be realized in the way of app.
This is the role of digital twin as a technical framework system and Middleware of digital industrial application architecture: it can start IOT data and connect the site; The industrial app is used to convey the insight and cognition of the site state and behavior, and support the decision-making of controlling the site.
Characteristics and functions of digital twins
At the level of digital production and operation, after more than 20 years of information development, manufacturing enterprises have adopted various design software and CAD in the process of product and process design, and realized the automation and information of business management process on the basis of original automatic production.
However, most industrial software is mostly used to manage business processes and planned production under normal conditions. In the production process, a lot of work needs to be done to make full use of a large number of equipment data on the production site, further reduce the failure rate of equipment, improve the overall equipment efficiency, improve product quality, reduce energy and material consumption, and ensure the compliance of the production process.
At this stage, the judgment of abnormal production, root cause analysis and coping strategies mostly rely on operation experience and manual operation. In the new round of digital development, whether in product and process design or in the production process, how to use the data from the physical world to make accurate judgment, intelligent decision-making and timely implementation has become the focus of enterprises.
Enterprises need to get through the efficient collaborative operation of multiple equipment upstream and downstream of the production process, integrate the principles and methods of lean management into the digital production operation management, and pursue the standardization, stability and continuous improvement of the production process. The birth of digital twin technology can make use of the digital virtual replication of physical entities to solve practical problems.
When we use the digital twin methodology to establish a new technical framework, from the perspective of OT, we take the equipment as the main object for modeling, define the characteristic data of the equipment, and establish an algorithm model to judge or predict the behavior of the equipment. In this way, the equipment experts can easily define the characteristic data of the equipment, and the experts who understand the equipment operation and production process can cooperate with the algorithm engineers to establish the algorithm model, so that the two sides can be integrated through the digital twin structure.
In this way, digital twins are established for various devices, and data definitions and algorithms are encapsulated into software components that can be reused in multiple plug-ins. Equipment suppliers familiar with the equipment or industrial design institutes familiar with the process can also independently build and provide digital twins of a certain type of equipment or production process, and can support multiple scenarios.
In addition, the digital twin itself is decoupled from other business software components (such as production rules or user interface) in the software structure. When the iterative algorithm of digital twins is improved, such as improving the accuracy of a certain calculation, it is only necessary to update the digital twins independently without modifying or changing the business software code.
In a word, using digital twins, users have the opportunity to build a technical framework that can effectively support the integration of digital technology and industrial knowledge, promote the precipitation, accumulation and continuous improvement of industrial knowledge, and promote the wide reuse and sharing of ecological industrial knowledge.
Make good use of “digital twins”: reuse and accumulation are the key
To realize the value of digital twin, we need the integration of multiple technologies and knowledge, especially the integration of digital technology (it) and industrial knowledge (OT). For manufacturing enterprises, they often encounter the dilemma of weak it technology resources and weak ot knowledge accumulation. There are many obstacles to using digital twins alone to promote digital production and operation management.
To establish an effective digital twin in the industrial field, we need the joint efforts of industrial ecological partners, including digital technology suppliers, equipment suppliers, industrial knowledge suppliers (such as industrial research institutes and design institutes) and enterprises as Party A to build and improve technologies and systems.
In this process, it is particularly important to promote the reuse of digital twins in the industry ecology. It is conceivable that in the scenario of each enterprise, the software is used to deeply and accurately characterize and map the characteristic states and behaviors of equipment and production process. The threshold of separate construction is high, the cost is high, the accumulation and promotion is slow, and the result is twice the result with half the effort.
Tips:
At present, many enterprises still use the traditional software project-based implementation management mechanism in building digital application systems, which belongs to one-time project implementation. After the solution supplier wins the bid for implementation and delivery, there is generally no consideration or opportunity for subsequent maintenance or update after one year of warranty. Next time, it will be difficult to accumulate and improve.
Enterprises need to maintain a long-term cooperative relationship with solution suppliers, provide corresponding investment to support the maintenance and improvement of software applications, leave a long-term way for the established digital solutions and avoid one-off trading. At the same time, enterprises can also consider using the products of third-party suppliers to avoid unnecessary self built projects. In this way, they can leverage the products of suppliers to maximize reuse and accumulate improved results.
If the corresponding digital twin sharing technology framework and value trading mechanism can be established in the industry to promote the reuse of digital twins, it will reduce the start-up cost, accelerate the improvement of its effectiveness, quality and performance, and improve the cost performance of digital twin applications.
On a broader scale, software can be said to be used. When a software application has more users, more scenarios and more problems, its functions will become rich and powerful, and its stability will be improved. This is the result of the accumulation of reuse.
This is why third-party software products are generally better than self-developed ones, which is caused by the differences in space (scene) and time accumulation, especially the advantages of cross space reuse. Therefore, reuse is not only beneficial to software suppliers, but also beneficial to software users.
Post time: Jan-17-2022