Here is a framework for digital twin’s application including conceptual design, elaborate design and finalized design. According to changes with the potential happen in each design stage, a modular approach is required to build flexible digital twin to save workload and time for developing new digital twin.
Through a case study by using the digital twin in factory design, hidden design flaws are discovered and solutions are proposed. Moreover, it can be concluded from this case that the duration for building digital twin model is greatly reduced by using modular approach, which improves the feasibility for applying digital twin to changeable factory design. Different industries have different characteristics and modules may different in different industry application.
Digital twin offers a practical way to study smart manufacturing by focusing on building digital twin for smart machine, smart workshop and smart factory. Not much attention has been focused on the application of digital twins to factory design. A lot of work has been done in production, but less so for layout and equipment configuration applicable to factory design.
Design frequently changes from initial version to final version. These characteristics carry challenges to digital twin introduced to factory design. Traditional factories are increasingly transforming into smart factories, and the demand for design new factories is increasing simultaneously
The concept of digital twin shop-floor has been introduced into the production floor, opening up the bottleneck between the physical and information worlds through a combination of physical and virtual models Operational control of Automated Guided Vehicles inside the shop floor can take advantage of digital twins to increase operating efficiency.
Smart assembly equipment and shop floor production lines can also increase effectiveness by using digital twin technology. There has been investigation into connections between people and digital twin in production area. This includes interactions between large amounts of data and information in production and the substitution of manual methods for a digital twin to improve planning and management in industrial production.
Digital twins also have a wide range of applications during the machining environment. Traditionally, the design of a factory is considered to be an optimization algorithm design problem with the capacity as the final optimization goal and the equipment or the area as the restrictions.
However, in a very dynamic real production environment, the traditional algorithm cannot effectively solve the optimization scheme. Therefore, more complex algorithms have been designed, but in a practical factory, the dynamic behavior of the manufacture system is difficult to be predicted in mathematic equations.
For this reason, the optimization is valuable only when the manufacture system is constrained by some assumptions. Consequently, these algorithms can solve simple dynamic problems, but are not very good at modelling complex problems.
Because of the complexity, there are studies of factory design using discrete event system simulation like integrated simulation model to predict the life span of individual parts or a simulation-based approach to integrate mathematical algorithms to balance the operation performance and planning cost.
Investigations have been applied to simulate the factory layout and help the designer to experience the designed virtual factory. However, factory layout is not the only issue of factory design. It also includes detail works such as capacity calculation, machine utilization, number of machines, designing of the logistics and calculation of its efficiency. etc.
These issues are difficult to present with traditional algorithms or just visualized by virtual reality technique. Regarding the design phase of a smart factory, the majority of businesses remain in a relatively undeveloped stage, using inadequate historical data to predict key performance indicator KPI, such as throughput, equipment usage and storage capacity.
Since high dynamic and random features will be carried out in the future production of the designed factory, the KPI prediction can only derived by limited calculation or empirical ways. This traditional factory design approach cannot utilize all historical production data, consequently may leads to design flaws in the design phase and might be huge problems in future production.
Consequently, digital twin might be a suitable solution because of its fidelity to physical world. Big problems may be avoided in early design stage and the performance shown by digital twin should be ideal.
Digital twin can undoubtedly increase the feasibility and quality of a design project. Digital twin forces the designer to design the factory in detail, which includes factory layout, material handling, buffer capacity, worker shift etc., coordinating these factors to make the designed factory efficient and strong.
Typically, factory design includes several main design stages: conceptual design, elaborate design and finalized design. Conceptual design is the first design stage, it focused on the designing the concept of new factory, which includes plant layout, capital investment and throughput predication. In conceptual design phase, digital twin can help the designers and shareholders to verify the design concept, predict throughput and rate of return on investment by simulating the process in a relative rough way.
Elaborate design is the second design stage and further conceptual design, includes machine configuration, process design, production line or production unit configuration, material handling system configuration and work shift configuration. In most of the cases, the objective of elaborate design is to further and validate conceptual design. Consequently, digital twin in elaborate design phase is to help the designers to elaborate design the factory and integrated validation.
Finalized design is the final stage and links to construction. In this stage, the machine and logistics unit control strategy will be designed and the whole manufacturing system needs to be integrated. Because the virtual factory corresponds to the finalized design is the most similarity to physical factory in the future, therefore digital twin has the most fidelity to physical world in this stage.
Digital twin connects with multiple control tools like manufacture execution systems and programmable logic controller. Due to the connection, digital twin emulates the manufacturing and logistics control strategy, and help the designer to debug the control and make decision.
The framework of digital twin’s application on three design stages enables designers to connect with suppliers, shareholders and design documents with considering the simulation results of digital twin. Based on the output of digital twin, designer evaluates the current design and decides whether it can be approved.
In conceptual design stage, digital twin mirrors the design concept and visualized the concept by animation. The animation helps the designer to further the concept in a comprehensive way and clearly states the design to shareholders and suppliers. In elaborate design stage, digital twin mirrors the detail configurations of the factory and validate whether the configurations can carry out plenty of throughput via simulation.
In finalized design stage, digital twin mirrors the final approved design and emulates it as a virtual factory. Due to the emulation, the control strategy and tool design can be confirmed.
Digital twin progresses from conceptual design to finalized design while the fidelity to physical world also moves forward stage by stage. Fidelity of digital twin High fidelity to physical world is one of the most important features of digital twin.
However in factory design, the physical world is vague and different in the three design stages. In conceptual design stage, the physical world is the uncertain concept hidden in awareness of designer concepts. It can be definite when the factory layout is produced. Traditionally, the factory layout is two dimensional diagrams, investment and throughput can only be calculated with basic calculation because of uncertainty.
Digital twin can make the concept solid with three dimensional animation and help the designer to think the concept in a relative detail way. Moreover, investment and throughput can be predicted with algorithms or knowledge based approaches embedded in digital twin.
Consequently, the fidelity of digital twin in this stage is mapping the uncertain design concept. Since most data is uncertain, 3D animation is the most useful feature of digital twin in conceptual stage.
The physical world in elaborate design stage is more definite than conceptual design stage. The factory layout, machine layout, material handling system, working shift, and even equipment efficiency are defined in this stage.
Consequently, the physical world in this stage is the product-making mechanism in the designed future factory. For example, the product can only be processed only if the product is transported to the destined machine. The corresponding machine, fixtures, tools and skilled workers need to be available so that the processing conditions are met. Otherwise the product has to wait in queue and may jam the working flow.
Many design parameters are critical in this stage, such as equipment number, worker number, and buffer capacity. These parameters are difficult to be accurately validated because manufacturing process is featured by mixed up dynamic behavior.
Embedded with discrete event simulation, digital twin can test different parameter combinations step by step. Consequently, accurate parameters validation can be offered and helps to make decision.
In finalized design stage, the physical world is the physical entity in future factory. The features of physical entities are not only appearance but also internal control strategy so the main connection between virtual model and physical entities is the control.
On the other hand, control is decentralized into different physical entities in smart manufacturing. The fidelity of digital twin in this stage is to emulate decentralized control of physical entities and the integration of them. Based on the emulation, digital twin helps the designer to evaluate control strategies and find the best one to match designed factory.
- Digital Twin provides links between real/digital world by mapping measured data, and virtual representation
- Digital Twin functions as element of product itself and delivered with components or as stand-alone service forming business models
- Digital Twin closes feedback loop back to real system and early lifecycle phases for development of new versions or product generations
- Digital Twins can be embedded in the component or device and accompany the real twin once manufactured
- Digital Twin of product combined with production system twins include engineering/virtual tasks and operation of the plant itself
- Digital Twin utilises sensor data and collected information on product use supports optimized application of product itself
- Digital Twins valid for multiple instances enhanced by production information collected to generate fleet data for applications and services
- Digital Twin template specified during the conceptual design phase to describe how different components are linked together and interact
- Digital Twin functionality identifies possible root causes of malfunctions in contrast to pure detection
- Digital Twin connection with operational data offers wide range of new services from failure detection and diagnosis to faster product improve/develop