The Data Centric Architecture of a Factory Digital Twin
This paper presents a framework for implementing factory digital twins ( FDTs ), focusing on four applications : ( 1 ) debottlenecking , ( 2 ) engineering design / retrofit analysis , ( 3 ) facility fit , ( 4 ) scheduling / planning . These provide substantial value for FDT implementation . The ability to " test-before-invest " fixes mistakes in the virtual realm before committing capital , significantly reducing risks and costs associated with real-world trials .
Our research demonstrates that FDTs serve as effective proxies , allowing complex manufacturing processes to be optimized – both existing and yet-to-be-constructed .
The paper explores how FDTs identify critical data , leverage optimization , and enable rapid scenario testing that would be impractical / impossible in physical systems . We summarize the key role of the Resource Task Network ( RTN ) which provides a structured means of organizing manufacturing data , describing the manufacturing process , and underpinning the FDT .
Importantly , we address the limitations of FDTs , acknowledging that a single digital twin cannot encompass all aspects of reality . Instead , we use various specialized digital twins to model different aspects of the manufacturing process .
We provide case studies demonstrating how this approach has been successfully implemented , resulting in improved efficiency , reduced downtime , and enhanced innovation .
This article contributes by offering a step-by-step approach to FDT implementation , focusing on applications that are impactful to manufacturing processes . Our findings have significant implications for both practitioners seeking to implement digital twin technologies and researchers exploring the future of digital manufacturing .
Keywords : Factory Digital Twins , Digital Thread , Manufacturing Innovation , Industry 4.0 , Optimization , Project Management
1 INTRODUCTION
An FDT accurately replicates the behavior of a manufacturing facility from raw materials to final products . The history of FDTs began in the early 1900s with process charting ( e . g . Gantt Charts ), but computers in the 1950s enabled FDT optimization and the power of FDTs has grown with the power of computing . Contemporary FDTs became recognizable in the 1970s – for historical perspective see Luyben 1973 [ 1 ].
An FDT can generate significant value by rapidly investigating scenarios that would be difficult , costly , or impossible to undertake using the real-world manufacturing facility . An FDT offers the possibility of looking over the multitude of ways a manufacturing facility can be operated to find the best ways that increase capacity , reduce cost , and overcome unexpected events .
An FDT complements digital twins ( DT ) at the unit operation / manufacturing activity level . A DT intimately depends on the process physics and often involves addressing spatial geometry , partial / ordinary differential equations resulting from the physics , and the underlying
98 February 2025