Operating Digital Twins Within an Enterprise Process

Abstract

The Digital Twin concept is an all-encompassing Industrial Internet of Things (IIoT) use case. It is an artificially intelligent virtual replica of a real-life cyber-physical system (CPS) useful in all phases of a system’s lifecycle. The Digital Twin is made possible by advances in physics-based modeling and simulation, machine learning (especially, deep learning), virtual/augmented reality, robotics, ubiquitous connectivity, embedded smart sensors, cloud and edge computing, and the ability to crowd source the domain expertise. These technologies have the potential to make Digital Twins anticipate and respond to unforeseen situations, thereby making CPS resilient. To realize resilient CPS, engineers from multiple disciplines, organizations and geographic locations must collaboratively and cohesively work together to conceptualize, design, develop, integrate, manufacture, and operate such systems. The refrain “model once, adapt with data and domain expertise, and use it many times for many different purposes” offers an efficient and versatile approach to render organizational silos extinct. This is accomplished by providing a “single source of the truth” representation of the CPS, to collaborate virtually, assess and forecast in evolving situations, and make adaptive decisions. Organizationally, Digital Twins facilitate situational awareness and effective organizational decision-making through the acquisition, fusion, and transfer of the right models/knowledge/data from the right sources in the right context to the right stakeholder at the right time for the right purpose. That is, the design, manufacturing, optimal operation, monitoring, and proactive maintenance of the CPS. This Chapter addresses the operation of Digital Twins within an enterprise process. The discussion is generally applicable, but illustrated specifically with examples from the Aerospace Industry. It begins with a vision for the enterprise Digital Twin methodology to provide timely and accurate information created during the initial conceptual design, product development, and subsequent operational life cycle phases of the product/system utilizing a comprehensive networking of all related information. All related partners share such information, thereby connecting product/system design, production and usage data with those human and non-human agents requiring this information. The Chapter further reviews the enterprise-wide product lifecycle phases and describes how the use of digital twin methodology allows quasi-static model-based systems engineering (MBSE) and Enterprise Resource Planning (ERP) business models to morph into a temporal information continuum, spanning the life cycle of the product or system. Specifically, the focus is on global information flow throughout the enterprise, and suggested DT-committed organizational changes affecting the enterprise. It is followed by a discussion of MBSE-based requirements analysis and platform-based design principles in the product’s conceptual design phase and related examples. This lays the groundwork for encouraging a range of conceptual design ideas, standardizing design, analytical and learning tools for superior coordination and integration of the information flow within the enterprise’s Digital Twin processes. Subsequent portions of the chapter discuss the product development phase via Digital Twin models and platform-based design principles using digital 3-D CPS models with specific examples from Sikorsky and GE as illustrations. An emphasis is placed on the improvement of computing capabilities, such as the introduction of hyper-efficient Graphical Processing Unit (GPU)-based computational capability that provided an order of magnitude improvement in design productivity. Multi-functional causal models are introduced to help uncover failure modes, their propagation paths, and consequent functional effects, and discuss how such digital twin models automatically generate fault trees for risk assessment analysis, and an initial Failure Modes, Effects, and Criticality Analysis (FMECA) report. The importance of domain knowledge and data-informed models is emphasized in how it can aid in product testing, qualification, and certification phase. A formal system of health modeling to test the severity of candidate faults and their effects to generate an updated FMECA model using the digital twin is also introduced. This enables design engineers to understand the potential faults in the system, their probabilities of occurrence, and their manifestation as functional failures (effects), monitoring mechanisms for making the effects visible, system level implications in terms of safety, customer inconvenience and service/maintenance implications, and so on. The Digital Twin methodology enables the FMECA and fault tree updates in real-time. The role of Digital Twins in product manufacturing, quality management and distribution phase is addressed next. It includes a description of an integrated process for additive manufacturing and an advanced 3D quality inspection process. The DT network links these highly accurate coordinate measurement processes with the complex 3D Cyber-physical models intended to define the product accurately. As part of the Digital Twin, an integrated on-board and off-board system health management, coupled with virtual/augmented reality, can improve customer experience and support via real-time monitoring, incipient failure detection, root cause analysis, prognostics, predictive maintenance, and training assessment. The S-92 helicopter’s data integration process serves as a constructive example of how a proactive digital twin-aided health management system can significantly improve product resilience, safety, and customer acceptance. The remaining portions of the chapter describe how the Digital Twin infrastructure’s ability to process enormous amounts of data into information and knowledge, aided by the Failure Reporting, Analysis and Corrective Action System (FRACAS) database, enables proactive product configuration management and active learning. The result is efficient product maturation and customer adaptation. The Digital Twin’s ability to support the design of environmentally sustainable products and how they can eventually be suitably disposed is also addressed. The successful adoption of Digital Twins has other consequences, including the need to revamp traditional organizational structures to be effective in a globalized environment. While Digital Twins offer great promise, it is also important to consider some cautionary thoughts on the need for accurate models, domain knowledge-informed machine learning, and awareness of human fallacies in implementing the Digital Twin methodology. Finally, in a business environment, commitment to the Digital Twin methodology hinges on an understanding of the value that Digital Twins provide and the steps that the enterprise must take to successfully adopt the methodology. The enterprise must accept significant structural and cultural changes to succeed with a DT methodology. The authors firmly counsel that adaptation of these necessary enterprise modifications will not be easy and, as such, will require top-down leadership to respond to structural and cultural changes with the necessary corporate resources (adequate and digitally literate staff, leadership, hardware-software computing and communication infrastructure, and budget).