Digital Twin Quality Assurance - PhD

Digital twin (DT) is a virtual representation of a physical asset, system, or process, that is connected to the real world through sensors and two-way synchronization. A DT can be used to monitor, simulate, and optimize the performance of the physical asset, and can provide insights into its behaviour and potential issues. DTs often rely on data from sensors, IoT devices, and other systems that may use different formats, protocols, and interfaces. In addition, DTs may need to interact with other systems, such as enterprise resource planning (ERP) systems, supply chain systems, or customer relationship management (CRM) systems, which may also use different standards and data formats.

There is growing interest in literature for this topic; Franciosa et al. [1] demonstrated the results of using the first fully digitally developed remote laser welding process for aluminium doors, which yielded a right-first-time rate of >96% for door assembly cell development. Soderberg et al. [2] emphasise that accuracy challenges in DTs, in the context of real-time geometry assurance, are driven by faster optimization algorithms, higher computer power, and increased amount of available data. Erdos et al. [3] propose that by improving the twin closeness through multilevel calibration methods, an accurate DT of the as-built cell can be established in which the sufficient accuracy of offline planned robotic operations is guaranteed. Preuveneers et al. [4] investigated the extent to which faults in DTs can compromise intelligent cyber-physical production processes, and suggested that DTs should be equipped with safety features such as feature toggles and software circuit breakers capable of intercepting local faults and preventing them from propagating and cascading to other systems. Stark et al. [5] highlight that real-time simulation takes place as the DT controls the manufacturing process with the information about the status of the process and the positions of every object. While there are gaps in terms of the mechanisms to increase the accuracy of DTs, Helu et al. [6] emphasise the need to complement simulation and support optimal decision making throughout the lifecycle, as means to align and contextualize as-planned and as-executed process data dynamically using semantically rich, open standards.

This PhD proposal focuses on measuring and dynamically delivering (where needed) the assurance of DTs. The PhD will apply the principles of resilience engineering for DTs to bridge the gap between the as-planned and the as-executed process (e.g. engineering, manufacturing). The focus is on how to increase the consistent and accurate representativeness of the DT (considered as the key measure for assurance) over time given the dynamic behaviour of the component, product, or system. The novelty of the work lies in the proposed innovative approach that integrates resilience and machine learning to self-learn and adapt the assurance of the DT.

Cranfield University is wholly postgraduate, and is famous for its applied research in close collaboration with Industry. At Cranfield, the candidate will be based within the Manufacturing theme at the Centre for Digital Engineering and Manufacturing (CDEM). The Centre hosts cutting-edge simulation and visualisation facilities. The student will have access to high-end computers and digital technologies in the Centre for ontology-based and knowledge-based systems development, Digital twin development, advanced dynamic modelling and simulations, AI, VR, AR developments. The candidate will work on his/her research individually and collaborate with other researchers in the Centre.

The existing literature presented emphasises the issues of DT accuracy, calibration, safety and reliability, simulation capability, and the impact of DTs on the resilience of the underlying system. However, the concept of resilience has not yet been applied to DTs themselves and represents the novelty of this PhD to enable DT assurance. The PhD will take the approach that resilience should be considered not only in physical manufacturing/engineering processes, but also in the digital assets including DTs, as it can enable the relevance of the digital support assets to be sustained with assurance in decision making over time. In this context, resilience refers to the ability to respond to disruptions experienced in the DT that may arise due to various reasons such as missing data, a temporary disruption in the data feed, or models that are not up to date given the evolving asset or its processes.

This PhD has the following objectives:

1. Capture current approaches to DT development, and assurance assessment. 
2. Right-time identification of the DT assurance status. 
3. Develop a dynamic framework to quantify the optimised assurance level for the DT with respect to the data feed and its models over time. 
4. Mechanisms to self-adapt the DT to meet optimised assurance.

Objective 1 is focusing on mapping the approaches to DT within BAE Systems and beyond in terms of DT design architectures, established assurance methods, current data types, structures, and interoperability challenges. This will lead to a detailed representation of the as-is model approaches to assurance of DTs. This will be developed through interviews, and cross case analysis, to capture trends, and a detailed risk analysis will be conducted to prioritise methodologies for assurance in DTs.

Objective 2 is focusing on measuring and assessing the DT assurance status. This will be achieved by developing a DT intelligence module that allows it to measure current state, and predict future behaviour, effectively. The learning capability will be used to detect anomalies, i.e., detecting when the behaviours of the real system and the DT diverge and result in assurance issues. The learning agent will be designed to monitor a set of properties of the real system - observables - and compare their values with those predicted by the DT.

Objective 3 is focusing on determining the optimised response to a deficiency in DT assurance. Here it will be important that the response is accurate and that the recovery is rapid in order to minimise the delta between the DT and the real system. An optimisation algorithm will be developed to determine the best possible DT assurance level, and the approaches to achieve it will be determined.

Objective 4 is focusing on how to address inadequate DT assurance level. It will build resilience-based approaches. Here resilience is defined through the resilience loss, which is the integral of the lost accuracy due to the disruption in assurance. We will investigate how the introduction of additional sensing could affect the DT accuracy and contribute to the correct identification of causes of inadequate DT assurance. A framework will be developed on how to parameterize the structure of DT as well, and to incorporate domain knowledge to assess its feasibility. As part of this we will investigate alternative ways to enable feedback from digital representation back to the real system. The proposed approach could be extended by introducing the ability to suggest changes that could improve the performance and resilience of the real system in the face of uncertainty and disruptions.

A PhD in Digital Twin Assurance offers unparalleled opportunities for global exposure, skill development, and professional networking with the close ties between Cranfield and BAE Systems. Candidates can present their research at leading international conferences, such as the IEEE Digital Twin Symposium or ASME Digital Twin Summit, where they engage with industry leaders and academics driving innovation in this field. Many programmes facilitate international collaborations, including research visits internationally, as well as industry visits to BAE Systems, and potentially global tech leaders such as Siemens and Dassault Systèms. Furthermore, students can enhance their expertise through external training in cutting-edge tools like ANSYS Twin Builder and Siemens’ MindSphere or obtain certifications in ISO quality standards for digital twin environments. These experiences not only enrich academic knowledge but also build a strong professional profile tailored for the future of digital transformation.

The student will gain from the experience in numerous ways, whether it be transferable skills in the technical area of Digital twin, ontologies and machine learning, or soft skills including presentation skills, project management, and communication skills. There are also numerous employability opportunities that the PhD will offer whether it be in Industry or in Academia.