Title Promoter Affiliations Abstract "The Twin Growth Project - Pathophysiology, diagnosis and outcomes of isolated selective fetal growth restriction in monochorionic twin pregnancies" "Liesbeth Lewi" "Woman and Child" "Monochorionic twins are monozygotic and during their intrauterinelife, they share a single placenta with anastomoses connecting theircirculations. Although identical, in about one in five pairs, one twin issignificantly smaller than the other. The best management ofselective intrauterine growth restriction (sIUGR) is unknown and stillposes daily clinical dilemmas, especially when the blood flow towardsthe smaller twin is disturbed and birth is not an option yet. In thesepregnancies, the growth-restricted fetus may die, which puts thenormally growing co-twin at risk of demise or ischemic brain damagedue to exsanguination across the anastomoses. Also, we do not fullyunderstand its pathophysiology, and current diagnostic criteria areinsufficient. Our research proposal consists of the multi-centerretrospective ""ALIGN"" study, which analyzes management-relatedoutcomes of severe sIUGR in the pre-viable period to inform apatient decision aid and an aligned management protocol. Second,the multi-center prospective 'TWIN GROWTH"" study aims to identifyadditional prenatal diagnostic markers to distinguish between sIUGRpregnancies with expected good outcomes and those at high risk offetal demise or requiring an extreme preterm birth. Finally, in the""TWIN CAKE"" study, we focus on the histologic differences betweenthe different types of sIUGR to increase our knowledge of theunderlying pathophysiology." "Elucidation of the role of aluminium in hydrogen induced degradation of TWIP steels" "Kim Verbeken" "Department of Materials, Textiles and Chemical Engineering" "Advanced high strength steels (AHSS) are very promising for e.g. structural components in car bodies due to their high strength combined with large elongations. These steel types are however very prone to hydrogen embrittlement (HE). HE is characterized by a decrease in the metalU+2019s mechanical properties. It might also lead to unpredictable failures due to hydrogen induced cracking (HIC). Hydrogen originates either from the production process or in use by contact with a hydrogen containing environment. The present project will focus on HE in twinning induced plasticity (TWIP) steel which is an austenitic AHSS. Austenite is the high temperature variant of iron, stabilized at room temperature by addition of over 17 wt% of manganese. The HE/HIC phenomenon and the responsible mechanism in TWIP steels is not well understood yet. Accurate hydrogen charging and test methods will be developed. A literature based TWIP steel will be used. Additionally, aluminium addition will be investigated as it is reported to have a beneficial influence on the HE sensitivity of TWIP steels. This effect might be a bulk effect, a surface effect or a combination of both. Alloy design, together with high resolution material and surface characterization techniques will be used to investigate the role of the aluminium in these materials." "From Virtual Sensing to Executable Digital Twin: Towards Multi-objective, Real-time Estimation Algorithms for Vehicle Dynamics through a Digital Twin Approach" "Frank Naets, Wim Desmet" "Mecha(tro)nic System Dynamics (LMSD)" "Driven by the digital revolution towards an industry 4.0 context, mechatronic industries such as the automotive sector have undergone major transformations in order to strengthen their flexibility and competitiveness on the market by enhancing vehicle performance and comfort. In addition, safety regulations are becoming increasingly stringent as car accidents happen evermore often due to the rise in population density, especially within large cities. All these aspects pose additional challenges towards the development of new vehicles, corresponding subsystems and control algorithms in order to increase performance, comfort, and safety, which usually presents a trade-off. This has led, amongst others, to the development of Advanced Driving Assistance Systems (ADAS) to enhance passenger vehicle safety, comfort, and performance.To further increase market flexibility and reduce vehicle development costs, Original Equipment Manufacturers (OEMs) are heavily investigating the capabilities of digital twins or virtual representations of physical assets. The idea has already been launched at the end of the 20th century, although thus far practical realizations are still lacking due to the challenging effort needed to synchronize available models and data with the physical asset. Therefore, one of the main goals of this research is to develop a practical realization of a digital twin framework in order to show the potential of virtual assets to reduce development costs and increase market flexibility. In order to achieve this, current state-of-the-art approaches related to vehicle dynamics modeling and virtual sensing must be improved, and novel parameter synchronization strategies have to be developed to ensure that virtual representations are sufficiently accurate. It should be noted that, although the framework was specifically developed for the automotive industry, the general principles are applicable to other sectors as well since this work adopts a physics-based approach.The digital twin as developed in this work is based on three main building blocks, namely (i) a family of vehicle dynamics models, (ii) a range of virtual sensing methodologies allowing communication between virtual models and the physical asset, and (iii) a parameter synchronization tool which provides up-to-date parameters between different models through a general synchronization strategy. Since different applications will require different model complexity, a modeling framework is developed which uses the general theory of classical mechanics to derive equations of motion and consequently provides various solving strategies depending on the complexity of the derived equations. Using this framework, a family of vehicle models is created with the application towards low-velocity driving scenarios and state estimation algorithms in mind. Furthermore, a virtual sensing framework is proposed, which includes an extensive observability analysis in combination with dynamic coupling analysis and a novel sensor selection methodology to allow communication between virtual models and the physical asset. Through the use of a Singular Value Decomposition (SVD), this work shows it is possible to identify unobservable states and to stabilize the estimator Ricatti equation using an observable transformation on the covariance equations, if and only if the targeted quantities of interest are independent from decoupled internal model states. This allows further increasing model complexity without having to increase the amount of required sensors for full observability. Last but not least, a novel sensor selection algorithm was developed which is capable of taking previous observability considerations into account. The selection methodology ranks the possible sensors according to their relative contribution to the targeted quantity of interest estimation performance, characterized by corresponding covariances as calculated by the Extended Kalman Filter (EKF). This approach is capable of handling different sensor types, alternative sensor locations, and sensor accuracy as indicated by their corresponding noise level, which is heavily linked to the associated cost.The combined result of this work is the practical realization of such a digital twin framework for vehicle dynamics applications aiming for the development of real-time estimation algorithms, validated on an internally built demonstration platform, dubbed as the LMSD Concept Car. The estimation methodologies were experimentally validated on several test vehicles including an electric Range Rover Evoque, a Ferrari 250LM, a Goodyear SightLine test vehicle, and the LMSD Concept Car platform. The latter vehicle platform was used to prove that real-time applicability of estimation schemes featuring reduced-complexity models is possible and provides accurate estimation results. Additionally, the work shows that the virtual asset is a powerful tool, enabling various non-trivial analyses using parts of the digital twin to evaluate different vehicle dynamics aspects such as design changes, new controller designs, different powertrain layouts, flexible coupling between subsystems and/or vehicle components and many more." "Urban Digital Twins as an instrument for demand driven, cross-domain cooperation and evidence-based policymaking." "Joep Crompvoets" "Public Governance Institute" "Digital Twins originated as a term and discipline in the manufacturing industry as a simulation tool for testing industrial designs two decades ago. A Digital City Twin or a Digital Urban Twin emerged more recently as an instrument to manage complex processes in a city. The concept of a Digital Twin of the city is in full swing in 2021. Notable examples in Asia, North America and Europe such as Singapore, Toronto, Helsinki, Rotterdam and Vienna mainly focus on an infrastructure and supply-oriented approach by realising BIM (Building Information Models), whether or not in combination with IoT sensor data. This research wants to start from a demand-based 'use case-oriented approach', in which the questions and possibilities of a Digital Urban Twin are examined to be able to simulate cross-policy policy questions in the city as input for policy decisions and communication / co-creation to the citizens." "Digital twin of a laser line scanner: uncertainty evaluation and scan path planning" "Wim Dewulf" "Manufacturing Processes and Systems (MaPS)" "According to the Guide of the Expression of Uncertainty in Measurement every measurement needs to be accompanied with an uncertainty. A digital twin of a measurement system can be used to determine the measurement uncertainty through Monte Carlo simulations. In the past, the state of the art has developed virtual coordinate measuring machines to determine the measurement uncertainty. However, this research was never fully extended towards a digital twin of an optical measurement system. One of the main reasons is the high number of different error contributors inherent to non‑contact probing. The goal of a digital twin is to mimic its physical counterpart. Therefore, the first objective of this thesis is the development of a digital twin of a laser line scanner mounted on a coordinate measuring machine. The considered error contributors are relevant for the measurement conditions in which this research is performed, namely a metrology laboratory.In order to accomplish the first objective, an equation to model the systematic error and the random error of the laser line scanner is formulated. This equation is a second degree polynomial function with variables dependent on the relative position and orientation of the probing system towards the object. The coefficients of the model are experimentally determined. The developed method is reproducible with the standard equipment available in a metrology laboratory. The proposed model and the incorporation of the constraints of a laser line scanner allow to simulate the probing device. Furthermore, an updated method of virtual coordinate measuring machine is proposed. This is essential for the digital twin, since the laser line scanner is positioned and oriented by a coordinate measuring machine. The introduced virtual coordinate measuring machine distinguishes itself from the state of the art by its clear description of the representation of the incorporated errors. The kinematic errors of the model are elaborated with a new‑defined equation. The combination of the virtual coordinate measuring machine and the model of the laser line scanner form the digital twin. The obtained results are validated through a series of experimental measurements in the metrology laboratory.The second objective is to determine the task‑specific measurement uncertainty, when measuring with a laser line scanner. The in‑house developed digital twin allows to perform measurements virtually, thus the measurands are virtually determined. As a consequence a variety of geometrical properties, with a known nominal value, can be scrutinized. Different measurement strategies are applied to measure a feature, in order to investigate the influence on the uncertainty budget. As validation the virtually obtained conclusions are compared with the experimentally obtained data.The third objective is to determine the scan path of the laser line scanner autonomously. The manufacturing industry tends to manufacture an increased amount of lot‑size‑one products with small tolerances. As a result of this tendency, the quality inspections become a bottleneck in the manufacturing process. Optical probes are able to reduce the measurement time compared to conventional tactile measurements. Digital twins allow to estimate the uncertainty offline without the need of repeated online measurements. However, the manual determination of the scan path is time‑consuming and is detrimental for the traceability of the measurement. The scan path generation for complex geometries is challenging, hence, the third objective was formulated. The proposed algorithm to determine the scan path autonomously takes the uncertainty budget of toleranced features and all the constraints of the measurement system into account. Making this task autonomously reduces the inspection time and allows the transition of the measurement system towards the Industry 4.0. The validation of the proposed algorithm is performed on two objects. The algorithm was able to autonomously generate a scan path which is able to cover the entire measurable surface and to determine the measurement uncertainty." "Time-varying vibro-acoustic digital twins towards a sustainable society." "Elke Deckers" "Mecha(tro)nic System Dynamics (LMSD)" "The current European energy crisis has re-established the importance of accelerating our transition towards a more sustainable society and industry. The sustainable design and monitoring of vehicles and machines is an important contributor to this transition. To improve sustainability, a delicate balance between the conflicting criteria of health, profitability and environment needs to be found. One of the most important health considerations towards this goal is noise pollution.  The recently emerged digital twin paradigm has the potential to help us find this balance. A digital twin is a digital copy of a physical asset that behaves the same as the physical asset. Thus, by using sensor data from the physical asset, it can make design/control/maintenance decisions. In the context of vibro-acoustics initial digital twin applications show great potential using the underlying assumption that the physics are time-invariant. However, many vibro-acoustic systems instead show time-varying behavior.Therefore, the goal of this project is to assess how the inclusion of time-varying behavior in the digital twin can improve performance towards sustainability, by looking at time-varying behavior at different time scales. In the project novel physics-based models, model order reduction techniques and digital twin formulations will be developed, looking at several use cases to demonstrate the potential of time-varying digital twins to tackle the sustainability challenges of the future." "Digital Twins for high-dynamic mechatronic systems -dynamic force, torque and stress estimation" "Wim Desmet" "Numerical Analysis and Applied Mathematics (NUMA), Mecha(tro)nic System Dynamics (LMSD)" "The dynamic forces, torques and stresses in function-critical parts and components play a crucial role in the economic profitability, in the environmental sustainability, and in the human health impact of machines and vehicles. Knowing these quantities - during the design as well as the operational lifetime of an asset - has therefore an important socio-economic impact and relevance. Unfortunately, these quantities are hard or expensive to measure directly in high-dynamic mechatronic systems.The goal of this project is the development and deployment of Digital Twin enabled force, torque, and stress estimation approaches in mechatronic systems.A Digital Twin is an evolving digital replica of an individual physical asset, including both its historical and current behaviour. The incoming data from sensors and actuators aboard of mechatronic systems enable the continuous improvement of the Digital Twin. The Digital Twin, in turn, delivers valuable insights to improve the system design, and provides (often hard to measure) information on the system operation.In this project, the Digital Twin combines heterogeneous models, available from engineering analysis, within a novel multi-model estimation scheme. Major challenges include the efficient mechatronic model evaluation through (parametric and non-linear) model order reduction, and their use in virtual sensing through (time stable) state-input-parameter estimation to (continuously) retrieve the aforementioned key dynamic quantities. Methodological validations and application demonstrations will be performed on dedicated gearbox, vehicle dynamics and high-speed weaving loom test rigs." "Integration of sustainability assessments in a biorefinery digital twin" "Karel Van Acker" "Center for Sustainable Catalysis and Engineering, Sustainable Metals Processing and Recycling" "The PhD position is part of the interdisciplinary  network project “Future chemical FActories: Sustainability through Smart Digitalisation (FFASSD)” in which 3 PhDs in 3 different research groups will collaborate intensively on the development of a digital twin concept for biorefinery technologies, including mechanical, chemical performance and sustainability aspects. A digital twin is a virtual replica of an asset fed with data from the physical set-up, which allows for cost- and time-effective analysis, optimization, up-scaling and virtual scenario assessment. The aim of the FFASSD project is to apply the digital twin concept on biorefinery technology with the objective to optimize the process in an integrated way for performance, energy and resource efficiency, environmental impact and cost. The project focusses on a unique biorefinery technology developed at KU Leuven that converts wood to plastic building blocks and commodity chemicals. To turn such a successful lab-scale process into an industrial one, a series of capital intensive and laborious steps is classically required, which impedes rapid industrialization and, hence innovation. The digital twin concept, as currently developed within mechanical engineering at KU Leuven, is an interesting key enabling technology to tackle this challenge. This PhD will extend the digital twin with environmental impact and economic analyses, based on life cycle sustainability assessments (LCA, LCC). Furthermore, these models will be further elaborated in order to correctly evaluate the effect of upscaled reactor sizes, from lab-scale to industrial scale, production efficiency will be analyzed, and the link with the economic and environmental context of biorefineries will be studied." "Digital Twin development for moulding and casting processes to evaluate final part quality" "Frank Naets" "Mecha(tro)nic System Dynamics (LMSD)" "This PhD project focuses on the development of a Digital Twin for moulding and casting processes to assess the quality of final parts. The Digital Twin is constructed by integrating data from both lab scale and industrial scale systems, enabling a comprehensive representation of the manufacturing process. The Digital Twin is driven through physical modeling of the component and process, utilizing finite element models based on the component's geometry as a backbone. A multiphysical approach is adopted, considering various physical phenomena and their interactions to provide a holistic understanding of the entire process. This research aims to enhance process optimization, reduce defects, and improve the overall quality of the final parts by leveraging the capabilities of the Digital Twin for comprehensive evaluation and analysis." "Development of a digital twin for pig growth and behaviour." "Tomas Norton" "Animal and Human Health Engineering (A2H)" "A livestock farmer needs to control and manage many different facets to run his business, and a lot of that complexity originates from the fact that it is dealing with living animals that are individuals in need of appropriate care. An animal is a complex organism, which in the context of livestock farming receives certain inputs (such as feed, water, housing) in order to produce outputs (meat, eggs, milk) as efficiently as possible, but at the same time the goal is to have a maximum welfare and health and minimal effect on the environment. A large range of both internal and external factors along with individual variation influence this interplay of production, health, welfare and environment. A lot of these factors are man-made or can be influenced by the farmer while the animal itself has little to no room for making adjustments. Therefore, the farmer needs to optimize the conditions in order to prevent decreased production, abnormal behaviour, disease, impaired welfare, etc. Early identification of suboptimal conditions or upcoming problems is crucial. However, the current welfare monitoring approaches focus on the detection of specific and relatively late, severe and apparent indicators or problems. Moreover most approaches use the detection of negative symptoms and conditions whilst we should also work on preventing them and increasing the positive occurrences. Therefore multi-purpose digital twin models should be developed for the prediction and early identification of positive and negatives states of the animal that are reflected by behavioural changes. The first step in the research would be to combine existing automated measurements of behavioural, physiological and environmental indicators that are useful to assess the animals internal state and are needed to build/train the digital twin model. The second step in the research is to combine existing models and novel AI algorithms to simulate the animals internal state (stress, hunger, joy, etc.) and to model positive and negative occurrences and interactions and the animals’ daily behavioural patterns. The final step in the research is thus to apply the digital twin model as a prediction model for the animals’ health, welfare and production. Because of its predictive power and the hybrid approach, the digital twin is able to replicate the animals internal state and thus its use can be tested and validated on several use cases, for example to predict the risk of occurrence of certain negative/positive behaviours and their consequences such as tail biting or free access to the feeder when the animal is hungry."