On this page

Program

Pre-FEARS workshops

All info about the pre-FEARS workshops on the workshop page.

FEARS 2023 program

13:00

Registration

All attendees are welcome from 1pm to enjoy a coffee. Presenters can set install their demo setups and posters.

13:30

(Parallel sessions)

Poster and demo session (Part 1)

Stroll through interesting posters and demonstrations of FEA research and take the time to get to know colleagues.
Detailed poster and demo schedule

Roundtable: topics 1-5

14:00-14:45

Dive deep into hot topics in engineering and architecture with leading companies at our Industry Roundtables.
More info on the roundtable page

Pitch session

Get inspired by a selection of two-minute pitches by FEA researchers.
Detailed pitch schedule

15:00

Coffee break

15:30

(Parallel sessions)

Poster and demo session (Part 2)

Stroll through interesting posters and demonstrations of FEA research and take the time to get to know colleagues.
Detailed poster and demo schedule

Roundtable: topics 6-10

15:45-16:30

Dive deep into hot topics in engineering and architecture with leading companies at our Industry Roundtables.
More info on the roundtable page

Infosession 1: Starting a PhD

Information session for students. We help you figure out whether a PhD is something for you and how you can start one.

Infosession 2: Valorise your research

An introduction to the different TechTransfer practices for postdocs and finalising PhD researchers.

17:15

Panel AI in Research

A panel discussion on "AI in Research," envisioning the future, exploring the challenges, and receiving invaluable advice on the power of AI in research! Confirmed members: prof. Eric Mannens (UGent), prof. Lieve De Wachter (KULeuven), Matthias Feys, PhD (CTO ML6) and Laurent Sorber (CTO Radix.ai).

18:00 - 20:00

Reception with award ceremony

We will close FEARS 2023 with a drink and some bites, while presenting awards for remarkable contributions to the symposium.

Pitch sessions

Pitch sessions are organized in the Calefactory. Please find the order of pitches below.

InSiDe: A Cardiovascular Screening Device Based on Silicon Photonics

Authors

Simeon Beeckman, Yanlu Li, Roel Baets, Patrick Segers

Abstract

Cardiovascular disease is the largest cause of death worldwide with a contribution of around 30% of global mortality. Screening for these diseases may be done by assessing the increase in arterial stiffness. This can be done by calculating carotid-femoral pulse-wave velocity (cf-PWV), a biomarker that can reliably be measured via, amongst others, laser-doppler vibrometry (LDV). A prototype utilizing this technology was developed within the scope of the CARDIS (& subsequently the INSIDE) project. By simultaneously measuring skin vibrations above main arteries such as the carotid and femoral arteries, cf-PWV can be estimated as PWV=∆x/PTT. The current prototype contains two handpieces equipped with an array of six laser beams each. A retroreflective patch is applied to the measurement site on which the laser beams are reflected, picking up the skin displacement information. From these time series, acceleration is derived. The peaks in the acceleration traces that correspond to the arrival of the heart-contraction induced pressure-wave are identified. The time delays between detected peaks that stem from the same heartbeat are called pulse-transit times (PTT). The distance between measurement sites is called ∆x. Ongoing work is being done on a new prototype lacking the need for the patches and on algorithms for both cf-PWV and more potential biomarkers present in the LDV data. New data is currently being measured by project partners at INSERM, within the scope of a clinical feasibility study in Paris. The potential use of chest vibration measurements for other biomarkers, relating to heart rhythm disfunction and carotid stenosis, is being investigated.

Pre-Operative Partial Nephrectomy Planning: From Four-Phase CT-Scan to 3D Renal Perfusion Model

Authors

Saar Vermijs, Pieter De Backer, Pieter De Visschere, Alexandre Mottrie, Charles Van Praet, Karel Decaestecker, Charlotte Debbaut

Abstract

Partial nephrectomy is the optimal surgery for early-stage renal tumours. Selective clamping (only clamping those arteries that perfuse maximal tumour and minimal healthy tissue) lowers the risk of excessive bleeding during tumour resection, while safeguarding the remaining healthy tissue’s renal function. To clamp successfully, insight in the patient-specific perfusion zones is necessary, which cannot be provided by CT-scans. Therefore, we propose a 3D renal perfusion model, which we validated in a retrospective study (De Backer, Vermijs et al. 2023). Twenty-five patients were included. Their four-phase CT-scan was segmented in Mimics (Materialise, Belgium) to create a 3D model. The arteries were simplified to a centerline using VMTK (vmtk.org) to label the branches. The labelled centerline points served as seeds in a Python-based region growing algorithm, executed in a voxelized bounding box around parenchyma and tumour. This resulted in a 3D renal perfusion model for each patient. To validate this, the clamped selective arteries were detected on the surgical video. Their perfusion zones, which were made visible during surgery using indo-cyanine green under near-infrared lighting, were compared to the calculated perfusion model using two metrics. The first metric, the total overlap of the perfusion zone contours, was scored by six independent urologists. This resulted in an average score of 4.28 out of 5 (median: 5; range: 2-5: interquartile range: 4-5). The second metric, focusing on the perfusion of the tumour, was evaluated using a scoring grid and resulted in an average score of 4.14 out of 5 (median: 5; range: 1-5; interquartile range: 3.5-5). The high scores for both metrics show that the proposed algorithm is robust. A limitation of this study is that indo-cyanine green only reveals perfusion at surface level. However, the results are promising and show that the perfusion model can assist in preoperative planning.

Toward Metal-Organic Framework Design by Quantum Computing

Authors

Kourosh Sayar Dogahe, Tamara Sarac, Delphine De Smedt, Koen Bertels

Abstract

We present a hybrid quantum-classical method for calculating Potential Energy Surface scans using quantum computers, which are essential for designing Metal-Organic Frameworks for Direct Air Capture applications. The results demonstrate, at a small scale, the potential advantage of quantum computing-based models. We aimed to define relevant classical computing model references for method benchmarking. The most important benefits of using the (PISQ) Perfect Intermediate Scale Quantum approach for hybrid quantum-classical computational model development and assessment are demonstrated. To investigate that, a quantum computing-based energy calculation model was constructed following the PISQ approach and a comparison of perfect and physical qubit-based implementations by Variational Quantum Eigensolver (VQE) algorithm was made in parallel with classical methods including RHF, CCSD, and CASCI computations. The potential of quantum computing has been demonstrated through well agreement of VQE-perfect qubits results with CASCI as the reference. However, an attempt to scale up the model to involve more molecular orbitals and hardware implementation revealed drawbacks of current quantum computing: low number of qubits available, erroneous behavior of qubits, and algorithm design challenges. Perfect qubit-based simulation proved to be a valuable approach to perform model evaluation. In addition, VQE-perfect qubits results were used to parameterize VQE-physical qubits implementation, which shortens hardware runtimes. Perfect qubit simulation was used to draft a pathway toward modeling 3D and full unit cell PES scans.

The Study of Conformational Dynamics of Choline Import Mediated by a Bacterial Membrane Transport Protein Using Modeling and Non-Equilibrium Simulations

Authors

Alen T Mathew, AhmedReza Mehdipour

Abstract

Background: LicB is a bacterial membrane choline transporter that plays a crucial role in virulence features & antibiotic resistance in S. pneumonia. Invitro studies involving inactivation of the licb gene, lead to nonviable S. pneumonia suggesting the therapeutic potential of targeting LicB, but currently the exact mechanism by which LicB imports choline into the bacterial cytoplasmic space is unknown. Objectives: Our objective is to explore the conformational changes of LicB during the choline import by reconstructing the free energy landscape of LicB to explain the thermodynamics & kinetics of the transport cycle. Methods and Results: To achieve our objective, we are using a set of computational techniques like biomolecular simulations coupled with enhanced sampling techniques. Using homology modelling approaches and non-equilibrium simulation techniques like steered molecular dynamics (SMD) & targeted MD (TMD), we will generate intermediate states, such as the inward-open, which are yet to be identified experimentally. Later, techniques like Markov state modelling & transition path sampling are used to explore the kinetics of the protein conformational changes during the transport cycle. Through our microseconds long unbiased simulation of the protein, we verified the stability of the experimentally identified intermediate structures of the protein. Later we were able to generate other intermediate structures like inward open, choline-bound outward open, which are yet to be identified experimentally, using a combination of Alphafold based predictions, homology modelling of evolutionary similar proteins and non-equilibrium simulations like SMD & TMD. Conclusion: We have identified a suitable way of generating intermediate structures of proteins involved in membrane transport cycle using a combination of modelling and non-equilibrium simulations. Next step involves identifying the free energy and kinetics associated with the transport cycle which will help in designing drugs targeting LicB against S. pneumonia.

Exploring Caveolae's Role in Oxygen "Buffering" Through Coarse-Grained Molecular Dynamics

Authors

Samaneh Davoudi, An Ghysels

Abstract

The "oxygen paradox" refers to the intricate interplay between two contrasting biological processes involving oxygen (O2) as a reactant. O2 is vital for aerobic metabolism, acting as a fuel for oxidative phosphorylation within mitochondria. However, an excessive supply of O2 can lead to the generation of reactive species that harm cellular health. Therefore, maintaining O2 homeostasis becomes crucial, requiring a delicate balance that prioritizes the former process while minimizing the latter. In earlier research, a hypothesis was proposed centered around specialized membrane invaginations called caveolae. These unique structures exhibit a curved morphology and are rich in cholesterol. It was postulated that caveolae play a pivotal role in regulating O2 levels within cells by efficiently absorbing O2 and attenuating its release to the mitochondria. However, the exact mechanism through which caveolae contribute to O2 buffering remains unclear, presenting an intriguing research question. To address this knowledge gap, our primary objective is to investigate how specific structural characteristics of caveolae, such as membrane curvature and cholesterol content, influence the local O2 abundance and membrane permeability. To accurately simulate a curved membrane resembling caveolae, we employed coarse grained molecular dynamics (MD) simulations. After conducting a comprehensive set of tests, we carefully selected one of the CG beads to serve as the CG O2 model. Liposomes of varying sizes were created to represent different levels of curvature, composed of phosphatidylcholine (POPC) and cholesterol. This approach allows us to observe associated changes in the O2 free energy profile and membrane permeability. By unraveling the influence of membrane curvature and cholesterol content on local O2 abundance and membrane permeability, we aim to deepen our understanding of the underlying mechanisms governing O2 homeostasis. Ultimately, this enhanced knowledge may pave the way for the development of novel therapeutic approaches targeting O2-related disorders and conditions.

Optical Sensor Integration in Architectural Glass

Authors

Shahryar Nategh, Jan Belis, Jeroen Missinne

Abstract

Over the past years, size and number of glass panels in buildings increased substantially. As glass strength is known to reduce over time as a function of e.g. the load history, it is important to know its evolution over the design lifetime. In this project, the objective is to fabricate invisible optical sensors, Bragg grating waveguide (BGW), enabling monitoring of architectural glass without jeopardizing its transparency. The investigated method uses laser technology to write invisible optical sensors, able to capture both strain and temperature by modifying the bulk of the glass. Bulk modification by laser is known for glasses used at small scales in e.g. micro-electronics, but is novel for soda-lime silica glass used at large scales in construction. In addition, with a view to next generation smart facades, the technology will be implemented on the thin chemically pre-stressed glass (aluminosilicate), which is lightweight and more resistant against scratches. Afterward, the impact of the integrated sensors on the strength of the glass will be examined using some flexural tests. In the final stage of the project, two glass components, one made of soda-lime silica and the other made of thin chemically pre-stressed aluminosilicate glass, both equipped with sensors, will be tested to demonstrate the technique’s functionality.

Computational Models to Investigate the Impact of Chiari Type 1 Malformation on Cerebrospinal Fluid Flow

Authors

Sarah Vandenbulcke, Joris Degroote, Patrick Segers

Abstract

This study focusses on the cerebrospinal fluid, which is a clear fluid circulating and interacting with the brain and spinal cord. Specifically, we aim to gain more insight into the effects of Chiari type 1 malformation, where the normal flow of this fluid is obstructed at the bottom of the skull. Importantly, this disorder frequently appears together with severe spinal cord complications, and thus sensory and motor problems. However, the mechanism leading to these complications in about 66% of the patients with Chiari type 1 malformation is unclear. Several studies have hypothesized that sudden actions such as coughing lead to abnormal fluid pressures and in that way are a driving force behind these spinal cord complications. However, measuring cerebrospinal fluid pressures is invasive and sudden actions are difficult to measure with non-invasive tools such as MRI. Therefore, we have developed a computational fluid dynamics framework to model the cerebrospinal fluid pressures and velocities. This framework allows us to investigate the impact of normal expansion and collapse of arteries as well the effects of a sudden action such as coughing on cerebrospinal fluid pressures and flow. Comparing the impact of coughing with and without obstruction can provide us with more insight into how Chiari type 1 malformation disturbs flow and pressure gradients and how coughing can even strengthen these effects.

Robust sensor selection for reconstructing thermal properties in electromagnetic devices

Authors

Faezeh Hosseini, Guillaume Crevecoeur, Hendrik Vansompel

Abstract

Electromagnetic devices have gained widespread use in various systems such as renewable energy systems, electrical motors, generators, transformers, etc. Nevertheless, due to the power losses resulting in thermal phenomena including heat transfer, it is crucial to acquire adequate insight into them to enhance energy efficiency. Furthermore, the optimal sensor placement in electromagnetic devices holds significant potential for optimizing energy efficiency. Despite utilizing state-of-the-art thermal modeling tools, there are differences in the measured thermal behavior between the prototype and the model ones. This research aims to address these challenges by primarily enhancing the thermal model of electromagnetic devices. This improvement is achieved through a combination of the finite element method and inverse modeling, enabling the reconstruction of thermal properties while considering parameter uncertainties. Additionally, the research seeks to determine the optimal number and locations of thermal sensors based on the sensitivity of thermal parameters. The problem of robust and optimal sensor placement in the presence of uncertain thermal parameters is addressed using a combination of the Gramian-based method and genetic algorithm. The experimental and simulation results demonstrate the effectiveness of the proposed approach in accurately determining the thermal parameters of the electromagnetic devices even in the presence of uncertainties associated with thermal parameters.

Poster and demo sessions

Poster and demo sessions are organized in the Chapter Room. You can find the poster IDs below.

EEG Correlates of Human-Rhythm Interaction

Poster ID

1

Session

1

Authors

Wannes Van Ransbeeck, Dick Botteldooren, Sarah Verhulst, Marc Leman

Abstract

Within the metaverse a tendency exists towards a seamless and intuitive interaction in the virtual world by an immersive and interactive environment. Capturing user sensation and experience during the interaction can help to build and maintain this immersion in any interaction, including music and rhythm. Here, besides physical user inputs, brain monitoring and bio-synchronization, can be a relevant tool for user analysis of the experience. For example, neural entrainment, being the unidirectional synchronization of neural oscillations to an external rhythmic stimulus, provides a pathway to studying these interactions and help characterizing them. Nevertheless, despite being a topic of major interest in neuroscience, correct non-invasive quantification of the entrainment and defining the emotion, experience and sensation without providing misinterpretation seem difficult. In this poster an experiment and pilot data are presented aimed at exploring potential brain markers in relation to synchronicity of a performance and associated questionnaire responses to identify bio-indicators of a good human-rhythm interaction and through it identify entrainment and user experience. A finger-taping experiment with variable beat patterns will be conducted on a cohort of 15 subjects where a subject synchronizes its tapping to the auditory presented rhythm or a variation of it. Here, both spectral band activity and the low rhythm-frequency content of the EEG activity will be evaluated locally in time and across the experiment. Two spatial filters, based on localizer trials, will allow separation of sensorimotor and perceptual activity to study their response separately. The quality of the rhythm interaction quality is defined on the basis of the physical performance, questionnaires and additionally builds on the theory of human embodiment within music interaction.

Design of High-Performance Composites via "Self-constructible Finite Element Material Library" Driven by Reinforcement-Based Machine Learning

Poster ID

2

Session

2

Authors

Ninghan Tang, Pei Hao, Francisco A. Gilabert

Abstract

Lightweight fiber-reinforced polymer composites offer a promising alternative to metal-based engineering solutions. However, understanding and predicting their complex nonlinear mechanical behavior poses challenges due to intricate microstructures and experimental limitations. Developing constitutive models for accurate Finite Element (FE) simulations demands significant expertise and time investment. This research proposes the integration of Artificial Intelligence (AI) into constitutive material modeling. Firstly, we establish a comprehensive database that combines fundamental experiments with FE-based data, enabling the categorization of elementary nonlinear thermo-mechanical features. Secondly, we develop a Neural Network (NN)-based architecture to identify nonlinear features in stress-strain responses. Lastly, we construct a self-consistent AI-based framework to determine the appropriate combination of physics-based rheological analogs needed to replicate the observed mechanical response of the material under various loading scenarios. This innovative approach harnesses existing experimental and simulation data, employing advanced AI algorithms to overcome traditional modeling challenges associated with composite materials. By digging into the existing experimental and simulated data, new material models are autonomously constructed. Furthermore, the prediction of mechanical behavior for materials under diversity of loading and environmental conditions are accomplished through this advanced method. The resulting framework serves as a valuable tool for guiding composite design and facilitating their integration across diverse engineering fields.

Prediction and Optimization of Surface Waviness in Wire and Arc Additive Manufacturing Using ANN+PSO/PSO Hybrid Model

Poster ID

3

Session

1

Authors

Jun Cheng, Wim De Waele

Abstract

This poster presents a novel hybrid model for the optimization and prediction of surface waviness of components produced by wire and arc additive manufacturing. It consists of an artificial neural network optimized by the rank-Gaussian particle swarm optimization (PSO) and combined with an RGPSO algorithm. The novelty is that RGPSO is not only used to optimize the hyperparameters of the ANN model improving its prediction performance, but also to solve the problem of optimizing surface waviness. Experimental data of waviness from literature are used to define training data, and the K-fold cross-validation strategy is applied to train, test, and validate the prediction model. The results indicate that the performance of the developed RGPSO/ANN model reaches a higher accuracy than a PSO/ANN model and an ANN model. The prediction errors of the RGPSO+ANN are within ±0.05 mm for all samples, while the PSO+ANN model has some errors that are outside of this range. The accuracy of the PSO/ANN prediction model is quantified as 0.019, 0.990, 0.013, and 3.46% in terms of RMSE, R^2, MAE, and MAPE. The RGPSO, PSO, and other optimization algorithms are applied to optimize the WAAM process parameters to reach the minimal value of waviness. The lowest value for waviness is obtained with the RGPSO algorithm and is equal to 0.1631 mm.