92 research outputs found
CFD assessment of shock flow evolution during oxygen pressure surge test
No full textpeer reviewedThe objective of this study is to perform detailed CFD assessment of the flow phenomena during the oxygen pressure surge test (OPST) at 30 MPa based on the ISO 10297 standard. Transient CFD simulations with high temporal resolution of two geometries have been carried out using the OpenFOAM software: (i) a 1000 mm long, 5 mm diameter pipe and (ii) a 750 mm long, 14 mm diameter pipe. A highlight of this study is the implementation of the opening behavior of the quick opening valve from test data as an input in the CFD simulation. This is a novel approach which results in a very realistic simulation behavior of the oxygen flow from the high-pressure driver section into the low-pressure driven section. Real gas properties, high-resolution meshing, and shear stress transport turbulence modeling are used to obtain a realistic prediction of the entire flow-field and the associated flow phenomena (e.g. supersonic shock flow, rapid compression at the end-wall, shock wave reflection etc.) during the OPST. It is observed that pressure and temperature surge near the end-wall is more pronounced in the 1000 mm length pipe implying that pipe geometry is an important factor influencing shock strength. This study thus aims to bridge the existing lack of knowledge about the flow phenomena during OPST using CFD oriented approach to aid engineers in designing oxygen products with a high degree of safety.9. Industry, innovation and infrastructur
Data driven insights into oxygen pressure surge testing
No full textThis abstract presents a comprehensive study on oxygen pressure surge testing (OPST)
in Mechanical Engineering, focusing on the critical role of data analytics and computational fluid dynamics (CFD) in enhancing safety and reliability. OPST, a vital procedure governed by the ISO 10297 standard, evaluates the performance of oxygen systems under high-pressure conditions.
In our research, we employ advanced data analytics techniques to analyze test data
collected during OPST experiments up to 100 bar. Leveraging large volumes of test data, we extract valuable insights into flow behavior and system dynamics. This data-driven approach
enables us to identify patterns, anomalies, and critical parameters that influence system safety.
Furthermore, we utilize computational fluid dynamics (CFD) simulations, conducted
using Ansys Fluent on a supercomputer cluster, to complement experimental findings. By integrating CFD with test data analytics, we gain a deeper understanding of flow behavior during OPST, particularly focusing on adiabatic compression effects and their implications for system safety.
Through meticulous analysis of CFD results and test data, we uncover nuanced insights
into pressure-temperature surges, supersonic flow characteristics, and pressure profile
variations. These insights play a pivotal role in informing engineers during the design phase, enabling them to develop safer and more reliable products.
Overall, our research underscores the importance of data-driven approaches in
Mechanical Engineering, demonstrating how the fusion of test data analytics and CFD
simulations enhances our understanding of complex phenomena like OPST. By leveraging these insights, engineers can make informed decisions and design oxygen systems that meet stringent safety standards and regulatory requirements
Synthesis and properties of water vapor sensing carbon nanohorn - cellulose sheets and nanometals
No full text
Performance Prediction Models for Deep Learning: A Graph Neural Network and Large Language Model Approach
Get PDFIn this thesis, we developed an advanced performance prediction model to estimate critical metrics of Deep Learning (DL) models, such as latency, memory consumption, and energy usage. These models are designed to support neural architecture search and efficient cloud deployment.
DL has transformed domains such as computer vision, natural language processing, climate modeling, and scientific computing. However, the increasing complexity of DL models introduces significant computational demands that require efficient hardware utilization and resource allocation. Accurate prediction of performance metrics is essential for optimizing hardware-specific compilers, enabling cost-effective cloud deployments, and minimizing environmental impacts.
To address these challenges, we present a comprehensive framework for performance prediction in this thesis. The work begins with a systematic benchmarking study of DL models, highlighting computational bottlenecks and establishing the necessity of performance prediction as a foundation for further development.
We introduce a Graph Neural Network (GNN) based performance prediction model capable of analyzing DL models from various software frameworks, including PyTorch and TensorFlow. This model predicts performance metrics and recommends NVIDIA multi-GPU instance profiles for efficient deployment. Building on this, we propose a semi-supervised performance prediction approach that leverages unlabeled data to accelerate training convergence. Using a graph autoencoder for unsupervised learning, we generate high-quality embeddings that enhance supervised training, leading to faster and more accurate predictions.
For Large Language Models (LLMs), which present unique challenges due to their extensive nodes and edges, we proposed a tree-based performance prediction model. This method significantly improves inference speed compared to traditional GNN-based techniques, making it particularly suitable for complex LLM architectures.
Finally, we explore multimodal learning by combining LLM with GNN to create a hybrid performance prediction model. This model quickly adapts to new hardware environments with sparse training samples, leveraging a novel three-stage training strategy to effectively integrate GNN and LLM for quick adaptation.U-AGR-8013 - INTER/EuroHPC/20/15077233/MAELSTROM - BRORSSON Mats Haka
A scientometric analysis of research publications on male infertility and assisted reproductive technology
No full textAssisted reproductive technologies (ART) are considered as one of the primary management options to address severe male factor infertility. The purpose of this study was to identify the research trends in the field of male infertility and ART over the past 20 years (2000-2019) by analysing scientometric data (the number of publications per year, authors, author affiliations, journals, countries, type of documents, subject area and number of citations) retrieved using the Scopus database. We used VOS viewer software to generate a network map on international collaborations as well as a heat map of the top scientists in this field. Our results revealed a total of 2,148 publications during this period with Cleveland Clinic Foundation contributing the most (n = 69). The current scientometric analysis showed that the research trend on ART has been stable over the past two decades. Further in-depth analysis revealed that density gradient centrifugation (46%) and intracytoplasmic sperm injection (59.2%) are the most reported techniques for sperm separation and ART, respectively
Fusing the facial temporal information in videos for face recognition
No full textFace recognition is a challenging and innovative research topic in the present sophisticated world of visual technology. In most of the existing approaches, the face recognition from the still images is affected by intra‐personal variations such as pose, illumination and expression which degrade the performance. This study proposes a novel approach for video‐based face recognition due to the availability of large intra‐personal variations. The feature vector based on the normalised semi‐local binary patterns is obtained for the face region. Each frame is matched with the signature of the faces in the database and a rank list is formed. Each ranked list is clustered and its reliability is analysed for re‐ranking. To characterise an individual in a video, multiple re‐ranked lists across the multiple video frames are fused to form a video signature. This video signature embeds diverse intra‐personal and temporal variations, which facilitates in matching two videos with large variations. For matching two videos, their video signatures are compared using Kendall‐Tau distance. The developed methods are deployed on the YouTube and ChokePoint videos, and they exhibit significant performance improvement owing to their approach when compared with the existing techniques
Update on the proteomics of male infertility: A systematic review
Get PDFObjective: To assess the role of differentially expressed proteins as a resource for potential biomarker identification of infertility, as male infertility is of rising concern in reproductive medicine and evidence pertaining to its aetiology at a molecular level particularly proteomic as spermatozoa lack transcription and translation. Proteomics is considered as a major field in molecular biology to validate the target proteins in a pathophysiological state. Differential expression analysis of sperm proteins in infertile men and bioinformatics analysis offer information about their involvement in biological pathways.
Materials and methods: Literature search was performed on PubMed, Medline, and Science Direct databases using the keywords 'sperm proteomics' and 'male infertility'. We also reviewed the relevant cross references of retrieved articles and included them in the review process. Articles written in any language other than English were excluded.
Results: Of 575 articles identified, preliminary screening for relevant studies eliminated 293 articles. At the next level of selection, from 282 studies only 80 articles related to male infertility condition met the selection criteria and were included in this review.
Conclusion: In this molecular era, sperm proteomics has created a platform for enhanced understanding of male reproductive physiology as a potential tool for identification of novel protein biomarkers related to sperm function in infertile men. Therefore, it is believed that proteomic biomarkers can overcome the gaps in information from conventional semen analysis that are of limited clinical utility
Advancing oxygen safety: A State-of-the-Art experimental facility dedicated to researching pressure surge in oxygen systems
No full textpeer reviewedRapid pressure surges in flow control equipment (FCE) like valves and regulators increases the risk of fire hazard in oxygen systems. This paper outlines the details of the newly established Oxygen Pressure Surge Test (OPST) facility at the University of Luxembourg. The facility is engineered to test a broad range of adiabatic compression scenarios in oxygen systems allowing for an in-depth analysis, characterization, and optimization of the FCE. The facility boasts a solenoid-actuated quick opening valve (QOV) designed for precision in adiabatic compression testing in oxygen systems. Capable of handling pressures up to 750 bar, the QOV’s operation can be tuned to have a range of opening times enabling the testing of various rapid pressure surge scenarios. The test features a robust safety framework to mitigate fire hazards. It is housed inside a secure container, equipped with advanced fire suppression systems, and includes a protective enclosure around the testing apparatus. Dynamic pressure and temperature sensors positioned close to the test sample capture critical data during rapid pressure and temperature surges, providing essential insights about the flow. The test data also serves as a validation database for CFD benchmarking which could improve the accuracy and reliability of CFD models in simulating complex flow behaviors. With the aim of performing detailed testing and characterization of oxygen FCE’s, the facility is expected to significantly contribute to enhancing the safety and reliability of oxygen systems
Adverse Drug Reaction Monitoring of Anticancer Drugs in Hematology Department
Get PDFBackground: Cancer is among the leading causes of mortality in India. Studies have reported antineoplastic agents as the common class of drugs causing Adverse Drug Reactions (ADRs). The present study aimed to conduct active surveillance of ADRs of anticancer drugs in the hematology department.Methods: A prospective observational study was conducted in 136 patients with cancer and the incidence and frequency of ADRs were assessed. The study was conducted in 6 months in a multispecialty hospital.Results: Among 136 cancer patients, All was more prevalent (39.70%); CLL, Non- Hodgkin’s Lymphoma were less prevalent (0.73%). ADRs were more prevalent in the Pediatrics department, i.e., 18.53% of ADRs were observed in patients aged <10 years. ADRs in male patients constituted 54.39%, whereas it was 45.60% in female patients. Cytarabine caused the highest number of ADRs (34.48%). The most prevalent ADR was anemia (25.60%).Conclusion: Multiple ADRs were detected in cancer patients. We found that hematological ADRs were more prevalent. Most of the ADRs were possible reactions according to Naranjo and the World Health Organization (WHO) scales
- …
