1056 research outputs found
Sort by
ConcreteShellFEA: A surrogate modelling dataset for the buckling and stress behaviour of concrete thin-shells
No full textConcreteShellFEA is a dataset designed for the training of deep learning models to predict buckling loads and stress fields in concrete thin-shell structures.
It contains 3 smaller datasets, which can be used for different use cases:
1. PerfectShell_LinearFEA: A dataset of 20,000 thin-shells (with various span, height, thickness, and Young's modulus), for which buckling factors and stress fields under design loads were determined using linear Finite Element analysis. The data is presented in three formats (tabular, image, graph) to enable different types of deep learning models (Multilayer Perceptrons, Convolutional Neural Networks, and Graph Neural Networks) to be trained.
2. ImperfectShell_LinearFEA: A dataset of 20,000 imperfect thin-shells (with various span, height, thickness, Young's modulus, and geometric imperfections), for which buckling factors and stress fields under design loads were determined using linear Finite Element analysis. The data is presented in two formats (tabular, image) to enable different types of deep learning models (Multilayer Perceptrons, Convolutional Neural Networks) to be trained.
3. PerfectShell_NonlinearFEA: A dataset of 25,000 thin-shells (with various span, height, thickness, and Young's modulus, and geometric imperfections), for which buckling factors under design loads were determined using Finite Element analysis. The buckling factors were determined using linear Finite Element analysis for 20,000 thin-shells, and using nonlinear Finite Element analysis for 5,000 thin-shells, to enable mixed-fidelity applications. The data is presented in a single format (tabular).The methods used to generate this dataset can be found in the associated thesis.The input data and output stress data are in HDF5 (.h5), and the output buckling data is in CSV (.csv). The input images are saved in HDF5 (.h5), and the input scalars in CSV (.csv). The graphs are saved in the PyTorch format (.pt).
The data was generated by conducting Finite Element simulations using the ANSYS software (version 2021 R1). The simulations' configurations are detailed in the associated thesis.
The Python scripts included in the dataset (load_tabular_datasets.py, load_image_datasets.py, and load_graph_datasets.py) rely on the dependencies listed in requirements.txt.The original folder structure is given in README.md. To reproduce it, create a folder "ConcreteShellFEA" with two subfolders, "datasets" and "scripts". Extract the contents of the datasets ZIP folders (datasets-01.zip to datasets-28.zip) in the "datasets" folder and move the three Python scripts (load_tabular_datasets.py, load_image_datasets.py, and load_graph_datasets.py) inside the "scripts" folder
Dataset for "Feasibility and acceptability of 7-day smartphone-based, activity-triggered Ecological Momentary Assessment among low-income older adults"
No full textSmartphone-based Ecological Momentary Assessment (EMA) is increasingly used to collect real-time data on physical activity behaviour. The current study aimed to assess the feasibility and acceptability of activity-triggered EMA in low-income older adults. For 7 days, 39 older adults (76.4 ± 8.5 years; 76% earning below £25,000/year) received EMA surveys, delivered via the movisensXS application (version 1.5.23, movisens GmbH, Karlsruhe, Germany) for Android operating systems, when they surpassed a predefined activity/inactivity threshold, or when two hours elapsed between prompts. Participants wore a Move 4 activity sensor (movisens GmbH, Karlsruhe, Germany) to measure their steps. A post-study questionnaire assessed perceptions of acceptability.
The dataset includes all quantitative data needed to replicate analyses in the article "Feasibility and acceptability of 7-day smartphone-based, activity-triggered Ecological Momentary Assessment among low-income older adults." The "Descriptives" sheet contains a unique participant identifier, demographic information, and responses to the post-study questionnaire. The "EMA" sheet contains a unique participant identifier (Participant_ID), age (Age_years), biological sex (Biological_sex), time of day (Time_of_day), day of week (Weekday), and EMA compliance (EMA_compliance; whether participants completed the EMA prompt or missed the EMA prompt) variables needed to perform the multilevel logistic regression models. It also contains the data necessary to limit the sample to participants with valid activity sensor wear and run Model 2, including the length of time in minutes that participants were not wearing the activity sensor in the 15-minute window before (Nonwear_before) and after (Nonwear_after) the EMA survey, and concurrent physical activity (Concurrent_PA; the number of steps in the ± 15-minute window around the EMA prompt).
Day of study (day number from 1 to 7), trigger type (whether participants received an activity-triggered, inactivity-triggered, or timeout EMA prompt), trigger time (absolute time of the auditory signal and/or vibration alerting participants that it was time to complete an EMA survey), EMA outcome (whether the EMA prompt was completed, not answered, or answered but incomplete), form start time (absolute time when the EMA survey was answered), form completion time (absolute time when the EMA survey was completed), observation number (variable that assigns the observation number to each row by participant ID), and observation counter (variable that assigns the number of total observations to each row of data for a given participant) variables are also provided to enable researchers to replicate all of the summary statistics presented in the article. A complete description of the variables, including the text of questionnaires (where relevant), is provided in the "Overview" sheet.Full details of the data collection methods are provided in the published article. If an EMA survey was delivered at the end of the introductory appointment or during an interim appointment, participants had the option of completing it as a practice survey with the research team. These responses were discarded from analyses and have been omitted from the current dataset. The DataAnalyzer software (version 1.15.1; movisens GmbH, Karlsruhe, Germany) calculated steps with a 60-second resolution. The raw data was aggregated (using the "SUMIFS" function in Microsoft® Excel® for Microsoft 365 MSO, Version 2502 Build 16.0.18526.20416 64-bit) to obtain the length of time in minutes that participants were not wearing the activity sensor in the 15-minute window before (Nonwear_before) and after (Nonwear_after) the EMA survey, as well as the number of steps in the ± 15-minute window around the EMA prompt (Concurrent_PA). This aggregated data is presented in the current dataset.Data were processed in Stata BE version 18.0 (StataCorp, College Station, TX), and multilevel models were performed in R version 4.4.1 with RStudio version 2024.04.2. Processing and analytic code (Stata and R) are available in the GitHub repository at https://github.com/OliviaMalkowski/EMA-feasibility.git. First, run the file "2025-08-19_Stata-do-file_v01" in Stata to replicate the summary statistics presented in the associated journal article and to prepare the datasets (i.e., limiting the sample to participants with valid activity sensor wear for Model 2) ahead of performing multilevel logistic regression modeling. Then, knit the file "2025-08-19_Feasibility_v01" in RStudio to run the multilevel logistic regression models (observations nested within persons) regressing EMA compliance on time-invariant (i.e., age, biological sex) and time-varying (i.e., time of day, day of week, concurrent physical activity) factors. The dataset is saved in XLSX format (given the multi-sheet capability of XLSX files) and can be opened with Microsoft Excel. The separate sheets (Overview, Descriptives, and EMA) are saved in CSV format and can be opened with any software that supports CSV files, including Microsoft Excel.Rather than uploading a 'Readme' file or questionnaire templates, all information that would assist understanding and enable reuse of the data has been provided in the "Overview" sheet of the dataset
Dataset for "Co-creation of an airflow and COVID-19 transmission risk model for shelter design"
No full textThis dataset underpins a journal article titled "Co-creation of an Airflow and COVID-19 Transmission Risk Model for Shelter Design." The paper introduces the first collaboratively developed tool designed to guide shelter design by ensuring adequate natural ventilation, optimal indoor air quality, and minimized airborne transmission risks. This study explores the development and application of this tool to promote healthier shelters and enhance the shelter design process.
Data was collected using the JISC online tool across two phases: the first before the tool's creation and the second after its implementation by participants. The dataset includes responses from online surveys conducted with participants from various global locations. It encompasses information on shelter designers' experience in shelter construction, their background knowledge of natural airflow and indoor air quality, and feedback on the usability of the co-created tool.The data have been collected through two online surveys using JISC online tool to comply with GDPR and Certified to ISO 27001 standards. All the participants were enrolled on a voluntary basis after they signed an informed consent. This research has received ethical approval from the University of Bath - ethics application reference number: 2448-2592.The data was collected using the JISC online tool across two phases: the first before the tool's creation and the second after its implementation by participants. The dataset, stored in a CSV file, includes responses from online surveys conducted with participants from various global locations. It encompasses information on shelter designers' experience in shelter construction, their background knowledge of natural airflow and indoor air quality, and feedback on the usability of the co-created tool. The data was further processed and analyzed using a Python script to derive insights and conclusions presented in the article
Dataset for "Pioneering Net Zero Carbon Construction Policy in Bath & North East Somerset: Evaluating the effectiveness of novel planning policies over time"
No full textThis data was collected as part of a continuing collaboration between the University of Bath and Bath and North East Somerset Council, exploring the impacts of (and reception to) pioneering sustainable planning policies for new buildings which were first introduced in January 2023. This project evaluates the success of the policies two years on, establishing long-term trends, opportunities for refinement, and the national policy implications of this unique policy case study.
The deposited data relates to two parts of the methodology.
The first is an analysis of incoming planning application, relating to the characteristics of proposed buildings and key parameters submitted to comply with the net zero energy requirements.
The second is the results of a questionnaire sent out to applicants.This dataset was generated through a mixed‑methods approach designed to capture both quantitative performance data and qualitative stakeholder perspectives relating to the B&NES sustainable construction policies. First, all eligible planning applications submitted between May 2024 and June 2025 were systematically reviewed. Energy summary data, SAP/PHPP modelling outputs and associated documentation were extracted to assess compliance with operational energy and embodied carbon requirements. These submissions were analysed at the individual‑plot level, enabling comparison of design parameters such as space heating demand, total energy use, U‑values and air permeability.
To complement this, a structured questionnaire was distributed to planning agents and consultants to gather insights into applicant experience, perceived challenges and evolving practice.Data was anonymised with any personal identifying information redacted
Dataset for "Fast structural analysis of concrete thin-shells using deep learning"
No full textThis dataset contains scripts and data supporting the following research article: Pollet, M., Shepherd, P., Hawkins, W., and Costa, E., 2026. Fast structural analysis of concrete thin-shells using deep learning. Computers & Structures, 320, 108042.
Concrete thin-shells are materially efficient structures, which can be used to reduce the environmental impact of concrete structures. Their shape is typically determined iteratively and evaluated through Finite Element Analysis (FEA). This research proposes the use of surrogate models as faster alternatives to FEA, thus enabling wider design space exploration.
This dataset contains deep learning models – Multilayer Perceptrons, Convolutional Neural Networks, and Graph Neural Networks – that have been trained to predict the buckling factor and stress fields of concrete thin-shells of various shapes under design loads. It also contains the Python scripts that were used to train these models and assess their performance. Running these scripts necessitates the associated ConcreteShellFEA dataset to be downloaded. Further details about this data can be found in the related research article.Full details of the methodology used may be found in the associated article.The data in the models and results folders was generated using the Python code in scripts folder. These scripts rely on the dependencies listed in requirements.txt.The original folder structure is given in README.md. To reproduce it, create a folder "FastStructuralAnalysisOfConcreteThinShellsUsingDeepLearning" and extract the "models.zip" and "results.zip" folders inside. Additionally, create a "scripts" folder and store all Python scripts inside.
The path to the ConcreteShellFEA dataset needs to be specified in each script, under the DATASET_ROOT variable
Dataset for Nonlinear viscoelastic models improve characterisation of 6 DOF intervertebral disc load response at low strain rates
No full textThe data presented in this file comprises the 6 DOF position control data from tests performed on 6 porcine lumbar isolated spinal disc specimens using a triangular displacement waveform at a frequency of 0.1Hz. Data is presented relative to the centre of the intervertebral disc. The final three cycles are presented here. For each of the six specimens and for each of the six axes, data is provided for the applied displacement and each of the 6 resulting loads (forces and moments). The principal elements (highlighted in the tables) occur when the direction of applied displacement and measured load are the same - for example, axial torsion displacement (RZ) and torsional load (MZ). The data has been filtered to remove unwanted noise but no other pre-processing steps have been performed on this data - for example, cycle averaging and offsetting the central point to the origin.The methodology can be found in the associated paper
Dataset for a framework for assessing the impact of geometric imperfections in concrete shell structures using deep learning
No full textThis dataset contains scripts and data supporting the following following thesis: Pollet, M. (2025). Rapid structural analysis of prefabricated thin concrete shells using deep learning (Thesis). University of Bath.
Concrete thin-shells are materially efficient structures, which can be used to reduce the environmental impact of concrete structures. However, geometric imperfections, which may occur during production can negatively impact their structural behaviour. While this impact can be assessed through Finite Element Analysis (FEA), a faster analysis method, such as surrogate modelling, could benefit concrete shell manufacturers.
This dataset contains deep learning models – Multilayer Perceptrons, and Convolutional Neural Networks – that have been trained to predict the buckling factor and stress fields of geometrically imperfect concrete thin-shells of various shapes under design loads. It also contains the Python scripts that were used to train these models and assess their performance. Running these scripts necessitates the associated ConcreteShellFEA dataset to be downloaded. Further details about this data can be found in the related thesis.The methods used to generate this data can be found in the related thesis.The data in the models and results folders was generated using the Python code in scripts folder. These scripts rely on the dependencies listed in requirements.txt.The original folder structure is given in README.md. To reproduce it, create a new folder and extract the "models.zip" and "results.zip" folders inside. Additionally, create a "scripts" folder and store all Python scripts inside.
The path to the ConcreteShellFEA dataset needs to be specified in each script, under the DATASET_ROOT variable
Dataset for "High freeze-casting cooling rates enhance the piezoelectric responses and reproducibility of porous lead zirconate titanate for sensing and energy harvesting"
No full textThis dataset is a part of the research article 'High freeze-casting cooling rates enhance the piezoelectric responses and reproducibility of porous lead zirconate titanate for sensing and energy harvesting'. It contains comprehensive characterization data for ferroelectric lead zirconate titanate PZT NCE51 ceramic, fabricated using a range of freeze-casting cooling rates ranging from 1 to 4 °C/min. This dataset contains hysteresis polarization-electric field loops, impedance spectroscopy data and scanning electron micrographs, which provide insights into the hierarchical relationships between processing, microstructure, and properties in freeze-cast ferroelectrics.
The dataset also contains the results from finite element modeling, demonstrating the effects of wall thickness (or mechanical clamping), pore channel defects (i.e., ceramic grains within pore channels) and wall defects (i.e., pores within ceramic walls) on bulk electromechanical properties. A model representing the residual stress state after poling, which demonstrates how residual stresses influence thermal stability in terms of piezoelectric properties of lead zirconate titanate near the Curie temperature.
This may be of interest to researchers focused on the design and characterization of advanced ferroelectric composites.Full details of the methodology can be found in Section 2 of the associated research article
Dataset for "Influence of Block Microstructure on the Interaction of Styrene-Maleic Acid Copolymer Aggregates and Lipid Nanodiscs"
No full textCopolymers between styrene and maleic acid are able to extract membrane proteins directly from cells, reconstituting lipid membranes into nanodiscs. RAFT copolymerisation was used to generate copolymers of equivalent molecular mass but inverted block sequences and end group termini. This dataset contains characterisation data for the copolymers (GPC, NMR, FTIR, UV-vis), included deuterated variants for neutron scattering experiments, as well as the structures formed in solution. Aggregates were assed by a combination of DLS and surface tension measurements, and nanodisc formation kinetics through UV-vis using both model DMPC vesicle and E.coli membrane suspensions. It was found that mismatched hydrophilic and hydrophobic end groups on the respective styrene block and alternating block, impeded membrane solubilisation. This highlights not only how the amphiphilic balance of these blocks is important for efficient nanodisc formation, but also how end groups influence these and may be optimised towards the extraction of more challenging MPs.Data collection methods are described in full in the publication "Influence of Block Microstructure on the Interaction of Styrene-Maleic Acid Copolymer Aggregates and Lipid Nanodiscs". Briefly, various copolymers between styrene and maleic anhydride were prepared by RAFT polymerisation, which, when using DDMAT, results in a relatively-large and hydrophobic SC12 end group (SMAnh-SC12). This block sequence was then inverted by first synthesising a poly(sty) macro-RAFT agent, from which a Sty:MA alternating block may be polymerised. A commercial variant, SMA2000, synthesised by free-radical polymerisation was also used for comparison. All copolymers were then hydrolysed to the acid form (SMA) before workup and purification.1H and 13C NMR:
Spectra were analysed using Mestrelab MNova 11.0 software where spectra were baseline corrected and line broadening used to allow accurate integration of peak area.
GPC:
Chromatograms were analysed in Agilent GPC/SEC software to extract Mn and PDI values.
UV-vis:
The presence of the SC12 end group can be monitored by the peak at 310 nm in UV-vis spectra. Resultant spectra were normalised by the styrenic absorbance peak at 262 nm.FTIR:
FTIR measurements were conducted on a Perkin Elmer ATR desktop spectrometer with solid-state polymer samples at room temperature.
1H & 13C NMR:
1H and 13C NMR spectra were recorded on an Agilent 500 MHz spectrometer at room temperature using d6-acetone (for anhydride species) or D2O (for acid species) as the solvent.
GPC:
GPC was conducted using an Agilent GPC 1260 Infinity chromatograph using two PLgel 5μM MIXED-D 30 cm x 7.5 mm columns with a guard column PLgel 5 μm MIXED Guard 50 x 7.5 mm. The column oven was maintained at 35 °C, with GPC-grade THF as the eluent at a flow rate of 1.00 mL/min and refractive index detection and polymer concentrations between 1.0 – 2.0 mg/mL. The system was calibrated against 12 narrow molecular weight polystyrene standards with a range of Mw from 1050 Da to 2650 kDa.
DLS:
DLS was conducted using a Malvern Zetasizer Nanoseries at theta = 173 degrees (backscattering) and wavelength = 633 nm.
Pendant Drop Tensiometry:
Tensiometry was conducted on a FTA 1000 contact angle/surface tension analyser and processed using FTA 32 surface tension image analysis software. Syringe needles were prepared by extensive washing before SMA polymers in PBS at variant concentrations were passed through these to produce a small hanging droplet which was imaged at a typical rate of 10 images per second for 10 seconds.
SANS:
SANS was performed at the ISIS Neutron and Muon Source (Rutherford Appleton Laboratory, Didcot, UK), on the SANS2D instrument (doi:10.5286/ISIS.E.RB2010215), using 1 mm quartz Hellma cells at 25 °C. Prior to experiments, samples were mounted in a temperature controlled multi-position sample changer. Data were subsequently reduced using Mantid software and the varying solution contrasts simultaneously fit using the NIST SANS analysis package within IgorPro
Dataset for Threshold screw insertion torque for carbon fibre-reinforced polyetheretherketone and titanium (Ti-6Al-4V) locking plate constructs
No full textDescription of the data file
This dataset contains the raw and processed mechanical testing data generated for the study investigating the influence of insertion torque on the performance of locking screw constructs in carbon-fibre reinforced polyetheretherketone (CFR-PEEK) and titanium alloy (Ti-6Al-4V) plates.
The data include results from single-screw and two-screw constructs tested across predefined insertion torque levels (0.5–3.0 Nm). Mechanical outcomes were assessed using axial push-out testing and cantilever bending testing. For single-screw constructs, additional video-based measurements of screw rotation during insertion are provided to quantify the relationship between applied torque and angular displacement.
Each entry records plate material, construct configuration (single- or two-screw), insertion torque, test modality (push-out or cantilever bending), and the corresponding mechanical performance metrics. Statistical groupings used in the analyses reported in the manuscript are identifiable within the dataset.
The dataset is provided in tabulated format and is sufficient to reproduce all analyses and figures presented in the associated manuscript, as well as to enable secondary analysis of torque–performance relationships in locking plate constructs.Data collection method
Data were generated through controlled laboratory experiments designed to assess the mechanical performance of locking screw–plate constructs under varying insertion torque conditions. Locking screws (3.5 mm diameter, Ti-6Al-4V) were inserted into either CFR-PEEK or Ti-6Al-4V locking plates using a calibrated torque-limiting device. Screws were inserted perpendicular to the plate at predefined target torques ranging from 0.5 to 3.0 Nm. Each construct was assembled under standardized conditions to minimise variability related to alignment, insertion angle, and operator technique.
Single-screw constructs were tested across six torque levels (0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 Nm). Additional two-screw CFR-PEEK constructs were assembled and tested at three torque levels (0.5, 1.5, and 2.5 Nm). For each construct, axial push-out testing and cantilever bending testing were performed using a mechanical testing system under displacement-controlled conditions. Load–displacement data were continuously recorded throughout testing until construct failure or predefined endpoint criteria were reached.
Screw insertion was recorded using high-resolution video for single-screw constructs, and video analysis was subsequently performed to quantify screw rotation during insertion. Rotation measurements were synchronised with applied torque values to characterise the relationship between insertion torque and angular displacement.
All testing was conducted at room temperature under consistent environmental conditions. Sampling was based on predefined experimental groups rather than random sampling, with replicate specimens tested at each torque level to allow statistical comparison of mechanical performance outcomes
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇