Data scientist with 12 years of experience of computational and experimental research projects. Experienced in processing complex and messy data sets, using modelling and machine learning to inform decision-making and policy interventions, including informing non-technical audiences. Specific expertise in environmental impact of domestic energy use and low carbon technologies.
I am passionate about driving future research and decision-making with transparent and reproducible analysis of data and evidence, and making the findings of this analysis accessible to both technical and non-technical people.
I currently work as a Data Scientist at the Jean Golding Institute, the central hub for data science and data-intensive research at the University of Bristol. I have an MSci in Physics (Natural Sciences) and am a Member of the Institute of Physics.
I am working on collaborative Urban Analytics projects across a range of University of Bristol groups (Geography, Engineering, UKCRIC Observatory, Bristol Digital Futures) and external partners and stakeholders (including industry, government and the third sector).
Two specific areas of focus are achieving carbon neutrality by 2030 and reducing health and well-being inequalities, including a range of urban mobility projects.
Familiar with Python, Numpy, Scipy, Pandas, Scikit-learn, Jupyter, Git, I complement my knowledge of data analysis, modelling and machine learning with hands-on experience of remote data collection, large-scale field trials, and efficiency and emissions testing both in the field and in the laboratory
BEIS Biomass Boiler Field Trial
Remote efficiency and pollutant emission measurement
I was project manager for Phase II of the BEIS Biomass Boiler Field Trial, studying 67 domestic and non-domestic biomass heating systems. I am lead author on the report, and led the detailed analysis of the performance of the systems, including the effects of thermal stores and controls.
I designed and implemented a bespoke algorithm to calculate the efficiency of the appliances. The inputs to the algorithm were various data points at one-minute intervals, for the duration of the trial (12–24 months). The algorithm determined the operating mode of the appliances each minute and then selected an appropriate set of equations to calculate their efficiency. The algorithm was implemented in Python, Numpy and Pandas, and accelerated with Numba; graphs were generated using Matplotlib.
I made intervention visits at domestic and commercial sites (space and water heating), and carried out performance testing at a site with several kilometres of underground heat main and many plant rooms.
I chaired two stakeholder engagement days. The attendees were from industry, trade bodies, academia and government. Our role was to present interim findings and gather feedback from those present. As chair, I ensured all the attendees were able to comment on the findings, but also had to clearly define the boundaries and time limits around discussions, to ensure all attendees were fairly represented. I also then led an analysis of the feedback gathered to determine if the project plan should be altered as a result.
SGN H100 Fife & Cadent Hydrogen Hubs
Feasibility modelling of integrated energy systems
I led the modelling of integrated energy generation and storage systems for two feasibility studies. These including consideration of varying wind generation, hydrogen production, storage requirements and demand for heat and transport.
A key challenge was matching the variable generation from wind turbines with the variable demand of a distribution network. Various electrolysers were evaluated, however each of these had their own constraints. I designed theoretical models of each electrolyser and used this to model various generation, demand and storage scenarios in terms of efficiency and cost.
I designed a bespoke processing pipeline to simulate one year of operation under varying conditions (best, average, worst case according to historical data and statistical distributions) using Python/Numpy and JIT-accelerated using Numba. Results were reported in fully-documented Jupyter Notebooks to allow for easy transfer of findings to reports, and re-use of code in future projects.
BEIS Hydrogen for Heat Programme (Hy4Heat)
Safety assessments using gas escape surveys and testing
I collected and analysed data in a safety assessment for the use of hydrogen gas in domestic properties, which required knowledge of the frequencies of different types of gas escape and how gas accumulates inside a property.
I designed and implemented a web-based system (using Python/Django) to collect data from First Call Operatives (FCOs) during emergency gas-escape call-outs. Working with Gas Distribution Network Operators (GDNOs), I deployed the system and conducted face-to-face and video training sessions. I analysed the surveys using both expert-system (implemented in Pandas) and machine learning models (using Scikit-learn), and devised and managed the process for quality assurance of the models using a team of experts, whilst providing regular dashboards to stakeholders.
I also programmed a system to extract and compare gas accumulation data with theoretical models of gas dispersion, both for technical review and quality assurance of the contractor-collected data and for the evaluation of the appropriateness of the theoretical models.
SGN H100 Consequences & DECC HyHouse
Natural gas and hydrogen gas escape and explosion safety testing
I designed the data collection and processing system for a series of hydrogen feasibility & practical safety trials in various domestic settings, for both SGN's H100 Consequences and DECC's HyHouse projects. I also led on several phases of the field work, data analysis and reporting.
My data processing system, written using Numpy, Scipy & Pandas, took data exports from high-speed voltage recorders (10,000 Hz) and applied various filters (low-pass, custom Fourier, median filters, etc.), before identifying and quantifying sudden changes; graphs were generated using Matplotlib. Additional graphs of low-speed incoming data were displayed on a low-powered laptop using a user-customisable Excel dashboard.
As part of the SGN H100 programme, I also designed and implemented the data acquisition system for a set of gas chromatographs and led the data analysis of a series of laboratory trials to demonstrate the effectiveness of odourant as a means of detecting hydrogen gas escapes.
Heat Pump Field Trial
Remote efficiency measurement and data analysis
I was a data analyst in the EST Heat Pump Field Trial (2013) and more recently led the analysis and authored reports in two heat pump field trials for manufacturers. Processing was carried out in both Python using Pandas, and Excel with Visual Basic.
A key objective was independently assessing the energy efficiency of the product at different system boundaries (according to the SEPEMO methodology).
Utility Meter Data
Smart meter data analysis
I led the analysis and co-authored the report on a review of gas and electricity data from over 1,000 domestic properties in a trial of prototype smart meters for a large utility company.
Key achievements included quantifying the energy performance of the properties and identifying those at which energy saving measures could be made, ranging from controls adjustments, to retrofit of newer heating appliances and improved insulation.
EU Emissions Trading Scheme (CO2)
I was Lead Verifier in the EU Emissions Trading Scheme Phase 3 (2015–2020), undertaking audit visits to industrial sites and verifying the application of mass-balance measurement systems for carbon emissions.
I am familiar with processes in the steel, glass, ceramics and lime industries, and carried out audit visits as a sole Lead Verifier, or leading a team of Verifiers.
I was a member of an internal technical review panel for applications to the Mayor of London Cleaner Heat Cashback scheme (2018–2020).
In 2016, I led a team of independent QA reviewers for the BEIS field trial of non-domestic Renewable Heat Incentive ground-source and water-source heat pumps.
I have also carried out internal audits, as a part of a testing laboratory accreditation under ISO 17025.
In 2015, I managed an expert investigation into boiler performance for a local authority, visiting 15 sites that were operated by a mix of personnel with differing technical knowledge, from caretaker to building services manager. Stakeholder appetite to the investigation varied and during the short site visits I had to quickly build a level of trust with them to enable sufficient data to be collected and technical drawings of the heating system to be made. I then had to build on knowledge from two technical experts and summarise the themes of performance and health & safety issues observed at the sites in a report for the client.
Also in 2015, I worked in a consortium of experts providing technical advice to Ofgem around the use of blended virgin and waste biomass fuels. Regular telephone conferences and exchange of comments on SharePoint were used to ensure all experts agreed with the technical output of the final report.
I performed practical and paper-based reviews of the performance of energy saving devices for manufacturers (2010–2012). These included advanced heating controls, central heating system debris filters, and water restriction devices, including one for the Carbon Emissions Reduction Scheme (CERT).
Outside of my professional role I develop websites for local non-profits and sports associations, with the aim of facilitating effective communication with these groups and their members.
I develop websites using Python, Django and Wagtail, and host them on virtual machines running Debian/Ubuntu.