Petroleum engineering undergraduate study
Petroleum engineer · PhD researcher
I am a petroleum engineer and PhD researcher specialising in drilling engineering, with a focus on water-based mud and biodegradable drilling additives. I also work in production and reservoir engineering.
I now combine my petroleum engineering background with machine learning and data analytics to turn difficult field and laboratory data into insights engineers can act on.
Petroleum engineering undergraduate study
Petroleum engineering master’s research
Petroleum engineering data analytics
PhD petroleum engineering, machine learning and AI

Beneficiation cuts fluid loss by up to 82 % and still never reaches the API limit of 15 mL, though it does lift most treated samples past the viscosity minimum. Only 12 of the 71 formulations were tested on both requirements, and the single one passing both is imported bentonite. The database also caught two arithmetic defects in the published tables, including a negative yield point.
QuestionCan a locally sourced biodegradable additive replace imported CMC or xanthan gum, and at what concentration?
MethodModels predicting plastic viscosity, yield point and fluid loss from additive type and dose, with feature importance used to rank which variables carry the performance.
Nov to Dec 2026
QuestionNatural polymers degrade downhole. How much viscosity and fluid-loss control is lost as temperature and aging time rise, and how deep can each additive be trusted?
Jan to Feb 2027
QuestionHow do green inhibitors perform against potassium chloride on linear swelling and shale recovery tests?
Feb to Mar 2027
QuestionHow have flaring volumes moved by field and operator, and what does the trend forecast?
Mar to Apr 2027

Spear grass (Imperata cylindrica) was pulverised and mixed into bentonite water-based mud, then tested for rheology, pH, mud weight and fluid loss at room temperature, 70 °C and 100 °C with salt contamination from 5 to 25 g. At the same 1 g loading it reduced fluid loss more effectively than xanthan gum and improved other mud properties under saline conditions.
doi.org/10.2118/221580-MSFollow-on study taking the Chrysophyllum albidum seed oil bio-lubricant through to a formulated oil-based drilling mud and characterising the resulting fluid.
DOI as printed on the paper: 10.47545/ertj.2021.6.2.083. Link pending verification.
A polyol-ester bio-lubricant was synthesised from African star apple (Chrysophyllum albidum) seed oil as a replacement for diesel in the continuous phase of oil-based mud. Density, viscosity, pour point and flash point met ASTM standards and compared favourably with diesel, with lower toxicity and better biodegradability.
doi.org/10.9790/264X-0602012732Nickel-iron and nickel-copper-iron composite electrocatalysts were electrodeposited onto copper foam and used as cathodes in alkaline water electrolysis, then compared on hydrogen yield, purity, overpotential, Tafel slope and stability. Hydrogen purity reached 88.36, 88.59 and 92.13 % for the copper foam, nickel-iron and nickel-copper-iron cathodes, the ternary composite performing best with 78.3 % energy efficiency after 20 hours.
doi.org/10.53982/ajerd.2026.0902.08-jA green corrosion inhibitor was built from waste cassava peel starch, chitosan from snail shells and green-synthesised silver nanoparticles, then tested on mild steel in dilute hydrochloric acid. The composite reached up to 97 % inhibition efficiency by forming a protective film, the nanoparticles improving stability and the biopolymers keeping it non-toxic.
doi.org/10.53982/ajerd.2025.080.13-jSemi-empirical models were fitted for cell voltage, Faraday efficiency and gas purity in alkaline water electrolysis, using lye from charred plantain peel and caustic soda from seawater at 25 to 35 g/L and 45 to 85 °C. Most models fitted well, with R² between 0.96 and 0.999.
doi.org/10.53982/ajerd.2025.0801.32-jHydrogen was produced by electrolysis of the urea in human urine, modelled empirically across three electrolyte concentrations and five temperatures from 45 to 85 °C. Cathode gas reached 99.88 % hydrogen, with output highest at 35 wt. and 65 °C, and most models fitting with R² between 0.96 and 0.99.
doi.org/10.53982/ajerd.2024.0702.39-jEthanol extracts of sugarcane bagasse and garcinia kola fruit pod were tested as green corrosion inhibitors for mild steel gas pipeline in acid, using a central composite design across concentration, exposure time and temperature. Best efficiencies were 92 % and 90 % respectively, with garcinia kola the stronger inhibitor overall.
doi.org/10.19184/cerimre.v7i2.50834A Taguchi orthogonal array was used to find the conditions that maximise hydrogen volume in alkaline water electrolysis with plantain-peel lye and seawater-derived caustic soda. The optimum was 30 g/L, 15 minutes, 85 °C and 7 V for KOH, and 35 g/L, 15 minutes, 45 °C and 7 V for NaOH.
doi.org/10.5281/zenodo.10049652Reports a machine learning approach for generating Poisson’s ratio maps, used to determine rock and drilling direction features. Machine learning is applied to inspect the quality of large volumes of drilling data, extract the critical records and forecast downtime, reducing the time and cost of checking that data. The conclusion drawn is that machine learning can fundamentally alter the decisions oil and gas managers and engineers make.
routledge.com ↗Sets out proven machine learning and deep learning algorithms for anomaly detection, and argues that AI enabled analytics yield prescriptive insight as well as prediction. Automation, orchestration and optimisation are presented as the competencies that bring process excellence to oil and gas operations, and with it the adaptivity, adroitness and affordability of those systems and services.
ISBN 978-1-119-98558-7Attributes the rise of machine learning in the energy industry to new technologies such as sensors and high performance computing services, among them Apache Hadoop and NoSQL, which make big data acquisition and storage practical across different fields of study. Big data is defined as a quantity too large to be gathered, stored and analysed with common tools, on the order of terabytes. The climbing publication count on OnePetro and with the AAPG is cited as evidence, with the expectation that adoption will widen as more companies see the value added.
shop.elsevier.com ↗The formulation data comes from peer-reviewed studies, extracted by hand. Where a dataset is partly synthetic, it says so.
Temperature, aging condition and concentration travel with each measurement. A model trained across mixed protocols learns laboratory differences and calls them chemistry.
IMRaD structure, references managed in Zotero, and a limitations section written to be read rather than skipped.
Each project ships its dataset, its notebook and a README, so the analysis can be checked and repeated.
I am currently exploring opportunities in oil and gas, drilling and machine learning, and always open to connecting with like-minded professionals. If you have a role, project or collaboration in mind, I would love to hear from you.
