David S. Fulford

Last updated October 28, 2026

Research artifacts

• Transient Hyperbolic Model (SPE-167242) November 2013 · SPE Unconventional Resources Conference – Canada

  Fulford and Blasingame 2013: the THM, its segment forms, the analytic solution from reservoir properties, the numerical cases, MFHW heterogeneity, a fit to your own data, and a radial THM with the Fetkovich type curve.

• Bayesian Derivatives & Optimization (SPE-201404) October 2020 · SPE Annual Technical Conference and Exhibition

  The Bayesian pressure and rate derivative: the latent log-log slope with its posterior bands, the regularisation path and NUTS for the paper's pressure, rate and qDb cases and your own data, checked against the PyMC 6 posteriors, computed in the browser by the bayesian-derivative WebAssembly engine.

• Multiphase Flowing Material Balance – Bayesian Pore Volume Estimation 2022 · after Thompson and Ruddick, URTeC 3718045

  Multiphase flowing material balance with a Bayesian latent slope: the density plot, the pore-volume trajectory, β*, rates, ratios and saturations for three simulated wells, the KITE field well and your own data, computed in the browser by the bayesian-derivative WebAssembly engine.

Open source packages

• petbox-dca Python · PyPI: petbox-dca

  Decline curve analysis: Arps hyperbolic, Modified and Transient Hyperbolic, generalized multi-segment hyperbolic, power-law exponential, stretched exponential and Duong models, with secondary and water yield models attachable to any primary phase.

• gwr Python · PyPI: gwr-inversion

  Numerical inversion of Laplace transforms by the Gaver-Wynn-Rho algorithm (Valkó and Abate, 2004) at arbitrary precision, with composable Laplace-domain reservoir models.

• combocurve-api-helper Python · PyPI: combocurve-api-helper

  A client for the ComboCurve REST API: projects, wells, production, forecasts and type curves, forecast runs and econ models, including batched forecast-parameter writes.

• tafra Python · PyPI: tafra

  A minimalist dataframe: named, typed numpy columns with SQL-style group-by and joins, a lightweight substitute for some uses of pandas.

Blog

• Bayesian Networks Are Decision Making Tools February 2020

  Bayesian networks for inferring missing or latent variables, worked through the Monty Hall problem.

• Convolution in the Oilfield October 2018 · updated February 2020

  Using numpy.convolve to build a rig-line production profile from a type well and a schedule of well start dates.

• Non-Linear Regression on Production Data January 2019

  Fitting non-linear models to production data directly, and why the choice of cost function matters more than least squares suggests.

• What's in a Mean? A Visual Primer on Linear Regression December 2018

  Linear regression from the ground up, starting from the arithmetic mean as its simplest case.

About