Application of Various Machine Learning Techniques in Predicting Total Organic Carbon from Well Logs

Table 1

Summary of different empirical correlations.

Authors	Model	Remarks
Schmoker [20]	where densities are in g/cm^3	Predicted TOC in volume percentage and used data from Devonian shale. ρ is the formation bulk density, and and ρB represents the organic matter free rock density.
Schmoker [21]		A revised model used Bakken shale’s data and predicted TOC in weight percentage. R is ratio between the organic matter and organic carbon. ρo is the density of the organic matter, and ρmi is the average bulk density.
Passy et al. [22]	where resistivity is in ohm·m and transit time is in μs/ft	Widespread model and known as the ΔlogR model. First, the logs separation (ΔlogR) is calculated from acoustic transit time (Δt) and formation resistivity (FR). Then, TOC is estimated from ΔlogR and the level of maturity (LOM).
Wang et al. [23]	where α, β, δ, and η are constants	Revised ΔlogR models to estimate the TOC developed for Devonian shale using ΔlogR, gammaray (GR), and the indicator of maturity (Tmax).
Zhao et al. [24]	where a, b, and c are constants	Revised Wang’s model. Do not depend on the level of maturity, vitrinite reflectance (Ro), or Tmax.