IIARD International Institute of Academic Research and Development

WORLD JOURNAL OF INNOVATION AND MODERN TECHNOLOGY (WJIMT )

E-ISSN 2504-4766
P-ISSN 2682-5910
VOL. 7 NO. 1 2023
DOI: https://doi.org/10.56201/wjimt.v7.no1.2023.pg67.86

---

Modelling Drilling Mud Viscosity Behaviour at Downhole Conditions Using Multi-Gene Genetic Programming

Godwin Samuel Edoho and Julius Udoh Akpabi

---

Abstract

This study developed a nature-inspired algorithm based on multigene genetic programming to predict downhole mud plastic viscosity for oil and water based muds, using data obtained from the field and from open literature. The initial mud plastic viscosity (IPV), downhole temperature (T) and downhole pressure (P) were used as the input parameters to the algorithm. To develop the model, 88 and 149 data points were used to develop downhole mud plastic viscosity models for oil based muds and water based muds respectively. To assess the performance of the models, four statistical error tools namely: the mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE) and determination coefficient (R2) were adopted. The results indicate that the model for the oil based mud had an R2 value of 0.9499 and an MSE of 0.2507, MAE of 3.12 and RMSE of 0.5. For the water based mud downhole viscosity model, R2 value of 0.8166 and an MSE of 0.1418, MAE of 2.25 and RMSE of 0.3766. In order to ascertain the parametric importance of the input variables used, the partial derivative sensitivity analysis was utilized. In this regard, the initial mud plastic viscosity had the highest influence for water based muds (70%) followed by the temperature (29.3%) while the pressure had the least effect (0.75%) on downhole mud plastic viscosity. For the oil based mud, down hole temperature had the highest influence (99.6%) followed by the initial mud plastic viscosity (0.3%) while downhole pressure had the least effect. In addition, the MGGP model was presented in an explicit form that makes it easy to be deployed in software applications, something rarely found in most machine learning studies. The study also assessed the computational speed of the developed models. This was necessary so as to know the efficiency of the model when deployed in software applications. With respect to execution speed, 18 and 16 floating point operations per second for OBM and WBMs w

keywords:

Multigene; Genetic Programming; Downhole; Mud Viscosity

---

References:

Abdullah, M. O., Yassin, A. A. M. and Kamis, A. (1991). Formulation of an Environmentally Safe Oil
Based Drilling Fluid. SPE Asia Pacific Conference, 4-7 November, Perth, Australia.

Abduoa, M., Dahabb, A., Abusedaa, H. and Elhossieny, M. (2016). Comparative study of using Water-
Based mud containing Multiwall Carbon Nanotubes versus Oil Based mud in High Pressure High
Temperature fields. Egyptian Journal of Petroleum, 25(4): 459- 464. doi:
10.1016/j.ejpe.2015.10.008.

Agwu, O. E., Okon, A. N. and Udoh, F. D. (2015). A Comparative Study of Diesel Oil and Soybean Oil as
Oil-Based Drilling Mud. Journal of Petroleum Engineering, Hindawi Publishing Corporation. 1–
10. Https://Doi.Org/10.1155/ 2015/828451.

DOWNLOAD PDF

---

Back