Professional Documents
Culture Documents
This paper was prepared for presentation at the Abu Dhabi International Petroleum Exhibition & Conference to be held in Abu Dhabi, UAE, 9 – 12 November 2020.
Due to COVID-19 the physical event was changed to a virtual event. The official proceedings were published online on 9 November 2020.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
Suitable drilling fluid with suitable properties is a critical factor in successful drilling and completion
operations. Drilling fluids consultant will need information such as formation lithology, wellbore geometry,
temperature, pressure to be able to formulate fluids for different sections. Drilling fluids consultants' job is to
aid drilling engineers and scientists to formulate effective drilling fluids systems for the entire well sections.
The paper describes a drilling fluid advisory system with demonstrating examples to the end user. The
advisory system includes a Bayesian decision network (BDN) model that utilize inputs such as temperature
and pressure and provide drilling fluids formulations based on Bayesian probability determinations.
A number of drilling fluid specialists/experts feedback were gathered to develop the drilling fluid
advisory system. The drilling fluid specialists/experts feedback were used to design Bayesian Network
Model that would allow the end-user to take an elementary data set to obtain optimum recommendations
in the area of drilling fluids.
Introduction
Shortage of expertise motivates researchers to gather and develop expert or advisory systems for knowledge
sharing and to ensure that the industry does not run out of experts of consultants. Expert systems are defined
as knowledge processing to allow computers to produce certain tasks similar or better than human specialists
do, Hayes-Roth (1987).
Another definition of Expert system can be as "An interactive computer-based decision tool that simulates
the thought process of a human expert to solve complex problems in a specific domain." "We need experts
system because of limitations in expertise, working memory, insufficient maintenance of significant data
and biased opinions", (Pandey and Osisanya 2001).
We need previous experience to be able to develop advisory system with a good degree of confidence. The
most important factor for drilling operations success according to the authors of this paper are experience
and knowledge transfer.
Al-Yami et al. (2012) have explain how different companies can come up with different methods or
manuals to setup guidelines that cannot be used by other operators. Experts' opinions such as Drilling Ligand
Mr. Rhem Bill and Dr. Jerome Schubert through their publications or consultation were used to develop
2 SPE-203334-MS
advisory systems for underbalanced drilling advisory systems and well control advisory system. They were
able to develop a drilling expert system for different area such as cementing, completion, well control, and
underbalanced drilling utilizing Bayesian network.
Al-Yami et al. (2012) have proposed a set of guidelines in different drilling specialties utilizing Artificial
Bayesian Intelligence where optimum practices collected from literature review and experts' opinions to
simulate likely scenarios of its use that will honor efficient practices when dictated by varying certain
parameters.
In Bayesian Networks, we use nodes to represent variables either discrete or continuous to show the
• To assess the risk from nuclear waste disposal, Lee et al. (2005).
• To model flow to select the model with greatest uncertainty at the boundaries, Abbaspour et al.
(2000).
• For Hydrodynamic behavior characterization, Ferraresi et al. (1996) and most probable areas of
salinity sources distribution in the Gaza aquifer, Ghabayen et al. (2006).
• For pipeline leak detection, Carpenter et al. (2003).
The work done by Alouhali et al. 2018 and other focused on discriminative modeling where the objective
is to know probability of an event given an input or parameter. Bayesian network allow us to do joint
distribution utilizing graphical models, Al-Yami et al. 2015a-g
Alouhali et al. 2018 have shown that Bayesian network model perform less accurately compared to
Decision Tree and KNN but still produce acceptable correlation. However, we believe that the only machine-
learning tool that allow us to deal with joint uncertainties is Bayesian network. It is way harder to learn joint
probability distribution of data set than learning about conditional probability. All machines learning model
including Bayesian Network allow observational inferences. However, only Bayesian network allow causal
SPE-203334-MS 3
interventions. Dashevski et al. (1999) and Stoner (2003) have shown that non-Bayesian network approaches
have the following limitations:
• Adding new rules or changes require programming again because knowledge database and
inference algorithms are inseparable.
• The previous approaches are limited in their ability to make decisions under uncertainty.
Bayesian network is useful to deal with missing data sets. The reason of that the modeling here is done
Al-Yami et al. (2016) has shown the difference between frequentist statistics and Bayesian statistics.
Al-Yami et al., 2020, demonstrated detailed proof of concept of Bayesian network in developing a small
drilling fluid advisory.
To design a Bayesian we need to do the following
The objective of this research is to propose models in the area of drilling fluids utilizing Bayesian network
to serve as training tools or a guide to aid drilling engineers and scientists.
The second uncertainty node allow the user to select potential hole problems, Fig.3. The third node is
related to the second, which is the formation name. Sometimes, the only available information for the drilling
engineer is the formation name. The model will inform the user about the expected potential hole problem
based on the formation name, Fig.3.
SPE-203334-MS 5
The user can select any potential hole problem he suspect to have for example formation damage, Fig.4.
The user can select a temperature range for example 220 to 330 °F, Fig. 5. Recommended drilling fluid
formulations are shown in Fig. 6.
If the selected conditions ended up with no options i.e. all zero that means that there is no recommendation
for the given inputs, Figs.7-9. In other words, we missed this possible scenario. This can be added to the
advisory system easily.
Fig. 10 shows the initial probabilities assignment of potential hole problem. If the user does not know
what to expect and he only knows the formation to be drilled he can select that from the options available
in the formation node. For example, Fig. 11 shows the potential hole problem based on selected formation
name.
A Software tools for user interface engineers was developed for the work in this paper. Which can be
used to show the formulation at any selected conditions, Fig. 12. The system can also tell the user that the
outcome does not exist because the developer did not include them as output, Fig. 13. The system also can
tell the user when he put non-compatible inputs, Fig.14.
Figure 14—The system also can tell the user when he put non-compatible inputs
SPE-203334-MS 11
We can change the states of probabilities if we have different opinions or feedback from drilling fluids
consultants. We can also update the model if we find out that we need to expand it to include other
uncertainties or inputs.
Conclusions
• The paper describes a drilling fluid advisory system with demonstrating examples to the end user.
The advisory system includes a Bayesian decision network (BDN) model that utilize inputs such
Nomenclature
API American petroleum institute
Bc Bearden units of consistency
bpm Barrels per minute
Bwoc By weight of cement
CCA Casing to casing annulus
DV Differential valve
0F Degrees fahrenheit
ft Feet
Gal Gallons
gal/sack Gallons per sack of cement
hrs. Hours
in Inches
lbf Pounds of force
lbm/ft3 Pounds per cubic foot
MD Measured depth
min Minutes
pcf Pound-force per cubic foot
psi Pounds per square inch
SBS Shear bond strength
SCP Sustained casing pressure
TVD True vertical depth
UCA Ultrasonic cement analyzer
WOC Waiting on cement
TOC Top of cement
in. 2.54* E 02 = m
ft × 3.048* E 01 = m
gal × 3.785 412E 03 = m3
lbm × 4.535 924E 01 = kg
References
Alouhali, K., Matta, V., Huggenberger, P., and Johnso, C. A. 2000. A Contaminated Site Investigation: Comparison of
Information Gained from Geophysical Measurements and Hydro-geological Modeling. Contaminant Hydrology 40:
365–380.
Alouhali, R., Aljubran, M., Gharbi, S., Al-Yami, A. S.: "Drilling Through Data: Automated Kick Detection Using Data
Mining," paper SPE-193687-MS presented at the 2018 International Heavy Oil Conference and Exhibition held in
Kuwait City, Kuwait, December 10-12.
Al-Yami, A. S., Schubert, J., Medina, C., and Ok-Youn, Y. 2010b. Drilling Expert System for the Optimal
Design and Execution of Successful Cementing Practices. Paper IADC/SPE 135183 presented at the Asia
Pacific Drilling Technology Conference and Exhibition, Ho Chi Minh City, Vietnam, 1-3 November. http://
dx.doi.org/10.2118/135183-MS..
Al-Yami, A. S. and Schubert, J. 2012. Drilling Fluids Consultation System: Development and Field Applications. Paper
SPE 152098 presented at the SPE Western North American Regional Meeting, Bakersfield, California, USA, 19-23
March. http://dx.doi.org/10.2118/152098-MS.
Al-Yami et al.: "Systems and methods for drilling fluids expert systems using Bayesian decision networks," US Patent
No. 9,202,169, December 1, 2015.
Al-Yami et al, "Systems and methods for an expert system for well control using Bayesian intelligence," US Patent No.
9,202,175, December 1,2015.
Al-Yami A. S, Al-Shaarawi, A., Al-Bahrani, H., Wagle, V. B., Al-Gharbi, S., Al-Khudiri, M. B.: "Using Bayesian Network
to Develop Drilling Expert Systems," paper SPE-184168-MS presented at the 2016 SPE Heavy Oil Conference and
Exhibition held in Kuwait City, Kuwait, December 6-8.
Al-Yami et al.: "Expert systems for well completion using Bayesian decision networks with a multilateral junction design,
and a junction classification decision node," US Patent No. 9,366,129, June 14, 2016.
Al-Yami et al.: "Expert system for well completion using Bayesian probabilities and a consequences node dependent on the
zonal isolation types, reliability level, cost level, productivity level, the completion type, and the junction classification
decision nodes," US Patent No. 9,366,130, June 14, 2016.
Al-Yami et al.: "Expert systems for well completion using Bayesian probabilities, and a packers consequence node
dependent on wellbore fluids, hydrocarbon types, completion fluids, packers decision, and treatment fluids nodes,"
US Patent No. 9,371,726, June 21, 2016.
Al-Yami et al.: "Expert systems for well completion using Bayesian probabilities, open hole gravel pack types, gravel
pack design details, open hole gravel packs, and completion types," US Patent No. 9,376,905, June 28, 2016.
Al-Yami et al.: "Expert systems for well completion using Bayesian decision networks including a fluids damage and
temperature effects uncertainty node, perforations considerations uncertainty node, a perforation analysis uncertainty
node, and a perforation type decision node," US Patent No. 9,382,791, July 5, 2016.
Al-Yami, A. S., Wagle, V., Al-Rubaii, M., Abubacker, Z.: "Development of Drilling Fluid Advisory System Utilizing
Bayesian Network," paper SPE-203334 presented at the 2020 SPE/IADC Middle East Drilling Technology Conference
and Exhibition.
Carpenter, P., Nicholas, E., and Henrie, M. 2006. Bayesian Belief Networks for Pipeline Leak Detection. Paper PSIG 0604
presented at the PSIG Annual Meeting, Williamsburg, Virginia, 11-13 October.http://dx.doi.org/10.2118/0604-MS.
SPE-203334-MS 13
Dashevski, D., Dubinsky, V., and Macpherson, J. D. 1999. Application of Neural Networks for Predictive Control in
Drilling Dynamics. Paper SPE 56442 presented at the SPE Annual Technical Conference and Exhibition, Houston,
USA, 3-6 October. http://dx.doi.org/10.2118/56442-MS.
Ferraresi, M., Todini, E., and Vignoli, R. 1996. A Solution to the Inverse Problem in Groundwater Hydrology Based on
Kalman Filtering. Journal of Hydrology 175: 567–581. http://dx.doi.org/10.1016/S0022-1694(96)80025-4.
Ghabayen, S., McKee, M., and Kemblowski, M. 2006. Ionic and Isotopic Ratios for Identification of Salinity
Sources and Missing Data in the Gaza Aquifer. Journal of Hydrology 318: 360–373. http://dx.doi.org/10.1016/
j.jhydrol.2005.06.041.
Giese, M. and Bratvold, R. 2011. Probabilistic Modeling for Decision Support in Integrated Operations. SPE Economics