Artificial Intelligence as a Solution to Fluctuating Accuracy in Well-Logging

Characterization
Introduction
Well logging is a technique in which different tools are lowered into the wellbore and then
pulled back up. As the tools pass through different formations, it measures different rock and
formation properties which include electrical properties, acoustic, and radioactive properties.
The tool known as sonde transmits measurements using wireline cable, where the data is
recorded in the computers for experts to analyze.
Well logging tools are not accurate as they should be, and different issues may occur that can
impact the quality of log data. “These issues are of two types, firstly, issues related to the
tool. Secondly, there are issues related to the borehole environment” (Dr. Zaki Bassiouni,
2012).
In the Kingdom of Saudi Arabia, different geological formations can affect the accuracy of
well logs like Jabal Rayah, Ash Shutbah, and As Su’ayerah. To make the well logging data
more accurate and to eliminate fluctuating accuracy, technology can be considered, which
includes the internet of things, i.e., artificial intelligence or machine learning.
History of Well Logging in Saudi Arabia
The first commercial quantity discovery of oil was made at Dammam well no.7 in 1938. It
was the first time that logging and mapping were used in Saudi Arabia. Nestor Sander was
the first responsible person who was assigned to do well logging and mapping of the Abqaiq,
south-west of Dammam dome (Micheal Q. Morton, 2018).
Dielectric logging is a type of logging that was introduced in the late 1970s to measure the
water-filled porosity in the zones flushed and free from Archie components and water
salinity. It was introduced in Saudi Arabia in the 1980s (Denis P. Schimitt, Ahmed Al-Harbi,
Pablo Saldungaray, Ridvan Akkurt, and Tianhua Zhang, 2011). Dielectric logging eventually
disappeared because of hardware complications and low accuracy.
As technology is advancing Saudi Arabia’s state-own company, Saudi Aramco is adopting
modern well logging techniques to remove the fluctuating accuracy in the well logging data.
Saudi Arab has done successful multilateral well logging, which is an innovative approach
for multilateral wells (Ali Al-Momin, Abdul Wahab Azark, Anton Burov, and Murat Zeybek,
2011). Saudi Arabia has a history of adopting techniques that are technologically advanced
and help in getting accurate data.
Conventional Techniques
Conventional techniques mean the logs the traditional logs that are run to obtain the
formation properties. These conventional techniques don’t use the latest or modern
technology like machine learning to improve data accuracy. There will always be a chance of
obtaining inaccurate or fluctuating data. Primary conventional, well logging tools are
discussed below.
Spontaneous Potential Log (SP)
The spontaneous potential curve measures the naturally occurring electrical potential. The
tools display a curve known as the SP curve, which reflects the difference between a movable
electrode fixed in the wellbore and a fixed reference electrode on the surface, as shown in
figure no. 1. There are different uses of Spontaneous potential logs, which include an
indication of permeability, shale volume, water salinity, and porosity. It is one of the simplest
conventional logging tools that is run in the wellbore.

Figure 1: Working Principle of SP Log

Image Credit: https://openei.org/wiki/Spontaneous_Potential_Well_Log

Spontaneous potential logging or SP logs have limitations. It is only used in water-based

muds, and it won’t work in oil-based muds. SP logging tool is not reliable when there is a
small contrast between mud filtrate salinity and water salinity. Whenever there is a small
contrast, the SP logging tool will result in either small deflection or no deflection at all.
Formation Density Log
Formation density logging is another conventional well logging tool which is used to measure
the formation’s bulk density and length of the borehole. The density logging tool was
developed in the 1950s. It uses is a radioactive source and detector which is lowered into the
borehole, and the source releases gamma-rays into the formation. The gamma-rays
intermingle with the electron present in the formation, and the detector detects these scattered
electrons, which is the indication of formation density, as shown in figure no. 2.
 Figure 2: Formation Density Log Tool

Image Credit: docplayer.net

The formation density log is a conventional logging technique that has its limitations. The
correction in density log is only valid if the mud cake or standoff has lower than 0.75” and it
is consistent along with the pad. Moreover, the formation density tool will show extremely
low density in washed-out holes or rough holes higher than 17-inches.
Neutron Logging
Neutron is another primary conventional logging technique that is practiced in the oil and gas
industry. The purpose of neutron logging is to evaluate the formation of porosity. The neutron
logging tool uses the neutron source to record the hydrogen index in the reservoir, as shown
in the figure. 3. Hydrogen index is defined as ratio of the concentration of hydrogen atoms
per centimeter cube (cm3) of the material to that of pure H20 at 75oF.

Figure 3: Neutron Logging Tool

Image Credit: omicsonline.org

Like SP logging and density logging, neutron logging has limitations as well. In shale
formation, the hydrogen bonding is extremely low, so the neutron emitted from the source is
captured by the hydrogen, and loss of neutron occurs. If the neutron log is run in the gas
reservoir, the gas molecules are too far from one another, which tricks the tool in showing
lower porosity than the actual porosity.
Real-Life Applications of Well Logging
Well logging plays a vital role during drilling of an oil or gas well. The logging technique is
also used after the well is completed during its production phase. The well logging has lots of
application because logs tell us the formation properties without logging data engineers are
blinds because they don’t know what’s in the hole. Different logging tools help in identifying
the oil and gas reserves, and based on the data, well testing is conducted and well is
completed. The figure no. 4 below is the real-life example of a well log identifying the
potential formation with oil and gas reserves.
 Figure 4: Well Logs Showing Presence of Hydrocarbon in the Formation

Image Credit: Research Gate

Other applications of well logging are to tell us about the composition of rock like types of
minerals, elements, and common strata types. It tells us about the characteristic and integrity
of rock. Moreover, well logging tells about the type of liquid present in the borehole, shape,
and size of the borehole as well.
Benefits & Drawbacks
The conventional well logging techniques have their benefits and limitations. These tools use
conventional technology, which has its advantages and limitations as well.
Benefits
The benefits of well logging are listed below:
i. The conventional well logging techniques help in continuous measurements and
evaluation of the formation. There is no disturbance in between, and the data is
recorded continually.
ii. The well logging tools are easy to use and quick to work with because they are
available for a long time. It takes less time for a trained crew to rig up the logging
equipment and run the tools into the wellbore.
iii. These tools help in obtaining real-time data, which decreases the time of data
acquiring.
iv. As compared to seismic data, the resolution and quality of well logging data are much
better.
v. Finally, the conventional well logging is economical, and it helps companies save lots
of money as compared to the latest and innovative tools.
Drawbacks
There are certain drawback and limitations of well logging which as well are listed below:
i. Indirect measurement is of the biggest drawback of well logging. The tool may give
fluctuating measurements because of different factors like malfunctioning and wrong
calibration.
ii. Well logging tools have limited specifications like pressure and temperature
limitations. These tools can perform in certain temperatures and certain pressures. If
the pressure or temperature is above the specified limit, the tool will not work, or it
will give inaccurate measurements.
iii. Well logging tools are affected by the environment of the borehole. The condition of
the borehole can affect the efficiency of the tool, which is yet another drawback of a
well logging tool.
iv. These tools have varying resolutions. Sometimes the resolution is very high some
time the resolution is poor.
Alternative Techniques
Well logging traditional tools have drawbacks and limitations. As the technology is
progressing, different techniques are being developed to help eliminate the limitations and the
fluctuating accuracy in the well logging measurements. Artificial intelligence or machine
learning can be an alternative technique that can help in eliminating the fluctuating accuracy
in the well logging data.
The Art of Artificial Intelligence
Artificial intelligence and the internet of things are the most powerful technologies of this
era, and when you combine both AI and IoT, the result is AIoT, which means the artificial
intelligence of things. The internet of things works as the digital nervous system, while AI
works as the brain of the entire system (Bernard Marr, 2019).
The artificial intelligence of things starts with IoT. Internet of things means different devices
and tools connected to the internet and other devices collect and process data. Artificial
intelligence means the system itself can learn from and execute different tasks without any
human involvement.
The artificial intelligence of things is being used in the oil and gas industry to get accurate
results in production, drilling, and reservoir study. In well logging, there are different
problems that AI can solve, and one such problem is fluctuating well-logging results. It
doesn’t mean that the tools and devices will be eliminated entirely, but those tools will be
connected to the internet, and AI models will be applied to make the tools smarter. The AI
models can learn from previous well logging data and deliver the well logging data free from
fluctuation.
Real-Life Application of Artificial Intelligence
Artificial intelligence is being used in the oil and gas industry, and research is underway to
create AI models that can ease the entire process of well logging, data interpretation, and
correlation. Here is the real-life application of AI taken from Hugoton and Panoma gas fields
located in the United States of America.
Well logging data was taken from nine wells that were assigned lithofacies by an expert.
When the machine learning model is trained on the dataset taken from nine wells, the AI
model will be used to allot facies to the wells that don’t have any labels. The dataset includes
3232 observations, the well log data that was fed to the model includes gamma-ray,
resistivity, photoelectric effect, neutron-density porosity logs. The model is trained using a
machine learning tool that optimizes ML pipelines using programming. The table no. 1 of
how dataset looks like is given below.
Table 1: Representation of Dataset in ML

Table Credit: Medium.com

After data processing and data fed into the machine learning the model training with pipeline
optimization tool TPOT is shown in the table no. 2:
Table 2: Training the machine learning model

Table Credit: Medium.com

The model accomplishes above 70% accurate predictions. Finally, the model used to
calculate and equate predictions result in the actual formations (Ali Sharaf, 2018). The log
data of machine learning model is given in the figure no. 5.

Figure 5: Comparison of actual log data and machine learning well logs

Image credit: Medium.com

Advantages & Limitations
There are different advantages and limitations of artificial intelligence as well because it is
still in the developing phase. Different models are being trained to get precise and accurate
results of well logging.
Advantages
The advantages of AI are listed below:
i. The improved result is one of the biggest advantages of machine learning. It helps us
achieve accurate results and helps in eliminating fluctuations. The tools are connected
to the internet, and the AI, the brain of the tools, learns from the previous data and
helps in eliminating any discrepancies in the well logging data.
ii. It saves time and removes any possible chance of human error. As compared to
conventional techniques that have lots of chances of human error, AI trains itself,
learns from the data, and executes the task eliminating any human error in the data
fluctuation.
iii. There not any chores when it comes to AI. Everything is generated and computed by
the machine. While conventional techniques have lots of chores like an expert has to
monitor data continuously when a well logging tool is inside the wellbore. AI
eliminates the need for continuous monitoring and helps automates the entire process.
iv. The models like artificial neural network (ANN) and automated learning model
(ALM) can identify and remove any parameters that are not required in the results.
Limitations
Like any other system, AI has its limitations as well, which are listed below:
i. The input data can sometimes be out of boundaries of the AI training model. If the
data is outside the boundaries of the AI algorithm, it won’t be able to compute and
give accurate results.
ii. In order to achieve the best result, you have to train the model by inputting a large
number of data to get better results. Entering years of data can be time-consuming.
Conclusion
As the world is advancing, the traditional technologies are getting obsolete the same is the
case in the oil and gas industry. The well logging techniques are traditional and their
drawbacks because they can give fluctuating data, and the solution to the problem is adopting
the latest technologies like AI or machine learning.
In order to achieve better results, we must go with machine learning. Train the models and
use the existing models to get accurate results. The only way forward is to adopt the new
technologies, improve the quality of data, and extract the hydrocarbons.
Recommendations
Based on the applications and comparison of conventional logging tools and machine-based
learning, the company should work on the following recommendation:
i. The machine learning model should be trained with the help of programmers and well
logging experts to remove the fluctuations from the well logging data.
ii. The creation of a model and training process of machine learning can take time. It is
recommended that conventional tools should be used.
iii. It is recommended that the latest technologies related to well logging available in the
market should be procured to deal with variations in the logging data.
 References
Ali, Ashraf. (2018). Machine learning in Oil & Gas: Identifying Geological Formations based

on well logs. Medium. Retrieved from https://medium.com/@alisharaf/machine-

learning-in-oil-gas-identifying-geological-formations-based-on-well-logs-

e5073c98a83d

Ali Al-Momin, Abdul Wahab Azark, Anton Burov, & Murat Zeybek. (2011). First successful

multilateral well logging in Saudi Aramco: Innovative approach toward logging and

open hole multilateral oil producer. SLB. Retrieved from

https://www.slb.com/-/media/files/premium-content/technical-paper/140/149079.ashx

Bernard Marr, (2019). What is the artificial intelligence of things? When AI meets IoT.

Forbes. Retrieved from https://www.forbes.com/sites/bernardmarr/2019/12/20/what-

is-the-artificial-intelligence-of-things-when-ai-meets-iot/#56b93648b1fd

Denis P. Schimitt, Ahmed Al-Harbi, Pablo Saldungaray, Ridvan Akkurt, & Tianhua Zhang.

(2011). One Petro. Retrieved from https://www.onepetro.org/conference-paper/SPE-

149131-MS

Dr., Zaki. (2012). Well logs quality control issues. Oilproduction.net. Retrieved from

http://oilproduction.net/files/Well%20Logs%20Quality%20Control%20Issues.pdf

Michael, Q. Morton. (2018). The first discoveries of oil in Saudi Arabia. Geo Ex Pro.

Retrieved from https://www.geoexpro.com/articles/2018/04/the-first-discoveries-of-

oil-in-saudi-arabia