Presented at The Students Technical Symposium & Exhibition Held at Obafemi Awolowo University, Ile Ife On The 22 of February, 2019

AI POWERED PIPELINE
SURVEILLANCE SYSTEM
Presented at the Students Technical Symposium & Exhibition
held at Obafemi Awolowo University, Ile Ife on the 22nd of
February, 2019.
Presented by
Osisiogu, Ukachi Oluwaseun (B.Eng)
Williams, Yerima Yohanna (B.Eng)
AUST SPE Chapter
Objectives
• Problem Description
• The Idea
• The Solution
• Recommendation & Future Work
The Problem
• Former Minister of Finance, Dr. Ngozi Okonjo-Iweala once lamented that Nigeria
was losing about 400,000 barrels of crude oil per day because of
• Crude Theft,
• Illegal bunkering
• Vandalism of infrastructure and,
• Halt in Production
The Problem
The Problem
• Former Minister of Petroleum Resources – Mrs. Diezani Alison-Madueke

once said
• “Nigeria might not recover from the negative

impact of crude theft and pipeline vandalism in
the next 20 years or more”
The Problem
• “Oil multinationals - Shell and Agip were

losing 190,000 barrels of crude daily from oil
fields in Southern Bayelsa State alone”
- Vanguard Newspaper
The Problem
• The Nigeria Natural Resource Charter when presenting a report in 2018 stated that:
• “In 2016, both government and oil companies

indicated that a combined total of up to N3.8 trillion
was recorded as the amount lost from crude oil theft,
sabotage and pipeline vandalism.”
Problem with the Oil theft Story…
Problems Possible Reasons

Lots of Controversies and Lack of Fool Proof measures to
uncertainties in actual loss figure monitor pipelines
Absence of real-time technology to
Lack of insight into the amount of determine actual production
loss by caused various events
Low enthusiasm towards the
policing of pipelines
Our Proposed Solution
IOT + AI = Solution
The Idea
To make pipelines smarter by equipping

them with:
The ability to feel (Vibration and Temperature)
The ability to be self-aware (Pressure and Mass

flow rate)
The Idea
Feedback with little or no human interference until

case is severe
IOT – Internet of Things
IOT – Internet of Things
Pressure Sensor
Specifications
Working Voltage: 5V DC
Output Voltage: 0.5 – 4.5V DC
Working Pressure Range: 0 – 1.2 MPa
Maximum Pressure: 2.4MPa
Pressure Sensor
Flow Sensor
Working Voltage 4.5V 5V – 24V

Maximum Operating Current 15mA
Working Pressure below 1.75Mpa
Flow Sensor
Flow Sensor
&
Pressure Sensor
connection
Vibration Sensor
Piezo Vibration Sensor is suitable for

measurements of flexibility, vibration, impact
and touch. When the sensor moves back and
forth, a certain voltage will be created by the
voltage comparator inside of it.
Returned values: from 0 (no strain) to 1023

(maximal strain).
Pressure,
Flow Rate
and Vibration
Sensor
Pressure,
Flow Rate &
Vibration
Sensor
connection
Temperature Sensor
• A very simple sensor to

wire - measuring
temperature in a huge
range, with high accuracy.
• Returned values: –55°C to
+125°C (–67°F to +257°F) -
with an accuracy of (+/-
0.5C).
Temperature Sensor
Temperature Sensor
Artificial Intelligence
Fuzzy Classification Algorithm

Uses degree of membership to perform classification tasks
VIBRATION
PRESSURE
Low Medium High

Low Vandalism No Vandalism Vandalism
Medium No vandalism No Vandalism Vandalism
High No Vandalism No Vandalism Vandalism
Table 1: Showing Classification based on sensor readings

Membership Function Plot for Pressure
Membership Function Plot for Frequency
Pressure (MPa)
Low Medium High
Low
Leakage
Medium
Frequency (Hz)
High
Explosion Explosion Explosion

DRONE SURVEILLANCE
• Will be used provide more

insight into the situation
for proper action
• Also uses Machine

Learning – Mask Region
Convolution Neural
Network to detect and
provide more information
on leakages and vandalism
in pipelines.
How it works
AI algorithms
runs on the
Cloud
Sensor
Trigger Monitoring
Data

case is severe
Lots of pipelines = More Data
More Data! = Big Data!
AI algorithms
runs on the
Sensor Data Cloud
=
Big Data! Trigger Monitoring

case is severe
Estimated Budget (Summary)
Purpose Phase Amount (N)
Smart Pipeline (excluding Phase 1 71, 791.59

Cloud Services)
Drone Phase 2 213, 143.40
Total 284, 934.99

Expected Impact
• With more knowledge discovery more insight can be gained
We can answer certain questions
Are the pipelines

operating under
normal conditions?
Now we know better!
• Or has a leakage
happened?
• Can we accurately
determine the events
that caused the
leakage?
More knowledge, more insight
• Was it
leakage
caused by
vandalism?
Predictive maintenance
• Did an explosion occur?

What events occurred
before such explosions
and how can we prevent
it from happening
again?
Reduced oil loss due to theft and leakage
2018 – over 400, 000 barrels daily
2028 – less than 1,000 barrels daily

Less Crude oil lost? Then what?
Increased Revenue
With reduced crude oil More energy for everyone
loss
More Advantages
• Faster Reaction to Damages and Spillages when

they occur
• Pattern recognition can be achieved to improve
predictive maintenance
• Safer Environment
• … and much more
Future Work
• Estimation of volume loss during vandalism or leakage using other machine learning algorithms
(Linear Regression/Support Vector Regression).
𝐹 = Force Applied
𝑚 = mass of fluid
𝑎 = acceleration
𝑡 = time
𝑃 = pressure
𝐴 = cross sectional area
𝑣 = velocity of the fluid
• 𝜌 = density of the fluid
𝑉 = Volume
𝐹 𝑉 𝑚𝑣
We have that 𝑃 = 𝑏𝑢𝑡 𝐹 = 𝑚𝑎; 𝑎 = ; 𝐹 =
𝐴 𝑡 𝑡
𝑚𝑣 1 𝑚𝑣
Hence 𝑃 = × =
𝑡 𝐴 𝐴𝑡
𝜌𝑉𝑎 𝜌𝑉𝛿𝑣
But 𝑚 = 𝜌𝑉 therefore 𝑃 = =
𝐴 𝐴𝑡
Future Work
• To test other machine algorithms – support vector

machine, probabilistic graphical models, neural networks in
order to choose the most effective, then we make further
experiments in order to perform optimisations on them.
• Improve on algorithms and collaborate with system

engineers to increase the number of classifications and
focus research in application specific areas.
Any Questions?
For further questions you

can contact us
[email protected]
[email protected]

Presented at The Students Technical Symposium & Exhibition Held at Obafemi Awolowo University, Ile Ife On The 22 of February, 2019

Uploaded by

Copyright:

Available Formats

Presented at The Students Technical Symposium & Exhibition Held at Obafemi Awolowo University, Ile Ife On The 22 of February, 2019

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Presented at The Students Technical Symposium & Exhibition Held at Obafemi Awolowo University, Ile Ife On The 22 of February, 2019

Uploaded by

Copyright:

Available Formats

AI POWERED PIPELINE

• Former Minister of Petroleum Resources – Mrs. Diezani Alison-Madueke

• “Nigeria might not recover from the negative

• “Oil multinationals - Shell and Agip were

• “In 2016, both government and oil companies

Problems Possible Reasons

To make pipelines smarter by equipping

The ability to feel (Vibration and Temperature)

The ability to be self-aware (Pressure and Mass

Feedback with little or no human interference until

Working Voltage 4.5V 5V – 24V

Piezo Vibration Sensor is suitable for

Returned values: from 0 (no strain) to 1023

• A very simple sensor to

Fuzzy Classification Algorithm

Low Medium High

Table 1: Showing Classification based on sensor readings

Low Medium High

Explosion Explosion Explosion

• Will be used provide more

• Also uses Machine

Feedback with little or no human interference until

Feedback with little or no human interference until

Purpose Phase Amount (N)

Smart Pipeline (excluding Phase 1 71, 791.59

Drone Phase 2 213, 143.40

Total 284, 934.99

Are the pipelines

• Did an explosion occur?

2018 – over 400, 000 barrels daily

2028 – less than 1,000 barrels daily

• Faster Reaction to Damages and Spillages when

• To test other machine algorithms – support vector

• Improve on algorithms and collaborate with system

For further questions you

You might also like