The document provides an overview of using data and AI technologies to optimize operations in the oil and gas industry:
- Machine learning and predictive analytics can be used to optimize drilling configurations to minimize costs per foot drilled and maximize rate of penetration based on lithology, bit type, drilling parameters, and other factors. This was demonstrated on a real dataset from the Permian Basin.
- Other applications discussed include predictive maintenance of equipment using sensor data, forecasting tank levels to optimize oil pickup scheduling, and identifying underproducing wells by analyzing production and service histories.
- Microsoft Azure provides the cloud infrastructure and services needed to build these types of AI solutions for oil and gas customers, including data storage, processing, machine learning
Barriers and issues with the creation of calibrated modelsDaniel Coakley
Presented at IES Seminar - 'Operational Energy Management of the Built Environment' at Guinness Storehouse on 18th November 2015.
Whole building energy simulation (BES) models play a significant role in the design and optimisation of buildings. Simulation models may be used to compare the cost-effectiveness of energy-conservation measures (ECMs) in the design stage as well as assessing various performance optimisation measures during the operational stage. However, due to the complexity of the built environment and prevalence of large numbers of independent interacting variables, it is difficult to achieve an accurate representation of real-world building operation. Therefore, by reconciling model outputs with measured data, we can achieve more accurate and reliable results. This reconciliation of model outputs with measured data is known as calibration.
See full paper: https://www.researchgate.net/publication/262678655_A_review_of_methods_to_match_building_energy_simulation_models_to_measured_data
Time Series with Driverless AI - Marios Michailidis and Mathias Müller - H2O ...Sri Ambati
This talk was recorded in London on October 30, 2018 and can be viewed here: https://youtu.be/EGVY7-Spv8E
Time series is a unique field in predictive modelling where standard feature engineering techniques and models are employed to get the most accurate results. In this session we will examine some of the most important features of Driverless AI’s newest recipe regarding Time Series. It will cover validation strategies, feature engineering, feature selection and modelling. The capabilities will be showcased through several cases.
Bio: Marios Michailidis is now a Competitive Data Scientist at H2O.ai He holds a Bsc in accounting Finance from the University of Macedonia in Greece and an Msc in Risk Management from the University of Southampton. He has also nearly finished his PhD in machine learning at University College London (UCL) with a focus on ensemble modelling. He has worked in both marketing and credit sectors in the UK Market and has led many analytics’ projects with various themes including: Acquisition, Retention, Recommenders, Uplift, fraud detection, portfolio optimization and more.
He is the creator of KazAnova, a freeware GUI for credit scoring and data mining 100% made in Java as well as is the creator of StackNet Meta-Modelling Framework. In his spare time he loves competing on data science challenges and was ranked 1st out of 500,000 members in the popular Kaggle.com data competition platform. Here is a blog about Marios being ranked at the top in Kaggle and sharing his knowledge with tricks and ideas.
Finally, Marios’ likendin profile can be found here, with more information about what he is working on now or past projects.
https://www.linkedin.com/in/mariosmichailidis/
Bio: A Kaggle Grandmaster and a Data Scientist at H2O.ai, Mathias Müller holds an AI and ML focused diploma (eq. M.Sc.) in computer science from Humboldt University in Berlin. During his studies, he keenly worked on computer vision in the context of bio-inspired visual navigation of autonomous flying quadrocopters. Prior to H2O.ai, he as a machine learning engineer for FSD Fahrzeugsystemdaten GmbH in the automotive sector. His stint with Kaggle was a chance encounter as he stumbled upon the data competition platform while looking for a more ML-focused platform as compared to TopCoder. This is where he entered his first predictive modeling competition and climbed up the ladder to be a Grandmaster. He is an active contributor to XGBoost and is working on Driverless AI with H2O.ai.
Linkedin: https://www.linkedin.com/in/muellermat/
Barriers and issues with the creation of calibrated modelsDaniel Coakley
Presented at IES Seminar - 'Operational Energy Management of the Built Environment' at Guinness Storehouse on 18th November 2015.
Whole building energy simulation (BES) models play a significant role in the design and optimisation of buildings. Simulation models may be used to compare the cost-effectiveness of energy-conservation measures (ECMs) in the design stage as well as assessing various performance optimisation measures during the operational stage. However, due to the complexity of the built environment and prevalence of large numbers of independent interacting variables, it is difficult to achieve an accurate representation of real-world building operation. Therefore, by reconciling model outputs with measured data, we can achieve more accurate and reliable results. This reconciliation of model outputs with measured data is known as calibration.
See full paper: https://www.researchgate.net/publication/262678655_A_review_of_methods_to_match_building_energy_simulation_models_to_measured_data
Time Series with Driverless AI - Marios Michailidis and Mathias Müller - H2O ...Sri Ambati
This talk was recorded in London on October 30, 2018 and can be viewed here: https://youtu.be/EGVY7-Spv8E
Time series is a unique field in predictive modelling where standard feature engineering techniques and models are employed to get the most accurate results. In this session we will examine some of the most important features of Driverless AI’s newest recipe regarding Time Series. It will cover validation strategies, feature engineering, feature selection and modelling. The capabilities will be showcased through several cases.
Bio: Marios Michailidis is now a Competitive Data Scientist at H2O.ai He holds a Bsc in accounting Finance from the University of Macedonia in Greece and an Msc in Risk Management from the University of Southampton. He has also nearly finished his PhD in machine learning at University College London (UCL) with a focus on ensemble modelling. He has worked in both marketing and credit sectors in the UK Market and has led many analytics’ projects with various themes including: Acquisition, Retention, Recommenders, Uplift, fraud detection, portfolio optimization and more.
He is the creator of KazAnova, a freeware GUI for credit scoring and data mining 100% made in Java as well as is the creator of StackNet Meta-Modelling Framework. In his spare time he loves competing on data science challenges and was ranked 1st out of 500,000 members in the popular Kaggle.com data competition platform. Here is a blog about Marios being ranked at the top in Kaggle and sharing his knowledge with tricks and ideas.
Finally, Marios’ likendin profile can be found here, with more information about what he is working on now or past projects.
https://www.linkedin.com/in/mariosmichailidis/
Bio: A Kaggle Grandmaster and a Data Scientist at H2O.ai, Mathias Müller holds an AI and ML focused diploma (eq. M.Sc.) in computer science from Humboldt University in Berlin. During his studies, he keenly worked on computer vision in the context of bio-inspired visual navigation of autonomous flying quadrocopters. Prior to H2O.ai, he as a machine learning engineer for FSD Fahrzeugsystemdaten GmbH in the automotive sector. His stint with Kaggle was a chance encounter as he stumbled upon the data competition platform while looking for a more ML-focused platform as compared to TopCoder. This is where he entered his first predictive modeling competition and climbed up the ladder to be a Grandmaster. He is an active contributor to XGBoost and is working on Driverless AI with H2O.ai.
Linkedin: https://www.linkedin.com/in/muellermat/
Intelligent Production: Deploying IoT and cloud-based machine learning to opt...Amazon Web Services
Alex Robart, CEO of Ambyint, presents their AI-driven production optimization platform for the Oil and Gas Industry.
Their IoT-based innovative hardware and software solution, delivers a revolutionary approach to monitoring Oil and Gas production operations, by updating traditional SCADA-based telemetry, cloud-enabling them, and bringing in Artificial Intelligence capabilities. Presented at the AWS Oil and Gas Industry Day in Calgary, 2017.
Energy Savings & Green Considerations in Motion Control WebinarDesign World
Everybody is talking more and more about “green” engineering these days. While its beginning to become clearer what this means in the fields of alternative energy and consumer energy usage, it’s still not at all clear what this means for the world of motion control. Some of the leading motion control companies discuss and explore your questions about “green” design principles and how they apply to applications in the motion control world.
By watching this special 1-hour free webinar you will gain a better understanding of the key factors involved in “green” motion control and their relation to the projects you’re working on now and in the future.
Electric motors and the systems they drive consume electrical energy more than twice as much as lighting, while lighting is typically considered a major low-lying fruit for energy reduction in a factory. It is estimated that motors account for between 43% and 46% of all global electricity consumption particularly from the manufacturing sector. The presentation will cover topics like the effort to ensure the motor efficiency in manufacturing sector; the application of variable speed drives and other factors which affect the motor efficiency such as power quality; and good maintenance regimes for energy reduction
SAS Global Coal-Fired Power Diagnostic Testing and Combustion TuningJustin Bennett
Due to recent strict EPA regulations, more stringent burdens will continue to fall upon our industry. Coupled with increasing competition, fossil fueled power plants are struggling to comply with government regulations and
competing in a turbulent market. SAS Global Power is the only firm that has the experience to accurately assess
your current operating conditions and provide the technology for you to effectively and efficiently produce power without exceeding emissions standards.
The SAS Global Performance Testing and Combustion Tuning Group specializes in the reliable examination of your fuel flows to the boiler, backpass emission mapping, visual flame conditions inside boiler, fly ash and coal analysis. Utilizing the collected data, a comprehensive report detailing current operating assessment will be provided. The report will include recommendations designed to improve combustion stoichiometry, while enhancing auxiliary efficiencies and reducing emissions.
The scope of the test report will depend upon your predetermined goals and system imbalances, which will be
determined from your own custom test program.
Our service is unique to the specific requirements of each plant, price quotes are prepared on a location-by-location
basis. Please contact us for a custom tailored proposal that meets all of your specific needs.
The demand of power is increasing exponentially results in installation of new stations whereas the sources of water are depreciating acutely. In future there may be a situation in which water sources may not cope up with this requirement.
Also the serious concerns of the regulatory authorities regarding usage of natural resources, definitely the norms will be further be tightened, which will curtail the freedom of usage of water in power plant.
In present scenario land acquisition is one of the toughest hurdles in plant installations which can be averted by locating stations in water scarce regions, by employing air cooled system which eliminates dependencies on water for CW.
Although dry cooling systems are costly technologies on techno-economic considerations, but foreseeing the future it is the need of hour to employ dry cooling system which offers possible solution for power plant installation eliminating the above mentioned challenges.
KK Wind Solutions presentation on Control System RetrofitRené Balle
KK Windsolutions presenting a strong retrofit case for Bonus, Siemens, Vestas and other turbines - showing significantly improved performance and short payback time.
This presentation was originally shown at the AWEA Wind Project O&M and Safety Seminar 2015 in San Diego.
For more information visit www.kkwindsolutions.com, nichr@kkwindsolutions.com or Phone: +4597221033
Developing a new generation of energy efficiency products for reciprocating e...Bowman Power
Learn how a new energy efficiency product gets made, from opportunity to concept, design, validation and production, with this free presentation from the 73rd Indonesia National Electricity Day & POWER-GEN Asia. #PGASIA
There is plenty of room for improvement in operations and maintenance activities. James Parle will tell how a team from Muir Data systems developed a system that lets wind tech report more thoroughly on their activities with easy-to-use portable devices.
And Grant Leaverton will report on how unmanned aerial vehicles can provides high resolution images of wind turbine inspections, especially blades, without rappelling from the nacelle.
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Subhajit Sahu
Abstract — Levelwise PageRank is an alternative method of PageRank computation which decomposes the input graph into a directed acyclic block-graph of strongly connected components, and processes them in topological order, one level at a time. This enables calculation for ranks in a distributed fashion without per-iteration communication, unlike the standard method where all vertices are processed in each iteration. It however comes with a precondition of the absence of dead ends in the input graph. Here, the native non-distributed performance of Levelwise PageRank was compared against Monolithic PageRank on a CPU as well as a GPU. To ensure a fair comparison, Monolithic PageRank was also performed on a graph where vertices were split by components. Results indicate that Levelwise PageRank is about as fast as Monolithic PageRank on the CPU, but quite a bit slower on the GPU. Slowdown on the GPU is likely caused by a large submission of small workloads, and expected to be non-issue when the computation is performed on massive graphs.
Intelligent Production: Deploying IoT and cloud-based machine learning to opt...Amazon Web Services
Alex Robart, CEO of Ambyint, presents their AI-driven production optimization platform for the Oil and Gas Industry.
Their IoT-based innovative hardware and software solution, delivers a revolutionary approach to monitoring Oil and Gas production operations, by updating traditional SCADA-based telemetry, cloud-enabling them, and bringing in Artificial Intelligence capabilities. Presented at the AWS Oil and Gas Industry Day in Calgary, 2017.
Energy Savings & Green Considerations in Motion Control WebinarDesign World
Everybody is talking more and more about “green” engineering these days. While its beginning to become clearer what this means in the fields of alternative energy and consumer energy usage, it’s still not at all clear what this means for the world of motion control. Some of the leading motion control companies discuss and explore your questions about “green” design principles and how they apply to applications in the motion control world.
By watching this special 1-hour free webinar you will gain a better understanding of the key factors involved in “green” motion control and their relation to the projects you’re working on now and in the future.
Electric motors and the systems they drive consume electrical energy more than twice as much as lighting, while lighting is typically considered a major low-lying fruit for energy reduction in a factory. It is estimated that motors account for between 43% and 46% of all global electricity consumption particularly from the manufacturing sector. The presentation will cover topics like the effort to ensure the motor efficiency in manufacturing sector; the application of variable speed drives and other factors which affect the motor efficiency such as power quality; and good maintenance regimes for energy reduction
SAS Global Coal-Fired Power Diagnostic Testing and Combustion TuningJustin Bennett
Due to recent strict EPA regulations, more stringent burdens will continue to fall upon our industry. Coupled with increasing competition, fossil fueled power plants are struggling to comply with government regulations and
competing in a turbulent market. SAS Global Power is the only firm that has the experience to accurately assess
your current operating conditions and provide the technology for you to effectively and efficiently produce power without exceeding emissions standards.
The SAS Global Performance Testing and Combustion Tuning Group specializes in the reliable examination of your fuel flows to the boiler, backpass emission mapping, visual flame conditions inside boiler, fly ash and coal analysis. Utilizing the collected data, a comprehensive report detailing current operating assessment will be provided. The report will include recommendations designed to improve combustion stoichiometry, while enhancing auxiliary efficiencies and reducing emissions.
The scope of the test report will depend upon your predetermined goals and system imbalances, which will be
determined from your own custom test program.
Our service is unique to the specific requirements of each plant, price quotes are prepared on a location-by-location
basis. Please contact us for a custom tailored proposal that meets all of your specific needs.
The demand of power is increasing exponentially results in installation of new stations whereas the sources of water are depreciating acutely. In future there may be a situation in which water sources may not cope up with this requirement.
Also the serious concerns of the regulatory authorities regarding usage of natural resources, definitely the norms will be further be tightened, which will curtail the freedom of usage of water in power plant.
In present scenario land acquisition is one of the toughest hurdles in plant installations which can be averted by locating stations in water scarce regions, by employing air cooled system which eliminates dependencies on water for CW.
Although dry cooling systems are costly technologies on techno-economic considerations, but foreseeing the future it is the need of hour to employ dry cooling system which offers possible solution for power plant installation eliminating the above mentioned challenges.
KK Wind Solutions presentation on Control System RetrofitRené Balle
KK Windsolutions presenting a strong retrofit case for Bonus, Siemens, Vestas and other turbines - showing significantly improved performance and short payback time.
This presentation was originally shown at the AWEA Wind Project O&M and Safety Seminar 2015 in San Diego.
For more information visit www.kkwindsolutions.com, nichr@kkwindsolutions.com or Phone: +4597221033
Developing a new generation of energy efficiency products for reciprocating e...Bowman Power
Learn how a new energy efficiency product gets made, from opportunity to concept, design, validation and production, with this free presentation from the 73rd Indonesia National Electricity Day & POWER-GEN Asia. #PGASIA
There is plenty of room for improvement in operations and maintenance activities. James Parle will tell how a team from Muir Data systems developed a system that lets wind tech report more thoroughly on their activities with easy-to-use portable devices.
And Grant Leaverton will report on how unmanned aerial vehicles can provides high resolution images of wind turbine inspections, especially blades, without rappelling from the nacelle.
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Subhajit Sahu
Abstract — Levelwise PageRank is an alternative method of PageRank computation which decomposes the input graph into a directed acyclic block-graph of strongly connected components, and processes them in topological order, one level at a time. This enables calculation for ranks in a distributed fashion without per-iteration communication, unlike the standard method where all vertices are processed in each iteration. It however comes with a precondition of the absence of dead ends in the input graph. Here, the native non-distributed performance of Levelwise PageRank was compared against Monolithic PageRank on a CPU as well as a GPU. To ensure a fair comparison, Monolithic PageRank was also performed on a graph where vertices were split by components. Results indicate that Levelwise PageRank is about as fast as Monolithic PageRank on the CPU, but quite a bit slower on the GPU. Slowdown on the GPU is likely caused by a large submission of small workloads, and expected to be non-issue when the computation is performed on massive graphs.
Data Centers - Striving Within A Narrow Range - Research Report - MCG - May 2...pchutichetpong
M Capital Group (“MCG”) expects to see demand and the changing evolution of supply, facilitated through institutional investment rotation out of offices and into work from home (“WFH”), while the ever-expanding need for data storage as global internet usage expands, with experts predicting 5.3 billion users by 2023. These market factors will be underpinned by technological changes, such as progressing cloud services and edge sites, allowing the industry to see strong expected annual growth of 13% over the next 4 years.
Whilst competitive headwinds remain, represented through the recent second bankruptcy filing of Sungard, which blames “COVID-19 and other macroeconomic trends including delayed customer spending decisions, insourcing and reductions in IT spending, energy inflation and reduction in demand for certain services”, the industry has seen key adjustments, where MCG believes that engineering cost management and technological innovation will be paramount to success.
MCG reports that the more favorable market conditions expected over the next few years, helped by the winding down of pandemic restrictions and a hybrid working environment will be driving market momentum forward. The continuous injection of capital by alternative investment firms, as well as the growing infrastructural investment from cloud service providers and social media companies, whose revenues are expected to grow over 3.6x larger by value in 2026, will likely help propel center provision and innovation. These factors paint a promising picture for the industry players that offset rising input costs and adapt to new technologies.
According to M Capital Group: “Specifically, the long-term cost-saving opportunities available from the rise of remote managing will likely aid value growth for the industry. Through margin optimization and further availability of capital for reinvestment, strong players will maintain their competitive foothold, while weaker players exit the market to balance supply and demand.”
Explore our comprehensive data analysis project presentation on predicting product ad campaign performance. Learn how data-driven insights can optimize your marketing strategies and enhance campaign effectiveness. Perfect for professionals and students looking to understand the power of data analysis in advertising. for more details visit: https://bostoninstituteofanalytics.org/data-science-and-artificial-intelligence/
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...John Andrews
SlideShare Description for "Chatty Kathy - UNC Bootcamp Final Project Presentation"
Title: Chatty Kathy: Enhancing Physical Activity Among Older Adults
Description:
Discover how Chatty Kathy, an innovative project developed at the UNC Bootcamp, aims to tackle the challenge of low physical activity among older adults. Our AI-driven solution uses peer interaction to boost and sustain exercise levels, significantly improving health outcomes. This presentation covers our problem statement, the rationale behind Chatty Kathy, synthetic data and persona creation, model performance metrics, a visual demonstration of the project, and potential future developments. Join us for an insightful Q&A session to explore the potential of this groundbreaking project.
Project Team: Jay Requarth, Jana Avery, John Andrews, Dr. Dick Davis II, Nee Buntoum, Nam Yeongjin & Mat Nicholas
2. Session Objectives
Today’s session objectives will be to
✓ Provide an overview of technologies driving data intelligence in the oil and gas industry
✓ Provide an overview of Microsoft and our partner’s innovation and gain an understanding of their
impact in this industry
✓ Gain an understanding of how business scenarios have benefited from these technologies through
stories and examples
✓ Discuss innovation challenges and barriers to adoption
3. Welcome, Agenda and Presenters
9:30 am Welcome & Introductions Tammi Warfield, Customer Success General Manager, Microsoft
10:00 am
Leveraging Machine Learning and AI for Oil and Gas
Intelligence
Bill Barna, Cloud Solution Architect – AI, Microsoft
11:00 am Injecting AI into Seismic Data Management
Liam Cavanagh, Principal Program Manager, Microsoft Azure Search
Olivier Lhemann, President, INT
12:00 pm Lunch & Networking
12:30 pm Predictive Maintenance
Ramkumar Krishnan Senior Program Manager, Microsoft
Naveen Vig, Principle Solutions Architect, Microsoft
1:15 pm Integrating AI with Spatial Analytics Omar Maher, Director of AI, ESRI
2:00 pm Pipeline Flow Quality Assurance on the Blockchain
Scotty Perkins, SVP of Product Innovation, Quisitive
Bryan Blain, Principal Consultant, Quisitive
3:00 pm Closeout Discussion and Next Steps Tammi Warfield, Customer Success General Manager, Microsoft
Today’Facilitator: Dania Kodeih, Industry Solutions Architect
4. Upstream Oil & Gas Analytics
Increasing production and optimizing drilling
May 21, 2018
Bill Barna
Cloud Solution Architect – AI
Microsoft Irving
bibar@microsoft.com
Jeff Johnston
Principal Consultant
Artis Consulting
jjohnston@artisconsulting.com
5. Microsoft Azure Provides the Building Blocks
• Reference architectures
• Everything fits together
• Fewer skills needed
• Flexible and agile
• Scalable
• Zero exit costs
Partial Listing of Azure Services
7. Production Costs in the Permian Basin
Subsurface Maintenance
*http://www1.salary.com/Roustabout-Salary.html
Secondary Oil Recovery
• Electricity ($19,990)
• Waste Water Disposal ($9,000)
• Rig Workovers ($19,710)
• Remedial Services ($8,930)
• Pumper Labor ($7,620)
• Roustabout Labor ($3,540)Targets
Data Science
Well Cost Data
• Data from 2009 Department of Energy Survey
• West Texas
• 8,000 foot well
Texas Production Wells
• 13 to 20 barrels per day in 2015 (all well types)
• $189k to $292k annual rev/well@ $40/barrel
• 190,000 to 250,000 oil wells in Texas
• 33,000 to 150,000 stripper wells in Texas
• Average stripper well produces < 2 barrels/day
2015 Median Annual Salaries
• Field Supervisor: $86,001/year
• Roustabout: $45,952/year
• Pumper: $42,600/year
Market Condition Metrics Annual Operating Costs
$13,170
$53,020
$3,010
$14,330
$13,120
$33,940
$0
$10,000
$20,000
$30,000
$40,000
$50,000
$60,000
$70,000
$80,000
$90,000
$100,000
Primary Secondary
Subsurface Maintenance Surface Maintenance
Daily Expense $0
$10,000
$20,000
$30,000
$40,000
Supervision Pumper Labor Chemicals Power & Water Supplies
$0
$3,000
$6,000
$9,000
Roustabout Labor Supplies & Services Equipment Usage Other
$0
$5,000
$10,000
$15,000
$20,000
Rig Workover Remedial Services Equipment Repair Other
Primary Secondary
Surface Maintenance
Daily Expense
8. Intelligent Pad
Azure AI-based pad monitors assets, predicts outcomes, and takes action
Wellhead
Free Water Knockout
Heater Treater
Waste Water Tank
Oil Tank
Gas Compressor
Monitoring Sensors
Monitoring Sensors
Monitoring Sensors
Monitoring Sensors
Tank Level Sensor
Tank Level Sensor
BS&W Sensor
Monitoring & Alarms
• Wellhead
• Free water knockout
• Heater treater
• BS&W sensor
Predictive Maintenance
• Gas compressor
Forecasting
• Tank levels
Azure Technologies
Dynamics 365 Field Service
• Automated ticket generation
• Oil & water haulers assigned
• Technicians dispatched
Tightly Integrated with Azure
• IoT asset monitoring
• Predictive maintenance
• Tank level forecasting
San Antonio
Datacenter
Technician
WALLIS 23-33-71 A 2H
Oil
Water
WALLIS 23-33-71 A 2H
9. Predictive Maintenance
Equipment sensor data can be used to improve maintenance schedules
Activity Cost
ESP Replacement $100,000
Workover Rig (3-days) $30,000
Trucking/Dock Services $6,000
Rental Tools & Equipment $5,000
Contract/Service Labor $4,500
Test/Inspect Pipe $2,000
Insurance (2% of total) $2,950
Contingency Costs (10% of total) $14,750
Total $165,200
Downtime 3 days lost
Event Classification Model Training by Component
Sensor Variables
• Revolutions per minute
• Amperage
• Temperature
• Vibration
• Etc.
Failure Risk
Motor
Pump
Normal
Anomaly
Support Vector
Machine
Automatic Detection
• Spikes
• Dips
• Level changes
• Trends
• Extreme values
Time
Sensor Signals
Anomaly
Electric Submersible Pump (ESP)
http://www.hexionfracline.com/flowback-causes-serious-issues-for-esps
Anomaly Detection Model Component Failure Prediction
Classification Anomaly Detection
Strength Known events Unknown events
Approach Supervised learning Unsupervised learning
Output ESP events Rate change precursors
10. Prevent Failures: Downhole Dynamometer Card
Machine learning can be used to avoid costly failures of oil well rod pump systems
ADF consumes aggregates
returns ML results
DocumentDB
Consumed
Produced
Machine Learning
Predictions
Daily notifications
generated
Notifications sent to
dispatch system
Data to Card Shapes
+ Area per shape
Quantify shape, match
to problem shapes
Data
No
Data Visualization
Shape Library
Source: Sage Oil Tools
Source: Lufkin Automation
Analysis Process & Diagnosis
11. Tank Level Forecasting
Machine Learning can be used to forecast and schedule oil tank pickups
• Machine learning used to forecast tank levels using
onsite sensor reading and historical data
• Tank forecasts used to schedule tank pickups and
minimize downtime at site
• Parallel model development in Azure ML allows for
rapid solution testing and development
• Advanced machine learning algorithms (Neural
Network Regression, Poisson Regression, Decision
Forest Regression) enable accurate prediction
Tank Forecast DevelopmentMultiple Algorithm Evaluation
Parallel Feature Testing
5
15
25
35
45
55
65
1 6 11 16 21 26 31 36 41 46
TankLevel
Day
Linear Regression
Random Forest
Neural Network
Level
Forecast Accuracy Impact
1 hour High Truck Route
1 day High Schedule
1 week Medium Schedule
Forecast Horizons
12. Corrosion Prediction
Reducing workover costs with proactive treatments
“If a little works, more will work better!”
– Production Engineer in Midland
Predict and treat corrosion before it
occurs.
14. Finding Production: Identifying Underproducing Wells
Using data mining to identify under producing wells because of misconfigured beam pumps
Why review configurations?
• Beam pump misconfigured at installation
• Downhole conditions change over time
• Stimulation changes system dynamics
• Equipment efficiency changes over time
Business impact of misconfigured wells
• More equipment failures
• Higher electricity consumption
• Oil production at less than 100 percent
Challenges to reviewing beam pumps
• Manual process
• Large volume of wells
• Many wells configured correctly
• Lack of skilled personnel
• Opportunity cost (higher priority work)
Asset Database Production
History
Service History
Gateway
Storage blob
Data FactoryHDInsight Machine
Learning
Data Lake
Solution
• Automated process
• Scheduled or ad-hoc
• Data Factory managed
• Hadoop preprocessing
• Data mining with ML
• Data Lake Storage
• PowerBI visualization
• Recommender engine
Are there beam pumps underproducing and/or at risk of failure?
• Production history
• Service history
• Asset inventories
• Configuration data
• Dynamometer cards
• Service history
• Water cut history
• Gas to oil ratio history
• Stimulation history
• Service bulletins
Data Sources
15. Reducing Electricity Costs: Beam Pumps
Machine Learning and Linear Programming optimize beam pump configurations to reduce electricity costs
Beam Pump Configuration Performance Metrics and Costs
Pumping Unit
Pump Size
and SPM
Rod String PIP
Max-Min
Loads
Load
Range
Gearbox Torque
Polish Rod
Horsepower
Monthly
Power Bill
Production
C-456-256-120 120" 1.50" - 10.20 SPM 86 100 25,576 8,594 16,982 478,700 20.85 $1,944.00 200 BFPD
C-456-256-120 103" 1.75" - 10.20 SPM 86 100 26,729 9,649 17,080 341,800 18.71 $1,482.00 200 BFPD
M-320-256-144 144" 1.25" - 10.40 SPM 86 100 24,608 5,910 18,698 305,400 25.46 $1,971.00 200 BFPD
C-320-256-120 120" 1.50" - 8.70 SPM 86087 100 18,097 6,022 12,075 289,600 18.97 $1,383.00 200 BFPD
Source: Lufkin
Production Forecasting
• Internal data
• Competitor/partner data
• Dynamometer cards
• Downhole logs
• Texas Railway Commission
• Frac Focus
(Oil and Water)
Regenerative Power
• Capture downstroke power
• Meter only runs forward
• Multiple wells per circuit
• Synchronization
• System balance
(Well Clustering)
Optimized Configurations
• Production forecast
• Regenerative power
• Stroke length vs torque
• Stroke length vs speed
• Equipment limits
• Efficiency & Durability
• Strategy over time
17. Drilling Costs in the Permian Basin
Reporting Companies: Energen, Concho, Laredo, EP Energy, and EOG Resources.
Drilling costs range from $88,000 to $109,000 per day
Trends in U.S. Oil and Natural Gas Upstream Costs, March 2016
18. Well Drilling Planning Process
Objectives
• Safe
• Minimum cost
• Usable
Drilling costs correlate directly to the effectiveness of the planning effort
Problem Type: Optimization
Source: Petro Wiki
Prospect Development
Data Collection
Pore Pressure Analysis
Fracture Prediction
Pipe Setting Depth Selection
Hole Geometry Selection
Completion Planning
Mud Plan
Cement Plan
Casing Design
Tubing Design
Rig Sizing and Selection
Drill Time Projections
Cost Estimate
Steps for developing a drilling plan
Constraints
• Geology
• Drilling equipment
• Temperature
• Casing limitations
• Hole sizing
• Budget
19. Drill Bits – Selecting and Grading Wear
“Historically, a driller would learn through experience how to examine a "dull" to determine what type of bit to run next, and how it should be run
(WOB, RPM, etc.). This was part of the art that separated the best drillers from the rest.” - SPE 23938 The IADC Roller Bit Dull Grading System
Drill Bits
- Many choices defined by IADC codes
- Most important variable in system*
- Decisions based on Driller’s experience
Drill Bit Wear
- Drillers use IADC grading system
- 2nd most important variable in system*
- Driller reads bits like tea leaves
Art becomes science
Driller empowered by AI
*In relation to lithology
20. Estimating Lithology for Model Inputs
No MWD available prior to planning a new drilling project
Permian Basin
• Layer cake formation
• Many lithologies
Seismic Data
• Can replace MWD (partially)
• Poor resolution and accuracy
Adjacent Well Data
• MWD logs from other projects
• Proxy data for current project
21. Drilling Optimization – An Evolutionary Process
SPE-133429-MS – Basic Drilling Cost Model and Optimization Theory (2010)
• SPE-169451-MS – Improves the bit wear model (2014)
• SPE-180518-MS – Improves the lithology variables (2016)
• SPE-178852-MS – Adds Bottom Hole Assembly variables (2016)
• SPE-181382-MS – Provides new AI models to evaluate drilling data to optimize ROP (2016)
• SPE-185909-MS – Improving ROP performance with steerable mud motors (2017)
• TBD - Delivering these solutions more easily, faster, and less expensively with Microsoft Azure
Oil & Gas Companies, Academia, and Microsoft all contributing to a solution
Solution Contributors
22. Optimizing Drilling Costs Per Foot
SPE-133429-MS: Cost-Per-Foot Reduction by Bit-Run Optimization: A Simulation Study (2010)
Rate of Penetration (ROP) Factors
• Bit type
• Bit tooth wear
• Formation characteristics (lithology)
• Drilling fluid properties
• Weight on Bit
• Rotary Speed
• Bit hydraulics
Lithology variables (feature inputs)
• Rock type (shale, lime, sand, chert)
• Fracture gradient
• Softness
• Abrasivity
• Gamma ray
• Resistivity
• Porosity
• Fluid type
• Pore fluid pressure gradient
Other variables (feature Inputs)
• Bit hydraulic horsepower
• Fluid type (water or oil)
• Mud weight (ppg)
• Mud flowrate (gpm)
Bit types and tooth wear by lithology
Original Costs
Optimized Costs
Costs down 17%
23. Minimize the Cost Per Foot Drilled in the Permian Basin
Storage blob
• Lithology
• Bit type
• Weight on Bit
• Rotary Speed
• Mud
• Hydraulics
Input Data
Machine
Learning
HDInsight
Power BI
IoT Hub
• Ingest
• Store
• Preprocess
• Model
• Visualize
Drilling Process Today
• Infinite configuration settings
• Limited pre-drilling data
• Experience dependent
• Driller overly conservative
• Hit & miss results
Future Drilling Process with AI
• Optimized configuration settings
• Improved data insights
• Experience augmented by math
• RoP balanced against TOOH risk
• Consistently better results
Microsoft Azure
Azure-based AI makes drillers smarter to increase Rate-of-Penetration (RoP) while reducing Tripping-out-of-Hole (TOOH)
Algorithm: Gradient Boosted Decision Tree
Tricone Bit PDC Bit
25. Artis Consulting ROP Demo
Data
• University Lands Data
• 3 wells from Andrews County
• MWD logs (includes lithology)
• Mud Logs
• Daily Drilling Reports
• ~4500 to ~14000ft depth modeled
Modeling
• 2 R Scripts were used
• one for regression model training &
prediction
• one for creating the permutations
• predictive model optimized to
protect against over-fitting
• one prediction every 2 feet of depth
Power BI Visualization
• Load well logs
• Load permutations data
• Load predictive model
output
• Allow user to configure
ROP Factors
Data
Engineering
Perms
Predictive
Model
Model
Output
Create
Lithology &
Bit Perms
Log Files
Cortana Gallery (Pending)
26. Public Data Used to Train Models
Predict Rate-of-Penetration (RoP) for different drilling configurations
