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.
4. Production Operations Pain Points
Manual data from
daily pumper visits
#3: too many wells, too little time
#1: low quality data + not enough data
#2: 20-year old control logic
30-minute data
via SCADAOR
POC
Gibbs Equation = 1960
VFD
Speed control logic = if/then
1 production engineer
150-200 wells
80/20 rule
20% of wells get 80% of attention
5. Lagging Automation Maturity in Artificial Lift
ESP
Rod
Pump
PCP
Plunger
Gas Lift
Automation maturity and growth Current Growth
Moderate adoption of
20-year old automation
technologies
Niche application means
lower technology adoption
Weak infrastructure and
well economics have limited
adoption
“Re-emerging” lift type and
complex system has seen
limited adoption
Higher value wells,
generally have telemetry
Growth
Growth
Growth
GrowthCurrent
Current
Current
Current
6. Every well, every day
Problem & Solution
Pump by exception
VFD + SCADA + XSPOC Adaptive Control SolutionTimer + Operator + Greasebook
Problem: Production Operations not set up for success
Solution: Take ‘pump by exception’ to the next level
7. Complementing & Advancing Legacy Systems
1990
XSPOC
SoftwareHardware
POC + VFD
SCADA
Modern Software UI
Big Data
Machine Learning
Artificial Intelligence
Smartphone
M2M Comms
Edge Computing
Distributed AI
Ambyint
201725+ years of innovation
Physics-Based Analytics
Anomaly Detection
Incident Characterization (ML)
Dynocard Screening (ML)
Torque Based Fillage Analytics
Well Lifecycle Management
8. +
Production Optimization Platform (POP)
SaaS Software Platform
Optimizing Production Operations through Physics-based
Analysis complemented with AI-Driven Monitoring & Control of
Artificial Lift Systems
High Resolution Adaptive Controller (HRAC)
Intelligent Devices
Automated & Intelligent Production Optimization
10. High-Impact Analytics: Need Physics + Data Science
Source: Terry Trieberg, Theta at ALRDC Gas Well Deliquification Workshop 2/22/17. Presentation titled, “Well Production Automation and Diagnostics: Past, Present, and Future”.
Physics-based
Analysis
traditional rod pump
optimization
Data Science &
Artificial Intelligence
“big data” statistics
Dover/XSPOC
Weatherford
GE/Lufkin
Flutura
Spark
Cognition
Ambyint
XSPOC key considerations for modern
analytics:
- Domain expertise (physics) is critical
- Low quality data limits analytics
Ambyint:
- Deep physics-based analytics expertise
- Proprietary hardware to generate better data
- Successfully deployed actionable modern data
science in artificial lift
- 5.8 x 1013
observations. 67,000x more than
nearest competitor
12. Ambyint solves key economic hurdle preventing adoption of conventional automation
in low value or remote wells
Changing the Game: Stripper Well Automation TCO
5 bopd well, no
automation, 70% uptime:
● Ambyint improves
uptime to 80%,
reduces Opex =
4-month payback
● Conventional
automation improves
uptime, reduced Opex
= NPV negative
$56,000
$3,000
13. Value Prop: Reduce Opex + Increase Production
Effective Base
Management
Increase uptime
Reduce failure cycle
time
Mitigate over-pumping
Reduce Intervention
Cost
Lower failure rates
Move from reactive to
proactive
Minimize over-pumping
Increase Operational
Leverage
Automate routing
workflows
Focus on higher-value
activities
Bridge field and office
14. Effective Base Management: Reducing Wellsite Visits
Customer Pre-Ambyint
● Visiting wellsites every day on set
routes, regardless of wellsite issues
● No remote surveillance to enable
pump-by-exception
● Heavily people-driven and
truck-reliant
● Heavily reliant on contract operators
Ambyint Impact
● Reduction of wellsite visits by 87%
● Redeployment/Reduction of contract
operator workforce for direct and
immediate Opex savings
● Reduction in windshield time improves
HSE
● Integration of leak detection device to
identify potential leak hazards
Operator calculating 3-4 month paybacks based on Opex reductions
15. Reduce Intervention Cost: Diagnose/Predict Gas Lock
Automatic Recommendation
Repeat
First
occurrenceAmbyint Screenshot
Normal
operation
Gas lock
operation
Ambyint Pattern Recognition
Operator calculating 1-3 month paybacks based on this value-prop alone
Customer Pre-Ambyint
● E&P consistently experiencing gas
lock across large portion of wells in
field
● Pump is actively pumping but not
producing
● Do not realize until you go out to
field and check production tanks
● Causes damage to equipment and
leads to catastrophic failure
● Can intervene to slow down pump, if
you can diagnose/predict
16. Increase Operational Leverage: Unlocking More Time for Well O
BBLs
Treatments
Failures
Strokes
Visits
Lever Direction
High Quality Data
Real time visibility
Remote on/off control
Logic control (fillage)
Physics analysis
Data science &
Predictive
Adaptive
control
Ambyint
only
Both
legacy &
Ambyint
Key well optimization
levers available
Legacy tools have
reached a performance
limit for optimization
due to data quality and
architecture
Ambyint overcomes
these obstacles by
providing a stronger
data foundation and
novel technology for
advanced well
optimization
Optimization ability, enabled by technology,
increases with more time/well
Ambyint enables
additional time/well
Routine Well Optimization
18. … years of data gathered from...
Exceptional Foundation
1000+
33M+
70/30
10+ … oil wells.
… dynamometer cards generated with expert classification!
… percentage of horizontal / vertical wells in our data lake.
5 ms … sampling rate of high resolution data.
20. SCADA Data Limits Data Science
Ambyint
Stroke Based
Observation
SCADA Fixed
Polling
Frequency
● Regular polling
frequency of SCADA
misses insightful events
● Uneconomic to ratchet
up polling without large
CAPEX spend
High resolution + stroke level + event-based data enables quality data science
● Ambyint HRACs capture
rich, stroke level data
● HRACs send insightful
events into data science
● Transmission costs
managed by Ambyint
Sticking?
Tagging?
Load
Time
Load
Time
Pumpjack
duty cycle
Ambyint
SCADA
21. Field data coming in
from Ambyint devices
IOT
Recommendation Engine
Data Analysis Big Data
- Ambyint
- WellExpert
- Public Data
Machine
Learning
Algorithms
1. Training sets
2. Models
3. Real data
Examples:
Gas lock / gas interference
Pattern recognition
Automated anomaly detection
Ambyint competitive advantages of data flows and analysis
1. Only integrated end to end adaptive control solution on the market
2. Continuous feed of new data with massive data lake from WellExpert to back-test new ideas
3. True closed loop machine testing and learning cycle
Predictive Analytics
22. Cognitive Card Recognition: A New Approach
Visual pattern recognition
(traditional approach) diagnosis
systems are inherently flawed
Ambyint data science utilizes
micro-pattern diagnosis +
expert-categorized data lake
to resolve this flaw
Humans only see macro-level
changes, missing the subtle
details hidden due to poor
data quality or high friction
1
2
3
23. Repeat occurrence:
well shut down
First
occurrence
Torque-Based Analytics
Normal
operation
Gas lock
operation
Ambyint screenshot
Ambyint pattern recognition
Operator calculating 1-3 month paybacks
based on this value proposition alone
Teaching the machine to predict and diagnose gas lock in oil wells.
24.
25. Creating a POC from an Electric Motor
Electric
Prime Mover
Current
Torque
Fillage Had to leverage our 33mm dynocards in our data lake; 22mm with torque
Challenge: How do you predict pump fillage from Torque?
1. Collect motor torque information
2. Need to determine when a stroke starts and ends
3. Apply physics based algorithms to stroke cycle
4. Machine learning model generated around features of torque patterns
a. Use max torque as up stroke, and voted method to decide starting
point
THIS IS HARD
26. ● Unsecured
local ports
● Susceptible
to DDoS
PLC/RTU Wireless Comm SCADA Server LAN / PC
● Clear text
● Sniffable
wireless
● Unencrypted
storage
● HDD
● Uptime?
● Backup?
● Unencrypted LAN
● No user level
security
● Susceptible to
malware/ransomw
are
Cybersecurity - Typical SCADA
27. ● Secure
device
● No local
ports
● DDoS attack
immune
Ambyint Device Satellite AWS Any Device
● Push
● Encrypted
●
● Encrypted
storage
● 99.99%
● Highly
available
● SSL Encryption
● Browser based
● Native apps
● User level and app
level security
Cybersecurity - Ambyint
29. Architecture Drivers
EC2/VPC:
- Auto scaling capabilities behind AWS ELB
- Automatically allocates compute capacity as needed
- Plan to expand to using ALB in the next 3-6 months
VPC
- Provides secure private subnets across availability zones which allow for
secure highly available systems
S3
- Cost effective data storage for data lake
CDN
- Globally distributed content delivery network for low latency delivery of
web assets
ECS
- Docker container management for microservice deployment of ambyint
microservices.