Our methodology is simple: Our expertise meets your expertise to empower your wells.
How we develop predictive models:
1. We run analytics on select well groups
We run & train models to analyze your wells’ operating states and display results in our system. We discuss & agree with
you on problems to look for.
2. Our oilfield experts review initial results
They view and validate the model results, limiting the burden on your team.
3. We ask you questions
We schedule Q&A sessions to better understand your data and operations and focus that knowledge on model improvement.
4. We review results with you
Our experts review and escalate results to your team. Reviews & all communication happen within OspreyVision.
5. We review models with your feedback
We run & train models to analyze your wells’ operating states and display results in our system.
6. You add more wells over time
As model results improve, well groups or “cohorts” get larger until all wells are online.
Learn about how our Predictions work
We monitor and model on multiple sensors
First, our models look for normal and indicative problem patterns across all signals and engineering models.
Your Artificial Lift experts and ours label problem states across this data, known as events.
We normalize your raw data to create modeling features
We determine a normalized signal for wells by type and across fields. This removes the need to analyze by individual
wells. Our models work across various wells regardless of size, allowing for rapid deployment across your fields.
Our Machine Learning finds labeled patterns
We employ different machine learning models to find the patterns labeled by the expert team. Machine learning greatly
automates this historically manual process.
We train & test to ensure models are working
We train and test to ensure models perform well on new data.
We track key measurements of Model Performance
We provide complete transparency to model effectiveness at every stage. We constantly work to improve model performance
by monitoring Accuracy, Precision and Recall metrics.
You can adjust Model Performance to your liking
Change your models’ performance thresholds to meet your business objectives.
The End Result: A powerful Predictive Platform that’s uniquely yours
As little as $1 per well a day
With our 30x ROIs for past clients, you won’t believe how affordable the OspreyData platform is.
See the Demo
Talk to sales to schedule a product demo & receive a bundled quote.
What problems can you conquer?
- GAS INFLUX, SLUGGING, LOCK
- HOLE IN TUBING
- BROKEN SHAFT
- RISING TEMPERATURE
- VOLTAGE IMBALANCE
- EXCESSIVE PUMPOFF
- HIGH RISK OPERATION
- PUMP FAILURE
- CABLE SHORT
- TUBING FAILURE
- PUMP FAILURE
- BAD LOAD CABLE
- IMPROPER POC SETTINGS
- LEAKING/STUCK TRAVELING VALVE
- LEAKING/STUCK STANDING VALVE
- EXCESSIVE PUMP OFF
- FLUID POUND
- GAS POUND
- GAS INTERFERENCE
- FLOWING WELL
- PUMP TAGGING TOP/BOTTOM
- OVER INJECTION
- UNDER INJECTION
- ICE PLUGGING
- MITIGATE NGL FLOW
- VALVE THROTTLING
- OPERATING OFF TARGET VALVE
- LEAKING VALVE
- STUCK/PLUGGED VALVE
- LIQUID LOADING
- FLOWLINE RESTRICTION
- VALVE FAILURE
- COMPRESSOR FAILURE
- BACK PRESSURE
- PLUNGER VELOCITY +/-
- WELLBORE RECHARGE
- SHORT TRIPPING
- AFTER FLOW +/-
- STATIC BOTTOM HOLE PRESSURE
- PUMPING BOTTOM HOLE PRESSURE
- FRICTION CHANGES
- GAS/OIL MIXTURE CHANGES