A HEOLOGIC COMPANY
FROSWELL FIELD SIMULATOR HAS THE HIGHEST DATA ACCURACY IN THE INDUSTRY
Dataset
Considering real-life situations, Froswell Simulator is designed to handle incomplete datasets that do not necessarily contain relevant information and in many cases misrepresent the characteristics and behaviour of reservoirs.
Here are some key components to consider for a data set:
​
Geological Data
the structure, stratigraphy, and rock properties of the reservoir: layers, formations, rock types, porosity, permeability, and fluid properties.
​
Well Data
well-specific data, including well locations, completion details, well trajectories, well logs (gamma-ray, resistivity, porosity, etc.), production rates, pressure data, and any historical production or injection data.
​
Reservoir Properties
reservoir properties such as initial fluid saturations, reservoir pressure, temperature, formation compressibility, and relative permeabilities.
​
Fluid Properties
viscosity, density, composition (oil, gas, water)
​
Boundary Conditions
the reservoir extent, any external pressures or constraints, and fluid flow interactions with neighboring reservoirs or boundaries.
​
Heliometric Data
wellhead monitoring, areal Helium Survey
Automated Data Matching
Froswell's automated data matching process identifies and matches relevant data from datasets based on specific criteria and parameters. It involves comparing data elements across multiple datasets to find matching records or establish relationships between them.
​
Matching Criteria
the specific criteria and parameters selected for matching the data (relevant attributes).
​
Data Preparation
the datasets cleaned and standardized to ensure consistency and improve matching accuracy.
​
Data Comparison
the data elements compared from one dataset with those from another to identify potential matches.
​
Matching Algorithm
tool employs different algorithms and techniques to efficiently compare and identify matching records based on the defined criteria.
​
Recording Linkage
the matched records from different datasets are linked or associated, establishing relationships or connections between them.
Validation
automated validation to ensure the correctness of the matching outcomes.
​
Post-Matching Processing
merging datasets, aggregating information, updating records, generating reports, or conducting additional data analysis tasks if required.
Physical Model
Froswell physical model is based on the most advanced "INSIM" ("Integrated Network Simulator for Integrated Models.)" It is a physical modelling approach used in the oil industry to simulate and analyze the behavior of complex interconnected systems in an oil field.
The INSIM approach involves creating physical models that represent the various components and processes within an oil field, such as reservoirs, wells, pipelines, surface facilities, and production networks. These physical models are typically constructed using scaled-down replicas or physical elements that mimic the behavior of the real system.
By utilizing physical models, the INSIM approach allows to understand the dynamics and interactions within the oil field system. It enables to assess the impact of different operational scenarios, optimize production strategies, evaluate the performance of equipment and facilities, and make informed decisions regarding field development and management.
INSIM is particularly valuable in studying complex oil field systems where interactions between reservoirs, wells, and production networks play a significant role in overall performance. The physical models used in INSIM provide a tangible representation of the system and allow for hands-on experimentation and analysis.
Field Development Forecast
Field Development forecasting involves using physical model, historical data of field and field development activities, heliometric data and machine learning algorithms to calculate different development scenarios, predict future events or outcomes of certain activities and well interventions.
Froswell Simulator utilizes patterns and relationships according to the following process:​
Data Collection of previous field development activities
relevant historical data related to the activity subject to forecast (type, time, location, well and reservoir conditions, results of previous activities etc. ).
Data Preprocessing
the collected data preprocessing by handling missing values, removing outliers, and transforming it into a suitable format for machine learning algorithms ensuring that the data is in a consistent and usable state.
​
Engineering
the most informative features from the collected data identified or new features created to enhance the prediction accuracy.
​
Physical Model Matching
the collected data matched with the physical model by means of machine learning algorithms referring to the specific requirements and characteristics of the activities and events being predicted.
Model Training
the machine learning model training sessions to learn the patterns and relationships within the data that can be used for event and activity result forecasting.
Model Evaluation
the performance of the trained model assessed using the validation data and measuring its accuracy, precision, recall, F1-score, or other relevant metrics to evaluate its effectiveness in predicting.
​
Activity result and Event Forecasting
the trained model applied to new or unseen data to make predictions feeding the relevant features or factors into the model and obtaining forecasts or predictions for the target activity or event.
Monitoring and Refinement
Froswell continuously monitors the performance of the forecasting model and updates monthly as new field development data becomes available.
Used machine learning model refined by retraining it periodically or incorporating new features or techniques to improve its accuracy over time.
