Predicting Reservoir Fluid Properties from Mud-Gas Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for predicting reservoir fluid properties, such as density, saturation pressure, and gas-oil ratio, are limited by their inability to accurately estimate properties influenced by oil-related components at an early stage of the drilling process.
Innovation Solution
A method is developed to generate a model for predicting reservoir fluid properties using mud-gas data, excluding samples with significant biodegradation to improve accuracy. This involves creating a computer-implemented method that utilizes a machine learning algorithm to predict properties like gas-oil ratio, fluid density, and formation volume factor based on measured mud-gas data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If mud-gas logging is used to determine reservoir fluid properties, then early stage analysis is possible, but the composition data is not comparable with actual reservoir fluid composition due to C1 richness and lacks oil component information
Solution Approach 1:
The patent uses machine learning models trained on comprehensive reservoir fluid composition data as an intermediary to translate mud-gas logging data into accurate predictions of reservoir fluid properties. The model acts as a mediator that converts the limited C1-C5 mud-gas composition data into reliable estimates of full reservoir fluid composition including oil components C7-C36+.
Solution Approach 2:
The patent transforms the mud-gas composition parameters (C1-C5 ratios) into reservoir fluid composition parameters (gas-oil ratio, density, saturation pressure) through machine learning models. This parameter transformation allows early stage mud-gas data to provide accurate predictions of properties that would otherwise require complete fluid sampling.
2Productivity
If LWD tools are used for continuous logging data, then probabilistic estimates of reservoir composition can be provided, but the tools are expensive and consume significant rig time
Solution Approach 1:
The patent creates a virtual copy of the expensive LWD logging capability by using machine learning models trained on comprehensive reservoir data. These models replicate the functionality of continuous LWD composition logging using the simpler and cheaper mud-gas logging data, providing continuous prediction of reservoir fluid properties without the high cost and rig time consumption of actual LWD tools.
3Measurement precision
If down hole fluid sampling is performed for accurate composition determination, then detailed fluid composition can be obtained, but the process is slow and cannot provide immediate feedback
Solution Approach 1:
The patent performs preliminary action by training machine learning models on comprehensive reservoir fluid composition data before actual field application. This pre-training allows the models to immediately predict accurate reservoir fluid properties from mud-gas logging data during drilling operations, eliminating the time delay associated with actual fluid sampling and laboratory analysis while maintaining high measurement precision.
Data Source
AI summary
The present disclosure relates to techniques for prediction of reservoir fluid properties of a hydrocarbon reservoir fluid, such as the density, the saturation pressure, the formation volume factor and the gas-oil ratio of the reservoir fluid. To predict the reservoir fluid properties, a model is generated by selecting a subset of available reservoir samples based on a degree of biodegradation of the samples, generating an input data set comprising input data and target data, the input data comprising measured or predicted mud-gas data; and generating a model using the input data. The application of this technique allows a continuous log of the selected property to be generated using mud-gas data collected during the well drilling process.