The directional steering of drilling is crucial to improving the recovery from petroleum reservoirs. Geosteering refers to the directing drilling and well placement in real time based on geology.
By inverting logging while drilling (LWD) data into quantitative descriptions of rock properties and 3-dimensional geological structures, we can gain a deeper understanding of the 3D geology around a wellbore, thereby improved well geosteering decision making.
There are several shortcomings to the current 3D well geosteering system, a) There is no automated, real-time, 3D geological interpretation and assessment of interpretational uncertainty while drilling (b) a standardized procedure is not available to identify all possible geological configurations along a planned well, (c) it is computationally expensive to calculate EM measurements obtained in 3D geomodel with the current inversion method that accounts for uncertainty, (d) the range of Machine Learning (ML) based technologies for geosteering is very limited, and we are unaware of any method that identifies geological features.
The primary objective is to develop and verify a workflow for automatic, real-time, around-wellbore 3D geological interpretation of LWD logs for optimal well placement decisions. The following secondary objectives will be addressed:
• Around-wellbore geomodelling: Generate a set of site-specific around-wellbore 3D geomodels that capture all known and expected geological features and configurations expected to be encountered along the well.
•Probabilistic 3D inversion: Robust, accurate and computationally effective ensemble-based inversion of LWD logs (including deep sensing borehole EM measurements) to quantify uncertainties.
• ML methods: ML-based methods for real-time geological interpretation to complement and expand on the analytical methods.
• Verification: Assess the quality of the around-wellbore interpretation under various geological conditions through case studies.
Geosteering decisions are based on geological models. To make optimal decisions while drilling, it is necessary that geological models for the near-wellbore region are well-calibrated against measurements received while drilling, and that uncertainties are quantified. The current work processes for geosteering suffer from shortcomings. (i) It is highly challenging to geologically interpret LWD measurements in real-time for complex formations in an objective manner. (ii) It is highly challenging to update geomodels from available interpretations. (iii) There is a lack of a transparent, systematic, consistent and effective workflow for quantifying complex geological uncertainties.
The primary objective of this project is to establish and verify a workflow for automatic, real-time, objective around-bit 3D geological interpretation of LWD logs for optimal well placement decision support while drilling in complex formations. The format of the interpretation supports future automatic updates of standard probabilistic geomodels.
To achieve this, we will address the following secondary objectives:
- Around-bit geomodelling. In the pre-drill phase, generate a set of site-specific geomodels that capture all possible geological configurations that can be expected along the well. This ensures that the basis for inversion is firmly grounded in the geological understanding of the drilling target, and constrains the spread of possible interpretation outcomes.
- Probabilistic 3D inversion. Robust, accurate and computationally effective, ensemble-based inversion of LWD logs (including deep sensing EM logs) to quantify uncertainties.
- ML methods. Machine learning methods for real-time geological interpretation to complement and expand on the analytical methods.
- Case studies. Assess the quality of the around-bit interpretation under various geological conditions, applied to both synthetic and real data.