US-12618324-B2 - Predicting formation pore pressure in real time based on mud gas data

US12618324B2US 12618324 B2US12618324 B2US 12618324B2US-12618324-B2

Abstract

A system for estimating a pore pressure value associated with a depth of a well subject to drilling operations may include a data repository for storing integrated mud gas and pore pressure data associated with one or more existing wells. The data repository may also store a machine learning (ML) engine. The system may also include one or more hardware processors configured to train a ML model using the ML engine and the integrated mud gas and pore pressure data, to estimate, during the drilling operations, the pore pressure value of a formation zone, at the depth of the well, using the trained ML model and mud gas data associated with a depth value that identifies the depth of the well subject to the drilling operations, and to update a drilling program for a production system based on the estimated pore pressure value.

Inventors

Fatai A. Anifowose
Mokhles M. Mezghani

Assignees

SAUDI ARABIAN OIL COMPANY

Dates

Publication Date: 20260505
Application Date: 20211217

Claims (17)

1 . A system for estimating a pore pressure value associated with a depth of a new well subject to drilling operations, the system comprising: a data repository for storing integrated mud gas and pore pressure data associated with one or more existing wells, and a machine learning (NIL) engine, wherein the integrated mud gas and pore pressure data is generated by integrating historical mud gas data with historical pore pressure data according to a correspondence comprising: one or more depth values that identify one or more depths of the one or more existing wells; and a nonlinear mathematical relationship determined by multiplying the historical mud gas data by a weight factor value, wherein the weight factor value is determined by a nonlinear degree of correlation between the historical mud gas data and the historical pore pressure data, thereby forming a depth-specific predictive relationship for an estimation of pore pressure; an access module configured to: access the integrated mud gas and pore pressure data, access the ML engine, and access real time mud gas data comprising a real time mud gas value and an associated depth value that identifies the depth of the new; and one or more hardware processors configured to: train a ML model using the ML engine and the integrated mud gas and pore pressure data, wherein the trained ML model is configured to receive the real time mud gas data associated with the depth of the new well and generate the estimated pore pressure value of a formation zone of the new well based on the real time mud gas data; and during the drilling operations, iteratively: obtain, via the access module, the real time mud gas data; input the real time mud gas data into the trained ML model to obtain the estimated pore pressure value corresponding to the real time mud gas value and the associated depth value; and update a drilling program for a production system of the new well using the estimated pore pressure value, wherein updating the drilling program comprises adjusting a density of drilling mud weight at a drill bit to stabilize a hydrostatic pressure of the new well and mitigate a risk of underpressure or overpressure.
2 . The system of claim 1 , the system further comprising: a separator tank configured to: collect returned mud that is returned to a surface during the drilling operations of the new well, and degas a gas mixture from the returned mud; and an analysis tool configured to generate the real time mud gas data based on analysis of the gas mixture.
3 . The system of claim 1 , wherein the real time mud gas data includes data pertaining to a light gas that is liberated from the formation zone during the drilling operations.
4 . The system of claim 1 , wherein the real time mud gas data includes data pertaining to a heavy gas that is liberated from the formation zone during the drilling operations.
5 . The system of claim 1 , wherein during the updating of the drilling program, the one or more hardware processors is further configured to: determine that the estimated pore pressure value deviates from a hydrostatic pressure value of the new well by at least a threshold value; and determine a cause of the deviation.
6 . A method for estimating a pore pressure value associated with a depth of a new well during drilling operations, the method comprising: during the drilling operations, iteratively: accessing real time mud gas data comprising a real time mud gas value and an associated depth value that identifies the depth of the new well; generating an estimated pore pressure value of a formation zone at the depth of the new well using one or more hardware processors and a trained machine learning (ML) model, wherein the trained ML model is configured to receive the real time mud gas data associated with the depth of the new well and generate the estimated pore pressure value of a formation zone of the new well based on the real time mud gas data; wherein the ML model is trained using integrated mud gas and pore pressure data associated with one or more existing wells, wherein the integrated mud gas and pore pressure data is generated by integrating historical mud gas data with historical pore pressure data according to a correspondence comprising: one or more depth values that identify one or more depths of the one or more existing wells; and a nonlinear mathematical relationship determined by multiplying the historical mud gas data by a weight factor value, wherein the weight factor value is determined by a nonlinear degree of correlation between the historical mud gas data and the historical pore pressure data, thereby forming a depth-specific predictive relationship for an estimation of pore pressure; and updating a drilling program for a production system of the new well using the estimated pore pressure value, wherein updating the drilling program comprises adjusting a density of drilling mud weight at a drill bit to stabilize a hydrostatic pressure of the new well and mitigate a risk of underpressure or overpressure.
7 . The method of claim 6 , further comprising: collecting returned mud that is returned to a surface during the drilling operations of the new well; degassing a gas mixture from the returned mud; and generating the real time mud gas data based on analysis of the gas mixture.
8 . The method of claim 6 , wherein the real time mud gas data includes data pertaining to a light gas that is liberated from the formation zone during the drilling operations.
9 . The method of claim 6 , wherein the real time mud gas data includes data pertaining to a heavy gas that is liberated from the formation zone during the drilling operations.
10 . The method of claim 6 , wherein the updating of the drilling program includes: determining that the estimated pore pressure value deviates from a hydrostatic pressure value of the new well by at least a threshold value; and determining a cause of the deviation.
11 . A method for training a machine learning (ML) model to estimate a pore pressure value associated with a depth of a new well subject to drilling operations, the method comprising: accessing a mud gas log associated with an existing well, wherein the mud gas log comprises historical mud gas data; accessing a pore pressure log associated with the existing well and corresponding to the mud gas log, wherein the pore pressure log comprises historical pore pressure data; and training, using one or more hardware processors, the ML model to: receive real time mud gas data associated with the depth of the new well and generate an estimated pore pressure value of a formation zone of the new well based on the real time mud gas data, wherein the ML model is trained using an integrated dataset comprising the historical mud gas data and the historical pore pressure data to determine one or more relationships between mud gas characteristics and pore pressure, wherein the integrated dataset is generated by integrating the historical mud gas data with the historical pore pressure data according to a correspondence comprising: one or more depth values that identify one or more depths of the existing well; and a nonlinear mathematical relationship generated by multiplying the historical mud gas data by a weight factor value, wherein the weight factor value is determined by a nonlinear degree of correlation between the historical mud gas data and the historical pore pressure data, thereby forming a depth-specific predictive relationship for an estimation of pore pressure; wherein, during the drilling operations, the one or more hardware processors is configured to iteratively: obtain the real time mud gas data comprising a real time mud gas value and an associated depth value that identifies the depth of the new well; input the real time mud gas data into the trained ML model to obtain the estimated pore pressure value corresponding to the real time mud gas value and the associated depth value; and update a drilling program for a production system of the new well using the estimated pore pressure value, wherein updating the drilling program comprises adjusting a density of drilling mud weight at a drill bit to stabilize a hydrostatic pressure of the new well and mitigate a risk of underpressure or overpressure.
12 . The method of claim 11 , wherein the outputting of the estimated pore pressure value includes: generating an output space of the ML model based on applying a function to an input space of the ML model, the input space including a product of the weight factor and a mud gas value included in the mud gas log, the output space including one or more estimated pore pressure values including the estimated pore pressure value.
13 . The method of claim 11 , further comprising: determining that a difference between the estimated pore pressure value that is output by the ML model and an actual pore pressure value that is included in a validation dataset exceeds an error threshold value; and adjusting a learning parameter of the ML model to a pre-determined value.
14 . The method of claim 11 , further comprising: determining that a difference between the estimated pore pressure value that is output by the ML model and an actual pore pressure value that is included in a validation dataset does not exceed an error threshold value; and identifying the ML model as being validated for receiving the real time mud gas data associated with the new well, as input to estimate a real time pore pressure log for the new well during the drilling operations of the new well.
15 . The method of claim 11 , further comprising: accessing the real time mud gas data associated with the drilling operations of the new well; accessing real time pore pressure data associated with the new well and corresponding to the real time mud gas data associated with the drilling operations of the new well; and calibrating the trained ML model based on the real time mud gas data associated with the drilling operations of the new well and the real time pore pressure data associated with the new well.
16 . The method of claim 11 , wherein the real time mud gas data includes data pertaining to a light gas that is liberated from the formation zone during the drilling operations.
17 . The method of claim 11 , wherein the real time mud gas data includes data pertaining to a heavy gas that is liberated from the formation zone during the drilling operations.

Description

BACKGROUND Conventionally, pore pressure analyses include three aspects: pre-drill pore pressure prediction, pore pressure prediction while drilling, and post-well pore pressure analysis. The pre-drill pore pressure is predicted using seismic interval velocity data in the planned well location, as well as geological, well logging, and drilling data in offset wells. The pore pressure prediction while drilling mainly uses logging while drilling (LWD) data, measurement while drilling (MWD) data, drilling parameters, and mud lithology data for analyses. Pore pressure may be calculated based on overburden and effective stresses. Overburden stress may be determined from bulk density logs, while effective stress is correlated to well log data, such as resistivity, sonic travel time or velocity, bulk density, and drilling parameters (e.g., D exponent). The post-well analysis may include analysis of pore pressures in the drilled wells using available data to build a pore pressure model. The pore pressure model may be used for pre-drill pore pressure predictions in future wells. Pore pressure is commonly estimated from shale properties derived from well log data which include acoustic travel time or velocity and resistivity. Pore pressure has also been estimated using other wireline logs such as true vertical depth (TVD), unconfined compressive strength (UCS), gamma ray, neutron porosity (NPHI), and bulk density (RHOZ). Further, pore pressure has been estimated from combined drilling parameters and log data, such as weight on bit (WOB), rotary speed (RPM), rate of penetration (ROP), mud weight (MW), bulk density (RHOB), and porosity. In addition, pore pressure has been estimated from only seismic data or from rock elastic properties. The conventional methods of estimating pore pressure may not have a high degree of accuracy because they may be based on empirical equations that assume linear relationships. Abnormal pore pressures may cause serious drilling incidents such as blowouts, kicks, fluid influx, pipe sticking, and lost circulation, and may greatly increase drilling non-productive time. To avoid such incidents, pore pressure needs to be accurately estimated and closely monitored while drilling. Accordingly, there is a need for a system that accurately predicts pore pressure ahead of coring, wireline logging, and formation testing activities, and without using seismic, surface drilling parameters, and wireline logs used in the current approaches. SUMMARY This summary is provided to introduce concepts that are further described below in the detailed description. This summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used as an aid in limiting the scope of the claimed subject matter. In general, in one aspect, embodiments disclosed herein relate to a system for estimating a pore pressure value associated with a depth of a well subject to drilling operations. The system includes a data repository for storing integrated mud gas and pore pressure data associated with one or more existing wells. The data repository also stores a machine learning (ML) engine. The system includes an access module configured to access the integrated mud gas and pore pressure data, to access the ML engine, and to access mud gas data associated with a depth value that identifies the depth of the well subject to the drilling operations. The system includes one or more hardware processors configured to train a ML model using the ML engine and the integrated mud gas and pore pressure data. The one or more hardware processors are also configured to estimate, during the drilling operations, the pore pressure value of a formation zone, at the depth of the well, using the trained ML model and the mud gas data associated with the depth value that identifies the depth of the well subject to the drilling operations. The one or more hardware processors are also configured to update a drilling program for a production system based on the estimated pore pressure value. In general, in one aspect, embodiments disclosed herein relate to a method for estimating a pore pressure value associated with a depth of a well, during drilling operations. The method includes accessing mud gas data associated with a depth value that identifies the depth of the well subject to the drilling operations. The method includes generating an estimated pore pressure value of a formation zone, at the depth of the well, during the drilling operations, the generating being performed using one or more hardware processors, the mud gas data, and a trained machine learning (ML) model. The method includes updating a drilling program for a production system based on the estimated pore pressure value. In general, in one aspect, embodiments disclosed herein relate to a method for training a machine learning (ML) model to estimate a pore pressure value associated with a depth of a well subject to drilling operations. The