System and method for estimating reservoir fluid contamination

Patent metadata

Field	Value
Publication number	US-12590940-B2
Application number	US-202318115367-A
Country	US
Kind code	B2
Filing date	Feb 28, 2023
Priority date	Mar 11, 2022
Publication date	Mar 31, 2026
Grant date	Mar 31, 2026

Abstract

A system and method for determining reservoir fluid contamination of a reservoir is disclosed. The system includes one or more sensors to detect field measures from reservoir fluid of the reservoir and includes memory and at least one processor to execute instructions from the memory to cause the system to perform further steps. A step by the at least one processor is to simulate relationship data having purity levels correlated to volumes or times associated with extraction of an applied or simulated fluid for varying modeled rock formations. A further step is to fit one or more of the relationship data to the field measures to generate errors. Yet another step is to determine the reservoir fluid contamination from one of the purity levels that is associated with the relationship data that is within a threshold error value or error range of the errors.

First claim

What is claimed is:

1 . A system for determining reservoir fluid contamination for a reservoir, the system comprising:
one or more sensors to detect field measures from reservoir fluid of the reservoir; and memory and at least one processor to execute instructions from the memory to cause the system to: simulate a plurality of relationship data comprising a plurality of purity levels correlated to a plurality of volumes or times associated with extraction of an applied or simulated fluid for varying modeled rock formations; fit one or more of the plurality of relationship data to the field measures to generate a plurality of errors; and determine the reservoir fluid contamination from at least one of the plurality of purity levels that is associated with the one or more of the plurality of relationship data that is within a threshold error value or an error range of the plurality of errors.

2 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
project the plurality of errors to a distribution; and determine the threshold error value or the error range from a mean of the distribution.

3 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
train one or more neural networks using at least the plurality of relationship data and the plurality of errors; and infer the reservoir fluid contamination for a new reservoir based in part on new field measures used with the one or more neural networks.

4 . The system of claim 1 , wherein the one or more sensors are adapted for sensing one or more of optical measures, sound speed measures, refractive index measures, density measures, and viscosity measures as part of the field measures from the reservoir fluid of the reservoir.

5 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
generate a modeled rock formation to comprise pore structure using one or more images of a physical rock formation and using a plurality of physical variables; and simulate purity measures and volume or time measures for the modeled rock formation, the purity measures and the volume or time measures to be comprised in the plurality of relationship data.

6 . The system of claim 1 , wherein the varying modeled rock formations are generated from input data comprising one or more of a simulated formation pressure, a simulated formation porosity, a simulated formation permeability, a simulated formation flow rate, a simulated invasion parameter, a simulated anisotropy ratio, and simulated packers.

7 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
apply a smoothening algorithm to the field measures to generate a smoothened version of the field measures; and fit an individual set of the smoothened version of the field measures to an individual set of the plurality of relationship data to identify a linear match, wherein the linear match is used in part to generate the plurality of errors.

8 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
determine a purity level based in part on an extracted applied fluid through an individual physical rock formation having an injected contaminant and created as part of one or more physical rock formations from the varying modeled rock formations, the purity level contributing to the plurality of purity levels and a volume or a time of the extracted applied fluid contributing to the plurality of volumes or times.

9 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
determine a volume or a time associated with the reservoir fluid from a sample taken from the reservoir; and determining the reservoir fluid contamination for the sample based in part on an extrapolation of the volume or the time to at least one of the plurality of relationship data after applying one of the plurality of errors.

10 . The system of claim 1 , wherein the at least one processor executes the instructions from the memory to further cause the system to:
extrapolate, using the reservoir fluid contamination, a further reservoir fluid contamination for future volumes or times of the reservoir fluid extracted from the reservoir.

11 . A method for determining reservoir fluid contamination for a reservoir, the method comprising:
detecting, using one or more sensors, field measures from reservoir fluid of the reservoir; simulating, using at least one processor, a plurality of relationship data comprising a plurality of purity levels correlated to a plurality of volumes or times associated with extraction of an applied or simulated fluid for varying modeled rock formations; fitting, using the at least one processor, one or more of the plurality of relationship data to the field measures to generate a plurality of errors; and determining the reservoir fluid contamination from at least one of the plurality of purity levels that is associated with the one or more of the plurality of relationship data that is within a threshold error value or an error range of the plurality of errors.

12 . The method of claim 11 , further comprising:
projecting the plurality of errors to a distribution; and determining the threshold error value or the error range from a mean of the distribution.

13 . The method of claim 11 , further comprising:
training one or more neural networks using at least the plurality of relationship data and the plurality of errors; and inferring the reservoir fluid contamination for a new reservoir based in part on new field measures used with the one or more neural networks.

14 . The method of claim 11 , further comprising:
sensing, using the one or more sensors, one or more of optical measures, sound speed measures, refractive index measures, density measures, and viscosity measures as part of the field measures from the reservoir fluid of the reservoir.

15 . The method of claim 11 , further comprising:
generating a modeled rock formation to comprise pore structure using one or more images of a physical rock formation and using a plurality of physical variables; and simulating purity measures and volume or time measures for the modeled rock formation, the purity measures and the volume or time measures to be comprised in the plurality of relationship data.

16 . The method of claim 11 , wherein the varying modeled rock formations are generated from input data comprising one or more of a simulated formation pressure, a simulated formation porosity, a simulated formation permeability, a simulated formation flow rate, a simulated invasion parameter, a simulated anisotropy ratio, and simulated packers.

17 . The method of claim 11 , further comprising:
applying a smoothening algorithm to the field measures to generate a smoothened version of the field measures; and fitting an individual set of the smoothened version of the field measures to an individual set of the plurality of relationship data to identify a linear match, wherein the linear match is used in part to generate the plurality of errors.

18 . The method of claim 11 , further comprising:
providing one or more physical rock formations from the varying modeled rock formations; injecting a contaminant into an individ

Assignees

• Baker Hughes Oilfield Operations Llc

Inventors

• Coskun Sefer
• Hunnur Anup
• Shahkarami Alireza
• Yang Junjie
• Hall Emiliano

Classifications

• G01V20/00

  Geomodelling in general · CPC title

• G06N3/08

  Learning methods · CPC title

• E21B49/08

  Obtaining fluid samples or testing fluids, in boreholes or wells · CPC title

• E21B2200/20

  Computer models or simulations, e.g. for reservoirs under production, drill bits · CPC title

• G01N33/2823 Primary

  Raw oil, drilling fluid or polyphasic mixtures · CPC title