Abstract

High frequency reservoir surveillance data become available in an increasing number of oil and gas fields.

Real-time data from both reservoir and surface facilities open the possibility to control and consequently optimize field production in real-time. This real-time control would be a step forward to the industry's next goal - The Smart Field.

An integrated software and data approach is presented based on Data Mining methods1. In this paper, the "Automation Task" concept is discussed which allows automation of data processing, event detection and user notification. An Expert System interface enables the surveillance engineer to program the software in order to run on a 24/7 basis. Time savings in routine reservoir surveillance and accelerated production through faster and better reservoir management decision were identified as premium goals.

Introduction

Data Mining methods have been added to conventional reservoir surveillance tools. Methods like Neural Networks provide capabilities which are especially of interest to provide an automated reservoir surveillance tool:

  • learning from data, i.e. deriving models

  • fast execution of trained models (real-time)

  • detecting trend violations in high-dimensional problems

  • error tolerant

  • can handle missing values

  • no predefined model architecture

Currently implemented spreadsheet solutions cannot provide a real-time solution. The huge amount of data, automation and time critical operations cannot be handled by such systems. Data handling work is still the doiminating time consumption in routine reservoir surveillance.

Spreadsheet solutions can hardly be maintained in a way, that all engineers use the same data and the same software tools (programs) beside their limited storage capacities.

Available SCADA systems provide information about individual parameter values. Nevertheless a single pressure gauge does not give the complete picture about field performance and cannot be used directly for production optimization. Therefore real-time surveillance models which include reservoir performance and surface facilities constraints are needed.

Value loop in reservoir management

SHELL's Smart Field group promotes the value loop concept as basis for the closed-loop control including the main steps of reservoir management

  • data acquisition

  • interpretation and modelling

  • generate and evaluate options

  • execution

Keywords:
upstream oil & gas, artificial intelligence, constraint, surveillance, increment, software, spe 83974, automation task, data mining, optimization
Subjects:
Information Management and Systems, Data mining, Artificial intelligence
This content is only available via PDF.
2003. Society of Petroleum Engineers
You can access this article if you purchase or spend a download.