Generally, we use a coocking recipe style for the operations of production and its evaluation. As a result, the opportunity of maximum production of a well is reduced which minimizes the common benefits.

The efficient design of productive oil wells by artificial lifting requires the analysis of the components that make up the general system. This analysis is commonly called Nodal Analysis.

The Nodal Analysis defines the propositions that will be used to optimize a production system, for oil or gas wells. Each component of the producing system is analyzed in order to reach the desired production rate as economical as possible.

Among the objectives of the Nodal Analysis we can bring out:

  • Determine the fluid rate that an oil or gas well will produce, taking into account the hole geometry and the completion boundaries.

  • Optimize the system to produce an objective fluid rate as economica1 as possible.

  • Allow the fast recognition of the possible ways to increase the production rate.

During the last years different Nodal System Ana1ysis have been extensively used, due to the development of the computer technology that allow fast calcu1ations of complex algorithms, helping to improve the completion and production technics.

The fluids found in a productive horizon must be produced through a complex system till reaching its destiny in the surface, this is called a productive system that covers: the reservoir, the producing well and the fluid lines (ΔP).

Nowadays to design a producting system the elements must be analyzed as a group or as a whole (globa1 analysis) that according to the time has an effect on a Dynamic Global Analysis.

It allows to predict the effect of a producing zone and of the artificial lifting system in relation to the production of oi1, gas and water.

Due to the fact that the Dynamic Globa1 Analysis involves the dimension of time, considers changes in the flowing bottom pressure, in water saturation, in the oil-gas relation and in the discharge of pressure of the subsurface pump.

To each time there will be a convergence to new flow rates which reflect the reservoir performance, the hydrodynamics and thermodynamics conditions (hydraulic correlations of the produced fluids through the tubing) and the operation of the subsurface equipment, to show it the Thrasher, Fetkovich and Scott approach has been used in this case.

If we tried to set up this analytic technique along with a conventiona1 computer program, the old routines would rarely be used without any modification. But modifications always involve a risk when introducing errors and, sooner or later, a continuous increase of the number of versions will result confused.

Therefore, a new approach on programming based on the human thinking model (brain physiology) has been introduced, that is it uses connections and doesn't consider all the system's details. Thus, for instance, the tubing string is absent in our mind as a group of variables and equations, but as a "black box" with its inlet and outlet limits.

The application of this technique or connectionist model (Artificial Neural Networks), that show usefu1 behaviors when learning, recognizing and applying relations among real 1imits, is shown through the development of a computer program for monitoring a system of production, with the advantage of not requiring excesive answering time (in real time) neither of convergence problems related to the combination of the reservoir and the artificial lifting system.

The purpose of this investigation is to show the use of this technique of programming along with the Dynamic Global Analysis, to help solving the existing monitoring problems. Besides justifying the fixing of a sand of gas or water, the success of such fixing and / or recompletion of a reservoir and ana1ize the impact of the given production of a wel1 with respect to other neighbor wells of the same system (reservoir) without representing a significant increase of the costs (automation and personal training) for analysing and notifying the abnormal situations.

The accuracy in the performance control is 95 per cent, with respect to the information stored in the artificial neural networks during its testing period.

Achieving this way the development of a Base Profile of Behavior that along with the application of Artificial Neural Networks (BAM Type) becomes in a new "alternative" versus the conventional methods in the monitoring and control of the dynamics of the reservoir-artificial lifts system thus adding value to the petroleum industry operations.

You can access this article if you purchase or spend a download.