Virtual-Intelligence Applications in Petroleum Engineering: Part 1--ANN

Today, I read the paper 'Virtual-Intelligence Applications in Petroleum Engineering: Part 1--ANN'. I know more about ANN and find something similar to what I think when reading other papers.

Summary

Virtual intelligence has been used to solve problems related to pressure-transient analysis, well-log interpretation, reservoir characterization, candidate-well selection for stimulation and other such areas.

The development of new learning algorithms, such as backpropagation, revives the growth of neural-network research and application.

The human brain, although a million times slower than common desktop PC’s, can perform many tasks at speeds that are orders of magnitude faster than computers because of its massively parallel architecture.

ANN are information-processing systems that are a rough approximation and simplified simulation of this biological process and have performance characteristics similar to those of biological neural networks.

A multilayer network usually consists of an input layer, one or more hidden layers, and an output layer.

Neural networks have shown great potential for generating accurate analysis and results from large historical databases, the kind of data that engineers may not consider valuable or relevant in conventional modeling and analysis processes.

Many more areas exist for application of neural networks in the oil and gas industry including field development, two-phase flow in pipes, identification of well-test-interpretation models, completion analysis, formation-damage prediction, permeability prediction and fractured reservoirs.

There are two tables shown as follows:

Table 1 shows the accuracy of this method when applied to the four wells being studied.

Table 2 shoes that recoverable-reserve calculations based on virtual MR logs are quite close to those of actual MR logs.

Similar to what I think

ANN have the potential to be used as an analytical tool for generation of synthetic MR logs from conventional geophysical well logs. Part of the well data should be used to train a neural network and the remaining well data are used as verification. This method is most useful for fields with many wells where only a handful need to be logged with MR tools. These wells can be strategically placed to capture as much reservoir variation as possible. Then, a virtual MR application can be developed on the basis of these wells and applied to the rest of the wells in the field.

Tomorrow, I will read more papers on ANN.