«Oilfield engineering»

Application of neural networks for LWD interpretation

UDC: 622.276:550.832
DOI: -

Authors:

¹ Institute of Geology and Development of Combustible Minerals, Moscow, Russia
² Tyumen Oil Research Center, Tyumen, Russia
³ Taas-Yuryakh Neftegazodobycha, Lensk, Russia

Keywords: LWD interpretation, neural networks, reservoir classification, horizontal wells, drilling process geological monitoring, geo-steering

Annotation:

The main goal of horizontal wells drilling petrophysical support is quick-look formation evaluation and determination of net reservoir thicknesses and formation fluid. Log interpretation results are used for estimation of geological objectives achievement and initial flow rate forecast. Reservoir classification is necessary for permeability estimation increase in case of drilling in high geological uncertainties. It leads to high quality geo-steering in high permeable intervals. However such complicated approach can have a negative impact on time spent on log interpretation reports during geo-steering process. The experience of neural networks approach used for quick-look LWD interpretation is represented in this paper. In particular neural networks were used for reservoir classes determination by LWD data. Reservoir classification was performed via core analysis and sedimentological description (J. Lucia approach). Trained Kohonen network is integrated in software in use and does not need any additional time expenditures. Patterns for recognizing reservoir classes via LWD data were obtained with sufficient accuracy. Further training of the neural network can increase the reliability of class identification. The results of comparison of permeability via J. Lucia adapted dependencies and permeability via core general equation are represented in this paper. The comparison of actual and expected initial flow rates for wells drilled in 2023 is represented as well. Such analysis demonstrates the increase of calculations reliability. Thus, new approach allows to increase the reliability of LWD interpretation, initial flow rates forecast and geo-steering and completion recommendations without any additional time expenditures.

Bibliography:

1. Lucia F.J. Carbonate Reservoir Characterization. – Springer, 2007. – 342 p.
2. Manzhula V.G., Fedyashov D.S. Neyronnye seti Kokhonena i nechetkie neyronnye seti v intellektual’nom analize dannykh // Fundamental’nye issledovaniya. – 2011. – № 4. – S. 108–114.