Artificial Neural Network Ability to Synthesize Effective Porosity and Permeability Components of Nuclear Magnetic Resonance (NMR) Log from Conventional Well Logs Data

Nuclear magnetic resonance (NMR) log provides the capability of reservoir characteristics measurement such as permeability, free fluid porosity (FFP) and bound fluid volume (BFV). However this tool is one of the most important logs in reservoir characterization but there are some limitations i.e. it is impossible running this log in cased holes. On the other hand, due to high costs of NMR logging, a few wells in a field have this log. To overcome these issues, it will be needed a new method for generating NMR log synthetically.Since permeability and effective porosity are the two fundamental reservoir properties which have a significant impact on fields operations and reservoir management; current study presents, from a practical point of view, an excellent approach to synthesize permeability and FFP components of NMR log in a useful and fast manner and at a much lower price compared to the NMR logging practice. The presented methodology incorporates a back propagation artificial neural network (BP-ANN) as its main tool. The proposed methodology is presented with an application to field information of a carbonate reservoir, located in Persian Gulf, Iran.Validity of the results is checked through the comparison of synthesized values to real amounts of a set of data which is not included in development of the optimized ANN model and the correlation is presented. The results demonstrate that artificial neural network is an efficient and trustworthy way in synthesizing NMR components with an acceptable degree of accuracy.

Keywords:

Nuclear magnetic resonance (NMR) log , Artificial Neural Network (ANN) , Free Fluid Porosity , Permeability , Carbonated Reservoir , Persian Gulf