Facies Modeling Based on Comprehensive Sedimentary and Diagenesis Study Assisted by Neuro Network for an Extremely Heterogeneous Carbonate Reservoir in UAE

Abstract

Facies model is the basis of the geological model and determines the overall reservoir appearance. Due to limited data and extremely strong heterogeneity, the distribution of superior reservoir was hard to predict in a carbonate reservoir in the UAE, which often leads to empty wells and severe economic losses. With the help of neural network, a systematic study of sedimentation and diagenesis was carried out. The main controlling factors of reservoir properties were identified, and rock types were defined. Based on this, the facies model was established. The paper proposed a facies modeling method based on comprehensive sedimentary and diagenesis study assisted by neuro network. Firstly, sedimentary facies study was conducted to clarify the sedimentary background and reservoir geological architecture. Then, the diagenesis study was performed. In general, the main controlling factors for superior reservoir were deposition and cementation. Thirdly, rock type definition was established based on the knowledge of deposition, cementation and MICP data. Fourthly, the logs were processed, including quality control, normalization, etc. Then the sedimentary, diagenesis and rock type interpretation for the limited cored wells were propagated to the uncored wells using the neural network IPSOM method. The trend of deposition, cementation and rock type distribution were depicted. Finally, the rock type distribution was studied based on the trend of deposition and cementation, then the facies model was established. The study area was mainly proximal carbonate sedimentary environment, including the subtidal inner ramp and intertidal tidal flat. Diagenesis mainly included compaction, cementation, dissolution, and dolomitization. For the main reservoir, which was mainly the inner ramp deposits, SRT1 and SRT3 were defined as grain-dominated deposits with weak-medium cementation and medium-strong cementation, respectively. SRT5 was non-reservoir. For the secondary reservoir, it was intertidal environment, and mainly micrite dominated deposits. It only forms reservoir with dolomitization. SRT2 was defined as dolomitized micrite dominated deposits with medium-strong cementation, and the reservoir properties had been greatly improved. SRT4 was defined as weak-medium dolomitized micrite dominated deposits with strong cementation, with limited improvement in reservoir properties. SRT5 in this sedimentary environment was also defined as non-reservoir. With the neural network IPSOM method, the interpretation of deposition, cementation and rock type for the cored wells were propagated to non-cored wells and the trends were depicted. Then the rock type trend was finalized. Based on this, the facies model was established, and the cause of the dry wells was well explained. For this highly heterogeneous reservoir, the core analysis data was very limited. The cores were studied in detail, then the understandings were propagated to the uncored wells, which greatly enhanced the database. Based on this, a sedimentary facies model is established, which characterizes the distribution of high-quality reservoirs and provides a solid basis for oilfield development.