The concept of uncertainty, risk, and probabilistic assessment is increasingly employed as a standard in the E&P industry to assist in optimum development and investment decisions. The studied Onshore Abu Dhabi field is a Cretaceous complex carbonate oil producing reservoir, which has more than 15 years of production history. This paper discusses an integrated static and dynamic workflow to create a range of probabilistic simulation models to forecast oil production under several production schemes. The study deals with quantitative identification and ranking of factors affecting volumetric and reserves uncertainty in the field. In order to quantify the uncertainties, the main uncertainty parameters and their respective ranges were first identified, selected, and analyzed using Experimental Design to generate a tornado plot which enables the selection of the most influential parameters on the objective function. Secondly was to build a Proxy Model that would help in defining the full probabilistic volumetric distribution on the stock tank oil initial in place (STOIIP) and Recovery Factor (RF). Five main static uncertainty parameters were selected to assess the STOIIP distribution namely structure, free water level, saturation height function, porosity, and formation volume factor. In addition, four dynamic uncertainty parameters were incorporated for reserves estimation specifically Sorw, Kv/Kh, relative permeability, and subdense layer communication. A cumulative distribution function was created in order to extract the probability cases of P10, P50 and P90 of the STOIIP. The simulation models were then built using the P50 volumetric case derived from the static model that was run with hundreds of realizations. Combinations of dynamic uncertainty parameters were simulated using Monte Carlo to define the Low, Base, and High Cases. This was done by comparison with material balance computations and streamline simulation. A stochastic combination of the STOIIP distribution and the RF sensitivities was done through an Experimental Design, Proxy Model, and Monte Carlo approaches. The Base Case model history-match was checked against the choice of parameters defining the Low and High sensitivity cases. The match data available included: oil rates, water cuts, GOR, WHP, flowing and static Pressure, and saturation profiles derived from open and/or cased hole logs. The sensitivity assessment showed that using currently available data, the two major factors affecting the volumetric uncertainty are the free water level and structure. In contrast, porosity possesses the smallest impact. In addition, Kv/Kh and relative permeability are the two main parameters affecting the RF. A number of appraisal wells will be drilled to reduce the structure uncertainty specifically in the flank areas, which will lead to further maturation of reserves. Economic calculations were performed to check that all projects pertaining to the reserves category would consider oil price, CAPEX profile, OPEX profile, well and facility life time.

The distribution of reservoir quality in tight carbonates depends primarily upon how diagenetic processes have modified the rock microstructure, leading to significant heterogeneity and anisotropy. The size and connectivity of the pore network may be enhanced by dissolution or reduced by cementation and compaction. Consequently, a clear understanding of the diagenetic process that responsible for the reservoir tightness would offer vital assurance on the spatial property distribution and future field development plan. In this paper, we have examined the factors which affect the distribution of porosity, permeability and reservoir quality in the Thamama Group, which is a prospective low permeability carbonate reservoir rock in Onshore Abu Dhabi. The dataset includes regional stratigraphy, well logs and core material from a number of wells, a suite of laboratory petrophysical measurements, seismic attributes, geomechanics, fracture study, and production history. Dataset analysis and interpretation suggested that the reservoir was deposited in shallow to deep marine low energy environment which led to deposition of fine to very fine grains (lime-mud supported) types of sediments. This, in turn, would produce poor reservoirs during compaction and finally leads to tightness. Because of the low permeability nature of this tight reservoir, it is quite challenging to obtain their complete reservoir properties and dynamic behavior. As in many other tight reservoir projects, a considerable area of the reservoir must be effectively stimulated during the hydraulic fracturing process to achieve economic productivity. In addition, development of tight reservoirs often faces challenges, for example, low initial production rates and high declining rate. This paper aims to frame all possible optimum development practices for tight reservoir in the studied field that should be considered for future development plan. We also investigated the application of new technology to enhance the poor oil recovery within the pool including horizontal drilling and multi-stage fracture completion technology. Furthermore, this paper also discusses well orientation relative to the far field principal stresses, hydraulic fractures treatment, fracture fluid selection, and nano-technology application. This, in turn, would provide valuable information on how to optimally develop this previously considered marginal and uneconomic reservoir.

The purpose of this paper is to communicate the experiences in the development of an innovative concept named "ASK Thamama" as an automated data and information retrieval engine driven by artificial intelligence techniques including text analytics and natural language processing. ASK is an AI enabled conversational search engine used to retrieve information from various internal data repositories using natural language queries. The text processing and conversational engine concept is built upon available open-source software requiring minimum coding of new libraries. A data set with 1000 documents was used to validate key functionalities with an accuracy of 90% of the search queries and able to provide specific answers for 80% of queries framed as questions. The results of this work show encouraging results and demonstrate value that AI-enabled methodologies can provide natural language search by enabling automated workflows for data information retrieval. The developed AI methodology has tremendous potential of integration in an end-to-end workflow of knowledge management by utilizing available document repositories to valuable insights, with little to no human intervention.