Abstract:
Oil production optimization has always been a challenging function in the petroleum industry. The search for effective production planning tools is an ongoing goal of many companies who are involved directly or indirectly in oil production.
The need for this research is to address a business problem related to the difference between the targeted production’s theoretical capacity and what is actually produced in large oil fields. The business problem is known as lost production opportunity which is defined as the difference between the reservoir system and the surface facility’s capabilities. This loss has impact on the profit as well as long term development plans for enhanced oil recovery. The objective for oil companies is to minimise this loss which is estimated at 15% in ADMA-OPCO (Abu Dhabi Marine Operation Company) and other companies (ref. Table 1-1 Lost production case studies).
In order to identify the reasons behind the loss in production opportunity, it is important to identify the key decision variables that contribute to production and include them in the production model. To achieve this goal, a survey with subject matter experts was conducted. The results of the survey identified a number of random variables which were not addressed in the current models or handled with specific assumptions. Examples of these assumptions are a steady well production or lift curve model which is subject to changes, the use of asset availability program in the absence of probabilities of failure, the use of constant separation models regardless of change in fluid characteristics, and ignoring stimulation activities which improves the well performance. A data collection and analysis was conducted on the defined decision variables. The information related to process flow diagrams and capacities were used to design and configure the simulation model parameters. The information related to production and operational activities were analysed for trending of polynomial or probability functions, in the case of random inferencing, and used to create material flow in the simulator.
