Production optimization involves subsurface understanding and flow optimization from downhole to surface facilities. It is a continuous process that requires a representative, easy-to-use, automated, repeatable, and updatable Well & Surface Network Modelling (W&SNM) workflow. This paper presents a novel W&SNM process for a giant brown oil field that demonstrates the power of data science utilized by petroleum engineering experts in maximizing production.
Python scripts are used to streamline a sophisticated W&SNM process for a mature North African oil field with over 300 wells and 60 years of production. The process involves preparing static and time-variable resources, including well-test and production databases, well categories and corresponding PROSPER templates, and a GAP network model. A PROSPER file is automatically created per well updating its IPR and VLP curves with the latest data, connecting it to its corresponding node in GAP and calculating its uptime. Wells with no production data are automatically inactivated. The resulting W&SNM is calibrated and optimized for the integrated production system.
The methodology presented produces an integrated model of well and surface networks which can be prepared in significantly less time than what a petroleum engineer would typically require.
Engineers are notified of any deviations from prior conditions via automated updates of the well models. This can indicate changes in the productivity index, reservoir pressure reduction, or a suboptimal artificial lift setup. Through automation scripts, hundreds of wells’ flowing bottomhole pressures can be estimated within minutes, thereby updating the well PI accordingly. The artificial lift design and optimization processes are executed with ease.
Automated well design simplifies development planning by enabling easy investigation of various types of future well completions and artificial lift. The process includes identifying workover candidates and reactivating them as required. By coupling the surface network with the reservoir model, a correct material balance is ensured by using optimized rates and pressures from GAP.
In this regard, hundreds of well models are updated based on reservoir pressures, water-cut, GOR, and productivity indices, and fed into the surface model for further calculations. Python optimization libraries are utilized to readjust the friction coefficient in flow correlations, enabling history-matching of pressure drops on hundreds of surface pipelines in a few minutes.
Well and Surface Network Modeling involves multiple manual and repetitive workflows, which not only consume time but may also be prone to errors. To address this issue, a new script-guided workflow is proposed, which minimizes human intervention in W&SNM. This approach enables engineers to streamline their modelling work and focus on more critical tasks. The process can be effortlessly replicated across various datasets, making it highly appealing for operators with multiple assets.
