Many industries, such as oil and gas, CO2 storage, geothermal or energy storage, use wells, not only for injection and production, but also for monitoring to ensure the quality and safety of the processes that take place in the reservoirs in the subsoil.
The possibilities for extracting data from the wells, such as flow rates, pressure, temperature etc., have been greatly improved in recent years. The signals can now be measured with high precision and stored in almost unlimited quantities. What has not developed similarly are the possibilities to efficiently analyse and interpret these data so that the inherent information becomes visible and can be used for monitoring and managing the wells in the most appropriate way.
Until now, such analyses have mainly been done manually, which is both time-consuming and labour-intensive, and often means that important information is overlooked, or arrives too late. This project aims to develop methods, models and algorithms, to effectively handle these large amounts of data, eliminate noise, and recognize patterns that may contain useful information. Potentially useful data will be passed on to advanced interpretation systems which can extract such information by combining traditional analysis methods (analytical and numerical models) with more data-driven approaches such as machine learning and big-data analytics.
The goal is an automated process where the data stream is processed continuously. In this way, abnormal behaviour, such as leaks, can be detected early, and important information about well and reservoir conditions can quickly reach decision makers.
The outcome of this project can be of great importance, not only for the costs and results of the relevant underground activity, but also for the environment and climate through less pollution, better energy utilization and lower greenhouse gas emissions.
Well surveillance and control remain key issues for industries like oil and gas (O&G), geothermal, geological carbon storage (CCS) and compressed air energy storage (CAES). Significant progress with real-time well measurements and automated data gathering has recently been achieved, while interpretation and well control optimization remain labour and time-consuming work mainly done manually.
The project aims at development and testing of a new methodology and tool (prototype software) for automated injection well monitoring and control based on real-time pressure, temperature and rate data. The basis is time-lapse Pressure Transient Analysis (PTA) providing capabilities to monitor well performance and interference, containment of injected fluids and safety of abandoned wells. Modern gauges provide high frequency and resolution data, however some challenges are unresolved. The project will focus on these challenges including recognition of well and reservoir response from noisy / biased data, interpretation of low-amplitude changes and combination with other transients like temperature. A separate issue is real-time data interpretation needed to optimize well control.
Advanced interpretation and automation will be achieved via combining model- (time-lapse PTA) and data- (big data analytics / machine learning) driven approaches. Real field data will guide the project tasks and priorities and serve as a testing and application environment for the research and development. Automated well monitoring and warning system will be integrated with injection well control. The control includes rate changes designed to send signals with real-time interpretation of the response (on-the-fly well testing) to optimize well performance and inter-well communication and prevent injection safety issues.
The automated solution developed in the project will reduce costs, improve efficiency and minimize environmental footprint in O&G and facilitate CCS, CAES and geothermal industries.