Real-Time Flow Stability Advisor

Abstract

An Aker BP operated oil field in the North Sea has occasionally experienced production flow instabilities in the production lines and risers. The oscillations in multiphase rates are kept within the process capacity limitation at the host installation typically by increasing backpressure (planned flaring is not allowed for on the Norwegian Continental Shelf). The heightened backpressure impacts the production potential of the field. The objective of the project described in this paper has been to develop and implement a new method for real-time production optimization providing an online assessment of slugging severity and suggested actions in order to mitigate slugging and increase production. The developed software tool has been validated using field data. A statistical approach based on the physical characteristics of the separator has been developed. A combination of transient multiphase flow simulations and data analysis has been employed in order to formulate the risk of exceeding separator constraints as a multidimensional function of the operational conditions. In order to generate a three-dimensional heat map of the risk related to the current state, operational data is continuously gathered from production sensors and transformed into pseudo-steady state values. A heat map is defined by a function where four relevant operational values can be selected. These values are: oil production rate, topside choke setting, gas lift rates and water cut. The software solution is run on a cloud infrastructure with an interactive web user interface. In a pilot program we have evaluated the ability of the stability advisor to continuously assess the severity of flow instabilities, identify measures to reduce the risk level and minimize associated production losses. The operator has identified valuable operational insights from the tool in a pilot program. The flow instabilities predicted by the model correlate well with observed data from the field. The tool is scalable to other fields with similar flow problems. Previous papers on slug flow prediction are in general conducted as offline study projects. There has been little success in making real-time scalable solutions available to continuous operations. This paper explains a method on how physical modelling of the flow system combined with statistical methods and access to real-time sensor data can provide a new approach for real-time slug flow prediction. The result demonstrates a scalable solution where output is presented in a format that can be applied by daily operations to act on and provide new and valuable production insights.