Preface:

To make sound investment decisions, stakeholders need accurate estimates of the uncertainties present in forecasts of reservoir production and reserves estimation. Traditionally, reservoir studies were not integrated, and dealt with in silos. Each discipline worked on its own separately and projects often followed a linear approach: geophysics, geology, and then reservoir engineering. Deterministic approaches were generally used to conduct an evaluation due to timescales, but it is accepted that they have a limited confidence in quantifying the ranges of uncertainty involved due to honoring of critical data and exploring the full range of uncertainty.
In this project, the probabilistic approach allowed for extensive review of profiles, whereby more reliable P90-P50-P10 scenarios could be established. Based on this investigation, in combination with an economical evaluation, a criterion was developed in order to understand the ‘true’ potential of an exploration or appraisal block, providing a robust methodology to support critical decisions that can be subsequently replicated in similar future studies.
This study focuses on the Cygnus Field in the Southern North Sea, operated by Neptune Energy and Spirit Energy. Cygnus is the UK’s largest single producing gas field and exports around 250 million standard cubic feet of gas every day. This field has been producing gas since 2016 and has a field life of over 20 years.

Workflow

The first stage of the project was to import three pre-existing static models into tNavigator. The structures were built based on different surfaces, facies and sediment distribution. In order to optimise run time, a sector model was extracted including only fault blocks south of Cygnus for this study. By using the sector model, the total active blocks was reduced from 1.16MM to 120-140K depending on the grid realisation, which in turn optimised the computational run time from 2.4 hours to 20 minutes.

Following extraction of the sector model, an uncertainty project was set up to evaluate the following parameters and their impact on the profile, namely: geological realisations accounting for different structural interpretations, horizontal and vertical permeability along with porosity variation, fault transmissibility, aquifer size and fluid contact.

In order to include the structural uncertainty, DEFINES option was used to define GRIDNUM 1 (low structure), 2 (base structure) and 3 (high structure) representing different deterministic cases. Using this approach meant it was possible to include as many structural realisations as needed. The remaining variables for the uncertainty study were also created under DEFINES option. Arithmetic statements were used to define how the hydrodynamic model would be impacted during the study.

Based on these parameters and their ranges, two experimental designs were created:

• Sensitivity Analysis: to investigate the model response to selected ranges of uncertainty (Tornado)
• Latin Hypercube: to determine P10–P50–P90 profiles pre and post MEFS truncation and determine criteria for economic success.

Using results from the Tornado study, critical variables with the biggest impact on project results were identified. Both permeability and fluid contact were proven to have a significant effect on initial rates and total gas production, whilst porosity, structural realisation and fluid contact were responsible for major changes in the in-place volumes. Additionally, the Tornado study was also used to understand if the range used on the variables was representing the uncertainty correctly.

After evaluating the Tornado study an experimental design (Latin Hypercube experiment) with 150 cases was created . By evaluating all cases and cumulative distribution function for the project, a representative P10-P50-P90 set was selected. Latin hypercube sampling method allowed to extensively yet efficiently explore the parameter search space for each uncertainty variable. Using the reservoir coupling functionality in tNavigator, these models were connected to the full field model and true incremental production associated with the project was calculated, taking into account the effect of back pressures in the production network.

The incremental production profile was used in the economical model to calculate the NPV of project. Based on this study it was decided that the main parameter impacting the project was gas water contact . An specific GWC was required in order to have an successful appraisal program and move to development.

Outcome:

By using the probabilistic approach to production and reserves estimation, a realistic P10-P50-P90 was generated, leading to a more robust analysis and evaluation during the appraisal process. By using an integrated approach and Assisted History Matching in tNavigator, it was possible to run 150 realisations in a short time frame, which allowed the project to evaluate a much wider range of uncertainty than would have been achievable when using the previous deterministic approach.