Sleipner 3layers

Benchmarking VE simulation methods against seismic history at Sleipner

What we do

In this project, we will perform history matching of the Sleipner plume data. The VESA simulator, an in-house simulator at NORCE that has been developed over many years will be used for this purpose. The code can be run from a Python script for easy implementation with ensemble-based data assimilation toolkit described above. Three main uncertainties will be explored are:

U1. Vertical transmissibility of migration pathways (permeability of “holes” in intraformational shales that allow CO2 to migrate vertically in the system).

U2. CO2 uptake in water column due to convective mixing (mass flux of CO2 into water per area of plume within range predicted by existing theory). Mass flux is a static input parameter

U3. Reservoir temperature due to variation in geothermal gradient (range in temperature within 2 degrees C from measured values). Reservoir temperature is a static input parameter that varies vertically according to a defined geothermal gradient. Plume temperature is modeled using an energy equation coupled to the VESA flow simulator.

The first uncertainty is the primary quantity of interest that will mainly control how the plume migrates vertically from Layer 1 to 9, while dissolution and reservoir temperature are secondary controls on migration. Therefore, most of the project focus will be on U1 followed by U2. If time allows, we will investigate reservoir temperature.

Why is this important?

VE methods have been shown to simulate Utsira plume with high accuracy at significantly reduced computational cost. Reduced computational times implies that a larger parameter can be sampled and additional processes can be tested, potentially leading to a more reliable estimate of the geological model.

The addition of convective mixing in the VE model enables an additional component to the history match that is missing in standard commercial simulation. The eventual history matched value of the strength of convective dissolution will give solid evidence for the role convection has to play at the field scale in real systems.


The project will be carried out in 4 stages:
M-1. Preparation of stochastic models, i.e. process benchmark data and prepare stochastically generated models sampled from above uncertainties to be run within the ensemble-based method.

M-2. Testing of history matching workflow: We take the plume outline as a starting point for history matching. The precise measure for goodness-of-fit will be determined in this phase to ensure the workflow can be run efficiently and robustly.

M-3. History matching to U1 vertical transmissibility: The tested workflow in M-2 will be applied to the set of stochastic models in M-1 where the vertical permeability of the “holes” will be the history matching parameter.

M-4. Joint with secondary parameters U2 and U3: the results of M-3 will be augmented with secondary parameters: CO2-uptake and reservoir temperature. First, we run the history matching joint between U1 and U2 uncertainties. The workflow with U3 (temperature) will only be carried out if time allows.