Skip to content

Building 3D Reservoir Models to Analyse the Impact of Key Uncertainties on Volumetric Ranges

Hyde, Jonathon

Engineering Honours Degree 2007

University of Adelaide


The aim of this project was to build 3D geological models of a recently discovered offshore oilfield and nearby prospect in order to analyse the impact of key reservoir uncertainties on volumetrics. The objectives  evolved during the study due to the concurrent drilling (by Well C) and failure of the prospect. Potential volumetric upsides of the discovered field were assessed to assist in determining the possibility of a stand alone development. The project was undertaken through the perspective of the field operator, Santos Ltd, who requested that some information associated with the report remain confidential.

The oilfield is located in the Dampier Sub-basin of the Carnarvon Basin, Western Australia, approximately 130km north-northeast of Dampier. Oil is trapped in two sands within the Calypso Formation, Callovian (or Middle Jurassic) age. The formation represents the transgressive flooding of the underlying fluvio-deltaic sediments of the Legendre Formation, which are of Toarcian to Callovian age. The lower can be further separated into two zones: the lowest zone being a storm dominated zone of shelfal origin while the higher zone is an amalgamation of tidally influenced mouth-bars. The upper of the two oil bearing sands is similar in nature to the upper-zone in the lower sand but of much poorer quality. The Calypso formation is subcropped by a basin-wide Oxfordian age unconformity, above which are Lower Cretaceous shales that provide the ultimate top and lateral seal.

Initial stages of the project involved understanding key uncertainties and how they would be managed. This was assisted by undertaking an @Risk simulation where the reservoir parameters provided by Santos were utilised to estimate a range of hydrocarbons-initially-in-place (HCIIP) and reserves. A calculated mean total reserves estimate of 7.2MMbbl was established with a P90 of 3.9MMbbl and a P10 of 11.4MMbbl. From the generated reserves distribution it was determined that there is a 7.92% probability that the oilfield contains greater volumes than the minimum economic pool size (MEPS), 12MMbbl. The reserves distribution is heavily weighted to the lowside of the input parameter distributions due to lognormal distribution types being utilised, for all parameters, in the simulation.

A sensitivity analysis was carried out on the HCIIP and reserve distributions and it was established that the recovery factor (RF) of the lower reservoir is the most uncertain parameter for reserves, followed by the gross-rock-volumes of the lower and upper sand respectively. Gross-rock volume (GRV) is the key uncertainty in the estimation of HCIIP.

Building the geological models involved the generation of probabilistic distributions of facies and reservoir properties. This assisted in the regional geological understanding of the play and also allowed an estimation of the volume of hydrocarbons-in-place. Through further analysis; such as model up-scaling, dynamic simulation and comparisons between development techniques, the economic viability of the discovery can be further assessed.

The 3D modelling software utilised in this project contains methods where probabilistic distributions can be assigned to properties or parameters, accounting for uncertainty in their values. Three models were built; a model based on depth grids prior to the drilling of the nearby prospect, a model based on depth grids tied to the nearby prospect once drilled and a model based on grids provided by Santos which were updated from Well C.

Data incorporated in the Petrel Project included: seismic data and interpretations, well data, well logs – MWD, LWD and wireline, and core data. Further data was obtained during the project from the drilling of the nearby prospect.

Depth maps were generated for the formation of interest by converting seismic interpretation (time) to depth. This was achieved through the multiplication of time grids with corresponding velocity maps and tying the resulting depth maps to well markers.

The method by which the project was undertaken followed the workflow described in ‘Petrel Workflow Tools: Petrel Introduction Course’, published by Schlumberger.
  Placement and correlation of well markers was undertaken by the analysis of mudlogs, well logs, well evaluation summary plots and wellbore data. Correct placement of well markers was critical as they were used in defining horizons, zones and layers later in the workflow.
  The 3D model boundary contains both the discovered oilfield and the identified prospect. The I and J grid directions were SE-NW and SW-NE respectively and their dimensions were approximately 17km by 4km respectively. The depth interval of the model was 300m.
  The stratigraphic organisation of the region of interest required the generation of six zones; three of which were shale zones and three reservoir zones. The thickness of layering within the sand dominated layers was governed by the need to capture fine scale detail. Well logs, petrophysical analysis, and the core description were used to conclude that the layers were to have a thickness of 0.5metres. Well logs of interest; facies, porosity, permeability, and water saturation, were up-scaled to this thickness.
  Facies modelling involved the propagation of well-log facies throughout the zones. In the sand dominated layers, where multiple facies were present, this was achieved by using Sequential Indicator Simulation with a spherical variogram, anisotropy ranges recommended by the geologist who completed the core description and global fractions from the up-scaled cell proportions. In the shale zones propagation was via assigning values of the shale facies only.
  Property modelling was completed in a similar manner to the facies modelling whereby the sand zones were populated with porosity values by Sequential Gaussian Simulation with a spherical variogram and the same anisotropy ranges used in the facies modelling. From this point the porosity within distributed shale facies (including the shale zones) was classed as undefined.

Three Petrel models were produced and these generated a variety of reserves estimations. A Prior to Well C Model produced Total Reserves of 11.5MMbbl, a Well C Updated Model calculated Total Reserves of 15.2MMbbl and a Santos Grids Model computed Total Reserves of 14.0MMbbl. These values are all significantly larger than the volumes estimated by the initial @Risk simulation. The @Risk volumes are more likely as the structural models provided to input to Petrel were generally on the high side of the distribution.

In the models, upside potential exists in the form of larger GRV, net-to-gross (NTG) and RF and from a near-field opportunity, to the south-east of the discovered field (adjacent to the Well A structure), bounded to the south by the Rosemary Fault Zone. Maximum reserves, including all upside potential, from the Santos Grids Model (most likely), were calculated at 24.3MMbbl. The near-field opportunity is regarded as extremely high risk as its difficult seismic interpretation, associated with its near-fault location, and its juxtaposition against the stratigraphically lower Legendre sands give little confidence of the structure containing hydrocarbons.

If further confidence can be attributed to the upside volumes calculated in this project (up to 24.3MMbbl total) then there may be the opportunity for a stand-alone development, however, further technical and economic evaluation will be required.

Australian School of Petroleum



T: +61 8 8313 8000
F: +61 8 8313 8030