Professional Documents
Culture Documents
ENG TrendModeling 4Q13
ENG TrendModeling 4Q13
Industry challenge
Trends are ubiquitous in reservoir facies and petrophysical properties. Transport patterns of depositional environments determine different Facies
associations and successions. Vertical profiles as well as the extent and direction of continuity of petrophysical properties such as porosity and
permeability also exhibit large-scale changes associated to diagenetic (i.e. compaction, cementation) and depositional geological processes.
Accounting for these trends is paramount in order to obtain reliable predictive models of facies and petrophysical properties and help reduce the
inherent uncertainty.
Many tools and algorithms have been developed in geostatistical modeling to account for systematic deterministic components in the spatial
distribution of discrete and continuous properties. Several methodologies are currently available in Petrel property modeling, including Co-kriging
with locally varying mean and Collocated Co-kriging. However, analyzing the available data to evaluate and ultimately build reliable trend models
remains a big challenge in geostatistical modeling.
If there is strong evidence of a trend in the available data that is consistent and geologically meaningful, and if, and only if, enough good quality and
consistent data is available to build the trend model (e.g. log data, seismic attributes, geological information), then a trend model can explicitly use
in the geostatistical method. If, on the other hand, you do not have strong evidence of a trend that is consistent and geologically meaningful or not
enough data is available to build a reliable trend model, then it is better to keep it simple and not model an explicit trend.
Fig 2. Optimal combination of the input trend models. The Estimate trend weights functionality perform a multivariate linear regression.
Fig 4. Different settings and input variables can be used on the different regions or facies codes.
Fig 5. The influence of each individual input trend of the output trend model can be assessed by changing the trend weights
Fig 6. Using the resulting optimal trend model as an input for the collocated co-kriging algorithm