Cgg Services Sas

Marathon Oil Company

Compagnie Generale De Geophysique

Raphael Bornard

Yves Lafet

Benjamin Roure

Pierre Lanfranchi

Jeffry G Hamman

Donald H Caldwell

Laurene Michou

Thierry Coléou

Thierry Crozat

Bernard Deschizeaux

Amélie Rodrigue Molle

Fabien Allo

Properties of underground formations are obtained by performing simultaneous geostatistical inversion of two or more seismic datasets acquired over the same area. These methods enable simultaneous estimating quantitative changes in a hydrocarbon-producing field.

Method and apparatus for simultaneous geostatistical inversion of time-lapse seismic data

Global inversion of multi-vintage seismic data uses simulated annealing to minimize a cost function simultaneously for all vintages and all angle stacks to yield values of geophysical properties. Each vintage is generated from an independent seismic survey of a subsurface structure conducted over a distinct period of time and includes seismic traces and angle stacks. An initial model of the subsurface structure is used and includes values for geophysical properties and time shift maps between vintages. The time shift map contains shifts in the seismic trace between vintages. The cost function includes a time shift map term for the difference between the time shift map and a calculated time shift of the seismic trace between vintages and is based on a proposed perturbation to at least one of the geophysical properties. The time shift map is also used as a global constraint on proposed perturbations of subsurface properties.

Time-lapse simultaneous inversion of amplitudes and time shifts constrained by pre-computed input maps

A method is provided for predicting a value of a designated rock or fluid property in a subterranean geologic volume. A first predicted value of the designated rock or fluid property is also assigned to a volume of a multi-dimensional, multi-scale model. A first predicted value of seismic response for the model volume is calculated from a response model using the first predicted value of the designated rock or fluid property, wherein the response model is responsive to changes in predicted values of the designated rock or fluid property. A synthetic trace is generated and iteratively compared to the corresponding trace obtained from one or more sets of actual seismic data to determine a difference while consistency is maintained between the types, scales and dimensions of values and data. The first predicted value of the designated rock or fluid property is adjusted in response to the difference to create a second predicted value of the designated rock or fluid property, wherein the second predicted value generally reduces the difference.

Method for predicting quantitative values of a rock or fluid property in a reservoir using seismic data

Growing community of inventors

Leiden, Netherlands

Raphael Bornard