Matching of Production History and 4D Seismic Data—Application to the Girassol Field, Offshore Angola

Abstract

fax 01-972-952-9435. AbstractThis paper presents an advanced history matching methodology for constraining 3D stochastic reservoir models to both production history and 4D seismic attributes. The proposed approach is based on an optimization loop which integrates geostatistical modeling, upscaling, fluid flow simulation and petro-elastic modeling in the same workflow. Simulated production history and 4D seismic attributes are compared to real data using a single objective function, which is minimized using a new optimization algorithm based on response surface fitting.Inversion parameters can be selected throughout the entire modeling workflow, to update the facies spatial distribution, the petrophysical properties and fluid flow parameters. Advanced parameterization techniques are used to constrain the fine scale geostatistical model. The gradual deformation method allows smooth transformations of the facies model realization while conserving the overall statistical characteristics. In addition, a new history matching technique has been implemented to constrain the average facies proportions over given regions. As an example, new geologically-based parameters like shale proportions in flow barriers or sand proportions in channel sequences can be considered in the inversion process.A successful application to the Girassol field is presented. Girassol is a large, complex and faulted turbidite field located offshore Angola. A detailed geostatistical model was built first from 3D high resolution seismic data. The major challenge was to integrate 4D seismic data to constrain the reservoir model in a quantitative way.The gradual deformation method has shown a major impact on production history, since the changes in the facies distribution affect the connectivity in sandy channels and the position of shaly barriers. A good match of water and gas breakthrough was obtained over one year of production data. The addition of new geological constraints allowed us to better honor the 4D seismic information (extent of gas injection and of depleted zones). This was achieved by constraining average sand and shale proportions over channel and levee regions. Improved predictions of future production and of a new seismic survey have been obtained using the matched model.

Geologic and seismic modeling of a coarse-grained deep-water channel reservoir analog (Black’s Beach, La Jolla, California)

“…As the net-to-gross ratio of the fill decreases, the control on fluid flow can be significant. Gonzalez-Carballo et al (2006) and Roggero et al (2007) have analyzed 4-D seismic data to demonstrate movement of injected water being strongly controlled by the position of such a mudstone-filled channel plug. Of course, the impact of this fine-grained channel fill is dependent on the 3-D geometry of the erosive channel and the underlying strata into which it is eroding.…”

Seismic History Matching of Nelson Using Time-Lapse Seismic Data: An Investigation of 4D Signature Normalization

“…The choice of parameterization is always important in history matching, and there are many other methods we could have chosen such as gradzone (Gosselin et al 2001), gradual deformation (Roggero et al 2007), probability perturbation (Castro 2007), discrete cosine methods (Jafarpour and McLaughlin 2009), and ensemble Kalman filter (Skjervheim et al 2007). Many of these methods alter the reservoir properties in a more global sense, and the outcome may have been different.…”

“…These involve computation of a misfit between predicted and observed data, and then a mathematical routine is used to choose new models either stochastically or deterministically. Many researchers have investigated different inversion algorithms, parameterization schemes, and appropriate methods to compare data (Huang et al 1997;Gosselin et al 2001;Aanonsen et al 2001;Arenas et al 2003;Lygren et al 2003;Kretz et al 2002;Mantica et al 2001;Vasco et al 2003;Portella and Emerick 2005;Jin et al 2007;Roggero et al 2007;Skjervheim et al 2007). Often, observed acoustic-impedance changes, which may be of undefined units, are compared to equivalent synthetic seismic data (Guerillot and Pianelo 2000;Clifford et al 2003;Mezghani et al 2004;Emerick et al 2007).…”

Seismic Assisted History Matching Using Binary Image Matching

“…d. As we are aware that amplitude decrease (softening) in the 4D seismic data is a consequence of gas, as well as pressure increase (Calvert et al, 2014), the amplitude decrease caused by an increase in pressure around a water injection well is removed from the analysis; however in the case of a gas injector well (where an increase in pressure and the presence of gas cause the same softening effect on seismic), the magnitude of the pressure and gas saturation will need to be determined in order to ascertain which has a more dominant effect on the seismic. To perform the history match, an initial ensemble of models is created using the Latin Hypercube Experimental Design (LHED) method (Roggero et al, 2007), as multiple models have extensive coverage of the search space and deliver robust results. The initial input parameter sampling is important, and is usually carried out using experimental design methods, such as Plackett-Burman, LHED or Factorial Design (Schulze-Riegert andGhedan, 2007, Zubarev, 2009).…”