Seismic-driven Reservoir modeling

The incorporation of seismic is a crucial element in the construction of earth models. While seismic data are routinely used to define the size and shape of the reservoir, utilization of the seismic information to constrain the modelling of the inter-well reservoir heterogeneity is still less widespread. Nevertheless, the interest for constructing seismic-constrained 3D property models is growing rapidly, due in part to the greater availability of high-resolution 3D and 4D seismic data, and powerful pre-stack inversion techniques. 

Some of the most widely used workflows in the industry for the integration of seismic in geomodelling involve a sequential process where seismic data are first inverted to elastic properties. Inverted data are then depth-converted and transferred into the earth model framework, where the seismic attributes are used to guide the interpolation of reservoir properties between the wells. Geostatistics provides a number of tools for this purpose, such as co-kriging, kriging with external drift, stochastic simulation with seismic constraint, and geostatistical inversion. [1]

 

Evolution of Seismic Earth Modeling Technologies

 

The incorporation of seismic is a crucial element in the construction of earth models. While seismic data are routinely used to define the size and shape of the reservoir, utilization of the seismic information to constrain the modelling of the inter-well reservoir heterogeneity is still less widespread. Nevertheless, the interest for constructing seismic-constrained 3D property models is growing rapidly, due in part to the greater availability of high-resolution 3D and 4D seismic data, and powerful pre-stack inversion techniques. 

Some of the most widely used workflows in the industry for the integration of seismic in geomodelling involve a sequential process where seismic data are first inverted to elastic properties. Inverted data are then depth-converted and transferred into the earth model framework, where the seismic attributes are used to guide the interpolation of reservoir properties between the wells. Geostatistics provides a number of tools for this purpose, such as co-kriging, kriging with external drift, stochastic simulation with seismic constraint, and geostatistical inversion. [1]

Evolution of Seismic Earth Modeling Technologies

 

Multi-attribute Bayesian Sequential Gaussian Simulation

Challenges

Seismic data have a good lateral resolution but poor vertical resolution compared to wells. The good vertical resolution of the seismic data may warrant integrating it as a continuous vertical variable informing local reservoir properties, whereas poor resolution warrants using only a single map representing vertically averaged reservoir properties. The first case represents thick reservoirs and/or high-frequency seismic data in soft rock and is usually handled using a cokriging-type approach. The second case represents a thin bed, or interbedded sandstone reservoir equal or below the seismic resolution. Due to the point-to-point nature of the co-kriging method, it can not be effectively used to model the vertical heterogeneity of the reservoir. Many mature fields that have entered the middle and late phase of field development face the challenge to enhance oil recovery.  [2][4]

Solution

Behrens, et al (1998) proposed a new multiple map Bayesian technique with variable weights for the modelling of the thin reservoir where a single seismic map cannot effectively represent the entire reservoir. This new technique extends a previous Bayesian technique by incorporating two seismic property maps generated at the top and bottom of the reservoir and also allowing vertically varying weighting functions for each map. Huang et al. (2018) extended the technique by incorporating multiple attributes and the entire 3D seismic volumetric information in the Bayesian framework to improve the modelling of vertical reservoir heterogeneity.  This method has been tested on numerous mature fields characterized by thin interbedded sandstone reservoir widely spread in China with proven results. [3][4]

Benefits

  • On average 3-times higher vertical resolution of thin beds
  • Unmatched accuracy and certainty in modelling vertical reservoir heterogeneity 
  • Significantly improved “Trueness” of the reservoir model, enabling confident decision making for the redevelopment of mature fields consisting of alternating thin reservoirs

 

Collocated Co-kriging Vs. Bayesian Sequential Gaussian Simulation

 

Multi-attribute Bayesian Sequential Gaussian Simulation

Challenges

Seismic data have a good lateral resolution but poor vertical resolution compared to wells. The good vertical resolution of the seismic data may warrant integrating it as a continuous vertical variable informing local reservoir properties, whereas poor resolution warrants using only a single map representing vertically averaged reservoir properties. The first case represents thick reservoirs and/or high-frequency seismic data in soft rock and is usually handled using a cokriging-type approach. The second case represents a thin bed, or interbedded sandstone reservoir equal or below the seismic resolution. Due to the point-to-point nature of the co-kriging method, it can not be effectively used to model the vertical heterogeneity of the reservoir. Many mature fields that have entered the middle and late phase of field development face the challenge to enhance oil recovery.  [2][4]

Solution

Behrens, et al (1998) proposed a new multiple map Bayesian technique with variable weights for the modelling of the thin reservoir where a single seismic map cannot effectively represent the entire reservoir. This new technique extends a previous Bayesian technique by incorporating two seismic property maps generated at the top and bottom of the reservoir and also allowing vertically varying weighting functions for each map. Huang et al. (2018) extended the technique by incorporating multiple attributes and the entire 3D seismic volumetric information in the Bayesian framework to improve the modelling of vertical reservoir heterogeneity.  This method has been tested on numerous mature fields characterized by thin interbedded sandstone reservoir widely spread in China with proven results. [3][4]

Benefits

  • On average 3-times higher vertical resolution of thin beds
  • Unmatched accuracy and certainty in modeling vertical reservoir heterogeneity 
  • Significantly improved “Trueness” of the reservoir model, enabling confident decision making for redevelopment of mature fields consisting of alternating thin reservoirs

 

Collocated Co-kriging Vs. Bayesian Sequential Gaussian Simulation

 

Direct Seismic Inversion in Geomodel Grid

 

Challenges

Doyan (2007) sets out the remaining challenges for seismic integration into earth model, one of which is the lack of a method for accurate and efficient integration of seismic inversion results in the earth model framework. In traditional approaches, pre-stack elastic inversion is typically performed in specialist software packages. Inversion results must be depth converted and exported to geomodelling software for integration. This integration process is often tedious and time-consuming. More importantly, even when inversion is done in the stratigraphic grid, the grid geometry is different from that of the reservoir model and inversion results must still be depth-converted and resampled into the reservoir grid. Another matter that is less talked about is that to accurately position the inversion results into the reservoir model grids, an accurate and detailed velocity model of the reservoir layers is required. This information is often not readily available. These reasons partially explain the limited use of seismic information in constraining 3D property models.  [1]

Solution

iRes-Geo’s proprietary single pass data casting (SPDC) technology is designed for this purpose. SPDC performs geostatistical seismic inversion directly in depth and delivers estimates of elastic properties directly in the reservoir grid. Uniquely, SPDC simultaneously inverts for detailed reservoir velocity field and properties from the seismic data. The results have shown unmatched accuracy in characterizing the fine-scale reservoir details compared to standard stochastic and geostatistical inversion methods. 

Benefits

  • Easy, fast and cost-effective construction of High Accurate, High Precision (HAHP) geo-model conditioned to seismic
  • An inexpensive alternative of Full Wave Inversion (FWI). The proprietary dual inversion technology generates highly detailed velocity model in the reservoir section. 
  • Unrivalled accuracy in modelling fine-scale reservoir heterogeneity below seismic resolution

Direct seismic inversion in geo-model grid & resulting “high-resolution” velocity field from dual inversion

 

Direct Seismic Inversion in Geomodel Grid

 

Challenges

Doyan (2007) sets out the remaining challenges for seismic integration into earth model, one of which is the lack of a method for accurate and efficient integration of seismic inversion results in the earth model framework. In traditional approaches, pre-stack elastic inversion is typically performed in specialist software packages. Inversion results must be depth converted and exported to geomodelling software for integration. This integration process is often tedious and time-consuming. More importantly, even when inversion is done in the stratigraphic grid, the grid geometry is different from that of the reservoir model and inversion results must still be depth-converted and resampled into the reservoir grid. Another matter that is less talked about is that to accurately position the inversion results into the reservoir model grids, an accurate and detailed velocity model of the reservoir layers is required. This information is often not readily available. These reasons partially explain the limited use of seismic information in constraining 3D property models.  [1]

Solution

iRes-Geo’s proprietary single pass data casting (SPDC) technology is designed for this purpose. SPDC performs geostatistical seismic inversion directly in depth and delivers estimates of elastic properties directly in the reservoir grid. Uniquely, SPDC simultaneously inverts for detailed reservoir velocity field and properties from the seismic data. The results have shown unmatched accuracy in characterizing the fine-scale reservoir details compared to standard stochastic and geostatistical inversion methods. 

Benefits

  • Easy, fast and cost-effective construction of High Accurate, High Precision (HAHP) geo-model conditioned to seismic
  • An inexpensive alternative of Full Wave Inversion (FWI). The proprietary dual inversion technology generates highly detailed velocity model in the reservoir section. 
  • Unrivalled accuracy in modelling fine-scale reservoir heterogeneity below seismic resolution

Direct seismic inversion in geo-model grid & resulting “high-resolution” velocity field from dual inversion

 

Applications

  • Enhance production from the thin inter-bedded reservoir 
  • Improve reservoir management of the mature field with production controlled by small-scale reservoir heterogeneities (thief zone, high-permeability conduit, etc)
  • Accurate identification of infill drilling targets
  • Make informed decision to better optimize production 
  • Integrated evaluation of Enhanced Oil Recovery (EOR) projects
Reference

[1] Seismic Reservoir Characterization: An Earth Modelling Perspective (EET 2), By Philippe Doyen, EAGE, 2007

[2]Incorporating Seismic Data of Intermediate Vertical Resolution Into 3D Reservoir Models, By Behrens R., et al, SPE, 1998

[3] Well and seismic joint velocity modeling based on the Bayesian sequential Gaussian simulation, K. Li,  Y. Hu, and X. Huang, Reservoir Geophysics, SEG, 2018

[4] Yin, X., He, W. and Huang, X., 2005,Bayesian sequential Gaussian simulation methodJournal of China University of Petroleum(Edition of Natural Science), 2005, 05, 28-32.