Joint history matching of seismic and production data

History matching is the process of building one or more sets of numerical models (representing a reservoir) which account for observed, measured data. The reason why we want to calibrate the subsurface model is that we want to use it to reliably predict future reservoir behaviour. 

Challenges

Reservoir properties are only truly known at well locations. In the geomodeling workflow, we use various geostatistic algorithms to populate models in-between and beyond wells. Standard history matching methods mostly match the models to production data or other data measured at wells, assuming that the reservoir models matched to these data shall produce a robust prediction. On production forecasting, this may be true if matched models produce equivalent flow behaviour measured at well locations.  However, if the objective is to draw definitive conclusions, such as identify the by-passed oil pockets in order to pin down the next infill target, it is not enough to have flow-equivalent models. Instead, the focus should be given to increasing the geological “trueness” of the models. 

Benefits

  • Unrivaled accuracy in locating the remaining oil enabling optimal selection of infill drilling targets, cost-effective production optimization and EOR activities
  • Reduce the cost and turnaround time of standard history matching projects by reducing the number of variables, and by allowing reservoir engineers to focus on the most important variables
  • Better decision-making through improved “Trueness” of the reservoir model and reduced uncertainties  (smaller ensemble) by seismic history matching

Promoters of sophisticated history matching packages often falsely claim that a particular method honours all reservoir data and accounting for all uncertainties.  However, the truth is even the latest sophisticated history matching methods do not completely honour seismic data. If one generates synthetic seismic cubes using the ensemble models matched with production data, it will not resemble the observed seismic data, suggesting the loop between the reservoir models and seismic not closed. 

 The discrepancies suggest low geological “Trueness” of the resulting ensemble models. As John Keynes here quoted quite a long time ago, “Accuracy is key to making a good quality decision.” In the context of geomodelling, this requires the interpretation to be both “Precise” (low uncertainty) and “True” (low bias). This is particularly true for the task of Locate the Remaining Oil in mature oilfields.

It is not enough to simply produce large arrays of models, we need to know which models should be carried forward to support definitive business decisions. This will also help reduce the number of ensemble models that are often impossible for the interpreter to manage effectively, and allow the decision makers to focus on what really matters. 

 

Approach

It is the common objective of all reservoir model updating projects to integrate all reservoir data (both static and dynamic). However, there is no single approach that is approximate for all circumstances. Depending on the quality of different types of reservoir data available, and the existing reservoir modelling strategy, there are many ways to condition reservoir models to seismic data while maintaining the matching with production data. 

Solution 1 – Manual updating

Despite an increasing number of sophisticated history matching tools emerging in recent years allowing users to generate a suite of realization rather than a single reservoir model in history-matching research. For many asset teams, updating the single reservoir model manually would still seem to be a sensible choice under practical conditions when taking into account the following factors:

  • Cost 
  • Time and effort taken to set up the sophisticated tool, train up the users, and examine the multiple scenarios and realizations
  • Lack of “Trueness” of the geostatistical reservoir models only matched to historic production data
  • Uncertainties are almost certainly under-estimated regardless of history matching methods
Joint product and seis

iRes-Geo leads the way in the development of software for joint production and seismic history matching. Closed-loop™ offers a suite of useful functionalities, such as co-visualization, data management and model updating functionalities, enabling asset teams to easily detect the problems with the reservoir model and make the necessary changes.

Approach

It is the common objective of all reservoir model updating projects to integrate all reservoir data (both static and dynamic). However, there is no single approach that is approximate for all circumstances. Depending on the quality of different types of reservoir data available, and the existing reservoir modelling strategy, there are many ways to condition reservoir models to seismic data while maintaining the matching with production data. 

Solution 1 – Manual updating

Despite an increasing number of sophisticated history matching tools emerging in recent years allowing users to generate a suite of realization rather than a single reservoir model in history-matching research. For many asset teams, updating the single reservoir model manually would still seem to be a sensible choice under practical conditions when taking into account the following factors:

  • Cost 
  • Time and effort taken to set up the sophisticated tool, train up the users, and examine the multiple scenarios and realizations
  • Lack of “Trueness” of the geostatistical reservoir models only matched to historic production data
  • Uncertainties are almost certainly under-estimated regardless of history matching methods
Joint product and seis

iRes-Geo leads the way in the development of software for joint production and seismic history matching. Closed-loop™ offers a suite of useful functionalities, such as co-visualization, data management and model updating functionalities, enabling asset teams to easily detect the problems with the reservoir model and make the necessary changes.

Solution 2 – Small-Loop approach

It is the common objective of all reservoir model updating projects to integrate all reservoir data (both static and dynamic). However, there is no single approach that is approximate for all circumstances. Depending on the quality of different types of reservoir data available, and the existing reservoir modelling strategy, there are many ways to utilize Closed-loop™ to condition reservoir models to seismic data. 

For all sorts of reasons – logistical, technical, management, contractual, it is not always easy for reservoir engineers to go back and adjust the geological model in a close cooperative environment. This approach is often referred to as small loop history matching. Under this circumstances, it is quite common for reservoir engineers to adjust the geological modelling without recoursing to the geologists by multiplying the porosity, the permeability, the anisotropy (kv/kh), the relative permeabilities, the well factors and many other parameters within their numerical world. Sometimes these factors can be large and global and probably outside the limits of the geological reality. When tied in with small-loop workflow, seismic Closed-loop™ is a good solution to reduce parameterization and ensure geological consistency of matched simulation models. 

 

Option 1Seismic Closed-loop™ + Small-loop history matching

Solution 2 – Small-Loop approach

For all sorts of reasons – logistical, technical, management, contractual, it is not always easy for reservoir engineers to go back and adjust the geological model in a close cooperative environment. This approach is often referred to as small loop history matching. Under this circumstances, it is quite common for reservoir engineers to adjust the geological modelling without recoursing to the geologists by multiplying the porosity, the permeability, the anisotropy (kv/kh), the relative permeabilities, the well factors and many other parameters within their numerical world. Sometimes these factors can be large and global and probably outside the limits of the geological reality. When tied in with small-loop workflow, seismic Closed-loop™ is a good solution to reduce parameterization and ensure geological consistency of matched simulation models. 

 

Option 1Seismic Closed-loop™ + Small-loop history matching

Solution 3 – Big loop approach

Some oil companies use big-loop, ensemble-based modelling workflows to generate vast arrays of models to explore the uncertainty space. This is to avoid having to make one single interpretation of an oilfield, with which the interpreter runs the risk of being “precisely wrong”.  However, the ensemble models are generated using highly probabilistic approaches, and without being conditioned to seismic data, the resulting ensembles can be both “imprecise” and “wrong”. 

iRes-Geo’s Closed-loop™ delivers the highest level of accuracy in simulator-to-seismic modelling with proprietary SmartGrid technology. Different simulation models can be hooked to Closed-loop™ platform.  Also, from Closed-loop™, the user can export updated simulation models and launch new simulation runs. iRes-Geo leads the way in its unique combined approach that allows for quick generation of synthetic seismic data for a large ensemble of reservoir models, and efficient model refinement if necessary. Also, Closed-loop™ offers a variety of methods to analyze the difference between observed and modelled seismic in both the seismic and the reservoir model domain.  

 

Option 2Seismic screening of the ensemble of matched reservoir simulation models

Different schemes made available by Closed-loop™ for mismatch evaluation

 

Solution 3 – Big loop approach

Some oil companies use big-loop, ensemble-based modelling workflows to generate vast arrays of models to explore the uncertainty space. This is to avoid having to make one single interpretation of an oilfield, with which the interpreter runs the risk of being “precisely wrong”.  However, the ensemble models are generated using highly probabilistic approaches, and without being conditioned to seismic data, the resulting ensembles can be both “imprecise” and “wrong”. 

iRes-Geo’s Closed-loop™ delivers the highest level of accuracy in simulator-to-seismic modelling with proprietary SmartGrid technology. Different simulation models can be hooked to Closed-loop™ platform.  Also, from Closed-loop™, the user can export updated simulation models and launch new simulation runs. iRes-Geo leads the way in its unique combined approach that allows for quick generation of synthetic seismic data for a large ensemble of reservoir models, and efficient model refinement if necessary. Also, Closed-loop™ offers a variety of methods to analyze the difference between observed and modelled seismic in both the seismic and the reservoir model domain.  

 

 

Option 2 – Seismic screening of the ensemble of matched reservoir simulation models