With a streamline simulator it is possible to create a surveillance model or a flow simulation model.  To summarize, a surveillance model produces streamlines and well-rate allocation factors at each timestep, but there is no transport step taken to move fluids along the streamlines.  With a flow simulation model, similar to classic simulation, streamlines are produced, but then a transport step which moves fluids along streamlines occurs.

  • Surveillance models are useful for determining well-pair interactions based on historical voidage rates.  These are light models that are easy to understand, compute, and require minimal data yet can yield the all-important well-rate allocation factors and injector pattern efficiency plots.  The allocation factors are based on historical flowrates, well locations, and whatever level of geology has been put in the model.  Meaning they are a huge step forward from geometric-based allocation factors. Furthermore, through the application of dynamic pattern material balance calculations to the historical injection/production, insitu saturations can also be estimated from an OOIP distribution. However a surveillance model cannot be used for forecasting as in a traditional flow simulation model.
  • Flow Simulation models also yield all the information of a surveillance model and more.  Because of the transport step additional information such as relperms, fluid PVT descriptions, detailed geology are needed such that the simulation model results will match history.  Thus a large portion of time in building a simulation model is the history matching process.  However, simulation models can then be used for many forecast scenarios, a more detailed understanding of flood performance, highlighting swept vs unswept areas, etc.

In 3DSL, skipping the transport step and running a surveillance model means setting the keyword SURVEILLANCE=ON.

For an example and a tutorial on how to convert a surveillance model to a flow simulation model see here.


Rod Batycky's picture

What is the difference between a surveillance and a simulation model?