Models: Overview

Related topics:

There two suites of models in WellTest, which we refer to as legacy models and advanced models. Only the legacy models support minifrac and non-conventional (e.g,. PITA test types). The advanced models include numerical models and horizontal multifrac analytical models, as well as functionality not available in the legacy models.

Models are used to simulate your data using the process of history matching pressure transient data based on mathematical principles. There are many different models available to match data, depending on the situation. Thus, it is important to analyze the pressure transient data before modeling because it forces you to think about the probable reservoir configurations, and provide good estimates of reservoir parameters. Models are not unique (different model types can match the same set of data), and as a result, we recommend that the choice of model type occur after the analysis step. 

The advantages of modeling pressure transient data are as follows:

  • Modeling makes use of all the information within a dataset. While analyzing data, the analyst might try and determine permeability and skin by analyzing the data points which make up the zero slope on the derivative plot and the semi-log straight line on the radial plot, but ignore the data points in the transition period between wellbore storage and radial flow. Models make use of the information contained in transition periods.
  • Modeling takes all flow regimes into account. In multi-rate situations, analyses depend on the superposition of the equation for a single-flow regime. For example, the derivation of Horner time includes the assumption that all flow, including the entire drawdown, is radial. Modeling doesn’t assume that only one flow regime has occurred.
  • With modeling, you can simultaneously analyze multiple flow periods, so that a single set of parameters can be found.

Parameter values obtained during the analysis step provide a good starting point for an appropriately chosen model type. Parameters can then be optimized by automatic parameter estimation (APE). Before using the APE method, corrupted data should be removed from the dataset to prevent the attempted match of invalid points.

Model Assumptions

There are many assumptions that go into the model itself, which can lead or mislead the analyst. Most models assume that a reservoir is homogeneous (dual porosity excluded). In nature, there isn’t a single reservoir that is actually homogeneous, but many reservoirs behave as homogeneous reservoirs.

For example, the composite model assumes that reservoir properties change at a certain radius from the reservoir. This phenomenon also doesn’t necessarily occur in nature, yet some reservoirs behave as though they are composite reservoirs. An example of this is an injection well, where the fluid properties (compressibility and viscosity) change at a certain distance from the wellbore. The composite model can be used to match many datasets, so it is important that it is not overused; however, there are cases where it is the appropriate model to use.