Abstract
A large amount of gas is trapped in the transition zones in tight, conventional reservoirs. These reservoirs tend to be prone to water production because the in-situ water saturation is at or exceeds the critical water saturation. The water-production issue is even more pronounced when wells are stimulated by hydraulic fracturing; the resultant water production might be interpreted to have its source in the aquifer below the free water level to a bad cement job or a connection to unknown faulting.
As an example, a typical Rotliegendes Sandstone well was considered. It was completed and subsequently fractured, and produces gas with a high water-gas ratio. The objective of this study was to demonstrate how to ascertain the most likely source of the water; whether it was produced from the aquifer or from high water saturation matrix regions in the reservoir.
Available fracture placement data were used to match analytically the pressures to confirm the stresses with the resultant fracture model compared with the one used for initial design. A box model was constructed in a reservoir simulator and the fracture was described using a Tartan grid. Using the analytical result as a base case the reservoir and fracture properties were adjusted using assisted history matching to replicate observed gas and water rates and flowing bottomhole pressure. The main matching variables were permeability, permeability anisotropy, stand-off to gas-water contact, relative permeability, gas and water end points saturations and fracture dimensions. The water saturation was derived from a saturation vs. height function; the power law factors in this equation can also be matched.
Analysis of the fracture delivery indicated that the dimensions of the fracture could be different from the dimensions derived from the design. The fracture height and half-length in the post-fracture model showed a higher fracture height and a smaller half-length connecting potentially with free water. The assisted history matching demonstrated that the water rate behavior could best be explained as a result of high water saturation in the reservoir, rather than direct connection to the aquifer through an inadequate cement bond or fault. A range of predictions was provided to describe the range in estimated ultimate recovery of a single typical well. The completion design will cater for active dewatering of the well.
Because great uncertainty exists about the propensity to water production, the optimal well design will be flexible enough to allow for de-watering throughout the life of the well.
