document preview

Abstract
This paper addresses insights obtained from dynamic simulation of detailed models of low permeability fluvial sandstones in the Williams Fork Formation, Piceance Basin, Colorado. The detailed static models and an overview of the dynamic modeling aspects were decribed previously (Pranter et. al., in press). The static models represent the complex stratigraphic architecture and connectivity of these fluvial sandstones as developed through outcrop analysis while the flow simulation provide a means to understand the dynamic connectivity. The Williams Fork Formation consists of fluvial channel sandstones, crevasse splays, floodplain mudstones, and coals that were deposited by meandering- and braided-river systems. Static connectivity analyses of 3-D outcrop-based architectural-element models have shown how relatively wide sandstone bodies enhance connectivity. The dynamic simulations illustrated in this paper shows that historical performance can be mimicked by considering only point bars, channel bars and marine sandstones (reservoir-quality sandstones) as pay. The reservoir simulations used an integrated approach to create 3D dynamic simulation models by combining detailed static geological, geophysical and petrophysical characterizations. We incorporated and calibrated hydraulic fracture properties at each well to approximate initial productivity, used the characterization process to match hyperbolic decline behavior and then investigated the volume influence of wells and the impact of geologic characterization on performance by predicting long-term gas recovery. These realistic and detailed reservoir models can honor the historical gas rate without applying the assumption of open natural fractures. This work demonstrates that an integrated approach can lead to realistic 3D geologic and dynamic models which are consistent with static data and historical performance. Such models are useful for estimating the impact of complex sandstone connectivity on early and long-time performance including well interference and optimal spacing. The paper also briefly discusses how seismic constraints can lead to more unique descriptions with regard to distributions of multi-story sandstone channels. Such methods combined with detailed geologic models as described here could be used for designing more optimal well locations and optimal spacing in less developed areas in tight gas reservoirs.

Introduction
Mamm Creek Field is located in the Piceance Basin, northwestern Colorado, in the United States. Most of the natural gas production in Mamm Creek Field comes from fluvial tight sandstone (~5000 ft deep) in the Williams Fork formation; however, marine sandstone in the Corcoran, Cozzette and Rollins members (~7000 ft deep) of the Iles Formation and the middle and upper sandstones of the Williams Fork Formation also contribute to production (Scheevel and Cumella, 2009). The Williams Fork Formation is well known as a low porosity and low permeability tight-sandstone within a basin center gas accumulation system. Deposition of these Williams Fork Formations include fluvial channel sandstones, crevasse splays, floodplain mudstones, and paludal coals that were deposited by meandering- and braided-river systems within coastal- and alluvial plain settings (Pranter et. al., in press). This complex fluvial depositional environment results in highly heterogeneous reservoir connectivity. From a production point of view, it is essential to understand the details of properties and trends within the system.

To better understand the gas production from basin centered tight sandstone systems, static sandstone-body connectivity is not the only issue with regard to effective well drainage. The near-well connectivity improvement by modern hydraulic fracturing must also be considered. Long-term drainage is affected by the sandstone-to-sandstone contact area and permeability. This long-term effective drainage is best evaluated through dynamic modelling.