High Performance Integrated Asset Modeling: A Giant Gas Field Case Study

Abstract

We present the development of a field-scale simulation tool coupling the model of Zohr, a super giant deepwater gas reservoir requiring a multi-million active cells grid with dual porosity/dual permeability formulation, to its gathering network. Deepwater field development often relies on a complex subsea gathering infrastructure, possibly evolving over time and leading to a complex topology. Here, facilities route the multiphase production stream to an onshore compression plant throughout a network with bifurcations. To perform integrated modeling of such assets, we have developed a Facility Network Solver (FNS) with flexible topology, whose formulation relies on a graph representation with continuity equations at nodes and tabulated constitutive equations at edges. FNS was designed to be integrated with the industrial grade GPU-based reservoir simulator used in the company, both jointly developed by Eni and Stone Ridge Technology, with emphasis on preserving usability and simulation performance. The correctness of FNS results in the presence of bifurcations was thoroughly assessed in standalone mode, through benchmarking against what is today considered a reference commercial network solver. FNS integration with the reservoir simulator enables forecasts where the back-pressure is taken into account. Engineers can better assess the viability of different development scenarios, including dynamic upgrades to the network topology, using reservoir simulation workflows they are well accustomed to. In particular, a single tool replaces often heterogeneous associations of third-party software, without impacting simulation time. Sensitivity analyses were performed on both coupling frequency and location. It was concluded that periodic coupling at the well-head was a satisfactory setting, yielding negligible performance overhead with respect to standalone reservoir simulations, thus enabling the integrated model to be used routinely as the sole simulation model.