Illumination of channelised fluvial reservoirs using geological well-testing and seismic modelling

Abstract

Fluvial reservoirs are amongst the most prominent hydrocarbon bearing deposits in the world. Complexity in channel networks, spatial pattern and internal heterogeneities are of the main challenges in characterising these types of reservoir. In this thesis, a novel geoengineering approach is implemented to integrate the multidomain information (e.g. outcrop and time-lapse seismic interpretation) and to describe the well-test response of certain fluvial deposits. Comprehensive modelling and numerical well-test simulations have then been employed to study the dynamic behaviour of such systems. These resulted in diagnosing a well-testing family that includes a new generalised "Ramp Effect" model that is presented for the first time. Using systematic geostatistical modelling, the ramp effect is elaborated in terms of spatial statistics. The ramp model has been demonstrated by a few real-life well-test examples in a variety of channelised environments. Sophisticated multi-point statistics modelling are utilized to capture the facies transitions producing the lateral crossflow transients that result in the ramp effect and to demonstrate how the response can be generated in a meandering and anastomosing fluvial environments. Internal cross-flow can be confused with external layer cross-flow and with other linear flow responses (e.g., parallel faults, natural or artificial fractures). The non-uniqueness in interpretation of the ramp effect is addressed by employing time-lapse seismic data, which help in detecting the spatial geological heterogeneities and constraining the welltest interpretations. The illuminating power of the time-lapse seismic data is illustrated by synthetic seismic modelling examples. The implications of a complex fluid (i.e. gas-condensate liquid drop-out) on altering the well-test and seismic responses are also discoursed. A compositional reservoir simulator is employed to mimic the complex fluid behaviour of the fluvial reservoirs showing the ramp effect. Application of compositional simulations highlights the limitations in current petro-elastic modelling that are unable to take the compositional changes into account. Therefore, a novel approach is also developed to improve the petro-elastic modelling which facilitates the synthetic seismic generation in the presence of continuous composition changes of the fluid. This leads towards a better description of the reservoirs under simultaneous effect of geological and fluid heterogeneities.