Seismic waveform simulation using hydro-mechanical reservoir models to assess time-lapse seismic attributes

Time-lapse seismic monitoring provides information about subsurface changes in reservoir fluid-saturations, stress and strain. Using multi-vintage surface seismic surveys, time-lapse seismic measurements, quantified through seismic attributes such as travel-time shifts and reflection amplitude differences, are serving as a valuable tool for reservoir and geomechanical model calibration and hence, improving reservoir predictions. In this thesis, I introduce and develop an integrated seismic and coupled fluid-flow and geomechanical workflow for computing time-lapse dynamic changes from surface reflection synthetic seismograms that focuses on reducing the error and uncertainty in time-lapse seismic analysis. The workflow is demonstrated using both an anisotropic ray-based algorithm and an isotropic finite-difference full-waveform algorithm on a suite of simple four layer reservoir models as well as two numerical geomechanical reservoir models: a two-fault graben structure model and a complex deep reservoir model undergoing depletion. Using these models, a variety of acquisition strategies and data processing methods are applied. First, I generate synthetic seismograms using an anisotropic ray tracing algorithm to investigate the effects of time-lapse subsurface changes on seismic attributes. The reservoir models with time-variant properties, constructed from output of coupled fluid-flow and geomechanical simulation and a stress-sensitive rock physics model, have geometry characterised by three compartments that are offset by two normal faults having high or low fluid-flow transmissibility. Travel-time shifts and reflection amplitude changes are used to evaluate physical changes within the reservoir system. The results suggest that compartmentalisation can be identified but that it is important to understand the stress path of the reservoir if quantitatively accurate estimates of velocity changes and strains are required. Next, I explore the feasibility of using time-lapse AVO and AVOA analysis to monitor reservoir compartmentalisation as well as to evaluate stress induced seismic anisotropy. Time-lapse seismic reflection amplitude changes are estimated using an anisotropic ray tracing simulation, as well as the exact and approximate reflectivity solutions. The time-lapse AVO and AVOA signatures display noticeable deviations between models experiencing isotropic and anisotropic TI (VTI and HTI) elasticity changes. The results imply that time-lapse AVO and AVOA analysis can be applied as a potential means for qualitatively and semi-quantitatively linking azimuthal anisotropy changes caused by reservoir production to pressure/stress changes. I then extend the integrated scheme using a geomechanical model for a complex deep reservoir undergoing compaction and an isotropic finite-difference full-waveform modelling algorithm to study the influence of overburden effective stress perturbations. The full-waveform synthetic waveforms are used to evaluate time-lapse seismic attributes resolution for more realistic synthetics. The time-lapse seismic travel-time shifts and time strains (from pre-stack and post-stack data) calculated from the synthetic seismograms are in a reasonable agreement with the respective input elasticity model. The results show that the time-lapse technique is reasonably accurate for predicting overburden velocity changes and hence geomechanical effects. Finally, I propose a new algorithm to measure time-lapse vertical travel-time shifts in seismic pre-stack shot and CMP gather data by tracking traces of a constant horizontal slowness in the tau-p domain. The approach is used to estimate layer vertical travel-time shifts, a 1D reservoir compaction-dilation coefficient, and hence calculate both velocity and thickness changes within the reservoir and overburden. I compare the estimates of layer interval vertical time-lapse travel-time shifts, and velocity and thickness changes with those of the input model. The results indicate that the new tau-p time-lapse method produces sufficiently accurate results compared to conventional methods. The results of the thesis indicate that time-lapse seismic monitoring in conjunction with reservoir fluid-flow and geomechanical simulations, rock physics models and seismic numerical modelling, has the potential to be a valuable tool for accurate measurement (both qualitatively and quantitatively) of time-lapse effects due to reservoir pore pressure induced geomechanical deformations. Time-lapse seismic data has the potential to help in the calibration of geomechanical reservoir models.