Pre-drill pore pressure and fracture gradient predictions obtained from seismic velocity data are an important component of any well design. Although there are significant uncertainties associated with these analyses, few methods have been proposed to quantify them. We present here a Monte Carlo approach to Quantitative Risk Assessment to model uncertainties in velocity, in the effective-stress - velocity relationships, and in the magnitudes of centroid and buoyancy effects. The approach can be applied to pore pressure prediction regardless of the mechanism that is responsible for the overpressure - in fact, uncertainties in the mechanism can be modeled if the relationships have the same functional form. The results reveal both the total uncertainty and the independent effects of the uncertainties of each input, from which it is possible to determine the critical input values for which a reduction in uncertainty is important. While uncertainties in the velocities contribute significantly, surprisingly large uncertainties in the predicted pore pressures result from uncertainties in relationships between velocity and density, and between velocity and effective stress, as well as from difficulties determining the parameters to predict pore pressures in reservoir sands using centroid and buoyancy models. Using geological constraints on the minimum stress derived from computed overburden and pore pressure, it is possible to establish limits on hydrocarbon column height based on frictional and hydraulic fracturing limits on reservoir seals. It is also possible to predict the relative number of casings required to reach target reservoirs utilizing appropriate mud windows. The results, including uncertainties, provide critical initial data for risk-based approaches that assess reservoir potential and drilling costs.
1. INTRODUCTION
It is standard practice when planning wells to utilize pre-drill seismic data to compute pore pressure and fracture gradient profiles to use as upper and lower bounds on required mud weights for safe drilling (see a recent review by Huffman [1] and Swarbrick [2]). Considerable uncertainties in the results are introduced not only by inaccuracies in velocity determinations but also in the transforms used to derive these upper and lower bounds from the velocities.
Improved mud window estimates are now being derived utilizing geomechanical principals to compute collapse and lost circulation pressures. Both of these are functions not only of the position of the well in the field but also of wellbore orientation [3,4]. In order to compute collapse and lost circulation pressures from velocities, it is necessary to utilize quite general relationships between velocity and other parameters such as rock strength, and to make simplifying assumptions about the in situ horizontal stresses. While this has the potential to improve the accuracy of mud window prognoses, it is unclear whether the increased uncertainty is small enough to justify the extra analyses.
By quantifying uncertainties in pore pressure and other predicted values, and more important by determining their origin, it is possible not only to begin to quantify the drilling risk but also to make decisions about how best to reduce that risk. For example, if uncertainties in the velocities used as input to the predictions contribute large uncertainties to the results, this may dictate reanalysis of the seismic data.
