Abstract

A key consideration in the development of geothermal resources is the placement of each new well to maximize production. A slight increase in production based on improved subsurface characterization can have a major impact on the economic success of a project. In this paper we describe a novel method to model the fracture network and generate permeability maps to site new drilling locations in the Darajat geothermal field in Indonesia, operated by Star Energy Geothermal. Despite the more than 50 wells drilled in this field, relatively high uncertainty remains in the success of future steam make-up wells, as the new wells will be in step-out (previously undrilled) locations. This project is an attempt to reduce the performance uncertainty in these future wells by modeling the natural fracture system, applying methods used successfully in oil and gas reservoirs, matching the known distribution of permeability in drilled portions of the field, and predicting its distribution in undrilled portions.

The modeling of natural fractures uses “fracture drivers” to generate and attempt to match fractures interpreted from borehole image logs. Each fracture driver generates a limited fracture set, which are combined into a full reservoir fracture model. The drivers depend on the interpreted distribution of lithology units, the largest faults, and igneous intrusions in the field and perturbed stress fields or empirical rules around these geological features to generate fracture sets. Two of the drivers are fault -related and two are intrusion-related. The geologic features are characterized using remote sensing methods, geophysical studies (gravity, resistivity, microseismic), subsurface geology (core, cutting), and surface geology. The drivers are applied in sequential steps, and fractures generated from the driver-specific constraints are used to define fieldwide volumetric fracture intensities. The intensity and orientation of fractures generated by each driver are combined, and these composite parameters are used as input for creating a discrete fracture network (DFN) across the field. The DFN models are used to calculate fracture permeabilities throughout the geothermal field using measured productivity indices to help calibrate fracture apertures and length distributions. The completed models show a heterogeneous permeability distribution in the field that shows promise for planning and optimizing future well planning.

This is the first known application of this fracture modeling technique to a large geothermal project. A second phase of this project has now been undertaken to incorporate uncertainty modeling. We envisage that the tools and methods developed for the natural fracture characterization in this study will be applicable to other geothermal fields, and also to oil and gas fractured reservoirs.