Water-Influx Modeling of the Svartsengi Geothermal Field, Iceland

Abstract

Summary. In this paper, lumped-parameter models are presented for confined and unconfined liquid-dominated geothermal reservoirs without recharge and compared to production data from the Svartsengi geothermal field in Iceland. They show that recharge or water influx is likely to affect the field's drawdown with time. Three petroleum reservoir water-influx methods are then presented for steady-state, pseudosteady-state, and unsteady-state aquifers encroaching into the geothermal reservoir. History matching shows that the reservoir/aquifer system is best described as unsteady state and radial, with the aquifer boundary taken at infinite distance. Introduction Geothermal and hydrocarbon reservoirs are in some respects similar, and yet different. In this paper, we examine one of their similarities: the recharge or influx of water from surrounding aquifers. Two types of high-temperature reservoirs represent most geothermal systems: vapor-dominated and liquid-dominated. The Geysers field in northern California is a typical vapor-dominated reservoir, and the Wairakei field in New Zealand is a typical liquid-dominated reservoir. The main feature that distinguishes these types of reservoirs is their pressure gradient. In vapor-dominated systems, the pore space is filled with steam, so the pressure gradient is vaporstatic. In liquid-dominated systems. pressure gradient is vaporstatic. In liquid-dominated systems. the pore space is liquid-filled, so the pressure at any depth is dominated by the water column weight to surface i.e. hydrostatic. The volume of high-temperature geothermal reservoirs must be large to sustain steam production for electric power generation, typically several cubic kilometers of power generation, typically several cubic kilometers of reservoir rock and fluid. This requirement and the volcanic nature of geothermal energy mean that exploited geothermal reservoirs must be thick, usually 1 to 2 km [0.6 to 1.2 miles]. In the natural state, geothermal reservoirs are dynamic rather than static. This means that there is convection within the reservoir and advection to the surface and surrounding formations, hot springs and fumaroles are manifestations of geothermal systems. The fluids lost from a geothermal system must be supplied from below or from adjacent aquifers. Therefore, aquifers surrounding, geothermal resources are likely to encroach into the reservoir when the pressure falls as a result of production. In this paper, we study the application of water-influx methods in geothermal reservoir evaluation. We use a lumped-parameter material-balance model and several water-influx models to match the production history of the Svartsengi geothermal field in Iceland. Basic Reservoir Models The modeling approach used in this paper is often described as lumped-parameter. An early application of this approach to geothermal systems is that of Whiting and Ramey, who studied the Wairakei field in New Zealand. A later lumped-parameter study is that of Brigham and Neri, who studied a part of the vapor-dominated Larderello system in Italy. Other similar studies are those of Castanier et al. and Fradkin et al . An overview of lumped-parameter models is given by Grant et al. and Grant. Olsen reviewed uses of lumped-parameter and water-influx models in geothermal reservoir engineering. There are two components to our model of the Svart-sengi geothermal field, the reservoir and the aquifer. They are brought together in the material-balance equation (1) where the current mass, Wc, equals that initially in the reservoir, Wi, minus what has been produced, Wp, plus any water influx, We. In lumped-parameter modeling, the reservoir is treated as one element with some average properties. Of primary interest in our work is the reservoir production primary interest in our work is the reservoir production mechanism. Is fluid produced because of expansion or because of a fall in liquid level? The initial fluid in place in liquid-dominated reservoirs may be compressed water. In this case, when the reservoir is penetrated by wells and produced, the water expands because of its compressibility. We call this a confined reservoir. For a reservoir of volume V, the fluid mass in place is given by (2) SPERE P. 77

Model predictions of the Svartsengi Reservoir, Iceland

“…It is also of interest to compare predicted water level decline using the present model with those calculated by earlier models (Figure 5 models developed by Kjaran et al [1980] and Regalado [1981]. Comparison between the present results and those presented by Gudmundsson and Olsen [1987] are given in Figure 14. Figure 13 shows that the present model predicts much lower water level decline then the models that assume one-sided recharge and somewhat larger decline than the model based upon the unit response function.…”

“…The production history is shown in Figure 1. In March 1983 30 x 109 kg of fluids had been extracted, with a total pressure drawdown in the reservoir of 8-9 bars (,• 100 m [Vatnaskil, 1983;Gudmundsson and Olsen, 1987]). [Georgsson, 1981;orgsson and Tulinius, 1983].…”

Lumped-parameter models for low-temperature geothermal fields and their application

“…Several lumped-parameter models and modeling methods have been reported in the literature (Grant, 1983;Whiting and Ramey, 1969;Brigham and Neri, 1980;Castanier et al, 1980;Brigham and Ramey, 1981;Olsen, 1984;Gudmundsson and Olsen, 1987;Axelsson, 1989;Alkan and Satman, 1990;Axelsson and Dong, 1998;Satman and Ugur, 2002). Axelsson and Gunnlaugsson (2000) discuss the usefulness of lumped-parameter models in interpreting monitored production data for low-temperature geothermal fields.…”