Proceedings World Geothermal Congress 2005 Antalya, Turkey, 24-29 April 2005
Advances in Subsidence Modelling of Exploited Geothermal Fields Philip J. White, James V. Lawless, Sergei Terzaghi, Wataru Okada Sinclair Knight Merz Ltd., P.O. Box 9806, Newmarket, Auckland, New Zealand
[email protected]
Keywords: Subsidence, modelling, prediction, Wairakei, Tauhara, Ohaaki, Kawerau
Two important but often overlooked features of subsidence bowls are: The bowls are rarely circular, at least those that are well defined by benchmarks (they are troughs rather than bowls) The bowls can move and expand over time Any theory to explain the development of subsidence bowls must be able to account for these features.
ABSTRACT One dimensional subsidence modelling that was standard 20 years ago has recently given way to more advanced two and three dimensional modelling using soil mechanics software packages, which now run on standard office computers. Accurate modelling requires a good knowledge of the subsurface geology, reservoir conditions, and rock properties, including porosity, permeability and compressibility. With good input data it is possible to obtain a close match to historical subsidence, and thus to produce reliable future predictions. The latter is however dependant on the reliability of future reservoir pressure predictions, which requires more detailed reservoir modelling than would normally be the case for geothermal resource estimation or management.
2. WAIRAKEI Geothermal power generation began at Wairakei in 1958, after several years of drilling and output testing. Electrical production peaked at 192 MWe in the 1960s, and is currently about 165 MWe (Contact 2003). The scheme is now owned by Contact Energy Limited (Contact), and extracts about 140,000 tonnes per day (tpd) of steam and hot water from a variety of areas and depths. Partial reinjection (begun 1996), now comprises about 30,000 tpd. Poihipi power plant on the western side of Wairakei (commissioned 1997) produces 4,800 tpd from a shallow steam zone, with all condensate reinjected outside the field. A development at Tauhara that will extract another 20,000 tpd (with full reinjection) should occur by 2005, and further expansion is planned for Wairakei (Contact 2001, Geotherm 2001).
Subsidence at Wairakei-Tauhara due to almost 50 years of geothermal fluid extraction was the first to be modelled by two dimensional finite-element analysis. The software accommodates variable rock properties, including nonlinear stress-strain behaviour, and pre-consolidation history. A good match to historical subsidence in time and space was achieved with a single set of rock properties for each geological unit, apart from two local zones with different permeability. Compared to previous 1-D subsidence modelling, this study showed greater sensitivity to changes in reservoir pressure and strong control over the location of rapid subsidence by the formation morphology. Analysis of Ohaaki and Kawerau geothermal fields indicates that the method can be applied to other systems where there is sufficient geological, geomechanical and pressure data.
2.1 Subsidence The past 47 years of geothermal power generation at Wairakei with little reinjection has caused extensive pressure decline within the reservoir (-25 bars), and subsidence of the ground surface. Subsidence was detected by repeat levelling surveys of benchmarks in 1956 during output tests prior to commissioning of the power plant (Hatton 1970). More benchmarks were installed, and by 1966, the pattern of subsidence at Wairakei was clear. Although there have been some changes in the rates since then, the overall pattern has remained relatively constant. Minor subsidence extends over a wide area (most of the area that has been monitored inside the resistivity boundary), within which there is a deep subsidence bowl about 1 km in diameter located to the north of the borefield, approximately 1 km west of Wairakei village (Figure 1). Subsidence rates reached almost 500 mm/year in the late 1970s, but have since declined to about 100 mm/year. Total subsidence (to 2001) exceeds 15 m near the centre of the bowl, more than at any other geothermal field in the world.
1. INTRODUCTION Subsidence due to exploitation has been documented at many geothermal fields, including The Geysers (Mossop & Segall 1997), Svartsengi (Eysteinsson 2000) and Cerro Prieto (Glowacka et al. 2000), but at rates that are typically an order of magnitude smaller than at Wairakei-Tauhara. Accordingly, little modelling or future prediction has been applied to geothermal subsidence outside of New Zealand. Much greater subsidence and more refined modelling has been applied to oil/gas and groundwater reservoirs, such as the Ekofisk, Beldridge and Lost Hills oil/gas fields (e.g. Chin et al 1993, Fossum and Fredrich 2000), and lessons from these examples have been applied in this study.
Horizontal movement rates of up to 130 mm/year have been measured near the Wairakei subsidence bowl, though these have slowed along with the decline in vertical rates. Horizontal movement surveys suggest that the centre of the Wairakei subsidence bowl has moved south by about 200 m in 10 to 20 years (Allis 2000).
Subsidence modelling of Wairakei-Tauhara was undertaken for Taupo District Council, for input into the Wairakei consenting process. Analysis of Ohaaki was undertaken to test the general validity of the subsidence model. A preliminary assessment was made of historical and potential future subsidence at Kawerau as part of a prefeasibility study for a possible new power development there.
1
White et al.
Figure 1: Location map of Wairakei and Tauhara, New Zealand
Figure 2: Geological cross section A-A’ (no vertical exaggeration) Because Wairakei geothermal field is hydrologically connected to Tauhara (Figure 1) the pressure decline has extended to Tauhara, causing up to 2.5 and 1.6 m of subsidence by 2001 in two separate subsidence bowls there (Figure 1). Since the late 1990s, a third area of subsidence has formed in southern Tauhara, at the edge of the Taupo urban area. Subsidence rates in this area (up to about 80 mm/year in 2004) are now possibly higher than anywhere else in Wairakei-Tauhara (White 2004). Although the Wairakei subsidence bowl has experienced the most subsidence, the three bowls at Tauhara are of more concern because of their proximity to the urban environment.
Past 1-D modelling by Allis and Zhan (1997) and others has been used to predict future subsidence. However, there are significant limitations with the 1-D method; hence the need for detailed 2-D modelling. 2.2 Geology The geology of Wairakei-Tauhara has been described by numerous authors, beginning with Grindley (1965). Using structural contour maps, geological cross sections were constructed. One of those cross sections is shown in Figure 2. The units that are most significant for this study are: 2
White et al. Waiora Formation: pumice breccia and ignimbrite layers, with interbedded sediments and interlayered extrusive rhyolite lava flows (including Karapiti Rhyolite). This formation is the main productive reservoir at Wairakei, and the major pressure decline due to production has occurred within this formation. In most of the field it is overlain by:
For subsidence modelling of Wairakei, historical reservoir pressure and temperature data was taken from reservoir modelling by O’Sullivan (1999), O’Sullivan et al. (2001) and Mannington et al (2002), with some modifications to fit field measurements reported by Clotworthy (2001), and geological controls.
Huka Falls Formation: lacustrine sediments and pumice breccias, the latter comprising pyroclastic flow deposits and their re-worked equivalents. Grindley (1965) distinguished four members: Hu1 (oldest) and Hu3: porous but low permeability mudstones, Hu2: moderately permeable unconsolidated pumice breccia that forms a shallow aquifer, and Hu4 (youngest): fine sandstone and mudstone, forming a partial aquiclude.
In selecting pressure input data for the Tauhara subsidence modelling, efforts have been made to achieve consistency between input data and measured data from wells. Because the reservoir model of O’Sullivan (2001) has a relatively poor match to historical well data, measured pressure data was used wherever possible, with the reservoir model being used only as a guide to lateral and temporal changes. Wherever possible, input pressures were selected which conform to the known Tauhara pressure profiles, and the historical trends for the deep wells.
Above the Huka Falls Formation are younger pyroclastics and minor lake sediments, which as a whole constitute groundwater aquifers, though locally perched.
Future subsidence was modelled for several possible future production/reinjection scenarios. Under the status quo scenario, pressures were assumed to remain unchanged over the next 50 years. Reservoir model predictions were used for other scenarios, including O’Sullivan’s (1999) prediction of a 2 bar incremental pressure decline at Tauhara for the 20,000 tpd development there.
2.3 Material Properties There is limited laboratory test data on the geotechnical properties of units in the Wairakei-Tauhara geothermal system (e.g. cohesion, friction angle, permeability, stiffness, void ratio, and stress-strain behaviour). An initial set of geotechnical properties was derived from previous studies involving similar materials (including Robertson 1984, Kelsey 1987, Allis 1999, Fairclough 2000, and Grant 2000). These properties were optimised to match the model subsidence trend from 1950 with subsidence measurements. A single consistent set of reference parameters (which are adjusted by the model to account for stress state, void ratio, and pre-consolidation pressure) was used throughout, with two exceptions. Beneath the Wairakei subsidence bowl, enhanced permeability was introduced to correspond with near-vertical permeable zones (faults or hydrothermal eruption vents) that previously fed hot springs there. A zone of low permeability was introduced to explain the delayed pressure response in southern Tauhara.
2.5 Modelling Eight 2-D models were developed using the finite element analysis code Plaxis Version 7.2 on the sections in Figure 1. The main advantages of 2-D over past 1-D modelling are:
2.4 Reservoir Pressures Since the 1950s, deep liquid pressures at Wairakei have fallen by about 25 bars over a large area. Pressures in the deep liquid at Tauhara are less well known, but have fallen by about 20 bars, at least in the area of the deep wells. Pressures in the deep liquid at Wairakei and Tauhara have stopped falling, and have risen slightly since reinjection began. Two-phase zones have formed locally at shallow levels, where pressures appear to be still dropping.
It is based more closely on the interpreted geological structure.
It allows more advanced definition of geotechnical properties (e.g. permeability varying with void ratio, non-linear stress-strain behaviour, and preconsolidation stress history).
It incorporates the coupled Biot Theory, modified to account for non-linearity, plasticity, and stress changes in the 2-D plane strain.
Fluid flow and pressure changes can be modelled both horizontally and vertically.
Horizontal and vertical permeability can be set independently. This is particularly necessary for the mudstones, which have strongly anisotropic permeability.
Horizontal Distance (km) 0
1
2
3
4
5
-0.50
Subsidence (m)
0.00 0.50 0
1.00 1.50 2.00 2.50 3.00 2001 Survey data
Model Match
Figure 3: Comparison of actual and calculated subsidence to 2001, Section A-A’ 3
6
White et al. Using the input geology, material properties and pressures, the model was initially tested against observed historical subsidence. The parameters were adjusted until a good match was obtained for all 8 model sections, using a single set of material properties for each geological unit. Once the match to past subsidence was reasonable, predictions of future subsidence were made.
the Wairakei subsidence bowl) correctly simulates the acceleration of subsidence in the early 1960’s and subsequent decrease in the subsidence rate towards the late 1980’s and early 1990’s, though the model generally overstates subsidence slightly. Benchmark 9734 (near the centre of the largest subsidence bowl at Tauhara) was first monitored in 1997, so prior subsidence was calculated by comparing with adjacent benchmarks. A good match was achieved to past subsidence.
Matches in space The model match to historical subsidence along one of the 2-D profiles is shown in Figure 3. A similar match was obtained on most model profiles, expect those which parallel the structural contours on the base of the compacting layer. The third dimension (out-of-plane) drainage that will result causes the model to under-estimate subsidence on these sections.
Future predictions Future subsidence at Wairakei and Tauhara was predicted to 2052, based on various development options, including the status quo, the proposed 20,000 tpd Tauhara development and the 245,000 tpd Wairakei expansion going ahead, and total shutdown in 2026. Predictions for benchmark P128 (Wairakei) are presented in Figure 5.
Although the magnitude of subsidence will depend on the thickness of the compacting layer, the rate of subsidence is controlled by the slope of the lower boundary. These models have shown that subsidence is most rapid where the consolidating unit(s) overlies a steep slope, so that fluid can flow laterally out of the mudstone into more permeable formations. This explains why the subsidence is greatest at specific locations; it is where the geological contacts on the consolidating units are steepest, or where those units are cut by permeable structures (e.g. faults).
Under the status quo scenario, the rate of subsidence will slowly decrease, but subsidence will continue to 2052 and beyond. Total (including past) subsidence to 2052 is predicted to exceed 26 m at P128 (Wairakei), and 5 m at 9734 (Tauhara). Any additional fluid extraction will increase subsidence rates and total subsidence significantly. A total shutdown would result in a small, gradual rebound, though most subsidence is not reversible. In contrast, based on 1-D modelling, Allis (1999) predicted that the 20,000 tpd Tauhara development would have no significant effect on future subsidence rates, and (Allis 2004) that full reinjection will have no significant effect.
Furthermore, the modelling indicates that subsidence is largely due to compaction of the Hu1 mudstone unit. Although the Hu3 mudstone has similar material properties, the pressure decline beneath Hu3 has been much smaller (about 15 bars maximum, compared with 25 bars beneath Hu1).
2.6 Effects At Wairakei, effects to date include damage to pipelines, wells and drains around the Wairakei subsidence bowl, and to nearby roads, electricity transmission lines, and Wairakei Hotel (Bloomer and Currie 2001). This is despite few structures within the worst area of subsidence, which is forested land. A lake has formed in Wairakei stream above the area of greatest subsidence.
With time, the subsidence bowls are predicted to shift and enlarge as the pressure change propagates further laterally into the Hu1 unit. Matches in time The model match with time for benchmark 9734 at Tauhara and P128 at Wairakei are shown in Figures 4 and 5. The model subsidence at benchmark P128 (near the centre of
The subsidence effects at northern Tauhara relate mainly to road damage and stormwater drainage, and have been dealt with through maintenance, albeit with increased frequency. Year
1950 0
1960
1970
1980
1990
2000
Subsidence (m)
0.5 1 1.5 2 2.5 3 Measured
Model Match
Figure 4: Historical match for benchmark 9734, Tauhara subsidence bowl 4
2010
White et al. Year 1950 0
1970
1990
2010
2030
2050
Subsidence (m)
5 10 15 20 25 30 35 Survey Data Prediction (Expanded Wairakei)
Model Match Prediction (Total Shutdown)
Prediction (Status quo)
Figure 5: Historical match and predicted future subsidence at benchmark P128 (Wairakei subsidence bowl) In southern Tauhara, houses, roads and footpaths have suffered damage since mid-2003. Although this is not structural damage in terms of affecting the structural integrity of buildings, there have been significant nonstructural effects, including loss of water-tightness. The nature of the damage is consistent with subsidence.
and south of the field. To minimise returns to the reservoir, fluid was reinjected into deep marginal locations, and into shallow cool rhyolite aquifers that appeared to be hydrologically separate from the geothermal system (Clotworthy et al. 1995). However, some reinjection wells were closed after reinjected fluid returned to the system, lowering reservoir temperatures.
Based on the predicted subsidence over the next 50 years, structures and infrastructure in all of the subsidence bowls at Wairakei and Tauhara are likely to be further damaged in the next 50 years. The extent of subsidence, and hence damage, is dependent on the future production and reinjection at Wairakei and Tauhara.
With commissioning, fluid withdrawal increased to about 45,000 tpd, and reinjection to about 28,000 tpd. Most of the remaining fluid goes to the atmosphere via the cooling tower, and some (about 2,000 tpd, or 5%) goes to Ohaaki Pool and the Waikato River. Surplus capacity was available until 1993, but production has since been declining. New wells have been drilled and wells have been deepened and cleaned to remove calcite scale, but steam supply has continued to decline. Output has dropped from 80 MW in 1995 to 50 MW in 2000 (Thain & Dunstall 2000), and 42 MW in 2002 (Contact 2002). In hindsight, the 116 MWe development was oversized for what is a relatively small field.
2.7 Mitigation Because subsidence rates are sensitive to small pressure changes at depth, and because subsidence locations are controlled by geology, targeted reinjection has been proposed to minimise future subsidence. Targeted, in this context, means beneath the subsidence bowls, into the formations with the greatest pressure drawdown.
3.1 Subsidence Subsidence was first detected about 1970, during the initial discharge testing, with rates of up to 150 mm/year (Allis et al. 1997). Subsidence slowed and stopped in the 1980s as pressures recovered from the discharge testing, but with little or no reversal of subsidence. Subsidence rates accelerated as pressures declined following commissioning of the power plant in 1988. By 1994, maximum subsidence rates were over 500 mm/year (Allis et al. 1997), but they declined rapidly again, so that four years later, maximum rates were less than 300 mm/year (Allis & Zhan 2000). Total subsidence at Ohaaki (about 3 m) is much less than at Wairakei, because of the much shorter time span.
There has been only limited modelling of reinjection, but the predicted pressure recovery, and thus subsidence, with full reinjection are similar to (but slower than) what would occur in the total shutdown scenario. The total shutdown scenario indicates that subsidence rates will decline rapidly, possibly with a slight reversal of past subsidence (Figure 5). 3 OHAAKI Ohaaki is located about 25 km northeast of Wairakei. Exploration of the Ohaaki geothermal system began in the 1960s, with the first well drilled in 1965. There was extensive field testing between 1967 and 1971, with extraction averaging 25,000 tpd, and no reinjection. Smallscale extraction from 1975-1980 was partly balanced by reinjection. In 1982, the decision was made to proceed with development, and the 116 MWe plant was commissioned in 1988.
As at Wairakei, subsidence affects the entire area within the resistivity boundary, but the area of rapid subsidence (>100 mm/year) is 1-2 km2. The subsidence “bowl” is crescent shaped at Ohaaki. Allis & Zhan (2000) observed that the centre of subsidence moved 500 m east between the testing phase (1970s) and the production phase (1990s) (Figure 6), strongly supporting the concept of lateral drainage.
Production wells are located within the central part of the field, with approximately half on each side of the Waikato River. Reinjection wells are located around the north, west 5
White et al.
Line of section (Figure 7) Centre of subsidence bowl 1969-1974
H338/1
Figure 6: Ohaaki subsidence bowl: horizontal (1995-96) and vertical (1988-95) movements (from Allis et al 1997) Horizontal movements of up to 150 mm/year were calculated from 1995-96 surveys, and movement was toward the areas of greatest subsidence.
1988, pressures again declined, by a further 20 bars by 1995 (Clotworthy et al 1995, Lee & Bacon 2000). The total pressure decline of about 25 bars is thus similar to that experienced at Wairakei and Tauhara.
3.2 Geology At Ohaaki, the Waiora Formation is overlain by Huka Falls Formation, and there are a number of buried siliceous domes. Waiora Formation pumice and lapilli tuffs comprise the main producing aquifer. The overlying Huka Falls Formation is largely composed of mudstone and siltstone, with minor gravel and sand (e.g. Allis et al. 1997). The Huka Falls Formation has not been subdivided at Ohaaki. The buried domes include Ohaaki Rhyolite in the northwest, and Broadlands Dacite in the southeast.
3.5 Modelling Changes in historical and future pressures were input into the subsidence model at 1 year intervals for the period 1988-1996. The subsidence and pressure decline prior to 1988 were not incorporated in this model because there is insufficient survey data and pressure measurements. Therefore the model matches post-1988 subsidence, not the total cumulative subsidence to date (Figure 8). There are similarities and differences between Ohaaki and Wairakei, but application of the SKM model methodology has successfully matched the subsidence history of Ohaaki. In particular, the SKM model can account for the rapid increase and then decrease in subsidence rates, and the eastward migration of the centre of subsidence. The east sloping surface of the Ohaaki Rhyolite beneath the Huka Falls Formation means that as pore fluids are lost from the mudstones, the consolidation front will continue to move eastwards.
The geology is summarised in the cross section in Figure 7, which passes through the centre of the subsidence bowl. This section was constructed from data presented as well logs and cross sections by Lovelock (1990). As at Wairakei, it is the compressible and relatively impermeable mudstones within the Huka Falls Formation that are compacting. Huka Falls Formation sediments are about 300 m thick in the east, but thin to nothing over the highest parts of the Ohaaki Rhyolite.
The period in the 1970s between discharge testing and production provides a real indication of the effect a longterm shut down would have on fluid pressures and on subsidence. The gradual recovery of reservoir pressures was accompanied by a slowing of subsidence rates.
3.3 Material Properties Preliminary modelling was carried out using the same set of material properties that was used for the Wairakei and Tauhara models. The model was then modified to incorporate material properties indicated by laboratory testing of samples from Ohaaki by Read et al. (2003).
3.6 Effects Although the subsidence rates have been comparable with Wairakei (for a few years at least), and the total subsidence is similar to Tauhara, there are few structures on overlying land, which comprises farmland and forest areas. The most serious effect is flooding of Ohaaki Marae and surrounding land by the Waikato River. Bloomer & Currie (2001) also reported tilting of separation plants 1 and 2 by up to 3%, and damage to pipelines and wells.
3.4 Reservoir Pressures Deep liquid pressures declined by up to 13 bars during the test discharge period between 1968 and 1971, but only by about 8 bars in shallow wells (Clotworthy et al. 1995). Most of this pressure drop (8 bars) was recovered over the next five years, largely by recharge from shallow aquifers (Allis 1982). After commissioning of the power plant in 6
Elevation (m rsl)
White et al.
Distance (m)
Figure 7: Geological cross section through Ohaaki (from Lovelock 1990) Year 1987 0.0
1988
1989
1990
1991
1992
1993
1994
1995
1996
Subsidence (m)
0.2 0.4
0.6 0.8
1.0 1.2 Measured
Model Match
Figure 8: Historical matching for benchmark H338/1, Ohaaki. The location of this benchmark is shown in Figure 6. A separate development for power generation is now under way in the eastern part of the field. Production drilling commenced in early 2004, and a 50 MW development is planned.
4 KAWERAU The Kawerau geothermal field is about 80 km northwest of Wairakei. This field has been in production since 1957. Steam is supplied to paper and saw mills and used for electricity generation (equivalent to 40-45 MWe total). Resource consents permit the taking of up to 44,400 tpd. Initial shallow production has gradually been replaced by deeper wells. Prior to 1991, all brine was disposed of to the Tarawera River. Reinjection began in 1991, increasing to about 25% of total brine flow by 1997. Reinjection is to two shallow (
