2. HEAT- AND MASS TRANSFER THROUGH LUSI
The heat- and mass transfer processes which occur at LUSI have to be defined to understand the importance of parameters which can be monitored. For this we use the analogue of treating the discharges of the mud volcano as features of a low temperature geothermal system where heat and mass is transferred from a reservoir to the surface involving buoyancy forces.
2.1 The Geo-Pressured Reservoir and Its Reservoir Fluids
Data from the BJP-1 well and the c. 1 km deep adjacent relief wells RW01 and RW02 (Mazzini et al., 2007, Davies et al., 2008, Tingay et al., 2008) were used to obtain a geological section of the geo-pressured reservoir encountered by the wells (see Figure 2). The data show that part of the reservoir consists of under-compacted and overpressured bluish-grey clay deposits (plastically deformable) which were encountered in the uncased section of the well between c.1.2 and 1.85 km depth. Here, formation pressures of c. 20 and c. 30 MPa respectively prevail. The associated anomalously high pore pressures are contained within a sequence of rather impermeable layers. Sonic and neutron logs point to layers with high porosities (up to 35 %) at these depths. Geo-pressured volcano-clastics with average porosities of c. 20% occur from 1.85 km depth to bottom (c. 45 MPa pressure and c. 135 deg C near bottom hole). Microfossils in the mud ejected at the beginning point to initial mud source depths of c. 1.2 to 1.8 km. The buoyancy force required to lift a geo-pressured mud mixture via a 1.6 km quasi-vertical feeder system to the surface is c. 27 MPa. This would allow discharge of a gas-rich mud/ liquid mixture with a density of < 1730 kg/m3. Assuming a particle density of 2,500 to 2,600 kg/m3, a total liquid content of > 50% is indicated for this limit.
All geo-pressured sediments are fully saturated. The free liquid fraction of the liquid/mud mixture discharged at the surface appears to have changed with time. Initially c. 60% (by volume) of the discharge was hot water; this proportion has decreased with time (c. 40 % in 2008). The values, however, are only estimates. Chemical analyses of water samples collected during 2006 have shown that the discharged liquids derive from marine pore fluids (their content of 20 g/kg NaCl is equivalent to a c. 60 % paleoseawater content).
The surface characteristics of the LUSI mud-volcano are anomalous if compared with those of other mud volcanoes exposed over geo-pressured systems elsewhere (Dimitrov, 2002; Kopf, 2002) where cooler, rather viscous mud discharges produce mounds of mud. Boiling temperatures at the surface have not been reported for any of these features. The temperatures at the bottom of the BJP-1 well are also anomalously high if compared with temperatures in other geo-pressure systems where temperatures of c. 135 deg C occur only below 4 km depth (Griggs, 2005).
Figure 2: Simplified geological section of strata encountered by the 2.83 km deep BJP-1 exploration well showing spot temperature and pressure data at given depths and inferred fluid discharge paths.
2.2 Surface Manifestations and Discharge Characteristics
Irregular discharge of hot liquid mud has been confined to the central eruption centre. A set of vertical satellite photos, published since 07.08.06 on the CRISP website, constitute now the most important monitoring resource data (see Figure 3). The volume of discharged steam can vary by an order of magnitude during a time span of a few weeks. Steam has always been discharged by a set of small, bunched steam vents that occurred within an inner area with a diameter di of c. 50 m towards the end of 2006. This inner area is surrounded by a circular, light-grey coloured mud ring caused by an up-flow of gas bubbles (see inset of Figure 3). The diameter (dm) of the upwelling mud area increased continuously from c. 50 to 90 m in 2006 and fluctuated between c. 75 and 130 m during 2007 and 2008. The discharge of steam from vents in the inner area (with diameter di) can be compared with that of fumaroles over high-temperature geothermal systems (Hochstein and Bromley, 2001). Wafting steam over the outer discoloured mud zone is the result of evaporation which is enhanced by the rough surface caused by the continuous ‘popping’ of gas bubbles. Minor and irregular mud eruptions of the central ‘crater’ have occurred during the whole time.
Gas is discharged by the central ‘crater’, by vents in the
flooded areas (concentrated discharge) and also outside the flooded areas through the soil (diffuse discharge). The gas discharge (mostly CO2 and CH4) has not been monitored yet. Within the flooded area, discharge is ephemeral. CRISP satellite photos show that during 2008 concentrated gas discharges waxed and waned at nine sites in the flooded area between the central crater and the W boundary dam.
Seven ebullient gas discharge centres covered a total area of c. 3,500 m2in October 2008. All gas discharges outside the central crater are cold discharges due to volume expansion near the surface.
Figure 3: Satellite (IKONOS) photo (05.06.07) of the area flooded by the LUSI mud flow. Shown are also contours of subsidence rate (m/yr) from an INSAR analysis of ALOS satellite data (taken from Abidin et al., 2009). The LUSI discharge centre is enlarged in the right hand inset showing the steam cloud plume, the outer gas-discharging mud ring (slightly lighter grey tone) and the horse shoe shaped inner, circular dam.
Surface temperatures of the upwelling mud in the central part of the crater are close to boiling; spot readings between 94 and 97 deg C have been reported several times. Daily temperature measurements during a 20 day period in June 2008 recorded mud temperatures between 88 and 110 deg C (mean of 94 +/- 5.5 deg C). The maximum was observed after a large, distant earthquake had occurred (Istadi, pers.comm.). The surface mud temperatures decrease rapidly from c. 94 deg C in the centre to c. 40 to 45 deg C at the edge (dm) of the de-gassing mud ring.
Temperatures at c 1.7 km and 2.7 km depth in the deep well point to mean vertical geothermal gradients of c. 42 and 39 deg C/km respectively. The large gradients are most likely due to a rather low thermal conductivity of the highly porous, liquid saturated reservoir rocks. The fast equilibrating (K/Mg) geo-thermometer, applied to analyses of fresh liquids from the crater (cited in Mazzini et al, 2007), yields equilibrium temperatures of c. 80 to 120 deg C which correspond to fluid temperatures prevailing between c. 1.2 to 2.3 km depth. Isotope data of the same samples show large δ O18shifts similar to those of young palaeo-pore fluids with high marine content as encountered, for example, in Gulf Coast (US) oil and gas wells (Faure, 1986).
Because of the affinity of observed isotope data with those of pore fluids of gas wells in non-volcanic settings, a volcanic heat input is not required to explain the anomalous temperatures in the BJP-1 well.
2.3 Steam Discharge
Discharge of the hot, liquid mud is associated with several heat transfer processes. One of the processes involves boiling of ascending pore waters at shallow depth beneath the crater and the discharge of a steam plume over the crater.
The plume contains not only steam and gas, but also air which enters it at the bottom during steam ascent. Studies of fumaroles in NZ by Hochstein and Bromley (2001) have shown that projected steam-cloud areas are proportional to the total steam-cloud volume Vc which, in turn, is proportional to the heat loss ∆Qc. Meteorological parameters, such as air pressure, air temperature, humidity, and wind speed, all affect the steam cloud volume.
A schematic view of the steam discharge over the crater is shown in Fig.4a; a summary of all parameters used for the analysis of heat transfer by steam and mud discharge is listed in Fig. 4b.
For a given constant heat output of ∆Qc, the steam cloud volume Vc varies with air pressure (p0), air temperature (T0), and relative humidity (rh0) from day to day.
However, these parameters did not change much between the days when the published IKONOS satellite photos were taken (always recorded during cloudless days and c. 90 min before meridian time).
Figure 4a: Perspective view of LUSI mud discharge centre and characteristics of mud flow and central steam-cloud used in the text.
Figure 4b: Legend of parameters and symbols used in Figure 4a.
Using meteorological data from the nearby Surabaya airport, it was found that p0 showed the smallest changes (less than
0.2 % with respect to an annual mean of 1010 mbar at noon).
The air temperature T0 also showed little variation since during nine different satellite crossings in 2007, for example, T0 differed at noon by less than +/- 2 deg C from the annual mean of 28 deg C.
Only the relative humidity rh0 showed slightly larger variations (c. +/- 5% with respect to a mean of 75%). For the steam cloud shown in the inset of Fig.3 the values at 10:30 on 05.05.07 were: p0 = 1009 mb, T0 = 26.5 deg C, rh0 = 79 %.
Hence, steam clouds shown on the IKONOS satellite photos taken at different months over the LUSI mud volcano can be compared allowing for reductions of Vc with respect to T0 and rh0.
The volume Vc can be approximated by:
Vc = F ac sc (1)
Where F is a dimensionless form-factor (significantly affected by windspeed), ac the cross-sectional area perpendicular to the long axis of the plume at mid height, and sc the slant distance of the long axis of the plume.
Assuming that F is close to unity, the diameter dc of area ac is proportional to the cloud volume Vc. Using the zenith angle αz, the projected length of the cloud on the ground (lc),and the shadow length (ls), the approximate height of the steam cloud (hc) can also be obtained.
Following the approach of Ryan et al. (1974), it was assumed that the heatflux due to evaporation at the base of the steam cloud is proportional to the vapour pressure difference (es – eb) where es is the saturation vapour pressure at the mud surface of the crater and eb the water vapour pressure in the atmosphere close to the base.
Assuming that es is almost constant, changes in Vc and ac are therefore proportional to changes with respect to eb of a ‘reference’ steam cloud with the average annual parameters of p0, T0, and rh0 listed above.
The reduction can be simplified by considering only changes applied to the diameter dc, a parameter which is also proportional to Vc.
The effect of time-variable changes in p and T was found to be small and was neglected.
The reduction of the rh0 effect can also be applied to the projected cloud area Ac, i.e the area visible in the satellite photo, which, in turn, is proportional to the heat loss ∆Qc (Hochstein and Bromley, 2001).
An assessment of satellite photos taken, for example, during 2006 at days with little wind showed that Vc values of the steam plume over the LUSI crater, reduced for humidity effects, varied between c. 1,000 and c. 500,000 m3on 09.10.06 and 06.12.06 respectively.
This points to a larger than two orders of magnitude variation within a 2-month period. The reduced area ac of the plume at mid-height varied between c. 100 and c. 3,000 m2for the two days. Using the zenith angle of the sun and shadow length of the steam clouds, the height hc of the plumes was found to be c. 10 and c. 250 m respectively.
Comparing these values with reduced NZ fumarole data, a heat loss ∆Qc on the order of 3 and 100 MW respectively is indicated for the two cases. Satellite photos show that the large value for the 06.12.06 steam cloud was exceeded only a few times in the satellite photo sequence during the next two years.
Large changes in Vc can occur even within a week as indicated by changes in steam cloud characteristics on satellite photos taken on
17.09.06 (Vc: 1,000 m3) and 25.09.06. (Vc: 150,000 m3). Such variations appear to be the result of pulsating ‘slug’ flow discharges.
2.4 Volume-and Mass-Flow Rates of Liquid Mud
Rates of discharged mud volume (in m3/d or m3/s) were estimated during the first 11 months from accumulated mud thicknesses in the flooded areas.
The first estimates appear to be the more accurate ones. The specific density of the mud increases with time due to dewatering and mud compaction.
Thickness estimates are also affected by subsidence which induces some groundwater to reach the surface. For June/July 2006, the original, average mud flow estimates were c. 0.5 m3/s which increased to c. 1.1 m3/s and 1.4 m3/s in August and September 2006 respectively (neglecting densification).
For May 2007 the last estimate was c.1.3 m3/s. Pulsating discharge activity was noticed between middle of August and middle of September 2006 when discharge rates could vary by half an order of magnitude during a period of a few days.
The LUSI mud management policy was aimed at changing the radial mud outflow to a sectorial outflow (to the south).
To achieve this, a high-standing, semi-circular dam was constructed (completed by the end of August 2006) which directed the flow to the south (see Figure 3); mud was allowed to accumulate to form a c. 7 m high, broad mound but increasing mud viscosity retarded outflows.
From mid 2007 onward the deposition pattern of the mud flows was changed when excavators were used on the inner dams to promote mud movement towards the Porong Canal (at the S margin of Figure 3).
Mud outflow was also controlled by maintaining a rather high (surface) level of the central crater by continuously raising the crest of the confining circular dam to compensate for ongoing subsidence.
Initially, separated brine was drained by a new canal in the south. After its silting, mud-slurry was pumped from stations near the S dam into the Porong Canal.
This required ‘thinning’ of the mud by adding river water to reduce the solid components of the slurry to c. 45% (per volume unit).
The total daily capacity of the slurry pumps was c. 120,000 m3/d when working in 15 hr long, daily shifts.
The removal of c. 100,000 m3/d of ‘original’ mud was equivalent to a maximum slurry removal rate of c. 1.15 m3/s.
Average monthly pumping data of mud slurry for the period of May 2006 to December 2008 (reduced by a factor of 0.87 to allow for mud dilution) are shown in Figure 5b.
Mudflow management by slurry pumping constitutes an important monitoring procedure.
Details in satellite photos from the CRISP website can be used to infer trends in energy transfer via the steam plume (∆Qc) and the outflowing mud column (∆Qm).
The equivalent diameter dc of the steam plume at mid-height and the quasi-circular, upwelling mud area (with diameter dm), are such parameters. Plotting the diameter dc of the reduced cross-sectional area ac together with the diameter dm of the upwelling mud versus time, shows an acceptable correlation (see Figure 5a).
These data can be compared with the inferred and observed mud discharge rate for the first 12 months of the LUSI activity (see Figure 5b).
It appears that the average mud discharge rates reached a peak in December of 2006 (c. 1.85 m3/s) as a result of increasing up-flows stimulated by a sequence of earthquakes.
It caused a huge mud eruption around 06.12.06 that breached the semicircular, inner dam.
Mud discharge rates probably fluctuated between 1 m3/s and
1.5 m3/s during 2007 and 2008 as indicated by the trend of the diameter dm of the upwelling mud from January 07 until December 08 (see Figure 5a).
Peak values of dm (c.120 m) were reached during March to June 2008 when mud discharge rates also reached their peak (probably c. 1.5 m3/s).
Discharge rates ∆Qc (via the plume) are proportional to the reduced cross-sectional area a c and its equivalent diameter d c and point to (order of magnitude) variations between c. 15 and 100 MW from September 2006 to August 2007 - with a generally declining trend from January to December 2008. In between (September to December 2007), heat transfer by steam also declined.
Using the inferred mud discharge rates shown in Figure 5b, approximate subtotals of c. 20 and c. 40 Mill m3/yr of discharged mud are indicated for the years 2006 and 2007/2008 respectively (i.e. total of c. 100 Mill m3from June 2006 to December 2008).
However, a total of c. 35 Mill m3of mud was removed from the originally flooded area by mud pumping during 2007 and 2008. The net volume of mud deposited in the flooded LUSI area until December 2008 was therefore c. 65 Mill/m3, equivalent to an average (unconsolidated) mud thickness of c. 10 m over a flooded area of c. 6.3 km2.
2.5 Heat Discharge and Transfer Rates
The total anomalous heat transfer of the LUSI mud-volcano is approximately given by the heat discharged via the steam cloud (∆Qc) and that by the hot, upwelling mud (∆Qm ).
The magnitude of evaporative and conductive losses from the hot mud surface can not be assessed yet. However, the anomalous heat discharged by the upwelling mud can be assessed separately for its liquid and its solid portion.
If we assume, for example, a mean surface discharge temperature of 95 deg C, a reference mean annual T of 28 deg C, a particle density of 2,500 to 2,600 kg/m3 for the solids, an initial (near surface) liquid proportion of 50 to 60%, and appropriate specific heat capacity values for liquid and solids, this points to an anomalous heat discharge ∆Qm of c. 215 MW per m3of liquid mud.
For an inferred maximum discharge rate of c. 1.5 m3/s, a heat loss of c. 320 MW is indicated. For the large eruption period around 06.12.06, when the inner circular dam was breached and ∆Qc was .>100 MW, a total of at least 400 MW would have been transferred to the surface.
Figure 5a: Plot of diameter dm of the upwelling hot mud ring (crater region of LUSI mud discharge) versus time (August 2006 to December 2008). Also plotted is the equivalent diameter dc of the cross-sectional area ac of the steam cloud, reduced for changes in relative humidity.
Figure 5b: Plot of average and inferred (mud) volume discharge rates (in m3/s) versus time (August 2006 to December 2008) of LUSI mud flow. Also shown are mud removal rates from pumping (reduced for mud ‘thinning’) versus time for the period June 2007 and December 2008 (courtesy Mr.Soffian, BPLS).
Although representative mud discharge rates are not known since April 2007, it can be inferred from earlier observations and trend data cited in this paper that the present-day heat loss is still > 200 MW.
The hot liquid mud, in turn, cools by evaporation and conduction. Together with expansion cooling of rising gas, this reduces surface temperatures from c. 95 deg C in the inner discharge area to c.40 to 45 deg C just outside the upwelling mud ring.
Near the pump stations it is further cooled by addition of cool river water. The mud cooling pattern is visible in IR satellite photos.
2.6 Gas Discharge
Gas discharge is significant, both as diffusive and concentrated discharge. Widespread diffusive discharge occurs in settlements along the W and SW boundary dams (i.e. outside the flooded area) and has led to their partial or total evacuation. Some concentrated gas discharge occurs not only at the central crater but also at several centres in water-covered stagnant pool areas.
In the western flooded pond area, water/gas upwellings could be observed, driven by escaping CO2 and CH4 gas (see example in Figure 6).
The total area of seven centres with ‘ebullient’ gas discharge in the northern ponds was at least 4,000 m2in October 2008. Ebullient discharges visible on satellite photos showed an increase during 2008 but disappeared in early 2009.
Smaller gas discharges have been observed at smaller sites which appear to be at or close to inferred fractures. However, none of the gas discharge centres in the western ponds, active during 2008, were close to documented smaller eruption sites mapped in June 2006.
Figure 6: Photo of ebulliant gas discharge centres near the western boundary dam (visible in the back);
October 2008.
Specific gas and steam flux data have not been obtained yet. Gas samples were taken during the second half of 2006 from the crater steam plume together with some gas seeps and were analysed in terms of non-condensable gas species but not for water vapour (Mazzini et al., 2007).
Other reports refer to estimates of non-condensable gases by ‘gas-sniffer’ instruments. Assuming that most gas over the crater comes from deep fluids where temperatures between 100 and 135 deg C and pore pressures between 30 and 45 MPa prevail, the solubility of both major gases in pore fluids can be estimated by using, for example, solubility data cited by Lu and Kieffer (2009). Allowing for salinity, a discharge of c. 1.5 m3/s of liquid mud could produce maximum CO2 and CH4 mass discharge rates of c. 2 kg/s and 0.7 kg/s respectively.
The likely total volume of greenhouse gases discharged by the LUSI mud-volcano is therefore large and should be assessed in more detail. A ground based gas-flux survey, especially for areas within already evacuated villages outside the dams, is indicated.