Product has been added to the basket

How to make a volcano

Drowned village - LUSI's victim. All photos courtesy, Dr Adriano Mazzini.

The world has a new mud volcano, and called it LUSI. But who has unleashed this terror – nature, or man? Ted Nield reports on the continuing debate.

Geoscientist 18.6 June 2008

It may not have been exactly paradise before, but at 0500am on Monday 29 May 2006 the village of Sidoarjo became an inferno as, just 200 metres from the head of a hydrocarbon exploration well in East Java, Indonesia, water, steam and gas at 100°C began erupting from the ground. A new mud volcano had been born. Dubbed LUSI, she derives the first two letters of her name from “Lumpur” (which means “mud”) and the last two from “Sidoarjo”, whose tiled roofs no longer show above the stinking mud that has engulfed them.

Since first breaking surface, LUSI’s eruptions have continued for two years, smothering Sidoarjo under as much as 180,000 cubic metres of toxic sludge every day. By May last year, mudflows had spread over 6.3 square kilometres, despite several makeshift dams. The number of persons displaced had risen to about 30,000; 25 factories had been abandoned, and rice paddies and shrimp ponds destroyed - all at an estimated cost of nearly 11 million Rupiah.

Like a lanced boil, as the mud, steam and gas erupted, the ground in the area subsided – in some places by several metres, fracturing the Pertamina gas pipeline and causing an explosion that killed 13 people and a fire whose heat was reportedly felt up to a kilometre away. Rail lines and motorways were cut and other infrastructure put out of action. Extending way beyond the area of the mudflows, since the eruption began, an elliptical zone 22 square kilometres in extent has been sinking by between one and four centimetres per day. 



Two days before the LUSI eruption began, the south coast of Java was hit by an earthquake of magnitude 6.3. Although somewhat eclipsed by subsequent news, this earthquake killed many more people than the subsequent eruption – over 6200 – and put over 1.5 million people on the streets. From the beginning opinion has been divided on the role of this catastrophe in subsequent events. To what extent might it have been responsible for the eruption? This is a question in which the operators of the nearby well, Banjar-Panji 1 (BP1), PT Lapindo Brantas, took a keen interest (see box).

With legal proceedings and a full-blown media campaign by the operator going on in the background, geologists’ efforts to understand the nature of the beast that had been unleashed on the citizens of Sidoarjo have been closely watched as they piece together the sequence of events leading up to the eruption of 29 March 2006. Early reports favoured the “drilling accident” scenario; but the “earthquake-trigger” hypothesis has not lacked geological supporters.

One thing everyone agrees on however is this. Whatever caused it, LUSI is bigger and nastier than any other mud volcano in the region, and interestingly combines elements of both mud volcano - and geyser. So what exactly do we know about the series of unfortunate events that led to this disaster?


LUSI is a “pioneer mud volcano” because none was previously known in the immediate area. This is rather worrying because many geologists believe that once established, a mud volcano system can become self-perpetuating – the subsidence of the overburden causing cracks to open up, which in turn develop into new mud volcanoes.

LUSI lies in the East Java Basin, an inverted extensional basin made up of a number of east-west striking half graben that developed in the Paleogene and which later underwent compression in the Miocene to Recent. Since the Oligocene, thick piles of shallow marine limestones and marine muds were laid down, some of which are overpressured. These were folded into gentle anticlines and synclines, with normal and reverse faults cutting the anticline crests.

One of these crests was the target of the Banjar-Panji 1 well, which was prospecting for natural gas in Kujung Formation carbonates (Oligo-Miocene). Passing first through Pleistocene Pucangan and Kabuh Formations, BP1 then entered ~1km of overpressured muds (Kalibeng Formation, also Pleistocene); a further 1.3km of interbedded sands and muds. What happened next is disputed, but proponents of the human-trigger theory believe the well penetrated the overpressured limestones of the Kujung Formation. At this point, all are agreed, no casing lined the lower 1743 metres of the well, allowing free communication between the borehole’s fluids and not only the overpressured limestone, but also the entire thickness of interbedded sands and mud above it, and the overpressured muds above them.

It is known from nearby wells that the pore pressure in the Kujung Formation limestone aquifer is 48 Megapascals (MPa) (6970psi) at 2597m. In BP1, pore pressures 700m above the limestone are known to have been 38MPa (5500psi). Professor Richard Davies (Durham University), believes that at the depth that the well pierced the Kujung Formation, the overpressure would have amounted to 21 MPa. Davies and co-workers went on the record in January last year1 proposing that the penetration of the overpressured Kujung Formation caused a “kick” in the well, as the limestone’s pore fluids forced their way into the drilling mud, finding a readymade conduit between the highly overpressured pore waters in the aquifer and the unconsolidated muds lying above. Recent (unpublished) studies suggest that this kick occurred after a slight delay, as the string was being pulled.

Thereafter, Davies et al. believe, the route to surface veered away from the well itself and passed through the surrounding overburden. High pore pressures are known to cause natural hydraulic fracturing whenever they exceed the rock’s fracture strength. This is highly likely when uncased wells pass through weak, poorly consolidated sediments, and probably lucky for the workers on the rig; as Davies and others believe that this prevented a straightforward blow-out. Such an event could well have resulted in every oilman’s worst nightmare - “cratering”, when the rig gets blown off the face of the Earth.

Richard Davies doesn’t think the earthquake could have been directly responsible for LUSI. “No other mud volcano eruptions were reported at the same time” he says. “The quake preceded LUSI’s eruption by two days, and when the earthquake struck nothing happened in the well. The well kick occurred the following day while they were puling the drill bit out of the hole.” On graphs of distance from epicentre versus earthquake magnitude, LUSI plots way above the line below which mud volcanoes are affected by seismic activity. Moreover, earthquake-induced liquefaction is seen in sands but less often in muds, which are more cohesive. The quake could have weakened the uncased well, but Davies thinks it would have been too big a coincidence for this to have also caused a fracture to surface 200m away, and provide the network of cracks extending down into the overpressured limestone that he believes would be needed to produce the eruption.

The combination of unconsolidated mud above an overpressured aquifer is a potent one. The aquifer provides a ready pressure-drive for the system, while the overlying sediment, easily entrained, can contribute an almost limitless supply of solids. Davies likens the system to the (mercifully much cooler and smaller-scale) mud springs at Wootton Bassett, near Swindon, England. Here, the pressured waters derive from Corallian limestones and entrain the Ampthill Clay on their way to the surface; but this connection is of course natural and did not require the intervention of either earthquakes or drillers to bring them together.

Boiling mud

Muddy water

Not everyone agrees with Davies, however. Dr Adriano Mazzini of the University of Oslo, and an international team of authors working in cooperation with the well’s operators, believe the case against the drillers is still unproven. Their paper, published in Earth & Planetary Science Letters in September 2007, emphasizes the role of natural conditions and processes in a back-arc extensional regime with high sedimentation rates, where thick overpressured piles of sediment are riddled with faults and folds, and replete with pressure-release structures leading to natural mud volcanoes. 

According to Mazzini et al., well-log data show no evidence that the all-important limestone, the Kujung Fmn., was ever penetrated – no trace of carbonate in the deepest cuttings, or calcimetry data – the latest of which recorded only 4% calcite. They also note that, days after the eruption, a fracture hundreds of metres long and tens of centimetres wide opened up in the vicinity of the (then still-operating) BJP1. This fracture was consistent with the local structural trend (mainly a NW-SE fault crossing the area, inferred from seismics and regional field observations).
Location map

Seismic trigger

The link between tectonic activity and the behaviour of geysers, methane emissions and mud volcanoes is well documented - even in areas thousands of kilometres distant from the epicentre. Delays of a few days have also been recorded between earthquakes and changes in eruptive activity. Faults and anticlines (plentiful in seismic profiles of the LUSI region) are commonly associated with vertical piercement structures, leading to natural mud volcanoes.

Mazzini et al. believe that a NW-SE fault, extending from the Penanggungan volcano, runs NE towards LUSI. Where this fault intersects a railway line, significant bending of the rails was observed after another earthquake, in September 2006, four months after the LUSI eruption. Seismic sections (acquired before the eruption) also show evidence of a vertical piercement structure, with surrounding strata bending upwards around the conduit zone. Mazzini et al. write that the available data support an earthquake-triggering hypothesis for LUSI. There may have been a long history of active vertical mud movement beneath LUSI, they say, and conclude that the May 27 event reactivated pre-existing fractures - which also explains a partial fluid loss observed in BJP1 10 minutes after the tremor. They also point out that neighbouring mud volcanoes showed variations in activity after the earthquake.

As Mazzini et al. see it, deep fracturing within already overpressured clay units occurred as a result of the quake. Fluids in overpressured units rose along these fractures; CO2 then exsolved, encouraging more vertical flow and leading to the bulk mobilisation of mud. Once the fluids reached 200m, they began to boil - resulting in self-sustaining surface eruptions that continue today, with geyser-like pulsation.

Mazzini believes it is important to distinguish between the "cause" and the "trigger" of the eruption. "We clearly showed that the cause of it is natural: old seismic profiles reveal that a subsurface piercing feature already existed at this location. This feature is situated on a fault that crosses the Java Island with a SW-NE direction. This fault hosts other mud volcanoes towards the north of the island. This type of setting and seismic [profile] is typical for mud volcanoes in other locations around the globe and is commonly targeted for hydrocarbon exploration. Azerbaijan is a classic locality where each mud volcano can accommodate hundreds of wells for hydrocarbon extraction" he says.

"The geochemical analyses … traced back the likely sources of water, gas and mud and, so far, there is no obvious evidence of deep fluids. There is little doubt that the earthquake affected the region and probably contributed to destabilise a system in an already critical condition. Propagation of fractures towards the surface allowed the overpressured fluids to rise, finally resulting in the LUSI mud eruption. Seismic events with the epicentre located few hundreds (and sometimes thousands) of kilometres away can trigger the eruption of “naturally prepared” mud volcanoes, and there is a large literature documenting this phenomenon. The continuous monitoring that we are conducting still shows a consistent increase in LUSI flow rate after earthquake activity."

"No kicks were recorded when drilling at 9297ft (at the bottom of the borehole). Davies's hypothesis is based on the fact that the drilling reached the supposedly overpressured limestone formation. Nevertheless there is no real data proving this" he says, adding however: "I cannot exclude that the drilling contributed to the eruption. Nevertheless I could not see convincing data to support this alternative."

Mazzini abides by the verdict of Mazzini et al., where the authors wrote: “no kicks were recorded at the bottomhole of BJP1”. Davies disputes this. “The kick is a fact – Lapindo confirm it occurred” he says, pointing out: The kick occurred when they were pulling out of hole after reaching 2834m – our ‘model’ does not require that they reached the Kujung – but we think that it’s the most likely source for the water. Our proposal that there was a kick…is not dependent on the Kujung limestone."
There goes the neighbourhood

Run and run

Rather like LUSI herself, this debate looks set to run and run. Meanwhile around Sidoarjo, plans proceed to minimise the damage. A “relief well” being drilled to intercept the flow and perhaps allow it to be staunched, has been abandoned before reaching its target depth – though not before Lapindo had sunk $40 million into it. For some time, erupting mud has simply been channelled into a nearby river (with unknown environmental consequences downstream). The latest plan is to encircle the world’s newest mud volcano in a curtain wall – known as “the bund”. This mighty structure will be 120m in diameter, with walls 10m thick and 50m high.

Will the bund prove enough to contain the beast unleashed on Sidoarjo for the rest of its natural life? Only time will tell. 

LUSI's Eruptive mix

The temperature of the erupting fluids indicate, says Professor Richard Davies, that they originate at between 1.5 and 3km down. Measurements of the temperature of the erupting mud peak at 97°C, which together with the boiling water and steam at the centre of the crater seems to confirm a top temperature of 100°C. The geothermal gradient is very high locally (42°C/km), which in view of the proximity of a volcanic arc is perhaps not surprising.

Extensive sampling and analysis by Mazzini et al. has revealed that the erupted gas consists mostly of water vapour. Of the other gases evolved, researchers found highly variable amounts of carbon dioxide (9% - 74%), methane (83-85%) and hydrogen sulphide.

The expelled water is noticeably less saline than seawater (39% lower), with sulphates and magnesium similarly reduced. However boron and calcium were both enriched. The gas isotopic composition supports the notion of a mixed biogenic and thermogenic origin; the carbon dioxide being microbial in origin. The H2S content is also suggestive of an origin in rocks rich in sulphates and organic matter.

However the rapid variation in the eruptive mix’s composition suggests, say Mazzini et al., a varying origin and a complex and constantly changing plumbing system. The mud mineralogy confirms a depth of between 1.6 and 1.8 km – and certainly no shallower than 1.2km. The low salinity of the fluid, is thought to be due to dilution by water derived from deep diagenetic rather than meteoric sources. Enrichment in 18O indicates that this deep diagenetic water comes from clay mineral dehydration rather than silicate alteration (which would have consumed rather than produced 180).

LUSI - Balls dropped

One of the most bizarre turns of events in the whole episode has been an unsuccessful attempt to strangle LUSI with a ball and chain….

“Volcano consumes concrete balls”

(Agence France Presse headline, 27/02/07 )

In February 2007, groups of four chain-linked concrete spheres, two 20cm and two 40cm in diameter and coated with anti-corrosive chemicals, were dropped into the erupting volcano in an attempt to narrow its neck and reduce the flow. Phase 1, which began on February 24, involved 374 groups of balls; and Mr Basuki Hadimuljono, in charge of the operation, was quoted as saying that up to 1000 such chains might be required before the technique would work. In the event only a further 25 clusters were dropped before the attempt was abandoned.

The balls were the brainchild of Bagus Nurhandoko, Satria Bijaksana and Umar Fauzi, of the Bandung Institute of Technology and cost Lapindo Brantas four billion Rupiah (US$ 440,000). They achieved a brief hiatus in LUSI’s eruptive pattern, lasting about 30 minutes; but although the attempt to stem LUSI’s flow was unsuccessful, the data retrieved were not without interest or utility. Many clusters plumbed depths greater than 300m, and some achieved the astonishing depth of over 1km.

LUSI - Smooth operator

The operating company on the Banjar-Panji 1 well was PT Lapindo Brantas, which owned 50% of the concession rights in the Porong District under the Brantas Production Sharing Contract, together with two other companies: Santos (18%) and MedcoEnergi (32%). Early support for the “earthquake trigger” theory came from one Aburzial Bakrie, at that time the Indonesian government Minister of Welfare. These statements caused little surprise in Indonesia. PT Lapindo Brantas is part of the industrial conglomerate PT Bakrie & Brothers Tbk - which has made the Minister’s family the sixth richest in Indonesia, with an estimated net worth of $1.2bn.

Later, the Bakrie conglomerate attempted to sell PT Lapindo Brantas to an offshore company in September 2006 for $2. This, and a subsequent deal involving a larger sum, was blocked by Indonesia’s Capital Markets Supervisory Agency.

The end

Further reading

  1. Birth of a Mud Volcano: East Java 29 May 2006. Richard Davies, Richard Swarbrick, Robert Evans & Mads Huuse: GSA Today February 2007 v 17 no 2, 4-9
  2. Triggering and dynamic evolution of he LUSI mud volcano, Indonesia. Mazzini, A, Svensen H, Akhmanov G G, ALoisi G, Planke S, Malthe-Sørenssen A and Istadi B. Earth & Planetary Science Letters v. 261, 3-4, 30 September 2007 pp375-388.
  • All photographs courtesy, Dr Adriano Mazzini