By Mark Tingay, @MarkTingay
|
The Lusi mud volcano, shortly after it's birth on the 29th of May 2006, with the TMMJ-04 rig that was drilling the Banjar Panji-1 well sited less than 150 meters away. |
Foreword:
This is the third part of a special blog to mark
the 10th anniversary of the ongoing Lumpur Sidoarjo mud volcano
disaster. Part 1 examined the background and geology of the Lusi mud volcano.
Part 2 examined the arguments for and against the proposed earthquake trigger
model for Lusi, and concluded that there is very little evidence to support
that hypothesis.
When I started this ‘Lusi trilogy’, I envisaged the third
part would look at all aspects of the ‘drilling trigger’ model, and the
evidence for and against this hypothesis. However, this part of the story
quickly grew and grew to an ‘epic’ length. Furthermore, I got side-tracked from
finishing this post by a long series of events that included being made
redundant (boo!), falling ill (boo again!), starting a new business (yay -
insert Rocky theme music here!), having my two PhD students submit back-to-back
(awesome effort guys – but reading and editing 13 thesis chapters in 3 weeks
was a big ask!) and seemingly half the journal editors on the planet swamping
me with a pile of really interesting manuscripts to peer review (seriously, how
did all you all know I was unemployed?). So, I have decided to ‘Harry Potter’
this final Lusi Blog part and break it into two sections, both because of the
length this is becoming and also the horrible delays I’ve had in writing this
blog.
The first section of this epic finale to the Lusi triggering
story will set the scene. Drilling accidents don’t happen because of one event.
Their root cause is always a number of decisions and events that may start
weeks or months prior to the disaster, but which ultimately culminate in things
going suddenly and spectacularly wrong. It is no different with Lusi. In this
blog post, I will go through the many key aspects of drilling the Banjar
Panji-1 well, in the months and weeks prior to the disaster, that resulted in
the borehole being in a precarious condition. Whilst my next and final Lusi
blog section will carefully dissect the key 48 hours leading up to the
disaster, and the 3 days after the Lusi mudflow started. This blog section herein is necessary because
it will highlight the key incidents and decisions that, ultimately, meant that
the major kick, when it happened, could not be controlled.
Note also that more details on mud volcanos and Lusi can be
found by looking back on my twitter feed from May 2016 (@MarkTingay,
#lumpurlapindo).
Introduction and Background
The Lusi mud volcano suddenly appeared and erupted in a rice
paddy on the morning of the 29th of May 2006. Less than 150 meters
away, the TMMJ-04 rig was drilling the Banjar Panji-1 (BJP-1) gas exploration
well, which had been experiencing significant well control problems over the
previous 2 days, including suffering a major kick that erupted over 360 barrels
(>15000 gallons) of mud and water at the wellhead less than 24 hours
earlier. The drillers on BJP-1 immediately assumed that the adjacent mud
eruption was caused by the well control issues that started the previous day,
and made several attempts to kill the Lusi mud flow, as well as calling in
experts from a major well control company. However, just 2 days later, the operator
of the BJP-1 well, Lapindo Brantas, decided to abandon the well, and started
the ~36 hour process of closing the well and pulling the rig off location.
Within a week, one of the partners in the well, Medco, had instigated legal
proceedings against Lapindo Brantas for gross negligence in the drilling
campaign, and the Indonesian government and police had also commenced
investigations into the rapidly unfolding disaster.
From my examination of media reports and early technical
reports, there seems to be little doubt that Lapindo Brantas felt initially
(for several weeks, up to perhaps two months) responsible for the disaster,
including agreeing to cover costs associated with initial damages. Yet,
sometime during the following months, Lapindo started making claims that the
Lusi mud volcano was entirely unrelated to the BJP-1 well, and was instead the
result of entirely natural processes triggered by the Yogyakarta earthquake 274
km away (see Part 2 of this blog). Perhaps it is entirely coincidental that
these new claims of innocence came around the time that Lusi became
spectacularly larger, more violent and vastly more destructive in August 2006,
not to mention being indirectly responsible for the deaths of 13 people (mostly
police) in a pipeline explosion in November 2006, and the significant political
connections of a senior Indonesian politician to Lapindo (or maybe I am just
being cynical).
The debate about Lusi triggering in the wider scientific
realm began with the publishing of the first major peer-reviewed paper on the
disaster in early 2007 entitled “Birth of
a mud volcano: East Java, 29 May 2006”, authored by a group of UK mud
volcano and petroleum geology experts lead by Professor Richard Davies (Davies
et al., 2007). In this key first paper, the limited data available at the time
was compiled and analysed, and the authors concluded that the disaster was most
likely the result of a blowout in the Banjar Panji-1 well. More drilling data
started becoming available throughout 2007, and, it must be said, Lapindo
Brantas were quite willing to share some data (which is exemplary and very
unusual after a potential oil industry accident). This new data resulted in the
first two major papers purely examining the triggering of Lusi (looking at both
the drilling and earthquake trigger hypotheses) being published in EPSL and Geology in 2008 (Davies et al., 2008; Tingay et al., 2008). 2008
also marked the first (and only) international ‘great debate’ on Lusi
triggering, which was hosted by the American Association of Petroleum
Geologists at their International Conference and Exhibition in Cape Town (if
you want to read more about this, I recommend this excellent Blog post by Chris Rowan, a fantastic blogger who actively wrote updates about the Lusi
disaster for several years). Finally, in 2009 and 2010, there was another
series of papers examining the drilling-trigger hypothesis. The first was
written by Lapindo Brantas drilling engineers and geologists responsible for
the well, and was their arguments for why drilling did not trigger Lusi (Sawolo
et al., 2009). This paper is very important, as it provided a lot of the raw drilling
data in a public and peer-reviewed source (previously, data had been made
available to examine, but not necessarily to use). The Sawolo et al. paper was
challenged on many fronts by Davies et al. in 2010, and countered with a reply
by Sawolo et al. (2010) (though, it is worth noting that this reply did not
actually address any criticism of Davies et al., and instead just repeated the
claims of the 2009 paper). Yet, I have to say that, looking back now with a few
years more experience and more detailed examination of available data, we
(Davies et al., 2010) were too quick to write our reply. We missed a number of
key things, and, in particular, failed to spot how so many of the claims made
by Sawolo et al. (2009) are completely contradicted by Lapindo’s own reports
and data that were buried in a digital appendix to the paper.
In this post, and Part 3B to follow, I aim to provide a
summary of all the different claims surrounding the drilling-trigger hypothesis
for Lusi. However, I will apologise in advance that this can be a difficult and
tortuous subject! The deeper you go ‘down the rabbit hole’, the more you
realise just how often statements and claims are frequently made without
evidence (or in contradiction to evidence), how often key events seem to be
ignored, how data may be erroneous or be potentially manipulated and, in
particular, how some data can have two wildly different interpretations. In the
drilling of BJP-1, there seems to be debate and uncertainty over just about
every single little bit of data or events – and this makes it a very difficult
subject to explore! Furthermore, the potential legal implications of this
disaster necessitate being very careful to cross-check and carefully explain
all the evidence.
Herein, I have tried to make it clear whether, and what,
evidence is used to make claims and statements, and to highlight the
contradictions when they occur. Please note that, herein, I have used the
following assumptions when assessing the reliability of data and reported
events.
1.
Evidence and data is stronger if it is supported
in multiple places.
2.
Reports of events/data collected ‘on the day’
are considered more reliable than those made years later (once the extent of
the disaster was seen).
3.
Interpretations based on normal, widely accepted
and peer-reviewed practices are more reliable than those using non-standard or
non-peer-reviewed methods.
4.
Interpretations done in accordance with standard
drilling practices (e.g. seen in well control handbooks), and those taught by
formal qualification providers (e.g. the International Well Control Foundation),
are more considered more reliable and favoured over alternative approaches.
The Drilling
Triggering Model for Lusi
In Part 2 of this blog I highlighted that both the earthquake
and drilling-trigger models for the Lusi eruption are actually far more similar
than most people realise. Both models suggest that ‘something’ caused a reduction in effective stress (stress minus
pore fluid pressure) at the Lusi location, which, in turn, caused a fault or
fracture to initiate/reactivate and allow overpressured water (containing
clays) to escape to the surface. However, under the drilling-trigger model, the
drop in effective stress is argued to be due to the increase in fluid pressure
that occurred during a major drilling kick in the BJP-1 well. More precisely,
the increase in fluid pressure occurred when the drillers closed the Blowout
Preventer (BOP) at the surface of the well during the kick, which caused fluid
pressures in the well to spike as pressures equilibrated. The drilling-trigger
model is, in essence, an uncontrolled version of hydraulic fracture stimulation
(fracking), in which the fluid pressure in a well is increased to such a high
level that it fractures the surrounding rock formations.
|
Original Drilling Trigger model by Davies et al., 2007. Note the model assumes tensile fracturing during the BJP-1 kick, with kick fluids coming from the deep carbonates (please note that there are some geological errors in this figure - the 'sandstone aquifer' is actually tight volcanics and volcanoclastics, and the carbonates are not Kujung FM (which are actually Oligiocene), but are rather Miocene age and likely of the Tuban FM). |
Another common aspect of all drilling-trigger models for
Lusi is that the primary initial source of erupted water for Lusi is considered
to be the deep Miocene carbonates, believed to be located at the final depth of
the BJP-1 well (~2833m). In all drilling-trigger models, the highly
overpressured carbonates form both the primary water source and driving force
for the Lusi eruption, with fluids initially flowing up the BJP-1 borehole from
the carbonates, before passing into a fault or fracture somewhere inside the
Kalibeng clays, at which point the water mixes with the clays to become mud
before flowing up to the surface to erupt at Lusi. However, despite the main
commonalities, there are some slight variations between the drilling-trigger models
by different authors, as well as some evolution of the models over time (as has
been seen in the earthquake-triggering model in part 2).
Probably the biggest evolution or difference in the drilling-triggering
models is that the original models (and most models up until ~2010) proposed
that Lusi fluids escaped from the BJP-1 borehole to the surface via a purely
tensile (mode 1) fracture. These models were actually based on the occurrence
of similar blowout-triggered eruptions offshore Brunei in the 1970s that I had
studied during my PhD (Tingay et al., 2003). The idea of tensile fracturing is
significant, as it had been a well-established industry practice to assume that
wells will likely fail in tension in well-control events, and many debates
about whether or not kick pressures in Lusi were sufficient to cause fracturing
were based on this assumption. However, since ~2010, I, and some other
researchers, have suggested that the high kick pressures in BJP-1 were more
likely to induce shear failure (mode 2 fracturing), and, thus, either initiate
or reactivate a fault, rather than create a simple tensile crack.
|
Updated model for drilling triggering of Lusi, proposing that the increased borehole pressures during the BJP-1 kick were sufficient to induce shear failure (from Tingay, 2010).
|
There are several arguments to support the idea that the
kick triggered shear failure rather than tensile failure. First, shear failure
is more favourable given that fracturing appeared to initially be along an ~050°N
oriented structure, which is optimally oriented for shear failure in the
existing present-day approximately strike-slip stress-state and ~020°N maximum
horizontal stress orientation (whereas tensile failure should occur
sub-parallel to the present-day maximum horizontal stress; Tingay et al., 2010).
Second, the present-day stress state is actually far more conducive to
initiating shear failure than it is for tensile. As will be discussed in Part
3B, basic geomechanical models indicate that shear failure requires
significantly smaller increase in borehole fluid pressure than is needed for
tensile failure. Indeed, an interesting aspect of later Lusi drilling-trigger
models involving shear failure is that they are even more similar to the
earthquake-triggering model, which also assumes shear failure. Again, it should
be reiterated that there really are, in essence, very little difference between
then drilling and earthquake trigger models – they really only differ about
‘what’ caused the drop in effective stress that initiated/reactivated a fault,
and on the initial source of water for the Lusi eruption (deep Miocene
carbonates for the drilling-trigger, Kalibeng clays for the quake trigger).
Have Blowouts Ever
Caused Mud Volcanoes?
One of the most common arguments against a drilling-trigger
for Lusi is the misbelief that things such as mud volcanoes cannot possibly be
human-triggered. It is remarkably common to hear statements along the lines of ‘Lusi is a mud volcano, and mud volcanoes are
natural features, therefore Lusi must be natural’, or ‘drilling has never triggered a mud volcano’. Yet, these statements
are both false and illogical. Indeed, they are similar to the erroneous logic
discussed in Part 2, where people have argued that Lusi must be triggered by an
earthquake simply because other mud volcanoes have been triggered by
earthquakes. However, as discussed in Part 2, prior natural causes for events
do not preclude current human causes for the same events. Classic examples are
anthropogenic climate change, as well as human-induced seismicity (sure,
climate change and earthquakes happen naturally, but this does not mean they
can’t also be induced by human activity).
So, have mud volcanoes been triggered previously by drilling
accidents? The answer is yes – many times! Indeed, I have found at least three
other examples in Indonesia alone, such as the Dieng-24 blowout in Central
Java, the Gresik mud volcano (near Lusi) in 2008 and a recent mud eruption
triggered by geothermal drilling in Sulawesi. Furthermore, if we consider Lusi
to be simply be the escape of high pressure underground water (the water just
mixes with clay en route to the surface to become mud), then there are numerous
global examples of major water blowouts from wells. As mentioned above, the
initial drilling-trigger models for Lusi were based on a similar, albeit
smaller, series of blowouts in the Champion Field, offshore Brunei. In 1974,
and again in 1979, wells being drilled in the giant Champion Field encountered
unexpected highly overpressured compartments and suffered major kicks. In both
cases, the blowout preventer (BOP) at the surface held, and prevented major
fluid eruption at the rig site. However, the fluids flowing into the wellbore,
rather than escaping to the surface, instead started flowing into a shallow
normally pressured compartment, in what is known as an underground (or internal)
blowout. Gradually, the flow of deep pressured water into the shallow
compartment increased the pressure in the shallow compartment (think of
connecting an inflated balloon to a deflated balloon via a straw – air will
flow from the high pressure to low pressure until both balloons reach pressure equilibrium). After a few days, the pressures in the
shallow compartment got so high that the overlying cap rock fractured, and
allowed fluid to flow to the surface (termed a surface blowout resulting from
an underground blowout). In both Champion blowouts, the fluids erupted in many
places along a linear zones (like in Lusi), with some eruptions up to 5 km from
the responsible well. In the Champion Field, they were able to quickly use
other wells, and drill more wells, to contain the pressures underground and
stop the surface eruption – but the underground blowouts actually continued for
about 20 years!
|
Dieng Field mud eruptions triggered by drilling activity in Central Java, Indonesia. |
|
Shallow seismic time-slice (essentially horizontal) of induced fractures and associated lines of eruption points generated by the Champion Field 1974 and 1979 blowouts (Tingay et al., 2003). This is just one of many examples highlighting how blowouts can also cause aligned eruption vents, as was seen in the first few days of the Lusi mud volcano. |
There have been many other similar industry accidents, such
as the Unocal Platform-A blowouts offshore Santa Barbara (USA), Chevron’s Frade
blowout offshore Brazil and other events in Brunei and Azerbaijan. Hence, it
cannot be said that similar events to Lusi have not ever been triggered by
drilling. Furthermore, these other examples dispel another common argument that
is still routinely made by Lapindo Brantas, and in papers authored by Dr Adriano
Mazzini, which can be paraphrased as ‘Lusi
cannot be due to drilling, because the mud flow did not erupt at the well, but
rather ~150m away’ and ‘Lusi must be
natural, because eruptions initially occurred in several places along a linear
zone’ (Sawolo et al., 2009; Mazzini et al., 2009). For now, let’s just
ignore the fact that >360 barrels of watery mud did erupt at the BJP-1 wellsite during the kick (before the
BOP was closed), and note that there are numerous examples of wellbore blowouts
causing eruptions at a distance from the well (up to 5 km distant), and that blowouts
have also caused eruptions along extensive linear zones. So, these arguments
against the drilling-trigger model are invalid. However, it must also be stated
that, just like for the earthquake-triggering hypothesis, prior examples of
drilling-induced eruptions cannot be used as direct evidence to support the
drilling-trigger model for Lusi, and rather these can only be used to provide
precedence that such events can potentially occur.
The Banjar Panji-1
Well
The Banjar Panji-1 (BJP-1) gas exploration well was
targeting a Miocene carbonate reefal mound that was originally believed to be
at ~2600m depth. This reefal build-up is the western-most of a series of five similarly
aged reefal mounds that occur along a ENE-WSW line (Kusumastuti et al., 2002).
All four other mounds had been previously drilled by the BD-1, KE-11E, KE-11C
and Porong-1 wells (stated in east to west order), which had discovered either
small (non-commercial) amounts of oil and gas, or only residual hydrocarbon
accumulations. It was noted by several operators that the overburden near the
crests of these reefal mounds tended to be highly faulted, which led to the
belief that all four of these previously drilled structures were breached
traps. Indeed, some gas fields are hosted in shallow Pucangan formation
reservoirs within just a few kilometres of Lusi, such as the Wunut and Tangulangin
Fields, and it is widely believed that the gas in these field migrated out of
the breached Porong structure. Interestingly, the Porong structure, 7 km from
Lusi, was even previously noted to be overlain by a large ‘circular collapse structure’ that, in hindsight, is probably caused
by an earlier version of Lusi. However, whilst the crests of other mounds were
highly faulted, it was noted that the Banjar Panji reefal structure did not
appear to have nearly as much, or as intense, faulting in the overburden.
Hence, the play concept for Banjar Panji-1 was quite simple – it was believed
that the structure was an un-breached version of the nearby Porong, Kedeco and
BD structures, and thus would hopefully hold commercial oil and gas
accumulations (Istadi et al., 2009). It can perhaps be considered a cold
comfort that the structure did not seem to contain large hydrocarbon volumes,
or else Lusi could have been a far worse environmental disaster!
|
The Porong structure, drilled by the well Porong-1, located 7 km from BJP-1 and Lusi. Porong is the adjacent Miocene reefal high to the structure targeted by BJP-1, and porong-1 and BJP-1 are essentially sister wells. However, the structure targeted by BJP-1 does not have intensive faulting overlying it's crest. BJP-1 was thus testing the hypothesis that the neighbouring, and relatively unfaulted, structure would hopefully trap hydrocarbons. Porong was found to be breached, and it is believed that the hydrocarbons from Porong migrated into the Pucangan and are today being produced from the Wunut and Tangglangin Fields. What is also fascinating about Porong is the 'circular collapse' feature that, in hindsight, now indicates that the Porong structure is a earlier version of Lusi and has undergone caldera collapse during it's mud eruption. |
In summary, BJP-1 was targeting what was hoped to be an
un-breached version of the Porong reefal build-up, just 7km away, and the BJP-1
well was essentially a sister well to the Porong-1 well. The Porong-1 well was
drilled by a different operator, but was in acreage taken over by Lapindo, and
all Porong-1 data was available, and used, in the planning of BJP-1. I want to stress this point – as many
definitive moments in the drilling of the BJP-1 well, and many of the (strange)
drilling actions taken, all seem to directly contravene the data, observations
and drilling experiences from the Porong-1 well. Furthermore, some actions
made by Lapindo are nonsensically explained away as the drillers assuming that
BJP-1 has the same conditions seen in offset wells >100km or more away, in geologically
different fields in the offshore East Java Basin, rather than using the data
and knowledge from the neighbouring sister well, Porong-1.
The BJP-1 well was spudded on the 8th of March 2006 by the
TMMJ-04 rig (as an aside, Lapindo Brantas and the company that owns the TMMJ
drilling rig were both ultimately owned by Abrurizal Bakrie, one of Indonesia’s
wealthiest people and the welfare minister at the time of the disaster). I have
provided a summary of the key drilling events of the BJP-1 well herein.
However, I will briefly summarize some key aspects herein as well.
26” hole section: The BJP-1 well began encountering
overpressures at very shallow depths of just ~350m below the surface, which led
to the 20” casing being set ~13m shallower than planned.
17.5” hole section: Increasing overpressure, combined
with high amounts of background gas and some wellbore instability (pack-offs)
in the 17.5” hole section resulted in the 16” liner being set at ~666m depth,
which was ~310m shallower than planned. Numerous indications that the 16” liner
was not well cemented, including gas bubbling up from annulus, top squeeze
cement job and need for bottom squeeze cement job after 16” liner shoe leak-off
test (LOT).
14.5” hole section: Drilled to 775m with few problems
(minor reaming). Pumps broke and took 16 days to repair. Drilled down to 1096m,
with numerous connection gases and flows in the Upper Kalibeng clays, as well
as wellbore instability issues. Further instability issues, flows and possible
ballooning when reaming hole after wireline logging. Final 13-3/8” casing set
at 1091m, 280m shallower than planned. Losses prior to cement job, and then
during cement job, with some ballooning back – 756bbls lost during cement job.
Indications of marginal cement job, including unusual LOT profile (discussed in
detail later).
12.25” hole section: numerous connection gases and
high overpressure indicators in the Kalibeng clays. Encountered unexpected
thick volcanics/volcaniclastics, which greatly reduced drilling speed to just a
few meters per hour. Drilled to 2667m, which is past the planned 11-3/4” liner
point (1992m) and 9-5/8” casing point (2650m). Ran logs and check-shot survey.
Decided to keep drilling until carbonates were reached. Experienced total
losses at final well total depth of 2833m. Pulled back while pumping LCM slug.
Losses temporarily stopped, but then continued during pull out of hole. Well
kicked during pull out of hole, BOP shut-in, pressure spiked, followed by 24
hours of well control before the Lusi mud volcano appeared. Two days of well
control occurred, including three attempts to stop Lusi by pumping high density
mud down hole. Lusi mud eruption noticeably decreased during all three kill
attempts, and then increased rate after kill attempts stopped (though Lapindo
claim there was no evidence of connection between BJP-1 and Lusi mud flow
during these tests). Decision made to cut string, place cement plugs and
abandon well.
Did BJP-1 have
Sufficient Casing?
Casing is a critical part of safe drilling practice. Casing
is steel pipe that is placed in the well and cemented into place in order to
seal off the upper sections of drilled rock formation, and to protect the rock
from damage, such as due to the invasion of drilling mud into aquifers and losses
due to fracturing. In particular, casing is used to maintain a safe ‘mud weight
window’ while drilling. In order to drill a well safely, the density (‘weight’)
of drilling mud must be kept within a range (‘window’) such that pressure in
all uncased sections of the wellbore is greater than the maximum pore pressure
and lower than the minimum formation fracture pressure (and also above the
collapse pressure’ required to keep the well stable, but let us leave this out
for sake of simplicity). The minimum safe ‘window’ for mud weight varies, but
in all wells it is required that there be some safety tolerance between the
maximum pore pressure gradient in an uncased section and the minimum fracture
pressure gradient (usually assumed to be the leak-off test value at the casing
shoe). Casing is set when the drilling mud weight window approaches a
critically narrow range. The drilled hole section can be strengthened and
protected by placing steel pipe along the open hole section, and then filling
the annular space between the outside of the pipe and the wellbore wall with
cement – preventing the cased section from being fractured (or overpressured
fluids from entering any cased-off formation) and thus widening the mud weight
window and allowing the well to be deepened safety.
Casing is so critical to well design that it is planned out
and carefully designed long in advance of drilling the well. A basic tenant of
drilling wells is to carefully plan the well and then stick to the plan.
Indeed, any significant changes need to be documented in great detail and fully
and carefully justified (usually via a detailed memo of the change with a full
risk assessment). Yet, this was not done in BJP-1. The well was planned to have
6 strings of casing or liner set along the well (Tingay et al., 2008), with a
casing shoe approximately every 610m (2000’). Yet, as described above,
unexpected high pore pressures, losses (fracturing) and wellbore instability
caused the 20”, 16” and 13-3/8” casing shoes to be set shallower, with the
final 13-3/8” casing shoe set 280m shallower than planned (Sawolo et al.,
2009). However, the real deviation from the drilling plan came with the
repeated decisions to delay, or abandon, setting the 11-3/4” and 9-5/8” casing
strings.
BJP-1 was planned to have an 11-3/4” liner set at ~2000m
depth, which was prognosed to be inside the Kalibeng clays. However, the well
instead encountered an unexpected layer of thick volcanics and volcaniclastics
from ~1850m depth. It is not certain why the 11-3/4” liner was abandoned. The
encountering of an entirely unprognosed geological unit would have been a valid
reason to set casing. However, it has also been stated by Lapindo that this liner
was just a contingency casing point. But, again, there is no reason given in
any of the drilling reports as to why the liner was skipped. It is likely that
it was not considered as needed, given that pore pressure gradients appeared to
have plateaued, and the deeper rock was exceptionally strong, and maybe
considered unlikely to fracture. Furthermore, it may have been a simple cost
saving measure, as the well was already well behind schedule at this stage,
following the earlier drilling issues and the 16 days downtime due to pumps, and
was getting even further behind schedule due to the extremely slow ROP in the
volcanics. Regardless of the reasons, which remain unknown and speculative only,
the 11-3/4” liner was never set in place, and the decision was made to continue
drilling to the 9-5/8” casing point.
BJP-1 was planned to have a 9-5/8” casing set at
approximately 2630m depth, which was the predicted depth of the carbonate
target reservoir. However, the depth to this reservoir was underestimated, due
to the presence of the unprognosed volcanic layer that, with its extremely high
velocity, resulted in erroneous depth conversion of seismic data. When the
planned casing depth was reached and slightly exceeded, the decision was
instead made to stop and run petrophysical logs and conduct a Vertical Seismic
Profile (VSP), to try and estimate the depth to the top of the carbonates. The
VSP results were inconclusive, and suggested carbonates could be as deep as
~2926m. The decision was made to not set casing yet, but to instead drill until
the carbonates were reached, or until a maximum depth of 2865m (though no
mention of this is made in the daily drilling reports – the only source is
Sawolo et al., 2009). Ultimately, the well would reach a final depth of ~2833m
and the 9-5/8” casing would never be set. This meant that the BJP-1 well had
1740m of open (uncased) wellbore when the well control issues commenced at the
end of May 2006.
Did BJP-1 have
sufficient ‘Drilling Window’?: Pore pressure and fracture gradient in BJP-1
Pore Pressures in
BJP-1
Last year, I published a detailed compilation of pore
pressure, fracture gradient and petrophysical data for Banjar Panji-1 and some
surrounding wells (Tingay, 2015). Note that I am happy to provide this data to
others that wish to examine the disaster. High pore pressures (overpressures)
are commonly observed throughout the region, and commenced at shallow depths of
only 350m (1150’) in BJP-1. Pore pressure gradients reached very high
magnitudes of 17.2 MPa/km (14.6 ppg; 0.76 psi/ft) at just ~1200m (3900’) depth.
Comparison of pressure data with newly updated geological information shows
that overpressures under the Lusi mud volcano occur in three distinct
formations: (1) shales of the Pleistocene Pucangan and Upper Kalibeng
formations (~350-1870m depth); (2) Pliocene to early Pleistocene volcanic and
volcaniclastic sequences (from 1870- ~2830m depth), and; (3) Middle Miocene
reefal carbonates (>~2830m depth).
|
Pore pressure, leak-off test and overburden gradient data for BJP-1 and nearby offset wells in the Porong and Wunut fields (from Tingay, 2015). |
The overpressures at BJP-1 start remarkably shallow, and
reach very high levels at shallow depths, which further highlights the
difficulties faced in drilling the BJP-1 well. Furthermore, whilst the
overpressures in the Kalibeng clays are ‘textbook’ examples of standard disequilibrium
compaction overpressure, it is highly unusual to see high magnitude
overpressures in volcanics and carbonate rocks. Indeed, there are not even any
well-established or reliable ways to predict pore pressure in these rocks,
which makes well planning and drilling even more difficult.
Fracture Gradient
in BJP-1
The fracture gradient in BJP-1 is reasonably well defined by
the leak-off tests in BJP-1 and surrounding wells, although it should be
stressed that, again, fracture gradients are notoriously difficult to predict
or estimate in volcanics and carbonate sequences. However, one of the most
controversial and debated aspects of the drilling of BJP-1 comes from the
interpreted leak-off pressure from the final 13-3/8” casing shoe leak-off test.
This is both because the interpretations of this test vary greatly in the
literature, but also because it is this value that is often seen as defining
the minimum pressure required during the kick to create the Lusi
drilling-trigger scenario.
Leak-off tests are a standard test done immediately prior to
drilling a new hole section, and are conducted for safety reasons to determine the
maximum pressure the well can tolerate in a kick (and thus the pressure the
borehole must be safely kept below). Leak-off tests are performed by first
drilling out the recently cemented casing shoe, drilling a few meters into new
formation, closing the annular BOP and then pumping drilling mud into the
sealed well until the pressure increases enough to fracture the rocks.
Pressures during a leak-off test should first increase in a linear fashion, as
more drilling mud is forced at a constant rate into the fixed volume of the
wellbore. The ‘leak-off pressure’ is generally interpreted as the point in
which there is a break in slope of the increasing wellbore pressure, which is
considered to represent the initiation of a small and growing fracture in the
rocks at the bottom of the well, and thus a slight increase hole volume.
The debate over the interpreted leak-off pressure in BJP-1
stems from two things: the pressure gauge used and the point at which leak-off
is interpreted. The most controversial part of Lapindo’s interpreted leak-off
pressure is that they chose to use the absolute
maximum possible value. First, Lapindo picked ‘leak-off’ as the maximum
pressure value reached during the LOT, and not the first break in slope of the
pressure increase. They cite an unavailable Unocal internal report as their
rationale for this, even though this interpretation is contrary to every single
interpretation methodology I can find in the literature (there are literally
dozens of textbooks and papers that agree with this). This deviation from all
known drilling practice, in my opinion, overestimates the leak-off pressure
gradient by 0.6 MPa/km (or 0.5 ppg).
|
Available data for the BJP-1 leak-off test conducted at the 13-3/8" casing shoe at 1091m depth. Note that three pressure data curves are available, from three different gauges. Lapindo drilling reports state that they used the data from the 'Drill Pipe - Rig Floor Gauge' to derive a 15.7 ppg leak-off test. However, this was later modified to a 16.4 ppg LOT using the 'Halliburton gauge' data. In both instance, the leak-off pressure is taken as the maximum pressure reached during the test at approximately 6 barrels pumped. This is significantly higher than using the standard 'first-break in slope' approach, where leak-off is considered to occur after approximately 2 barrels pumped (data curtesy Lapindo and Sawolo et al., 2009). |
The data from the leak-off test was collected on three
gauges – a Halliburton drill pipe pressure gauge, a second drill pipe pressure
gauge (unspecified) and a casing pressure gauge (also unspecified). It is not
known where each gauge was located, and it is possible that the ‘Halliburton
gauge’ is on the cementing unit, which may then be appropriate to use, but this
is uncertain (sadly, it is also unknown how the test was lined up and executed
on the day). Hence, it remains difficult to establish which gauge is most
likely the more reliable. What is known is that Lapindo originally used the
non-specific drill pipe pressure gauge data and then, several days later,
modified their interpretation by using the Halliburton gauge. The use of the
Halliburton gauge data (which showed higher pressures than all other gauges),
combined with picking leak-off as the maximum pressure reached (and not the
break in slope pressure), yielded Lapindo the maximum possible leak-off
pressure value out of a wide range of possible interpretations.
There are several reasons why this is a dubious practice.
First, I polled a number of geomechanics and drilling experts who argued that it
is often the casing pressure that is more reliable during leak-off tests,
rather than the drill pipe pressure. Second, and more importantly, the leak-off
pressure represents a critical safety threshold that should never be crossed in a well control situation – and as
such, it is common practice to ‘err on the side of caution’ and select the
lower range of possible interpretations (and then place an additional safety
margin on top), rather than to select the absolute extreme high case. Hence,
Lapindo argue for a leak-off pressure gradient at the 13-3/8” show of 19.2
MPa/km (16.4 ppg). However, is more correct to say that the leak-off pressure
gradient is within a range from 17.9-19.2 MPa/km (15.3-16.4 ppg), which is
equivalent to a 19.5-20.9 MPa (2829-3037 psi) pressure range at the casing shoe,
and that safe well control should utilise the lower end of this range.
The uncertainty of the 13-3/8” leak-off test represents a microcosm
of data unreliability in BJP-1. Normally, a leak-off test value is something
about which there is little debate – there are standard practices for the
performing and interpretation of these tests, and it is usually just a single
value that gets reported. Yet, even this fairly simple and standard test
results in a significant degree of uncertainty, and thus major scientific and
legal ramifications. The same can, unfortunately, be said for just about every
other facet of BJP-1. Almost every single aspect and event that occurred has a
degree of uncertainty and unreliability, or variation in possible
interpretation – all of which clouds the waters in trying to determine the
cause of the Lusi disaster.
Implications of Not
Setting Casing on the ‘Drilling Window’ in BJP-1
It remains uncertain as to why the 9-5/8” and 11-3/4” casing
strings were not set at their planned depth. However, what is perfectly clear
from all the drilling reports and Sawolo et al. (2009) is that the driller’s
intention was to drill into the target carbonates
prior to setting the 9-5/8” casing. This is stated clearly in the drilling
reports, and is Lapindo’s often stated reason for continuing to deepen the well
beyond the planned casing point (though no memorandum of change or risk
assessment for deepening the casing point has ever been provided). Yet, this
decision makes no sense at all – and if this was the plan, then it is this decision to drill into the
carbonates before setting casing that likely represents the primary root cause
of the disaster.
In Sawolo et al. (2009), it is stated that the drillers
believed that the carbonates would contain very low pore pressures, and thus
that it would be better to set casing just inside the carbonates (it is common
to drill a little way into lower pressure reservoirs before setting casing).
Sawolo et al. (2009) state that this idea of a low pressure carbonate reservoir
was based on observations of the Kujung formation being commonly normally
pressured in many offshore fields. Yet, as described in Part 1 of this blog,
the carbonates targeted in BJP-1 are not the Kujung carbonates, and are
actually a completely different age. Furthermore, these statements also indicate
that Lapindo were basing their
prognosis on wells located >100km away offshore, rather than on immediately
nearby wells in their own acreage, and which were used to design the well
(particularly Porong-1). This seems oddly contradictory, to say the least!
I have previously highlighted that the BJP-1 well was
extremely similar to the Porong-1 well. Indeed, the entire reason for drilling
the well was the hope that it would be an unbreached version of the Porong
structure. Yet, the decision to drill
into the carbonates before running casing is almost suicidal when one considers
the pore pressures and drilling history of Porong-1. In Porong-1, the
casing was set just a few meters before the top of the carbonate target
reservoir. This is the usual standard practice when drilling into a reservoir thought
to contain highly overpressured gas, as it generally gives you a wide drilling window,
and only a short section of uncased formation that may need well control. Yet,
even with this proper precaution, Porong-1 suffered significant problems
drilling into the carbonates. The Porong-1 well experienced several days of
continuous well control delays, with alternative large losses and kicks when
the highly overpressured reservoirs were penetrated (Tingay, 2015). Direct
wireline pressure tests confirmed the data from several kicks: namely that the
carbonates were water wet (minor residual hydrocarbons only) and highly
overpressured (~18.5 MPa/km; 15.7 ppg). The drilling of even this short, ~50m,
hole section in Porong-1 was further complicated by the narrow drilling window
(leak-off test of ~19.2 MPa/km; 16.3 ppg), and the presence of high permeability
thief zones in the carbonates (Tingay, 2015). But, eventually, the well was
brought under full control, and Porong-1 was drilled to its final depth ~40m
inside the carbonates.
The lessons from Porong-1, as the perfect offset well for
BJP-1, are quite clear – the carbonates in BJP-1 are expected to be highly
overpressured. As such, casing should be set prior to drilling into the
carbonates. Yet, this was clearly not the plan in BJP-1. Despite the pore pressure measured in Porong-1 carbonates, and the
thinking that the BJP-1 structure may be unbreached, and thus possibly even
more highly pressured (and containing gas), the drillers repeatedly state that
they intended to penetrate the carbonates before setting casing. Even
stranger, the drillers made no effort to raise the mud weight prior to drilling
into the carbonates – and thus apparently planned on using 14.7 ppg mud to
drill into a gas reservoir target that their best offset well indicated
contained pore pressures of 15.7ppg or more!
In short, the BJP-1
well, in its final days, was being deliberately drilled:
·
into a
likely highly overpressured (≥15.7 ppg) gas reservoir with 14.7 ppg drilling
mud (an ~1.0 ppg mud weight underbalance);
·
with an
open hole (uncased) section that was ~1740m long, despite the well being
planned to have no more than ~600m open hole section, and;
·
into
expected ~15.7 ppg pore fluid pressure gradients that were close to, and very
possibly exceeded, the leak-off pressure at the 13-3/8” casing shoe (15.3-16.4
ppg).
This course of action really makes no sense to me – it
absolutely boggles my mind. Such actions are either:
· based on horribly incorrect assumptions (e.g. using
offset wells >100km away, rather than the nearest and most relevant offsets,
which makes little sense given the evidence that Porong-1 was used in the well
planning, including fracture gradient prediction);
·
terribly negligent (not doing proper well
planning as is usually required by law);
·
Suicidal (knowing the likely conditions the well
would encounter, but doing it anyway), or;
·
all of the above!
No valid explanation seems to be given for why the 11-3/4” liner
was not set as planned, nor has any detailed documentation for defending this
decision ever been provided. The
failure to set casing, and the planned drilling into the carbonates before
setting casing, breaches numerous drilling safety standards. Furthermore, it is
quite possible that the Lusi disaster could have been avoided had either the
11-3/4” or 9-5/8” casing been set as planned.
Concluding Summary
We have now reached a point the BJP-1 drilling story that is
around the 26th of May 2006 – which is about 3 days before the Lusi
mud eruption first began, two days before BJP-1 experienced its major kick, and
the day before BJP-1 experienced total losses at TD, as well as the day before
the Yogyaykarta earthquake. I hope readers can recognize the hazardour
condition the BJP-1 well was in at this point. The lack of casing, and decision
to drill underbalanced into rocks that have pore pressures close to (and
possibly exceeding) the 13-3/8”leak-off pressure, is the drilling equivalent to
driving at high speed along a dark mountain road, at night and with no
headlights. At this point in the story, nothing bad has happened – for example,
driving unsafely does not mean that an accident will definitely occur, nor does
it automatically mean an accident is the dangerous driver’s fault. However, it
does highlight that the scene has been set for a disaster to occur – and that
normal safety conditions and procedures were not in place. In the next Lusi
blog post, I will go through, in detail, the drilling events that I believe
triggered the Lusi disaster, and the evidence from the well that supports (and
the evidence that argues against) the drilling-trigger theory.
References
Davies, R. J., R. E. Swarbrick, R. J. Evans, and M. Huuse, 2007, Birth of a mud volcano:
East Java, 29 May 2006: GSA Today, 17, 4–9
Davies, R., M. Brumm, M. Manga, R. Rubiandini, R. Swarbrick,
and M. Tingay, 2008, The east Java mud volcano (2006 to present): an earthquake
or drilling trigger?: Earth and Planetary Science Letters, 272, 627-638.
Davies, R. J., M. Manga, M. Tingay, S. Lusianga, and R.
Swarbrick, 2010, Discussion on: “The LUSI mud volcano controversy: Was it
caused by drilling?”, N. Sawolo, E. Sutriono, B. P. Istadi and A. B. Darmoyo,
authors: Marine and Petroleum Geology, 27, 1651-1657.
Istadi, B., G. Pramono, and P. Sumintadireja, 2009, Modeling
study of growth and potential geohazard for LUSI mud volcano: East Java,
Indonesia: Marine and Petroleum Geology, 26, 1724–1739.
Kusumastuti, A., P. van Rensbergen, and J. Warren, 2002,
Seismic sequence analysis and reservoir potential of drowned Miocene carbonate
platforms in the Madura Strait, East Java, Indonesia: AAPG Bulletin, 86,
213-232.
Mazzini, A., A. Nermoen, M. Krotkiewski, Y. Podladchikov, S.
Planke, and H. Svensen, 2009, Strike-slip faulting as a trigger mechanism for
overpressure release through piercement structures. Implications for the Lusi mud
volcano, Indonesia: Marine and Petroleum Geology, 26, 1751-1765.
Sawolo, N., E. Sutriono, B. P. Istadi, and A. B. Darmoyo,
2009, The LUSI mud volcano triggering controversy: Was it caused by drilling?:
Marine and Petroleum Geology, 26, 1766-1784.
Sawolo, N., E. Sutriono, B. P. Istadi, and A. B. Darmoyo,
2010, Reply on “Discussion of: The LUSI mud volcano controversy: Was it caused
by drilling?”, Davies, R. J., M. Manga, M. Tingay, S. Lusianga, and R.
Swarbrick, authors: Marine and Petroleum Geology, 27.
Tingay, M., Hillis, R., Morley, C., Swarbrick, R. &
Okpere, E., 2003. Pore pressure/stress coupling in Brunei Darussalam –
implications for shale injection. In: Van Rensbergen, P., Hillis, R.R.,
Maltman, A.J. & Morley, C.K. (eds.) Subsurface Sediment Mobilization.
Geological Society of London Special Publication, London, 216, 369-379.
Tingay, M., O. Heidbach, R. Davies, and R. E. Swarbrick,
2008, Triggering of the Lusi mud eruption: earthquake versus drilling
initiation: Geology, 36, 639-642.
Tingay, M., 2010, Anatomy of the ‘Lusi’ mud eruption, East
Java: Australian Society of Exploration Geophysicists 21st International
Conference and Exhibition, 1-6, doi:10.1071/ASEG2010ab241.
Tingay, M., 2015, Initial pore pressures under the Lusi mud
volcano, Indonesia: Interpretation, 3(1), SE33–SE49,
doi:10.1190/INT-2014-0092.1.