2009 - The LUSI Mud Volcano Triggering Controversy Was It Caused by Drilling

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Marine and Petroleum Geology 26 (2009) 1766–1784

Contents lists available at ScienceDirect

Marine and Petroleum Geology


journal homepage: www.elsevier.com/locate/marpetgeo

The LUSI mud volcano triggering controversy: Was it caused by drilling?


Nurrochmat Sawolo, Edi Sutriono, Bambang P. Istadi*, Agung B. Darmoyo
Energi Mega Persada, Wisma Mulia 22nd Floor, JI. Jend. Gatot Subroto 42, 12710, Jakarta Indonesia

a r t i c l e i n f o a b s t r a c t

Article history: Following the Yogyakarta earthquake on May 27th, 2006, the subsequent eruption of a mud volcano has
Received 25 April 2008 been closely observed and analyzed by the geological community. The mud volcano, known as LUSI,
Received in revised form began erupting near the Banjarpanji-1 exploration well in Sidoarjo, East Java, Indonesia. LUSI offers
6 April 2009
a unique opportunity to study the genesis and development of a mud volcano.
Accepted 7 April 2009
For the first time, this paper presents all raw and interpreted drilling data, so any interested party can
Available online 18 April 2009
perform their own assessment. Our study suggests that LUSI mud volcano was a naturally occurring mud
volcano in an area prone for its mud volcanism. Pressure analysis done on the Banjarpanji well shows
Keywords:
Mud volcano that the pressure exerted at the well is lower than the fracture pressure at the last casing shoe, and
LUSI suggests that the well was intact and did not suffer an underground blowout. This is further supported by
Underground blowout evidence and observation made during drilling (such as circulation was done on an open BOP) and
Earthquake subsequent relief wells (Sonan and temperature log runs).
East Java This study offers a different alternative to earlier hypothesis that events at the Banjarpanji well were the
Watukosek trigger for the LUSI mud volcano. More work is needed by the scientific community to study the
sequence of events in order to explain and clarify the real trigger of LUSI.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction engineering data and analysis on the Banjarpanji well. It analyzes


the bottom hole pressure and rock strength data to test if the
On May 29th, 2006 at around 05:00 h an intermittent eruption pressure induced by the kick was sufficient to fracture the shoe and
of hot water and steam was observed some 200 m from the Ban- caused an underground blowout. It also examines the evidence
jarpanji-1 well. The location map of Banjarpanji is shown in Fig. 1. surrounding the well and during the re-entry program after the
The eruptive bursts of hot water and steam were dramatic with mud eruption. The analysis results suggest that the well remained
a distinct geyser-like cycle of active and passive periods. This intact and did not suffer any underground blowout.
marked the birth of a new mud volcano known as LUSI in East Java, This paper also presents relevant data from 8750 ft to 9297 ft
Indonesia, shown in Fig. 2 (Cyranoski, 2007; Satyana, 2007). (2667 m–2834 m) while drilling to look for the top of the Kujung
The cause of LUSI is controversial. The essence of the contro- formation. This includes the daily drilling report, daily geological
versy is whether the mudflow i) Originates from the wellbore, thus report and daily mud loggers report. Real time data plot for the
an underground blowout (Davies et al., 2007, 2008; Tingay et al., critical period between the times when the BOP was closed to the
2008) or ii) Originates from an eruption of overpressured shale time that the mud eruption was reported is also attached for critical
through reactivated faults as conduits (Mazzini et al., 2007a,b), or analysis by interested parties.
iii) Originates from geothermal activity, where superheated
hydrothermal fluids at high temperature and pressure were
released through a fault zone or fracture network as the conduit 2. Regional geology of East Java
(Sudarman and Hendrasto, 2007).
Only the first hypothesis, the underground blowout hypothesis, The Eocene and Early Oligocene Sequence of East Java back arc
will be discussed in this paper. Hypothesis based on fault reac- basin is associated with rifting where clastic deposition and
tivation and geothermal activities is beyond the scope of this paper carbonate buildup of the Ngimbang Formation took place. The Late
and will not be discussed. This paper lays out the drilling Oligocene and Miocene sequence is separated from the underlying
sequence by an unconformity which served as the foundation of
ENE-WSW oriented carbonate trends. This platform development,
* Corresponding author. Tel.: þ62 8 1188 3266. which is known as the Kujung limestone, occurred in the late
E-mail address: bambang.istadi@energi-mp.com (B.P. Istadi). Oligocene while the Prupuh and Tuban reefal development took

0264-8172/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marpetgeo.2009.04.002
Fig. 1. Banjarpanji-1 location and its offset wells. Banjarpanji-1 is located 30 km south of Surabaya, in the island of Java, Indonesia. Banjarpanji offset wells, both onshore and
offshore, provided good drilling information and lessons learned to help design the well. These include the setting of casing shoe in the carbonate formation to anticipate a pressure
regression in Kujung carbonate and the use of oil base mud to mitigate the overlying highly reactive shale.

Fig. 2. Initial LUSI eruptions at five different locations aligned along Watukosek fault lines. River bends and escarpment aligned with the Watukosek Fault System. (photo inset)
taken on 29th May 2006, 9 am four hours after mud eruption was reported showing the geyser-like mud eruption approximately 200 m away from the Banjarpanji 1 well location.
1768 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

Fig. 3. Mud volcanoes in Eastern Java. Map of Eastern half of Java showing locations of mud volcanoes. In particular, there are at least five mud volcanoes near the Watukosek fault
on the north eastern part of the island.

place in the Early to Middle Miocene. Such reefs have been the of the Kalibeng Formation is overpressured in most parts of the
target for several exploratory wells including the offset Porong-1 basin where rapid pressure transition occurs.
well. The Plio-Pleistocene sequence overlies an unconformity. In Several tectonic models have been proposed to explain the
some places this unconformity removed the entire Middle and Late complexity of East Java Basin, and the understanding of the tectonic
Miocene section such as in Porong and the KE-11-C area. Subse- and basin development of the area is still subject to ongoing debate.
quent Pliocene and Pleistocene sedimentation consisted of an A continental fragment model is favored, a continental fragment
eastward-prograding mudstone-dominated volcaniclastic wedge where possibly detached from the Gondwana super-continent
of Kalibeng and Pucangan Formation, with thickness of 8000 to which collided with the eastern margin of the Sunda Microplate
10,000 feet (2438–3048 m). The volcaniclastic materials derived and uplifted a mélange complex in Central Java. This tectonic model
from the Java volcanic arc south of the Sidoarjo area. The mudstone was proposed by Hamilton, 1979, and has been followed by every

Fig. 4. Mud Volcanoes along Watukosek faults. LUSI and five other known mud volcanoes are located along the Watukosek fault zone. The concentration of mud volcanoes near
Watukosek fault confirms that weak zones adjacent to the fault are conducive and prone to mud volcanism.
Fig. 5. Seismic section (Top) and Stratigraphy of Banjarpanji-1 (bottom). Banjarpanji-1 logged section shows presence of low density intervals within the shale unit which
correspond with the low velocity interval between 40000 and 60000 , correlatable with ‘‘bluish gray clay’’ of Upper Kalibeng Formation Pleistocene in age. Correlation based on:
i) foram fossil index plankton Globoratlia truncatulinoides and nanno fossil index Gephyrocapsa spp. assemblages similarities. The sediments were deposited in Middle to outer
shelf environment.
1770 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

worker since e.g. Daly et al. (1987), Hall (2001), Longley (1997), parallel to the direction of the collisional suture. Prasetyadi et al.,
Sribudiyani et al. (2003). The model was finally proven by Smyth 2006, continue this line of thought and propose a micro-continent
et al., 2005, 2007. The grain of the continental basement influenced as the core of southeast Java (‘‘East Java Micro Plate’’).
basin trends. The dominant basement grain is E–W, in eastern part The complex geology and presence of overpressure sediments
of East Java Basin, and at the collision zone NE–SW direction, results in the many mud volcanoes in the area (Fig. 3).

Table 1
Drilling Operations.

Key event Date (2006) and Time


Performed Leak Off Test at 13-3/800 casing shoe. LOT at 3580 ft (1091 m) was 16.4 ppg May 6th
(19.29 MPa/km)
Drill 12-1/400 (31.1 cm) hole section to 8750 ft (2667 m) May 6th–May 23rd
Prognosis of the top of carbonate was at 8500 ft (2591 m) Drill to 8750 ft (2667 m) without
encountering any carbonate section. Log the well to better determine the carbonate top.
Log well. May 23rd–May 25th
Result of VSP log suggested that the expected depth of carbonate section can be as deep as 9600 ft (2926 m).
At 10 bbls and 0.5 ppg (1590 l and 0.59 MPa/km) pre-determined kick tolerance, the deepest
that it was safe to drill was 9400 ft (2865 m). Decided to continue drilling to top of
carbonate or 9400 ft (2865 m) maximum depth.
Continue to drill 12-1/400 (31.1 cm) hole to 9297 ft (2834 m) May 25th–May 27th
Yogyakarta earthquake May 27th 05:55
Well recorded a 20 bbls (3180 l) mud loss 7 minutes after main earthquake May 27th 06:02
Well recorded a total loss of circulation and 130 bbls (21670 l) mud loss less than 2 h after two major May 27th 12:50
aftershocks (Fig.12). The proximity of the times suggest that the earthquake had an impact down hole in the well.
Pumped 60 bbls (9540 l) of Loss Control Material to stop losses. Losses cured. Well static for 7 h May 27th 13:00–22:00
without any further loss or kick, and is deemed to be safe to start pulling out of hole.
Decided this will be the casing point.
Pulling out of hole May 27th 22:00
After sufficient new mud is made and well static, start pulling out of hole. Pulling rate
w5 min per stand, pumping mud every 4–7 stands, no apparent drag. Unlikely to swab.
Well kicked, shut in and kill well May 28th 07:30
Well kicked, H2S content 500 ppm. Shut BOP (Blow out preventer) to stop further influx. ISIDP 450 psi May 28th 07:50
(3.10 MPa) and ISICP 350 psi (2.41 MPa) volume of influx w360 bbls. (57240 l). ISIDP, ISICP and influx
are reading from Real Time Chart.
Kill well by using volumetric method, applied twice and well dead. Maximum SICP 1054 psi (7.27 MPa).
Attempted but not able to burn influx. Circulated out the kick and found influx to be saline water of
8.9 ppg. (10.47 MPa/km) density
Open BOP but found drill string stuck May 28th 11:00
Drill string stuck, but still able to circulate, BHA appeared to be differentially stuck. Stretch calculation May 28th 11:00–14:20
suggests it was stuck at 4182 ft (1275 m)
Fish stuck drill string while circulating at high rate, unsuccessful. Ability to circulate stopped at around May 28th 14:20–21:30
14:30; the well appeared to have caved in. Jar stop functioning. Wait for fishing tool while circulating
through trip tank.
Released trapped pressure. Pumped 40 bbls (6380 l) soaking fluid to try to dehydrate mud cake, pumping May 28th 21:30–23:00
with no return. Shut in to let soaking fluid to work.
Rig up free point indicator (FPI) to cut the drill string. Bleed off trapped pressure in the drill pipe. May 29th 02:00–04:00
While rigging up tool, 35 ppm H2S gas detected at the rig floor. Evacuate personnel and abandon FPI.
Mud eruption started May 29th 05:00
Villagers reported a hot water flow in their field w200 m from the well.
Hot water and steam blew intermittently to around 25 ft (8 m) height geyser like with 5 min intervals
between bursts.
Checking for any connection or channel between the well and the mud eruption May 29th 10:00–23:30
Pumped first batch of mud 185 bbls of 14.7 ppg (29415 l of 17.29 MPa/km) down hole with a pressure of
700 psi (4.83 MPa)
Second batch of mud 200 bbls of 16 ppg (31800 l of 18.82 MPa/km) with Loss Circulating Material with a
pressure of 900 psi (6.21 MPa)
Initially suspected that injection reduced the eruption intensity. But high pressure indicated the absence of
any channel between the well and the eruption. This absence of channel and connection to the eruption
was interpreted as safe to continue fishing job.
Continue to fish stuck drill string May 29th–June 2nd
As an added safety measure while fishing, a barrier of cement will be laid below the fish to isolate it from
the open hole below.
Perform injection test, with 2.5 bpm rate at 370 psi (398 l/m at 2.55 MPa) This third injection test again May 30th 04:00–05:00
showed a high pressure injection above the LOT pressure meaning the absence of any channels.
Pump two batches of cement to isolate the fish from the open hole below. Injection test after laying May 30th 05:00–June 1st 04:30
cement plugs showed 1000 psi suggesting that the cement plugs were in place.
Ran free point indicator and found that the stuck point has moved upward from 4182 ft (1275 m), the original
stretch measurement, to 3200 ft or 2600 ft (between 980 m and 790 m)
Ran string shot and finally the drill string was backed off at 2989 ft (911 m). Pull out of hole with the drill string.
Temporarily abandon well and move out rig June 2nd–4th
Condition at the rig was critical with mud starting to enter location, more eruptions and developing new cracks
at the rig site.
To safe lives and properties, decided to Temporary Abandon the well and move the rig out. This will give time to
assess situation and plan intervention program
Rig released June 4th 00:00
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1771

The overpressure is due to high sedimentation rates in a rapid 4. Operational summary of Banjarpanji #1 well
subsidence and burial basin (Willumsen and Schiller, 1994; Schiller
et al., 1994) and maturation of the organic materials. The East Java The drilling operation of Banjarpanji was done by PT Tiga Musim
geosyncline has thick Tertiary sediments of more than 6000 m Mas Jaya, a reputable drilling company in Indonesia. The key
(Koesoemadinata, 1980) with an estimated sedimentation rate of services (the mud chemical services, the electric logging and the
2480 m/ma in the vicinity of LUSI (Kadar, 1986). Mud volcanism is cementing services) were provided by M-I, Baker Atlas and Halli-
known to be associated with highly under compacted over- burton, all in the top five of their respective fields. Other ancillary
pressured shales and there is a clear relationship between mud services are provided by PT Elnusa (mud logging), Weatherford
volcanoes breaching the surface and tectonic movement along (equipment rentals) and Sperry-Sun (for any directional drilling
faults as evidenced in the mud volcanoes aligned along the and correction purposes). Apart from mud pumps equipment
Watukosek fault zone (Fig. 4). problems, there were no major operational issues and the team
performed their specific duties as good as, or better than, expected.
In the upper hole sections, the well lithology is not significantly
3. Geology of LUSI mud volcano different from that prognosed. The only difference is in the well
pressure where the depth of the transition zone to the over pres-
LUSI mud volcano adds to the many mud volcanoes existing in sure was found to be shallower than planned requiring the upper
the area, such as the Porong collapse structure (NE of LUSI), Kalang casing shoes to be set accordingly. The end result is such that
Anyar & Pulungan (Sedati, Sidoarjo), Gunung Anyar (UPN campus, subsurface drilling problems in the upper hole section of the
Surabaya), Bleduk Kuwu & Keradenan (Purwodadi), Wringin Anom/ Banjarpanji well were almost non existent (Sutriono, 2007).
Pengangson (Gresik), Semolowaru (Unitomo campus, Surabaya), Once the 13-3/800 (34.0 cm) casing was set, the operation in the
Dawar Blandong (Mojokerto), Sangiran (Central Java), Socah 12-1/400 (31.1 cm) hole section did not go according to plan and
(Bangkalan, Madura) and others. drilling parameters were adjusted accordingly. Lithology and dril-
All of these mud volcanoes occurred naturally; some of which ling parameters are different from the deep offset Porong well. The
are still active and still erupting mud. The existence of mud bottom most w3000 ft (910 m) of the well was a solid deposit of
volcanoes in East Java have been mapped as early as 1936. laharic sandstone section instead of shale. The pressure and the
(Duyfjes, 1936, 1938). In the eastern part of Java, in the East mud weight used to drill this section is lower, 14.7 ppg (17.29 MPa/
Kendeng zone, a number of these mud volcanoes exhibited km) compared to 15.5 ppg (18.23 MPa/km), at the same depth in
a pattern that follows the Watukosek fault. This major fault offset Porong well. The Leak Off Test, on the other hand was found
started from the Arjuno – Welirang volcano complex along the to be higher than prognosed (16.4 ppg vs. 16.0 ppg). The stronger
SSW/NNE direction. LUSI and other mud volcanoes that lie along shoe and lower pore pressure are the key reasons why the 9-5/800
this trend are shown on Fig. 4. (24.4 cm) casing shoe was able be set deeper, thus a longer open
LUSI mud volcano is unique, as scientists can observe the hole section was able to be drilled with the same safety factor.
evolving geological processes from its birth. The solid matter in Drilling operations of the 12-1/400 (31.1 cm) hole section until the
LUSI’s mud is ‘‘bluish gray clay’’ from the Upper Kalibeng Forma- mud eruption is summarized in Table 1.
tion, which is Pleistocene in age, as confirmed from mud samples,
cuttings and side-wall cores taken from the Banjarpanji-1 well
5. Was LUSI caused by drilling?
from a depth section of 4000–6100 feet (1219–1859 m). Specifically,
i) Foraminifera index plankton Globorotalia truncatulinoides and
LUSI gives the unique opportunity for the geoscientists to study
nannofossil index Geohyrocapsa spp. assemblages similarity. The
the birth of a mud volcano. At the same time the drilling commu-
sediments were deposited in a Middle to outer shelf environment
nity was also interested in the birth of LUSI, but from a different
(Kadar et al., 2007) as shown in Fig. 5; ii) Kerogen composition is
perspective, namely what triggered LUSI. Was it caused by drilling
correlatable with side-wall cores taken at a depth of 5600 feet
or simply another mud volcano that occurred naturally?
(1707 m); iii) Thermal maturity profile shows positive vitrinite
reflectance correlation with cuttings & side-wall cores taken at
depths from 5100 to 6300 feet (1554–1920 m). The presence of 5.1. Underground blowout hypothesis
overpressured shale is evident from the logs and was identified as
drilling hazard in the well prognosis. Initially, the complete Banjarpanji-1 drilling data set was not
The source of fluid, however, is still debated. Mazzini et al. publicly available. Early technical papers (Davies et al., 2007, 2008;
(2007a,b) used geochemical data to suggest that the overpressured Rubiandini et al., 2008; Tingay et al., 2008) were published based
fluids are primarily sourced from clay diagenetic dehydration on limited data and concluded that an underground blowout was
within the Upper Kalibeng Formation. Davies et al. (2007) suggest the cause of LUSI.
that the fluids are primarily sourced from the Kujung carbonate These papers proposed a classic underground blowout resulting
formation. Whereas Sudarman and Hendrasto (2007) suggest from an unsafe act (swabbing while pulling out of hole) combined
a deeper fluid source, where geothermal activity induced the mud with an insufficient safety factor (kick tolerance) during the drilling
eruption by the release of superheated hydrothermal fluids at high of the well. Their conclusion was that the resulting shut in pressure
temperature and pressure through fault zone or fracture network after the kick fractured the deepest casing shoe and caused an
as the conduit. underground blowout, which eventually reached the surface and
In the early stages, the eruption consist of hot water, steam caused the LUSI mud volcano.
followed by erratic and intermittent explosions of mud. The mud At the time, this underground blowout hypothesis was the only
consists of clay and salt water in slurry state at temperatures explanation for the LUSI mud volcano. As a result the public under-
between 80 and 100  C. The gas bubbles consist primarily of stood that an underground blowout in the Banjarpanji well was the
methane, N2, CO2 and small percentage of higher hydrocarbons root cause of the mud eruption. However, as more data from the field
indicating the presence of thermogenic oil. The gas composition became available and analyzed, it became clear that the analysis and
indicates similarities with gas from the adjacent Wunut Field deep field observations do not support the underground blowout
horizons. hypothesis (Table 2 and Fig. 8A). The following chapters explain why
1772 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

Fig. 6. Four key events that must take place if the mud eruption was triggered by the well. These four key events are i) Occurrence of an uncontrolled kick, ii) Pressure sufficiently
high to fracture the weakest formation, typically the casing shoe, iii) An underground blowout occurred, iv) Sustained fracture propagation pressure to extend the fracture to the
surface.

an underground blowout did not occur and suggest an alternative on the pressure analysis that was done on the ‘drill pipe side’.
explanation whereby LUSI was triggered by natural causes. Pressure analysis on the drill pipe side is commonly done
and perfectly legitimate provided that the well behaves as a perfect
5.2. Casing shoe was intact and not breached U-tube (Section 5.4.1, Field Evidence, Real Time Data). One must
ensure that the well bore pressure can be read without any
Pressure analysis shows the level of pressure that the well is restriction on the drill pipe side. In practice this is done by slowly
subjected to, and compares it to the formation strength to indicate pumping through the drill pipe, thus ensuring that the float valve
if the formation is fractured and the well-bore integrity compro- (Fig. 7) is kept in the open position. Unfortunately, this slow
mised. Mostly, the analysis is done at the deepest casing shoe since pumping was never performed in Banjarpanji, therefore the float
it is typically, but not always, the weakest portion of the well and valve is likely to be in a closed position and correct well pressure
the point where the fracture pressure is known. cannot be accessed from the drill pipe side.
Early papers claimed that the casing shoe was subjected to Performing pressure analysis in the annulus side is more direct
a pressure higher than its Leak Off Test and failed. This was based but it is more complex. In this particular case, the annulus is the

Fig. 7. Mud circulation system showing float valve at bottom of drill string. The use of non-ported float valve in the drill string is standard in Lapindo-Brantas, Inc. This float valve
restricts pressure communication between the annulus and drill pipe unless it is kept open by slowly circulating through.
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1773

Fig. 8. Casing setting depth check (top). Actual casing setting depth is checked using a leading commercially available casing design software with a kick tolerance of 10 bbls and
0.5 ppg gas influx and found to be safe including a planned 9-5/800 casing at 9300 ft TD. (Bottom) Pressure profile in wellbore and sensitivity analysis. The pressure data are plotted
that shows that the well is safe. Sensitivity analysis is done with a bottom hole pressure of 14.7 ppg, and a LOT of 15.8 ppg. Even at such extreme, the wellbore pressure at any depth
is always below the minimum formation strength meaning that the wellbore is always intact.

preferred leg to perform any pressure analysis as the well was not ii. Fluid in the upper part of the hole (influx fluid) ¼ 8.9 ppg
completely packed off and the fluid density and Bottom hole (10.47 MPa/km)
pressure of the well were measured. The annulus pressure is the iii. Fluid in the bottom part of the hole (mud) ¼ 14.7 ppg
better representation of the well-bore pressure. (17.29 MPa/km)
The data to perform this analysis is: iv. Bottom hole pressure (BHP) ¼ 12.8 ppg (15.06 MPa/km)
v. Leak off test (LOT) at the casing shoe (3580 ft or
i. Maximum casing pressure ¼ 1054 psi (7.27 MPa) 1091m) ¼ 16.4 ppg (19.29 MPa/km)
1774 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

Appendices C and D show how these data were obtained. 3053 psi (21.05 MPa). It shows that the formation strength is higher
The graphical pressure analysis at the casing shoe is described in than the fluid pressure in the well that it was subjected to; there-
Appendix E, and the graph is shown in Fig. 8B. fore, the casing shoe is likely to remain intact.
The resulting pressure that is acting at the shoe is 2710 psi Sensitivity runs were performed around the softer data, using
(18.68 MPa) whereas the formation strength at the casing shoe is a LOT of 15.8 ppg (18.58 MPa/km) and a Bottom hole pressure of

1200

Max casing press


Drill Pipe and Casing Pressures
= 1054 psi
1000

Bleed off Casing press

800
Drill Pipe pressure
Casing press
Casing Pressure
Pressures (psi)

600
Pumping
Pumping

400
Circulate out influx
Drill Pipe press
OPEN BOP
200
Bleed off
Shut BOP trapd press

Well died
0
-20 0 20 40 60 80 100 120 140 160 180 200

-200 Time after BOP shut in (minutes)

REAL TIME DATA PLOT


DP AND CSG PRESSURE FLOW IN AND OUT ACTIVE PIT
DP Pressure, psi CSG Press, psi FLOW IN, GPM FLOW OUT, GPM ACTIVE PIT

PRESSURE, psi FLOW, gpm VOLUME, bbls


-200 200 600 1000 1400 1800 0 400 800 1200 1600 2000 0 400 800 1200 1600 2000
0 0 0
KILL WELL

120 120 120


Kill well kick using Volumetric Method
Pressure anomaly
240 240 240
Well dead, Annulus and DP
press = 0 psi
360 360 360
Attempt to fish stuck drill string

480 480 480


Ability to circulate through bit ceased
FISH STUCK DRILL STRING

600 600 600


Pressure anomaly?
TIME, MINS

TIME, MINS
TIME, MIN

720 720 720

840 840 840


Bleed off drill pipe prior
Pump soaking fluid
to pumping soaking fluid
960 960 960

1080 1080 1080

1200 1200 1200


H2S gas at surface

1320 1320 1320

1440 1440 1440

1560 1560 1560

Fig. 9. Banjarpanji-1 Real time data dated May 28th to 29th, 2006. The top graph shows the pressure in both Drill Pipe and Casing during the shut in period. It shows that the kick
was successfully killed in less than three hours. The bottom graph shows the RTD data after the well was killed until the time that the mud eruption was reported. Apart of a brief
pressure increase, the fluid volume stays constant suggesting that there is no hydraulic connection with the eruption.
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1775

Table 2
Trip sheet data. Pulling out of hole trip sheet from the Real Time Data, showing the time, number of pipe stands pulled, its displacement and the amount of mud pumped.
Pulling out was done slow, around five minutes per stand, and excessive fluid was pumped in the hole exceeding drill pipe steel displacement that negates the possibility of
swabbing.

Time Activity PVT Change Disp. Pumped Cum PVT


23:15 Start POH 551
23:17 POH 1 Stands 551 0
23:58 POH 4 Stands 556 5 2.8 7.8
0:53 Pump out 4 stands 483 73 2.8 107.2 102.2
1:12 Pull 2 stands 486 3 1.4 97.8
1:17 Pump 470 16 15 113.8
1:42 Pull 3 stands 473 3 2.1 108.7
1:56 Pump 439 34 28.6 142.7
2:47 Pull 7 stands 444 6 4.9 131.8
3:00 Pump 418 26 32.9 157.8
3:40 Pull 5 stands 447 29 3.5 148.3
3:58 Pump 408 39 33.8 187.3
4:32 Pull 5 stands 411 3 3.5 180.8
4:44 Pump 372 39 33.8 219.8
5:23 Pull 6 stands 378 6 4.2 209.6
5:40 Pump 339 39 33.9 248.6
6:17 Pull 7 stands 344 5 4.9 238.7
6:25 Pull 2 stands, well flowing 377 34 1.4 203.3
6:56 Pump, Pull 1 stand 469 92 0.7 42.3 110.6
7:03 Pump, Pull 1 stand 450 19 0.7 25 128.9
7:19 Call Co Man 453 3 34.2 125.9
7:53 Shut In Well 819 366 124.5 240.1
8:14 Final PVT reading 1074 255 495.1
Bleed of Gas
8:17 Pump Mud 1045 29 25.4
8:24 Bleed off Gas, Transfer mud 851
8:59 Pump Mud 817 33 52.5
Bleed off Press, Well Dead 929
9:13 Circulate 800 93

14.7 ppg (17.29 MPa/km) and found to be safe. These showed that 2710 psi (18.68 MPa), which is much less than the formation
the casing shoe remained intact and not compromised even if strength of 3053 psi (21.05 MPa) (Appendix E).
‘worst case’ inputs were used. - After the mud erupted, the high pressure obtained during
injection tests (Section 5.4.2, Field Evidence, High Injection
5.3. Four key events Tests) showed the wellbore was totally isolated from the mud
eruption, meaning that the casing shoe was not fractured and
For an underground blowout to occur and breach to the surface, was still intact.
four key events must take place (Fig. 6). If any of this sequence of iii. Sufficient pressure and drive to cause an underground
events did not occur, an underground blowout that breach to the blowout. Observations in the field did not support this
surface is very unlikely to occur. These are: underground blowout claims.
- Circulation in the well was recorded in the RTD until May 28th,
i. Occurrence of an uncontrolled kick. In Banjarpanji, there was
no uncontrolled kick and the well was dead a day before the 2006 at 13:00 h (Fig. 9). In an underground blowout situation, no
mud started to erupt. circulation is possible since it will be sucked by the cross-flow
- With the BOP opened starting at 11:00 h May 28th, 2006, there
- The kick was controlled three hours after shut in at 11:00 h
(Fig. 9). Both the casing and drill pipe pressures were bled off was no pressure in the well bore. With the well dead one
to zero pressure. cannot have an underground blowout situation.
- With the BOP opened, the path of least resistance is up through
- The BOP was opened at 11:00 h (Drilling morning report of
May 29th, 2006), the well was dead and operations were the well-head not through the formation. Nothing flowed from
underway to fish the stuck drill string by circulating and the well-head while a major eruption was blowing nearby.
jarring up. No such operation is possible during a kick. iv. Sustained Propagation pressure to extend any fracture to the
ii. Kick pressure that fractures the weakest part of the well, surface. No sustained pressure existed in the well bore, since:
- With the well dead and BOP opened starting at 11:00 h May
typically the casing shoe. This did not happen, as:
- The pressure exerted by the kick was too low to fracture the
28th, 2006, there was no pressure in the wellbore able to
casing shoe. The maximum pressure at the casing shoe is propagate any fractures to the surface.

Table 3
Laboratory analysis of mud properties taken randomly from different sites near the mud eruption. Data analyzed by on site MI Mud Engineer at BJP – 1, 1/06/2006 report.

Date 31 May 2006 31 May 2006 31 May 2006 31 May 2006 1 June 2006 1 June 2006
Time 23:45 23:45 23:50 24:00 05:00 05:30
Weight 10.5 ppg 8.7 ppg 10.6 ppg 10.0 ppg 10.7 ppg 11.0 ppg
pH 7 7 7 7 7 7
Cl 15.500 mg/l 14.500 mg/l 13.500 mg/l 14.700 mg/l 14.400 mg/l 14.600 mg/l
Water 100% 100% 100% 100% 100% 100%
Oil Nil Nil Nil Nil Nil Nil
1776 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

Fig. 10. Banjarpanji Bottom Hole Pressure (BHP) estimation. The bottom hole pressure (BHP) is estimated based on a number of methods with differing reliabilities. The most likely
BHP is around 12.8 ppg and the maximum BHP of 14.7 ppg (for sensitivity analysis purpose).

Fig. 11. Leak Off Test (LOT) Banjarpanji well at 3580 ft. (1091 m) depth. The LOT result was 16.4 ppg (19.29 MPa/km) (Left). The shape of the curve is typical of LOT done using oil
based mud due to a higher compressibility factor compared to a water based mud system. The resulting LOT is compared to other nearby wells (Right). The formation pressure and
the LOT in the shallow section of Banjarpanji resembles Wunut wells, since it is within the same closure.
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1777

Fig. 12. A 20 bbls loss of mud after the main earthquake (top).The left portion showed the seismograph reading of the Yogyakarta earthquakew06:00 WIB 27 May 06 at Tretes BMG
station about 15 km away. The top right picture showed the 20 bbls loss from the mud logger’s real time data that happenned seven minutes after the main earthquake. The bottom
right shows 130 bbls complete loss of circulation from the wellbore that happenned two hours after two aftershocks. These losses that happenned after the earthquake showed
a compelling argument that a temporal connection exists between the earthquake and Banjarpanji well.

- Samples of mud collected at the eruption site were analyzed explain the status of the well, whether it is breached or stays
without any traces of synthetic oil based mud. Results of the intact.
mud analysis are shown in Table 3. Had it been from the well,
the drilling mud used will distinctly show its oil based mud 5.4.1.1. During the well control incident (while the BOP was shut
signature. in). With the drill bit off bottom when the kick was taken, the
preferred well control method was the Volumetric Method
(Abel et al., 1994). This method involves lubricating a volume of
5.4. Field evidence drilling mud (not necessarily a heavy kill mud) and bleeding off
a certain amount of gas. This method is not intended to kill the
Facts and evidence were collected during final days of drilling kick, but simply to lower the shut-in pressure sufficiently to allow
and subsequent relief well project. This evidence do not support the snubbing the pipe back down to bottom where the well can then
underground blowout claims. These include: be killed conventionally. But instead of the expected gradual
lowering of the shut-in pressure, the well died after two cycles of
5.4.1. Real time data lubrication.
The most important piece of data to confirm the status of the After circulating the influx out and ensuring that the fluid in
well is the Mud logger’s real time data (RTD) that include the both drill pipe and annulus are full of 14.7 ppg. (17.29 MPa/km)
pressure data, mud volume data, pumping data, gas data and mud, the drill pipe pressure remained at around 75 psi (0.52 MPa)
the drilling parameter information. Analysis of this data helps while the annulus was 0 psi. The trapped pressure at the drill pipe
1778 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

Fig. 13. Sonan (left) and Temperature (right) logs taken during re-entry operations, 2 months after eruption did not show any anomaly. The absence of anomalies suggests that there
is no flow close to the casing. If the flow originates from the well, erratic noise or distinct temperature changes would be registered in the log. This suggests that flow may not
originate from the well.

was bled off and the BOP was opened. The well was confirmed dead done based on the drill pipe pressure, and must be performed on
at 11:00 h. This is shown in Fig. 9A that plots the pressure data, both the annulus side that is in direct contact to the open formation and
drill pipe and annulus, during shut in from the RTD. The most the shoe.
pertinent data for pressure analysis is the annulus shut in pressure.
The annulus pressure continued to rise after shut in and reached 5.4.1.2. After the BOP was opened until the mud erupted
a plateau at 1054 psi (7.26 MPa). This pressure is used as the input (Fig. 9B). The second piece of information from RTD is shown on
data in the Pressure Analysis section. Fig. 9B. It shows the operation on the well from the time that the
The second is the anomalies at the drill pipe leg of the well. well had died and the fishing of the drill string until the mud
These are: eruption. The critical information here are:

i. At the time of the shut in, drill pipe pressure was higher i. Pumping through the bit was still possible until w15:00 h on
than annulus pressure. The drill pipe pressure was around May 29th, 2006, despite the fact that the drill-string was
500 psi (3.45 MPa) and the annulus pressure was 280 psi stuck. The drill-string appeared to be differentially stuck but
(1.93 MPa). not completely packed off at the time.
ii. After the first bleed off period, the annulus pressure remained ii. The ‘pressure trap’ phenomenon in the drill pipe
at 0 psi whereas the drill pipe pressure stayed at a higher prevails, which showed that the Drill Pipe pressure
pressure. around 480–550 psi when there were no pumping (15:00
through to 21:00 h). This pressure trap was finally bled
The operation just before the BOP was shut in was pumping off just prior to spotting the soaking fluid to unseat the
mud down hole. Therefore, the drill string was full of 14.7 ppg fish at 21:30 h.
(17.29 MPa/km) mud whereas the annulus fluid was of unknown iii. After the pressure was bled off, there was no sustained
composition at the time of killing. However, this is in direct contrast pressure in the well.
to the pressure readings at shut in where the drill pipe pressure was
higher than the annulus. The plausible explanation on why the
pressure of the drill pipe can be higher than the annulus is the float 5.4.1.3. The information from the mud logger’s RTD can be summa-
valve that isolates the drill pipe pressure reading from the open rized as follows.
formation (Fig. 7). This suggests a ‘pressure trap’ phenomenon or, in
drilling engineering, ‘the well does not behave as a perfect U-tube’. - The maximum pressure in the annulus during the well
The consequence is that any accurate pressure analysis cannot be control was 1054 psi (7.27 MPa). This pressure reading is
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1779

valid since the well was still able to be circulated through These two tests were done with a high pumping rate as the
and not packed off. intent was to kill the eruption if possible. The initial report after the
injection tests suggested that the eruption intensity decreased but
- Float valve in the drill string appears to create a trapped further observation showed that this may not be the case; the
pressure in the drill pipe. This trapped pressure made readings continued erratic and intermittent nature of the eruption afterward
in the drill pipe invalid unless it is removed beforehand by suggests this was coincidental.
circulating slowly that kept the float valve open. The high injection pressure, higher than the Leak Off Test,
confirmed that there were no channel formed between the well
5.4.2. High injection tests and the mud eruption. It was then decided to continue to fish the
On the first day of the eruption, the first operational priority was stuck drill string, and as an added safety measure, cement plugs
to ascertain if there was any connection between the well and the would be set in the open hole below the fish.
mud eruption and attempt to kill it. Three injection (pumping)
connectivity tests were therefore carried out: iii. Third injection test before cementing; the injection rate was
2.5 bpm (397.5 l/m) at 370 psi (2.55 MPa). (Daily drilling
i. First injection test with 185 bbls (29,415 l) of 14.7 ppg report May 30th, 2006)
(17.29 MPa/km) mud. The injection pressure was at 700 psi
(4.83 MPa). (M-I Swaco report date May 29th, 2006). These high injection pressures confirmed that the shoe was
ii. Second injection test with 200 bbls (31,800 l) of 16.0 ppg not fractured and there was no channel formed between the
(18.82 MPa/km) mud loaded with LCM material. The initial well and the mud flow. If such a channel existed, the injection
injection pressure was 1200 psi (8.27 MPa) with a final pressure pressures would have been lower than the leak off test
of 900 psi (6.21 MPa)) (Drilling morning report May 30th, 2006). pressures.

Fig. 14. With BOP open, well cannot be in an underground blowout situation. If BOP is opened, the path of least resistance is through the wellbore instead of fracturing the
formation and breaching to the surface. Yet nothing came out of the well (A). In the Underground Blowout Hypothesis several scenario would happen: the hole would enlarged (B);
the fish Fall to the bottom of the well (C); the very high flow rate, the flow behind casing would be easily detected (D). None of the above happened. The fish was found in its original
abandoned depth.
1780 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

5.4.3. Other field evidence against underground blowout claims Analysis and its sensitivity test presented in this paper shows that
the weakest point in the well, the deepest casing shoe, remained
i. The drilling assembly stuck in the open-hole part of the well intact and was not breached. Evidence further suggests that:
did not fall deeper into the well. If an underground blowout
had occurred, the high mud flow rates would erode and - The kick was killed within three hours and the well was dead.
enlarge the well-bore, causing the drilling assembly to fall to The well no longer had any pressure to support an under-
the bottom of the well (Fig. 14). However, it was confirmed ground blowout process.
that the fish was still at its original position during the re- - The well was circulated on an opened BOP; this is not possible
entry project. in an underground blowout situation.
ii. LUSI has been spewing mud for more than 2 years at - With the BOP opened the path of least resistance is up through
impressive rates; initially around 50,000 m3/day of mud, the well head and not through the formation. Nothing flowed
increasing to around 156,000 m3/day, and currently around from the well head while a major eruption was blowing
80,000 m3/day. Speculations that the fluid is coming from the nearby.
Kujung Formation are inconsistent with the known reservoir - High injection test pressures on the well confirmed that the
properties and water chemistry of the Kujung Formation. The shoe was intact and there were no channels formed between
productivity of LUSI is in the order of at least 150 times that of the well and the eruption.
Kujung reservoir (Nawangsidi, 2007). - The well remained full and did not sustain any drilling mud
iii. Temperature and Sonan logging was carried out during the losses throughout the eruption. Chemical analysis from the
relief well campaign to look for evidence of an underground erupted mud did not contain any drilling mud particles. It
blowout. These logs were run on 20th of July 2006 (over 50 indicates that the mud eruption did not originate from the
days after the first mud flow) to the top of fish (2984 feet or well.
910 m) in the re-entered Banjarpanji-1 well. The Sonan log - The mud eruption rate was at least an order higher than the
was ‘very quiet’ which indicated the absence of fluid flow reservoir ability to flow into and up through the well bore. At
behind casing (Fig. 13A. The temperature logs showed 60  C, this flow rate, the flow is more likely to pass through a fault
and did not record any abnormal shift or anomaly within the plane instead of a well.
Banjarpanji-1 well (Fig. 13B). If LUSI originated and was - Temperature and Sonan logs result showed the absence of
flowing near the well, the temperature would show a marked a near casing fluid flow that is characteristic of an underground
increase given the high temperature of the erupting mud blowout.
(95  C). Both the Sonan and Temperature logs did not suggest
any near well bore fluid flow. Operational data are now opened to the public and scientific
community and presented for the first time in this paper. Our data
leads to the conclusion that LUSI was not triggered by an under-
5.5. Well reports ground blowout. The authors welcome future studies based on this
data that will help improve our understanding on the origin of mud
Key operational data and daily reports are attached in the volcanoes.
Appendix G to enable interested readers to perform their own
assessment on the events in the rig during the critical period. These Acknowledgements
are actual operational data from a real life complex drilling oper-
ation. Some of these operational data are incomplete and often The authors express appreciation to the management of BPMI-
conflicting with each other and can be interpreted differently GAS, Lapindo-Brantas Inc. and its partners for their support and
depending on how much effort and drilling experience the indi- permission to publish the paper. Numerous constructive discus-
vidual reader possesses. sions and inputs from drilling engineers and geoscientists are also
The critical data include: appreciated.

- Daily drilling reports Appendix A: Banjarpanji-1 well design


- Daily geological reports
- Daily mud loggers reports The Banjarpanji-1 well was drilled as an exploration well with
- Real time data plot (Fig. 9) the Miocene Kujung or Prupuh to Tuban carbonates as its primary
objective. These carbonate reservoirs are proven and prolific oil and
gas reservoirs and the target of most deep exploration wells
6. Conclusion throughout East Java. A seismic section through Banjarpanji-1 is
shown in Fig. 5.
LUSI is a new mud volcano in a region prone to mud volcanism. Banjarpanji-1 was recognized as a High Temperature and High
Along the vicinity of the Watukosek fault, where LUSI is situated, Pressure (HTHP) well early in the planning phase, and this neces-
there are at least five other known mud volcanoes. sitated an increased focus on offset geological and operational
The Banjarpanji-1 well was planned and drilled according to benchmark data to ensure a safe and efficient design and operation.
standard industry practice for high pressure exploration wells. Key For Banjarpanji-1 well planning, the key offset wells were:
learnings from offset wells were incorporated into the design of the
well. The safety factor applied was consistent with accepted explora-  Huffco Porong-1, 6 km away. Porong-1 provided the most
tion well standards, and in fact was higher than offset wells targeting relevant drilling information and geological data in the deep
the same Kujung carbonate formation, that have longer open hole portion of the well. The well experienced fluid kicks in the
sections. Drilling operations were performed by a qualified drilling overpressure zones and losses while penetrating the deeper
contractor and supported by quality service companies. The result was carbonate section.
minimal down-hole drilling problems right until the time of the  Huffco Wunut-2, 1.5 km away. Wunut 2 supplemented
Yogyakarta earthquake, where a serious loss of mud problem occurred. information on the shallow section, as it is the closest well to
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1781

Banjarpanji-1 and both wells are within the Wunut anticline The Geological constraint is the carbonate formation. As per
structural closure. Lessons from this well were applied to agreed well plan and lessons learned from offset wells, the shoe
combat the highly reactive overpressured shale. was to be set 10–20 feet (3–6 m) into the tight hard pan on the top
 Mobil Oil BD-1 and BD-2, over 50 km away. The BD-1 and BD-2 of the carbonate. The top of the carbonate section is generally found
are twin wells with two completely different results; one full of to be tight with a thickness of around 50 ft (w15 m) where no fluid
drilling problems and the other trouble free. The main design loss is expected. Below this cap, a pore pressure reversal is observed
difference is that in the second well, the 9-5/800 casing was set in offset wells where loss circulation is likely to be encountered.
inside the carbonate section. These two wells provided much This setting of 9-5/800 casing inside the carbonate cap allows the
insight into how deep carbonate wells in East Java should be over pressured zone overlying the Kujung Formation to be isolated
planned and executed. such that drilling could then continue into the lower pressured
 Kodeco KE-11E and KE-11G, 30 km away. The KE-11E well carbonate formation using a lower mud weight.
successfully drilled the Kujung carbonate by setting the 9-5/800 This carbonate section was not found at the prognosed depth at
casing at the top of the carbonate section. The KE-11G was drilled 8500 ft (2591 m), but based on predictive electric log result, the top
to over 15,500 ft (4730 m) but did not find the objective. These of Kujung formation and hence the setting depth of the 9-5/800
wells are good examples of how deep wells should be drilled. casing shoe could be as deep as 9600 ft (2926 m). Setting the casing
 Santos Jeruk-1 and Jeruk-2, 28 km away. Jeruk-1 encountered at the very top of the Kujung Formation was an important part of
kick and loss problems and provided good lessons for design well plan and accordingly frequent bottoms up circulations of
and drilling of later wells. Jeruk-2 was drilled successfully by cuttings were carried out to check if the carbonate formation had
having its 9-5/800 casing set inside the top of the Kujung been penetrated.
Formation, and provides further benchmark data on There were no other drilling constraints in this well. The hole
a successful drilling strategy. was in an excellent shape as evidenced by several trips (three bit
trips and six short trips in the hole section) made without any drag
Operational insights from the offset wells were incorporated or fill. The mud weight was sufficient to contain the formation
into the design, operational procedures and risk mitigation plans pressure as there was no unduly high formation gas and connection
for Banjarpanji-1. The two key learnings incorporated in the well gas observed from the well. The Static Influx Test conducted at
design were: 9010 ft (2746 m) confirmed that the mud weight used was above
the formation pressure.
 The importance of setting the 9-5/800 casing inside the top of All systems were set to drill to the top of the Kujung carbonate
the Kujung Formation, and or 9400 ft (2865 m) whichever was encountered first. However, the
 The use of synthetic oil based mud to drill the highly reactive loss of circulation at 9297 feet (2834 m) signaled that the well
and overpressured shale sequence. condition had changed substantially and that a casing string must
be set at this depth once the loss problem had been stabilized. The
safety of this deeper than proposed casing setting depth was later
Appendix B: length of the open hole section verified using commercially available casing-design software, and
found to be acceptable, as shown in Fig. 8.
The length of the open hole in a well is typically dictated by The physical length of open hole itself is not an issue provided
three constraints, which are: that the above three criteria are met. As an example two recent
offset wells drilled by other operator nearby to the same objective
1 Kick Tolerance limit (safety consideration) have a longer open hole section (up to 6700 ft or 2042 m) compared
2 Geological constraints to 5717 ft (1743 m) of this well.
3 Other drilling constraints In summary, the operating procedure and well plan followed in
this well is fairly standard and in line with that followed by other
Kick tolerance is the amount of volume and pressure of gas operators in the area is proven by the number of offset wells. It is
influx at bottom-hole conditions that can be safely taken, shut-in therefore believed that factors other than drilling mechanics
and circulated out of the well without fracturing the weakest point, caused the mud eruption.
generally at the casing shoe. The volume (measured in bbls) of the
kick refers to the amount of gas influx that is allowed to be taken Appendix C: bottom hole pressure estimation
into the well-bore. The intensity (measured in ppg) of the kick
tolerance refers to the increase in mud weight required to balance In Banjarpanji, direct methods from the well were used to
the formation pressure. There are no industry standards for Kick estimate the bottom hole pressure. These methods have a high
Tolerance as it is very much case specific. Companies set their own degree of reliability since the factors involved are directly measured
standards based on their experience of drilling in the area (Unocal or observed from the well. The result from these methods are
Operating Guideline, 1998). In Banjarpanji well, the operator and its compared with values from calculated methods such as Electric
partners agreed to use a kick tolerance of around 10 bbl (1590 l) and logs and Drilling Dc-exponent (Bourgoyne et al., 1984) to get
0.5 ppg (0.59 MPa/km) gas kick, while nearby, a company with a better estimate of the bottom hole pressure.
more drilling experience drilling the Kujung formation, used kick
tolerances as low as 0 bbls and 0 ppg (0 L and 0 MPa/km). Engineer’s fill up
In Banjarpanji well, there was no constraint due to Kick Toler-
ance since the LOT was high (16.4 ppg or 19.3 MPa/km) and the This method is based on a physical phenomenon that during
Mud Weight was lower than expected (14.7 ppg or 17.3 MPa/km). a loss circulation event, the fluid level in the well will fall to a level
Therefore, with the agreed pre-defined Kick Tolerance limit it was that represents the pressure of the well. When pumping is resumed,
deemed safe to drill to around 9400 ft (2865 m) total depth the amount of mud pumped in the well until the first sign of fluid
provided there were no increases in mud weight. In this particular return is a good estimate of the volume necessary to fill the ‘void’.
case, the need to increase mud weight with depth was unlikely Knowing the capacity of the drill pipe and the annulus, one can
since the formation drilled were permeable sandstones. estimate the height of the fluid column in the pipe. The weight of the
1782 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

remaining mud column is inferred as the formation pore pressure. 1. The best estimate of the bottom hole pressure is around
This method of calculating the bottom hole pressure is widely used 12.8 ppg (15.06 MPa/km).
by field engineers as a quick and reliable bottom hole pressure 2. The bottom hole pressure is unlikely to be over 13.7 ppg
measurement. (16.12 MPa/km), with the highest possible bottom hole pres-
Using this procedure, the volume pumped when the first sign of sure of 14.7 ppg (17.29 MPa/km)
fluid return was observed was 2342 pump strokes or 220 bbls
(23980 l). The empty column height is therefore 1571 ft (489 m). In the Pressure Analysis section, the bottom hole pressure used
Assuming a fill up efficiency of 75%, the estimated pore pressure is is 12.8 ppg (15.06 MPa/km). The sensitivity test is performed with
12.8 ppg (15.06 MPa/km). As a sensitivity test, with a fill up effi- a maximum bottom hole pressure of 14.7 ppg (17.29 MPa/km).
ciency of 70% and 80%, the estimated pore pressure is 13.0 and
12.7 ppg (15.30 and 14.94 MPa/km) respectively. Appendix D: other pressure data

Static influx test Shut in Casing Pressure

This method is used with good success by a number of The maximum Casing Pressure of 1054 psi (7.27 MPa) is based
companies to estimate the upper limit of the formation pressure. In on the mud logger’s real time data (RTD) of May 28th, 2006. This
the new literature, this is referred to this as ‘‘micro-influx’’ method. casing pressure is considered reliable, as it is stable around 36 min
The mechanism of the method is by simulating a swabbing after shut in, and remained constant until it was bled off as part of
condition. Drilling process is halted; the drill pipe is then pulled to the well control procedure, as shown in Fig. 9 (upper picture).
generate swabbing conditions that lower down-hole pressure. A The (lower) pressures recorded at the choke and obtained
lowering of pressure by 0.5–1.0 ppg (0.59–1.18 MPa/km) is expected minutes after the well was shut-in are not considered valid indi-
as a result of this induced swabbing. If the mud weight in the well- cations of well bore pressure. For example, the Initial Shut In
bore was in close balance with the formation pressure, then an Casing Pressure (ISICP) reading of 350 psi (2.41 MPa) was not
under-balance condition will develop that induce an influx of stable and tended to increase. This increasing trend is believed to
hydrocarbon in the well-bore. The well is then circulated to observe be due to the migration of influx fluid up the wellbore replacing
the characteristics of the fluid at bottoms out. Unusually high gas the drilling mud.
content would indicate a close balance between the pore pressure
and the mud weight, and would provide a good estimate of the
Fluid density of the influx
formation pressure.
In Banjarpanji-1, the shut-in test was conducted several times
The well took a fluid influx of around 360 bbls. (57,240 l), which
with the deepest at 9010 ft (2746 m), all with negative results (no
represents approximately 30% of the hole volume. Accurate
influx observed). Since the mud weight in the wellbore was
measure of the total influx is difficult to estimate because a number
14.7 ppg (17.29 MPa/km), it was inferred that the formation pore
of operation that was ongoing at the time.
pressure is of the order of 13.7 ppg (16.12 MPa/km) or lower.
This influx fluid migrated to the upper section of the well
because of its lighter density. When the annulus pressure reached
Mud weight its stable period, it is likely that the whole volume of influx has
reached the surface and that the exchange of mud and influx is now
The mud weight used to drill to total depth was 14.7 ppg completed.
(17.29 MPa/km). At total depth, the well suffered a loss, which During the kill process, this influx was circulated out and the
means the pore pressure of the well must be lower than and cannot well was bled off. The influx fluid was found to be saline water with
be any higher than the mud weight. If the pore pressure is higher a density of 8.9 ppg (10.47 MPa/km), as shown in the well’s IADC
than the mud weight, the well would have suffered a kick instead of report and the drilling morning report dated May 29th, 2006.
a loss. Therefore 14.7 ppg (17.29 MPa/km) is the absolute upper end
of the pore pressure.
13-3/800 casing shoe Leak Off Test

Indirect methods A casing shoe leak off test (LOT) measures the strength of the
formation at the casing shoe, which is the ability of the open-hole
The bottom hole pressure estimation using these methods are well-bore to resist fracturing. To calculate the LOT, the data needed
shown in Fig. 10 (upper picture). These are the result from other is the casing shoe depth, the mud weight and the surface pressure.
sources, with the following ‘most likely’ pore pressure at the Traditionally, a LOT involves injecting mud into the formation
bottom of the hole: until it ‘leaks’. The pressure when the formation starts to leak is
called the ‘Leak Off’ pressure. However, when compressible oil-
a. Dc exponent (source Elnusa) ¼ 13.5–14.0 ppg (15.88–16.48 based mud is used, this traditional method is less reliable as
MPa/km) repeatability is poor and choosing the leak-off pressure is
b. Resistivity log (source Singh and Dusseault) ¼ 12.5–13.5 ppg subjective.
(14.71–15.88 MPa/km) To have a better reliability, The LOT is done is by injecting mud
c. Sonic log (source Schlumberger) ¼ 12.8–13.0 ppg (15.06–15.30 into the formation until the injection pressure stabilize, stop the
MPa/km). pump, and measure the ‘closing pressure’. This ‘closing pressure’ is
equal to the ‘opening pressure’ of the formation and thus the Leak
Off Test pressure (Unocal Operating Guideline, 1998). This tech-
Summary of bottom hole pressure estimation nique has better repeatability and reduces any subjectivity in
picking the leak off point. This LOT technique, however, may
The results of these pressure estimation methods, is shown in present a safety issue in the hard rock country since the fracture
Fig. 10 (bottom picture). These results suggest the following: caused may be difficult to heal.
N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784 1783

In Banjarpanji well, at the 13-3/800 casing shoe the ‘closing pres- At 9277 feet (2828 m), the first mud loss of 20 bbls (3180 l) was
sure’ was 400 psi (2.76 MPa) (Drilling morning report May 6, 2006). recorded at 06:02 h on May 27th, 2006, some seven minutes after
With the casing shoe depth of 3580 feet (1091 m) and the mud weight the 6.3 Richter Scale magnitude Yogyakarta earthquake, (Fig. 12A).
of 14.2 ppg (16.71 MPa/km), the LOT is 16.4 ppg (19.29 MPa/km) The tremors from the earthquake were felt at the rig site. The mud
(Fig. 11, left picture). The quality of this data is high since it was losses healed, and drilling continued. Cutting samples circulated
measured at the cementing pump and witnessed by the drilling from bottom at this depth showed the gas readings, lithology and
supervisor who had access to both pressure and volume readings. biota were unchanged, and the calcimetry was constant at 4.4%
This value is consistent with LOT results from nearby Wunut wells carbonate.
(NW, approximately 1.5 km away) and Tanggulangin wells (NE, At 9297 ft (2834 m), the second mud loss was recorded at
approximately 3 km away). This was expected, as the shallow section 12:50 h on May 27th, 2006 (Fig. 12B), less than two hours after two
of Banjarpanji-1 lies within the Wunut anticlinal structure. This value major aftershocks that followed the main earthquake. The well
is however, significantly higher than the shallow section of Porong experienced a total loss of circulation.
well (NE, approximately 6 km away). Leak Off Test results of these The timing of the earthquake and two mud loss events suggests
fields including Banjarpanji well are shown in Fig. 11 (right picture). that despite the 260 km distance to the earthquake epicenter, the
earthquake had an impact down hole in the Banjarpanji-1 well.
Appendix E: pressure analysis – graphical method

The data to perform this pressure analysis is obtained from Not a typical carbonate mud loss event
Appendices C and D, as follows:
The second mud loss was a complete loss of return, with a total
a. Maximum casing pressure ¼ 1054 psi (7.27 MPa) loss of mud of approximately 130 bbls (20,670 l). In order to cure
b. Fluid in the upper part of the hole (influx fluid) ¼ 8.9 ppg the loss, a pill of 60 bbls (9540 l) of loss circulation material (LCM)
(10.47 MPa/km) was pumped, and the losses cured. The bit was pulled four stands
c. Fluid in the bottom part of the hole (mud) ¼ 14.7 ppg off bottom (to 8737 feet or 2663 m) and the mud was circulated
(17.29 MPa/km) while monitoring the condition of the well.
d. Bottom hole pressure (BHP) ¼ 12.8 ppg (15.06 MPa/km) After the well suffered a complete loss of returns, the rig crew
e. Leak off test (LOT) at the casing shoe (3580 ft or 1091m) ¼ 16.4 anticipated and was ready for a kick that typically follows such
ppg (19.29 MPa/km) a major loss of mud. However, a kick did not eventuate even after
eight hours of close observation. This was not a usual loss-and-kick
Fig. 8B graphically plots the various pressures and gradients by: sequence that typically occurs in carbonate formations such as that
recorded at Porong-1
i. Plotting the maximum annulus surface pressure of 1054 psi The other unusual characteristic of the major loss event was that
(7.27 MPa) at the surface. it was easily cured; simply by pumping 60 bbls (9540 l) of loss
ii. Drawing a line down from the 1054 psi (7.27 MPa) surface circulation material (LCM). If this major loss had occurred in
pressure, using the 8.9 ppg (10.47 MPa/km) influx fluid a permeable formation such as a fractured carbonate, then curing it
gradient. would typically require much more time and involve multiple pills
iii. Plotting the BHP of 12.8 ppg (15.06 MPa/km) at 9297 feet and different types of treatments including cement plugs. For
(2834 m) (bottom of well). example, in the offset Porong-1 well, curing the loss of circulation
iv. Drawing a line up from the BHP, using the 14.7 ppg required multiple pills of LCM and finally a bentonite squeeze was
(17.29 MPa/km) oil-based mud gradient. used.
v. These two lines intersect at around 6000 feet (1829 m). This This suggests that the loss may not be to a carbonate formation
depth is consistent with the large amount of water influx but due to other causes, such as to a suddenly reactivated fault
taken into the well. system creating an isolated fracture through the Banjapanji-1
vi. Graphically, the pressure at the casing shoe depth is found to wellbore. If the earthquakes created a fracture in this well, then it is
be 2710 psi (18.68 MPa). very likely that they created many other fractures and opened
existing fractures/faults in the area. Such re-activation of regional
Mathematically, the wellbore pressure at the casing shoe is as faults may have triggered the LUSI mud volcano.
follows:
Pressure at casing shoe ¼ Maximum casing pressure
þhydrostatic pressure of fluid Pulling out drill string and kick
P@3580 ¼ 1054 þ ð0:052  8:9  3580Þ
¼ 2710 psið18:68 MPaÞ The trip out was done slowly, on an average of 5 min per stand,
and no over-pull was recorded. This lack of over-pull is consistent
This is equivalent to a fluid hydrostatic pressure of 14.6 ppg with a good hole penetrating hard volcanoclastic sandstones and
(17.1 MPa/km), much lower than the LOT pressure (16.4 ppg or shale drilled with a synthetic oil base mud. Similarly, multiple trips
19.29 MPa/km). Therefore the shoe was still intact and unlikely to in and out of this hole section had not reported any drag or fill.
have been compromised. Electric logs run at 8750 feet (2,667 m) had shown ‘gun barrel’ hole
conditions. The trip details extracted from the real time data are
Appendix F: observation summarized in Table 2.
The pumping of mud to compensate for drill pipe displacement,
Timing of Earthquake and Drilling Mud Losses the slow five minutes per stand pulling speed, the absence of any
over-pull and the excellent condition of the wellbore, make it very
The Banjarpanji-1 well suffered two mud losses which coin- unlikely that the pulling out of hole operation provoked an influx
cided with the time of the main earthquake and its after-shocks. (swabbing).
1784 N. Sawolo et al. / Marine and Petroleum Geology 26 (2009) 1766–1784

Unconventional influx 2007, Indonesia Agency for the Assessment and Application of Technology,
Jakarta.
Koesoemadinata, R.P., 1980. Geologi Minyak dan Gas Bumi. Penerbit ITB, Bandung.
During the pulling out of hole operation, a large water influx Longley, I.M., 1997. The tectonostratigraphic evolution of SE Asia. In: Fraser, A.J.,
entered the wellbore. The combination of high influx rate and late Matthews, S.J., Murphy, R.W. (Eds.), Petroleum Geology of Southeast Asia.
BOP shut in resulted in a high influx volume of around 360 bbls Geological Society Special Publication, No. 126.
Mazzini, A., Akhmanov, G.G., Svensen, H., Planke, S., 2007a. Pulsating mud volca-
(57,240 l) that was taken into the wellbore. Accurate measure of the nism at LUSI. In: Proceedings of the International Geological Workshop on
total influx is difficult to estimate because a number of operations Sidoarjo Mud Volcano, Jakarta, IAGI-BPPT-LIPI, February 20–21, 2007, Indonesia
were ongoing simultaneously. Despite all this, this kick was easily Agency for the Assessment and Application of Technology, Jakarta.
Mazzini, A., Svensen, H., Akhmanov, G.G., Aloisi, G., Planke, S., Malthe-Sørenssen, A.,
killed. Istadi, B., 2007b. Triggering and dynamic evolution of LUSI mud volcano,
The real time data showed that the kick was killed and the well Indonesia. Earth and Planetary Science Letters 261, 375–388.
died within three hours after the BOP was shut in (Fig. 9). This Nawangsidi, D., 2007. In search of theory on flow mechanism of wild mud blow. In:
Proceedings of the International Geological Workshop on Sidoarjo Mud
behavior was in stark contrast to the Kujung Formation loss and Volcano, Jakarta, IAGI-BPPT-LIPI, February 20–21, 2007, Indonesia Agency for
kick incident in Porong-1, which took six days to kill. This again the Assessment and Application of Technology, Jakarta.
suggests that the influx may not be from a carbonate formation but Prasetyadi, C., Suparka, E.R., Harsolumakso, A.H., Sapiie, B., 2006. The occurrence
of a newly found eocene tectonic Melange in Karangsambung area, Central
similar to one that caused the losses of drilling mud a day earlier. Java. In: Proceedings. IAGI 35th Annual Meeting and Exhibition, Pekanbaru,
October 2006.
Appendix G. Supplementary data Rubiandini, R., Maarif, A.S., Wahid, S., Sukiadi, T.K., Wijoyo, S., Wibisono, Y.,
Saleh, M., Sumanta, K., SuhartoPrastiyo, Rohi, P.A., Julianto, M.D., 2008. In:
Haryono, M.Y. (Ed.), To Kill Lapindo’s well (Indonesian version). GMLL, Jakarta.
Drilling data during the critical times from May 25th to June 4th, Satyana, A.H., 2007. Geological disaster on the falls of Jenggala and Majapahit
2006 include 1. Daily drilling morning reports, 2. Daily geological Empires: a hypothesis of historical mud volcanoes eruptions based on historical
reports and 3. Daily mud logger’s reports which can be found in the chronicles of Kitab Pararaton, Serat Kanda, Babad Tanah Jawi; Folklore of Timun
Mas; Analogue to the Present LUSI Eruption; and Geological Analysis of the
online version, at doi:10.1016/j.marpetgeo.2009.04.002. Kendeng Depression – Brantas Delta, Joint Convention Bali 2007, the 36th IAGI,
the 32nd HAGI, and the 29th IATMI, Annual Convention and Exhibition, Bali,
References 13–16 November 2007.
Schiller, D.M., Seubert, B.W., Musliki, S., Abdullah, M., 1994. The Reservoir Potential
of globigerinid Sands in Indonesia. In: Proc. 23rd Annual Convention Proceed-
Abel, L.W., Bowden Sr., J.R., Campbell, P.J., 1994. Firefighting and Blowout Control.
ings of Indonesian Petroleum Association, vol. 1, p.189ff.
Wild Well Control, Inc. 52–56.
Smyth, H., Hall, R., Hamilton, J., Kinny, P., 2005. East Java: Cenozoic basins, volcanoes
Bourgoyne Jr., A.T., Chenevert, M.E., Millheim, K.K., Young Jr., F.S., 1984. Applied
and Ancient basement. In: Proceedings 30th IPA Convention, pp. 251–266.
Drilling Engineering. In: SPE Textbook Series, vol. 2 252–285.
Smyth, H., Hamilton, P.J., Hall, R., Kinny, P., 2007. The deep crust beneath island arcs:
Cyranoski, D., 2007. Indonesian eruption: Muddy waters. Nature 445, 812–815.
inherited Zircons reveal a Gondwana continental fragment beneath East Java,
Daly, M.C., Hooper, B.G.D., Smith, D.G., 1987. Tertiary plate tectonics and basin
Indonesia. Earth and Planetary Science Letters 258, 269–282.
evolution in Indonesia. In: Proceeding 16th IPA Convention.
Sribudiyani, Prasetya, I., Muchsin, N., Sapiie, B., Ryacudu, R., Asikin, S., Kunto, T.,
Davies, R.J., Swarbrick, R.E., Evans, R.J., Huuse, M., 2007. Birth of a mud volcano: East
Harsolumakso, A.H., Astono, P., Yulianto, I., 2003. The collision of the East Java
Java, 29 May 2006. GSA Today vol. 17, 4–9.
Microplate and its implication for hydrocarbon occurrences. In: The East Java
Davies, R.J., Brumm, M., Manga, M., Rubiandini, R., Swarbrick, R., Tingay, M., 2008.
Basin, 29th Annual Convention Proceedings of Indonesian Petroleum
The East Java mud volcano (2006 to present): an earthquake or drilling trigger?
Association.
Earth and Planetary Science Letters 272, 627–638.
Sudarman, S., Hendrasto, F., 2007. Hot mud flow at Sidoarjo. In: Proceedings of the
Duyfjes, J., 1936. Zur Geologie und Stratigraphie des Kendeng-gebietes zwischen
International Geological Workshop on Sidoarjo Mud Volcano, Jakarta, IAGI-
Trinil und Soerabaja (Java). De Ingenieur in Ned. Indie, Sect IV (8), 136–149.
BPPT-LIPI, February 20–21, 2007, Indonesia Agency for the Assessment and
Duyfjes, J.,1938. Toelichting bij Blad 116 (Sidoardjo), Geologische Kaart van Java Schaal
Application of Technology, Jakarta.
1: 100.000. Dienst van den Mijnbouw in Nederlandsch-Indie, Bandung. 79 hal.
Sutriono, E., 2007. Pemboran Sumur Eksplorasi Banjarpanji-1. In: Proceedings of the
Hall, R., 2001. Cenozoic reconstructions of SE Asia and the SW Pacific: changing the
International Geological Workshop on Sidoarjo Mud Volcano, Jakarta, IAGI-
patterns of land and sea. In: Metcalfe, I., Smith, J.M.B., Morwood, M.,
BPPT-LIPI, February 20–21, 2007, Indonesia Agency for the Assessment and
Davidson, I.D. (Eds.), Faunal and Floral Migrations and Evolution in SE Asia,
Application of Technology, Jakarta.
Äı̀ Australasia. Lisse: A.A. Balkema (Swets & Zeitlinger Publishers), pp. 35–56.
Tingay, M., Heidbach, O., Davies, R., Swarbrick, R., August 2008. Triggering of the
Hamilton, W., 1979. Tectonics of the Indonesian region. U.S. Geological Survey
Lusi mud eruption: earthquake versus drilling initiation. GEOLOGY 36 (8),
Professional Paper 1078, 345.
639–642.
Kadar, D., 1986. Neogene Planktonic Foraminifera1 Biostratigraphy of the South
Unocal Drilling Department Operating Guideline and Procedure: Casing Design
Central Java area, Indonesia. In: Spec. Publ., vol. 5. Geol. Res. Dev. Centre,
Parameter and Standard Leak Off Test (LOT) Procedure, 1998, revision 2.0.
Bandung. 1–82.
Willumsen, P., Schiller, D.M., 1994. High quality volcaniclastic sandstone reser-
Kadar, A.P., Kadar, D., Aziz, F., 2007. Pleistocene stratigraphy of Banjarpanji#1 well
voirs in East Java, Indonesia. In: IPA, 23rd Annual Convention, vol. I, pp.
and the surrounding area .In: Proceedings of the International Geological
101–111.
Workshop on Sidoarjo Mud Volcano, Jakarta, IAGI-BPPT-LIPI, February 20–21,

You might also like