Seafloor Hazards

A team approach to mitigating shallow water flow risk

Since two wells were previously abandoned at Mississippi Canyon 849, a team of geohazards specialists alleviated the risk, allowing the drilling of a successful exploration well

Robert Bruce, Noble Energy, Inc., James McKeown, Fugro GeoServices, Inc., Tim Sargent, Noble Energy, Inc., Randal Garrett, Sierra Engineering

Wells drilled near the Mars, Ursa and Europa basins within the Mississippi Canyon Area, Gulf of Mexico, have experienced varying degrees of shallow water flow problems. Two wells drilled in Mississippi Canyon Block 849 (MC 849) penetrated overpressured sands within the shallow section, causing catastrophic shallow water flow. Both wells were subsequently abandoned. A team of geohazards specialists, geologists, geophysicists and engineers worked together to identify and mitigate this SWF, and successfully drilled the Slam Dunk exploration well in MC 849.


Mississippi Canyon Block 849 is located 145 mi southeast of New Orleans, Louisiana, in about 3,600 ft of water, Fig.1. The Pliocene target of the two wells previously drilled on the block, the Norcen-1 and UPRC-2, is a well-defined series of vertically stacked amplitudes within a thickening wedge of seismic reflectors. These amplitudes are stratigraphically trapped on the flank of a large salt ridge, which separates the prospect from the Mars basin to the east. While attempting to drill this prospect, both wells were lost due to shallow water flow (SWF). A seafloor amplitude map generated from a subsequent 3D geophysical survey by TGS over MC 849 showed areas of high seafloor amplitude around the Norcen and UPRC wells, which were interpreted to represent sand "mined" by the SWF. These areas are not visible in the original 3D seismic data acquired by Western Geophysical (Figs. 2 and 3). The SWF encountered at the Norcen and UPRC wells was so severe that during the planning of the Slam Dunk prospect, it was debated if drilling a successful well was feasible.

Fig 1

Fig. 1. Seafloor rendering showing Mississippi Canyon and MC Block 849.

Fig 2

Fig. 2. Seafloor amplitudes showing amplitude anomalies at the Norcen and UPRC wellsites.

Fig 3

Fig. 3. Seafloor amplitude map prior to drilling of Norcen and UPRC wells.

SWF is defined as water or sand and water flowing within and around the outside of structural well casing to the seafloor. 1 Isolating the specific stratigraphic interval where SWF was previously reported, and delineating the extent and distribution of the potential SWF sands were key factors in reducing the risk of encountering SWF at the Slam Dunk prospect.


The geohazards investigation was based on interpretation of reprocessed 3D seismic exploration data acquired by TGS. The 3D data cube contained 2-ms sample-rate data to a record length of 3.5 sec. TWT below sea surface. The exploration data was reprocessed with spectral whitening to enhance the higher frequencies in the tophole section. The dominant frequency of the 3D seismic data set was estimated to be about 50 Hz in the shallow section, which corresponds to a limit of separability of about 27.5 ft, assuming an average velocity of 5,500 ft/sec for the shallow sediments.

Offset well data provided for this study included LWD logs from the Norcen and UPRC wells, which allowed geophysical correlation to the proposed Slam Dunk prospect. Digital versions of the LWD logs were provided and integrated into the 3D seismic data with a velocity function generated for the Slam Dunk prospect. Drilling reports from the Norcen and UPRC wells were also utilized.


The Mississippi Fan is part of a larger accumulation of sediments deposited by the Mississippi River system during the late Pliocene and Pleistocene along the outer shelf, slope, and deep basin of the eastern Gulf of Mexico during glacial low-stands of sea level. Complex seismic facies within the fan reflects deposition by a wide variety of processes including turbidites, mass transport complexes, slumps, channel deposits and channel-overbank deposits. Primary fan deposition is apparently related to submarine canyons that cut across the slope and funneled sediments to the deep basin. The Mississippi Canyon is the youngest of these canyon systems. This canyon formed about 25,000 - 27,000 years B.P. and infilling commenced about 20,000 years B.P. The canyon, therefore, formed in a very short time, removing a minimum of 1,500 - 2,000 km 3 of materials in about 7,000 years. 2

Large-scale slumping on an unstable continental shelf-slope area represents the best explanation for the formation of the canyon based largely on the many large cuspate failure scars along the canyon walls. Some of the sediment masses on the present canyon floor can be traced back to those scars. 3 Thus, the Mississippi Canyon resulted from shelf-edge failures, followed by retrogressive slumping of shelf material around canyon margins. Rapid shelf progradation during lowering sea level probably produced a large mass of rather weak sediments. During the period 10,000 - 20,000 years B.P., a series of Late Wisconsin delta lobes almost filled the canyon with prodelta clays. 3 Hemipelagic sedimentation has been dominant during the last 10,000 years, and probably accounts for only tens of feet or less of the Quaternary section of the fan.

Multifold industry seismic data confirm this interpretation of the formation of the Mississippi Canyon and the depositional history of the Mississippi Fan. The availability of a relatively dense grid of industry multifold seismic data has allowed for a much more detailed analysis of the Middle Fan lying just south of the deformed slope. 4 In this area, 17 Quaternary sequences have been defined and mapped, each characterized by channel-levee-overbank systems.

Cyclic depositional patterns within each sequence probably reflect deposition in response to Pleistocene sea-level cycles. The lower, less organized seismic facies represent sediment eroded from submarine canyons during the fall in sea level, while the more organized channel-levee-overbank facies represent material funneled directly down the canyons to the deep basin during low stands and the following rise. The most recent lowstand was about 15,000 years ago and reached a minimum level near the shelf break at about 500 - 600 ft below present sea level. 5 Near present sea level was reached about 7,000 years ago.

At MC 849, two buried-channel complexes occur in the shallow stratigraphy within 2,000 ft below the water bottom. The most recent is the Young Timbalier Channel Complex, which underlies the Mississippi Canyon floor and has seafloor expression. Deeper in the section, is the Old Timbalier Channel, which extends from the northwest to the southeast through MC 849 and is probably composed of multiple cut-and-fill episodes that have crossed this part of the upper slope, Fig. 4.

Fig 4

Fig. 4. 3D seismic data example showing shallow geologic features within study area.

Deposition within the channels is very complex and variable, while some internal side-wall slumping of sand-prone strata toward the channel axis (thalweg) may have occurred during and after channel depositional events. The Old Timbalier Channel was probably a high-velocity system, as evidenced by its straight, relatively deep channel, Fig. 5. As a result, coarse-grained sediments may not have settled out of suspension within the portion of the channel that transects MC 849, and may have been deposited further down slope.

Fig 5

Fig. 5. 3D seismic data example showing Old Timbalier Channel incision through MC 849.

Underlying the Old Timbalier Channel Complex is an onlapping slope fan sequence. The fan steepens to the southeast and onlaps a diapiric ridge in the northern portion of the MC 849. A portion of this upper fan sequence was removed by the incision of the Old Timbalier Channel, Fig. 4. The fan is interpreted as a predominantly sand-prone unit. Seismically, the fan exhibits low-to-high amplitude, subparallel, discontinuous and chaotic reflectors.


Four subsurface horizons (10, 20, 30 and 40) were mapped in MC Block 849 to a depth limit of 1 sec TWT (~2,860 ft) below water bottom. These horizons separate four stratigraphic sequences (1 through 4) of distinct seismic and inferred lithologic character. Gamma ray and resistivity LWD logs from the Norcen and UPRC wells were correlated with the 3D seismic data and used to calibrate the lithologic interpretation of the shallow section in the vicinity of the proposed Slam Dunk wellsite.

Sequence 1 consists of mostly undisturbed, continuous parallel-stratified hemipelagic clays. Sequence 2 is interpreted to represent parallel-stratified clays with some interbedded sand-prone intervals and thin landslide deposits. The gamma ray logs indicate that the lower portion of Sequence 2 exhibits a fining-upward, sand-prone character. These wells were drilled along the northwestern margin of the Old Timbalier Channel. However, within the thalweg of the Old Timbalier Channel, reflectors display a low-to-moderate amplitude, discontinuous-to-chaotic character, interpreted as predominantly clay-prone deposits with some sandy lenses.

Within Sequence 2, as the Old Timbalier Channel incised into the underlying sand-prone slope fan, coarse-grained sediments may have been deposited through channel-margin failure. The steep, straight and, thus, high-velocity nature of the Old Timbalier Channel (Fig. 5), as well as the lack of high amplitude internal reflectors, suggests that coarser-grained sediments were probably deposited farther downslope of MC 849.

Sequence 3 is divided into upper and lower units. The upper unit of Sequence 3 comprises sand-prone slope fan deposits. The gamma ray and resistivity logs indicate sand-prone deposits within upper Sequence 3. The lower unit of Sequence 3 is interpreted as predominantly fine-grained submarine landslide/ debris flow deposits that may contain thin, discontinuous sands. Sequence 4 comprises fine-grained landslide/ debris flow deposits with some discontinuous sands. The gamma ray and resistivity logs indicate primarily clay-prone deposits with some interbedded sands within lower Sequence 3 and Sequence 4.


After reviewing the drilling data from the Norcen and UPRC wells and tying the well logs to the 3D seismic data, it was evident that the Old Timbalier Channel margins and the slope fan sequence presented the greatest potential for SWF. SWF was first observed in the Norcen well within a seismically dim sand-prone interval within Sequence 2 at about 4,600 ft below sea surface. SWF was observed in the UPRC well at the 20-in. casing annulus, which was open to the Sequence 2 sand interval. The slope fan sequence below the Old Timbalier Channel incision (Figs. 4 and 5) was also assessed as having a high potential for overpressure due to its interpreted high sand content and rapidly deposited, clay-prone overburden.

Sequence 2 sand. SWF began at the Sequence 2 sand in the Norcen well. The LWD logs and drilling reports from the Norcen and UPRC wells indicate a thin sand at about 4,600 ft below sea surface (~1,000 ft BML). In the 3D seismic, this interval correlates to a seismically dim trough reflector in Sequence 2 above the Old Timbalier Channel incision. The seismic attributes of this sand initially implied that it was potentially sand-starved, poorly developed and laterally truncated. However, current research indicates that it is difficult to seismically image SWF sands due to the relatively low sand/ shale contrast in acoustic impedance in the shallow section, 6 Fig. 7.

Fig 7

Fig. 7. 3D seismic record showing seismic/ well log correlation at MC 849 wellsites.

A 3D amplitude map of the Sequence 2 sand indicates what appears to be a channel levee depositional system, Fig. 6. The Norcen and UPRC wells are located in relatively higher amplitude associated with the levee on the northeastern margin of the system. There is a large, relatively dim amplitude area associated with the thalweg of the Old Timbalier Channel, indicating either the absence of, or poorly developed, Sequence 2 sand.

Fig 6

Fig. 6. Amplitude map of Sequence 2 sand showing channel/ levee complex.

Slope fan sequence. The slope fan sequence comprises subparallel, discontinuous and chaotic, low-to-high amplitude reflectors interpreted to represent a sand-prone slope fan with some interbedded clay-prone deposits. Gamma ray logs support this interpretation, but display a clay-prone response in the lower portion of the sequence (Fig. 4). This fan onlaps a salt diapir in northern MC 849, and slopes to the southeast. A portion of the fan has been removed by channeling events associated with the Old Timbalier Channel, which extends through MC 849 from the northwest to southeast (Figs. 4 and 5). This package is approximately 250 milliseconds thick at the Norcen and UPRC wells (Fig. 7). In the 3D seismic, the slope fan dips to the south at a high angle as the interval thickens dramatically.


After identifying the potential SWF intervals, a plan was developed to mitigate the risk. Accordingly, corrective measures would be taken to ensure that overpressured zones, if encountered, would not be allowed to flow. Mitigating the SWF potential while drilling would yield a stable wellbore and aid in successful cement isolation of the SWF zones. Additionally, it was recognized that if an SWF was encountered it would likely cause wellbore enlargement and undermining of the surrounding formation. SWF containment in this situation is problematic, with catastrophic wellbore failure common. Once SWF started at the Norcen and UPRC wells, it could not be controlled, and the wells were subsequently plugged and abandoned.

The drilling data from the Norcen and UPRC wells did not indicate whether the SWF was confined to the Sequence 2 sand or if there was additional flow from the deeper slope fan sequence (Figs. 4 and 7). Due to the interpreted high sand content, it was assumed that the slope fan could have contributed to the SWFs in both the Norcen and UPRC wells. Therefore, penetrating the slope fan sequence at its thinnest incised point would be favorable. Based on the drilling data, and due to the catastrophic nature of the SWF encountered at the Norcen and UPRC wells, the greatest SWF risk and drilling challenge was determined to be the Sequence 2 sand. It was believed that flow initiated in that sand and, once penetrated, the slope fan sequence could also flow. It was assumed that if these problematic zones began flowing, it would not be possible to contain them.

At the Norcen well, it was believed that the initial flow from the Sequence 2 sand undermined the surrounding formation causing a large washout in the wellbore, which contributed to insufficient isolation of the 20-in. casing annulus. Since the wellhead annulus was not sealed, the flow was able to continue unabated.

The UPRC well was drilled through the Sequence 2 sand with no indications of flow out of the 20-in. casing annulus until after the 16-in. casing was cemented below the slope fan sequence. The flow continued outside the 20-in. casing while the well was drilled to a total depth of 8,606 ft below sea surface. The 16-in. casing shoe was cement squeezed and a subsequent casing test proved unsuccessful. Communication with the existing flow behind the 20-in. casing indicated a mechanical failure of the 16-in. hanger seal or a casing connection, adding uncertainty as to whether the SWF was only from the Sequence 2 sand.

An amplitude map generated on the Sequence 2 sand indicated a channel/ levee depositional system, Fig. 6. The lack of relatively high amplitude within the thalweg of the Old Timbalier Channel was interpreted as relatively sand-poor deposition, which would eliminate much of the risk of SWF from the Sequence 2 sand interval. The straight, high-velocity nature of the Old Timbalier Channel was interpreted to have transported coarser-grained sediments farther downslope leaving the thalweg primarily clay-filled. The Old Timbalier Channel was sufficiently close to the prospect targets and would not require a high-angle directional well to test the prospect. Additionally, a surface location to the southwest and within the Old Timbalier Channel thalweg provided enough distance from the Norcen and UPRC wells to eliminate possible problems with higher pressured intervals associated with the SWF. Based on drilling data from the Norcen and UPRC wells, the slope fan interval presented the second highest risk for SWF. The slope fan dips to the southeast at about 28 to 30 and thickens dramatically basinward, giving the sands in this location the potential for increased hydraulic pressure from the basin to the south.

Within MC 849, the entire slope fan is incised by Old Timbalier Channel unconformity. The Slam Dunk surface location would have to "thread the needle" through the intervals identified as having high risk for SWF, penetrating each interval in the lowest risk position possible. A location 3,000 ft to the southwest of the Norcen and UPRC wells satisfied these criteria. At this location, the well would be in the thalweg of the Old Timbalier Channel and within an area of low Sequence 2 sand amplitude, interpreted as having the least sand potential. Also, it would be significantly down dip and far enough away from the Norcen and UPRC wells.

Two locations were permitted within 1,500 ft of each other. The rig would set up between the two locations and set anchors. Once anchors were set, the rig could then winch to the "A" location and drill. If problems were encountered, it would be possible to winch the rig to the second permitted location without having to re-set anchors.


The well design and operational plan for drilling through the shallow section focused on eliminating any SWF and lost circulation events. All available drilling data from the Norcen and UPRC wells, including LWD and pressure-while-drilling (PWD) information was analyzed for SWF and mud loss limitations. A favorable surface location was chosen which minimized the risk of SWF, while casing points were selected according to pore pressure and fracture gradient estimates as calibrated from the offset wells. Both the Norcen and UPRC wells flowed behind the 20-in. casing after cementing. As a result, emphasis was placed on casing design, cement seal design, and annulus sealing wellhead equipment for the Slam Dunk wellsite.

The Slam Dunk casing program required "jetting-in" the 36-in. conductor casing to 300 ft below water bottom for added wellhead structural stability. The 26-in. casing was set at 1,000 ft below water bottom in order to "top set" the Sequence 2 sand interval. The 26-in. casing was cemented using nitrified cement with 150% annulus excess to ensure returns to the surface. The SWF intervals were planned to be drilled using a riserless mud drilling technique ("pump and dump"), taking mud returns at the seafloor. The use of 11.2 ppg mud, in conjunction with the seawater gradient above the wellhead, yielded a hydrostatic equivalent of 9.8 ppg at TD and 9.3 ppg at the Sequence 2 sand interval, which was sufficient to contain the higher pressures within the SWF interval. Cement calculations were planned for 150% annulus excess to ensure returns to the seafloor using nitrified cement. To help eliminate possible "behind pipe" flow after cementing, a wellhead with four annular ball-valves was utilized. After cementing, the ball-valves were closed, thus greatly reducing the risk of an SWF.


After repeated attempts to drill the Pliocene prospect in MC 849, the Slam Dunk well was drilled as designed, under budget, and without SWF or loss-of-return events. The gamma ray and resistivity logs through the shallow section at the Slam Dunk well indicated very little sand development, especially within the Sequence 2 interval. The relatively high amplitude of the Sequence 2 sand along the Old Timbalier channel margins (Fig. 6) was representative of well-developed channel/ levee sands capable of SWF, while the channel thalweg was predominantly clay-prone, Fig. 7.

This project emphasizes the importance of a multi-disciplined team working together to successfully overcome potentially catastrophic events. This synergistic approach is becoming more essential as oil companies move into deeper water where well costs are high.  WO


3D seismic data examples are courtesy of TGS and Western Geophysical. Ralph Baird of Baird Petrophysical Int'l. Inc. provided the velocity function for the shallow subsurface. Thanks to Noble Energy, Inc., and Fugro GeoServices Inc. for permission to publish this paper. This article derives from OTC15249, which was presented at the 2003 Offshore Technology Conference held in Houston, Texas, May 5 - 8, 2003, and is presented with permission.


1 Alberty, M. W., M. E. Hafle, J. C. Minge and T. M. Byrd, "Mechanisms of shallow waterflows and drilling practices for intervention," Proceedings, Offshore Tech. Conference , OTC 8301, 1997.

2 Bouma, A. H., C. E. Stelting and J. M. Coleman, "Mississippi Fan: Internal structure and depositional processes," Geo-Marine Letters , v. 3, pp. 147 - 153, 1984.

3 Coleman, J. M., D. B. Prior and J. F. Lindsay, "Deltaic influences on shelf edge instability processes," in D.J. Stanley and G.T. Moore (eds.), The shelf break, critical interface on continental margins. Society of Economic Paleontologists and Mineralogists Special Publication, Vol. 33, pp. 121 - 137, 1983.

4 Weimer, P. "Sequence Stratigraphy, Facies Geometries, and Depositional History of the Mississippi Fan, Gulf of Mexico: American Association of Petroleum Geologists Bulletin , Vol. 74, pp. 425 - 453, 1990.

5 Ballard, R. C., and E. Uchupi, "Morphology and Quaternary history of the continental shelf in the Gulf Coast of the United States," Marine Science , Vol. 20, pp. 542 - 549, 1970.

6 Ostermeier, R. M., Pelletier, et al., "Dealing with shallow-water flow in the deepwater Gulf of Mexico" The Leading Edge , Vol. 21, pp. 660 - 668, 2002.



Robert C. Bruce earned a BS degree in geology from The College of Charleston, Charleston, South Carolina. He is primarily a prospect generator but is also responsible for hazard interpretation and SWF problems. He is currently a geophysical advisor in the Deep Water Business Unit of Noble Energy, Inc., in Houston. He is a member of SEG and AAPG and is a licensed geologist.


James M. McKeown holds an MS degree in geology from The University of Mississippi, Oxford, Mississippi, and is a licensed geologist in Texas. He is currently a project geologist with Fugro GeoServices, Inc., in Houston, where he is responsible for interpretation and mapping of 3D seismic exploration data and 2D high-resolution geophysical data, to assess potential geologic hazards and to predict conditions in the tophole section.


Tim Sargent earned a BSME degree in 1975 from Clemson University and immediately began his career in the oil industry at Amoco Production Co. Tim is currently employed at Noble Energy, Inc. as a sr. engineering advisor with supervisory responsibilities for GOM Deepwater and Shelf projects involving drilling, completion, or workover operations.