The North Sea Central Graben offers rich rewards in the form of large gas condensate accumulations. However, these pools are usually at depths below 16,000 ft, and are associated with extremely high pore pressure and formation temperatures. The operational success of wildcat exploration wells to test these HP/HT (high pressure/high temperature) prospects depends on the accurate prediction of pressure and temperature in the prospect area.
The pressure regime for a Central Graben well was predicted using a model developed with Geopop, an oil industry-funded research group coordinated by Durham University in the UK. In addition, the range of geological outcomes was quantified and risked, providing a risk envelope for the planning and management of the drilling operation.
At the 1998 American Association of Drilling Engineers Forum, Onyia pointed out that in drilling, 90% of unscheduled downtime is related to pore pressure, fracture pressure, and pressure depletion (Swarbrick, 1999). The Central Graben HP/HT fairway consists of gas condensate in Jurassic and Triassic sandstone reservoirs, often in tilted fault blocks, and sometimes intimately associated with salt diapirs.
Pore pressure often exceeds 18 ppge (lb/gal mud weight equivalent) and approaches the commonly assumed lithostatic gradient of 1 psi/ft. The pore pressure encountered at any well location depends on several factors, the main ones being depth and rate of burial, strength of the overlying seal, reservoir relief, hydrocarbon type, hydrocarbon column height, and vertical distance from the structural crest.
Time and tectonics are important. Rapid subsidence traps more overpressure in shales, whereas time allows equilibration of pressure between permeable reservoirs and adjacent seals, and the slow bleeding of pressure from the system. In the Central Graben, pressure regimes can vary greatly, even in adjacent fault blocks, reducing reliance on offset data.
The pressure prediction derived was for the "expected" geological outcome. In reality, a range exists for each of the main prospect parameters such as reservoir, trap, and presence or absence of hydrocarbon. The range for each parameter was used to model the risk envelope adopted for planning the operational stages of the well.
The conventional practice of designing a well for the worst possible case was challenged, and replaced to a certain extent by applying risk based design methods. Peak Well Management was contracted by Enterprise Oil to design the well, in accordance with Enterprise Oil's recommended practices.
The prospect comprised Late Jurassic, shallow marine sandstone (LJ2M) as the primary target, below 16,000 ft. In addition, there was a 25% probability of a 30 ft turbidite sandstone being present above the LJ2M, at the well location. A blowout and fire had occurred in a nearby field when mud weight was increased to control pore pressure in one of these turbidites, leading to fracturing of the underlying LJ2M sandstone.
Furthermore, RFT (repeat formation tester) pressure data from offset fields, such as Erskine, show different hydrocarbon water contacts in the LJ2M and the underlying Pentland Formation reservoir, which are separated by a 120 ft shale unit.
Seismic mapping defined the trap as a tilted fault block, assumed to contain gas condensate. The reservoir was sealed by updip abutment against shale, across a normal fault, and by dip in all other directions. In an alternative structural model, updip seal was provided by abutment against a salt diapir, putting the well location further downdip from the structural crest.
A pressure prediction model was developed for the pre-Cretaceous section in the prospect area. The main contributions to overpressure are assumed to be under-compaction due to rapid burial, lateral transfer of pressure from deeper in the basin via the permeable reservoirs, and the buoyancy effect of the hydrocarbon column.
BasinMod software was used to calculate the sedimentation rate and the under-compaction coefficient. The contribution of pore pressure due to disequilibrium compaction (1-D model) was estimated using algorithms published in Mann and McKenzie (1990). The model was tested on nearby fields, where crestal pore pressures were predicted close to values derived from RFT data. The pressure profile for the Cretaceous and younger section can be derived more confidently from offset data because the relative absence of faulting limits abrupt lateral variations.
In applying the model to our prospect, the first step was to ascertain the strength of the overlying seal, since this dictates the maximum pressure that the reservoir can have. It was assumed that leak off test (LOT) data provides a close approximation of the pore pressure required to fracture the seal and breach the prospect.
Offset LOT data was analyzed, particularly those from the Cromer Knoll Group, in which the critical 9 5/8-in. casing was to be set. The relationship between LOT and depth was found to be approximately linear, increasing by 0.1 ppge per 350 ft down to 14, 500 ft TVDSS (true vertical depth subsea), and by 0.1 ppge per 600 ft below that depth.
Crestal pore pressure
Viewed in less detail, the line of maximum horizontal stress appears to approach the lithostatic gradient asymptotically, with increasing depth. The seal strength displayed a range of 19.0-19.5 ppge, with 19.3 ppge expected at the depth of our reservoir's crest. The crestal pore pressure in nearby fields was derived by extrapolating RFT gradients.
Only in the Shearwater Field was crestal pore pressure found to be close to the seal strength. In the other offset fields (75%), it was at least 1.2 ppge below the point of seal failure. In our prospect, pressure predictions were run assuming the crestal pore pressure to be 18.9 ppge for the expected case and 19.2 ppge for the worst case, (just below the 19.3 LOT prognosed).
We used a gas condensate density of 0.2 psi/ft, derived from RFT data from nearby wells. The hydrocarbon column height we used reflected the range of 600-1200-1800 ft observed in the gas fields nearby.
Although the Late Jurassic sediments plunge below 25,000 ft TVDSS off structure, the LJ2M reservoir was expected to shale out, limiting the depth from which pressure could be transferred. That depth is difficult to predict but the input used was consistent with our estimate of crestal pore pressure.
In the expected case (18.9 ppge crestal pressure, 1500 ft gas column), the well was located 564 ft below the structural crest. A pore pressure of 18.4 ppge (15,277 psi) was predicted in the turbidite sand at the well, if present, giving a margin of 711 psi between the pore pressure and fracture gradient. The corresponding prediction for the LJ2M at the well was 17.9 ppge (15,289 psi). In the worst case scenario (19.2 ppge crestal pressure, 1800 ft gas column), pore pressure of 19.0 ppge was calculated for the turbidite sand at the well, reducing the pressure margin to 150 psi.
Encountering pore pressure close to the fracture gradient, because of the geology or by drilling too close to the structural crest, was identified as a potential hazard, during a preliminary risk assessment carried out jointly by Enterprise Oil and Peak. In response, minimum, expected, and upside pressure models were constructed for the turbidite sand and LJ2M reservoirs separately.
The expected case for the turbidite sand was estimated to be 4.5 times more likely than the minimum or upside cases. For the LJ2M, the expected case was three times more likely than the minimum and 4.5 more likely than the upside (worst) case.
The traditional industry approach is to design casing for the worst possible circumstance. Given the range of possible outcomes on the prospect, this could have lead to an "impossible" design. The work carried out by the subsurface team allowed the range of possible outcomes to be narrowed. It was then possible to design the well for the upside or worst case.
The design was then carried out in the traditional fashion and according to Enterprise Oil policy. Despite the work done to limit the design envelope, the resulting casing program was still very heavy. The combined load of the 14 in. (94 lb/ft) casing and 10 3/4-in. (101 lb/ft) by 9 5/8-in. (53.5 lb/ft) composite casing string was the heaviest ever landed on a DrillQuip wellhead.
The design of the tie-back string was approached differently. In the event that the well was tested, two solutions were envisaged. During testing, the annulus fluid could be either the mud used for drilling the 8 1/2 in. hole or seawater.
Underbalanced testing, using seawater as an annulus fluid, is becoming more accepted. This technique reduces stress on the production casing and increases the reliability of testing tools, which may be affected by the high solid content of heavy drilling mud. The designer had to decide whether a liner tieback string was required for this well.
High specification casing commonly has an extended delivery period. Accordingly, it must be ordered well in advance of the operation, and in the event of a dry well may be unused. Moreover, there are problems associated with tieback strings, such as increased risks of swabbing when pulling the DST (drillstem test) string, and settling of the mud behind the tieback, which would in effect, negate its' supposed benefit of strengthening the well structure.
A quantitative risk analysis was carried out. The values used came either from published data or were derived using the experience of the people associated with the project, erring on the side of caution whenever no hard data was available. The analysis predicted the probability of catastrophic failure of the well as less than 10-4, which is considered an acceptable risk level in most industries. It was therefore concluded that a tieback string was not required for this well.
It has become standard industry practice to carry out a risk assessment exercise for all wells. A panel was assembled, with representatives of all the parties involved: Enterprise's asset team, Peak Well Management's engineering and operations management team, the drilling contractors, and all the principal contractors.
The well program was reviewed in detail, and possible unplanned events were identified. The team then assigned a probability value to each event occurring, as well as assigning a value to the severity of the consequences of such an event.
By multiplying the two numbers, a value is assigned to the risk associated with that event. The results surprised the team, emphasizing the influence of the human factor in the outcome of such a well. The riskiest events are listed in the table below.
- Inadequate 3rd party QA/QC
- Wellhead/casing wear
- Failure of primary cement job
- Human failure to detect a kick
- Team in different locations (London and Aberdeen)
- Lack of competence of critical personnel
- Personal injuries due to hot mud
- Running heavy casing strings
- Thick coals in Pentland Formation
- Lack of stress management
- Excess working hours.
All issues were addressed during the planning of the well, with a particular emphasis being placed on the human factors. Selection of key personnel was very thorough, and training sessions were organised to ensure that the crew was fully familiar with the program and procedures.
The well was drilled without incident and operationally was extremely successful, reaching total depth in 69 days. Some casing wear (identified as a major risk) occurred in the 14 in. casing, but this had no adverse consequences. The LOT at the 9 5/8-in. casing shoe, in the Cromer Knoll Group, was 19.4 ppge. The turbidite sand was absent due to erosion, and the pore pressure in the LJ2M reservoir at the well was reasonably close to the 17.9 ppge predicted.
The model developed for the Central Graben HP/HT area provided an acceptable prediction of pore pressure gradient from first principles. However, the maximum pore pressure gradient that is possible at the crest of an accumulation must be derived by rigorous examination of the offset well data to estimate seal strength. As well as seal strength estimate (LOT), offset data allows a range of pressure scenarios to be modeled and risked, and the appropriate well design and drilling strategy to be adopted.
Mann & Mackenzie (1990) "Prediction of pore fluid pressures in sedimentary basins," Marine and Petroleum Geology 7.
Swarbrick (1999) AADE forum, "Pressure regimes in sedimentary basins and their prediction," Marine and Petroleum Geology 16.
The authors wish to thank BP Exploration, Enterprise Oil, Texaco, Total, and Peak Well Management, for their permission to publish this article.
Ron Daniel is Staff Geologist with Enterprise Oil in London, where he has worked in International New Ventures and the Central North Sea Asset Team. He holds a BS from the University of Durham and began his 14-year oil industry career at BP Exploration.
Jean-Roch Graulier is a Director and Senior Well Engineer with Peak Well Management in Aberdeen. He started in the industry in 1980 with Total as a production technologist and drilling engineer, then worked at Geoservices and the Expro Group, before moving to Peak in 1997. He holds an MS in Offshore Engineering from Ecole Centrale des Arts et Manufactures in Paris.