New model improves deepwater fracture gradient values off Norway

Feb. 3, 2003
A new fracture gradient model correlates well with all deepwater drilling off Norway.

A new fracture gradient model correlates well with all deepwater drilling off Norway.

The model is based on water depth and overburden stress. Because bulk densities vary significantly at different deepwater locations, the model includes a new water-depth normalization method to correct for varying overburden density.

The Gulf of Mexico and offshore Brazil have had deepwater drilling for some time, but it is fairly new off Europe, with the first deepwater well drilled off Norway in 1997. On the UK side, deepwater drilling has mainly taken place along the Atlantic margin, close to the Shetland and the Orkney islands. More activity is expected, however, towards the Faroe Islands and Norway.

Generally, the deeper the water, the smaller the margins during drilling. The average overburden weight is low, resulting in a low fracture gradient. This implies that in very deep water (exceeding 10,000 ft), water is the best drilling fluid.

Several new concepts have overcome this problem. In the Gulf of Mexico, a "dual density" approach involves pumps installed at the subsea wellhead to provide additional lift up the marine riser. Use of this concept can reduce the number of casing strings in the drilling of a given well. 1

In Norway, one concept analyzed is a buoyant wellhead installed in shallower water. In this case, the wellhead rests on a 20-in. long marine riser atop a buoyant element kept in position by lines to the sea bottom. This allows wells to be drilled with standard drilling rigs.

There are two major groups of problems in the deepwater scenario.

First, problems such as circulation losses, hole collapse, and water inflow lead to costly stuck pipes, sidetracks, and top-hole abandonment.

With narrow margins, the setting depth of casing strings is an important issue.

The second group of problems relates to the design of the operation. The dual-density approach reduces the number of casing strings in a well, but it requires a costly pumping system at the wellhead and a drilling rig with deepwater capability.

The shallow buoyant wellhead can be used with ordinary drilling rigs, provided they have dynamic positioning systems. Both concepts are feasible.

This article focuses on the geomechanics of deepwater drilling and, more specifically, on the modeling and accuracy of the fracture gradient.

Deepwater fracturing pressure

Shallower sediments in deep water are usually in a relaxed state. The overburden and the horizontal stresses are caused by compaction alone.

Under such conditions, the fracture pressure correlates with the overburden stress.

Click here to enlarge image

The fracture pressure approaches the overburden at shallow seabed penetration, based on shallower North Sea data.2 A similar analysis, using data from the US Gulf Coast and Brazil, showed that on average, fracture pressure was equal to overburden stress.3 The model is based on an exponential compaction model.

Another study adapted the same philosophy, but normalized all data to seabed, removing the effects of different water depths.4

A good correlation was found by applying this method on deepwater wells.

Also, a water depth-normalization method was developed based on the same model. Estimating the overburden stress employed a model initially derived by Eaton.5

New results

A "worldwide" fracture model for deepwater applications has been based on field data from Norway, the US Gulf, the Shetland Islands, Angola, and Nigeria.4 Several new wells have been drilled off Norway recently, and the model has been revisited.

Several new observations have been made. First, the lithology varies considerably from well to well, resulting in different overburden stresses. Second, Eaton's overburden stress model overpredicts, at least off Norway.5 Based on this, the fracturing model will be modified.

Five recent deepwater wells off Norway have been analyzed in detail. The lithology and the bulk densities were analyzed to provide an accurate overburden stress curve in each well. This stress will serve as a fracturing gradient reference (Figs. 1-5). The overburden curves vary considerably.

Click here to enlarge image

For example, the very low overburden stress in the upper section in Fig. 5 results from a 600-m thick sequence of ooze sediments with a bulk density of 1.4-1.6 sp gr. This low density leads to a low overburden stress throughout the well.

All wells considered show different bulk density profiles, and hence slightly different overburden stress curves. This also explains the variation of the leak-off data.

Click here to enlarge image

Comparing the measured leak-off data to the overburden stress revealed that the best correlation was obtained when the fracture prognosis was about 97% of the overburden stress. Equation 1 (Equations box) shows the best-fit fracture model for deep water off Norway.

Click here to enlarge image

The specific overburden stress for each well is so. This equation is valid for seabed penetrations of 403-2,586 m. One assumption for application is normal pore pressure.

Figs. 1-5 show the results of this correlation.

Click here to enlarge image

The measured leak-off data points match the 97% overburden curve closely. The error is on average 2%, but it varies from zero to 4%. This uncertainty is considered acceptable.

Normalization methods

We propose to use Equation 1 as a fracture prognosis for new wells off Norway. In other places, we recommend the same equation to estimate fracturing and to compare measured leak-off data to the model. If reference wells are drilled in the same area, data from these might be used for comparison as well.

Click here to enlarge image

Because deepwater wells are drilled at different water depths, comparing their data requires they be normalized to the same reference depth. Assuming a similar bulk density profile (overburden stress) led to a normalization equation (Equation 2).4

In that equation, Index 1 refers to the initial data set, whereas Index 2 refers to the new (normalized) data set.

Click here to enlarge image

These equations refer to the fracture pressure, but other parameters that depend on the overburden stress and reference height can be normalized as well, such as in situ stresses and pore pressure.

In other words, the frac pressure gradient is now normalized both for the water depth and for the drill floor elevation. Example 1 (Example box) shows how it works.

The overburden stress varies, as shown when comparing Figs. 1-5. Therefore, the normalization method must be modified to take this into account.

Click here to enlarge image

Assume that Index 1 still refers to the initial data and Index 2 to the normalized data. In addition, assume that the overburden stress gradient below seabed (measured from the seabed) is dob1, whereas the stress gradient for the normalized condition is dob2.

The normalization equations become, for this case, Equation 3. Example 2 shows how this works.

We observe that a lower bulk density in Well 2 leads to a decrease in overburden stress, resulting in a lower fracture prognosis in deeper water.

Fig. 6 plots the two examples. In addition to normalizing the data to a water depth of 1,000 m, another example normalizes the same data to 2,000-m water depth.

The average overburden stress decreases with increased water depth because in the latter case, water constitutes most of the total overburden stress.

Acknowledgment

The authors would like to thank Flemming Stene at Norsk Hydro for providing data to the project, and Terje Magnussen at Atlantis Deepwater Holding for providing information about the artificial seabed system.

References

1. Herrmann, R.P., Smith, J.R., and Bourgoyne, A.T., "Application of dual-density gas lift to deepwater drilling," Deepwater Technology, October 2001, pp. 17-22.

2. Aadnoy, B.S., Soteland, T., and Ellingsen, B., "Casing point selection at shallow depth," Journal of Petroleum Science and Engineering, Vol. 6 (1991), pp. 45-55.

3. Rocha, L.A., and Bourgoyne, A.T., "A new simple method to estimate fracture gradient," SPE Drilling and Completion, September 1996, Vol. 2h, No.3, pp. 153-59.

4. Aadnoy, B.S., "Geomechanical analysis for deep-water drilling," Paper IADC/SPE 39339, presented at the IADC/SPE Drilling Conference, Dallas, March 1998.

5. Eaton, B.A., "Fracture gradient prediction and its application in oilfield operations," Journal of Petroleum Technology, October 1969, pp. 1353-60.

The authors

Click here to enlarge image

Bernt S. Aadnoy has been a professor of petroleum engineering at Stavanger University in Norway since 1994. He began his career with Phillips Petroleum Co. in 1978 and later worked for Rogaland Research, Statoil, and Saga Petroleum. He holds BS and MS degrees in mechanical and control engineering from the Universities of Wyoming and Texas, and a PhD in petroleum rock mechanics from the Norwegian Institute of Technology.

Click here to enlarge image

Roar Saetre holds a BS degree in petroleum engineering from Stavanger University. He recently completed an MS there, also in petroleum engineering. His specialty has been deepwater rock mechanics.