Reliability-Based Casing Design Unlocks Reserves in a High-Pressure Gas Field

Fig. 1—Liner-collapse scenario with high-pressure water external and low-pressure gas internal.

You have access to this full article to experience the outstanding content available to SPE members and JPT subscribers.

To ensure continued access to JPT's content, please Sign In, JOIN SPE, or Subscribe to JPT

Increasing recovery was considered by lowering the reservoir-abandonment pressure below the initial design value. Design assumptions and equipment ratings were reviewed systematically to determine which aspects factored into the decision to change reservoir management. Collapse loading of the 10-in. production liner was identified as a key variable. This paper presents work performed to characterize the probability of collapse as a function of reservoir-abandonment pressure using reliability-based design (RBD).


For the high-pressure gas field under consideration, future-field-development studies were conducted alongside execution of the development. One such study investigated the potential application of gas compression to the field. Gas compression reduces the pressures downstream of the wellhead to allow depleted reservoirs to flow at lower pressures. The potential benefit is increased reserves recovery with effectively no increase in well cost.

The reservoir design and operating limit (RDOL) defines the lower allowable pressure from a reservoir perspective. Engineering studies have shown that this limit is set by the well design and operating limit (WDOL). If compression options are pursued, proposed depletion pressures should be checked for consistency with the limits prescribed by the RDOL and WDOL. With compression, when the wellhead pressure is reduced, the wellbore pressures may approach the lower end of the WDOL. This increases the pressure differential across the 10-in. liner section, which can result in a collapse failure, as illustrated in Fig. 1 above.

The section of the 10-in. liner acts as a barrier to high-pressure water-bearing sands below and possibly above the production packer. If a well is put on compression, the wellbore pressure will be reduced, thereby increasing the differential pressure across that section of the 10-in. liner. This induced high pressure differential can result in a collapse failure of the liner, and, subsequently, high-pressure water may enter the well, compromising well integrity and the ability to produce from the well.

The field basis of design stated that the wells should be able to withstand a minimum bottomhole pressure of 2,500 psi. This limit reflected collapse ratings for the 10-in. liner. At this downhole pressure, the feasibility of a compression project is marginal, with the incremental recovery insufficient to support the capital investment. Reservoir-­management studies indicate that, if the WDOL can be reduced to 1,000 psi, the compression phase of the project could be economical, with recovery potentially increasing by several hundred million BOE. The challenge was to find a way to lower the WDOL while maintaining well integrity and reliability, considering that many of the wells had already been drilled.

RBD Methodology

Working-stress design is the most common method used to evaluate casing and tubing. The acceptance criteria in ­working-stress design involves the computation of a design factor.

A yield-based performance rating is used for uniaxial design factors, and the material minimum yield strength is used for triaxial design factors. The calculated design factor is compared with minimum values set by company policy to determine whether the tubular meets design conditions.

Limit-state design follows a similar approach but uses a failure-based criterion rather than a yield-based rating. The ultimate limit state refers to the load that results in catastrophic failure. A limit-state design can use the same failure criterion but reflect specified minimum rather than actual values for input variables.

The design factor efficiently describes an acceptance criteria, albeit one that can be somewhat arbitrary. A shortcoming of the design factor is that it does not provide context around risk.

RBD can include uncertainty in casing strength, in load magnitude, and in frequency or likelihood of occurrence. Statistical methods for casing design have been used for years under the names probabilistic design, quantitative risk analysis, and load-resistance-factor design. Though specific implementations may differ, the fundamental principle involves a comparison of a distribution of strength to a distribution of load. The intersection of these distributions defines the probability of failure. Quantifying the load distribution can be difficult, particularly if the load is unlikely to occur. An alternative is to characterize the strength distribution and compare it with a single design load. The second approach calculates a probability of failure if the load case occurs.

Two statistical methods are used to predict the distribution of 10-in. casing collapse. The direct method builds a distribution of strength using sample data of physical collapse tests. This method is straightforward and easy to interpret, yet the underlying data may not be readily available. In contrast, the indirect method uses a limit-state equation and distributions for input variables to construct a distribution of strength. This method is more complicated to initialize, yet distributions of the input data—such as outside diameter, wall thickness, and yield strength—can be obtained easily. Please see the complete paper for a thorough discussion of the direct and indirect methods.

Initial Screening

The WDOL for the high-pressure gas wells was governed by the above-/­below-packer load. This load involved a leaking packer that results in the fluid in the A annulus falling to balance the pressure at the packer. This pressure is equal to the bottomhole flowing pressure with a gas gradient derived from the pressure/volume/temperature conditions up to the packer setting depth. The design load is illustrated in Fig. 2, which shows two completion options for the primary packer location. Design 1 results in a higher differential load across the 10-in. liner because the reduction in internal pressure above the packer reflects the completion fluid gradient. In contrast, Design 2 has the same bottomhole pressure but a lighter gas gradient inside the 10-in. liner.

Fig. 2—Collapse design load with two alternatives for packer setting depth.


The internal pressure gradient depends on the assumed level of compression and the associated flowing bottomhole pressure. The gradient is calculated using a standard equation-of-state model and a seven-component hydrocarbon that reflects field samples. External pressure is pore pressure reflecting a high-pressure water sand. The pore pressure behind the 10-in. liner varies across the field. The internal- and external-pressure profiles are combined to calculate the collapse load as a deterministic input for comparison with the distribution of pipe strength.

Wells for six areas of the field were checked for loads above the packer (Zone 1) and below the packer (Zone 2). The results for Completion Design 2 mirror those shown for Completion Design 1 Zone 2. For each zone and each well location, the highest design load was calculated for the interval of interest at 500‑psi increments of reservoir pressure.

The screening results were very encouraging. In all parts of the field, the capacity appears to exist to lower the abandonment pressure without raising the probability of collapse above the previously assumed 1/200 failure rate.


Conventional design-factor methodology did not provide context around the risk of decreasing operating pressure. A data-driven RBD model was built to support the decision to pursue compression and a lower WDOL. Well-specific analyses can be made to determine the operating limits for a particular well on the basis of factors such as completion geometry and casing wear.

Observations from the ongoing project include the following:

  • A focused program of data acquisition is key to reducing uncertainty. This involves data from collapse tests and from quality-control measurements.
  • A direct RBD model requires a significant number of collapse tests to narrow uncertainty. The present database of 99 tests reduces uncertainty to between 100 and 200 psi.
  • The largest variance in the indirect model is attributed to model uncertainty, which is the ratio of tested to predicted collapse strength. This finding underscores the importance of periodic collapse testing.
  • Incorporation of correlation between input variables, when available, appears to align the indirect model prediction with physical test results.
  • The small-data-set indirect model is somewhat cumbersome to apply and did not dramatically alter the prediction provided by a large-data-set Monte Carlo model, particularly when correlated variables are included.
This article, written by Special Publications Editor Adam Wilson, contains highlights of paper IADC/SPE 189669, “Unlocking Reserves in a BP-Operated High-Pressure Gas Field Through Reliability-Based Casing Design,” by Richard A. Miller, SPE, Rishi Ramtahal, SPE, and Oladele O. Owoeye, BP, prepared for the 2018 IADC/SPE Drilling Conference and Exhibition, Fort Worth, Texas, USA, 6–8 March. The paper has not been peer reviewed.

Reliability-Based Casing Design Unlocks Reserves in a High-Pressure Gas Field

01 June 2018

Volume: 70 | Issue: 6


Don't miss out on the latest technology delivered to your email weekly.  Sign up for the JPT newsletter.  If you are not logged in, you will receive a confirmation email that you will need to click on to confirm you want to receive the newsletter.