Well integrity/control

New Approach Tests Cement-Sheath Integrity During Thermal Cycling

Operators in the North Sea have been concerned about the ability of the cement sheath to maintain sealing integrity because of the increasing number of reported failures in mature wells. This paper presents results from a new laboratory setup to visualize the source of issues.

jpt-2016-05-fig1newapproach1.jpg
Fig. 1—Cement thermal-cycling cell.
Source: SPE173871

Well cement is placed into the annulus between casing and formation to provide structural support and zonal isolation throughout the well life cycle. Nevertheless, operators in the North Sea have been concerned about the ability of the cement sheath to maintain sealing integrity because of the increasing number of reported failures in mature wells. A new laboratory setup is designed to allow visualization of the development of possible leak paths throughout the cement sheath when exposed to pressure- and temperature-related varying loads.

Experimental Setup

The new laboratory setup presented in this work allows pressurization while maintaining the capability to perform a detailed study of the creation and propagation of cement-sheath failures upon thermal cyclic loads.

The cell is shown in Fig. 1 (above).

The applied formation materials were Saltwash North sandstone and Mancos shale.

Some of the current features of the cell are specified as follows:

  • Cell is X-ray transparent.
  • Temperature is controlled (from inside the casing) during cement setting and cyclic tests.
  • There is independent control of pressure inside casing on cement and confining pressure (around the rock) during cement setting and cyclic tests.
  • Maximum pressure is set at approximately 35 bar, while the allowable temperature range is set between −1°C and 150°C.
  • All components of the cell can be reused for several test samples.

In particular, the X-ray transparency is important because it allows for visualization of leak paths without the need to release pressure from the cell, or remove the internal test sample for a computed-tomography (CT) scan, or perform the traditional invasive method of cutting the test sample cross section. This was achieved by constructing the cell from aluminum, which has suitable radiolucent characteristics and allows the radiant energy to pass toward the test sample with little attenuation. The material also provides a lightweight cell to facilitate handling.
The pressurized cell has the potential to accommodate other types of measurements and to test diverse wellbore conditions. For instance, casing eccentricity and mud filter cake may be included in the tests. Moreover, the annular space between the rock and the surrounding sleeve provides room for the installation of additional measurement tools during thermal loading, such as acoustic emission sensors and thermocouples.

Testing Protocol and Sample Preparation

The aim of the current work was to introduce the testing protocol of the new pressurized cell and compare the thermal-cycling resistance of two well sections, Saltwash North sandstone and Mancos shale. The testing procedure for each sample is presented here:

  1. At ambient pressure, the casing, rock, and cell components are exposed to 66°C (cement-curing temperature).
  2. Typically after 1½ hours of heating the cell, cement is poured by gravity into the annulus through ports that are closed immediately afterward.
  3. Pressure around the rock and on the cement is increased to 35 bar.
  4. Pressure and temperature are maintained for 5 days to cure the cement.
  5. The cell is cooled to ambient temperature (approximately 16°C) while pressure is kept at 35 bar.
  6. The cell is disconnected from the hydraulic system and then placed on a CT scanner for visualization of the cement-sheath defects.
  7. The cell is coupled again with the laboratory setup to prepare for subsequent thermal cycles.
  8. Thermal-cycle profile and number of repetitions are programmed and executed.
  9. The cell is CT scanned as many times as required following Steps 4 through 6.

Thermal-Cycling Details

Heat is transferred by conduction from a thermal plate into the copper rod. The heat is nearly radially transferred to the pipe/cement/rock sample through a thin oil layer.

In this study, temperature has not been experimentally monitored at the cement sheath. However, numerical simulations of transient heat transfer have been performed to estimate the temperature changes through the sample. 3D finite-­element simulations were performed with commercial software. In the calculations, heat transfer from the thermal platform to the copper rod and then toward the sample and all cell components has been considered. Heat transfer was estimated as a conduction process among all the constituents of the test, including the oil film and nitrogen gas, in order to reduce the complexity of the transient model. Heat convection from the system to the surrounding environment was also considered.

Monitoring by CT

In order to obtain detailed 3D information on the size, geometry, and location of the cracks and debonded volumes in the cement sheath, CT scanning is applied for initial and post-cycling sample analysis.

To investigate the effect of a single thermal cycle or consecutive cycles on cement-sheath failures, the samples are taken to room temperature in order to create well-known and uniform temperature distributions for CT scanning.

Results

Both sandstone and shale rock samples were subjected to 10 thermal cycles. Each sample was investigated by CT at its initial condition and after the first and 10th thermal cycles. Such a comparison can reveal which sample defects arise during thermal cycling and which were initially introduced by the cementing procedure.

Bonding percentage is calculated by finding the surface of direct casing/cement and cement/formation contact for each sample. The area of this surface is further divided by the total available bonding area. This means that 100% bonding corresponds to the surfaces being in direct contact everywhere, while 0% bonding means that there is no direct contact between the surfaces. The cracks/voids percentage is defined as the ratio of their volumes to the total annulus volume.

By comparing the leak-path volumes for cracks and the bonding percentages, it is clear that crack propagation is the most significant failure mode when ­thermal-cycling loads are applied. Casing/cement debonding and cement/formation debonding are less significant, although more severe for the sandstone sample. For the thermal loads applied in the experiments, the increment in debonding has been the result of linking between adjacent debonded areas and cracks/voids that propagate tangentially from the casing toward the rock. Several of these failures were found along the cement sheath, although with greater amount in the sandstone sample. For both samples, only a few tensile radial cracks were seen to propagate from the casing toward the rock. Most of the failures were found to initiate at and propagate from locations where voids existed initially.

Discussion

Effect of Hydrostatic Pressure on ­Cement-Job Quality. Cement/casing and cement/rock bonding has been enhanced especially by hydrostatic pressure. The presence of pressure seems to aid in accommodating cement volumetric shrinkage, typically within 2.5 to 3% for Portland G cement, while maintaining the adhesion of cement/casing and cement/formation.

The sample cured at ambient pressure led to a critical situation of poor cement/casing bonding, with fairly clean microannuli that disable the initial cement-sheath integrity of this sample. Another interesting observation is that the condition of poor cement/casing bonding appears to benefit the adhesion to the formation. Therefore, the presence of hydrostatic pressure during the curing process is recommended to achieve a cement-sheath condition similar to that expected from a wellbore condition, with effective cement-placement design and casing centralization.

Cement-Failure Mechanisms Upon Thermal Cycling. It should be noted that most of the failures upon thermal cycling found from both pressurized samples have been initiated at and propagated from initial defects within the cement sheath. Failures propagate more rapidly in the cross section than in the longitudinal direction, with cracking being the most prevalent failure mode found along the cement sheath. The main origin of cracking seems to be shear (compressive) failures for both samples. A few tensile radial cracks were found in cement sheaths of samples and seem to be generated from initial voids along the casing. Debonding between cement and casing has not been significant for any sample. Furthermore, debonding between cement and formation appears to be related to propagation of shear failures toward the rock, as well as to links between initial voids along this interface that formed larger and continuous debonded areas. These interpretations are the result of the inspection of 3D reconstructions.

Debonding between cement and formation and cracking increase upon thermal cycling. The first thermal load delivered the largest detrimental effect to the cement sheath, whereas the effect of the subsequent 10 thermal cycles spread these failures with a lower but still significant rate. In addition, the cement-sheath damage appears to be more severe for the sandstone sample. Even though the cement sheaths obtained in the shale and sandstone samples do not correspond entirely in initial status, the relatively larger increment of cracks in the sandstone suggests that the sandstone is more vulnerable to failure propagation upon thermal cycling.

This article, written by Special Publications Editor Adam Wilson, contains highlights of paper SPE 173871, “Cement-Sheath Integrity During Thermal Cycling: A Novel Approach for Experimental Tests of Cement Systems,” by J. De Andrade, SPE, and S. Sangesland, SPE, Norwegian University of Science and Technology, and J. Todorovic, SPE, and T. Vrålstad, SPE, SINTEF Petroleum Research, prepared for the 2015 SPE Bergen One Day Seminar, Bergen, Norway, 22 April. The paper has not been peer reviewed.