It is well known that cement shrinks during hydration leading to a drop of stresses in the cement sheath below the hydrostatic pressure applied right after cement placement. This phenomenon might affect the integrity of the cement sheath under pressure and thermal loads taking place during the well lifecycle.
A standard practice in the industry is to add to the cement expansion additives to balance the effects of shrinkage. When designing the cement recipe, a recurrent question is the percentage of additives by weight of cement (BWC) that needs to be added to fulfill technical requirements, yet at the lowest possible cost. It is believed for example that exaggerated expansion could be counterproductive because of the development of too high stresses that might fracture the set cement.
Another important question is whether expansion can be activated without external water or pore pressure supply, which is the case if the cement is in contact with a shale formation or it is isolated from the reservoir by an impermeable mud cake or if the cement is placed between two casings. Cement permeability itself becomes an important parameter if the activation of expansion do require a source of water and/or pore pressure supply.
The API RP 10B-5 (ISO 10426-5) recommends to use either the annular ring test or the membrane test to measure shrinkage/expansion of well cement formulations at atmospheric pressure. In the case of the ring, the cement specimen is in direct contact with water while in the membrane test it is not. Many companies modified the protocol of ring test by applying a water pressure to mimic the hydrostatic well pressure and to be able to increase the temperature. The ring test can be considered to simulate the case of a cement isolating a permeable reservoir and the membrane test the case of a cement placed either in front of an impermeable formation (shale for instance) or between two casings. In practice, most of the time, expansion is evaluated in the ring setup without paying attention to its validity outside the conditions of this test.
In the recent years, Total has developed advanced cement testing devices that allow continuous measurement during hydration of volumetric strains, e.g. shrinkage/expansion, as well as water supply under realistic stress, drainage and temperature conditions. For the purpose of the work presented in this paper, three types of testing protocols were performed:
Drained tests in which the pore pressure is kept constant and the resulting in water inflow/outflow is monitored.
Undrained tests meaning zero water flow inducing changes of pore pressure that can be monitored by pressure sensors put at the two ends of the tested sample.
Hybrid tests starting by an undrained followed by a drained phase with the aim to test the cement under various levels of effective pressure, defined as the difference between confining and pore pressures.
In parallel, API annular ring tests, with and without pressure, were performed for the sake of comparison. Moreover, a theoretical model was modified on the light of the results of all these tests. This approach brought new understanding of the way the expansion is developing and most importantly its sensitivity to the effective stress and to water supply which vary significantly during cement hydration and possibly after and depends on the mechanical properties of the cemented formation. The results show clearly that API tests are insufficient to fully characterize shrinkage and expansion of a cement sheath.
The purpose of this paper is to first describe the advanced experimental set up and to compare its results with the ones obtained from tests recommended by API. Then, a theoretical model which simulates the process of hydration and subsequent shrinkage as well as expansion will be presented. It will be shown that to reproduce the observed behavior during laboratory tests and the differences between various testing protocols, it is necessary to introduce the concept of expansion force and to account for pore pressure and for water supply. From there, the model would be able to predict the efficiency of expansion additives and to optimize the expansion additive percentage BWC, should the expansion is believed to be active under local downhole conditions.