Summary During the last few decades, several laboratory investigations and field studies have been conducted in an attempt to find solutions to the problem of gas migration after primary or remedial cement jobs. This article reviews the primary or remedial cement jobs. This article reviews the general findings of previous investigators and offers an updated explanation of the mechanism of gas migration. Results of our laboratory studies show that "mobility" of the fluids in the pore spaces during the early life of the cement, particularly after the cement structure becomes load-bearing at a given hydrostatic pressure, is the main factor that must be controlled to minimize gas migration within the cement lattice. We also show that fluid-loss control alone, though helpful, is not sufficient to stop gas migration. This paper also describes an "impermeable" cement system developed by applying the principles outlined in the laboratory investigation. This impermeable cement has been used in field applications in areas where severe gas migration problems have been experienced after cementing. Thus, gas migration through the cement has been prevented when this new approach is used. Several of prevented when this new approach is used. Several of these case histories are presented and discussed. Introduction Literature Review. For many years the petroleum industry has recognized the problem of gas invasion of wellbores after cementing. In the early 1960's, Evans and Carter showed the importance of the condition of the pipe surface (roughness and wettability) in obtaining an pipe surface (roughness and wettability) in obtaining an effective bond at the casing/cement and cement/formation interfaces. In 1964, Bearden et al. introduced a special mechanical device that could be attached to the casing to control interzonal communication. The device consisted of a sealed ring of deformable rubber molded between two steel flanges, one of them movable. In 1966, Scott and Brace reported that primary cementing was improved by running resin-coated casing through completion intervals. The first published attempt to explain the problem of gas communication by means other than leakage at the casing/cement and cement/formation interfaces was presented by Carter and Slagle in 1970. The concept of presented by Carter and Slagle in 1970. The concept of the "inability of the cement column to effectively transmit full hydrostatic pressure" was formally introduced to the industry in that paper. In 1974, Stone and Christian used laboratory scale models to show that when the gas pressure was higher than the hydrostatic pressure after the cement had taken an initial set, a channel would form and gas would continue to migrate even after decreasing the formation gas pressure. In their recommendations, the authors brought up the need for good mud and cement placement practices as well as for the usage of cement placement practices as well as for the usage of cement slurries with good fluid-loss control and short setting times. The industry as a whole had been very aware of the need for proper displacement of the cement slurry to achieve good primary cement jobs. Even as early as 1948, Howard and Clark dealt extensively with the factors to be considered for proper casing cementing. Following the steps of previous investigators, Christian et al. in 1975 wrote a paper emphasizing the need to use cement slurries with good fluid-loss control to prevent gas migration. Their research indicated that prevent gas migration. Their research indicated that premature dehydration of cement slurries, resulting from premature dehydration of cement slurries, resulting from lack of fluid-loss control, may be the primary cause of gas communication. They proposed that fluid-loss additives effectively tie up the water required for hydration of cement and slowly release the water during the entire hydration process, as well as minimize the ability of fluids to flow through the cement porosity. In 1976, Garcia and Clarks ran a series of experiments and reported that annular gas influx was seen to occur if cement fluid-loss or uneven slurry setting occurred high in the hole such that hydrostatic head communication no longer existed between the bottom of the hole and the mud column above the set cement point. They indicated that while the cement slurry remained fluid, gas flow between zones was controlled. However, sometime after the cement set, gas flow began. Cook and Cunningham in 1977 presented an improved method for evaluating the fluid-loss requirements necessary to obtain successful liner or casing cementing jobs. They recommended the use of maximum fluid-loss control in cement slurries when cementing across zones of varying pressure to minimize gas leakage, since increased fluid-loss control resulted in less gas invasion and lower cement permeability. Another way to improve gas migration control, as reported in the literature, is to use expanding cements to promote better bonding at the casing/cement and promote better bonding at the casing/cement and cement/formation interfaces. One of the most recent papers dealing with this subject was presented by Griffin et al. in 1979; they discuss an expanding cement system that can provide superior bonding and zone isolation. A paper containing a series of practical techniques to control gas migration was written by Levine et al. in 1979. A graphical technique was introduced that predicts the potential of annular gas flow after cementing. Also in 1979, Tinsley et al. introduced, for the first time, a new cement system intended primarily to control gas migration at the cement/formation interface. JPT P. 1041

Postanalysis often cannot be applied because of lack of prediction and real-time performance data.Proper postevaluation includes collection of actual job performance data and comparison with predicted cementing performance. A comprehensive cementing simulator is required that accounts for fluid properties, well geometry, changing displacement and mud return rates, and the free-fall effect. Such a simulator has been used successfully in the design of many cementing operations and quite frequently in actual jobs to monitor the progress of the operation. In this paper, case histories are presented to show how this model and on-site measurements can be used for postevaluation of cementing jobs that did not perform as predicted. Two examples illustrate job problems caused by a restriction either in the annulus or in the casing; a third case points out problems associated with use of improperly calibrated pressure gauges during a cementing job.157

This paper describes a comprehensive circulation and shutin well-temperature-profile simulator capable of accounting for free fall during cementing operations. The program can predict temperatures during circulation of drilling mud in a well or during cementing. The system can also simulate periods of shut-in at any point during a given run. The simulator was verified by use of exact analytical solutions to certain problems. Comparison of simulator runs with field-measured well-temperature data indicated that the program does a good job of simulating well temperatures during circulation and shut-in.

This paper deals with the development of a mathematical model to describe the miscible displacement of drilling muds by cement slurries under laminar flow conditions. The model accounts for the effects of differing properties, geometry, and displacement rates. The model assumes that "mixing" in the displacement zone by molecular diffusion is minimal, and uses the Robertson-Stiff model to describe the rheological properties of both the drilling fluid and the cement slurry. The application of the model to a range of displacement conditions (densities, viscosities, yield stresses, displacement rates, etc.) indicates the conditions under which optimal or near optimal displacements are possible, and hence, provides a basis for designing efficient cementing operations from simple material property characterizations. Of special interest is the effect of the yield stress. These parameters are founded to strongly affect the displacement efficiency, particularly the formation of cement channels. Such results are described quantitatively in the paper as well as the effects of the other rheological properties, the densities, and the displacement rates. Field application cases are also included in this paper. Introduction Although many papers have been written on the subject of drilling mud displacement from wellbores during cementing operations, there are still many unresolved fundamental and practical questions. In particular, both laminar and turbulent flow conditions can produce good displacements; however, it is still not clear which represents the most effective displacement mechanism. Also, the displacement achieved under laminar conditions can vary greatly depending on the material properties of the fluids involved and the flow conditions. In some cases, very stable, high efficiency displacements are achieved, while in others, unstable fingering of the displacing phase occurs resulting in extremely low efficiency displacements. Even in the stable displacement cases, it has been only recently that efforts have been reported relating the displacement efficiency to the cement and mud material properties (densities and rheological properties) and the displacement rates. In the unstable cases, turbulent flow displacement would probably be dictated, but, as yet, there is no basis for estimating the displacement expected under different rates, nor the amount of the displacing phase which would be required. Further, the critical conditions separating stable and unstable displacement regimes have only been partially defined, and these relate only to Newtonian fluids, whereas drilling fluids and cements are highly non-Newtonian. It is clear that our knowledge of the fundamentals of the displacement processes in cementing is still quite limited, and, as a result, it is doubtful that many cementing operations are as effective or efficient as they might be if this information were available. Under normal conditions, satisfactory cement jobs are possible with suboptimal displacements; however, under possible with suboptimal displacements; however, under more demanding operations, such as displacement in permafrost zones, high efficiency displacements of permafrost zones, high efficiency displacements of the water base fluids are required. Clearly, in these latter situations a more accurate description of the displacement process is necessary, and such descriptions must influence the selection of the mud and cement properties to be used. properties to be used. In the present paper, we cannot consider all of the questions just raised. Instead, we focus on those aspects relating to the dependence of laminar flow displacement efficiency on the densities and rheological properties and the displacement rates. The fluids are considered to be non-Newtonian, and the displacement zone is taken as the narrow gap annuli between concentric cylinders, or equivalently, the region between two parallel plates. The approach is analytical and similar to that used previously by Flumerfelt; however, the rheological descriptions are more complete.

This paper evaluates the Robertson-Stiff rheological model for cement slurries and drilling muds. The model is compared with the Herschel-Bulkley model and is found to be an improved model for cement slurries. The Robertson-Stiff equations for volumetric flow rates in narrow annuli and tubes are shown to be limited to fluids with no yield stress. This paper develops more general equations that include the Robertson-Stiff results as special cases. Introduction In a recent paper, Robertson and Stiff presented a new model for describing the rheological behavior of drilling fluids and cement slurries. The principal advantages claimed for this model areit provides better fits of rheological data than other three-constant viscous models, andit gives explicit relations for the velocity fields, wall shear rates, and flow-rate/pressure-drop relations for flow in tubes and annuli. The latter advantage is not possible with comparable models such as the possible with comparable models such as the Herschel-Bulkley model. In this paper we provide an independent evaluation of the model with regard to these characteristics. First, we checked the accuracy of the model in fitting cement-slurry data and found that it does provide a good match to the experimental data. Fig. provide a good match to the experimental data. Fig. 1 shows the experimental data for a cement slurry along with fitting curves for the Robertson-Stiff and Herschel-Bulkley models. It can be seen that the Robertson-Stiff model is somewhat superior to the Herschel-Bulkley model in representing the data. This is also true in Fig. 2, which represents results for a different cement slurry of higher yield. Here again, the performance of the Robertson-Stiff model is somewhat better. Based on these observations, as well as others not presented here, we agree with the authors that their model is an improved model for cement slurries. With regard to the claim that the model provides explicit relations for the wall shear rate and pressure-drop/flow-rate relation, we have found pressure-drop/flow-rate relation, we have found certain errors in the Robertson-Stiff analysis that invalidate this claim and make their final equations applicable only in certain special cases. The basic error in their paper is that Robertson and Stiff overlooked the existence of a plug flow region in the center of the pipe or narrow annulus. Because of this, the equations derived are strictly valid only for the case of fluids with zero yield stress. Since most drilling fluids and cement slurries show some finite yield stress, the equations provided by Robertson and Stiff are of limited provided by Robertson and Stiff are of limited application. In the remainder of this paper we derive the correct relations for the model; the Robertson-Stiff relations appear as special cases. SPEJ P. 97