MONITORING OF BARITE SAG IMPORTANT IN DEVIATED DRILLING

Arild Saasen, Craig Marken, Njil Sterri
Rogaland Research
Stavanger, Norway

Jon Jakobsen
Phillips Petroleum Co. Norway
Tananger, Norway

Very low shear rate and oscillation rheometry techniques provide insight into the properties of drilling fluids that are associated with barite sag observed during drilling operations.

A detailed study of the rheological behavior of four field muds was completed with a controlled-stress rheometer. The techniques verified that detailed Theological studies of muds are needed to explain barite sag.

Barite sag can be the source of severe drilling and well control problems during the drilling of deviated wells. In a deviated well this phenomenon results from the gravitationally induced settling of the barite to form either a density gradient or a barite sedimentation bed.

Barite sag results if the rheological properties of the drilling fluid are inadequate to keep the weighing agent suspended. Improved rheological characterization of drilling fluids leads to a better understanding of barite sag and to the improvement of fluid properties that prevent sag.

Hanson, Trigg, Rachal, and Zamora have shown that barite sag occurs not only in a static annular fluid but, more importantly, also during low flow rate fluid circulation.1

Barite sag also occurs by a process described as Boycott settling in which a density gradient, and hence a pressure gradient, in a deviated well results in internal fluid circulation. Barite settling occurs more rapidly in a deviated well than in a vertical one. This phenomenon was first described in 1920 in studies unrelated to the oil industry.2

Jamison and Clements have observed that conventional viscosity parameters for drilling fluids may not be reliable indicators for predicting barite sag.3 Their tests found that the drilling fluid plastic viscosity, yield point, and gel strengths showed only minimal relationship to their measurements of sag.

Additives improved the sag problem but had only minimal effect on conventional rheological properties. Advances in rheology measurements of liquids and slurries, particularly at low shear rates, have expanded the characterization of these materials. These advances emphasized the complexity of describing non-Newtonian fluids.

Furthermore, the techniques also emphasized that the methods used to determine drilling fluid properties at the rig site and at operational laboratories do not reveal the true characteristics and potential performance of these fluids. This is particularly true for investigations of very low shear rate phenomena on such as barite sag. At these shear rates, which are typically less than 2 sec-1, the effects from yield stresses and gel formation or breakup processes become important.

BARITE SETTLING

Barite settling can occur prior to the formation of an adequate gel structure in the fluid.

If mud circulation is stopped, the mud will normally start to gel. Without sufficient gel development, particle separation (Boycott settling) can occur in deviated holes. The particle moves vertically down in an inclined hole.

This results in a lower density slurry in the high side of the hole and a more dense slurry at the low side of the hole. Hence, the dense mud will flow downwards and light mud will move upwards to balance local hydrostatic pressures.

In this manner, large beds of barite particles can be formed and will either remain at the low side of the hole or slump farther down the well. As a result of the increased density, viscosity, and gel strength within such a concentrated particle bed, difficulties can arise in resuspending the barite in the circulating mud.

Because the barite concentration has decreased in parts of the deviated well, hydrostatic pressure may be lost. If there is a slow movement of the fluid, gel formation can be prevented or minimized. Thus, particle separation can accelerate in deviated sections.

It is presumed that the circulation of the mud by the mud pump will keep the weighing agents suspended. However, under slow mud circulating conditions in a deviated well, barite can settle out. The slow circulation may be inadequate in removing the sedimented barite. In a static fluid, barite sag can be accelerated by slow movement or disturbances of the mud in connection with running logging tools. These movements can breakup the gel sufficiently to enhance Boycott settling.

During circulation, barite sag can be minimized through the use of a fluid with high viscosity at low shear rates. Likewise, settling of barite can be minimized with turbulent flow.

The high viscosity at low shear rates can be achieved with a fluid having a sufficiently high yield stress. This yield stress value is generally very different from the yield point as calculated using American Petroleum Institute (API) recommended procedures.5 Without large particle concentrations the conventional yield point is normally only an indication of a shear thinning behavior of the fluid.

Viscoelasticity and elongational viscosity are important factors that are not generally anticipated or evaluated in the drilling fluids industry. Viscoelastic fluids have a great ability to suspend particles.4 These elastic properties significantly affect particle settling in ways not explainable by conventional viscous properties. Barite sag should be minimized in a viscoelastic fluid.

The low shear rate viscosity and viscoelasticity of drilling fluids should affect barite sag. An implication has been made that low shear rate rheological measurements have been used in sag studies, but no observations or results were presented.3

TERMINOLOGY

Many who work in the drilling fluids industry are familiar with Theological procedures and gel tests recommended by API.5 Much of the terminology used in these recommendations is adequate but can lead to misunderstanding in more advanced interpretations.

The oil field terms such as yield point, plastic viscosity, and gel strength are used according to convention. However, additional terminology is necessary for a more accurate description of the fluid properties relevant to understanding the mud behavior.

For the purposes of the present discussion, yield stress will describe the stress at which the fluid starts to flow. Therefore, this yield stress will be a measured value and not a calculated value like yield point.

There is some debate among theoretical rheologists and engineers whether a yield stress of a fluid actually exists or remains just a useful construction for the interpretation of physical observations.6 For drilling fluid applications the yield stress concept can be valuable.

Similar to the yield stress, a gel stress is defined as the stress at which a gelled fluid starts to flow. This gel stress is determined directly at the first nonzero measured shear rate. The gel breakup is dependant on the strength of the gel.

A fragile gel is a gel that is broken immediately and totally when a stress larger than the gel stress is reached. Hence, the breakup behavior of the gel is described through fragility characteristics of the gel.

The gel strength as measured using the API procedures can never indicate lower gel stresses than the stress at 3 rpm on the Fann V-G meter. Experience with gel tests has shown that the AP[ gel strength numbers are generally much larger than the gel stresses. Thus, they must include parts of the fragility as well.

Little information is gained on what is actually measured by the API gel tests. For example, some information may be gained about gel formation while little is learned about the stresses required to break the gel.

RHEOLOGY MEASUREMENTS

The application of nontraditional Theological techniques can improve the understanding of drilling problems and application of drilling fluids.

These improved rheological techniques are most valuable as a supplement to the established techniques used in the laboratory and at the rig site.5

For the current studies, the rheology measurements were done with a Carri-med CSL 50 controlled-stress rheometer. A concentric cylinder geometry with a 1.950-mm gap was chosen over other available options, such as cone and plate configurations.

The gap size was required so that suspended particles as large as 120 R would not adversely affect the fluid flow and hence the measured drilling fluid properties. The gap must be more than ten times larger than the particles to prevent particle interference. The cone and plate geometry did not permit a gap of sufficient size to prevent particle interference and still obtain meaningful Theological data.

A thin film of oil was added on top of the measurement annulus to prevent sample fluid evaporation. Because of limits in the availability of pressurized instruments, all measurements were done at ambient pressure and 20 C.

For the comparison of the detailed Theological analysis, the properties of the fluids were measured following standard oil field techniques in the API recommended practices.5 To maintain homogeneity of the fluids, the muds were conditioned for 16 hr in a roller at room temperature prior to the measurements. The comparison measurements for mud rheology included 10-sec, 10-min, and 30-min gels from a 12-speed Fann V-G meter.

The controlled-stress rheometer operates differently than the Fann V-G rheometers commonly used by the drilling fluids industry. With the concentric cylinder configuration, the inner cylinder (the bob) rotates while the outer cylinder (the sample cup) remains stationary.

The rotational velocity is limited by a specified input torque. That is, the shear rate of the cylinder continuously changes to produce the torque limit set by the instrument. Hence, the shear rates are measured as a function of the shear stress (the set torque).

With this method, any yield stress values can be measured accurately because the stress results from the initial rotation of the inner cylinder. This instrument can record very low shear rates-as low as 0.05 sec-1. At the lower shear rates the measurements can be affected by the gel structure of the drilling fluid. Therefore, these controlled-stress techniques are useful for following the formation and breakup of gel structures.

SHEAR STRESS

To apply this controlled-stress technique, the maximum shear stress of the measurements must be defined (or even restricted by the limits of the instrument). For a comparison of shear stress vs. shear rate curves from a Fann V-G meter and the Carri-med instrument, the shear stress must be set. This stress (to, Pa) for the Carri-med instrument is established by setting the stress to the minimum of either of the low values defined in Equation 1:

[SEE FORMULA (1)]

In Equation 1, Q300 is the Fann V-G viscometer dial reading at 300 rpm, and 46-199 Pa is the maximum shear stress allowed with the present concentric cylinder geometry of the rheometer. The constant 0.45 was chosen so that shear rate would not exceed 582 sec-1, the limit with the concentric cylinder geometry.

The shear stress is fixed by the maximum shear rate in the first case or by the maximum torque in the latter case.

To limit the kind of data obtained and to ensure reproducibility, the controlled-stress rheometer was programmed with a set routine. For the shear stress vs. shear rate determinations, the following procedure was followed. Prior to any measurements the muds were sheared at the shear rate defined in Equation 1 for 2 min. The first shear rate vs. shear stress measurements were continuous over a shear stress range of 0 to to. Ramp up time from shear stress of 0 to to was 2 min followed by a 30-sec hold at maximum shear stress and then a 2-min ramp down time.

The gel characteristics of the mud were investigated by similar measurements. A maximum shear stress of t0/4 was set as the limit. After shearing the fluid for 2 min at the shear rate defined in Equation 1, the fluid was allowed to gel for different periods. Thereafter, the gelled fluid was sheared to the maximum shear stress of t0/4.

In these measurements, ramp up and ramp down times were both 1 min. There was no peak hold time. The gel times used for this study were 0 sec, 30 sec, 1 min, 5 min, 10 min, 15 min, and 30 min after the initial shearing was stopped.

VISCOELASTICITY

The Carri-med rheometer also has an oscillation function for investigating viscoelastic properties. Most drilling fluids should exhibit some viscoelastic properties that will impart some solid-like characteristic to the fluid to resist deformation.

The elastic properties cause the fluid to resist deformation when the inner cylinder is oscillated within the stationary outer cylinder. By this technique, the preset force of the instrument makes the cylinder oscillate with a strain rate taking the form -Y = Y0sin(wt), where w is the frequency, Y is the strain amplitude, and t is the time.

In response to the elastic deformation of the fluid, the torque oscillation differs from the cylinder oscillation by a changed amplitude and an added constant phase angle. This difference is used to calculate the linear viscoelastic properties as measured by the storage modulus, G', and the loss modulus, G". The resultant shear stress is given by Equation 2:

[SEE FORMULA (2)]

G' accounts for the stored elastic energy of the fluid, and G" accounts for the dissipated energy. The dissipated energy, or friction loss, is lost through the viscous properties of the fluid. For a nonelastic fluid G'= 0 and G" 0.

With the oscillation option, time sweep measurements were used to determine G' and G" for the drilling fluids. The frequency was set to 1 hz and the amplitude to 10 milliradians.

The 30-min measurements monitored the gel formation while the potential for breaking the gel was limited. The ratio G'/[(G')2 + (G" )2]0.5 indicates the solid-like nature of the fluid. If this ratio is zero, there are no elastic properties. If it is unity, the fluid has only the solid-like behavior. Gel formation is observed when this ratio increases with time.

CASE STUDIES

Low shear rate rheological measurement techniques were applied to four field muds used in drilling operations in the Gullfaks field in the North Sea. Of the four field cases, one well had severe barite sag and the other three had no major problems.

All four fluids were formulated with polyhydroxyl alcohols containing a mixture of polyglycerines and propylene glycol, as described by Peterson.7 8 The fluids also contained a shale inhibitor developed to work in conjunction with this polyhydroxyl alcohol mud system.

The occasional observation of barite sag in wells that used this mud system led to the investigation of low shear rate rheology measurements to minimize the occurrence of sag.

Table 1 contains the mud formulations used in these four wells along with the typical properties measured at the rig site.

CASE HISTORY 1

In Gullfaks Well A27-A, 548 ft of the 8 1/2-in. section were drilled with a drilling fluid containing polyhydroxyl alcohol. The deviation of this section of well was 23-24. Approximately 1,400 ft above this section the deviation was 58.

The mud as formulated (Table 1) proved stable to total depth and required minimal chemical treatment during these drilling operations. No problems were experienced with excessive torque or drag. Three wire line logs were run in the open hole with only minor overpull.

Severe barite sag was detected by determining the mud weight at 5-min intervals while circulating bottoms-up prior to running casing. The sag was believed to be a result of low gel strengths. In this section of the hole the 0-sec gel strengths were reported to be 3 lb/100 sq ft, and the 10-min gel readings ranged from 5 to 6 lb/100 sq ft.

CASE HISTORY 2

The second well, Gullfaks Well A24-A, was drilled using a similar drilling fluid in the 8 1/2-in. section. The angle of deviation was 42-46 in this section. About 3,900 ft above this section the deviation angle was 53.7.

During the drilling of the 649 ft of the 8 1/2-in. section, the mud system performed well. The 0-sec and 10-min gels had been increased from an initial 3 and 12 lb/100 sq ft to 10 and 30 lb/100 sq ft, respectively. Barite sag was negligible for this drilling operation.

This mud system was not used during the logging operation because well control problems related to water influx necessitated a change to oil-based muds.

CASE HISTORY 3

Similarly, Gullfaks Well A29 was drilled using this kind of drilling fluid. The 8 1/2-in. section was drilled for 895 ft with an angle of deviation of 20-26. A deviation of 56 occurred approximately 5,100 ft above this section.

Although the mud had been previously contaminated with cement, it proved extremely stable and required minimal chemical treatment. The 0-sec and 10-min gels were increased from an initial 6 and 29 lb/100 sq ft to 12 and 32 lb/100 sq ft, respectively. Barite sag was negligible for this section. The logs were run to bottom and the section logged without unusual overpull.

CASE HISTORY 4

The fourth well, Gullfaks Well A30, was also drilled with an 8 1/2-in. section. The angle of deviation was approximately 16 throughout the drilling of the 649-ft section. An 18.7 deviation was recorded about 2,035 ft above this section.

The section was drilled without problems, and there was negligible barite sag. This was not unanticipated, because the angle of deviation for this well was less than that of the other three cases.

The 0-sec and 10-min gels were increased from an initial 8 and 24 lb/100 sq ft to 11 and 29 lb/100 sq ft, respectively. There were no problems associated with the logging operations.

RESULTS

For an analysis of the nontraditional rheology measurements of drilling fluids, a comparison with the data from the standard measurements is required. Fig. 1 shows the shear stress vs. shear rate as measured on a 12-speed Fann V-G meter for the four muds.

Table 2 contains the rheological data and the gel strengths obtained in the laboratory for these field muds. Note that these data differ from the rig site data shown in Table 1. The rig site data were measured at 50 C., whereas the laboratory data were obtained at ambient temperatures.

Fig. 2 shows the shear stress vs. shear rate for the mud used in Well A27-A. Four curves are shown with measurements from 0-sec, 5-min, 10-min, and 15-min gels after shearing at the maximum shear rate. Similar measurements for the muds used in Wells A24-A, A29, and A30 are shown in Figs. 3, 4, and 5, respectively. In these figures, the measured gel times were 30 sec, 10 min, and 30 min after shearing the mud at maximum shear stress.

For the sake of convention, the shear rate is given as the horizontal axis. These figures also show the same kind of data at different gel times. The 0-sec gel time shows the base line curve that corresponds to the Fann V-G meter measurements.

These data were determined by measuring the shear rate as a function of shear stress on the controlled-stress rheometer (the inverse of the Fann V-G meter readings).

The continuous gel formation, measured by the oscillation rheometer as the storage modulus G' and the loss modulus G", for these field muds is represented in Fig. 6. The ratio G'/[(G') 2 + (G")2]0.5 is shown as a function of gelling time for the different muds.

With the oscillation option, time sweep measurements were used to determine G' and G". This indicates the solid-like structure of the gel as a function of gelling time.

GEL BREAKUP

As an example of the gel formation and breakup processes, the 30-min gel curve of Fig. 3 will be analyzed. One must remember that the stress on the bob surface is varied, and the shear rate is measured. The maximum shear stress is 11.81 Pa. There are approximately 100 equally spaced measurement points up to the maximum stress.

As shown in the figure, hardly any recognizable shear rate was measured until the stress reached approximately 6-7 Pa. At this point a significant, albeit small, change in shear rate occurs. This led to a considerable change in an apparent viscosity of this highly gelled fluid.

Hence, this point will be taken as the gel stress for this mud. As the stress is increased further, there is a steep shear-like-thinning curve which continues until the maximum stress is achieved. This "high viscous" region depicts the initial breakup process of the gel. If the gel had been broken fully after reaching the gel strength value, any increase of stress would result in a shift to the normal shear stress vs. shear rate curve.

Such a change would be represented as a nearly horizontal curve because shear rate is measured as a function of the shear stress. Immediately after reaching the maximum stress, the ramp down time is started. Without any gel, a curve very similar to the ramp up curve should be followed. As seen here, the shear rate increases with decreasing shear stress. This is a result of a continuing breakup process of the gel.

The point of maximum shear rate is not an indication of completion of the breakup process. The breakup process continues until the measurement curve reaches the ramp down curve of the nongelled mud. In the present example with this procedure, it seems impossible to shear the mud enough to reach the ramp down curve, as was the case for the 15-min measurements for the A27-A mud (Fig. 2) and the 30-min measurements for the A29 mud (Fig. 4).

Hence, complete breakup of the gel was not achieved with this measurement procedure for the 30-min gel of the A24-A mud.

The mud used in Well A27-A was more viscous at low shear rates than the mud used in Well A24-A. The actual shear stress curve from the controlled-stress rheometer should best be represented by the average between the up and down ramp curves for each mud.

The differences between the up and down ramp curves originates from using a measurement time span that is too short to compensate for inertia effects of the fluid. These inertia effects can lead to spurious results from a controlled-stress rheometer.9 The inertial effects from the measured fluid cannot be neglected just as steady state flow conditions cannot be totally achieved.

However, the positive contribution in the ramp up phase should approximately equal the negative contribution in the ramp down phase. Hence, the average should give the most accurate values.

The controlled-stress rheometer showed that the A27-A mud and the A24-A mud had nearly equal shear stress at a shear rate of 350 sec-1. At lower shear rates, less than 50 sec-1, the A27-A mud has greater shear stresses (see the 0-sec gel curves in Figs. 2-3).

At lower shear rates, the A27-A mud should exhibit better barite suspension properties than A24-A mud because of its higher viscosity at these shear rates. However, the severe barite sag was observed in the mud used in Well A27-A. These observations suggest that the barite sag should have occurred in this case when the mud was not circulating.

GEL STRESS

The gel buildup of the A27-A mud is shown in Fig. 2. There is little gel buildup within the first 10 min. However, a weak gel is developed slowly.

The gel stress is less than 1 Pa for gel times less than 5 min. At 10 min the gel stress value is 2.5 Pa, and at 15 min the gel stress is 3.3 Pa. The 10 min gel strength, as determined by the API procedure, is less than 5 Pa (Table 2).

The ramp down rheology curve very rapidly converges towards the 0-sec gel shear stress/shear rate curve. This indicates that the gel structure seems to be totally broken rather quickly.

Although the A24-A mud is less viscous at low shear rates, its gel development is very different as shown by Fig. 3. After 30 sec, the gel stress is greater than 1 Pa, followed by a very shear-thinning-like region up to a shear rate of 5 sec-1. This illustrates a part of the gel breakup process.

Additional measurements showed that after 1 min the gel stress had reached 2.2-2.3 Pa. The gel stress at 10 min has a value between 6.5 and 7 Pa followed by a strong shear-thinning-like region where gel breakup continues. The increased stress within this region, i.e., the low fragility characteristics, is probably one of the reasons why the API test gives such large values as 1 0 and 15 Pa at the 10 and 30-min gel times, respectively.

The gel is not broken immediately, and the gel continues to break even while reducing the shear stress during ramp down time. Little difference in the gel strength is observed if the gelling is allowed to continue for 15 or 30 min. The 15 min data are not illustrated. However, the shape was similar to the 10 min data but with a higher gel stress.

The gel breakup, however, is different. The different breakup properties of the 30-min gel can be seen from the increase in shear rate while reducing the shear stress and from the very low shear rate achieved at the maximum shear stress.

It is clear that the shearing cycle defined by the measurement procedure is insufficient to break the gel completely as the measurement values never approximate the 0-sec gel curve.

The 10-min gel characteristics of the muds used in Wells A29 and A30 were not dissimilar to those of the mud used in Well A24 A. The A29 mud had a gel stress of approximately 3 Pa at 10-min gel time (Fig. 4). This is 2 Pa less than that for the A24-A mud. Still, it is 1.5-2 Pa higher than the gel stress of the A27-A mud at the 10-min gel time.

The mud was significantly more fragile than the A24-A mud. This can be seen in Figs. 3-4 by comparing the breakup procedure which shows that A29 mud is probably completely broken.

This was not the case with the A24-A mud.

The A30 mud was rheologically similar to the A29 mud, except for having both lower viscosity and gel stress (Fig. 5).

Even though the muds used in Wells A29 and A30 were more fragile than the A24-A mud, they should have better properties than the A27-A mud regarding barite suspension. The gels are not broken immediately after reaching the gel stress. The curves of these muds have a strong shear-thinning-like region following the gel stress point.

This was not the case with the A27-A mud. Hence, the A27-A mud was much easier to break than the muds used in A29, A30, and A24-A.

OSCILLATION RHEOMETRY

Another way of illustrating the gel buildup is through the use of oscillation rheometry. The mud samples were evaluated using an oscillation of 1 hz frequency at 10 milliradian amplitude.

This represents a movement of the bob of 0.23 mm, which is approximately 12% of the gap width.

With the gelled mud oscillating within this gap under these conditions, the movement should not be large enough to break the gel. The gel formation over time can be followed by monitoring the linear viscoelastic parameters G' and G". As the gel develops, the mud should be more viscoelastic and have better barite suspension properties. With a stronger gel the relative value of the storage modulus should increase.

If the measurement is normalized, using G = [(G')2 + (G")2]0.5, the gel build up can be shown (Fig. 6). The vertical axis is the storage modulus normalized by G.

Hence, a value of zero represents a liquid-like fluid, indicating there is no gel. If the value is one, the gel has become completely solid-like and there is no possibility of flow within time scales shorter than 1 sec.

A good quality gel with respect to barite sag should have an immediate increase in the G'/G ratio with time. Then it should approach more slowly a value probably larger than 0.9 but never equal to 1.

The gel buildup measurements with the oscillation option demonstrate that the A24-A mud has a better gelling characteristic than the A27-A mud. However, the A27-A mud has a larger G'/G factor at short times after shearing.

This is probably a result of the chemical composition giving the enhanced shear thinning property of this mud at low shear rates.

The A27-A mud does not develop a gel with the same speed and strength as the A24-A mud. Actually, some of the increase in the G'/G values after 10 min are believed to result from a dryout at the sample surface in the rheometer. Likewise, the oscillation tests also show that the gel structure develops much faster for both the A29 and A30 mud than for the A27-A mud.

ACKNOWLEDGMENT

The authors thank International Drilling Fluids AS, Sandnes, Norway, for supplying the drilling fluids and field data for this article and the Norwegian Council for Industrial and Scientific Research, OT 41.23560, for its financial support.

REFERENCES

Hanson, P.M., Trigg, T.K., Jr., Rachal, G., and Zamora, M., "Investigation of barite 'sag' in weighted drilling fluids in highly deviated wells," SPE paper No. 20423, 65th Annual Technical Conference, New Orleans, Sept. 23-26, 1990.
Boycott, A.E., "Sedimentation of blood corpuscles," Nature, Vol. 104, p. 532, 1920.
Jamison, D.E., and Clements, W. R., "A new test method to characterize setting/sag tendencies of drilling fluids used in extended reach drilling," ASME 1990 Drilling Tech Symposium, PD Vol. 27, pp. 109-13, 1990.
Motley, T., and Hollamby, R., "Novel milling fluid saves time, cuts costs," World Oil, pp. 32-36, March 1987.
"Recommended practice standard procedure for field testing waterbased drilling fluids," API RP 13B-1, first edition, Washington, D.C., June 1, 1990.
Astarita, G., "The engineering reality of the yield stress," Journal of Rheology, Vol. 34, pp. 275-77, 1990.
Peterson, T.E., "Drilling and completion fluid," U.S. patent No. 4780220, Oct. 25, 1988.
Peterson, T.E., "Drilling and completion fluid," European patent No. 293191, Oct. 30, 1988.
Krieger, I.M., "Bingham award lecture 1989, the role of instrument inertia in controlled-stress rheometers," Journal of Rheology, Vol. 34, pp. 471-83, 1990.