LIGHT WEIGHT THERMALLY STABLE CEMENT COMPOSITIONS AND METHOD OF USE
Field of Invention
This invention relates to cement compositions for lining a borehole. More particularly, the invention relates to light weight thermally stable cement compositions and their method of use.
Background of the Invention A cement blend has been developed in response to potential uses, in particular, in
Northern Alberta, Canada. The object is to design a cement system for weak formations with low fracture gradients to be applied at low bottom-hole temperatures with the anticipation that these wells will be steam flooded at a future date to extract out heavy oil from the ground. It is also specified that the distance between the point of application of steam and this subject cement blend system will be such that this cement around the production liner may experience a net temperature of very close to 230 deg C (446 deg F). One object of the invention is for this new blend, when mixed at a density of 1400 kg/m3 (11.69 Pounds Per Gallon, PPG), to be thermally stable at this temperature. In other words, at a specified light weight density this blend must develop sufficient compressive strength (CS), as required by the local regulatory board, yet must also be durable enough to sustain a high steam temperature for the rest of the life of the wdL
With the objects in mind as outlined above, and with the theories in mind that underlie thermal stability, a system has been designed where, first, the net lime/silica ratio woidd be iiπώed to a maximum of 1.0. The ratio is hdd to not exceed this value in order to assist in maintaining thermal stability. A CaO (limeySi02 (silica) ratio exceeding 1 0 may lead, according to cement chemistry, to a product called alpha dicalcium silicate hydrate. Such may comprise an undesired product because alpha-C2SH (C2SH is the cement ehwnjjtts abbreviation for a hydrated dicalcium silicate) is believed to be thermally unstable and may lead to a break down of the cement matrix. Most of the literature references are extensive in their treatment of normal density/high density cement designs (> 1800 kg/m3 or 15.0 PPG) in connection with applications for high temperature environments, including steam, fire flooding or geothermal wells. Seldom, however, does the literature deal with light weight cernent designs, particularly as light as dealt with in this invention. The only references found, tn fact, with respect to light weight designs, were the following:
(1) Gallus, J. P.; Pyle, D. E.; and Waiters, L. T.: Terfoπnance of Od-WdJ Cementing Compositions in Geothermal Wells," paper SPE 7591, 1978; and
(2) Galtus, J. P.; Pyle, D. E.; and Moran, L. K.: "Physical and Chemical
Properties of Cement Exposed to Geothermal Dry Steam," paper SPE 7876,
1979. The lowest density dealt in these articles is 1470 kg/m3 (12.3 PPG)for a system containing 35% silica flour, 8% perlite and 4% bentonhe.
Relating to the above references, two points should be emphasized vis-a-vis the present design:
A Bentonhe may accelerate deterioration of the hydration product (responsible for thermal stability) in the presence of any C02. See reference (3) Milestone, N. B.; Sugama, T.; Kukacka, L. E.; and Cardello, N.: "Carbonation of Geothermal Grouts-Pt 3 CO2 Attack on Grouts Containing Beπtonite,H Cement & Concrete Res. (1987) 17, 295- 306.
B. Perlite, an expanded crushed volcanic glass, has open and dosed pores. Under hydrostatic pressure, the open pores fill with water and the closed pores get crushed. As a result, the perlite gets heavier and the resultant density is higher than reported by the reference (2) above. Reference (2), however, mentions that up to 100% silica (by weight of cement) may be necessary in lower density systems to ensure slurry stability.
Another point of interest to be mentioned in regard to the cement blends of the present invention is the potentially beneficial use of blast furnace slag (BFS), a by-product of manufacture of iron from iron ores, for oil well cementing,
In the above references, the Galhis et al family of artides teaches Class G cement plus 35% silica flour yidding good compressive strength and low permeability for two (2) years at 460°F in geothermal dry steam. The Terformance of Oil Wdl Cementing Compositions in Geothermal Wdls" artide disdoses experiments with Class G cement and 40% to 100% suka of varying "20 mesh to 170 mesh" particle sizes. The artide teaches leveling of compressive strength and peπneability curves after about 6 months and that Class G cement apparently can be made effective by adding sufficient silica, 40% to 80%, to facilitate formation of desirable truscottite.
Other references in the area comprise: (4) "Wdl Cementing" edited by Erik Ndson and published by Elsevier Science
Publishers B. V. and distributed by Elsevier Sdence Publishing Company Inc .
New York, N. Y Chapter 9 titled Thermal Cements"
(5) Eilers, L. R; Ndson, E. B.; and Moran, L. K.; "High Temperature Cement Composttions-Pectolite, Scawtite, Truscottite, or Xonotfite: Which Do You Want?" Jour. Of Petr. Tech (July 1983) 1373-1377.
(6) Dillenbeck, R. L. D3; Mudler, D. T.; and Orr, B. R.; "The Effect of Miσosilica
On The Thermal Stability Of Lightwdght Cement Systems", CIM/SPE 116,
(1990). The 1980 SPE article by Eric B. Nelson, "High
Temperature Cement Compositions" teaches or suggests a lime to silica ratio less than or equal to 1.0 and a Portland cement plus a fine silica of at least 35%. Chapter 9 of the "Thermal Cement Book" edited by Erik B. Nelson tends to teach away from the present invention. The chapter teaches no real solution to the problem of adequate thermal recovery cements. The articles discuss a silica stabilized cement phis "microspheres" and a silica stabilized "foamed" cement (Portland phis 35% silica flour). Each procedure had problems. The article gave no solution for the particular application. The artide points out the necessity for a low density slurry. The effect of micro-silica (silica fume) on thermal stability of "light weight cement systems" is disdosed by experiments with silica fumes substituted for silica flour and a cement phis 35% silica mixture. Compressive strength goes progressively down as fume is substituted for flour. No permeability data is given. Test were at 177°C (350°F). The artide concludes a substitution of fume for flour "shows promise". The artide points out the substitution of fume for flour does give low density.
Summary of the Invention Novelties of the blends of the present inventions include:
* They are designed for placement at low bottom-hole temperatures and yet develop compressive strength at a reasonable amount of time, TTTJnirnizing the rig time, yet when the well is subjected to steam flood, this cement composition is capable of withstanding the severe temperature conditions
* Inclusive of BFS to contribute to the reaction kinetics towards strength development and thermal stability in the blends * High Silica concentration of the magnitude used here to meet the theoretical expectations
The invention comprises a thermally stable low density cement composition having a cσaφnaάw strength of greater than or equal to 3-45 MPa (500 psi) and permeability of less than or equal to 4 mp after at least 100 days of subjection to temperatures at or at above 200°C; and wherein the composition has a slurry density of less than 1400 Kg/m3 (12 ppg) and a lime/silica ratio of less than or equal to one. The composition comprises cement and between 50% to 100% silica (by weight of cement), cornprising silica flour and silica fume. Preferably, the composition includes 30% to 100% BFS (by weight of cement). Preferably, the composite exhibits a coinpressive strength of greater than or equal to 3-45 MPa (500 psi) after setting for at least 48 hours at a temperature less than 40°C (104°F).
The invention of a light weight thermally stable cement composition suitable for cementing well bores includes an embodiment of cement of Class C or the like; 33% to
50% BFS (by wright of cement), 33% to 80% silica flour (by wdght of cement), 13% to
50% silica fume (by wdght of cement), a lime to silica ratio of less than 1.0; and water sufEάent to produce a slurry having a density of 1400 Kg/m3 (<12.0 ppg).
Description of Preferred Embodiments One successful blend that tended to meet the criteria is as follows: Blend 1:
ASTM Type m (API Class C) Cement + 33% BFS (by wt of cement) + 33% Fine Silica flour (by wt of cement) + 20% Silica fume (by wt of cement). This blend was mixed with water at a resultant slurry density of 1400 kg/m3 and gave a yidd of 1.283 m3/tonne of the blend. The base blend was tested at the following temperatures to determine the amount of time (hours) needed to devdop 3.45 MPa (500 psi) compressive strength (local regulatory board requirement):
40 deg C 30 deg C 25 deg C
24 Hours 48 Hours 306 Hours The result of the study at a temperature of 230 deg C:
3-day 7-day 28-day
CS in MPa/Perm (md) CS in MPa/Perm (md) CS in MPa/Perm (md)
4.2 (609psiy2.45 3.5 (508psi)/2.89 2.9 (421ps.)/2.80
Though the trend of the compressive strength value at 230 deg C seems to indicate a possible strength retrogression for this blend, aO literature studies reflect a similar pattern for as long as up to the first 90 days after this sudden and extra ordinarily harsh environmental exposure to high temperature. For these systems, once they get accustomed to this situation, the general trend is a rise of the CS and decreasing perrneabihty.
One thing to notice in this BLEND 1 is that the permeability has been stabilized. Anodier encouraging result is that this 28-day sample has also been analyzed by X-ray diffraction (XRD) and Scanning Electron Microscopy (SEM). The resuh indicates the formation of a mineral known as Xonotlite in more than 80% (by wt of the hydrated cement) which is reported in the literature as a thermally stable (up to 315 deg C) cement hydration product, to be desired in a steam flooded/fire flooded/geoώeπnai wdl.
When BLEND 1 was modified by adjusting die ratio between silica flour and silica fume towards better thermal stability, by increasing silica flour (from 33% to 40%) and decreasing silica fume (20% to 13%), leaving the total Si02 content the same, the following results were observed:
3 -day 7-day 24-day*
CS in MPa/Perm (md) CS in MPa/Perm (md) CS in MPa/Perm (md)
2.75 (399psi) 4.24 2.54 (368psi)/4.10 4.00 (580psi)/3.88 * This test was stopped four days earlier due to a leak observed in the pressure pump in die curing chamber
Though die permeability observed in this modified BLEND 1 is higher than that of die original, the trend still seems to be coming down to die equilibrium state after this "shocking" period. The compressive strength of tiiis modified blend definitely looks much superior than the original. BLEND 2
ASTM Type HI (API Class C) Cement + 100% BFS (by wt of cement) + 50% Fine Silica Flour (by wt of cement) + 50% Silica Fume (by wt of cement). This blend was mixed with water at a resultant shiny density of 1400 kg/m3 for a yidd of 1.602 m3/tonne of the blend. The foUowing properties of this blend (BLEND 2) have been obtained.
By adjusting other adrnixtures in this blend, die following data on die compressive strength have been, thus far, generated:
(a) 40 deg C (4.0 MPa or 580 psi in 24 hours); (b) 230 deg C (6.50 MPa or 943 psi in 72 hours) When BLEND 2 was modified, similarry to BLEND 1 reported earlier, by merely changing the proportion of silica flour and fiime fincreasing silica flour from 50% to 80% by wt of cement and decreasing die silica fume from 50% to 20% by wt of cement), the following results were observed: (a) 40 deg C (4.8 MPa or 551 psi in 48 hours); (b) 25 deg C (3.5 MPa or 508 psi
in 179 hours). At an elevated temperature of 230 deg C, the following results were obtained:
3-day 7-day 24-day*
CS in MPa/Perm (md) CS in MPa/Perm (md) CS in MPa/Perm (md)
6.51 (944psi)/0.68 5.93 (860psiy0.46 7.60 (1102psiy0.50)
* The test was stopped four days earlier due to a leak observed in die pressure pump in the curing chamber It is obvious from this test result that die permeability has almost reached an equilibrium at 230 deg C in 24 days and the compressive strength is far superior than die otiier blends reported here. X-ray diffraction and scanning dectron micrograph, however, shows almost 81% Tobermorite crystals (C5S6H5 approximate) and the remaining 19% was quartz. General Discussion
The lime/silica ratio has been maintained at less than 1.0 in all four cement compositions reported above. BLEND 1 and its modified form have CaO/Si02 ratio of about 0.88 whereas BLEND 2 and its modification design have this lime/silica ratio of about 0.64. So, theoretically all these four systems should be thermally stable, suitable for pumping into wdls of bottom-hole temperature of up to 315 deg C. However, most of the literature data are for normal density cement slurries and not for low density systems A lower lime/silica ratio (i.e. higher silica concentration) has been suggested for lower density cement composition. The limited amount of data generated so far does support this theory. The percentage of silica, say in BLEND 1 and modified BLEND 1, is about 53%, whereas it is 100% in BLEND 2 and to rnocΗed version (both by wt of cement) It can be argued that the actual value of das silica concentration in these blends (all four) may be higher than what h may appear. The argument behind this statement is that slag though not considered as a pozzolanic material, instead is considered as a hydraulic material But it is weU lmownthat u^ rrydrauUc proo The silica contribution from BFS is qώe significant. So, a certain amount of this contribution should be considered in determining the total percentage of silica.
Normally, an alkaline system is needed to activate slag. 'Such activator may be available from die hydration product of Portland cement when a blend of cement and ilag is used. Sodium metasilicate has been reported as an activator for BFS, and a snail amount of sodium metasilicate has been included in "otiier admixtures" package used tn
'One such common activator is a mixture of sodium hydroxide and sodium carbonate
each of die four blends discussed here for controlling die set time or the slurry stability or other properties. Yet, a competition between tins sodium silicate and the Ca(0H)2 generated from die hydration of Portland cement in tiiese blends should occur. If any portion of this Ca(0H)2 is used up to activate slag, that will also slow down die overall reaction rate towards the formation of die thermally stable minerals. Sodium silicate may play another role in future stage of hydration to facilitate die formation of Pectolite, another very strong and stable crystal.
High silica concentration is needed to form Tobermorite (approx. formula C5S6H5 in die Cement chemist's notation) which is a strong binder. There is more tiian one crystal form of Tobermorite reported in literature. They are termed based on tiieir X- ray spacings. Most common of them are 14 Angstrom Tobermorite (C5S6H9, forms at around 60 deg C), 11.3 Angstrom one with approximate composition of C5S6H5 (forms between 110-140 deg C). The other Tobermorite (9.3 Angstrom one) of approx. composition C5S6H forms between 250-450 deg C. The 11.3 Angstrom Tobermorite with time and proper reaction environment will convert to Xo∞tiite (C5S5H at 3.65 Angstrom spacings and between 150-400 deg C) which reacts with excess quartz to form Truscottite (C6S10H3 at 19 Angstrom X-ray spacings and between 200-300 deg C). The majority amount of die 28-day hydration at 230 deg C of the BLEND 1 composition has been identified as Xonolite. If coarser than 325 mesh silica flour were used, instead of Xonolite, afwilike (C3S2H3), ldlchoarate, or cddo-Chcndrodhe would have been the product. None of these are as strong materials as Xonotiite. The major product in die modified BLEND 2 after 24 days of hydration at 230 deg C has been identified as 11 3 Angstrom Tobermorite instead of Xonotiite or Truscottite. The reason behind not finding Xonotiite or Truscottite in this blend can be explained as follows. BLEND 1 has more cementitous materials per unit mass compared to modified BLEND 2. More cementitous materials in die reactants should favor the formation of calcium silicate of the type suitable for that particular reaction environments. In addition to Λat, the reaction kinetics in BLEND 1 is defintdy going to be much faster than the modified BLEND 2 composition because the former contains relatively more finer particles per unit mass than the later These two reasons, just cited, should be enough to accderate die whde hvdration kinetics at some constant temperature (say 230 deg C) for BLEND 1 more than that of the modified BLEND 2. Needless to mention that the modified BLEND 2 was hydrated for 96 hours less at 230 deg C than that of the BLEND 1.
The reason for sdecting different kinds of silica (silica flour and silica fume) is as follows. The average partide size of silica fume is less than c*e miαon and that of sbca flour is about 15 micron. Coarser than silica flour is unsuitable for high temperature
stabilhy. So, a balance between die flour and fiime must exist. Silica fume will facilitate low temperature compressive strength development which is a necessity in this application to abide by die local regulatory board as well as to minimize the rig time. Silica fume also helps in bringing some stability of the slurry preventing excess free water separation. However, dry blending of high levels of silica fume in most of the blending procedures is difficult because of the inherent nature of these fumes of sticking to the surface of die vessd instead of having free flow character due to the generation of high statical dectric αirrent. Also, there are some scattered beliefs that a high concentration of silica fume does cause micro cracks in set cement. The effect of silica flour (325 mesh) for high temperature stability is wdl documented. This is the main reason for looking at a compromise between die levds of flour and fume so tiiat a suffident strength can be developed at relatively lower bottom-hole temperatures during placement, yielding minimum lwarting-on-cement' time while yet the composition has a sufficiently large amount of silica flour to favor the reaction towards die formation of thermally stable materials at devated temperature when the wefl is subjected to steam flooding. n,-y„gsinrt p#> BFS, in rwtm.%, and, rttwr \mιss
The advantages of including BFS are tiiat (a) because of its 1:1 lime:silica ratio, it is more diermalry stable than Portland; cβmenfr fh) it can accept the Kme pmAiced from the hydration of Portland cement and convert it to thermally stable Tobermorite, and other materials; (c) in addition to being a source of silica, BFS is synergistic with additional silica (externally added) to form thermally stable system at devated temperature; (d) instead of being just a source of sϋica, it is more reactive than silica or fly ash or any pozzolanic materials. The disadvantages are that pofff^hlf" "rti/flT fOri"1!!! iftniT present in BES are known to be dejnmsBal in "the fluid loss contror property. It is a source of alica and does enter into the calculation of the lime/silica ratio.
Additives are important in adjusting the Thkkaapf Time Rheologv. Fhiid Loss property, etc. needed for each particular operations and are not in the matrix of die base Blend Composition.
The rime/alica ratio of the batch of Class C cement we are using will be 65.3/21 9 = 2.98; to compute the lime/silica ratios of BIL-ErΦS 1 and 2 respectively:
BLEND 1 BLEND 2
Lime Silica Lime Silica
Class C cement (100 gms) 63.3 21.9 Class C Cement (100 gms) 65.3 21 9
BFS (33%BWOC; 33 gms) 11.9 12.5 BFS (100% BWOC; 100 gms) 36.2 37 8
Silica Flour + Fume Silica Flour + Fume
(53%BWOC; 53 gms) 53.0 (100% BWOC; 100 gms) 100.0
TOTAL m sii Ml 159.7
Ratio = 77.2/87.4 = .88 Ratio - 101.5/159.7 - 64
The above assumed both silica flour and silica fume are 100% SiO2.
The composites can be blended without BFS and still thev will work. Due to the higher cost of BFS, commercially, h may not become attractive to increase H further, or even to use it, at least in all situations. Compositions with a lime/silica ratio of as low as 0.64 are rare and have not been used in oil well cementing.
Use of more tiian 48 hours to devdop 3.45 MPa is undesirable. A maximum of 72-80 hours may be acceptable in very rare and difficult situations. Otiier than some local regulatory board requirements, longer time to devdop compressive strength means higher rig time which is expensive. Normally, as soon as sufficient compressrve strengdis is devdoped from the cement setting (called Waiting on Cement Time, WOC) to support die casing, drill crew can go back to die rig for further drilling.
If one were to monitor the compressive strength devdopment and the permeability of the set cement up to 180 days (6 months) for some sdected compositions, one would expect the general trend of the compressive strength to increase and die rjermeabilitv to decrease, based on die available known chemistry. From die compositional point of view, one can forecast the reaction kinetics from die available chemicals in the system and die reaction corxirtiona. Both the compressive strength and the permeabihty of the expected mineralogy is known and thus can be predicted. Them i« nn minimum permeability requirement. Depending on the area where that cement is located in die wdls, less than one millidarcv is considered acceptable. In some cases it may be 4=5. md and in others it may be 0.1 md or of similar order of permeability is needed.
Compressive strength of 3-45 MPa (500 psi) is not needed everywhere. Some bdieve that h is required for holding the casing. Others bdieve that even 0.7 MPa (100 psi) is adequate to hold a casing. But generally, no matter what one befieves and practices, up to a minimum of 3-45 MPa (500 psi) for the rest of die well's life is desirable.
The "additive package" is from the chemicals, calcium chloride, hgnosulfonates.
sodium metasilicate and polyvinyl alcohol. These chemicals are used for adjusting thickening time sfuπy stability (i.e., prevent from settling particles instead of maintaining a homogeneous sytem), fluid loss control and development of earfv compressive strength etc. The base composition of die cement blend remains die same and one can adjust the concentration or the level of hgnosulfonates, sodium metasilicate and poryviπyle alcohol for desired properties.
A thermally stable 1400 kg/m3 density cement design can be made without any BFS. In that case, one will have to recalculate the lime/sutca ratio to compensate for die loss of lime and silica from BFS, to maintain the same lime/silica ratio as was in die blend with BFS. This will change die kinetics of the hydration reaction. Based on limited study, it is predicted that it will take longer time to devdop compressive strength (CS), particularly at low bottom-hole temperature of cement placement.
Depending on die minimum CS required within a certain amount of time either by die local regulatory board or the customer, one will have to adjust the main blend composition whh other chemical additives (additive package). These additives are used in a relatively small amount compared to die cornponents of the blend and does not normally enter into lime/silica ratio calculation, nor in the calculation of the slurry density. At present, CaC12 and anhydrous sodium metasilicate are used to accderate the set of die cement slurry and/or to enhance the early ∞rnpressive strength There are other chemicals known to assist in these two properties just described.
Portland cement is hydraulic material that is h sets hard and devdops CS in water. Silica (flour or fume) is called pozzalanic (pozz) materials, that is it reacts with lime and produces cementitious product. (So, pozz becomes hydraulic when comes in contact with lime.) BFS is called a hydraulic material. If one can measure the degree of hydraulicness, then BFS wifl definrtefy be categorized as a material whh much less hydraulic property than cement.
The
of BFS in these blends is that h is thermaUy more stable than Portland cement because the lime/silica ratio of BFS is more dose to the tiieoretical desired value of 1.0. Also, BFS can react or consume die Hme produced from the hydration of Portland cement faster and convert h to a higher degree of thermally stable tobermorite than the pozz can react with in-shu lime. BFS will thus, assist in the devdopment of earlier compressive strength more than pozz. It is wefl known tiiat alkali is needed to activate BFS, Lime, Ca(OH)2, produced during the initial hydration of C3S of Portland cement can and will activate the BFS which in turn wifl be more prone to enhance or accderate the kinetics of BFS hydration. Also, sodium metasilicate used in the blends of this invention can hydroryze to NaOH(strong base) and siϋdc add(weak acid
)
Sodium hydroxide can also activate the BFS and a faster hydration can occur. Silicic add (tiiough very small in amount) is a pozzolanic material and cannot be a detrimental addrtive.
The uniqueness of die blend is not vanished even if one omits BFS from the composition. Uniqueness does not lie on mere placement at low temperature and then subjecting it to a high temperature. The uniqueness is in placing a low density slurry at low temperature requiring a certain amount of CS devdopment within a time frame which is later subjected to high temperature. If any similar inddents are found in the literature, die slurry is extended (lowered the density) invariably whh a material that is crushed under pressure (perihe, spheres, etc.) whereby the downhole resultant density of die slurry, subjected to die bottom-hole temperature, is much higher than die reported density. This higher density hdps in getting higher CS in the same amount of time compared to a lower density system.