Expanding for Primary Cementing

L. G. CARTER MEMBER A/ME HALLIBURTON CO. H. F. WAGGONER DUNCAN, OKLA. C. GEORGE

Abstract lated oompounds, have been studied by Klein and Troxel1.2 The preparation of anhydrous calcium alumino­ The expansion of and the effect of various to produce expanding cement was reported by Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 expansive aids upon oil cementing compositions have them, and in subsequent work Halstead and Moore3 been investigated to determine the amount of expansion were able to determine the struoture of this feasible and to observe the stability of cements dis­ compound. Typical compositions of these expansive ce­ playing increased expansion. ments and methQds of manufacture are revealed in two Linear expansion measurements of 1 X 1 X lO-in. ce­ recently issued patents:' 5 mem specimens have revealed that admixtures - sodium The mechanism by which expansion can occur in sulfate, sodiLlm chloride, and combinations of concrete as a result of the develQpment these - will effectively increase the expansion of cement. of by was described in a papeT by Since sodium sulfate solutions have been known to be Hansen: The physical properties of expansive cement were deleterious to fresh cement slurries after having recently investigated by Monfore7 who states: "If the ex­ set, the addition of sodium sulfate to the cement at the pansion can be controlled SQ that it takes place when the time of mixing was not at first considered as a satis­ concrete has developed some strength but is still extensible, factory and practical means of increasing cement ex­ the concrete may accommodate the expansiDn with a mini­ pansion. Considerable expansion was realized with no mum 'Of cracking. Such a cement might be termed a use­ visible sign of deterioration of the cement in a period fully expansive cement. But if the expansiDn occurs after of nine months. the concre'te ha!\ lost extensibility and has become SD brittle Laboratory tests have shown that bonding of cement that it can nD longer accDmmodate the expansion without between concentric sections of pipe was considerably im­ serious cracking or disintegratiDn, the cement is properly proved when expanding cements were used. Measurements termed unsound." of the thermal expansion of cement have been made and The principal c'Oncern of this investigation has been t'O coefficients of expansion calculated for several tempera­ determine the effect of s''Ome expansive agents upon tures. A moderate decrease in cement expansion was cementing comp'Ositi'Ons ror use in oil and gas , noted with an increase in curing pressure with time. predicaJted on the improved bonding of cement t'O both casing and formati'On. Alth'Ough g'Ood placement tech­ Introduction niques are of the utmost impDrtance in accomplishing Since as early as 1920, numerous investigators have a satisfactory 'Oil well cement job, there are occasions been searching for a means of counteracting the shrink­ when under even the most ideal conditions microflow age of concrete that occurs when it loses moisture under channels may result from pressure differential between dry curing conditions. For some applications such as c'Ompletion and production fluids in the casing, or be­ pre-stressed concrete, improved bonding in confined spaces cause of thermal changes during the setting of cement. and for inducing tensile stress in restraining steel, it has In such instances there is a probability that fluid or been more desirable to develop an expansive cement gas migration in the pipe-fDrmation annulus can be designed to more than compensate for any potential reduced by the use 'Of an expanding cement. shrinkage. Expansive cements have been reviewed by With regard t'O oil well cement slurries, it has been Lafuma' who discussed the factors affecting volume determined that a slight expansion will occur when the changes in concrete and cited references to earlier work cement is cured under moist conditi'Ons. These same dealing with this problem. slurries, if air cured, will exhibit shrinkage due to loss More recently, the composition of expanding cements 'Of m'Oisture. Since cement slurries are not generally and the chemical reactions involved, particularly the role subjected t'O a drying envir'Onment under down..:hole con­ pl'ayed by calcium sulfoaluminate admixtures and re- ditions, they consequently would be expected to ex­ pand. However, the degree 'Of expansi'On will vary s'Ome­ Original manuscript receivoo. in Society of Petroleum Engineers ,)ffice Aug. 4, 1965. Revised manuscript of SPE 1235 rece}ved March what with various classes, brands or batches 'Of cement 25, 1966. Paper was presented at SPE Annual Fan Meetmg held in Denver, Colo., 'Oct. 3-6, 1965. due to differences in chemical compositions, particularly lReferences given at end of paper. tricaJcium aluminate c'Ontent. J>escription of Test I'roccdures asbestos sheeting immediately when removed from the bath and measurements were taken as rapidly as possi­ Linear Expansion ble. The specimens were returned to the baths at test Cement slurries containing various expanding agents and comparative length measurements were were mixed according to API RPIOB', and duplicate repeated at 3, 7, 14, 28 and 60 days. After the 60-day cement test bars were prepared by curing in molds under measurements, the samples were placed in the SOF bath water for 24 hours -at the specified test temperature. The and allowed to cool for 24 hours with final measure­ cement bars, 'having a I X I-in. cross..section and an ments being obtained. effoctive gauge length of 10 in., C()ntained stainless steel pins in each cnd for accuracy in measuring the Volumetric and Linear length of Ithe set specimens. Measurements to the nearest Expansion l"f ensuremenls 0.0001 in. werc accompl ished by means of thc Cenco­ Volumetric expansion measurements were made by Menzel Comparator (Fig. I). Specimens were submcrgcd mixing 1.5 gal of cement slurry on a large Waring in a constant -tcmperaturc water bath during the time Blendor for one minute and then pouring I gal of the interval between measurements. slurry into a plastic bag, A strain gauge element* was then Per ceDt linear expansion was calculated from the placed in the center of the cement sl urry and a thermo­ average length change of the bars at various periods of couple was also inserted to measure 'temperature. The time based on initial reference measurements at 24 hoUrs plastic bag was ,then immersed in a I \OF water bath after slUr ry preparation. This procedure is simi lar to with expansion and temperature measurements bcing taken throughout the entire curing period.*+ designation C 15 1-6O in ASTM Standards on Cement, Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 except modifications were made to accommodate oil well Effeet of Pre6sure on Expansion cement slUrries, and specimens were cured at atmospheric In this series of -tests, duplicate cement bars were used pressure, 10 evaluate the effoot of pressure on cement expansion. Cements of API Classes A, B-MSR and C-HSR: as The selected for this study were 0, 500, 1,000. well as pozzolan cement, were used for these comparative 2,000 and 4,000 'Ib /sq in. at a temperature of 100F. The tests with various additives. Also, a commercially manu­ cement slurries were first cured in molds for 24 hours factured expanding cement wa.-. used in 1his test series under water in the test chamber al the desired temperature which was conducted at atmospheric pressure. and pressure. The bars were then removed from the Thermal Expansion molds and returned to the tCSt chamber at the same Thennal expansion data were obtained by first curing curing conditions with the initial reference measure­ the samples at 80F for 24 hours. The bars were then ment being- recorded, Subsequent measurements were removed from the bath and initial length measurements taken on the cement bars for a period of 14 days. were recorded and tL'>ed as the basis of further com­ The test equipment consisted of an autoclave with an parative measurements. Duplicate specimens were then insulated heater jacket on the oUlside and a thermostat placed in the 80, I DO, 140 and 180F water baths for to control the temperature. Attached to the lid of the 24 hours and remeasured. Samples were wrapped in autoclave was a form to hold 1he duplicate cement bars in place, with two calibrated screw contact pins mounted through the lid. These pins are rotated downward to contact the cement bar and the length of the bar is observed from the external scale, which is calibrated to read 0.025 in.l revolution. Expansh'c Effect of Various Addiliv"", To study the effect of sodium sulfate upon cement expansion, various weight percentages of this chemical were added '10 API Class A, Band C cements and 50:50 pozzolan Class A cement for linear expansion tests at SOF. Particular note should be made of the different tricalcium aluminate (C.A) contents of the API cements (Table I). Thus far, measurements have been for nine months and the specimens have been retained for longer term evaluation. Per cent linear expansion of the speci mens, as com­ pared with the initial 24-hour measurement, was deter­ mined for 7, 14, 28, 90, 180 and 270 days (Fig. 2, Table 2). These data not only show the degrce of ·expansion allainable, but also indicate the upper limit in expansion that may be achieved with sodium sulfate in Class A cement without deterioration due to excessive expansion under non-pressu re conditions, The graph indicates that with 15 per cent sodium sulfate the rate of expansion increases rapidly after 90 days, resulting in disintegration and failure of the specimens. However, with \0 per cent or less sodium sulfate ,the rate of cement eX(ydnsion de­ creases and the cement remains competent, The patented expanding cement which contains a

· SH· ~ V~lore Con ~r ete F.rnhedment Type ES-9·S. Fig. l-Cenco-Menzel Comparalor used 10 measure length • 0Tlte instntm"nt USwaters encountered in some under­ and 50:50 pozz'Olan cement c'Ontaining 2 per cent ben­ gr'Ound f'Ormati'Ons can have a deleteri'Ous effect upon t'Onite. H'Owever, the additi'On 'Of 15 per cent sodium certain cementing c'Ompositions that have been mixed with sulfate t'O pozzolan cement has not resulted in deteri'Orati'On fresh water and allowed t'O set. H'Owever, when sodium sulfate in less than critical am'Ounts is added t'O the dry cement at the time 'Of mixing, it does not appear de­ TABLE 1-CHEMICAL ANALYSES trimental since expansion occurs at a c'Omparatively early Analysis Per Cent Clinker Composition Per Cent age bef'Ore the set cement bec'Omes extremely hard and Commercial Expanding Cement inextenslible. After cement has rigidly set, expansi'On pro­ Loss on ignition 2.9 Tricalcium 41.0 19.9 Dicalcium Silicate 28.0 duced by infiltration of sodium sulfate 'Or an excess of Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 Calcium Oxide 61.5 Tetracalcium Aluminoferrite 12.0 additive sulfate cannot be accommodaJted. Hence, cracks Magnesium Oxide 1.6 Tricalcium Aluminate 10.0 Iron Oxide 3.9 Calcium Sulfate 7.0 devel'OP and deteri'Oration results. Aluminum Oxide 6.3 Magnesium Oxide 2.0 Sulfur Trioxide 3.8 Total 100.0 As an expanding aid f'Or cement, sodium sulfate can Potassium Oxide 0.5 be used al'One 'Or in c'Ombination with 100.4 Total t'O obtain different degrees 'Of expansi'On. T'O 'Observe the API Class A Cement effect 'Of sulfate water upon API Class A cement loss on ignition 1.1 Tricalcium Silicate 64.9 Silicon Dioxide 21.4 Dicalcium Silicate 13.2 (C,A = 8.1 per cern) with and with'Out sufficient salt t'O Calcium Oxide 66.0 Tetracalcium Aluminoferrite 8.6 Magnesium Oxide 2.2 Tricalcium Aluminate 8.1 saturate the mixing water, duplicaJte bars were prepared Iron Oxide 2.8 Calcium Sulfate 3.3 and one specimen of each c'Ompositi'On was placed in fresh Aluminum Oxide 4.8 Magnesium Oxide 2.2 Sulfur Trioxide 1.9 Total 100.3 water and the other in 15 per cent sodium sulfate solu­ Total 100.2 ti'On and cured f'Or nine m'Onths. The latter is an accelerat­ API Closs B Cement ed test since underground f'OrmaJtion waters seldom Loss on ignition 3.3 Tricalcium Silicate 61.0 c'Ontain more than 0.5 per cent sodium sulfate. The Silicon Dioxide 21.0 Dicalcium Silicate 16.3 Calcium Oxide 63.0 Tetracalcium Aluminoferrite 12.6 results 'Of these comparative expansive measurements in­ Magnesium Oxide 2.7 Tricalcium Aluminate 3.9 Iron Oxide 4.0 Calcium Sulfate 3.2 v'Olving identical· specimens cured under fresh water vs Aluminum Oxide 4.0 Magnesium Oxide 2.8 Sulfur Trioxide 1.8 Total 99.8 Total 99.8

Special API Class C Cement (high sulfate resistant type) loss on ignition 1.1 Tricalcium Silicate 76.4 Silicon Dioxide 21.5 Dicalcium Silicate 4.8 Calcium Oxide 67.1 Tetracalcium Aluminoferrite 11.1 Magnesium Oxide 0.6 Dicalcium Fluoride 3.9 Iron Oxide 5.9 Tricalcium Aluminate 0.0 Aluminum Oxide 2.3 Calcium Sulfate 4.1 Sulfur Trioxide 2.4 Magnesium Oxide 0.6 Total 100.9 Total 100.9

TABLE 2-EXPANSIVE EFFECT OF SODIUM SULFATE UPON VARIOUS CLASSES OF CEMENT Curing Conditions

Fresh Water Temperature-80F Atmospheric Pressure Linear Expansion: Per Cent (curing time in days) Sodium Fig. 2-Effect of sodium sulfate (cement-API Class A Sulfate (C,A=8.1 per cent); water-5.2 gal/sack; curing pressure (per cent) 7 14 28 90 180 270 -atmospheric curing temperature-80F). Commercial Expanding Cement (C;JA = 10.0 per cent); Water-Cement Ratio: 5.2 gal/sack 0.0 0.049 0.061 0.071 0.102 0.122 0.131 API Closs A Cement (CiA = 8.1 per cent); Water-Cement Ratio: 5.2 gal/sack 0.0 0.027 0.035 0.045 0.067 0.078 0.083 2.5 0.028 0.039 0.053 0.089 0.112 0.120 5.0 0.063 0.081 0.106 0.145 0.174 0.189 10.0 0.097 0.131 0.169 0.244 0.272 0.292 15.0 0.107 0.149 0.210 0.373 1.24W· F** API Closs B Cement (CaA = 3.9 per cent); Water-Cement Ratio: 5.2 gal/sack 0.0 0.008 0.012 0.019 0.030 0.035 0.039 2.5 0.014 0.022 0.029 0.053 0.070 0.078 5.0 0.034 0.046 0.058 0.089 0.115 0.129 Special Closs C <:ement (C,A = 0 per cent); Water-Cement RatiO: 7.6 gal/sack 0.0 0.014 0.018 0.023 0.029 0.031 0.033 2.5 0.015 0.019 0.024 0.031 0.032 0.035 5.0 0.023 0.028 0.035 0.044 0.047 0.049 Fig. 3-Effect of soditim sulfate (cement-50:50 pozzolan *W-Specimen extremely warped and cracked. **F-Bar disintegrated. cement; -2 per cent; water-5.75 gal/sack; cur­ ing pressure-atmospheric; curing temperature-80F). a 15 per cent sodium sulfate solution are presented in TABLE 3-EFFECT OF CURING CEMENT IN 15 PER CENT SODIUM SULFATE SOLUTION Table 3. API Class A Cement (CaA = 8.1 per cent) When cured under 15 per cent sodium sulfate solution, Water·Cement Ratio, 5.2 gal/sack Class A cement with and without sufficient salt to Sodium Linear Expansion: Per Cent (curing time in days) satunllte the mixing water showed appreciably more Sulfate (per cent) 7 28 90 180 270 expansion than when correspondingly cured under fresh Curing Conditions water, and specimens revealed the characteristic cracking Fresh water Temperature, SOP: Atmospheric pressure due to sulfate attack. For comparison with these ex­ o 0.031 0.041 0.055 0.067 0.081 0.089 pansive values resulting from curing in sulfate solution, O' 0.080 0.100 0.124 0.155 0.179 0.196 it is to be noted that this Class A cement directly con­ Curing Conditions taining 10 per cent sodium sulfate gave a linear ex­ 15 per cent sodium sulfate solution Temperature, 80F Atmospheric pressure pansion value of 0.292 in 270 days, but showed no o 0.021 0.030 0.045 0.091 0.153 0.180X·· evidence of deterioration. O' 0.073 0.099 0.126 0.194 0.257 0.290X·· Use of sodium chloride has many applications in * Saturated salt water. oil well cementing and many oil producing companies **X-8ar cracking along edges (characteristic of sulfate attock). increasingly have been using saIt compositions for pri­ mary cementing. While generally improving the per­ formance of cementing slurries as described by Slagle deterioration after nine months was attained with poz­ Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 and Smith,'" saIt in varying concentrations also increases zolan cement containing 10 per cent sodium sulfate the expansion of set cement. Laboratory bonding tests," and mixed with saturated salt water. All compositions as well as acoustic attenuation data (bond logging), in­ appear to be approaching a limiting value of Sltabilized dicate excellent bonding where salt cement slurries are expansion. used, and field experience has shown this economical addi­ Since the preceding tests have been conducted at at­ tive most helpful in improving the success of primary mospheric pressure and a temperature of 80F, the ex­ and squeeze cementing jobs. pansive values obtained should he regarded as inherently A comparison of the expansive effect of various con­ showing a relative comparison of the various cementing centrations of salt, and combinations of saIt and sodium slurries and additives evaluMed. Under actual well con­ sulfate in pozzolan cement with 2 per cent bentonite and ditions the expansive values may change but the n.~lative in API Class A cement is shown in Figs. 4 and 5, order of these expansive cementing compositions would respectively. Increasing concentrations of salt in the not be expected to do so. Cementing compositions which pozzolan slurries (Fig. 4) produced increased expansion showed excessive expansion and deterioration in this of the set cement after nine months. A combination evaluation may not give the same result when confined of 2.5 and 5.0 per cent sodium sulfate with saturated salt under pressure in the annulus between pipe and forma­ water for Class A cement (Fig. 5) resulted in still higher tion in a well. A limited number of tests have been expansion. Highest expansion with no indication of performed confining these cementing compositions between concentric sections of casing, and no deterioration was observed.

Effect of Pressure The effeot of pressure on expansion of cement was investigated using 50:50 pozzolan cement containing 2 per cent bentonite, and the results indicate that an in­ crease in pressure produces a moderate decrease in ex­ pansion (Fig. 6). The specimens were cured at a temperature of 100F and pressures of 0, 500, 1,000, 2,000 and 4,000 psi, with measurements made at in­ tervals up to 14 days. Even though the tests indicated a slight decrease in expansion with an increase in pressure, it is well to remember that the specimens still exhibited expansion Fig. 4-Effect of sodium chloride (cement-50:50 poz­ zolan cement; hentonite-2 per cent; water-5.75 gal/sack; under pressure. Also apparent (Fig. 6) is the rapid growth curing pressure-atmospheric; curing temperature-80F). that occured in the first three or four days.

Fig. 6-Effect of pressure upon expansion (cement-50: Fig. 5-Effect of additives upon expansion (cement-API 50 pozzolan cement; bentonite-2 per cent; water-5.75 Class A; water-5.2 gal/sack; curing pressure-atmos­ gall sack; curing 'pressure-various; curing pheric; curing temperature-80F). temperature-IOOF). Effect of Temperature al'ld reference measurements varied from 8 to 20 hours depend­ Reference Measurement Time ing on composition and temperature. Upon Cement Expansion Expansion data on commercial expanding cement, 50:50 To determine the effect of high curing temperature pozzolan cement containing 2 per cent bentonite and upon cement expansion, bar specimens of API Class G API Class A Cement (the latter two cementing com­ cement with and without 18 per cent salt water, and con­ positions with and without saturated salt water) were taining 40 per cent silica flour to prevent strength retro­ obtained at 70, 80 and llOF. Reference measurements gression, were cured at 80 and 400F (l,000 psi curing were made as soon as the set 1 X 1 X 10-in. bar speci­ pressure) with initial and final reference measurements mens could be removed from the molds and after the being made after curing for 1 and 10 days, respectively. specimens had been cured for one and three days. Linear Linear expansion of these composItions in 10 days was expansion values derived from the earliest reference meas­ much greater at 400F than at 80F (Table 5) due, at least urement as well as those based on a 24-hour initial refer­ in part, to the greater reactivity of the sHica flour at the ence measurement are given in Table 4. higher temperature. Expansion of Class G cement with 40 per cent silica flour was also appreciably increased by These values show that expansion was taking place 18 per cent salt water at both 80 and 400F. from the time of the earliest measurement and that, gen­ erally, pozzolan cement displayed increasing expansion Expansive Effect of Gas ~ith increasing temperature, whereas commercial expand­ Producing Additives Ing cement and Class A cement predominantly showed The gas forming agents , magnesium, iron and alu­ minum powders have been used for years to produce decreasing expansion with increasing temperature for the Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 time and temperature ranges shown. These trends were cement expansion. The technique of adding small quan­ influenced to some extent 1>y the earlier reference time tities of finely powdered aluminum that reacts with the of the more rapidly hydrating cements. Time of initial alkali in the cement slurry to form tiny bubbles of hydro­ gen gas that are not entirely eliminated by pressure is an TABLE 4-EFFECT OF TEMPERATURE AND REFERENCE MEASUREMENT TIME economical means of achieving an expanding cement for UPON CEMENT EXPANSION Curing Conditions shallow well applications. Results of expansive measure­ ments of various percentages of aluminum in API Class Fresh Water Atmospheric Pressure A cement are given in Table 6. Initial linear Expansion-Per Cent Reference Time Curing Time The reaction is controlled by the fineness and amount ~)- 1 Day of aluminum, the fineness and chemical composition of Commercial Expanding Cement (water-~ement ratio: 5.2 gal/sack) the cement, temperature, pressure, cement mixture, etc.

Curing Temperature, 70F 12 0.066 0.088 24 0.022 TABLE 5-EFFECT OF HIGH CURING TEMPERATURE UPON CEMENT EXPANSION

Curing Temperature, 80F Curing Conditions 12 0.035 0.061 Fresh Water 80F: Atmospheric Pressure 24 0.026 4ooF, 1,000 psi Pressure Curing Curing Time Linear Expansion Curing Temperature, 110F Temperature (days) (per cent) 8 0.014 0.028 24 0.014 API Class G Cement, 40 per cent Silica Flour 50:50 Pozzolan Cement, 2 per cent Bentonite (water-cement ratio: 6.6 gal/sack) (water-solids radio: 5.75 gal/sack) 80F 10 0.026 400F 10' 0.246 Curing Temperature, 70F 16 0.016 0.033 API Class G Cement, 40 per cent Silica Flour, 18 per cent Salt Water 24 0.017 (water-cement ratio: 6.6 gal/sack) 80F 10 0.043 Curing Temperature, 80F 400F 10' 0.342 12 0.028 0.048 *Specimens were cured for one day at 80F before initial reference measure­ 24 0.020 ments, and then eight days at 400f, followed by curing for one day at 80F prior to 10-day expansion measurements. Curing Temperature, 110F 2! 0.033 0.053 50:50 Pozzolan Cement, 2 per cent Bentonit;'-Saturated Salt Wate~·019 (water·solids radio, 5.75 gal/sack) TABLE 6-EXPANSIVE EFFECT OF POWDERED ALUMINUM IN CEMENT

Curing Temperature, 80F API Class A Cement-5.2 Gal Water/Sack 20 0.012 0.050 24 0.038 PRESSURE· TEMPERATURE THICKENING TIME TESTS Depth Aluminum Thickening Time Curing Temperature, 110F i!!L-. (per cent) {hours:minutes} 17 0.030 0.072 6,000 0.00 ],58 24 0.042 1,55 API Class A Cement 6,000 0.50 (water-cement ratio: gal/sack) 5.2 24 HOUR COMPRESSIVE STRENGTH, psi Curing Temperature, 70F (Temperature: 80F) 12 0.036 0.045 Aluminum Curing Pressure 24 0.009 (per cent) o psi 3,000 psi Curing Temperature, 80f 0.00 1,290 1,615 1,625 12 0.019 0.030 0.05 1,080 24 0.011 0.10 1,085 1,480 0.25 1,120 1,540 1,495 Curing Temperature, 1l0F 0.50 500 1.00 425 1,060 8 0.010 0.022 24 0.012 Volume Expansion, Per Cent (80F) API Class A Cement, Saturated Salt Water (water-cement ratio: 5.2 gal/sack) Aluminum Curing Pressure Shear Bonding Strength, psi (p&r cent) o psi 3,000 psi (80F and 3,000 psi) Curing Temperature, 80F 0.00 148 20 0.034 0.082 0.05 11.84 0.712 24 0.048 0.10 17.90 0.917 0.25 24.00 1.64 170 Curing Temperature, 110F 0.50 56.51 2.64 188 12 0.049 0.078 1.00 57.19 5.17 194 24 0.029 In oil well cementing, heat and pressure will probably be volumetric and linear expansi'On produced by the addition the moot important factor,s affecting ultimate results. At of salt to pozzolan cement. The linear expansion in one atmospheric (75 to 80F), the reaction starts day of the salt-saturated pozzolan cement was 1.73 times at the time 'Of mixing and may continue for 'One and one­ greater :than pozzDlan cement without salt. The 46-day half to four hQurs. At temperatures above 90F, the reac­ measurement shDwed it to be 2.28 times greater. Volum­ tion may be completed in 30 minutes. At 40F, the reac­ etric expansion 'Of saIt-saturated pozzolan cement was 1.96 tion may be delayed -for several hours. Approximately and 1.99 times greater than pozzolan cement without salt twice as much aluminum is required at 40F as at 70F to in 1 and 46 days, respectively. produce the same amount 'Of expansion, with quantities Linear expansion data at 11 OF using pozzolan cement generally applicable ranging from 0.005 to 0.02 per cent without saJ,t showed it W'Ould increase in length fairly by weight 'Of cement. Higher concentrations may be used rapidly up t'O 14 days where it tended t'O level 'Off. SaJt­ where greater variations in density are required and ulti­ saturated pozzolan cement, under some test conditions, mate strength is nQt a factor. Cementing compositiQns required approximately 40 days before showing any signs containing practical amDunts of finely powdered aluminum 'Of leveling 'Off. to provide expansion have been determined to have per­ Volumetric expansi'On 'Of pozzolan cement with and meabilities 'Of less than 1 md in 24 hoors. wIthout salt indicates that the most rapid increase in length occurs in the first 21 days. H'Owever, b'Oth cement­ Thermal Expansion ing c'Ompositions continued ,to grow until they exceeded Thermal expansiDn 'Of cement will vary slightly with any the limits of the strain gauges used in making ,the meas­ Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 given cementing slurry because of the change in expan­ urements. sion caused by cement reactions taking place at different temperatures and time intervals. Data presented in Fig. 7 One interesting phen'Omen'On occurred when using strain show the per cent linear expansi'On 'Of API Class A cement gauges embedded in the cement slurry to measure vol­ vs time and temperature. The last meaJSurements at 61 umetric expansion. This was the increase in strain gauge days were 'Obtained after cooling the cement bar specimens readings due t'O growth of the cement slurry caused by back tQ initial curing temperature 'Of 80F. the cement reactiDns taking place prior ,t'O the cement setting. Thermal expansions for one-day measurements varied £rDm 0.014, 0.053 and 0.069 per cent when the tempera­ Application of Cement Expansion ture was increased from 80 tD 100, 140 and 180F, res­ The mDst significant relati'Onship of the cement expan­ pectively, and no d'Oubt were influenced partIy by accel­ sion data is Its application tD the microannulus effect that erated chemical reactiQn expansion created by higher has been experienced during ac'Oustical IDgging. The mi­ temperatures. croannulus is created by the c'OntractiDn 'Of the casing Calculated values of the coeffioient 'Of linear expansi'On prior to the cement acquiring a sufficient tensile bond to 6 6 at 100, 140 and 180F were 7 X 10- , 8.8 X 10- and h'Old the casing. 6 6.9 X 10- , respectively. Various c'Ombinations 'Of factors may cause casing con­ Thermal contractions at the c'Onclusion of these tests tracti'On; h'Owever, this discussion will be limited to 'Only were 0.010, 0.034 and 0.061 per cent for the three elev­ tW'O c'Onditi'Ons. The first applies t'O a decrease in pressure ated temperature conditions, indicating to what degree inside the casing caused by a decrease in density 'Of the chemical reaction influenced 'One-day values. At 180F it completi'On or logging fluid 'as compared with ,that 'Of the appears that expansiDn due ,to chemical reaction ,is modi­ displacing fluid. This represents the case where a 12.5 fied ,and that different hydration products are enc'Ountered, Ib/ gal drilling mud is used to displace :the cement sJurry resUlting over-all in less reaction expansion. dU'Pingthe primary jQb and then later replaced with a 10 lb/ gal completion fluid. The second conditi'On applies Volumetric Expansion When the well is initially closed in under pressure, and VDlumetric and linear expans~on resuIts using pozzolan after the cement has set, the pressure is released. This cement with and withQut salt are given in Fig. 8. Tests decrease in pressure allows casing contraction, resUlting showed vDlumetric expansion to be approximately 1.5 in a lowering 'Of cement-cas,ing bond and possibly creating times greater :than linear expansion in two-day measure­ a microannulus if the b'Ond between the cement and cas­ ments with specimens cured at a temperature of 11 OF ing is broken. A subsequent period of time is then neces­ under atmospheric pressure. This ratiD increased to abDut sary for the cement tD expand ,to seal ,this microannulus.. 2.6 ,times greater in 46 days. Under certain conditiDns the c'Ontracti'On of the casing Also shDwn in Fig. 8 is the excellent increase in both

Fig. 8-Volumetric and linear expansion (cemenl-50:50 Fig. 7-Thermal expansion (cement-API Class A; water pozzolan cement; hentonite-2 per cent; water-5.75 -5.2 gal/sack; curing pressure-atmospheric; curing gal/sack; curing pressure-atmospheric; curing tempera- temperature-80F initially). ture-llOF). may be greater than the expansion of the cement; there­ To calculate the expansion of any given cementing com­ fore it would be impossible for the cement to seal the position at specific time, Eq. 2 was used: microannulus. L = Lp (t,.) (2) Expansion with ,time of pozzolan cement mixed with 100 fresh and saturated salt water vs the contraction of 4lh, 5112 and 7-in. J-55 casing (11.6, 17.0 and 26.0 Ib/ft, res­ where pectively) with a change in internal pressure is shown in L = cement expansion, in. Fig. 9. This illustration allows evaluatiQn of the length of Lp = cement expansion at the specific time, per cent time required by the cementing compositiQn to seal the t, = cement sheath thickness = microannulus. FQr example, to determine how much the DJI-D, . 4lh -in. casing will contract radially when the internal ---=2--' In. pressure is lowered 1,500 psi, follow a line parallel to the base of the graph from 1,500 psi Qn the right side until it DJI = hole diameter, in. intersects the line marked 4lh -in. casing; then Proceed D, = outside casing diameter, in. from this point along a line parallel to' the right side of the graph until it intersects the left side which shQWS the con­ Example 2: HQW many days will it take fresh water pozzolan ce­ traction to be 0.0008 in. To find how many days it will ment to seal the micro annulus in Example I? (Assume a take the cement to expand to seal this microannulus, re­ hole size of 7% in.) turn to the intersection of 0.0008-in. expansion with either = 7.875 - 4.5 = 1 6875 . cement curve and then proceed parallel to the left side of t, 2 . m. Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 the graph down to the base. For the fresh water and satu­ rated salt water pozzolan cement the respective time in­ Pozzolan cement expansion under pressure for three days tervals are five days and one day. equals 0.041 per cent; for seven days, 0.051 per cent; and To calculate the radial contraction or expansion of the for 10 days, 0.057 per cent. casing, Eq. 1 was used: L (3 days) = (0.041) (1.6875) = 0.00069 in. t::.PD' 100 t::.R = (I) 4.7 Et L (7 days) = (0.051) (1.6875) = 0.00086 in. where 100 t::.R = change in casing radius, in. L (10 d ) (0.057) (1.6875) = 0.00096 in. = D = arithmetic mean diameter, in. ays 100 t = casing wall thickness, in. Therefore, it would 'take approximately five days to seal E = modulus of elasticity of steel, psi the microannulus. A similar calculation for salt-saturated t::.P = change in pressure, psi. pozzolan cement indicates it would take only approxi­ mately one day to seal the same microannulus. Example I: How much radial displacement will 4lh -in., 11.6 Ib/ ft, Field Results J-55 casing display if the pressure inside the casing is decreased 1,500 psi? To determine the ultimate value of any material in a laboratory examination it is necessary to evaluate its per­ DR = (4.25)' (-1,500) = -0.0008 . formance when applied in the field. Field application (4.7) (0.25) (30,000,000) m. shows that benefits have been achieved using expanding The negative value indicates contraCtion. additives in cements on primary completion jobs. In one area of the country a comprehensive field inves­ tigation has been made using cements with and without additives to increase expansion. Early jobs in this field indicated that when wells were completed using cements without expansive aids, internal pressure had to be applied to the casing at the time of running a bond log to obtain an evaluatiQn of the cementing job. The reason for this phenomenon was accredited to existence of a microannu­ Ius between the casing and cement caused by a change in pressure inside the casing. Normally, a higher density displacement fluid was used during primary cementing and then exchanged for a low density fluid prior to IQgging for final cQmpletion. Calcu­ lations similar to those previously shown indicated that in some instances it may be necessary ,to apply as much as 2,000 psi pressure to the casing to overcome the microan­ nulus effect during logging. Examination of the acoustic log with the application Qf this pressure proved to more accurately indicate the actual well condition for this situa­ tion. After evaluating laboratory expansion data it was de­ cided that salt-satura,ted pozzolan cement containing 2 per Fig. 9-Radial casing contraction vs cement expansion cent bentonite be used on one well to see if the increase (casing siz-4 %, 5 % and 7 in.; cement thickness- in expansion would eliminate the microannulus effect ob­ 1.6875; cement-50:50 pozzolan cement; bentonite-2 per cent; water-5.75 gal/sack; curing pressure-l,500 served on previous jobs, and result in better bonding. psi; curing temperature-IIOF). Results on numerous jobs have indicated that better bond- 109 was obtained and it was not necessary to run the bond Symposium on the Chemistry of f:ement, London. England log under pressure to evaluate the cementing job. (1952) 581. The introduction of salt-saturated pozzolan cement for 2. Klein, A. and Troxell. C. E.: "Studies 0f Calcium Sulfo­ aluminate Admixtures for Expansive Cements", Proc., ASTl\f alleviating sloughing shales and fresh water-sensitive for­ (1958) 58,986·1008. mations has resul·ted in greatly improved cement-forma­ 3. Halstead, P. E. and Moore, A. E.: "The Compo,iLion and tion bonding.'· Because the increased expansion of this Crystallography of an Anhydrous Calcium Aluminosulfatc cementing composition has produced an improvement in Occurring in Expanding Cement", jour. 0/ Applied Chem. cement-casing bonding, this combination of factors has (Sept., 1962) 12, 413. resulted in a noticeable reduction in remedial cementing 4. Armstrong, T. C. and Whitehurst, B. M.: "Sulfoaluminate operations. Cement", U. S. Patent No. 3,147,129, issued Sept. 1, 1964. 5. Klein, A.: "Calcium Aluminosulfate and Expansive Cements Summary Containing Same", U. S. Patent No. 3,155,526, issued Nov. 3, 1964. Expansion of oil well cementing composItIOns can be 6. Hansen, \V. c.: "Crystal Growth as a Source of Expansion increased by the addition of expansive aids such as sodium in Portland· Cement Concrete", Proc., ASTM (1963) 63, 932. sulfate, sodium chloride, pozzolan or combinations of these 7. Monfore, G. E.: "Properties of Expansive Cement Made with admixtures. Cements showed various degrees of expansion Portland Cement, and Calcium Aluminate Cement". after curing under moist conditions due to differences in jour., Portland Cement Association Research & Development Laboratories (May, 1964) 6, No.2, 21·19. chemical composition. 8. "Recommended Practice for Testing Oil Well Cements amI Some cementing compositions showed an increase in Cement Additives", Div. of Prod. Prac. API, Dallas, Tex. Downloaded from http://onepetro.org/jpt/article-pdf/18/05/551/2224321/spe-1235-pa.pdf by guest on 29 September 2021 expansion with increasing temperature for a temperature 9. "API Specification for Oil Well Cements and Cement Addi· range of 70 to 110F, whereas other compositions (parti­ tives", API Std. lOA., Div. of Prod. Prac., API, Dallas, Tex. cularly those having a more rapid hydmtion rate) gener­ 10. Slagle, K. A. and Smith, D. K.: "Salt Cement for Shale and ally showed a decrease in expansion with increasing tem­ Bentonitic Sands", jour. Pet. Tech. (Feb., 1963) 187. perature. However, expansion of Class G cement with 40 11. Evans, G. W. and Carter, L. G.: "Bonding Studies of Cementing per cent silica flour at a curing temperature of 400F was Compositions to Pipe and Formations", Drill. & Prod. Prac .. many times greater than that obtained for the same com­ API (1962) 72. *** position at a temperature of 80F. Gas-forming additives such as finely powdered alumi­ num have not been widely used for application in oil well cementing and do not appear as favorable as the other expansive additives. It is realized that more information is needed which will permit an accurate correlation between laboratory meas­ urements and actual well conditions. It would be very helpful in designing expanding cementing compositions for various well conditions if the required amount of im­ posed stress on the casing necessary to provide reliable bonding could be correlated with cement volumetric ex­ H. F. WAGGONER (right) is a development chemist con­ pansion data under pressure. cerned primarily with oilwell cementing materials. He re­ ceived his BS degree in chemistry from The U. of Okla­ Acknowledgments homa and his experience in the chemical and petroleum industries includes the fields of fuels, high and The authors wish to thank Halliburton Co. for permis­ petroleum practices. CHARLES GEORGE (left) is a senior sion to publish this paper and those in the laboratory who chemist with Halliburton. He received his BS degree in assisted in its preparation. chemistry from Northwestern State College and his MS from Oklahoma State U. A picture and biographical References sketch of L. G. CARTER were published in the April, 1966 1. Lafuma, H.: "Expansive Cements", Pro c., Third International issue of JOURNAL OF PETROLEUM TECHNOLOGY.