SETTING OF PORTLAND CEMENT Thermal Characteristics during Setting at Elevated Temperatures
Downloaded via UNIV OF SOUTH DAKOTA on June 17, 2018 at 20:42:23 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
F. W. JESSEN AND C. E. WEBBER Humble Oil and Refining Company, Houston, Texas A method has been developed for a study of the thermal effects in the setting of cement under conditions similar to those encountered in the cementing of oil well casing. A 40 per cent slurry of a standard brand of portland cement was used throughout the investigation. High formation temperatures lead to a rapid setting time of ordiThis increased rate of nary portland cement. setting is due to the high formation temperature to which the cement is subjected and also to the resulting higher temperature due to the rapid heat evolution during setting. The latter effect contributes quite an appreciable quantity of heat at elevated formation temperatures. The heat of hydration of a 40 per cent cement slurry has been measured up to temperatures of
150 ° F.
As evidenced by the heat of hydration and the initial setting time, it seems that definite compound formation takes place up to the time of the initial set of the cement. No great temperature effect was noticed during the final set of the cement. The results of this investigation may well account for the temperature difference found between cement and formations after 12 to 36 hours. Results of this investigation show that at formation temperatures above 120° F. the rapid setting and accompanying heat evolution of the cement make it probable that the cement slurry begins to set well before it is pumped into the desired position. Such a condition would undoubtedly account for many
ever-increasing trend toward deeper oil production levels necessitates the placing of cement at temperatures considerably higher than any heretofore encountered. At elevated formation temperatures numerous complications arise in cementing casing, due to the extreme rapidity with which the cement slurry attains its initial set. The purpose of the present investigation was (a) to evolve a method for the study of the thermal effects accompanying the setting of cement slurries at temperatures comparable to those encountered in oil well cementing, (b) to determine the maximum temperature to which a 40 per cent portland cement slurry might rise when placed in formations of relatively high temperature, and (c) to measure the rate of heat evolution from such a slurry at temperatures up to 150° F. Thermal data on the rate of heat evolution was anticipated up to a temperature of 200° F.; however, it became unduly difficult to mix the cement and water to a uniform slurry at 175° F. because of the rapid setting of the cement. Hence, no measurements have been made at temperatures above 150° F. The introduction of special high-temperature cements for oil well cementing has decreased the tendency of rapid initial setting, and it should be possible in the future to determine the rate of heat evolution from these cements at temperatures up to 200° F. The present practice of this laboratory is to determine the thermal properties of cement slurry as well as the physical character of such slurry as an additional guide in selecting oil well cement. When completed, this study of the thermal properties will be the basis for further results to be presented on “special oil well” cements already developed and others now in process of development.
faulty
cement
jobs.
phenomena of the volume decrease (contraction) and heat evolution accompanying the setting of cement at room temperature. A Carlson vane-type calorimeter (2) was used in the experiments to determine the rate of heat evolution; the results thus obtained compared favorably with the customary Dewar flask technique. The heat-of-solution method for evaluating the heat of hydration was employed by Lerch (10) and others (16). In studying the generation of heat in the hardening of cement used in cementing bore holes, Grozdorskaya and Yaichnikova (6) determined that equilibrium between the heat released by the cement and the heat transferred to the surrounding medium is established in 1 to 3 hours. Most of the previous work reported shows that the rate of heat evolution on mixing cement with water attains a maximum value after a time (generally 8 to 15 hours) characteristic of the given cement (14)· There appears to be no simple relation between the composition of the cement or degree of fineness and the maximum rate of heat evolution (7).
THE
Method of the Investigation Oil well conditions were simulated in that thick masses of cement were avoided, the thickness of the cement wall being from 5 to 8 cm. (2 to 3 inches). This thickness corresponds to the annular space generally employed in oil well cementing practice. A standard brand of portland cement was used in all tests. This cement had the following chemical analysis and physical characteristics: ,-Chemical Analysis-. Silica 20.60% 9.86 Iron and aluminum oxides 66.86 Calcium oxide 1.27 Magnesium oxide 1.38 Sulfur trioxide 0.03 Water
Previous Investigations Numerous investigators (1, 3, 5, 8, 11,12, IS) have studied the effect of temperature on the setting time of cement and the tensile and compressive strengths. Hemeon (7) studied the
Loss on ignition Total
745
0.07 100.07
^-Mesh On 100
Passed 100 Passed 200 Passed 325
Fineness-» on
on
200 325
Normal consistency
2.6%
12.0 71.0 14.5 100.0 24%
746
INDUSTRIAL
AND ENGINEERING
The cement met all requirements of the A. S. T. M. specifications for Portland cement. The apparatus (Figure 1A) used for determining
the maximum
temperature to which the cement might rise was extremely simple: A 2-gallon, wellinsulated stone jar served as a thermostatic bath. It was equipped with an electric heater, and a a stirrer, sensitive thermoregulator for controlling the temperature of the water bath. In the bath was placed a tin container, 10 cm. in diameter and 15 cm. deep, which served as a mold for the cement. The water bath surrounding the mold was maintained at a constant temperature, the greatest variation being ±0.2° F. or approximately 0.1° C. All temperature
CHEMISTRY
VOL. 31, NO.
6
as a thermometer well. Water was used as the thermometer well fluid. The temperature of the cement was recorded at regular intervals until the final set had been reached. Six separate experiments were made, one at each of the following temperatures: 80°, 100°, 120°, 140°, 160°, and 180° F. No correction due to evaporation of water from the cement slurry was made, since any loss of water by evaporation could have taken place only while the cement was at a greater temperature than that of the surrounding bath. By the time this occurred, the cement had in every case practically reached its initial set. The results of this investigation at the various temperatures are presented by Figures 2, 3, and 4.
inches) served
In order to study the actual increase in the rate of heat evolution with temperature, the following method was employed :
A 40 per cent portland cement slurry was used in all instances, the amount of cement used being 150 grams. Figure IB illustrates the apparatus. The calorimeter consisted of a 500-cc. Dewar flask. A 250-cc. volume of water was placed in the flask to provide an adequate water bath. The calorimeter was equipped with a copper coil, through which cooling water could be circulated, a heating element, and a motor-driven variablespeed stirrer to maintain proper thermal conditions. A copper container was used; it was 4.5 cm. (1.75 inches) in diameter and 8.25 cm. (3.25 inches) high, and was made of thin sheet copper which allowed maximum conductivity from the inside of the setting cement mass to the bath. This container fit snugly inside the cooling coil so that any heat emanating from the cement mass was immediately dissipated to the bath. Cooling water flowed through the copper coil, the rate being regulated by means of a pinch clamp. The inlet and outlet temperatures of the circulating water were measured to within 0.05° F. with accurately calibrated thermometers. The inlet water was maintained approximately 10° F. below that of the calorimeter by means of an auxiliary heated and thermostatically controlled bath. Heat losses were minimized by placing the Dewar flask in a water bath which was kept at the same temperature as the calorimeter. Under these conditions this type of calorimeter was considered to give results as accurate as those obtainable with the Carlson vane-type calorimeter (2). In later experiments to test the reliability of the data at measurements were 125° F., the cooling coil was eliminated from the calorimeter, read on a calibrated and water was allowed to flow directly into the Dewar flask. thermometer with The Dewar flask was closed with a tight-fitting rubber stopper, of an the center of which was cut so that the copper container for the accuracy ±0.1° F. cement could be slipped into place through the center hole. One thousand After the cement had been lowered into the water in the calorimgrams of cement eter, the center hole was stoppered tightly. Since the cement and four hundred slurry in the mold was completely covered with water through the test period, no water evaporated from the cement. A grams of water mixed were mercury seal was provided on the stirrer to prevent water from thoroughly in the entering the Dewar flask. The whole calorimeter was then mold at room temsubmerged in the outer bath. A sketch of the setup is shown in perature, giving a Figure 1C. This procedure also necessitated a change in the which filled mass method of maintaining the cooling water at a constant temthe mold to a depth perature. A large copper coil was immersed in the outer bath Immeof 9 cm. in which the calorimeter was placed, to assure the inlet water a diately upon being temperature within a few tenths of a degree of the calorimeter mixed, the mold temperature. The inlet and outlet water temperatures were was measured in the same way as before. Such a small temperature placed in the water bath. A 6difference between the inlet water and the calorimeter meant mm. glass test tube that a greater volume of fluid had to be passed through the which extended calorimeter. The results of this test checked the data obtained into the cement to a in the first series of tests and are given in Figure 5. The desirability of having the cement, mixing water, and depth of 5 cm. (2 calorimeter as nearly as possible at a uniform temperature is obvious; hence, prior to making any determinations of the rate of heat of hydration, they were brought to the same temperature. The cement slurry (150 grams of cement and 60 grams Figure 1 of water) was mixed by means of a small spatula A. Apparatus for Determinin the copper mold, and the container was immeto ing Maximum Temperature diately submerged in the water of the caloCement Might Which the rimeter. This procedure eliminated any transfer Rise of the cement slurry and therefore reduced heat losses before placement in the calorimeter. The for Studying Apparatus B. greatest variation in temperature between the in Rate of Heat Increase cement paste after mixing, and the water of the with Temperature Evolution calorimeter was 2° F. The adiabatic heat evolution of the cement was observed by means of the calorimeter that was of CalorimC. Arrangement used in the previous tests, and a heating coil of Test Reliability eter to was inserted in the water bath surrounding the 125° F. Data
INDUSTRIAL AND ENGINEERING
JUNE, 1939
Figure Figure
2.
Temperature Changes in Cement Allowed Set at Various Bath Temperatures
FORMATION
Figure 4. Initial
Effect of Temperature and Final Set of Cement
mass. measure
Figure 4 illustrates the time required for the initial and the final sets determined with the Gilmore needle. The time of the initial set in each case just follows the time required to reach the maximum temperature.
Discussion of Results As Figure 2 shows, the temperature of the cement rises reaches the temperature of the bath.
fairly uniformly until it
Heat
of Cement Evolution Temperatures
at
Elevated
to
A voltmeter and an ammeter were placed in the the amount of energy required to maintain the water in the calorimeter at the same temperature as that of the cement. In this particular test the temperature of the cement and the water at the beginning were both 110° F. The duration of current flow through the heating coil was measured with a Two stop watch. At times the current flow was intermittent. hundred grams of cement and eighty grams of water were used to form the slurry in this instance. Three hundred cubic centimeters of water were used in the calorimeter. The temperature of the calorimeter rose slowly at first, and then increased rapidly until at the end of 180 minutes the temperature was 210° F. The test was discontinued at this point, the cement having reached an initial set far earlier in the course of the determination. The result of this experiment to determine the heat evolution of cement under adiabatic conditions is shown in Figure 6. cement
5.
747
TEMR, «F.
Figure 3. Temperature Attained by Cement at Various Formation Temperatures
line to
CHEMISTRY
on
Figure 6. Heat Evolution of Cement under Adiabatic Conditions
A rapid increase in temperature then occurs with the result that the temperature of the cement passes through a maxi-
and then drops gradually to the temperature of the bath, remaining there for the time required to attain the final set. As the temperature of the bath is raised, the rate of rise of the cement temperature increases, and in each case a maximum is reached which is above that of the bath; this maximum is 30° F. when the formation temperature is 180° F. No appreciable increase in temperature was noticed while the cement reached its final set. The rather rapid decrease in the temperature from that of the maximum, indicated by the curves in Figure 2, does not actually occur in practice. It is considered that the maximum temperature reached, as indicated by these experiments, is approximately the true value; however, it is readily understood that the decrease of the temperature from the maximum value to that of the formation is not as rapid as Figure 2 indicates. That this decrease in temperature is much slower is shown by Figure 7, which plots the temperature decrease with time of a mass of cement comparable to that used in previous tests when placed in a wellinsulated sand bath rather than the water bath. At the end of 18 hours there was still a difference in temperature of 3 0 F.; after 27 hours this difference had reached 1° F. The low heat conductivity of the cement and the formations generally mum
INDUSTRIAL AND ENGINEERING
748
surrounding the cement, which results in the rather long period in which a temperature difference between the cement and formation exists, has been the basis of numerous schemes that have been devised for measuring the temperature in wells in order to locate the top of the cement behind the casing after the casing has been cemented.
TIME
Figure
7.
IN MINUTES
Cooling Curve
of
Cement
in
Sand Bath
The rate at which the cement reaches its maximum temperature increases rapidly as the temperature of the formation becomes higher. A line drawn from the origin to the maximum temperature attained by the cement represents approximately the average net rate of heat transfer up to the time the cement reaches the maximum temperature. This average net value of the heat transfer is so called because the Qement at first absorbs heat from the surrounding bath at a rate above that of the rate of heat evolution due to hydration; later, as the temperature of the cement becomes higher and higher, this rate of hydration is accelerated and an increasing quantity of heat is evolved, owing to the faster hydration of the cement, until the temperature developed in the cement mass is greater than that of the surroundings. Heat then Sows from the cement to the bath. The temperature developed in the cement at a bath temperature of 80° F. except for the rather rapid rise in the first 15 minutes (which may be attributed to the initial wetting or hydration of the cement) is quite moderate. The rate of this temperature rise is also very low. It thus appears that with a temperature somewhat lower than 80° F., the average net heat transfer rate mentioned above would be zero, and no maximum value of temperature would be reached. That is, the rate at which heat is evolved in the cement would be equal to the rate at which heat is dissipated from the cement, with the result that the temperature-time curve would be a straight line. This is very nearly the case at 80° F. However, when the cement is allowed to set at an elevated temperature, the time required for the same amount of hydration is less, which results in heat being stored up within the cement mass and the consequent rapid rise in temperature above that of the bath. The data obtained allows a calculation of the maximum temperature to which the cement will rise when placed in a formation having a known temperature. Such a calculation presupposes conditions generally attained in oil well cementing practice and is, of course, not applicable to large masses of cement more common to general construction work. Figure 3 was constructed by plotting the maximum temperature attained in the cement at the formation temperature against the temperature of the formation. The equation of this curve is given by:
CHEMISTRY
-92 + 2.72T 0.00586 2 Tmax. formation temperature, ° F. =
where T
=
VOL. 31, NO. 6 -
The preceding curves show that the temperature rise of a cement slurry in a well in which the temperature is above 120° F. would be greater than that normally expected, because of the increased rate of hydration (and the accompanying increase in heat evolution) of the cement. The result is that the temperature of the cement is raised by the high formation temperature and also by the heat of hydration. The rate of hydration at elevated temperatures is rapid, and the heat thus liberated is capable of building up a considerable amount of heat within the cement mass. The method of evaluating the rate of heat evolution of cement takes into account the fact that at no time is the actual temperature of the cement mass allowed to exceed the temperature of the water in the calorimeter by more than 0.2° F. A slight error is introduced (particularly during the first 1 of 2 hours of setting at elevated temperature) when the heat evolution from a cement slurry is measured by means of a calorimeter such as the Carlson vane type (2) which assumes a constant specific heat for the cement mass and calculates the heat stored in the sample and the rate of heat storage from the heat capacity per gram of cement. It is to be pointed out, however, that in recent types of the vane calorimeter the results are largely independent of specific heat. It is obvious that the specific heat and the heat conductivity of the cement are continually changing in the course of the setting of the slurry. The heat conductivity of set portland cement is reported at 0.00071 (9). Sheard (IS) measured the thermal constants of setting concrete prior to 3-day age by observing the heat flow between coaxial cylinders; the value of the heat conductivity, K, thus obtained was 0.006. Davey and Fox (4) stated that for setting concrete the heat conductivity equals 0.0045.
Figure 8.
Rate
of
Heat Evolution
of
Cement
Figure 5 shows the heat evolution in calories, per gram of cement at 100°, 125°, and 150° F. The total heat evolution up to the time of the final set is much the same in each case; however, since the setting of cement varies with the temperature, and because the setting is accompanied by a certain evolution of heat, quite a difference exists in the actual amount of heat liberated at various time intervals. Thus an equal amount of heat is liberated in 2 hours at 150° F. and 4.5 hours at 100° F., and it is easily recognized why the cement set is accelerated at elevated temperature. Figure 8 shows that the rate of heat of hydration increases rapidly between 100° and 125° F. Between 125° and 150° F. the rate of increase is not as great as that between 100° and
INDUSTRIAL
JUNE, 1939
AND ENGINEERING
125° F. This characteristic of this portland cement is also noticeable on Figure 2, where there is a more decided increase in the slope of the curves from 100° to 120° F. (each after reaching bath temperature) than there is from 120° to 140°. Since the slope of the curves after they reach the temperature of the bath is dependent mainly on the heat of hydration, it seems that there is a somewhat greater rate increase between 100° and 120° F. than there is between 120° and 140° F. Likewise the slopes of the succeeding curves at higher temperatures increase only slightly. These observations tend to indicate that definite compound formation was taking place at the initial set, and that this compound formation was accompanied by an amount of heat which characterizes the reaction. This is in accord with the observations of Woods, Steinour, and Starke (16) who established the fact that there is a relation between the heat of hydration and chemical composition of the cement. Also, the rate of this reaction, as evidenced by the heat evolution, is rather constant, and is not influenced appreciably by temperatures between 125° and 150° F. On Figure 6 is plotted the heat evolution from cement when allowed to set under adiabatic conditions. It is immediately discernible that the total amount of heat evolved under these conditions is three to four times that evolved at 150° F. in the same length of time. This behavior would be expected, since no heat leaves the cement mass. This fact, in conjunction with the data given in Figure 7, well accounts for the low
CHEMISTRY
heat flow from cements which set at temperatures of 150° to 200° F.—i. e., under oil well conditions.
A cknowledgment The authors wish to thank W. T. Ilfrey for assisting with the experimental work and the Humble Oil and Refining Company for permission to publish this paper.
Literature
FOREST A. HOGLAN
AND EDWARD
BARTOW
and Walker (1) prepared inositol from the water obtained in the manufacture of starch from steep BARTOW corn without the use of acids. This work was con-
tinued for the purpose of obtaining larger quantities of inositol, of improving the process of preparing inositol, of finding other sources of inositol, and of studying the preparation and properties of its derivatives.
Preparation of Inositol Inositol was prepared by the simple method of Bartow and Walker (1). The raw material (obtained from Penick & Ford, Ltd.) was the precipitate made by adding lime to steep water. It had the consistency of thin paste and contained about 20 per cent of solids. A 225-kg. batch of this wet material, equivalent to 45 kg. of dry substance, was heated in a large autoclave for 10 hours under 5.6 kg. per sq. cm. steam pressure. After being heated, the material was neutralized with milk of lime (to phenolphthalein alkalinity), diluted with water to about 680 liters, then agitated, and boiled for 5 minutes by passing in steam. After the sludge settled, the supernatant liquid was removed by decantation. The sludge was diluted with water, heated with steam, and
Cited
Barta, Chimie et industrie, 35, 792-6 (1936). Carlson, Proc. Am, Soc. Testing Materials, 34, Pt. II, 322-8 (1934). Davey, Concrete and Constructional Eng., 26, 359-64 (1931). Davey and Fox, Dept. Sel. Ind. Research (Brit.), Bldg. Research, Tech. Paper 15 (1935). (5) Davis and Traxell, Proc. Am. Soc. Testing Materials, Reprint
(1) (2) (3) (4)
62, 1-19 (1931). (6) Grozdorskaya and Yaichnikova, Navosti Teckhniki Bureniya, 4, No. 2, 5-6 (1936). (7) Hemeon, Ind. Enq. Chem., 27, 694-9 (1935). (8) Killig, Cement, 8, 499-501 (1919). (9) Lees and Charlton, Handbook of Chemistry and Physics, 21st ed., p. 1358, Cleveland, Chemical Rubber Publishing Co., 1936-37. (10) Lerch, Eng. News-Record, 113, 523-4 (1934). (11) Maeda, Bull. Inst. Phys. Chem. Research (Tokyo), 14, 714-19 (1935). (12) Pershke, Chem. Zentr., 1931, I, 2379. (13) Sheard, Proc. Phys. Soc. (London), 48, 498-512 (1936). (14) Swietoslowski and Rosinski, Przemysl Chem., 18, 590-4 (1934). (15) Timms and Whithey, J. Am. Concrete Inst., 5, 159-81 (1934). (16) Woods, Steinour, and Starke, Eng. News-Record, 109,404 (1932).
Inactive Inositol and Its and preparation Derivatives State University of Iowa, Iowa City, Iowa
749
properties
allowed to settle, and the supernatant liquid was again decanted. This process was repeated and the combined liquids from the three decantations were evaporated to a small volume in large open evaporators; they were filtered while hot to remove precipitated impurities. The inositol which crystallized out on cooling was removed by filtration, the mother liquor was further concentrated, and the residual inositol was precipitated by adding glacial acetic acid and alcohol. The average yield for the first nine experiments was 12.00 per cent of the dry matter in the material used. A lower yield of inositol was obtained when an attempt was made to remove the liquid from the sludge by filtration. Only 10.00 per cent of inositol was obtained when the liquid was separated from the sludge by filtration on an Oliver filter. Miore washing on the filter should increase the yield. Effect of pH on Yield feom Pbecipitates. Precipitates were carefully prepared by the Clinton Company by adding milk of lime to steep water until the desired pH values were reached. Five samples precipitated at pH values ranging from 5.5 to 7.5, inclusive, gave a yield of inositol decreasing regularly a,s the pH increased (Table I). About 3 kg. of the precipitate (dry) were used in each of these tests. The amount of precipitate obtained from given quantities of steep water at the lower pH values is small; hence it is less practical to use material precipitated below a pH of 6.5. The precipitate obtained at the higher pH values probably contains such impurities as CaCOs or CaHPO
