Changes Occurring within a Portland Cement Kiln WILLIAMN. LACEYAND HOWARD E. SHIRLEY,California Institute of Technology, Pasadena, Calif.
L
ACEY a n d Woods ( 1 )
latter was again r e t u r n e d t o A method of following changes in the comhave described a method the large can which was then bination of lime has been applied to samples of obtaining samples of r e s e a l e d . These precautions f r o m a n operating dry-process rotary Portland charge a t a number of points were f o u n d necesfiary on accement kiln, under the conditions prevailing along the length of an operatcount of the hygroscopic nature during a 5-hour period. ing rotary Portland cement kiln. of the samples, p a r t i c u l a r l y They showed curves portraying those s h o w i n g high free-lime The well-dried charge lost its combined water the changes in the kiln charge content. rapidly in the Jirst 30 feet of the 102-foot kiln. as it passed through the kiln. The decomposition of CaC03 began to be apANALYTICAL METHODS These changes were followed by preciable 15 feet f r o m the feed end of the kiln means of analyses of the samples T h e d e t e r m i n a t i o n s of and proceeded steadily until completed at about for loss on ignition and for free water, C o n , and loss on ignilime (uncombined CaO). The tion were made by heating a 20 feet f r o m the discharge end. Recombination former determination gave inweighed sample in a boat inof the CaO freed f r o m CaC03 did not begin acs e r t e d i n a s i l i c a tube and dication of the progressive loss of tively until the charge was nearly half way sweeping the evolved gas and volatile matter, while the latter through the kiln and was not completed until water vapor into weighed abshowed the excess of CaO freed within 10 feet of the discharge end. The specijc sorption bottles. The gas used by calcination over that which for s w e e p i n g w a s passed had recombined with the other gravity of the charge increased at a rate correthrough a preliminary purifynonvolatile constituents of the sponding roughly to the evolution of Con. ing train before entering the silica charge at any particular stage of the process. tube. I f i s the object of the present authors to report an extension The purifying train consisted of the following, in the order of this method of treatment, capable of giving a more complete stated: (1) bubble counter, (2) trap, (3) CaClz tower, account of the changes occurring. The results of analyses of a (4) soda lime tower, ( 5 ) desicchlora tube, (6) ascarite tube, similar set of samples obtained from the same kiln that gave (7) desicchlora tower, (8) ascarite tower, (9) large dry rise to the samples studied by the above authors are presented bottle, and (10) U-type weighing tube for checking the by way of illustration. By analyzing separately for water and effectiveness of the purification train. The large dry bottle for COZ as well as for the total loss on ignition and for free served as a reservoir to store a quantity of purified gas which lime, i t becomes possible to follow directly changes in the could be drawn upon for sweeping purposes. Frequent checks upon this system showed that, if the tubes were occasionally state of combination of the lime. shaken to prevent channeling, no corrections were necessary SAMPLING for water or COz entering with the purified gas. The silica tube was '/g inch inside diameter and l ' / ~ inches The kiln sampled was at the Crestmore plant of the Riverside Cement Company near Riverside, Calif. The plant was outside diameter. A zone 9 inches long was heated in a small handling a dry raw mix containing a small percentage of furnace by three Fisher burners, using a natural gas-air returned dust. The fuel was natural gas, and the gases leaving the feed end of the kiln passed STATIONS @TO @ 5HOWk ALONG AXIS silicm through waste-heat boilers. The dimensions of DIMINSION5 ARE GIVEN IN FEET FROM DISCHARGE END OF KILN the kiln and positions of the sampling stations were the same as those given by Lacey and Woods. They are shown again here, for conCONVEYORS venience of reference, in Figure 1. 7 87' Samples of feed, of clinker, and of charge a t 4'7'lQ 141s' 24i24'334 421 54 6d twelve different stations (indicated by numbers along the kiln axis in Figure 1)were taken nearly STACK simultaneously at half-hour intervals. All the CHAMBER individual samples for a given location were composited in a large tin can with t i g h t l y fitting cover. The period of continuous SamFIGURE 1. L O C l T I O N S FOR SAMPLIVG KILNC H A R G E pling extended over 5 hours. After the run the sample-can covers were sealed with paraffin until the samples were removed for grinding. After being mixture. The heated portion of the tube was protected from ground to pass a 60-mesh screen, each sample was well mixed, direct contact with the flame by a closely fitting shield of sheet quartered, and one quarter placed in a small tin sample can nickel. However, in spite of this protection, the silica slowly with a tight cover. Each of these cans, together with the re- devitrified, lasting for 6 months of continuous use. The maining threequarters of the sample, was returned to the entrance end of the silica tube was connected to the purifying larger sample can which was then resealed with p a r a h . After train by a long rubber stopper, one end of this fitting the end a sample for analysis was withdrawn from the small can, the of the tube, the other end fitting into a piece of large rubber 332
March, 1932
INDUSTRIAL AND ENGINEERING CHEMISTRY
tubing which was taped around the outside of the silica tube. This gave a satisfactory gas-tight joint. It was determined that the temperature of the inside surface of the stopper did not go above 40" C. as a result of heat radiated from the interior of the tube.
@lsTANIE FROM D I ~ L H A R G EEND OF K I L N FIGURE 2. RESULTS OF ANALYSES
The exit end of the silica tube was drawn down, and a smaller piece of clear quartz tubing was sealed to it. The end of this tube was ground to fit roughly into a flare on the end of one side arm of the first Fisher absorption bottle. This junction was made gas-tight by using a piece of pure gumrubber tubing over it. The absorption train consisted of four Fisher absorption bottles followed by a trap, a bubble counter, and an aspirator bottle to pull gas through the system. The first two absorption bottles were filled with desicchlora (anhydrous barium perchlorate) to absorb water driven from the sample for weighing. The third and fourth absorbers contained ascarite (asbestos impregnated with NaOH) for absorbing COz, and also some desicchlora to prevent loss of moisture from ascarite. The temperature of the gases coming from the exit end of the silica tube was found to be about 38" C. for the rate of approximately 2 liters per hour used in the determinations. Since this is safely below the decomposition temperature of the hydrate of barium perchlorate (120" C.), no difficulties were experienced from this source. The method used in making a determination was as follows: The Fisher absorbers were carefully wiped with a clean, dry, lintless towel and weighed. They were then inserted in the ignition train. A sample of the material to be analyzed was weighed into an alundum boat which had been previously ignited and weighed, The boat and sample were pushed into place in the heating zone of the silica tube by means of a long wire. The end of the silica tube was closed up again, stopcocks were opened, and the aspirator adjusted so that air was pulled through the train a t the rate of approximately 2 liters per hour. h vacuum of 15 mm. of mercury was necessary to accomplish this. The heating zone of the silica tube was then brought up to 1000" C. during the course of an hour. The temperature was measured with a chromel-aluniel thermocouple. It was found that 4 hours were necessary for the COz to be completely liberated and swept out of the tube. At the end of this period, the aspiration was stopped, and the absorbers were allowed to come to atmospheric pressure, removed from the train, and reweighed to ascertain the amount of water and COzevolved. The boat was allowed to cool in the silica tube, the purification train protecting it from water or COSmeanwhile. The loss in weight of the boat and sample was determined and reported as the loss on ignition. The free-lime determinations were made by the method of ammonium acetate in absolute alcohol, used by Lacey and
333
Woods (1). Specific gravities were determined with a standard Le Chatelier specific-gravity bottle. The MgC03 present in the feed was determined by a method described by Meade (2). The mean of several determinations gave a value of 3.9 per cent M&03 in the feed, corresponding to 1.9 per cent COz obtainable from this source. I n the calculation of CaO liberated from CaC03, this amount of COz has been taken into account as coming from MgC03 before any CaC03has decomposed. Since the amount of MgCO3 present is small compared to the CaC03 and on account of the much greater tendency of MgC03 to decompose at moderately high temperatures, this assumption involves only a small error in the calculations. The free moisture in the feed was determined by holding a sample at 105" C. for 5 days, which showed a loss of only 0.09 per cent. The analysis of the clinker was as follows: 64.3 per cent CaO, 3.8 per cent MgO, 23.1 per cent &Oz, 4.6 per cent AlzOa, 2.2 per cent Fe&. The results of the analyses are given graphically in Figure 2, which shows changes in the charge in relation to its distance from the discharge end of the kiln. Since the water evolved is small, a plot on enlarged scale is shown in Figure 3. Checks to the nearest 0.1 per cent were obtained on all values of loss on ignition, water remaining, and COS remaining which are reported. The check specific-gravity determinations agreed to 0.003.
DISCUSSION OF ANALYSES It was found that the values for loss on ignition and for COZevolved are in close agreement throughout. This would be expected to be true after most of the water in the charge had been driven off, but is surprising for samples at and near the feed end of the kiln. A comparison between the loss on ignition and the sum of the water and COZ evolved indicates that the loss on ignition of the feed sample is distinctly low. A possible explanation of this discrepancy is offered by the absorption by the feed of oxygen from the air used for sweep
DOTANCEr R O M Dl5CtiARGE END OF KILN FIGURE 3. WATEREVOLVED
ing the gases from the silica tube. To test this hypothesis, four determinations were made, replacing the air with nitrogen from a cylinder. The results are given in Table I. The second column shows ignition-loss values determined in previous analyses using air in the tube.
TABLE I. ANALYSES OF SAMPLES QAMPLB
Feed Feed 8 8
In IQNITION air In nitrogen Loss
% 31.7 31.8 9.4 9.4
% 32.2 32.2 9.3 9.2
COa
Ha0
% 31.3 31.2 8.9 8.8
1.1 1.2 0.2 0.2
DIFFERENCB BBT WCOz E E N LOSS HrOA N D
+
% -0.2 -0.2 0.2 0.2
It will be seen that the use of nitrogen results in a higher value of ignition loss and a lower discrepancy in the case of the feed.
334
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 24, No. 3
TABLE11. DATAON Loss ON IGNITION SAMPLE
Loes
ON
4 hours in tube
IONITION"
12 hours in crucible
%
a
%
12 30.7 10 23.1 8 9.4 6 3.2 Percentage of original sample.
31.2 23.5 9.9 4.1
Figure 3 shows that the combined water was driven out of the charge rapidly. As stated above, only 0.1 per cent water was free moisture, removable by drying a t 105" C. The slight increase in water remaining in the samples from stations 7 and 8, although not large epough to be much beyond experi-
IGNITION TME,IN HOURS FIGURE4. RATEOF Loss DURING IGNITION This smaller discrepancy might be explained by the fact that the nitrogen was produced from liquid air and was not purified from oxygen before use. No great difference was found for the sample from station 8 whether air or nitrogen was used. The higher value for loss on ignition of the feed has been used in Figure 2 and in subsequent calculations. A series of determinations of loss on ignition was made by the customary method of heating the sample in a platinum crucible over a blast burner, no change in gas supply being made during a run. The object of this series was a study of the rate of weight loss during ignition. The rates for three samples are shown in Figure 4,the total loss a t the end of 4'/~ hours of ignition being called in each case 100 per cent to give
DI5TANCE FROM DISCHARCXEND OF KILN FIGURE6. RELATION OF INCREASE IN SPECIFIC GRAVITY AND COz EVOLUTION
mental accuracy, might be attributed to the great tendency of samples from this section of the kiln to absorb moisture. This tendency is illustrated in Table 111. The samples cited therein were stored in tightly covered tin cans but were not sealed with paraffin, In a similar set of samples stored for a year, stations adjacent to No. 7 showed water-remaining values of about 9 per cent, while the feed gave only 1.5 per cent. MOISTURE TABLE111.' TENDENCY OF SAMPLES TO ABSORB SAMPLE
Feed 10 9 8
7 6
4 2 Clinker
DISCHARGLENDOF KILN OF CaO IN FIGURE5. STATEOF COMBINATION KILNCHARGE AT VARIOUS POINTSIN THE KILN
-Loss Initial % 31.8 23.1 15.4 9.4 6.4 3.2 0.3 0.3 0.1
O N IONITION4 months later Difference
% 31.7 23.4 17.2 12.0 10.3 8.2 1.7 0.4 0.6
% -0.1 0.3 1.8 2.6 3.9 5.0 1.4 0.1 0.5
FREELIME CONTENT
% 0.9 11.7 19.6 26.1 26.3 28.2 4.2 0.3 0.1
DI5TANCE F ROn
an easier basis of comparison. It will be seen that the samples vary in the rapidity with which volatile matter is given off. Two of the samples shown had evidently not reached constant weight a t the end of 41/zhours. Twelve hours of continuous igniting of four samples resulted in greater losses than those found in 4 hours in the furnace tube as shown in Table 11. The fairly constant increase in loss during the last 8 hours, irrespective of the variation of magnitude of the total loss from one sample to another, might indicate a gradual volatilization of alkalies. This gradual increase in loss on ignition makes it very difficult to know when to stop the process when highly accurate results are desired.
DISCUSSIOK OF RESULTS Before considering the changes occurring in the charge as it progresses through the kiln, it becomes necessary to bring the analytical results to some common basis for comparison. It has been assumed that the nonvolatile material (i. e., 100 per cent minus the percentage of loss on ignition) of the feed passes through the kiln without loss or gain in weight. This involves the assumption that the material volatilized by igniting the feed in a crucible is the same in amount as that volatilized by a partial ignition in the kiln followed by final ignition in the crucible. This constant amount of nonvolatile material then serves as a "tracer" to assist in the determination of the amounts of the samples taken at different stations which correspond to a given-size sample of original feed. Such calculations have been made on a
March, 1932
I ii D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R ’Y
basis of a sample of 100 pounds of feed, and the amounts of certain constituents present a t various stages of the process found. In these calculations the loss on ignition of the feed has been taken as 32.2 per cent, which is the value obtained by ignition in an atmosphere of nitrogen, as described above. The initial free lime plus all the CaO liberated from CaC03 during the process is less than the total CaO present, the difference representing CaO which was originally in combination with other acidic constituents than COZ. Figure 5 shows three fields indicating the state of combination of the CaO in the kiln charge a t various points in the kiln. It will be noted that CaC03 persisted in appreciable quantity to within 20 feet of the discharge end of the kiln. Free lime was not reduced to a satisfactory value until the clinker was within 10 feet of the discharge end. Only small amounts of lime were recombined while the charge was in the upper half of the kiln. The most rapid recombination of lime occurred near the 20-foot point where the last of the C02 disappeared from the charge. The last of the lime recombination was a comparatively slow process. The data presented in Figure 6 show that, although no
335
strict correlation exists, the increase in specific gravity of the charge follows in a general way the evolution of COz from the charge. The decided drop in the specific-gravity curve a t stations 3 and 2 might be explained by the presence of /3dicalcium silicate which changed, upon premature removal from the kiln, into the y form which has about 10 per cent greater volume. There is no direct evidence in favor of this explanation, and it is merely suggested. ACKNOWLEDGMENT The authors wish to express their indebtedness to the members of the research department of the Riverside Cement Company of Los Angeles, Calif., who greatly assisted in the taking of samples, and who made the free-lime determinations as well as the analysis of the clinker. LITERATURE CITED (1) Lacey and Woods, IND. EXQ.CHEM..21, 1124 (1929). (2) Meade, R. K., “Portland Cement,” pp. 526-9, Chemical Publishing Co., 1926. RECEIVED October 26. 1931.
Displacement of Crude Oil and Benzene from Silica by Aqueous Solutions F. E. BARTELL AND F. L. MILLER, Chemistry Department, University of Michigan, Ann Arbor, Mich.
I
N AN e a r l i e r p a p e r t h e The object of this investigation has been data showi1lg the effect of differauthors ( 2 ) reported some to obtain data which would give information ent salt solutions upon the recovery of crude oils by displaceresults obtained in a study concerning the function of “.flooding agents” It is the opinion of the of the displacement of different crude oils from finely powdered used in the displacement of oil f r o m oil-bearing above-mentioned workers that sands. The results obtained indicate that neither only the salts of strong bases and and, compressed silica by water. the surface-tension values of the liquids nor the weak acids are effectiveas floodA determination of the percentage of oil which could be interfacial-tension of the liquid-liquid ing a g e n t s . A c i d i c salts or neutral salts are not suitable. recovered from oil-wetted silica system are the dominant factors. Effective w a s n o t a t t e m p t e d , but the THEORIES RELATING TO FUNCdisplacing agents appear to be those that alter magnitude of the pressures deOF FLooD1sG veloped in the displacement of the aqueous solution-silica interface either through a high degree of adsorption or through chemical A number of theories have the oils from silica was deterreaction at the interface. been advanced to explain the mined. Inthe Present Paper the behavior of such aqueous soluproblem of the displacement of tions, but as yet none of them is oil by a q u e o u s s o l u t i o n s is generally accepted. It has been suggested that the oil is considered. It is well known that water will displace organic liquids and displaced from the sand because of a differential capillary crude petroleum oils from silica. It has been observed, effect; i. e., since the water has a higher surface tension than however, that water itself is comparatively ineffective as a the oil, it was thought that the difference in these capillary displacing agent when used in actual oil-field work. The forces might result in the displacement of the oil. Salts, water, whose displacing ability is originally very effective, which when added to water cause an increase in the surface soon becomes sluggish in its action, and the rate of displace- tension of the water, would, then, produce a greater difference ment of the oil becomes low. This effect may be due in part between the two surface tensions and would thus effect B to the saturation of water with some of the more soluble greater displacement of the oil from the sand. This explaconstituents of the oil or to highly adsorbed constituents of nation is hardly acceptable in view of the fact that many the oil on the silica surface which are not easily removed by different salts will increase the surface tension of water, and the water. Experience has shown that certain aqueous yet only a few of them are efficient as flooding agents. It is solutions are more effective than pure water in removing the also true that the increase in surface tension of the water is oil from the silica. Such solutions are generally known as slight. flooding agents. The salt solution most commonly referred Other investigators have attributed the displacement to in this connection is sodium carbonate, or soda ash. Prac- tendency to a chemical reaction. Nutting (11-14) is one tically, the use of this salt has apparently not met with sue- who has advanced this idea. He suggested that an attraction cess. Uren (15) and Beckstrom and T7an Tuyl (8) have exists betn-een the silica particles (Si02) and the hydrogen used a number of different salt solutions in their researches ions of the water, so that a layer of siloxyl (SiOOH) radicals is and have determined the relative percentages of oil recovered formed. These radicals would be acidic in character and by each. Uren and Fahmy (16) have also recently published would have one free bond by means of which they might unite
