%la), 1937
I N D U S T R I h L .4 N D E N G I N E E €3 I N G
(Figures 6 and 7 ) were constructed for the paraffins from isobutane to nonadecane. Curves for methane, ethane, and propane are also given in Figure 6, but they were not calculated by this method. Wiebe and Brevoort’s values (12) for ethane were used while Wiebe, Hubbard, and Brevoort’s determinations ( I S ) for methane and those of Dana, Jenkins, Burdick, and Timm (6) for propane were extrapolated to the critical pressure by plotting against hexane a t the same reduced pressures.
HEATSO F
~ . i P O R I Z . i T I O N O F OTHER SCBSPASCES
While the derivation outlined above was limited to hydrocarbons, the rule, that the ratios of the molal heats of vaporization (or latent heats) of two substances are constant at the same reduced pressures, is apparently more general. This is illustrated by Figure 8 where the molal heats of several pairs of compounds, selected at random, are plotted in this manner. Although greater accuracy will probably be secured if the compounds are chemically similar, this rule holds very well in the case of a polar liquid (water) plotted against nonpolar hexane.
C H EM I ST R Y
505
ACKNOWLEDGMEST The author wishes to thank W.E. Robinson whose suggestions and criticism were very helpful. LITERATURE C I T E D
(1) Ashworth, J. Inst. Petroleum Tech., 10, 787 (1924). (2) Calingaert and Davis, IXD. ESG. CHEW,17, 1287 (1925). (3) Coats and Brown, Univ. of blich., Eng. Research Pub., C‘irc. Series 2 (1928). (4) Cope, Lewis, and Weber, I F D . EXG.CHEJI.,23, 887 (1931). ( 5 ) Cox, Ibid., 15, 592 (1923). ( 6 ) Dana, Jenkins, Burdick, a n 3 Timm, Refrigereting Eng., 12, 387 (1926). (7) FroUch, ISD.ENG.CHEM.,21, 1117 (1929). (8) Krafft, B e d . Ber., 15, 1687 (1882); 29, 1323 (1896); 32, 1632 (1899). (9) Leslie, J. Inst. Petroleum Tech., 13, 546 (1927). (10) Mills, J . Am. Chem. Soc., 31, 1099 (1909). (11) White, ISD.EKG.CHEM.,22, 230 (1930). (12) Wiehe and Brevoort, J . Am. Chem. Soc., 52, 622 (1930). (13) Wiebe, Hubhard, and Brevoort, Ibid., 52, 611 (1930). (14) Wilson and Bahlke, IND.ENG.CHEM.,15, 592 (1923). (15) Young, S., Sci. Proc. Rou. Dublin Soc., 12, 374 (1910). RECEIVEDLlarch 10, 1932
H e a t Transmission in Convection Sections of Pipe Stills C. C. MONRAD,Standard Oil Company (Ind.), Whiting, Ind. ESPITE the importance There are several equations Data are presented on heat-transfer coefficients of p i p e s t i l l s i n t h e available in the literature for in coni~ecticmbanks of pipe stills. These tests operation of tlie modern c a 1c u l a t i n g heat-transfer coefon several types of furnaces under utere made refinery, no really satisfactory ficients from gases to staggered caried conditions of gas velocity, excess air, and methods have been published for tubes. Of these probably the temperatures. Gas temperatures were taken with best known is t h a t p r o p o s e d the design of this type of equipb y C h a p p e l l a n d 1IcAdams ment from the s t a n d p o i n t of high-ve1ocit.y thermocouples in order to eliminate heat transmission. This is par( I ) , which is based on data for radiation errors found aith static thermocouples. single pipes and wires, as well as ticularly true in the so-called Comparison of these data with calculated heatradiant section, but is true also banks of tubes. This equation transfer coeficien ts based on equations for conto some extent in the convection is as follows: ztection and gas radiation f r o m hot gases io stagor backpass section. No reli0.8 T ,1 ‘ 3 G(0 60 + 0 08 log D ) h , = ___ able data on this general subject D0.63 gered tubes shows good agreement of theory with are available in the literature, practice. The estimated probable error in com[l (1’ and in particular no correlation puting heat-transfer coe@cients for such cases is of theoretical calculation with Keilier (4) has proposed equaabout 10 per cent. experimental r e s u l t s has been tions based o n d i m e n s i o n a l Simplijed formulas for computing heat transpresented. analysis of the form (using film Accordingly, i’t was thought fer and pressure drops f r o m gases flouting past properties) that a n attempt to c o r r e l a t e banks of staggered f i d e s are giz!en. plant data on pipe-still furnaces would be useful and desirable. Since the methods by means of which heat is transferred in convection sections are simpler to study and understand, Reiher evaluated the constants C and n of Equation 2 from this part of the furnace was investigated first. A t the same his data and the earlier data2 of Rietschel (a), but did not obtime a considerable amount of data on the other parts of the tain good agreement. The actual data obtained by both observers fall on a reasonable curve, as s h o m later. Reiher furnace could be taken for future use. It is generally known that in the convection section of a found that the constant C increased with the number of tube pipe still, heat transmission takes place from the gas to the roms up to about five rows, and then remained constant tube wall by two main mechanisms-namely, forced convec- Thus, in a bank of tubes, the effect of the first four would be tion, and gas radiation from carbon dioxide and nater. Heat negligible after about ten tubes. This effect in the first few r o v is in agreement with data obtained by Reiher and is then conducted through the metal wall and thence to the moving oil. In general t l i c x resistance to heat flow is largely others on the type of gas flow past a bank of tubes of this from gas to tube n all, and the surface of the tube is often only kind. The gas does not become thoroughly turbulent for from 5’ to 15” higher than the oil temperature. Over-all some distance down the bank. Reiher’s final equation for a bank of tuiici with the clearlieat-transfer coefficients from 2 to 10 B. t . 11. per sq. ft. per F. per hr. are grneraIly found. and this 1s e-entiaIly equal to the ance equal to tlie diameter or to one-half the diameter may l ~ e coefficient on the gas side, which is the sunmation of heat expressed in dimensionally homogeneous unit? as follows: transfer by convection, gas radiation, and reradiation from 1 uaI taken a t minimum free area of tube banh hot furnace walls. 2 ‘ I he d a t a of Rietscliel uere not available and a e r e read from eqitatinns
+ ];
I
I
1
INDUSTRIAL AND ENGINEERING
506
h,
d kf
=
0.15
du,,. ( 7) pf
0.69
(3)
for mure than ten rows. The constant varied from 0.10 to
0.15 as the tube rows increased from 2 to 10, but substantiallJ7 no effect of tube spacing was noted.
CHEMISTRY
Vol. 24, No. 5
Figure 1 shows the data from different sources plotted as h, versus G2/aT0‘3/D1/3.I n addition, data obtained by the author on pipe-still convective sections are plotted in a similar manner. These data are based on the actual heat transmission observed after deducting calculated gas radiation and wall correction. as shown later. The best line through the experikental data appears to be somewhaclower than previous data obtained on smaller and smoother tubes. The equation for this line is
An inspection of Figure 1 shows that present data do not justify the use of a constant varying with the number of rows unless the number is less than four. Reiher’s own data are somewhat scattered, although there appears to be a definitely lower coefficient in the first few rows. 20 However, since most industrial installations contain a large number of rows, this point is not rV very important. Methods of calculating heat transmission by IO gas radiation from flue gases to cold surfaces 9t I , I I ! l I I ! l I have been developed largely by Schack (6) in Germany and Hottel ( 2 ) in this country. This radiation is due to the carbon dioxide and water in the gases and occurs in definite bands of the spectrum. Simplified charts for determining the heat transferred in this manner have been published by Hottel. These charts show the heat radiated as a function of the partial pressure ( P ) , temperature (T), and effective thickness ( L ) of the gas radiating. Suitable corrections for superimposed radiation of carbon dioxide and water may be made. Since in general the tube diameter is equal FIGURE2 . HEATTRANSFERBY CONVECTION TO STAGGERED PIPES to the clearance, L may be taken as 2.8 times the clearance. No correction for superimposed The equation proposed by Chappell and hh4darns was radiation of carbon dioxide and water is necessary because found to give results somewhat in disagreement with both this is very small, but a correction must be made a t the the Reiher and Rietschel data. Moreover, it is apparent that top of the tube bank owing to increased thickness of gas this equation is not dimensionally sound since, if G is taken layer. Such a correction may be made approximately by to the nth power, D must be to the (1 - n)th power. The assuming this radiation to be essentially that to a plane effect of tube spacing indicated by this equation does not equal to the area of the top of the tube bank a t PwL = POL= 1 appear justified in view of the results obtained by Reiher between the temperature of the gas above the bank and the with two spacings, and by Rietschel who used a bank of tubes temperature of the tube. The assumptioh of P L = 1 is justified, since the radiation does not change markedly a t with a very small clearance. It was thought desirable to attempt to correlate present high values of PL. However, since the radiation has already available data by means of a dimensional equation similar been calculated for PwL and P,L based on 2.8 times the to that proposed by Reiher, simplifying the results for engi- clearance, the added gas radiation is equivalent to that a t the neering purposes as much as possible. Since the thermal top gas temperature between PwL = P O L= 1 and P,L and conductivity and viscosity of air and flue gas are nearly the POLbased on the clearance. The radiation determined to same and may be expressed by the functions k = aTb and the plane area of the top of the bank can then be prorated over all the tubes in the bank. The coefficient of heat transfer p = d T 6 ’ (where T i s the absolute gas temperature and a, a l , b, b1 are constants), it is possible to simplify the Reiher equa- determined in this way has been called h’ra in this paper tion considerably. Moreover, since the exact power of the t o differentiate it from the average gas-radiation coefficient Reynolds criterion is rather indeterminate at present, it hrg . appeared reasonable to simplify this. In the convection section of a pipe still, the tube bank is The final equation correlating all the available data in the generally four to twelve tubes wide, and the area of the walls literature on staggered tubes is as follows (using gas stream surrounding the bank is a fairly large fraction of the tube temperatures) : areas. These walls pick up heat from the gas by convection 1.75G2I3TO.3 and gas radiation and reradiate it to the tubes by black-body h, = (4) D1/3 radiation according to Stefan’s law. Although this heat is where h, = heat-transfer coefficient (B. t. u. ft./sq, ft./’ F./hr.) not a large amount, it is not negligible and a correction should G = mass velocity in minimum cross section (lb./ser./ be made allowing for it. Although a theoretically correct sq. ft.) equation can be deduced for this case, it is very cumbersome T = gas temperature (’ F. absolute) and a much simpler method may be used. D = tube diameter (inches) 111
I
I
I
INDUSTRIAL AND ENGINEERING CHEMISTRY
May, 1932
507
TABLEI. DATAON PLANT TESTSOF Two FURNACES
--
F
--+AS-
'rEST
A1 A2 A3 A4 A5
Out 'F . 480 454 532 487
G'
In LD./sec./sq It F 0.095 1231 0.089 1305 0.129 1310 0.077 1320 0.138 1205
566
TEMPERATURES -OIL-Log In Out MTD 'F . F . F. 368 300 580 288 385 561 613 312 415 321 562 391 321 626 436
Av. oil
O F .
440 425 463 442 474
CARBON
Av.
DIOXIDE
gas
In
Out
F 606 810 878 833 912
%
%
..
.
.. .. ..
P,L
PvL
hc
hr"
h'rg
WALL EFFEFTheal.
hot,.
%
9.5 9.5 8.5 12.0 4.9
0.08 0.08 0.07 0.09 0.05
0.14 0.14 0.13 0.15 0.06
1.8 1.8 2.3 1.7 2.4
1.5 1.5 1.5 1.5 1.4
0.1 0.2 0.1 0.2 0.1
7 7
3.7 3.7
4.0 3.6
7 7
34 . 52 4.2
34 . 53 4.1
9.8 6.2 5.7 6.3 5.5
0.09 0.08 0.06 0.06 0.06
0.11 0.10 0.08 0.08 0.08
2.2 2.1 3.6 3.9 3.4
0.7 0.7 1.1 1.0 1.0
,
.. ... ... .. .. .,
16 12 12 12 14
3.3 3.1 5.3 5.5 5.1
6.3 3.7 5.6 5.3 4.9
ERROR % -8
-3
+; -2 EFF. o r
Bla 0.165 637 B2 0.121 1080 83 0.320 1030 B4 0.300 1070 1060 86 0.287 'a Ratio of outside area of
557 312 462 200 387 567 .. 640 161 265 635 213 848 . 810 292 433 557 363 920 . 835 298 437 565 368 953 . 948 , 308 835 436 576 372 total surface t o t h a t of plain tube = 3.8
If one disregards the difTerences in heat-transfer coefficients and the tubes as well as such factors as reabsorption nf heat, the following development may be made: to the wall
[h,
+ h,,][t, -t,]A, heat to nalls = heat to tubes from walls =
nalls
=
hrb[tw- & ] A ,
== heat
from
:. % ' increase in heat absorption by tubes above that received
directly -
h,b[tur-~t]Aw
[hc
x
100 -
+ h7,1[t,-ttlA,
h,t, [ t u - ILIA, X 100 At1 [hc + h w l [ t , - t m l [ho h r , l [ L - 411
+
+
Since h, and h,, must be calculated in order to get the direct heat transfer from gases to tubes, it is necessary merely to calculate hrb, which in turn may be obtained simply from the Stefan-Boltzman equation and is equal to hrh = 0.00688 p
[A]'
p is generally 0.95 and may be assumed equal to unity.
T
may be taken equal to tube temperature unless a more exact correction is desired. The total heat-transfer coefficient on the gas side, therefore, may be given by the following equation: L o .
=
(100
+ % wall correction) [h, + h,,l + h'w 100
(7)
I n all the experimental work presented later in this paper the over-all coefficient of heat transfer was assumed equal t o the coefficient on the gas side, since the effect of the tube resistance and the inside oil film was negligible, as indicated above. For cases such as superheating steam or heating air, this is not permissible, and it is necessary to calculate both coefficients. EXPERIMENTAL DATA
At the tirue this work was undertaken there were available a large number of scattered data on plant tests of various furnaces. These tests were generally incomplete since they were made for a specific purpose, and many data important for this type of work were omitted. I n particular, gas-temperature measurements were made with static thermocouples, involving errorb from 10" to 200" F. Carbon dioxide analyses and fuel consumption were iiot extremely accurate in most of these tests. It was therefore thought advisable to conduct :I series of tests on various plant units and insure that the data were consistent and a h accurate as possible. Since considerable work with high-velocity thermocouples had already been done
GILLS 190 119 106 95 97
under varied conditions by ti. E. Dewey of this laboratory the problem of obtaining accurate gas temperatures was quickly solved. The velocity therniocouples used for this purpose were constructed as follows: A long 0.75-inch pipe was fitted a t one end with a small steam ejector and at the other with a short nipple about 6 inches long. The inside of this nipple was heavily insulated, and the thermocouple proper placed firmly inside the insulation, the hot junction being embedded in a small metal head. This head was perforated so as to allow a certain aniount of gas to be sucked past it at a very rapid rate, the rate being increased until the temperature remained constant. At the same time the head damped out minor fluctuations in gas temperatures. The amount of gas drawn past the thermocouple Tvas not large enough to affect the general flow of gas appreciably and hence the temperatures observed are probably quite accurate. Data obtained with this instrument indicated corrections from static temperature from zero to over 200" F. in the convection section. Attempts to correlate this correction with calculations based on a balance between heat absorbed and emitted by the couple were not very successful. The correction varied considerably, even in the same positiun of the thermocouple, depending on gas-flow rates, etc. Explorations across a bank of tubes, particularly near the top, showed that the temperature beneath the tubes was less than that of the gas between tubes, confirming the results obtained by many investigators that there is a "dead-space" behind the tubes on the first rows, resulting in lower heat-transfer coefficients. Table I shows the results calculated and observed, as obtained from plant data already available on two furnaces before this work was begun. Corrections were made on the ga> temperatures in accordance with similar data obtained later. The first half of this table is from data obtained on the entire convection section of a furnace. The log mean temperature difference between the oil and gas was added t o the arithmetic mean of the oil temperatures to get the average gas temperature. Frorn a knowledge of the carbon dioxide analysis and the hydrogen-carbon ratio of the fuel, P,L and P,L were computed. h, was found from Equation 5, hr, and h'?, from the Hottel Charts, and the wall correction from Equation 6. hobs.was obtained by dividing the heat absorbed b y the oil (based on sensible heat only, since evaporation and cracking mere negligible) by the outside area of the tubes and the log mean temperature difference. The agreement of theory with practice is fairly good in this case even though the data may not be very reliable. The specific: heats of the oil were determined by means of the Fortsch and Whitman equation. The second half of Table I is of interest in showing the effect of gills or fins on the outside of tubes. The calculated coefficient was determined for a plain tube of the same size and compared with the actual coefficient based also on the
A
INDUSTRIAL AND ENGISEERING CHEMISTRY
508
Vol. 24, No. 5
TABLE11. DATAOBTAINEDUSING\-ELOCITY THERMOCOUPLES G
TEST
-GAS-
In
Out
TEMPERATURES -OILLog Av. In Out M T D oil
Av. gas
. DIOXIDE CARBON
Lb./sec./ sq. /l. O F . O F . O F . OF. O F . 0.195 875 576 421 502 250 0.193 1657 875 502 713 616 220 C2; 0.162 832 557 426 490 0.160 1584 832 490 694 575 C3P 0.139 509 420 465 189 815 0.131 815 465 616 1670 678 C4; 0.133 50 1 415 177 456 772 456 652 587 1636 u 0.129 772 42 1 482 247 858 C51 0.169 574 698 605 0.158 1610 858 482 506 274 422 CS? 0.231 925 590 734 595 506 0.218 1545 925 465 C7? 0.137 169 430 785 505 665 567 0.130 1575 465 785 501 267 423 CS; 0.208 894 592 1507 894 u 0.193 716 572 501 D1 0.487 1250 905 618 741 388 D2 0.168 1200 791 657 713 273 D3 0.749 1127 906 665 757 305 D4 0.443 1144 850 636 741 298 El 0.155 1820 532 472 712 360 1079 870 620 660 333 F1 0.306 a 1 and u are, respectively, lower and upper sections
Clln
4
Out
In
%
70
462 712 8.0 608 1124 8.3 458 678 8.9 592 1167 9.3 632 1 1 . 3 443 572 1187 1 2 . 1 613 1 1 . 6 436 554 1141 1 2 . 1 699 452 9.7 590 1195 1 0 . 4 738 464 6.7 620 1215 7.3 448 617 1 1 . 5 565 1132 1 1 . 7 462 8.0 729 609 1181 s . 5 680 1068 685 958 615 711 1016 5.4 689 987 7.6 592 952 640 973 6:s of tube bank
7.6 8.0 8.5 8.9 10.6 11.3 11.2 11.6 9.0 9.7 6.6 6.7 10.9 11.5 7.5 8.0 7.3 6.0 4.9 6.0 7.5 5.6
O F .
O F .
plain area. The ratio of the actual to the calculated is a measure of the “efficiency” of the gills. This particular furnace had tubes with a ratio of total surface to plain tube surface of 3.8. The efficiency decreased steadily from the beginning of operations until a t the end of a year the gills were actually less than 100 per cent effective. This is undoubtedly due to loosening of the gills with continued temperature variations, since in all cases the tubes themielves were clean. Other finned tubes were found to give more favorable results, but in general they do not appear to be satisfactory unless they are firmly attached. Table I1 shows a series of data obtained on several furnaces using velocity thermocouples and with more complete and accurate information on gas analysis, fuel consumption, etc. These furnaces were in general divided into upper and lower parts, and separate calculations were made for the two parts.
0
PERCENT EXCESS AIR 40 60 HEATTRAKSFER I N COKVECTION
20
FIGURE 2.
FURNACE C
80
SECTION
OF
The data indicate that the calculated coefficients are reasonably in accordance with practice. In the most accurate and complete series (furnace C), there were minor fluctuations in the upper and lower sections in some cases, but these are due somewhat to inaccuracies in measuring the average midtemperature of the oil. At this point there were two streams in parallel, whereas a t the top and bottom there was only one stream. An arithmetic average was assumed on the two temperatures a t the mid-point.
PcL
PwL
hc
hro
0.07 0.0s 0.08 0.09 0.09
0.11 0.12 0.12 0.13 0.11 0.13 0.14 0.14 0.12 0.13 0.09 0.10 0.14 0.15 0.11 0.12 0.10 0.11 0.09 0.12 0.10 0.09
2.7 3.0 2.5 2.8 2.2 2.4 2.1 2.3 2.5 2.7 3.1 3.4 2.2 2.3 2.9 3.1 5.0 2.4 6.7 5.0 2.6 3.7
1.3 2 2 1.2 2.2 1.0 2.2 1.0 2.4 1.0 2.4 1.1 3.2
h’ro
JVALI.
EFFECThcsio.
hobs.
ERROR
4.6 6.1 4.3 6.2 3.4 5.9 3.2 5.7 3.5 6.2 4.7 6.8 3.0 6.0 4.2 6.6 7.4 4.6 9.7 7.8 5.1 7.1
-2 +3 -5 -2 f 3 -3 f 6 0 +8 -2 0 0 f20 -5 +7 0 +4 +2 -2 0 0 -9
%
0.10 0.10 0.10 0.08 0.08
0.06 0.06 0.10 0.10 0.07 0.08 0.08 0.07 0.06 0.08 0.06 0.06
1.1
2.4 1.2 2.4 2.2 1.9 2.1 2.2 1.8 2.0
...
0.4
...
0.4
...
0.3
...
0.3 ,..
0.4
...
0.3
...
0.3
6.6 ...
..,
... ...
0.5
. ..
12 15 12 15 11 15 11 15 11 15 12 15 11 15 12 15 8 10 8 8 6 15
% 4.5 6.3 4.1 6.1 3.5 5.7 3.4 5.7 3.8 6.1 4.7 6.8 3.6 5.7 4.5 6.6 7.7 4.7 9.5 7.8 5.1 6.5
Figure 2 shows the data obtained on furnace C plotted as heat absorption versus the per cent excess air (using gas fuel in all cases). Such a plot is not strictly accurate, since several factors besides excess air affect the results, but the general trend is that generally observed-namely, that increasing excess air shifts the heat load from the radiant to the convection section, with a consequent increase in stack loss. The calculated and observed heat transmission were quite close and a fair check on this was obtained also from the loss in heat by the gas, allowing for air infiltration. However, such a check is not very reliable, since a slight error in gas analysis will directly affect the heat loss calculated in this manner, whereas it will affect only the convective heat-transfer calculations to the two-thirds power and will not affect gasradiation calculations appreciably. Table I11 shows pertinent data on the sizes of the convection sections used in these tests. It will be noted that these furnaces vary considerably in tube dimensions, number of tubes, and heating surface. It is felt that the experimental results on these furnaces indicate that calculation of heat transmission in the convection section by the methods outlined will give reliable results. It is of interest that calculations of the heat transfer in the first two rows of tubes, assuming 70 per cent of the added gas radiation at the top going to the top row and 30 per cent t o the second, check closely with observed results. This is roughly in agreement with the calculations of Hottel (3) on black-body radiation to two rows of tubes. I n addition to the heat-transfer studies in convection sections, a n attempt has been made to correlate pressure drops on the gas side with equations presented in the literature. It is extremely difficult, however, to obtain the true pressure drop through such a bank of tubes, since this is small in comparison with the stack effect, which is in turn complicated by variations in gas temperatures. Reiher has presented equations for the pressure drop through banks of tubes, and Walker, Lewis, and McAdams ( 7 ) have derived empirical formulas based on some Sturtevant data. These equations may be correlated into a general one of the Fanning type using the concept of “hydraulic radius.” If Cd is the equivalent diameter, this equation may be expressed in ft.-1b.-sec. units as
wheref
=
0.085
[Cdy.
p]-”?‘ -
C is 1.6, 1.7, 2.0, and 2.2 for .V = 5 , 10, 20, and 30 rows of tubes, respectively. Correlation of this equation with plant
I S D U S T R I A L -4S D E N G I N E E R I N G C H E >I I S T R Y
Ma), 1932
data has not been made to date with any degree of accuracy, although the results are of the correct order of magnitude. It is hoped that reliable data for such cases may be obtained in the near future. TABLE111
D.iT.4 O V
FURSACES USED
Total
In.
In.
3.75 2
3
4
:
33.75 .5
Lou-er Baaed on plain tube.
= 3.8. b
Ft.
3.5 3.5 3.5 3.75 3 4.38
16.75 14 24 24 22.5 16.75 13.75
24 12
14 15 20 24 8
12 5 6 6 8 12 6
Sq. ft. 4720 735 2087 2260 4620 4720 865
In. 3.75
P 8
Sg.fl. 68.5 20.5 53.8 53.8 83.6 68.5 31.6
10 3.75 9 Ratio of actual surface t o plain tube surface
Same center-to-renter spacing in same r o t i or from row t o row. S03IESCLA.CUHE
area, sq. ft.; C = constant = diameter of tube, ft.; D = diameter of tube, in. = friction factor = heat-transfer coefficient, B. t. u./sq. ft.1’ F.!hr. = convection coefficient, B. t . u./sq. ft./’ F.!hr. = gas-radiat,ion coefficient, B. t. u./sq. ft./’ F./hr. = added gas-radiation coefficient at top, B. t . u./sq. ft./ =
0
k
=
n
P p Ap
t
T
IN ‘rESTS
NUMBER TUBE F~RTCBEDIMEMIOSSOF TUBES OUTSIDESpac- FREE ISGb AREA S.ACE SECTIOY0 . D. I. D. Length Vert. Hor. SURFACE Total Lonera Upper Lower Lower
L .1;
IL..
P ./ llr.
thermal conductivity, B. t . u. ft./sq. ft./’ F./hr.
u
G It’ Z p p
t
509
effective gas-layer thickness, ft. number of tube rows = partial pressure or radiating gas, atmospheres = emissivity of surfaces = pressure drop, lb./sq. ft = temperature, ’ F. = temperature, ’ F. abs. = velocity, ft./hr. or ft./sec. = mass velocity, lb./sec./sq. ft. = tube spacing, in. = viscosity, centipoises = density, lb./cu. ft. = viscosity, lb./hr./ft. = tube; w = wall; f = film = =
LITERATURE CITED (1) Chappell and LMcAdams, Trans. Am. SOC.Mech. Engrs. (Dec., 1926). (2) Hotte1,’Im. EKG.CHEX.,19, 888 (1927). (3) Hottel, Mech. Eng., 52, 7 (1930). (4) Reiher, 2. Ver. deut. Ing., 70, 47-52 (1926); Mitt. Po,schunqsarbeitungen, Xo. 269. (5) Rietschel, Mitt. Prujungsanst. Heiz und Luftungseinr. der T . H . Berlin. 3. No. 3 11910). (6) Schack, 2.tech. Ph&k, 5; 267 (1924); “Der Industrielle IVimieiibergang,” Stahleisen, 1929. (i) Walker. Lewis, and ~Mcddams,“Principles of Chemical Engineering,” p. 89,McGraw-Hill, 1927. RECEIVEDMarch 7 . 1932
(.:orrosion Protection in Cracking Equipment J WQUE C.
C
~ ~ O R R E LALN D
GUSTAVEGLOFF, Universal oil Products Company, Chicago, 111.
ORROSIOS is encountered in practically all refinery stills, particularly in oil-cracking installations. From our present knowledge of the subject, hydrogen sulfide is the chief, if not the sole, cause of corrosion in these operations; hence corrosion in cracking units is more destructive than in ordinary refinery equipment because of the increased formation of hydrogen sulfide from the decomposition of sulfur compounds present in the oil. The extent of corrosion is not a direct function of‘ the sulfur content of the oil undergoing cracking but rather of the type of sulfur compounds, since sulfur compounds such as mercaptans, sulfides, disulfides, etc. , undergo decomposition to form hydrogen sulfide to a greater extent under cracking conditions than thiophenes and similar compounds. The hydrogen sulfide content of gases from the cracking of high-sulfur oils may be 1.5 per cent or higher. Aside from the danger to life and health by rupture of corroded elements of the unit and economic losses by replacement, the capacity and gasoline yields of the plant are materially lowered, owing to the necessity of reducing operating pressures of the cracking system. Figure 1 is a close-up view of the type of rorrosion which occurs in an unprotected reaction chamber; Figure 2 shows a s x t i o n of a new heating tube, and Figure 3 a similar tube in an advanced stage of corrosion. The use of chemicals, principally of a n alkaline character, such as sodium hydroxide, lime, soda ash, and ammonia, has not proved entirely satisfactory. Plating with metals, such as chromium, is expensive and has proved unsatisfactory, as have also sprayed metallic coatings. Although i t is certain that the use of cofiosion-resistant alloys, such as chrome-nickel-steel, for the entire vessel would be successful, this method of protection is too expensive for general adoption. Ordinary steel linings applied in the field have given considerable trouble, in most cases because of buckling and loosening. The following methods of protecting vessels which are
internally heated by hot oil and vapors have proved generally successful.
CEMEXTLISISGS There are two types of cement linings which will be referred to as ganister and nonmetallic, respectively. The ganister lining is made from cement mixed with crushed firebrick (ganister) or quartz sand. The nonmetallic lining is composed of mixtures of furnace cement, quartz sand, and asbestos with sodium silicate binder. The binder in the first group is cement and in the second group is sodium silicate or “water glass.” For purposes of the present classification the first group will be referred to as ganister lining and the second group as nonmetallic lining. A successful protective coating of such mixtures for cracking vessels should have the following properties: 1. Resistance to corrosive substances such as hydrogen sulfide at high temperatures. 2. Permanency under rapid heating or cooling-i. e., having a Coefficient of expansion, crushing and tensile strength, and elasticity of properly related values. 3. Strong adhesion to metal and hardness to withstand hammering and cleaning of the inside surface of the chamber. 4. Ease of removal of adhering coke, pitch, and other materials which accumulate during the processing of the oil.
GANISTER LINIKG. The interior of the chamber to be lined is thoroughly cleaned and sandblasted to remove coke, oil, rust-scale, and other foreign material. The reenforcement for the ganister lining is next applied. This consists of a 2.5-inch diamond-mesh expanded metal, or preferably No. 13 gage cyclone, or similar steel wire fencing. The reenforcing material is cut into segments and fitted in place. It is tackwelded to the chamber to hold it in place, whereas the horizontal spacer bars of 0.25-inch round reenforcing rods are welded on. The reenforcing material is then broken loose from the chamber and securely wired to the spacer rods.
