(5) Knuth, E. L.. ARS J . 32, 1424 (1962). (6) Knuth, E. L., Ind. Eng. Chem. 52, 688 (1960). (') Knuth, E. L., Phys. Fluids 2, 84-6 (January-February 1959). ( 8 ) Knuth, E. L., Univ. Calif. (Los Angeles) Dept. Eng., Rept. No. 59-39,pp. 19-20. June 1959. (9) Ibid., No. 60-11, February 1960. (10) Othmer, D. F., Benenati, R. F., Goulandris, G. C., Chem. Eng. Progr. 57, 47-51 (January 1961). (11) Othmer, D. F., Benenati. R. F., Goulandris, G. C., DechemaMonuerabh 47. 73-98 (1962'). (12) Thomas, P'. H., U. S. Patent 2,803,589 (Aug. 20, 1957).
(13) Tribus, M., Asimow, R.. Richardson. ?;., Gastaldo, C., Elliot, K., Chambers, J., ELans, R . B., Univ. Calif. (Los Angeles) Dept. Eng., Rept. KO. 59-34,September 1960. (14) Tribus, M., Evans, R. B., Zbtd., No. 62-53,February 1963. (15) \Voodward, H. T., Chem. E n g . Progr. 57, 32-7 (January 1961). (16) \Voodward, H. T., 144th Meeting, ACS, Los Angeles, Calif., April 1963. RECEIVED for review February 26, 1963 ACCEPTED July 8, 1963
DRYING AIR WITH ACTIVATED ALUMINA UNDER ADIABATIC CONDITIONS R O B E R T J. G E T T Y '
AND W.
P. A R M S T R O N G
Chemical Engineerin8 Department, TlhJhington Ciziiersity, St. Louis 30. .\lo.
Sorption of water vapor from air at 1 atm. under dynamic adiabatic conditions was studied to improve understanding of industrial gas dryer operation. A new type of behavior was caused b y a temperature wave which preceded the sorption wave under certain conditions. A multiple correlation yielded a high correlation coefficient and a regression equation which expressed operating time below 0" F. effluent dew point as a function of four independent variables: temperature, moisture content and flow rate of inlet air, and desiccant bed depth. The adiabatic operating capacity of a desiccant bed is much less than that calculated from isothermal equilibrium data. The regression equation for adiabatic capacity here presented can be used for design of large industrial gas dryers which operate without internal cooling.
of gases by physical adsorption has become a n important industrial operation. Mathematical treatments of this problem have been attempted (2, 4-6, g ) , but solutions were obtained only for isothermal operation. Since many industrial adsorption systems operate adiabatically, a real need for adiabatic data has existed. The p i m a r > - purpose of this work was to study the main operating variables affecting dynamic sorption under adiabatic conditions and to develop relationships of these variables \vhich could be used by the design engineer. A design equation was obtained and a new type of sorption behavior, peculiar to adiabatic operation, was observed. Air at atmospheric pressure and of adjusted temperature and humidity was passed through a well-insulated fixed bed of activated alumina. Temperatures were recorded at several points in the bed and the effluent air dew point was determined at frequent intervals. T h e effect on effluent dew point of four independent variables-bed depth, air flow rate, inlet air moisture content, and inlet air temperature-was measured as a function of time. .4multiple correlation of the data yielded the following regression equation, which expresses the operating time below a 0" F. effluent dew point. RYING
\vhere
X, = 2.999 + 1.163 loglo e - 0.00943 t - 51.94 H A', = log of time to 0" F. effluent dew point
e
= contact
time, seconds (equals bed depth/air velocity through empty tower) t = inlet air temperature, O F. H = inlet moisture content?pounds of water per pound of dry air T h e limits of applicability of this equation are: bed depth greater than 0.8 foot, air velocity between 10 and 100 feet per minute? temperature between 50" and 120' F., and inlet moisture content between 0.005 and 0.0195 pound of water 60
I & E C PROCESS D E S I G N A N D D E V E L O P M E N T
per pound of dry air. T h e linear correlation coefficient for the multiple regression was 0.9983. indicating that the four independent variables completely define the length of a run. For inlet moisture contents above 0.0195 pound of n a t e r per pound of drv air, the regression equation becomes:
X. =
2.298
+ 1.163 loglo e - 0.00943 t - 16.0 H
The time of operation to a -20" F. deiv point \vas found to average 90 to 927, of the time to a 0" F. de\\- point. Therefore, if a particular application requires that the effluent dew point shall not exceed -20' F., the time as calculated by the regression equation should be multiplied by 0.90 to get the correct operating time. Comparison of the recorded bed temperatures and the effluent air de\v points revealed a new effect not previously reported in the adiabatic operation of gas dryers; at low inlet moisture content the sorption \vave (zone of active interphase transfer of moisture) moved through the bed more slo\vlv than the temperature wave produced by the transport of the adsorption heat by air flow through the bed. Heretofore only the congruent movement of both the sorption and temperature Lvaves ivhich occurs at high inlet moisture content has been reported. Il'hereas the latter behavior results in only a single break point in effluent dew point LIS. time. the former new type of behavior causes tivo break points to appear. corresponding to the arrival of: first, the temperature Lvave and then the sorption Lvave at the bed exit. Betcveen the t\vo break points the effluent deiv point remains constant at an intermediate level.
1
111.
Present address, Alcoa Research Laboratories, East St. Louis.
Comparison of these adiabatic results bcith the isothermal data usually obtained i‘rom laboratory adsorption tests showed that the sorptive capacity of a desiccant is much less in adiabatic than in isothermal operation because of the temperature wave moving through the bed. Experimental Method
Air of adjusted temperature and humidity was passed through a fixed bed of activated alumina and the effluent moisture content of the air was measured us. time. All runs were made using lj’4-inch to %mesh, A c o a Grade F-1 activated alumina from the same drum, reactivated at 400’ F. Compressed air a t 90 p.s.i.g. was filtered to remove entrained moisture and dust and was then split into t\vo parts, each passing through a hand-controlled pressure regulator and a flowmeter. One of these air streams passed through a saturator, consisting of a 4-foot aluminum pipe with flanged ends filled \vith 2l,, 2 feet of ‘?-inchceramic Berl saddles and flooded Lvith Lvater. Heating coils wound around the outside of the saturator supplied the heat of vaporization. ‘The \vet and dry streams tvere then recombined and passed to a separatormixing chamber. The resulting air of desired temperature and moisture content \vas then led to the sorption cell. Installed in the line betrveen the separator and the sorption cell was a dew point-measuring device of the silvered cup type. The outlet air from the sorption cell \vas passed through gas meters for measurement. The main portion was metered through a dr!, gas meter and the other portion through a wettest meter doimstream from a second dew point-measuring device. Figure 1 is a detailtd diagram of the sorption cell. Cellulose acetate sheeting. 0.005 inch thick, was used to form the internal cell wall. This material had sufficient mechanical strength to resist breakage. and was light and thin so as to give very lo\\- heat capacity and longitudinal heat transfer. Aluminum tubing, 3’ 4 inches in outside diameter, was used a t each end of the acetate cylinder to provide a rigid structure for inlet and outlet connections. Polystyrene foam, chosen for low density as \vel1 as low thermal conductivity, was installed on the outside of the acetate cell to act as a n insulator. ‘Thermocouples \vert: located along the axis of the cell in order to study the heat \vave passing through the desiccant bed. Enameled 28-gage Constantan \vires were Lvelded onto a 20gage common iron wire and were led out the bottom of the cell through a packed tee and connected to a 12-point temperature recorder.
METAL HOLDER
-
PERFORATED SUPPORT PLATE
T C BUNDLE -EXIT
Figure 1.
Table 1.
=
log of time to 0 ” F. dcw point
Z = bed depth, feet V = linear velocity. feet/min.
’ F. inlrt de\+.point, O F.
i = temperature,
t’
=
Source of T’uriniict
Experimental Design
V t
The independent variables selected for study \