I n d . Eng. Chem. Res. 1988,27, 1545-1547
nozzle. The inside diameter of the nozzle used was 27.6 mm. Ducts of four different radii and lengths (Table I) were used. The nozzle tip was set on the cross section of the duct inlet. The velocity profile within the duct was measured. A simple empirical correlation was obtained between the rates of air entrainment and jet flow. The correlation agreed with measurements to within f3.170.
Nomenclature
D, = diameter of cylindrical duct, m D, = diameter of nozzle, m H = length of cylindrical duct, m Q1,Q2= rates of air entrainment and jet flow,respectively, m3/s 2 = axial coordinate, m
Literature Cited Albertson, M. L.; Dai, Y. B.; Jensen, R. A.; Rouse, H. Trans. ASCE 1950. 115. 639. Choi, D. W.;’Gessner, F. B.; Oates, G. C. J. Fluid Eng. 1986,108,39.
1545
Habib, M. A.; Whitelaw, J. H. J. Fluid Eng. 1979, 101, 521. Hinze, J. 0. Turbulence, 2nd ed.; McGraw-Hill: New York, 1959; Chapter 6. Islam, S. M. N.; Tucker, H. J. J. Fluid Eng. 1980, 102, 85. Joseph, D. D.; Nguyen, K.; Matta, J. E. J. Fluid Mech. 1983, 128, 443. Perry, R. H.Chemical Engineers’ Handbook, 6th ed.; McGraw-Hill: Singapore, 1984; Chapter 5. Rajaratnam, N. Turbulent Jets; Elsevier Science: Amsterdam, 1976. Ricou, E. P.; Spalding, D. B. J. Fluid Mech. 1961, 11, 21. Schneider, W. J . Fluid Mech. 1985, 154, 91. Schlichting, H. Boundary Layer Theory, 6th ed.; McGraw-Hill: New York, 1968; Chapter 24. Wall, T. F.; Subramanian, V.; Howley, P. Trans. ICE 1982,60,231.
Tetsuo Akiyama,* Taketoshi Marui Department of Chemical Engineering Shizuoka University Ha mama t s u 432, Japan Received f o r review March 30, 1988 Accepted April 18, 1988
Steaming of Activated Carbon Beds When activated carbon is used in a thermal swing adsorption process, i t may be regenerated by a hot inert gas or by steam. Preliminary experiments with steam (and without removal of an adsorbate) showed intermediate axial temperatures to be higher than inlet or exit temperatures. It was concluded t h a t the heat of adsorption of water is the primary source of energy for the regeneration step of thermal swing adsorption involving steaming. Activated carbon is employed extensively as an adsorbent in both liquid- and gas-phase separations. Its affinity for nonpolar over polar adsorbates makes activated carbon particularly suitable for water purification and for the removal of organics from high-humidity gas streams. It is commonly employed in the area of solvent recovery. Steam regeneration is the most widely used in-situ method of activated carbon bed regeneration. Because of partial condensation and adsorption, steam can provide significantly more energy than hot inert gas. While regeneration is the energy-intensive phase of an adsorption cycle, it is rarely the time-limiting step. This may explain, at least in part, the lack of discussion of steam regeneration in the open literature. Published experimental data are extremely limited, and modeling has not been addressed a t all. Wankat and Partin (1980) and Partin (1977) collected some steam regeneration data as part of a study of a new solvent recovery process. Steaming was not considered a critical step, however, and therefore not discussed extensively. Scamehorn (1979) limited discussion of his steam regeneration data because of difficulties encountered in controlling the operating conditions. He did present plots of total desorbed material versus time and concluded that both the heating and purging functions of the steam were important. The present paper reports the results of an experimental investigation of heat transfer within an adsorbate-free bed of activated carbon. Temperature profiles within the bed were recorded as the bed was purged with low-pressure steam. This work, undertaken as part of a larger study of carbon bed adsorption and regeneration (Schork, 1986), is intended as a preliminary step in the investigation of steam regeneration.
Experimental Equipment The experimental equipment and procedures used in this study have been described elsewhere (Schork, 1986;
Schork and Fair, 1988). Broadly, the equipment included a gas feed system, a 7.44-cm i.d. by 30.5-cm-long column, and extensive temperature monitoring and recording equipment. The activated carbon used was 8 X 10 US mesh Witco JXC, a petroleum-based cylindrical extrudate material. The column was wrapped with electrical heating tape and insulated with about 5 cm of Fiberglas. Pressure gauges and thermocouples were installed above and below the column. Four thermocouples were located in the carbon bed, two a t 10 cm from the top and two at 20 cm from the top. These entered the column through Cajon Ultra-torr fittings which allowed insertion to any desired radial position. Four surface thermocouples were attached to the exterior of the column, evenly spaced along the length. The thermocouples were connectea to an Omega 2700A digital thermometer through a relay system controlled by a TRS Color Computer 2. All 10 temperatures were read within a 5-s period and stored in computer memory every 15 s.
Discussion of Results The steam required for these experiments, about 3 lb/h, was drawn from a large building header. This presented considerable control problems which caused several runs to be abandoned. The remaining runs are summarized in Table I. The notation “preheated indicates that the bed was heated externally to approximately the inlet steam temperature before steaming was begun. The notation “residual water” indicates that there was already some water on the carbon at the start of the run. Complete data sets for all runs are available in tabular form (Schork, 1986). Data presented here represent temperatures measured a t the center axis of the bed. Temperatures recorded during run S10 are plotted in Figure 1. Data presented in Figure 2 represent the heating of the same experimental system with a nitrogen purge. Several features of the data differentiate steaming from
0888-5885/88/2627-1545$01.50/0 0 1988 American Chemical Society
1546 Ind. Eng. Chem. Res., Vol. 27, No. 8, 1988
Table I. ExDerimental Steam Heating Runs P, TSAT,OTIN,* G, run Dsig "C "C mollcm2/min initial conditions 0.026 residual water, S5 4.2 107 113 T = 18 "C 112 0.026 residual water, 108 S6 4.5 4.7 4.2
108 107
113 114
0.026 0.026
S10 6.9 S11 6.9
111 111
120 121
0.042 0.041
S8 S9
preheated to about 105 "C T = 24 " C preheated to about 116 "C T = 23 "C preheated to about 108 "C
T ~ A T= temperature of saturated steam a t system pressure. Complete initial temperature profiles are available (Schork, 1986). LI
* T I , = temperature of inlet steam. 1
---
__
~~
7 -
-
-
~
_ -. ~.-~_ r _- ~ -~ 7
2 O l L ,
1
20
,
1
4C
I
1
60
1
-A
ac
I ?
rlME (mil)
Figure 3. Effect of residual water on steam heating.
Figure 1. Steam heating run S10. Initial bed temperature = 23 "C.
Figure 2. Nitrogen heating run N3: 6 psig, 3.9 x mol/cm2/min.
X
8 US mesh carbon, 21.2
nitrogen purge heating. First, during steaming, temperatures within the bed rose above the inlet value. This demonstrates the importance of the heat of adsorption of water in the overall energy balance. During nitrogen purging, the primary source of energy is the sensible heat of the gas. Thus, bed temperatures cannot rise above that of the inlet gas. The data of Wankat and Partin (1980) and Partin (1977) show no significant rise above inlet temperature during the steaming of carbon beds loaded with n-octanol. Instead,
the bed temperature rose rapidly and then leveled off near the inlet value. The heat of adsorption of the water was a t least partially offset by the heat of desorption of the octanol. Additionally, there was possibly some water already adsorbed on the carbon a t the start of these runs. These researchers employed a four-step cyclic process: (1) ambient nitrogen purge; (2) adsorption; (3) high-tempeature replacement regeneration; and (4)steaming. Water was found in the effluent from steps 2 and 3, indicating that the bed was not completely dried in step 1. Scamehorn (1979) analyzed the drying of activated carbon beds fairly extensively. He concluded that the water desorption rate is such a strong function of temperature that high temperatures are required to prevent excessively long drying times. That conclusion is supported by the present study. Upon external heating, significant quantities of water were driven off a bed which had already been cooled under nitrogen purge and then evacuated for several hours. The bed was prepared for runs S 5 and S6 by ambient purging and evacuation. Based on the observation noted above, it is assumed that the carbon still contained some adsorbed water a t the start of these runs. This is supported by analysis of the temperature histories of these runs versus those of runs S8-Sll. Prior to the latter runs, the bed was purged while being heated slowly to about 175 " C and then evacuated for about 2 h as it cooled. The effect of residual water upon bed temperature is shown in Figure 3. The exact quantity and distribution of water on the bed at the start of run S5 is not known, so a complete discussion is difficult. It is clear, however, that the maximum temperatures obtained during S5 are significantly lower than during S8. This further supports the conclusion that the heat of adsorption of water is an important source of energy for heating the bed. Returning to Figures 1 and 2, the second feature of interest is the rate of heating. The temperature rise with steaming was almost instantaneous compared to that with nitrogen purging. Upon contact with a cold bed, the inlet steam immediately condensed and the system pressure dropped to atmospheric. The heat of condensation certainly contributed to rapid heating, but only briefly. As the data presented in Table I1 show, bed pressure began to rise within minutes, and system conditions quickly reached those of superheated steam. As seen in Figure 4,preheating the bed caused the bed temperatures to rise even more rapidly. The high-temperature peaks of run S9 indicate that there was a very
Ind. Eng. Chem. Res., Vol. 27, No. 8, 1988 1547 Table 11. Steam Heating Pressure Data run S6
run S5 t , min P, psig 0.0 3.0 3.3 3.5 3.8 4.1 4.4 5.3 5.8 6.3 7.3 14.5
t , min
0.0 0.0 0.6 1.7 2.2 2.7 3.2 3.7 3.7 3.9 4.1 4.2
run S8 t , min P, psig 0.0 0.0 1.1 0.0 2.1 0.6 2.3 1.7 2.8 2.7 3.6 3.7 4.8 4.3 5.8 4.5 8.0 4.7
P, psig 0.0
0.0 0.8 1.1 1.8
0.6 2.7 3.7 3.9 4.1 4.2 4.5
2.8 3.8 4.5 6.4
0 RUN
0
-
INITiALLI COOL BED
, C' -
120
1 L 20
40
60
80
0.0
0.0
0.5 1.0 1.2 1.4 1.8 2.8 3.8 5.3
0.0 1.5 2.1 2.7 3.3 3.7 3.9 4.1
run SI0 t , min P, psig 0.0 0.0 1.5 2.5 1.7 3.3 1.8 3.7 2.1 4.7 2.4 5.5 2.7 5.7 3.5 6.2 4.7 6.5 5.4 6.5 6.4 6.7 11.3 6.9
run S11 t , min P, psig 0.0 0.0 0.5 0.6 0.8 2.7 0.9 3.7 1.3 5.2 1.5 5.7 2.0 6.5 2.6 6.7 3.3 6.9
the initial peak, the temperatures in the preheated bed dropped fairly rapidly to approach those of the initially ambient bed. The results of runs S10 and S11 are consistent with those of runs S8 and S9 and serve primarily to confirm the observations made during the discussion of the latter.
55 - PREHEATED BED
RlJN 5 8
run S9 t, min P, psig
100
TIME (min)
Figure 4. Effect of initial bed temperature on steam heating.
high initial rate of water adsorption. The temperature effect was dampened by the heat capacity of the column and carbon when the bed was not preheated. Perhaps the most interesting features of the steam heating data are the axial temperature gradients. At the start of run S8, there was almost no axial gradient a t all. During run S9, there was a clear reversal of the gradient from a negative to a positive value. In both cases, the bed temperatures rose almost simultaneously since the adsorptivity of steam, unlike the sensible heat of an inert purge, is not diminished by passage through the bed. The formation of a temperature plateau during adsorption is predicted by equilibrium theory (Basmadjian et al., 1975). Nonadiabatic operation has been shown, however, to cause the plateau height to decrease (Basmadjian et al., 1975; Schork, 1986). The region of the bed in which the temperature is constant, or relatively constant, is a region of relative inactivity. There must, of course, be a transfer zone in front of this region in which the heat of adsorption is sufficient to raise the temperature of the inlet stream. As the carbon near the entrance reaches saturation, the temperature there begins to drop and the transfer zone moves down the bed. This explains why a gradient eventually developed in the bed. The existence of the sharp temperature peak during run S9 can only be explained as a transient result of finite transfer rates. The bed temperatures rose sharply when the large amount of energy generated upon the initial contact of the steam and carbon was not dissipated quickly. The high temperatures could not be sustained, however, because of the temperature dependence of the carbon's water capacity. It is interesting to note that, after
Conclusion In summary, it can be said that the primary source of energy for desorption and heating during steam regeneration is the heat of adsorption of water. While carbon is generally regarded as hydrophobic, it does have a significant capacity for water. The heat generated by the adsorption of water cannot be neglected in analysis of carbon bed regeneration. A thorough study of steam regeneration would necessarily include determination of isotherms for the water/adsorbate/carbon system. Analysis of the 50-100-min steam runs discussed here is of value in developing a better understanding of the physical phenomena. In industrial practice, however, a steam regeneration cycle would be much shorter. A rule of thumb often cited is 4 lb of steam per pound of adsorbed organic. Based on this, only about 10 min would be required to regenerate the lab-scale bed saturated to 10% loading. Total bed pressure was not constant until after the first 5-10 min of the experimental runs. Since it is this initial period which is of interest, a model of steam regeneration should include a momentum balance as well as the mass and energy balances. For an initially cold bed, it may be possible to neglect the axial position functionality of the bed temperature and deal with a single bed temperature variable. Registry No. C , 7440-44-0.
Literature Cited Basmadjian, D.; Ha, K. D.; Pan, C. Y. Znd. Eng. Chem. Process Des. Deu. 1975, 14, 328. Partin, L. R. "Activated Carbon Solvent Recovery Process". Ph.D Dissertation, Purdue University, West Lafayette, IN, 1977. Scamehorn, J. F. Ind. Eng. Chem. Process Des. Dev. 1979,18,210. Schork, J. M. "Thermal Regeneration of Fixed Adsorption Beds". Ph.D Dissertation, The University of Texas at Austin, 1986. Schork, J. M.; Fair, J. R. Ind. Eng. Chem. Res. 1988, 27, 457. Wankat, C.; Partin, L. R. Ind. Eng. Chem. Process Des. Dev. 1980, 19, 446.
Joan M. Schork, James R. Fair* Department of Chemical Engineering T h e University of Texas a t Austin Austin, Texas 78712 Received for review October 7, 1987 Accepted April 4, 1988