Environ. Sci. Technoi. 1092, 26, 1341-1349
(7) Furlong, E. T.; Carter, D. S.; Hites, R. A. J. Great Lakes Res. 1967, 14, 489-501. (8) Hamdy, Y.; Post, L. J. Great Lakes Res. 1985,11,353-365. (9) Thornley, S.; Hamdy, Y. A n Assessment of the Bottom Fauna And Sediments of the Detroit River; Ontario Ministry of the Environment Toronto, ON, Canada, 1984; pp 40-42. (10) Strachan, W. M. J.; Eisenreich, S. J. Mass Balancing of Toxic Chemicals in the Great Lakes: the Role of A t mospheric Deposition; Report to the International Joint Commission on Great Lakes Water Quality: Windsor, ON, Canada, 1988. (11) Shiraishi, H.; Carter, D. S.; Hites, R. A. Biomed. Environ. Mass Spectrom. 1989, 18, 478-483. (12) Jaffe, R.; Hites, R. A. Environ. Sci. Technol. 1986, 20, 267-274. (13) DiToro, D. M.; Thomas, N. A.; Herdendorf, C. E.; Winfield, R. P.; Connolly, J. P. J. Great Lakes Res. 1987,13,801-825. (14) Carter, D. S. Ph.D. Dissertation, Indiana University, Bloomington, IN, 1992. (15) Lewis, P. J. Severe Storms Over the Great Lakes: A Catalogue Summary for the Period 1957 to 1985; Atmospheric Environment Service, Environment Canada: Downsview, ON, Canada, 1987. (16) Robbins, J. A.; Edgington, D. N.; Kemp, A. L. W. Q. Res. 1978,10, 256-278. (17) Gregory, W. J., Atochem North America, Inc., personal communication, 1991. (18) Hermanson, M. H. Geochim. Cosmochim. Acta 1990,54, 1043-1051. (19) Rea, D. K.; Bourbonniere, R. A.; Meyers, P. A. J. Great Lakes Res. 1980, 6, 321-330. (20) Paris, D. F.; Wolfe, N. L.; Steen, W. C. Appl. Enuiron. Microbiol. 1982, 44, 153-158. (21) Swisher, R. D. Surfactant Biodegradation, 2nd ed.; Marcel Dekker: New York, 1982; pp 415-457. (22) Hamblin, P. F. J. Great Lakes Res. 1987, 13, 436-453.
central and eastern basins, indicating that the western basin is somewhat isolated from the rest of the lake. The central and eastern basins have similar fluxes, and we conclude that transport of sediment from the central to eastern basin is not significantly restricted. Because the concentrations of 24DP in all unmixed sediment cores maximize in 1968 f 2 years, we conclude that the rate of sediment transport throughout the lake is 2 years or less. From a compartmentalized mass balance model, we estimate that 70-75% of Detroit River derived pollutants accumulates in the western basin, 20% in the central basin, and 5 % in the eastern basin, and 2% flows out of the lake via the Niagara River. Each basin retains -70% of the sediment that enters it. Acknowledgments
We thank the Canada Centre for Inland Waters for help in obtaining Lake Erie sediment cores; the Toledo Department of Public Utilities for help in obtaining Maumee River surface sediments; Kent Orlandini of the Argonne National Laboratory and Kevin Komisarcik of the Indiana University Cyclotron Facility for help with radiometric dating; Ilora Basu, Mike Howdeshell, and David Webb for technical assistance; and Hiroaki Shiraishi for many helpful discussions. Registry No. 24DP, 120-95-6;24DP6Cl,42350-99-2;2B4P6C1, 122269-06-1.
Literature Cited (1) Hileman, B. Chem. Eng. News 1988, 66(6), 22-39. (2) EPA. Great Lakes, America; U.S. Environmental Protection Agency: Washington, DC, 1980. (3) Bedford, K. W.; Abdelrhman, M. J.Great Lakes Res. 1987, 13,628-648. (4) Mortimer, C. H. J. Great Lakes Res. 1987, 13, 407-435. (5) Kemp, A. L. W.; Thomas, R. L.; Dell, C. I.; Jacquet, J. M. J . Fish Res. Board Can. 1976, 33, 440-462. (6) Carter, D. S.; Hites, R. A. J. Great Lakes Res. 1992, 18, 125-131.
Received for review November 20, 1991. Revised manuscript received March 9,1992. Accepted March 17,1992. The Charles Stewart Mott Foundation funded an International Association for Great Lakes Research Fellowship to D.S.C. This project was funded by the United States Environmental Protection Agency (Grant R81-6743).
Applying the Two-Resistance Theory to Contaminant Volatilization in Showers John C. Little Indoor Environment Program, Lawrence Berkeley Laboratory, Berkeley, California 94720
rn The two-resistance theory is applied to the transfer of volatile contaminants from shower water to indoor air by means of two transient mass balance models. Masstransfer coefficients are calculated from reported experimental data for five full-scale shower systems. Liquid- and gas-phase coefficients differ substantially from one shower system to another although both appear to increase with water flow rate. Mass transfer in showers appears to be strongly influenced by the type of showerhead but is unaffected by the presence of a showering individual. The calculated mass-transfer coefficients enable predictions to be made of the accumulation of volatile contaminants in shower and bathroom air during a typical shower. Exposures are calculated for volatile organic compounds representative of the range of volatilities found in water supply systems. Introduction
The major sources of human exposure to volatile organic 0013-936X/92/0926-1341$03.00/0
compounds (VOCs) occur indoors rather than in the outdoor environment (1). For example, levels of certain VOCs present in indoor air have been found to be more than 10 times higher than outdoors (2). One potential source of VOCs in indoor air is transfer from contaminated tap water during residential water use in, for example, showers, dishwashers, and washing machines. McKone (3) has shown that the daily indoor inhalation exposure attributable to contaminated tap water may be as much as 6 times higher than that incurred by consuming 2 L of the same water. Of the total inhalation exposure, more than half was projected to occur in the shower with an additional one-third occuring in the bathroom (3). Andelman ( 4 ) has summarized most of the early work on volatilization from household water in a recent review of exposure to VOCs in potable water via pathways other than ingestion, namely, inhalation and uptake through skin contact, He pointed out the need for refining present estimates of exposure to VOCs by, inter alia, more accu-
@ 1992 American Chemical Society
Envlron. Sci. Technol., Val. 26, No. 7, 1992
1341
rately accounting for the chemical characteristics that affect the rate and extent of volatilization. In order to account for the different properties of VOCs, McKone (3, 5,6)proposed a relationship which adjusts the measured transfer efficiency (the fraction volatilized) for radon to that for any VOC by using the Henry's law constant and liquid and gas diffusivities. Although the relationship was only intended to be approximate, it does not adequately account for gas-phase resistance (7).More recently, however, the results of five experimental studies on full-scale showers have become available (6,8-12). These results present an opportunity to determine mass-transfer coefficients for the various shower systems and then to use these to account in a consistent fashion for variation in contaminant volatility. In this paper, the classic two-resistance mass-transfer theory (13)is applied to the volatilization of contaminants from showers. Experimental data from five full-scale shower systems are used to dculate liquid- and gas-phase mass-transfer coefficients using transient mass balance models. These models account for variation in volatility, mass-transfer driving force, water and air flow rates, and volume of the shower stall and bathroom. The results from the five studies are compared, and measured mas-transfer coefficients are used to predict shower stall and bathroom exposures during a typical shower. Theoretical Development The version of Henry's law used to describe equilibrium between water and air is y = mc (1) where y is the gas-phase contaminant concentration in equilibrium with c, the aqueous-phase contaminant concentration, and m is a dimensionless Henry's law constant. The temperature dependence of m is commonly described (14) by the expression
m = (l/r)lO'-Jln (2) where J is a temperature correction coefficient and T is absolute temperature. The two-resistance theory (13,15)gives the overall resistance as the sum of two resistances in series, one for each of the phases, or
where KoL,KL,and KGare the overall, liquid-phase, and gas-phase mass-transfer coefficients, respectively, and A is the interfacial area available for mass transfer between the water and the air. For very volatile compounds (large m), the liquid-phase resistance controls while gas-phase resistance becomes significant as volatility decreases. An idealized and schematic representation of an experimental shower stall is given in Figure la. The volumetric flow rates of water (63and air (Qh)are assumed to be constant with time and the shower air volume (VJ is assumed to be well mixed. The air entering the shower has constant contaminant concentration ye, (usually, y., = 0), and at time t = 0, the concentration of the contaminant in the water entering the shower changes from zero to cin As the water falls through the shower stall, it loses contaminant at a rate proportional to the concentration driving force existing between the water and the air, or (4) dc/dt = -KoL(A/VL)(C- y./m) where V, is the volume of water present in the shower. Assuming that the water passes through the stall in plug 1342 Envlron. Sd.Technol.. Vol. 26, No. 7, 1992
Qr.
Flgvn 1. Idealized schematic repreSamatlon of (a. left)shower stall and (b. right) shower stall and L m m r m . showing ak and water fbws. Symbols described In text.
flow and that the shower air contaminant concentration y. is constant during the relatively short residence time of the water (the validity of this assumption will be checked in a later section), eq 4 may be integrated to give cout= c, exp(-N) + CY./m)(l- exp(-N)) (5) where N = (Ko,,A)/QL is a dimensionless overall masstransfer coefficient. In reality, the water will not pass through the stall in perfect plug flow and there will be wme distribution of residence times as shown by Tancrede et al. (12). This liquid-phase dispersion will be incorporated into the values of N (or KO& estimated from the experimental data. A transient mass balance on the air in the shower stall results in (dYs/dt)V. = QL(c~.- cOu3- QGCV. - Y.~J (6) Substituting eq 5 into eq 6 and integrating from an initial concentration ysiat t = 0 gives on rearrangement y. =
+
(ySi- i)exp(-bt)
(7)
+
where a = (QLc, (1 - exp(-N)) Q,Q~,)/V, and b = ((QL/m)(l - exp(-N)) Q&/V.. Quation 7 is similar in form to that presented by Giardino and Andelman (11). Equations 5 and I simplify under certain limiting conditions. When t m, y. a/b, which is the steady-state concentration in the shower air. Also, when ys = 0, covt = ci. exp(-N). Equations 5 and 7 give c, and y. as functions of time, a single overall mass-transfer coefficient, and experimentally measurable parameters. The m&ss-transfer coefficient (KO& may therefore be determined by measuring c, and either tout or y. at some point in time. A more reliable estimate may be obtained by measuring both tout and ye over a time period and then determining KOLAby means of a least squares fit to both data sets in turn. Since the model assumes a perfect mass balance, any discrepancy in the experimental mass balance will result in different values of the mas-transfer coefficient behg obtained from the water and air data. A check of the experimental transient mass balance may he obtained for any time period tl to t, using
+
- -
where the left-hand side represents the amount of contaminant volatilized from the water and the right-hand side represents the net amount of contaminant transferred to the air. Figure l b shows a schematic representation of the same shower stall located inside a bathroom where the volu-
metric flow rate of bathroom air (QGb) is also constant and the bathroom air volume ( v b ) is assumed to be well mixed. A transient mass balance on the air in the shower stall now results in (9) (dy,/dt)Va = QL(Cin - C o d - Q G , ~ , - Yb) where Yb is the contaminant concentration in the bathroom air. A similar balance on the bathroom air yields (dYb/dt)Vb = QGbbbin - Yb) - Q G s b b - YJ (10) Rearranging eq 10 and substituting eq 5 into eq 9 yields (11) dy,/dt = A1 + A& + A o y , and (12) dyb/dt = B1 + Boy,+ B o b where A1 = (QLCin)(l - ex~(-N))/Va A 2 = QG,/V, A 3 = ((-QL/m)(l - exp(-N)) - QGJ/V,
B1 =
(QGbYbin) B3
=
/v b
(-QGb
BZ = QGs/ - QG,) / v b
Vb
Table I. Experimental Parameters of Five Full-scale Shower Studies
studies
T (22) G (12) H (IO) M (6)
parameter
v.,L vi,L
1481
QL, L/min ACH., l / h QGs,L/min ACHb, l / h QGb, L/min
13.5 1.4 34.8
5 2.1-3.3 42-66
33, 42 1.7 12
42,46 1.8 11
T,,"C
1200
2800
8100
height, m duration, min
13.7 2.4" 110" 0.28 37.8 40 1.6 10
2300 9600 9.5 212 2460 -3 -480 22, 37 1.6 20
8.7 f 1.1
low low 40 f 2 1.8 10
Calculated, not measured.
Mass-transfer coefficients typically depend on diffusivity according to a power relationship (16,17) or K L a D L P and KG a DGq, where D L and D G are the liquid and gas diffusivities, respectively. Factoring the dependence on diffusivities into eq 3 gives
A solution to these two simultaneous differential equations, using the initial conditions at t = 0 of ysi and &,i is (17)
and
and
The value of tout is calculated from eq 5 as before. When ysi,Ybi, or both are exactly zero, they should be set to a finite, but very small number. Also, if QGb is set equal to a very large number, then eq 13 gives the same results as eq 7. Finally, the steady-state values are Y, = (A@i - AiBs)/(A& - A 2 B J (15) = - A3B1) / - A&) (16) The above models describe the volatilization of VOCs from a shower in terms of measurable experimental parameters and an overall mass-transfer coefficient K O L A . Mass-transfer coefficients determined in this fashion are typically found to vary with liquid and gas flow rates (which influence the degree of turbulence in the liquid and gas "films" as well as the interfacial area), temperature, and the liquid and gas diffusivities of the transferring chemicals. If two or more chemicals of differing volatility are transferred under identical hydrodynamic and temperature conditions and through the same interfacial area, then all of these variables remain constant except for the diffusivities. This provides a basis for separating the liquid- and gas-phase mass-transfer coefficients (16-18). Yb
where i represents the individual VOCs and r a selected reference compound. When compounds are chosen with similar air and water diffusivities, eq 17 reduces to eq 3, and the individual mass-transfer coefficientsKJ and K& can be evaluated from the intercept and slope of a plot of l/KoLA vs l l m for the various compounds. Earlier work on spray-type mass-transfer equipment suggests that there are four regions of mass transfer: drop formation, a period of drop acceleration to terminal velocity, the fall of the drop at terminal velocity, and coalescence on impact (19). Clearly, the mass-transfer coefficient for a shower will lump the contributions of the different regions into a single average parameter. There is considerable evidence that internal circulation within drops, and hence the mass-transfer rate, is largest during drop formation, release, and acceleration, and a theoretical analysis shows that K L a DL1/' during drop formation from orifices (19). Theoretical and experimental studies of mass transfer in packed towers and bubble columns also find that K L a DL1f2 (16,17,20). On the other hand, correlations for gas-phase mass transfer from a drop to a gas show that K G a D(32f3(19). Typically, the dependence of K G on the power Of D G has been found to lie between ' I 2and 2 / 3 in other mass-transfer studies (16, 17,20). Analysis of Experimental Data Tancrede et al. (12), Giardino and Andelman (11), Hodgson et al. (IO),McKone and Knezovich (6), and Jo et al. (8,9) (also referred to as studies T, G, H, M, and J) have all obtained experimental data from full-size shower systems. The main parameters of these five studies are listed in Table I while complete experimental details may be obtained from the original references. Further details concerning the analysis of the experimental data may be found elsewhere (21). ACH, and ACHb are the air-exchange rates in the shower and bathroom, respectively, expressed as a number of volume changes per hour. The physical properties of the various VOCs examined in the five studies are given in Table I1 and were obtained from a compilation of such constants (14,20). Studies T and G were performed in specially constructed experimental shower units with the shower air mixed by means of a fan, Environ. Sci. Technol., Vol. 26, No. 7, 1992
1343
Table 11. Physical Properties of Volatile Organic Chemicals (20 "C)
1 0 9 / ~ ~1,0 5 ~ ~ ,
voc
J, K
mz/s
mZ/s
1030 1820 1770 1990 1960 1930 1510 1.2-dibromo-3-chloro~rouane (DBCP)0.0056 2350
0.90 0.92 0.90 0.85 0.94 0.97 0.79 0.76
0.84 0.72 0.80 0.77 0.84 1.01 0.73 0.56
m
trichlorofluoromethane carbon tetrachloride l,l,l-trichloroethane (TCA) tetrachloroethylene (PCE) trichloroethylene (TCE) chloroform 1,2,3-trichloropropane (TCPA)
3.0 0.88 0.57 0.55 0.32 0.12 0.012
while studies H, M, and J were done in shower stalls located in residential bathrooms. Tancrede et al. (12)measured the simultaneous volatilization of five VOCs in four experimental runs at varying water flow rates and temperatures. The data from the runs at 33 and 42 "C were selected for analysis here with salient details reported in Table 111. The value of cin is the average of measurements at 2,6, and 11min, while that of cout is the average of measurements at 8 and 12 min. The contaminant air concentration ys was measured after 10 min of shower operation. Using these data, the percentage recovery is calculated (see eq 8) as the ratio of the mass of VOCs accounted for in the air to the mass volatilized from the water during the first 10 min of operation and reported in Table 111. In calculating the masses leaving in the water and air streams (the two integrals in eq 8), the value of tout was assumed to be constant with time while that of ys was assumed to be increasing linearly with time from zero (the validity of these assumptions will be checked in a later section). The values of KOLA obtained from the water and the air data are reported in Table I11 together with their ratios. Giardino and Andelman (11) reported data for the volatilization of three VOCs in an experimental shower system, and details of the three runs are given in Table IV. The contaminant concentration in the outlet water was calculated from the percentage volatilization quoted in the original reference and represents an average tout for the 11-min shower period. The mass balance could be checked for TCE and CHC1, as measurements of the air concentration with time were given (11) for these two compounds. KOLA was estimated in an iterative fashion for the water data (using eqs 7 and 5). The air data given
for TCE and CHC1, also enabled an estimate of KOLA to be obtained (using eq 7) by a least squares fit to the data points. The percentage recoveries, the values of KOLA determined from the water and the air data, and their ratios are also given in Table IV. Hodgson et al. (10) measured the volatilization of VOCs in a residential shower within a bathroom, and selected results are given in Table V. The air-exchange rate between the house and the bathroom was measured using SFGas a tracer; however, the exchange rate between the shower stall and the bathroom was not determined. Two 10-min runs were performed, one with and one without a showering individual, and measurements of the bathroom air were made rather than of the air inside the shower stall. The values of tout and Yb represent average values of the outlet water and bathroom air during the course of each 10-min shower. The initial contaminant concentration in the bathroom air, Ybi, (attributed to the background VOC level in the house) was also determined and is reported in Table V. For the present calculations, the initial shower concentration ysi and the concentration entering the bathroom Y b h were assumed equal to Ybi. An estimate of the air flow rate between the bathroom and the shower was initially obtained by finding the values of KOLA and Q G s which resulted in the best fit (using eqs 5 , 13, and 14) to the values of coutand Yb in Table V. Q G s was found to be 110 f 50 L/min (mean f standard deviation) and was then fixed at 110 L/min for all the data, and the values of KOLA were recalculated using only coUv The recomputed values of KOLA did not change from those estimated initially and are listed in Table V. McKone and Knezovich (6) measured the transfer of TCE during eight 20-min showers at two water temperatures. The inlet contaminant concentration cin was -100 pg/L throughout, and the transfer efficiency was found to be 58 f 11% at 22 "C and 63 f 9% at 37 "C. No statistical difference was found in the transfer efficiency with time or with temperature. Equations 13 and 5 were used to find KOLAvalues for each of the average transfer efficiencies quoted above. For the purposes of the present calculations, cinwas set at 100 pg/L and cout was set at 42 and 37 pg/L, respectively, for the two temperatures. Note that the value of KOLA is independent of cin since the mass-transfer model assumes that volatilization is a
Table 111. KOLAValues Calculated from the Data of Tancrede et al. (12)
CCl, PCE TCE CHC13 TCPA CCll PCE TCE CHC13 TCPA (I
33 33 33 33 33
13.5 13.5 13.5 13.5 13.5
0.104 0.26 3.1 1.5 90
0.025 0.086 1.2 0.70 71
0.0046 0.0083 0.118 0.050 0.66
72 60 77 81 43
20.9 16.3 15.4 14.6 4.3
10.9 7.2 9.3 8.4 1.7
1.9 2.3 1.7 1.7 2.5
42 42 42 42 42
13.4 13.4 13.4 13.4 13.4
0.094 0.24 2.8 1.4 89
0.022 0.078 1.0 0.67 73
0.0043 0.0078 0.129 0.056 1.03
14 60 93 91 82
21.0 16.2 14.7 12.9 6.0
11.4 7.1 12.3 9.9 3.1
1.8 2.3 1.2 1.3 2.0
Calculated from water data. Calculated from air data.
Table IV. K O L AValues Calculated from the Data of Giardino and Andelman (11)"
voc TCE CHC13 DBCP (I
QL = 5 L/min.
1344
T,, "C 46 42 42
QGs,L/min 66 52 42
tin,
pg/L
920 580 1680
tout,
rg/L
170 260 1300
Calculated from water data. Calculated from air data.
Environ. Sci. Technol., Vol. 26, No. 7, 1992
rec, % 102 128
L/min KOLA: L/min KoLAb/KoLAc 8.9 4.4 1.9
9.2 7.4
0.97 0.60
Table V. KOLAValues Calculated from the Data of Hodgson et al. (IO)"
VOC
person present
CC13F CC13F TCA TCA PCE PCE TCE TCE
no yes no yes no yes no yes
cinr
Coutt
Ybivb
~b~~
pg/L
pg/L
pg/L
pg/L
11 9.4 3.4 3.1
1.0
0.0040 0.0045 0.0014 0.0066 0.0060 0.0036
0.011 0.012 0.0066 0.0096 0.025 0.022
nd nd
nd nd
1.8
0.5 0.8 3.2 4.2 0.4 0.4
18 29 2.7 3.0
KOLA: L/min
20
34 25 29 21 26 27 30 31
n " 0
"QL = 13.7 L/min. bnd, not determined. cCalculated from water data. Table VI. KOLAValues Calculated from the Data of McKone and Knezovich (6)" VOC
T,, O C
TCE TCE
22 37
QL
Cin,
rg/L
100 100
Coutt
pg/L
42c 37c
KoLA,~ L/min 8.6 9.7
1
-
2
4
6
8
1
0
1
2
Time (minutes)
Figure 2. Ftt of eq 7 to the measured concentrations of VOC In shower air for study G. The circles represent CHCi3 data, while the squares represent TCE data [after Giardino and Andelman ( 7 7)]. h
.5
= 9.5 L/min. bCalculated from water data. CAverageof
four runs.
first-order process. The values of QGs and QGb given in Table I are used in the calculation even though they are derived from what appear to have been only estimates of the residence times (6). These results are presented in Table VI. Jo et al. (8, 9) measured the transfer of CHC1, from showers in a study which compared inhalation and dermal exposure while showering. As part of that study, 13 runs were performed without a showering individual present and a further 6 runs with a person taking a shower. The tap water CHCI:, concentration ch varied between 5.3 and 35.9 bg/L over the course of the 19 runs but was assumed constant for the duration of each run. The average CHC13 concentration in the shower air was measured during each 10-min shower period. The flow rate of air through the shower stall was not measured, but efforts were made to keep it to a minimum (8). Equation 7 was used to find KOLAvalues for each of the runs, with QGs = 0, giving the following results: KOLA = 2.4 f 0.7 L/min (without a person) and KOLA = 2.6 f 0.7 L/min (with a person), confirming the results of Jo et al. (8),who reported no statistical difference between the CHC1, air concentrations with or without a person present. The average value of KOLA for study J is therefore 2.5 f 0.7 L/min. Finally, it should be noted that a commercially available watersaving showerhead was used in study J (8).
Discussion of Results Figure 2 shows the fit of eq 7 to the concentrations of TCE and CHCI3in shower air, measured in study G, which yielded the values of KOLA shown in Table IV. A similar figure was given in the original reference by Giardino and Andelman (11). The measured air concentrations follow fairly closely the exponential growth predicted by eq 7, providing some justification for using the transient mass-transfer models to describe volatilization from showers. The mass balances for study T shown in Table I11 range from 43 to 93 5% indicating that, for all experimental runs, more VOC was volatilized from the water than was recovered in the air. This discrepancy is reflected in the ratio of KOLA values, which varies from 2.5 at 43% recovery to 1.2 at 93% recovery. The reasons for the fairly low recovery are unclear, but could be due to unidentified sinks
9
3
0.0
0
20
40
60
3
l/(Henry's law constant)
Figure 3. Plots of l/K,A vs llm where the circles represent the data from study T at 33 O C , the squares represent the data from study T at 42 OC, and the triangles represent the data from study 0.
for the VOCs (12),imperfect mixing of the shower air, or inaccuracies in experimental measurement. Notice, however, that the steady-state mass balances used by Tancrede et al. (12) accounted for only 9-19% recovery. This shows the importance of using a transient mass balance (eq 8) which accounts for accumulation in the shower air over time. The recovery for the data of study G (Table IV) is almost perfect for TCE, but 128% for CHCl,, and again this is reflected in the ratio of the values of KOLA determined from the water and air data. These data suggest that it is not an easy task to obtain good mass balances in this type of shower experiment. The values of KOLA determined from the water data of study T, and shown in Table 111,decrease as the volatility of the VOCs decrease. Neglecting the influence of diffusivities for the moment, these data are plotted according to eq 3 in Figure 3. The intercept and slope of the two lines are used to find values for KLA of 18 and 17 L/min and for KGp of 320 and 380 L/min at 33 and 42 "C, respectively. Earlier workers have found that the KG/KL ratio appears to be reasonably constant under similar conditions for a given mass-transfer system (for example, see ref 171, and once known, it may be used to obtain an estimate of KGp or KLA when only the other is known. This ratio is found to be 17 and 22 for the runs at 33 and 42 "C, respectively. An overall, but irregular, decrease with m is also observed for KOLA obtained from the air data of study T, as shown in Table 111. It appears that at least some of the irregularity may be explained by the variation Environ. Sci. Technol., Vol. 26, No. 7, 1992
1345
in recovery. The Kelp values for the air data of study G shown in Table IV together with the value for DBCP are also plotted according to eq 3 in Figure 3. Even though the data were obtained at different air flow rates and water temperatures, the points again appear to be linear. The slope and intercept give values for KCp of 130 L/min and for KLA of 9.5 L/min with a KG/KLratio of 13. In using eq 3 to calculate the individual mass-transfer coefficients, the influence of DL and DGhas been neglected. A rough check of the validity of neglecting changes in diffusivity may be obtained by assuming that KL and KG vary according to the and the 2 / 3 power of the diffusivity, respectively. Using the values of DL and DG given in Table 11, the terms in eq 17 involving diffusivity are found to vary from 1.00 to 1.13 for (DLi/DLr)o.5 and from 0.76 to 1-00for (DGr/DGi)0.67(DLi/DLr)o.5, where DBCP is the reference compound. Neglecting these terms would introduce an error of approximately f12%, which appears to be justified when compared with the relatively large uncertainties associated with the experimental measurements in showers (11,12). The values of KOLA obtained from study H are all for fairly volatile VOCs, suggesting a small gas-phase resistance. The average value for KG/KL obtained from studies T and G is 17, and assuming that this holds for study H, eq 3 can be used to show that the gas-phase resistance amounts to no more than -8% of the total resistance for TCE, the least volatile VOC examined. This means that the KOLA values in Table V are all essentially equivalent to KLA values which, if the influence of liquid diffusivity is neglected, should all be the same. Averaging gives KLA = 28 f 4 L/min (mean f standard deviation) with a coefficient of variation of 14%. This variation is -2 times higher than the influence of gas-phase resistance and the variation arising as a result of differences in DL. One surprising and useful result demonstrated by Hodgson et al. (10) and Jo et al. (8) is the statistically insignificant influence on mass transfer when a person is standing under the shower spray. This is reflected in the mass-transfer coefficients given in Table V as well as those calculated from the data of study J and lends confidence to exposure predictions based on mass-transfer coefficients determined from experiments without a showering individual present. If the KG/KL ratio of 17 is assumed to hold for the shower system of McKone and Knezovich, then gas-phase resistance accounts at most for -15% of the overall resistance in their studies. Therefore, the error introduced by neglecting KCp is about the same as the variation in experimental precision (6),and the KOLA values in Table VI are essentially equivalent to KLA values. A similar analysis shows that the KOLA value calculated from the results of Jo et al. (8) is roughly equivalent to KLA. A summary of the mass-transfer coefficients determined from the five studies is given in Table VII. The calculated mass-transfer coefficients differ quite substantially among the five shower studies, although they appear to increase with increasing water flow rate. The influence of water temperature on mass-transfer coefficients appears to be smaller than the variation in experimental precision as observed by Tancrede et al. (12) and McKone and Knezovich (6). By measuring water concentrations at different levels in their experimental shower, Giardino and Andelman (11) found that -60% of the TCE volatilized during the initial spray formation with little mass transfer taking place during the time of fall of the drops. Since the spray was not wide enough to contact the shower walls, they ascribed 1346
Environ. Sci. Technol., Vol. 26, No. 7, 1992
Table VII. Individual Mass-Transfer Coefficients Calculated from the Data of Five Studies Q L ~
study
L/min
Tw,"C
G J M
5 8.7 9.5 9.5 13.4 13.5 13.7
42-46 40
M T T H
22
37 42 33 40
KLA, L/min
Kd, L/min
KG/KL
9.5 2.5 8.6 9.7 17 18 28
130
13
380 320
22
17
the remaining 40% to mass transfer from the pool of water around the shower drain. These observations appear to be consistent with the results of this and previous studies, which find that the mass-transfer coefficient is not influenced by a person standing in the shower, since the majority of the mass transfer takes place either above or below the showering individual. Taken together with the low mass-transfer coefficient found for study J, these results further suggest that the extent of volatilization in showers can be reduced by using showerheads with low masstransfer characteristics. It may also be possible to reduce volatilization in showers by careful design of the shower stall and drain and by reducing the water flow rate.
Behavior of Transient Mass Balance Models The experimental parameters of the study by Hodgson et al. (10) comprise the most complete set of conditions representative of an actual residential shower. These parameters will be used as a basis for examining the behavior of the mass balance models. In addition to the parameters listed in Table I, &,, Ybi, and ysiare all set to zero. All results are normalized with respect to cin since the models are first order and hence independent of initial concentration. KLA is taken as 28 L/min since this is the average value found from the data of study H. Using the ratio KG/KL = 17 results in an estimate for KGA of 480 L/min. Equation 3 is used to calculate KOLA, and hence contaminant concentrations in the air and water streams. The effect of increasing the shower air exchange rate is also examined by increasing ACHs from 2.4 to 12 h-l, the value used in study M. Panels a-c of Figure 4 show the influence of volatility (m)and shower air exchange rate (ACHJ on water, shower air, and bathroom air concentrations with time. CC13F, CHC13,and DBCP are chosen as representative VOCs since they cover the entire range of volatilities examined. The most obvious impact of the decrease in m is the dramatic reduction in mass transfer which results in a substantially higher tout and lower y, and Yb. Also of interest is the change with time in both air and water contaminant concentrations. For CC13F, the most volatile VOC, tout is practically constant with time, while for DBCP, the VOC of lowest volatility, tout increases fairly substantially with time. The curtailment of volatilization which occurs at low m is due to the reduced driving force for mass transfer (see eq 4) resulting from the increase in y, and the low m. The effect of increasing ACH, from 2.4 to 1 2 h-I has little influence on the water concentration, but a strong influence on both shower and bathroom air contaminant concentrations. Tancrede et al. (12) measured the residence time distribution of the water passing through their experimental shower and found that the bulk of the water passed through the shower stall within -30 s. Examination of Figure 4b shows that the shower air contaminant concentration will, under most conditions, be reasonably
.e
Y
e,
Y
a 0
2
0.0'
'
0
"
2
'
"
6
4
'
"
8
E
' " 10 12
3 b
2
Time (minutes)
.3
0.041
'
8
'
8
'
'
'
I
'
8
'
d
0.00
'
'
"""'
'
.1
""""
'
1
'
10
Henry's law constant
8
I
.e Y
2 8 0
d
BY a €0 e
8l
Time (minutes)
Ee,
4
d
.3 m b
0
2 8 LI
0.006 -
Y
8
.
0
0
8
l
~
i
Figure 5. Estimate of potential exposure showlng the Influence of volatility on the average contaminant (a) shower alr concentration and (b) bathroom air concentration during a 10-mln shower. Concentrations are normalized with respect to the concentration of the Inlet water. Plot A shows ail eight VOCs glven in Table I1 and uses the reference conditions (Henry's constants evaluated at 40 "C). Plot B shows the effect of increasing the shower air exchange rate from the reference level of 2.4 to 12 h-'. Plot C shows the effect of decreasing the water flow rate from the reference level of 13.7 to 5 L/min.
.e
m
0.004
-
e
B
d
0
Henry's law constant
2
4
6
8
10
12
Time (minutes) Flgure 4. Normalized contaminant concentrations in (a) outlet water, (b) shower air, and (c) bathroom air with time. Concentrations are normalized with respect to the contaminant concentration in the inlet water. Plots A-C are for CCI,F, CHCI,, and DBCP at the reference conditions, respectively. Plot D shows the effect on (a) the DBCP concentratlon and (b and c) the CHCI, concentration of increasing the shower air exchange rate from the reference level of 2.4 to 12 h-'.
constant during this short liquid residence time, justifying the assumption made in an earlier section. Finally, the assumptions of constant tout (see Figure 4a) and of linear increase in ya(see Figure 4b) with time, used in the analysis of the data of Tancrede et al., appear to be excellent for all compounds except TCPA, the least volatile VOC used in that study. By assuming a constant c,,~, the amount of TCPA in the outgoing water was overestimated by lo%, which means that the recoveries for TCPA tend to be overestimated. The amount of TCPA in the outgoing air is so small that any error introduced by the assumption of linear increase in yais negligible. One final comment should be made concerning the use of the shower models to estimate mass-transfer coefficients.
-
The steady-state solutions can also be used to estimate KOLA from experimental data, however, for lower volatility VOCs, the steady-state concentrations become relatively insensitive to the value of Ked. Therefore, it is important to use the earlier transient data when KOLA is being estimated for compounds of low volatility such as TCPA and DBCP. The models described here should provide a useful tool for determining suitable experimental conditions in future research.
Assessment of Potential Exposure The set of reference parameters is now used to estimate the average shower stall and bathroom exposure during a 10-min shower. These reference conditions represent an upper bound for the potential exposure because they include the highest water flow rate and mass-transfer coefficients for the five shower systems. The influence is also shown of increasing ACH, from 2.4 to 12 h-l and of decreasing the water flow rate from 13.7 to 5 L/min, the value used in study G. In the latter case, KLA and KGAwere taken as 9.5 and 130 L/min, respectively, since the mass transfer coefficients depend on water flowrate. Panels a and b of Figure 5 show the average normalized concentrations in the shower and bathroom air during the 10-min shower period. The influence of volatility on exposure is significant, with the average normalized concentration for Environ. Sci. Technol., Vol. 26, No. 7, 1992
1347
CC1,F being -3 times that for the least volatile DBCP in both the shower and bathroom air. Assuming a breathing rate of 20 m3/day (3) and considering only the most volatile compounds, the inhalation exposure during a single 10-min shower is equivalent to 1.5times the ingestion exposure incurred by drinking 2 L of the same water. Decreasing the water flow rate to 5 L/min results in an exposure which is about one-third that at a flow rate of 13.7 L/min for all volatilities examined. At a shower air exchange rate of 2.4 h-l the shower exposure is -20 times higher than the bathroom exposure. Increasing the exchange rate to 1 2 h-l results in a reduced shower exposure, but an increase in the average bathroom concentration. The influence of volatility on exposure will be even more dramatic if the shower is left running for a long period. The low-volatility compounds reach steady state rapidly, but the highly volatile VOCs take a long time to reach steady state during which the concentration in air builds to comparatively high levels. For example, under the reference conditions, the normalized steady-state shower and bathroom air concentrations for CC13F are 0.39 and 0.29, respectively, taking -16 h to be achieved. The equivalent values for DBCP are 0.016 and 0.012, in this case taking only -3 h to attain steady conditions. The high contaminant concentration for the more volatile VOCs has serious implications for institutional shower facilities such as health clubs where showers may be operated intermittently for periods of up to 16 h at a time.
-
Summary and Conclusions The two-resistance theory was applied to contaminant volatilization in showers by means of two transient mass balance models. Overall mass-transfer coefficients were calculated using previously reported data from two experimental and three residential, full-scale shower systems. The simultaneous volatilization of VOCs of widely varying volatility in the two experimental systems enabled individual liquid- and gas-phase mass-transfer coefficients to be determined. This provides a means of accounting for variation in contaminant volatility more accurately than has previously been achieved. However, the results are strongly influenced by inconsistencies in the mass balance between the amount of VOC volatilized from the water and that recovered in the air. Measured liquid-phase mass-transfer coefficients, K J , range between 2.5 and 28 L/min while the gas-phase coefficients,KCp, vary from 130 to 380 L/min. Until more reliable data become available, a ratio of gas-phase to liquid-phase mass-transfer coefficients of 17 is recommended for shower systems. This ratio can be used to estimate KCp when only KLAis known. The mass-transfer coefficients differ substantially from one shower system to another, although they appear to increase with water flow rate. Allowing for the variation in the Henry's law constant, the influence of water temperature on the mass-transfer coefficients is smaller than the observed experimental precision. More accurate and precise data are needed as well as data more evenly spread over the entire range of contaminant volatilities found in water supply systems. The influence on mass transfer of the showerhead, shower stall and drain design, and water flow rate should be more closely examined, since these factors appear to offer the greatest potential for reducing the inhalation exposure. The average inhalation exposure occuring during a 10min shower is calculated for a wide range of contaminant volatilities using the observed conditions which result in the highest potential exposure. The inhalation exposure 1348
Environ. Sci. Technol., Vol. 26, No. 7, 1992
-
in the shower stall for the most volatile compounds is equivalent to 1.5 times that incurred through ingestion of 2 L of the same water. Shower stall exposures for the most volatile compounds are higher by a factor of 3 than those for the compounds of lowest volatility.
Acknowledgments I thank Joan M. Daisey, Alfred T. Hodgson, Tsair-Fuh Lin, William W. Nazaroff, and Robert E. Selleck for their thoughtful reviews of the manuscript.
Glossary Dimensions of variables are given in parentheses: L is a unit of length, M is a unit of mass, T is a unit of time, and 8 denotes temperature. constant defined in eq 7 constants defined in eq 13 effective air/water interfacial area (L2) constants defined in eq 11 air changes per hour in bathroom (1/T) air changes per hour in shower (l/T) constant defined in eq 7 constants defined in eq 12 concentration of contaminant in water (M/L3) concentration of contaminant in inlet water (MIL3) concentration of contaminant in outlet water ( M / L 3 ) diffusion coefficient for contaminant in air (L2/T ) diffusion coefficient for contaminant in water ( L 2 / T ) temperature correction coefficient for Henry's law constant (8) gas-phase mass-transfer coefficient (L/T ) liquid-phase mass-transfer coefficient (L/ 5") overall mass-transfer coefficient (liquidphase basis) ( L / T ) Henry's law constant (dimensionless) overall mass-transfer coefficient (dimensionless) volumetric air flow rate in bathroom (L3/T) volumetric air flow rate in shower ( L 3 / T ) volumetric water flow rate ( L 3 / T ) constants defined in eqs 13 and 14 time ( T ) absolute temperature (0) water temperature in centigrade (0) volume of water in shower (L3) volume of air in bathroom (L3) volume of air in shower (L3) concentration of contaminant in air (M/L3) concentration of contaminant in bathroom air ( M / L 3 ) initial concentration of contaminant in bathroom air ( M / L 3 ) concentration of contaminant in influent bathroom air ( M / L 3 ) concentration of contaminant in shower air (MIL3) initial concentration of contaminant in shower air ( M / L 3 ) concentration of contaminant in influent shower air ( M / L 3 ) Registry No. TCA, 71-55-6; PCE, 127-18-4; TCE, 79-01-6; TCPA, 96-18-4; DBCP, 96-12-8; CClsF, 75-69-4; CC11, 56-23-5; CHCl3, 67-66-3.
Environ. Sci. Technol. 1992, 26, 1349-1353
Literature Cited Wallace L. A. J . A m . Coll. Toxicol. 1989, 8, 883-895. Pellizzari, E. D.; et al. Environ. I n t . 1986, 12, 619-623. McKone, T. E. Environ. Sci. Technol. 1987,21,1194-1201. Andelman, J. B. In Significance and Treatment of Volatile Organic Compounds i n W a t e r Supplies; Ram, N., Chistman, R., canotr, K., E&; ~ ~ publishers, w i ~hc.: Arbor, MI, 1990, pp 485-504. ( 5 ) McKone' T*E*;Knezovich' J' Presented at the 82nd Meeting of the Air & Waste Management Association, Anaheim, CA, 1989; Paper 89-80.6. (6) McKone, T. E.; Knezovich, J. J. Air Waste Manage. Assoc. 1991,40, 282-286. (7) Little, J. C. Environ. Sci. Technol. 1992, 26, 836-837. (8) Jo, w, K.; Weisel, c. p,; Lioy, p, J. Risk Anal. 1990, 575-580. (9) J ~ w. , K.; weisel, c, p.; ~ ip. J,~Risk~ , 1990,10, 581-585. (10) Hodgson, A. T.; Garbesi, K.; Sextro, R. G.; Daisey, J. M. Lawrence Berkeley Laboratory Report No. LBL-25465; Berkeley, CA, 1988. (11) Giardino, N. J.; Andelman, J. B. Poster paper presented a t the Annual Conference of the American Water Works Association, Philadelphia, PA, June 1991. (12) Tancrede, M.; Yanagisawa, Y.; Wilson, R. Atmos. Environ. 1992,26A, 1103-1111.
(13) Lewis, W. K.; Whitman, W. G. Ind. Eng. Chem. 1924, 16, 1215-1220. (14) Selleck, R. E.; Marinas, B. J.; Diyamandoglu, V. Sanitary Engineering and Environmental Health Research Laboratory, UCB/SEEHRL Report No. 88-3/1; University of California, Berkeley, CA, 1988. (15) Treybal, R. E. Mass Transfer Operations, 3rd ed.; McGraw-Hill: New York, 1980. (16) Cho, J. S.; Wakao, N. J. Chem. Eng. Jpn. 1988,21,576-581. (17) Munz, C,; Roberts, p. V. Water Res. 1989, 23, 589-601. (18) Little, J. C.; Selleck, R. E. J.-Am. Water Works Assoc. 1991, 83(6), 88-95. (19) Sherwocd, T. K.; Pigford, R. L.; Wilke, C. R. Mass Transfer; McGraw-Hill, Inc.: New York, 1975; pp 218-231. (20) Little, J. C. Ph.D. Dissertation, University of California, Berkeley, 1990. (21) Little, J. C. Lawrence Berkeley Laboratory Report No. LBL-31452; Berkeley, CA, Oct 1991.
(1) (2) (3) (4)
Received for review September 24, 1991. Revised manuscript received February 27,1992. Accepted March 2,1992. This work was supported by N I E H S Grant P42 ES04705 and by the Director, Office of Energy Research, Office of Health and Environmental Research, H u m a n Health and Assessments Division of the U.S. Department of Energy under Contract DE-ACO376SF00098 through Lawrence Berkeley Laboratory.
Solidification/Stabilization of Hazardous Waste: Evidence of Physical Encapsulation Amltava Roy, Harvlll C. Eaton," Frank K. Cartledge, and Marty E. Tlttlebaum Colleges of Engineering and Bask Sciences, Louisiana State University, Baton Rouge, Louisiana 70803
A synthetic electroplating sludge was solidified/stabik e d in ordinary portland cement. The microstructure and microchemistry were studied by scanning and transmission electron microscopy, optical microscopy, energy-dispersive X-ray microanalysis,and X-ray diffractometry. The sludge contained 86.6 mg/g Ni, 84.8 mg/g Cr, 18.8 mg/g Cd, and 0.137 mg/g Hg. The water to cement ratio was 1.43, and the cement to sludge (dry) ratio was 1.2. An ordinary portland cement sample with a similar water to cement ratio, but without sludge, was also prepared for comparison. The sludge consisted of submicrometer-sized particles of complex heavy metal salts, calcium hydroxide, and calcium carbonate. Ellipsoidal particles of sludge, hundreds of micrometers long, were found in the cement matrix of solidified/stabilized material. Large concentrations of waste elements were found in the sludge particles, and low levels were also present in the matrix. The solidified/stabilized waste was a mechanical mixture of sludge and binder. Introduction Solidification/stabilization ( S / S ) is an economical process for the disposal of many types of hazardous wastes (1, 2). The method involves mixing liquid or semisolid wastes with binders to produce a solid which is structurally sound and relatively impermeable. Binders often consist of Type I portland cement (ordinary portland cement, OPC) or OPC plus fly ash, kiln dust, other pozzolanic and industrial byproducts. Sometimes polymers, by themselves or in various combinations, are used (3). The mechanisms of S / S are very incompletely understood (1, 4). The process usually involves addition of a heavy metal waste to a cementitious binder, with or without pretreatment with lime. At the resulting high pH, 0013-936X/92/0926-1349$03.00/0
heavy metals are expected to precipitate as their respective insoluble hydroxides since many heavy metals reach their lowest solubility at ca. pH 10 (5). The hydroxides are subsequently immobilized in the dense matrix of binder, where ionic transport is decreased as the porosity and permeability are reduced (6). Leaching studies by Cat6 (7) however indicate that the leaching from S/S binders cannot be explained on the basis of the solubility characteristics of simple hydroxide species of the heavy metals involved. Observed leachate concentrations from solidified metal salts are either greater or smaller than that expected for the pure hydroxide and vary with the nature of the metal and with pH. Most studies of S/S in OPC have concentrated on the mechanical properties (such as unconfined compressive strength) and leaching behavior of the treated products (2, 7, 8). Very few detailed microstructural and microchemical studies of OPC with complex wastes exist, and the waste itself has been investigated even less. The present research examines the nature of a synthetic electroplating waste (EPA classification F006,40 CFR 261.31) and its S/S mechanisms by investigating the microscopic morphologies and microchemistries of the S/S products.
Experimental Methods The sample preparation procedure has been discussed in detail by Cullinane et al. (9). The heavy metals were precipitated as their respective insoluble hydroxides by the addition of lime to a solution containing the nitrates. The precipitate was concentrated to ca. 25% solids, yielding heavy metal concentrations of 86.6 mg/g Ni, 84.8 mg/g Cr, 18.8 mg/g Cd, and 0.137 mg/g Hg. OPC was added to the sludge to attain a final mixture with a cement to water ratio 1.43 and a cement to sludge (dry) ratio of 1.2. A sample of portland cement with the same water to cement
@ 1992 American Chemical Soclety
Envlron. Sci. Technol., Vol. 26, No. 7, 1992
1349
