Table 6. Contribution of experimental factors to
accepted as statistically insignificant. Liquid flow
ShallowTray stripping efficiency of TCE, PCE and
rate was somewhat significant to TPH removal
TPH (% contribution as determined by analysis of
but the main controlling factor was liquid tem-
variance [ANOVA]).
perature, which contributed to 74.3% of the TPH
stripping differences observed at the six factor/
Parameter
TCE (%)
PCE (%)
TPH (%)
level combinations. The interaction of the two
Flow rate
2.8
0.0
14.9
experimental factors was insignificant and the er-
Liquid temperature
64.3
15.7
74.3
ror or unexplained variability was low (8.4%).
Interaction
The low error term indicates that all variables
(Flow temp.)
15.6
60.1
2.4
significant to TPH stripping were controlled dur-
Total error
17.3
24.2
8.4
ing the experiment.
Sampling error
3.24
4.24
10.4
Analytical error
6.56
8.86
6.3
Unexplained error*
7.5
11.1
0
Treatment process
power requirements
* Total error (sampling + analytical error) = unexplained
Total treatment costs involve many variables,
error
and the specific capital, operating and mainte-
nance costs of the batch process presented here
the compound's volatility (PCE is slightly more
would be difficult to use for predicting large scale
volatile than TCE).
or continuous flow treatment systems. However,
The VOC error terms are both >15%, indicat-
it is instructive to examine the power consumed
ing that experimental error or unexplained vari-
per volume of water treated as stripping and wa-
ability was unacceptably high. Approximately
ter heating components. Power consumption rates
45% of this unexplained variability was due to
can be compared between widely different strip-
analytical and sampling error. The mean relative
per types and treatment methods to give a rough
standard deviation (RSD) of combined analytical
economic comparison for initial design choices.
and sampling variability is 9.8% (TCE) and 13.1%
Air stripper power consumption data were
(PCE) (see Table 6). The majority of the unex-
plotted, yielding an inverse relationship between
plained variability, 55%, is due to factors that
power consumed (kW/3785 L treated) and flow
were not fully controlled during the experiments,
rate (Fig. 8). The equation describing this rela-
in particular the VOC influent concentrations. Be-
tionship is the predictor equation for stripping
cause the highly effective VOC stripping produced
(only) power requirements:
consistent effluent concentrations over a narrow
[
]
-1
kW
= 0.02 (QL ) + 0.0157
range, the wide range of influent concentrations
(9)
3785 L treated
(which were not a function of the experimental
factors but of treatment process handling) cre-
where QL = liquid flow rate (L/min).
ated a large variability in VOC removals which
was independent of the controlled variables. This
Water was heated for treatment during six tri-
"process handling" variability is particularly no-
als, each time raising a different volume from a
ticeable in the duplicate trial data. Influent VOC
different initial temperature to either 15.56C or
concentrations of the two pairs of duplicate trials
22.2C. The kilowatts (kW) required to heat the
differed from 1.136.8% RSD. Effluent concentra-
liquid (using the equipment previously described)
tion variability was even greater (20.184.1% RSD)
were recorded. When these data were plotted as
due to the extremely low concentrations that ex-
kW/C increase vs. liters of water heated, the
aggerate even the smallest of differences. The net
linear relationship was described by a first-order
result of this uncontrolled error was to make it
regression:
impossible to correlate the VOC data into a VOC
removal prediction model.
[
]
kW
= 0.0034 (VL ) - 4.035
The statistical estimate of the main effects and
(10)
C increase
factor interactions during ShallowTray stripping
of TPH was also determined by analysis of vari-
where VL = liquid volume (L)
ance (ANOVA) of removal data (R). A summary
of experimental parameter contributions to TPH
From eq 10 it can be seen that when VL = 3785 L
(1000 gal.), 8.83 kW/∆C are required. This vari-
removal is shown in the right-hand column of
Table 6. A contribution below 15% is generally
able can be added to the equation for stripping
12
Previous Page
