The amounts of Benzene (B), Toluene (T), Ethylene (E) and Xylene (X) were measured in 7 different reactors in the 84-day and are shown in Fig. 1.
When all graphs are evaluated together, the maximum gas output was measured as Xylene and the least gas output was measured as ethylene. In general, gas output concentrations in SWCNT, MWCNT and AC are parallel to each other and take same values. The difference between SWCNT and MWCNT is not at a level that can be observed from the graph. However, it can be explained by statistical evaluation. There was a relative decrease in gas output in the 5th Reactor (Fig. 1e), where MWCNT was added by increasing it by 2 times. When all materials are evaluated together, the reactor with the least gas output is the 2nd reactor (Fig. 1b). Accordingly, as a result of the graphical evaluation, it can be listed as GO > Graphite > AC > MWCNT(2) > MWCNT(1) > SWCNT from high efficiency to low efficiency in terms of gas adsorption efficiency. There are a few studies about BTEX removal in air using GO, AC, CNT and Graphite (Yu et al., 2018; Hwang et al., 2013; Quynh Ann and Tuan, 2019; Isinkaralar et al., 2022). Adsorption capacity of different materials is given Table 1.
In the Fig. 2–5, the changes in the amount of BTEX in reactors are considered cumulatively in the time series. Benzene concentration in Fig. 2 peaks on the 7th and 41st days. At day 7, benzene concentrations were determined from high to low as graphite, AC, MWCNT(2), MWCNT(1), SWCNT, GO and raw, respectively. At day 41, benzene concentrations were determined from high to low as graphite, AC, MWCNT(2), MWCNT(1), SWCNT, GO, and, raw respectively. Peak benzene formation in 7 days decreased to the lowest level in all reactors in 14 days, and after reaching the peak value again on the 41st day with a slow increase, the benzene concentration was zero in all cabins at the end of the 84th day. In the study of Yu et al. (2018), GO and rGO showed to be effective adsorption of benzene and tolunene. However, due to its lower oygen containing groups, rGO showed higher adsorption capacity than GO.
Toluene concentration in Fig. 3 peaks on the 32nd days. At day 32, toluene concentrations were determined from high to low as graphite, AC, MWCNT(2), MWCNT(1), SWCNT, GO and raw, respectively. At day 41, toluene concentrations were determined from high to low as GO, SWCNT, raw, MWCNT(1), AC, graphite, and MWCNT(2), and respectively. Similar to the normal distribution curve, toluene measurements showed a low trend, after reaching the peak value at 32nd day from low concentrations at the beginning, reaching the lowest toluene concentrations ranging from 0 to 8 µg.m− 3 in all reactors on Day 84.
Ethylene concentration in Fig. 4 peaks on the 7th days. At day 7, ethylene concentrations were determined from high to low as SWCNT, MWCNT(1), MWCNT(2), GO graphite, raw, and AC, respectively. Ethylene concentrations reached the lowest level on the 84th day by decreasing gradually after reaching the peak level on the 7th day in parallel in all reactors.
Xylene concentration in Fig. 5 peaks on the 25th days and 56th days. At day 25, toluene concentrations were determined from high to low as AC, MWCNT(2), MWCNT(1), graphite, SWCNT, raw, and GO respectively. At day 56, xylene concentrations were determined from high to low as graphite, AC, raw, GO, MWCNT(1), SWCNT, and MWCNT(2), respectively. The xylene values showed a slight decrease in the beginning, then reached the peak level on the 25th day and then decreased until the 56th day. After the 56th day, a very small increase was observed, and the gas output was completed on the 84th day with a low trend again.
When each reactor is evaluated separately for pollutants; It is seen in Fig. 6 that the benzene, toluene and xylene values show a positive skewed distribution in all reactors. However, it shows a nearly symmetrical distribution in all reactors of ethylene.
ANOVA test for Benzene, Toluene, Etylene and Xylene for the results obtained in 7 different cabins was performed. As can be seen in the Table 4, the null hypothesis cannot be rejected since all p values in the ANOVA table are greater than the Type 1 error of the test (α = 0.05). F critical value was found to be 4.95 for v1 = 6, v2 = 5 at 95% confidence interval. Therefore, no F value in the Table 4 is greater than the critical value. In this case, the null hypothesis is rejected. There is a significant difference between reactor systems.
Table 4
One-Way ANOVA Results of Parameters for Seven Different Cabins
Parameters
|
F-Value
|
P-Value
|
Benzene
|
0,90
|
0,501
|
Toluene
|
0,80
|
0,574
|
Ethylene
|
0,85
|
0,536
|
Xylene
|
0,84
|
0,541
|
The normality test results of the BTEX values in each reactor are given in Table. 5. Since the Skewness/Std.error value is greater than 1.96 for all samples for alpha 0.05, it is concluded that the distribution is not normally distributed. In addition, the normality assumption was not provided for the samples with the Shapiro Wilk normality test result smaller than alpha = 0.05 significance level. For values that do not show normal distribution, data transformation was performed by inverting the data according to excessive positive skewness, but it was determined that the data that were not normally distributed even as a result of the transformation were not normally distributed as a result of the transformation.
Table 5
The normality test results of the BTEX values
Compounds
|
Reactors
|
df
|
Skewness / Std. Error
|
Shapiro-Wilk Statistics
|
Sig.
|
Normality
(N/UN)
|
Benzene
|
1
|
16
|
3.929
|
0.265
|
< 0.001
|
NN
|
2
|
16
|
3.612
|
0.287
|
< 0.001
|
NN
|
3
|
16
|
3.500
|
0.313
|
< 0.001
|
NN
|
4
|
16
|
2.621
|
0.181
|
0.011
|
NN
|
5
|
16
|
2.890
|
0.264
|
< 0.001
|
NN
|
6
|
16
|
3.372
|
0.449
|
< 0.001
|
NN
|
7
|
16
|
5.902
|
0.323
|
< 0.001
|
NN
|
Toluene
|
1
|
16
|
1.624
|
0.888
|
0.052
|
N
|
2
|
16
|
1.567
|
0.825
|
0.006
|
NN
|
3
|
16
|
2.014
|
0.887
|
0.051
|
N
|
4
|
16
|
0.651
|
0.975
|
0.910
|
N
|
5
|
16
|
-0.138
|
0.942
|
0.376
|
N
|
6
|
16
|
1.489
|
0.901
|
0.084
|
N
|
7
|
16
|
0.569
|
0.961
|
0.675
|
N
|
Ethylene
|
1
|
16
|
0.661
|
0.964
|
0.729
|
N
|
2
|
16
|
2.537
|
0.825
|
0.006
|
NN
|
3
|
16
|
2.603
|
0.799
|
0.003
|
NN
|
4
|
16
|
1.379
|
0.868
|
0.025
|
NN
|
5
|
16
|
1.131
|
0.813
|
0.004
|
NN
|
6
|
16
|
-0.392
|
0.949
|
0.480
|
N
|
7
|
16
|
0.190
|
0.932
|
0.262
|
N
|
Xylene
|
1
|
16
|
2.213
|
0.907
|
0.106
|
NN
|
2
|
16
|
1.968
|
0.868
|
0.026
|
NN
|
3
|
16
|
1.973
|
0.865
|
0.023
|
NN
|
4
|
16
|
3.963
|
0.791
|
0.002
|
NN
|
5
|
16
|
3.512
|
0.796
|
0.002
|
NN
|
6
|
16
|
5.252
|
0.653
|
< 0.001
|
NN
|
7
|
16
|
2.613
|
0.868
|
0.026
|
NN
|
N: Normal distribution, NN: Nonnormal distribution
|
It was concluded that non-parametric tests should be applied to all data sets, assuming that the data are not normally distributed. Accordingly, Kruskall Wallis one-way analysis of variance test was applied for sample sets that were not normally distributed.
As a result of the Kruskal Wallis H test for benzene (\({\chi }^{2}\left(6, n=112\right)=\text{6,253};p=\text{0,396}>\text{0,05})\), there is no significant (alpha=0.05 level) difference between the results of different reactor groups. Since there is a normal distribution for toluene, non-parametric tests are not applied. Instead, the F test (ANOVA) is applied. As the significance value (Sig.) was 0.442>0.05 as a result of the one-way ANOVA test, there was no significant difference between the toluene measurements of the reactor groups. As the result of the kruskal wallis H test for ethylene (\({\chi }^{2}\left(6, n=112\right)=\text{8,811};p=\text{0,185})>\text{0,05})\), there is no significant (alpha=0.05 level) difference between the results of different reactor groups. Since the Kruskal Wallis H test for xylene was (\({\chi }^{2}\left(6, n=112\right)=\text{8,427};p=\text{0,208})>\text{0,05})\), there was no significant (alpha=0.05 level) difference between the results of different reactor groups.
For benzene, MWCNT was added in the 3rd and 4th reactors. While 0.9999 g of MWCNT was added to the 3rd reactor, 1.9999 g of MWCNT was added to the 4th reactor. The Mann-Whitney U test was applied to these two data sets, which did not show normal distribution as two independent groups. It is presented with the Wilcoxon test statistics, W of 260.00 and the statistical significance (2-tailed p-value) of this test, which is 0,894 (the p-alue is adjusted for ties) and equivalent to the Mann-Whitney U test). As the p-value is greater than 0.05, it can be concluded that there isn’t a statistically significant difference in median engagement between two groups; 3 and 4 reactor for benzene. As a result, the addition of 1 gram of MWCNT does not have a statistical significance in benzene values.
For Toluene, the effect of MWCNT was investigated in the 3rd and 4th reactors. Since Toluene values show normal distribution, t-test was applied for two independent samples. The null hypothesis is not rejected because the significance value calculated in the t test is (p = 0.757 > 0.05). According to this; Adding more than 1 g of MWCNT in the 4th reactor does not cause a significant difference in Toluene values.
In comparison of the 3rd and 4th reactors for ethylene with Mann Whitney U test (W = 243;p = 0.438 > 0.05), it can be concluded that there isn’t a statistically significant difference. As a result, the addition of 1 gram of MWCNT does not have a statistical significance in ethylene values.
In comparison of the 3rd and 4th reactors for xylene with Mann Whitney U test (W = 246.5;p = 0.522 > 0.05), it can be concluded that there isn’t a statistically significant difference. As a result, the addition of 1 gram of MWCNT is not statistically significant in xylene values.