Experimental design
RSM was used to evaluate the independent, interactive, and quadratic input parameters (temperature, pH, and hydrogen peroxide concentration). The I-optimal method was also used for finding optimum points for the highest methane production and organic matter solubilisation. In mixture experiments, the proportions of the components of a mixture are the studied factors. The special nature of the factors makes special kinds of regression models necessary and special kinds of experimental designs. Although mixture experiments usually are to foresee the response(s) for all potential formulations of the mixture and to recognize optimal amounts for each of the ingredients, little research has been done regarding their I-optimal design. That I-optimal designs lower the average variance of prediction is so surprising and, consequently, looks more suitable for mixture experiments than the commonly used D-optimal designs, focusing on an exact model estimation rather than exact predictions[24].
Considering the results of previous studies [4, 25, 26] and the effectiveness of the pre-tests performed in this study, the proposed input range, as shown in Table 2, was applied. For statistical analysis, variable levels were normalized to three levels low (0), medium (1), and high (2).
Table 2
Levels and code of variables
Variables | Symbols | Range and value of variables |
0 (Low) | 1 (Medium) | 2 (High) |
Temperature (°C) | A | 25 | 75 | 90 |
Alkalinity (pH) | B | 8 | 10 | 12 |
H2O2 Concentration (mg/ g TS) | C | 0 | 30 | 60 |
Table 3: Experimental design with response surface methodology and the results
No.
|
Input Variables
|
|
Results
|
Methane Production
|
|
Microbial
|
Temp. (°C)
|
pH
|
H2O2
concentration
(mg/ g TS)
|
Cumulative MP
(mL/ g VS)
|
Increase
MP
(%)
|
COD
Solubilization
(%)
|
Increase
of sCOD
(g/L)
|
Decrease of VSS
(g/L)
|
Increase of Protein
(g/L)
|
Increase of
Polysaccharide
(g/L)
|
1
|
25
|
8
|
0
|
|
314
|
0
|
|
3.65
|
2.32
|
1.95
|
0.52
|
0.01
|
2
|
25
|
8
|
30
|
|
373
|
18.79
|
|
8.09
|
5.14
|
5.11
|
0.95
|
0.03
|
3
|
25
|
8
|
60
|
|
358
|
14.01
|
|
9.05
|
5.75
|
4.25
|
1.35
|
0.04
|
4
|
25
|
10
|
30
|
|
410
|
30.57
|
|
18.67
|
11.86
|
10.61
|
2.16
|
0.27
|
5
|
25
|
10
|
60
|
|
426
|
35.67
|
|
18.97
|
12.05
|
12.24
|
2.53
|
0.22
|
6
|
25
|
12
|
0
|
|
444
|
41.4
|
|
14.56
|
9.25
|
8.67
|
1.66
|
0.17
|
7
|
25
|
12
|
60
|
|
471
|
50
|
|
22.59
|
14.35
|
13.88
|
2.94
|
0.18
|
8
|
75
|
8
|
30
|
|
402
|
28.02
|
|
16.72
|
10.62
|
10.98
|
1.68
|
0.22
|
9
|
75
|
8
|
30
|
|
415
|
32.17
|
|
16.34
|
10.38
|
11.21
|
1.84
|
0.2
|
10
|
75
|
10
|
30
|
|
480
|
52.67
|
|
21.67
|
13.76
|
14.35
|
2.87
|
0.38
|
11
|
75
|
10
|
30
|
|
469
|
49.36
|
|
21.82
|
13.86
|
14.51
|
2.46
|
0.34
|
12
|
75
|
10
|
30
|
|
488
|
55.41
|
|
21.56
|
13.69
|
14.2
|
2.81
|
0.28
|
13
|
75
|
10
|
60
|
|
506
|
61.15
|
|
23.43
|
14.88
|
15.72
|
2.97
|
0.33
|
14
|
75
|
10
|
60
|
|
502
|
59.87
|
|
23.51
|
14.93
|
15.99
|
3.02
|
0.23
|
15
|
75
|
12
|
0
|
|
492
|
56.69
|
|
17.86
|
11.34
|
12.01
|
2.39
|
0.17
|
16
|
90
|
8
|
0
|
|
418
|
33.12
|
|
15.24
|
9.68
|
8.08
|
2.14
|
0.18
|
17
|
90
|
8
|
60
|
|
524
|
66.88
|
|
22.17
|
14.08
|
15.64
|
2.85
|
0.31
|
18
|
90
|
10
|
0
|
|
479
|
52.55
|
|
21.1
|
13.4
|
14.55
|
2.74
|
0.36
|
19
|
90
|
12
|
30
|
|
615
|
95.86
|
|
30.37
|
19.29
|
19.47
|
3.81
|
0.52
|
20
|
90
|
12
|
30
|
|
621
|
97.77
|
|
30.17
|
19.16
|
19.71
|
3.91
|
0.59
|
Design-Expert® software (version 11.0.5.0) was used for statistical analysis and to find the optimal answer. The three considered factors and their interactions were analyzed using the ANOVA table.
Pretreatment effect on organic matter solubilisation
Table 3 shows the effects of different pretreatments on the solubility of organic matter in waste activated sludge. A significant increase in COD solubility in all treated samples was observed compared to the controls, varying from 3.65–30.37%. The highest increase in sCOD was due to the triple combination of A2 + B2 + C1 (19th test) pretreatment, which increased the sCOD to 19.29 g/L. Similar results are recorded in previous research [25, 27]. On the other hand, VSS variations are closely correlated with the sCOD. The highest VSS reduction, which was equal to 19.71 g/L, was also obtained from the combined pretreatment of A2 + B2 + C1. According to the ANOVA table for COD solubility, which is provided in the supplementary file, a second-order model with respect to R2 > 0.98 and p-value = 0.0003, which is smaller than the acceptable value of 0.05 has been suggested. According to the analysis and by eliminating the unwanted terms, the following equation is proposed for COD solubility:
COD solubilisation (%) = -67.011–0.166 x A + 13.481 x B + 0.210 x C
− 0.017
x A x B + 0.004 x A2 − 0.509 x B2 − 0.003 x C2 (1)
According to the ANOVA table for VSS, each of the input variables had a significant effect on VSS solubility with R2 > 0.98 and p-value < 0.05. Interaction effects of BxC and CxA were removed because of their inappropriate p-value (> 0.05). According to the statistical analyses performed by the RSM and by eliminating the unwanted terms, the following equation is proposed for the VSS solubility:
VSS Solubilisation (%) = -130.034–0.180 x A + 25.442 x B + 0.343 x C
− 0.023 x A x B + 0.006 x A2 − 1.001 x B2 − 0.005 x C2 (2)
In this study, the amount of soluble protein and soluble polysaccharide were measured before and after the treatments. According to the data obtained, the amount of soluble protein and soluble polysaccharide increased significantly when the pre-treatments were applied to waste activated sludge. This enhancement was more considerable when a higher concentration of chemicals and higher temperature were used. The highest enhancement in soluble protein and soluble polysaccharide obtained from the bioreactor pre-treated with 20th test, in which they respectively reached to 3.91 g/L and 0.59 g/L. These values are considerably higher than those obtained from the control bioreactor with 0.52 g/L (soluble protein) and 0.01 g/L (soluble polysaccharide). The amount of soluble protein and soluble polysaccharide experienced a slight increase in the control bioreactor without any chemical addition or pH control, confirming a slight solubilisation in the control due to natural activity of the organisms. A similar enhancement was observed in previous studies [12, 28]. For protein changes, the quadratic model is proposed with R2 > 0.97 and p-value = 0.0008, which is a very good value that represents the power of the model. The proposed model with the elimination of unwanted terminology is equal to:
Increase of Protein (g/L) = -8.071–0.042 x A + 1.728 x B + 0.012 x C
+ 0.002 x B x C + 0.001 x A2 − 0.070 x B2 − 0.0002 x C2 (3)
Figure 1 shows the normal probability figures of COD and protein solubility. According to these forms, the less the dispersion of the existing data, the closer the results are to the normal line. In other words, when the R2 value tends to 1, the proposed model is stronger. Contour and 3D curves for COD and protein solubility at various concentrations of 0, 30, and 60 mg H2O2/g TS are shown in Figs. 2 and 3. According to Fig. 2, COD solubility is predicted to increase up to 32% in combined pre-treatment of 60 mg H2O2/g TS, 90 °C, and pH = 12, while the highest solubility enhancement from experimental results was 30.4% (19th test). The highest predicted VSS solubility was 55% in combined pre-treatment of 60 mg H2O2/g TS, pH = 12, and 90 °C, while the highest amount obtained from experimental results was 50% obtained from the 20th test (30 mg H2O2/g TS, pH = 12, and 90 °C). The addition of hydrogen peroxide to pretreatment demonstrated its function in breaking down the cell walls, as well as the EPS, caused by the release of radicals (hydroxyl radicals, hydroperoxyl radicals), converting organic matter to soluble phase [14].
Daily biogas production
As shown in Fig. 4 and Fig. 5, the highest daily biogas production was obtained between third and fifth days through the bioreactors. During the first days of the digestion process, the amount of biogas production in the control bioreactor was higher than most of the pre-treated bioreactors. This can be either attributed to the presence of inhibitory factors of the pre-treatment or due to the high organic loading available to anaerobic organisms and excessive volatile fatty acid [29].
Cumulative methane production
The amount of methane production from different bioreactors was measured regularly during the anaerobic digestion process. For normalizing the data, the produced methane (mL) was divided by the added volatile solids (gram) to each bioreactor. The effects of different pretreatment types on the cumulative yield of biogas production are shown in Table 3. The amount of cumulative methane yield was considerably enhanced when combined pre-treatment methods were used. The highest enhancement in methane production (97.77%), compared to the control, was obtained when the combination of A2 + B2 + C1 pre-treatments was employed. This is 30.89% higher than the highest increase achieved from the bioreactors with dual treatments (A2 + C2), corroborating the effectiveness of triple pre-treatment compared to individual and dual pre-treatments. As in previous studies, the methane enhancement from the individual pre-treatment used in this study was between 10% and 30% [13, 26, 30]. In this study, the methane increase from individual pre-treatment was between 14% in C2 and 33% in A2. Methane enhancement from dual pre-treatments was between 28% and 66%, while in previous studies, it was around 20–70% [12, 25].
Analysing the results of cumulative biogas production in Table 3 using RSM, a quadratic model with R2 > 0.98 and p-value < 0.0001 was suggested. By eliminating unacceptable terms and incorporating acceptable terms, the following equation was achieved:
Increase MP (%) = -39.897–1.798 x A + 8.408 x B + 0.678 x C
+ 0.005 x A x C + 0.0181 x A2 − 0.006 x C2 (4)
Figure 6 shows contour and 3D graphs related to the increase of cumulative methane production in different concentrations of hydrogen peroxide. According to these diagrams, the highest percentage of cumulative methane production is about 115%, which abstained from the bioreactors pre-treated with 90 °C, pH = 12, and 60 mg H2O2/g TS. It is important to note that the highest increase in cumulative biogas production, which was 98%, was observed in the 20th test.
Optimum pretreatment condition
A scenario was written for achieving the optimum condition, whereby temperature, pH, and H2O2 were minimized and the methane production, sCOD, soluble protein, and soluble polysaccharide were maximized. In the scenario, methane production was allocated the highest importance factor. The best-suggested desirability was 0.673. Table 4 shows the information pertaining to the optimization process. Another test with the suggested inputs was carried out for verifying the predicted phenomena in the model, in which the methane enhancement of 71% was achieved in the pre-treated bioreactor (Table 5). The difference achieved can be attributed to different sludge characterizations in the verification test. Despite the important results this study represents, applying new systems to anaerobic digestion of sewage sludge entails precise economic and feasibility assessments. Thus, in prospective studies, economic assessments for full-scale application of the pre-treatments are crucial, in addition to investigating possible side-effects of the pre-treatments on microbial communities and behavior in long-term exposure.