The Application of Response Surface Methodology in the Study of Photodegraded Industrial Dairy Effluents by the Photo-Fenton Process: Optimization and Economic Viability

This study presents results from an application of Photo-Fenton process for organic-load reduction in dairy effluents. Process efficiency was evaluated in terms of percentage dissolved organic carbon, chemical oxygen demand, and biochemical oxygen demand (DOC, COD, and BOD, resp.), whose initial values were 1658 ± 145mgO 2 L, 9500 ± 500mgO 2 L, and 2400 ± 100mgO 2 L, respectively. We applied a statistical design represented by Box-Behnken factorial design inclusive of Fenton’s reagent, the power of applied radiation (W), and pH factors. The set temperature value was 30C with a reaction time of 60min. The maximum efficiency obtained was at pH = 3.5, Fenton reagent in the proportion of 35 g H 2 O 2 + 3.6 g Fe, and ultraviolet radiation potency of 28W. The results obtained for DOC, COD, and BOD were 81%, 90.7%, and 78.8%, respectively. Regarding the cost/benefit evaluation, the variables and their levels should be the following: pH 3.5, 35.0 g H 2 O 2 /Fe 3.6 g, and 28WUV, obtaining a reduction in concentration of 79.5% DOC.


Introduction
Dairy industry effluents are characterized by high volumes of water consumption and elevated organic contents and inhibited recalcitrancy for a conventional treatment [1].The escape into the aqueous environment of such pollutants, defined as industry-relevant organic components and BOD, COD, pH, fats, and phosphates, among others, demands treatment formulas that minimize the devastating impact of effluent pollution inclusive of degraded waterways, harmed environment, and a general detriment to animal and human health [2,3].
Economically advantageous biological processes are typically used for dairy effluent treatments regardless of a series of practical limitations [4,5].A common problem in this methodology is the oscillation of the organic load in the dairy effluent, resulting in expanded sludge volume and compromises to the efficiency of biological processes [5].
POAs are defined as processes with considerable capacity for hydroxyl radical (•OH) production [6,7].A high standard reduction potential (see (1)) for this radical is capable of oxidizing a wide variety of organic compounds to CO 2 , H 2 O, and inorganic ions from heteroatoms: Among the POAs, the use of Fe 2+ /Fe 3+ in the presence of hydrogen peroxide under irradiation, called a Photo-Fenton reaction, is considered the most promising for remediation of effluents containing a variety of toxic nonbiodegradable organic compounds [8,9].
Using a process parameter optimization, the experimental design has been widely employed to improve product quality through the application of engineering concepts and statistical models [12].The Box-Behnken models are a class of models rotational or almost rotational second order based on factorial models incomplete of three levels.The number of experiments () required for the model development Box-Behnken is defined as follows: where  is the number of factors and   is the replicate number of the central point.
This study focused on the application of the Photo-Fenton reaction to a Box-Behnken matrix in treatment of dairy effluent.

Sampling and Preservation.
The effluent studies were performed at an industrial dairy in the city of Guaratinguetá, Vale do Paraiba Region, in the state of São Paulo.Samples were collected from the production line feeding the treatment plant before the acid pretreatment.400 L of raw effluent was collected and held in refrigerated storage at 4 ∘ C.
Due to complex characteristics, some methodologies were adapted to improve accuracy and analytical precision as suggested by Lima et al. [13].Determinations of COD in dairy effluents required adjustments to 5220 D Closed Reflux colorimetric method of APHA-AWWA [14].Winkler's modified method was used to measure BOD.Preliminary tests with diluted deionized water were specific to concentrations for each methodology for COD and BOD determinations.Dissolved organic carbon determination was performed in a TOC-VCPH Total Organic Carbon Analyzer (high sensitivity) from Shimadzu on high temperature catalytic oxidation.

Degradation of Dairy Effluent by Photo-Fenton Process.
Photochemical treatment was undertaken in a semibatch reactor with a plug flow reactor model GPJ-463/1 with nominal volume of 1 L. Mercury lamp was used of model GPH-463TJL with UV radiation in 254 nm and intensity of 15 W and 28 W protected by quartz tube.Fenton's reagent was added at the following concentrations: 0.82 mol L −1 of FeSO 4 ⋅7H 2 O, H 2 O 2 at 30% m/m.After thermal conditioning of the effluent, a simultaneous addition of ferrous ions and H 2 O 2 solutions were added into the system.These additions were performed using a dosing pump during 20 or 30 min intervals through 1 h of reaction time.Sulfuric acid and NaOH (both at 5.0 M) were used to maintain a consistent medium acidity.This control was performed by a borosilicate glass-electrode potentiometer.The electrode was kept in the reaction bath.Figure 1 presents a detailed treatment schematic.The operational stages with Photo-Fenton process were as follows: (1) 3.0 L of dairy effluent was kept at room temperature, homogenized, and placed in a glass container; (2) thermostatic bath and centrifugal pump were turned on; (3) temperature was adjusted according to experimental design; (4) pH was regulated according to experimental records; (5) reactor was started to emit UV radiation; (6) in parallel, ferrous and peroxide solution were added during 50 min of a 1 h reaction; (7) after adjusting all parameters and setting the time control for 1 hour treatment, pH was kept constant; (8) aliquots (20 mL) were taken every 10 minutes; (9) the pH of both rates was adjusted from 8.0 to 9.0 for ferrous ions precipitation and filtered in quantitative filter paper; (10) each sample of dairy effluent was submitted to determine concentrations of COD, H 2 O 2 , and DOC.Dilutions owing to pH adjustment (Advanced Oxidation Process-AOP-and precipitation) were taken into account to the concentration calculation in all analytical determinations.Table 1 illustrates control factors and levels for treatment with AOP according to Box-Behnken matrix (used in response surface).

Box-Behnken Matrix
Table 2 presents the Box-Behnken matrix operating on three levels for planning experiments with AOP dairy effluent.

Characterization of Raw
Effluent.The characterization of dairy effluent was carried out according to the most relevant physicochemical aspects, such as pH, COD, BOD 5 , total phosphorus, ammonia-nitrogen and organic nitrogen, DOC, turbidity, color, total dissolved and fixed solids, settled solids, oils and greases, and the ratio of BOD 5 /COD.
Table 3 shows physical/chemical data for dairy effluent in natura and after the AOP and chemical treatment (precipitation in pH 8 and 9) standards as per Articles 18 and 34 of Brazilian Statute CETESB CONAMA 357/05 and values from respective references.
The actual process of degradation which occurred in the analysis of both DOC (81.0%) and COD (90.7%) is evident since in the marked reduction of the same after the photochemical treatment.
The laws of the State of São Paulo and Federal government, in relation to water amd receiving waters, do not have a specific value of COD for effluents in receiving waters.We recommend the value of BOD < 60 mg L −1 or a reduction in the minimum efficiency of treatment processes at least 80%.In general, in both parameters (DOC and COD), treatments by photocatalysis homogeneous were effective.
The ratio BOD 5 /COD parameter is commonly used to verify the biodegradability of the effluent.The effluent in natura, as a crude pollutant, had a BOD 5 /COD ratio of 0.25.This value indicates that the effluent presents high recalcitrance, making biological treatment impractical without a due pretreatment.The BOD 5 /COD ratio shows the effectiveness of oxidation in destroying certain organic loads.Biodegradability was evaluated as described by Jardim and Canela [16], noting that a BOD 5 /COD ratio of >0.4 is characteristic for biodegradable effluent.Hence, Table 3 shows a ratio result from AOP (photonic and Photo-Fenton) of 0.65, asserting that the photochemical treatment increased the biodegradability of the dairy effluent.
The results of color and turbidity proved to be quite satisfactory in reducing 93.55% of color and 98.83% of turbidity.
Table 3 shows copper at concentrations 4 times higher than the allowed (1.0 mg L −1 ) as well as an increase in concentration after the photochemical treatment.This increase is due, possibly, to the use of cooling system of serpentine pipes made from copper.Leaching of this metal in systems with acid pH, as used in the photochemical process, is common in increasing the concentration of ionic copper.
Phosphorus percentages were reduced from their high after photochemical treatment.In the analysis of treated effluent precipitates after alkalinization, phosphorous was at 98.11 mg L −1 , possibly attributed to the formation of lowsolubility phosphate compounds after oxidation concurrent with alkaline solubility at a lower value [17].
Chlorides presented a 98% reduction.Precipitate analysis showed 967.80 mg L −1 of chloride in the effluent sample that can form insoluble chlorides in alkaline medium, in the presence of cations as Cu 2+ , Cd 2+ , and Pb 2+ that could be present in the effluent.The ammonia-nitrogen and organic nitrogen also showed a reduced percentage with similar values of 99%.A possible gaseous oxidation of ammonia, a very stable form, could be the cause [18].
The results of total and volatile solids presented a percentage reduction in the order of 50% and 64%, respectively.
Cadmium and manganese showed values close to the allowed standard (2.0 and 0.2, resp.).Carbonate and bicarbonate ions were not analyzed because of a tendency to transform into CO 2 and H 2 O at acidic pH levels.The Photo-Fenton process was used in an acid medium.It is well known that oxidative processes can only oxidize metallic ions to a less soluble form in relation to the reaction pH and consequently lead to a decrease in its respective concentrations.4 shows the results of percentile reduction of DOC obtained with the experimental conditions anticipated in the Box-Behnken matrix.Fifteen experiments with replicates were performed using three factors (pH, Fenton reagent, and UV) at a maintained temperature and reaction time.Generally, the results of Table 4 ranged from 55 to 80% on average, depending on factor levels.Experiment 5 showed the highest percentage of DOC reduction at 80%, similarly reflected in the counterpart of experiment 8.

Box-Behnken Matrix Results for Total Organic Carbon (TOC). Table
The Box-Behnken matrix analysis by graphics of effects (Figure 2) indicated that the Fenton reagent factor (m H 2 O 2 + m Fe 2+ ) and ultraviolet radiation strength must operate at maximum levels.Intermediate level (3.5) pH data indicated elevated effluent-degradation performance.These values are consistent with those described in the literature regarding data indicating the potentials in applying an oxidative process for dairy effluent treatment [8].
As per the discussions of the Box-Behnken matrix effects, a pH in the range of 3.5 favors the Photo-Fenton process.A greater amount of Fenton reagent potentiated with UV radiation may produce improved kinetics for the formation of OH radicals and mineralized organic matter in the effluent and gives the highest percentage of reduced DOC.The adverse effect, which occurred with the intermediate concentration of Fenton reagent in percentile relation to reduction degradation, can be a function of a mass ratio where intermediate reactions of action reducing hydroxyl radicals were predominant, attributed to the antagonistic ratio between the quantity H 2 O 2 and Fe 2+ .Systematic errors should be disregarded due to the results of the similar percentage reductions of DOC between the experiments and their replicas.
Table 5 shows the percentage of reduction for DOC, COD, and BOD 5 and their relationships in each Photo-Fenton processes sample to the optimum experimental data in the Box-Behnken matrix.
The increases in biodegradability occurring in AOP treatments are clearly seen in the DOC/COD relationship confirming the efficiency of photochemical treatment process.
The resultant effects and interactions are obtained for the DOC parameter response under optimal interpretation using photocatalytic variance oxidation process, as shown in Table 6.

AOP Degradation Rate.
The percentages of DOC, COD, and BOD reduction were verified photochemicaly after the treatment at 81.0, 90.7, and 78.8%, respectively.The values in Table 5 show the relations of BOD 5 /COD and DOC/COD, after AOP and chemical treatments with the reduction of BOD 5 at 78.81%.This demonstrates that a majority of the organic load present in the dairy effluent was degraded by the photochemical processes, meaning a significant AOP contribution for the degradation of dairy effluent.Summary analysis of Table 4 data shows a greater influence from photochemical pretreatment in the percentages of reduction for both concentrations of DOC proportional to those of COD and BOD 5 .

Economic Evaluation of Photo-Fenton Process.
The economic evaluation (energy consumption and reagents) of the treatment process for dairy effluent, second Box-Behnken, was performed taking into consideration only the chemical process (Photo-Fenton), that is, omission of physical/chemical analyses.
As detailed above, the Photo-Fenton process was performed using the energy consuming equipment: tubular photochemical reactor Germetec Model FPG-463/1 (nominal volume of 1 L and Hg lamp low-intensity irradiation GPH-463T5L, emitting maximum radiation at 253.7 nm and power of 15 W and 28 W, protected by a quartz tube without causing dispersion), the Quimis Thermostatic Bath Model of Q214S (nominal volume of 10 L of H 2 O, with unit refrigeration with capacity of 3000 BTU/h and power of 2100 W at 220 V), the centrifugal pump of Bomax Model NH-30 PX-T (0.013 HP and 220/380 V), the metering pump from Hanna Instruments Model BL 1.5 (output 1.5 L/h and 13 bar, and 220/240 VAC at 50/60 Hz), and pH meter Hanna Instruments Model HI 2221 (output 500 m, 12 VDC, 220 V, and 50-60 Hz).The estimated value of energy used by this equipment was determined by experimental measuring using power/consumption meter from ICEL, Model ME-2500 (220 V and 60 Hz), with Certificate of Conformity no.201111011512, as per Table 6.Energy and temperature controls for all experiments were initially calibrated to a water bath thermostat at 20 ∘ C, taking into consideration the time to reach working temperature.
As per the data in Table 6 (60 min duration), the temperature parameter is the highest energy consumption, requiring the highest value among the other variables (>95%).The second highest energy consumption is the low-intensity Hg lamp and centrifugal pump (approximately 1% to 2%).Thus, the process cost is significantly dependent on the value and respective control of temperature.
The percentage of degradation was lower in previously performed formularies, that is, 3 L of dairy effluent, 100 g of H 2 O 2 , and 11.9 g Fe 2+ , adding these two reagents added in the course of the reaction, with control of 3 pH and without temperature control (only the effluent cooled to 20 ∘ C and in a still water bath).Percentage of degradation observed was lower.Hence, a temperature control is required.
Values were also considered for an estimated consumption of the following reagents as Box-Behnken modeled experiments: H 2 SO 4 (98% w/w), H 2 O 2 (30% w/w), NaOH (98% w/w), and FeSO 4 ⋅7H 2 O (98% w/w).Table 7 shows the amount of reagents in the Box-Behnken model used in the dairy effluent treatment process for Photo-Fenton process.
To calculate the final consumption of energy and reagents in relation to a cost/benefit ratio (lower is better), each experiment conformed to the Box-Behnken model with optimal time profiles in 3 L of effluent in reduced DOC concentrations as shown above and in Table 8.
Evaluation of the results in Table 8 concluded that the best qualitative result corresponds to the highest cost/benefit ratio.This is exemplified in experiments 5 and 8 of Box-Behnken model that obtained 79.50% degradation of the concentration of DOC at a cost/benefit ratio of 11.02.Another relevant factor is that the cost of reagents was higher in relation to energy for all experiments.
A separate analysis among several variables of the Photo-Fenton process and related cost/benefit provided two specific evaluations depending on the disposal of the treated effluent.According to national regulation, if the effluent is deposited immediately into rivers, the costs will differ from those of experiments 5 and 18, to present a revised cost/benefit ratio among intermediate experiments.
Specifically, if the effluent is treated by Photo-Fenton after biological treatment (e.g., hybrid AOP-activated sludge), the UV variable should be evaluated carefully as the most expensive process.Treatment with UV light should be the least significant factor in the process.Thus, the experiments 23 and 24 should be highlighted as the most advantageous cost/benefit ratio (9.10), with the second best value obtained from the experiments of the Box-Behnken model with pH at intermediate level, reagent at a low level, UV radiation at high level, and a reduced concentration at 76% DOC.

Conclusions
It was concluded that physical-chemical characterizations of the raw dairy effluent with the parameters N (158 mg L −1 ), phosphorus (108.7 mg L −1 ), BOD 5 (2300 to 2500 mg L −1 ), COD (9000-10000 mg L −1 ), and DOC (1313 to 1663 mg L −1 ) with concentrations above legal limits indicated that dairy wastewater must be treated and made suitable for disposal.It was found that the degree of recalcitrant organic matter present in the dairy effluent at BOD 5 /COD < 0.25 complicates the conventional biological treatment, which may justify the use of AOPs in pretreatment or treatment.An experimental design and response surface Box-Behnken matrix obtained a percentage DOC-reduction index of 81% and COD of the order of 90.7% after the photochemical treatment.In the design, these factors showed significant effects in reducing DOC.
Two aspects were addressed regarding the cost/benefit evaluation.First, if the effluent treated by AOP is released directly into surface waters (rivers), the variables and their levels of the process should be as follows: pH 3.5, 35.0 g H 2 O 2 /Fe 2+ 3.6 g, and 28 W UV, obtaining a reduction in concentration of 79.5% DOC.If the posttreatment effluent undergoing Photo-Fenton process is subsequently treated biologically, optimal experimental data shows that the cost/benefit ratio should include the following variables and levels: pH 3.5, 30 g H 2 O 2 /Fe 2+ 2.1 g, and UV irradiation at 28 W to achieve a reduction in concentration of 76% DOC.

9
Application.The Box-Behnken matrix optimizes the Fenton reagent pH values in relation to the L Taguchi array values obtained by Loures et al. [15] at a fixed temperature of 30 ∘ C.

Figure 2 :
Figure 2: Primary effects in relationship to mean responses (percentage reduction of DOC) of factors used in the treatment of dairy effluent as per the Box-Behnken matrix.

Table 1 :
Factors and respective levels used for the experiment matrix for the Box-Behnken in the treatment of dairy effluent by Photo-Fenton process.

Table 3 :
Results of physical/chemical analyses in natura and posttreatment dairy effluents.

Table 4 :
Percentage reduction variations of the DOC as per factors and levels to the array of Box-Behnken array for Photo-Fentontreated dairy effluent.

Table 5 :
Average percentage reductions of DOC, BOD 5 , and COD obtained in the treatment of dairy effluent by Photo-Fenton processes according to the Box-Behnken matrix for optimal experimental results (5).

Table 6 :
Energy consumed in Photo-Fenton reactions data from Box-Behnken experiments at a volume of 3 L of dairy effluent and 60 min of reaction.

Table 7 :
Amount of reagents used in Photo-Fenton processes as Box-Behnken modeled experiments, considering the volume of 3 L of dairy effluent and a 60 min reaction duration.

Table 8 :
Reagent energy consumption in 3 L of dairy effluent treated by the Photo-Fenton process with each experiment optimized for the Box-Behnken model.