ELSEVIER Desalination 162 (2004) 85-91

DESALINATION

Biological treatment of meat industry wastewater

Ewa Sroka a, Wtadyslaw Kamfliski b, Jolanta Bohdziewicz a* alnstitute of Water and Wastewater Engineering, Silesian University of Technology,

ul. Konarskiego 18, 44-100 Gtiwiee, Poland Tel. +48 (32) 237-1698; Fax +48 (32) 237-1047; email: jolaboh@zeus.polsl.gliwice.pl

bFaculty of Process and Environmental Engineering, Technical University of Lodz, uI. Wolezanska 215, 90-924 Lodz, Poland

Received 17 July 2003; accepted 11 August 2003

Abstract

The work aimed to determine the effectiveness of the treatment of wastewater generated by the meat industry in a hybrid system combining biological methods of activated sludge (in an SBR reactor) and reverse osmosis. The tests carried out on the wastewater from the meat processing plant Uni-Lang in Wrzosowa showed that the biological treatment resulted in sufficient removal of contaminants from the wastewater, which consequently could be discharged into receiving water. In order to make it possible for the wastewater to be reused in the production cycle, it was additionally treated with reverse osmosis. The research was described mathematically by the program MATLAB using artificial neural networks. The program enables a prediction of the results for the treatment ofwastewater over a range of tested values.

Keywords: Activated sludge; Reverse osmosis; Neutral networks; MATLAB; Meat industry wastewater

1. Introduction

Meat processing plants use approximately 62 Mm3/y of water. Only a small amount of this quantity is a component of the final product; the remaining part is wastewater of high biological

*Corresponding author.

and chemical oxygen demand, high fat content and high concentrations of dry residue, sedi- mentary and total suspended matter as well as nitrogen and chlorides. Since the wastewater contains substantial amounts of proteins, it putrefies easily and gives off nasty smells. It may also contain disease microorganisms, eggs of ascaris and intestinal parasites.

Presented at the PERMEA 2003, Membrane Science and Technology Conference of Visegrad Countries (Czech Republic, Hungary, Poland and Slovakia), September 7-11, 2003, Tatranskd Matliare, Slovakia.

0011-9164/04/$- See front matter © 2004 Elsevier B.V. All rights reserved pII: S0011-9164(04)00030-X

86 E, Sroka et al. / Desalination 162 (2004) 85-91

The contaminant loading of the wastewater discharged from meat processing plants varies seasonally, daily or even on a shift basis. In order to reduce wastewater contamination, the pro- duction cycles of the meat processing plants which are run properly deal with the separation and utilization of solid waste [1].

2. Experimental 2.1. Materials

The wastewater was sampled from the Uni-Lang meat processing plant in Wrzosowa (southern Poland) whose activity covers the slaughter and processing of pigs. It was charac- terized by considerable pollutant load, substantial amounts of suspended matter and high con- centrations of total nitrogen and phosphorus. The values of the basic and eutrophic pollution indexes ranged widely during the whole pro- duction cycle. The wastewater was red and brown in colour, smelled nasty and tended to foam and putrefy. The characteristics of the raw wastewater are presented in Table 1.

2. 2. Apparatus

Treatment of the wastewater was carried out biologically applying the activated sludge method

in a 40 drn 3 chamber. The chamber was equipped with two aeration pumps: Maxima R manu- factured by Elite, whose average capacity was 420 dm3air/h; and an RZR 2020 stirrer manu- factured by Heidolph, with an adjustable rotation velocity ranging from 40 to 2000 rpm.

Reverse osmosis (RO) was conducted in a high-pressure apparatus equipped with a plate and frame module produced by Osmonics, with an active membrane area of 155 cm 2. The system operated in the crossflow mode.

2. 3. Methods

The research consisted of the following two basic phases of wastewater treatment: biological treatment in an SBR and post-treatment applying RO. The activated sludge used during the biological treatment was taken from the bio- logical wastewater treatment plant of the Uni- Lang meat processing plant in Wrzosowa, which ensured that the bacterial microflora had already been adapted to the treatment of this type o f wastewater.

The tests were carried out at a constant dry weight of 5 g/din 3 in the chamber, aeration inten- sity 840 dm 3 air/h, residence time o f 12 h, and tSt, + ta of 0.3 (t,, stirring time; t~, aeration time).

Table I Pollution indexes of raw wastewater which, after the treatment, can be returned to the natural receiving waters

Pollution indexes Concentration of pollution in raw wastewater, mg/dm 3

Range Mean value

Load, kg/d Permissible (mean value) standards, mg/dm 3

COD 2780-6720 4584 309.2 150 BOD 5 1200-3000 2100 126.8 30 Total nitrogen 49-287 198 13 30 Total phosphorus 15-70 32 2.1 5 a Total suspension 112-1743 396 26.1 50 Detergents 7-21 11.3 0.75 5

~For a treatment plant of wastewater flow below 2000 m3/d.

E. Sroka et al. / Desalination 162 (2004) 85-91 87

These parameters were selected on the basis of previous research [2].

An attempt was also made to determine the most favourable loading of activated sludge for the wastewater tested. The measurements covered the range of 0.05-0.75 COD/ga~ ~ x d.

Subsequently, with the help of the MATLAB program and neural networks, we developed a mathematical model which enabled us to find a correlation between the loading of activated sludge in SBR and final degree of removal of contaminants from wastewater. The program calculated specific parameters: COD, total nitro- gen, and phosphorus on the basis of a given activated sludge loading (across the range of tested values).

In the last phase of the investigations, the wastewater treated biologically under optimum operating parameters of the activated sludge was additionally treated with RO, applying a flat com- posite DS3SC 1206366 membrane (Osmonics). The membrane process was conducted at a pressure of 2.0 MPa and linear flow velocity of 2.0 m/s.

The obtained results showed that the purified wastewater could be discharged into receiving water because it met the requirements of the Regulations of the Ministry of Environmental Protection, Natural Resources and Forestry, dated 5 November 1991, and none of the pollution indexes exceeded the permissible standards. The combination of activated sludge in the SBR bioreactor and RO enabled a reuse of the wastewater in the production cycle.

2. 4. Analytical procedures

COD, concentrations of phosphorus, and total ammonium and nitrate nitrogen were determined using an SQ18 photometer (Merck) [3], whereas BOD5 was assayed using OxiTOP measuring cylinders produced by WTW [4]. The dry matter of the sludge was determined by means of the gravimetric method [5].

3. Results and discussion

3.1. Biological treatment o f wastewater

The biological treamaent of wastewater from the meat industry aimed at decomposing organic matter and removing biogenic compounds, i.e., nitrogen and phosphorus. An increase in the loading of activated sludge in the bioreactor chamber brought about a decrease in contami- nant removal. COD depended on the changes in activated sludge loading over the range of tested values (0.05 gCOD/gdwtX d + 0.75 gCOD/gawtXd). Maximum COD removal was reached at the sludge loading of 0.05 gCOD/gdwt x d, and the raw wastewater had a value of 57.2 gO2/m 3. The lowest COD removal of 90.2% was achieved for the sludge loading of 0.75 gCOD/ga,~ x d. Its value for purified wastewater was 509.6 gO2./m 3.

It was also found that BOD 5 was not dependent on sludge loading. The removal per- eentage remained constant for all applied activated sludge loadings and ranged from 99.1% to 99.6%.

The research showed a strong dependence of biogenic compound removal in the purified wastewater on activated sludge loading. For total nitrogen, it reached the highest value of 98.2% for the activated sludge loading of 0.15 gCOD/ g ~ wt. siu~ex d . Whereas its concentration in the raw wastewater was 530 gN~t/m 3, in the purified wastewater it decreased to a leve 1 of 9.5 g Ntot/m 3. The lowest removal of total nitrogen was 82.1% (raw- wastewater 236 g Nto t/m3; purified waste- water 42.2 g N~t/m 3) for the loading of 0.75 g COD/g ~y~. ~l~x d.

The results for phosphorus removal showed that it depended mainly on the sludge loading. The highest removal of this biogene was achieved by applying an activated sludge loading of 0.15 gCOD/g~y wt. ~u~ x d. It amounted to 87.3% (purified wastewater 4.8 gP/m3), while the lowest degree of removal was obtained for a loading of 0.55 gCOD/gd~. ~ludg~x d, which was 59.4%. It is the equivalent of phosphorus removal

88 E. Sroka et al. / Desalination 162 (2004) 85-91

Table 2 Effectiveness of wastewater treatment applying the activated sludge method under the most favourable operating con- ditions (activated sludge loading: 0.15 gCOD/g ~.,luago x d, aeration intensity: 840 dm 3 air/h, residence time in the bio- reactor: 12 h, ratio t,/t~ + t~: 0.3)

Pollution indexes Raw wastewater, Retention, Wastewater after activated Permissible mg/dm 3 R, % sludge bioreactor, rag/din 3 standards, mg/dm 3

COD 5300 98.1 102 150 BOD 5 2900 99.6 10 15 Total nitrogen 557 98.2 9.5 30 Total phosphorus 37.8 87.3 4.8 5 Ammonium 2.0 95.0 0.1 6

in the purified wastewater to the level of 20.9 gP/m 3.

In further tests, the activated sludge loading of 0.15 gCOD/g ~wt sludge X d, regarded as the most favourable, was used and produced the best results in wastewater treatment. Table 2 shows the results obtained in the investigations.

An analysis o f the obtained results showed that the purified wastewater could be discharged into receiving water because it met the require- ments o f the Regulations of the Ministry o f Environmental Protection, Natural Resources and Forestry, dated 5 November 1991, and none of the pollution indexes exceeded the permissible standards.

There are multitudes of ways to describe research mathematically. We decided to choose artificial neural networks (ANNs) for their universality.

ANNs originated from an interdisciplinary synthesis o f the traditional sciences of biology, physics and mathematics. Their dynamic devel- opment took place only recently as a result of computational capabilities in computer science and electronics.

The basic characteristic which distinguishes ANNs from the programs using algorithmic data processing is the ability to draw generalizations from new data, unknown before, i.e., not intro- duced during learning. It can be also expressed as

EJ

Fig. 1. Schematic of neural networks.

Output: layer [

the ability of ANNs to approximate the values of the function of multiple variables. Another description is also possible, e.g., expert systems usually require a complete knowledge of the problems they are going to solve, whereas ANNs require a single learning process and are tolerant of imperfect data, distortions or even gaps in data sets. This makes it possible to use ANNs to solve problems when other methods fail.

In this work, three neural networks were used, each of them having input, hidden and output layers (Fig. 1). The sludge loading was the input layer, whereas COD, total nitrogen and phos- phorus in the purified wastewater were the output layer.

The number of hidden neurons, being grad- ually increased, was determined through compu- tation. Two hidden neurons appeared to be sufficient for the network to solve a given

E. Sroka et al. / Desalination 162 (2004) 85-91 89

Table 3 Initial data file

Sludge loading, COD Total Total gCOD/ removal, phosphorous nitrogen gaw~a~oxd % removal, % removal, %

0.05 98.9 87.8 95.5

0.10 98.4 85.0 94.1

0.15 98.1 87.3 98.2

0.20 98.2 82.4 89.7

0.25 98.1 69.6 90.4

0.30 97.7 66.8 87.5

0.35 97.3 65.0 88.5

0.40 97.4 62.9 87,6

0.45 95,7 64.1 87.4

0.50 94,3 62.9 87,2

0.55 94.1 59.4 86.4

0.60 92.5 62.3 84.8

0.65 91.8 62.0 84.4

0.70 90.9 61.4 84.3

0.75 90.2 59.9 82.1

problem. This number enabled the network, which is connected with a small amount of weights, to make generalizations. The network was trained on the basis of initial data (Table 3). Learning from a randomly selected system of weights was repeated 20 times memorizing the best result in terms of error mean square. The following training parameters were applied: maximum number of epochs in the learning process, 4000; permissible error SSE, 0.01; learning coefficient, 0.005.

The program applied enabled a determination of correlation between the loading of raw wastewater, treated in SBR, and final removal of contaminants from it. Each index had an error mean square calculated.

Figs. 2-4 show a correlation between sludge loading and removal of particular contaminants (COD, total phosphorus and nitrogen).

The developed artificial neural network enables a prediction of wastewater purification

99

98

97

- - e j ~

O

o ° 6~

62

' I 1 I

~2

~2

6,4 8,6 '8.7 6,6 S l u d g e l o a d i n g [ g C O D / d T : S . * d ]

Fig. 2. Correlation between sludge loading and COD removal, x: measurement points, o: points learnt by the network { [A'-datal C,2)]'* [A'-datal C,2)1 }/15; error mean square for COD = 0.0460.

8~

E

E 75

z a

-t ','.> ~> x

Sludge loading [gCOD/gT.&*d] 8;7

Fig. 3. Correlation between sludge loading and phos- phorus removal, x: measurement points, o: points learnt by the network { [A'-datal (:,3)]'* [A'-datal(:,3)] }/15; error mean square for phosphorus = 1.0667.

180

"~ ~4

E 52

Z

8A

82

¢) <> 6,

8,7 S l u d g e l o a d i n g NCOD/gT.S.*d]

Fig. 4. Correlation between sludge loading and total nitrogen removal, x: measurement points, o: points learnt by the network. {[A'-datalC,4)]'*[A'-datalC,4)]}/15; error mean square for total nitrogen = 0.9517.

90 E. Sroka et al. / Desalination 162 (2004) 85-91

over the range of tested values. Entering a specific sludge loading results in a prediction of COD, total nitrogen and phosphorus removal. Thus, it is possible to predict the extent of wastewater purification.

Table 4

Effectiveness of wastewater treatment through RO after it pretreatment using the biological method

Pollution Wastewater after Wastewater after indices activated sludge RO process

bioreactor

3.2. Additional treatment ofwastewater by means of pressure driven membrane operations

As the results of the investigations showed, wastewater from the meat industry can be puri- fied only to the extent which enables the waste- water to be discharged into receiving water. Since the meat industry uses huge quantities of water, as mentioned herein, and thus produces highly loaded wastewater, an attempt to treat it additionally so that it could be reused in the production cycle was made. In our work, we determined a correlation between the volume permeate flux and its recovery (Fig. 5).

It was found that during RO, the volume permeate flux depended on its recovery to a small extent. At a permeate recovery of 20%, it decreased to 1.67<10 -6 m3/m 2 x s, i.e., by a mere 10.6%.

The effectiveness of additional treatment of wastewater during RO after it was treated bio- logically applying the activated sludge method is given in Table 4.

% ×

eo ~j~

>.

2.5

2 .........................................................................................................................................

III water i 0.5 .................................................................................... A RO "cross-flow" ' ]

/ 0

0 5 10 t5 20 Permeate recovery, degree [%]

Fig. 5. Dependence of volume permeate flux on its recovery for reverse osmosis of wastewater after traditional treatment.

Cone., Conc., Retention, rag/din 3 rag/din 3 %

COD 76 10.8 85.8

BOD 5 10 5.0 50.0

Total 3.6 0.09 97.5 phosphorus

Total nitrogen 13 1.3 90.0

The degrees of contaminant removal obtained during RO were as follows: phosphorus was removed to a value below 0.1 gP/m 3, the con- centration of total nitrogen was 1.0 gNog/m ~, COD and BOD s were relatively low - - 10 g 02/m 3 and 5 gO2/m 3, respectively. It was con- cluded that the purified wastewater could be then reused in the production cycle of a plant.

4. Conclusions

Wastewater from the meat industry is very difficult to purify due to its specific charac- teristics; irregular scatter; and considerable amounts &organic, mineral and biogenic matter. This type of wastewater can be treated bio- logically by means of activated sludge applying a low sludge loading of 0.15 COD/g ~y~. ~lu~e xd, aeration intensity of 840 dm 3 air/h, constant sludge concentration in the chamber of 5 g/dm 3, residence time ofwastewater in the bioreactor of 12 h and a ratio of the stirring time to the sum of the stirring and aeration times of 0.3. The wastewater thus treated meets the requirements of the Regulations of the Ministry of Environ- mental Protection, Natural Resources and Forestry, dated 5 November 1991, and can be discharged into receiving water.

E. Sroka et aL / Desalination 162 (2004) 85-91 91

Wastewater from the meat industry can also be satisfactorily treated so that it can be reused in the production cycle o f a plant. In order to achieve this degree of wastewater purification, a hybrid process combining the biological method of activated sludge and RO should be used.

References

[1] B. Koziorowski and J. Kucharski, Industrial waste- water, WNT, Warsaw 1980.

[2] J. Bohdziewicz, E. Sroka and E. Lobos, Application of the system which combines coagulation, activated sludge and reverse osmosis to the treatment of the wastewater produced by the meat industry, Desali- natior% 144 (2002) 393-398.

[3] User's Manual, Photometer SQ 118, Merck. [4] User's Manual, Determination of BZT using respiro-

metric method, Oxi Top, WTW. [5] W. Hermanowicz, ed., Physicochemical Testing of

Water and Sewage, Arkady, Warsaw, 1998.