Cleaner production option in a food (Kimchi) industry

Journal of Cleaner Production 9 (2001) 35–41www.cleanerproduction.net

Cleaner production option in a food (Kimchi) industry

Hosang Yi, Jaehong Kim, Hoon Hyung, Sangho Lee, Chung-Hak Lee*

College of Engineering, Department of Chemical Technology, Seoul National University, Kwanak-Gu, Shinlim-Dong 56-1,Seoul 151-742, South Korea

Received 12 October 1999

Abstract

In the Kimchi (a salt-pickled and fermented food) manufacturing industry, the process of brining and rinsing the raw vegetableproduces a vast amount of wastewater of high salinity. Instead of the expensive and low-efficient conventional treatment system,a brining wastewater reuse system was developed using hybrid chemical precipitation/microfiltration. In the microfiltration of chemi-cally treated brining wastewater, a comparison of flux, backwashing frequency and energy consumption was made between dead-end and crossflow filtration modes. The optimum location of the neutralization step in this system was also discussed in connectionwith the microfiltration performance. The quality test ofKimchi prepared by the reuse system confirmed the new approach to thecleaner production option was successful in terms of water/raw material (salt) savings and wastewater reduction. 2000 ElsevierScience Ltd. All rights reserved.

Keywords:Brining wastewater; Chemical precipitation; Dead-end filtration;Kimchi; Microfiltration; Wastewater reuse; Cleaner production

1. Introduction

Cleaner production has been spreading to almostevery industrial sector and many researchers havedevoted themselves to the development of clean techno-logies in the food industry [1].

Kimchi, a salt-pickled and fermented vegetable, is oneof the most popular foods in Asian countries, such asKorea, and more and more people in the world get usedto its taste. There are more than 200Kimchi manufactur-ing industries in South Korea and the annual productionof Kimchi reached about 2.1 million tons in 1997. Inthe Kimchi manufacturing industry, the brining of rawmaterials (e.g. Chinese cabbage) and successive rinsingof brined raw materials generate a large amount of brin-ing wastewater and rinsing wastewater, respectively. Tocomply with the effluent discharge limit the two kindsof wastewater are mixed together and passed through aconventional end-of-pipe wastewater treatment system,

* Corresponding author. Tel.:+82-2-880-7075; fax:+82-2-874-0896.

E-mail address:[email protected] (C.-H. Lee).

0959-6526/01/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.PII: S0959-6526 (00)00029-9

presented in Fig. 1. However, these multiple treatmentsteps are not only very complicated but also unfavorablefrom an economic point of view. Moreover, this systemsuffers from some severe problems especially in the acti-vated sludge process due to the high salinity of the brin-ing wastewater [2]: (1) difficulties in biological floc for-mation; (2) reduction of biochemical oxygen demand(BOD) removal efficiency; (3) poor settling character-istics of biological flocs, etc. Furthermore, loss of reus-able salts and process water is inevitable in this end-of-pipe treatment system. The purpose of this study was todevelop a new clean technology in theKimchi industrythrough the internal recycling of brining wastewater,which enables a zero-discharge system at least at thebrining step as well as salt and water recovery.

1.1. Brining wastewater recycle through hybridchemical precipitation/microfiltration system

The schematic diagram of brining wastewater reusesystem developed in this study is shown in Fig. 2, whichconsists of chemical precipitation and microfiltration.For the brining wastewater recycling, hybrid microfil-tration with chemical precipitation was applied toremove the organics responsible for tastes and odorproblems from the brining wastewater. Because sun-

36 H. Yi et al. / Journal of Cleaner Production 9 (2001) 35–41

Fig. 1. Flow chart for the conventional wastewater treatment.

Fig. 2. Schematic of a brining wastewater recycle system.

dried bay salt is usually used to make the brining water,about 1800 ppm of magnesium free ion remains in thebrining wastewater. Accordingly, the addition of sodiumhydroxide generates magnesium hydroxide precipitateswhich can easily adsorb or enmesh both the organic andcolloidal impurities including microorganisms. Theeffluent from the chemical precipitation tank was furtherpurified by the membrane process (microfiltration) priorto the recycle to the brining process. The concentratesfrom microfiltration (MF) is returned to the storage tank

and combined with the brining wastewater. As the pHof the supernatant in the settling tank is more than 10,it is compulsory to neutralize the chemically treated brin-ing wastewater prior to its reuse. In this study, it wasintended to find out the optimal process option and oper-ational conditions in the MF/precipitation hybrid systemwith respect to energy consumption, quality of recycledwastewater and taste of the final product,Kimchi.

37H. Yi et al. / Journal of Cleaner Production 9 (2001) 35–41

2. Materials and methods

2.1. Characteristics of brining wastewater

Brining water becomes turbid and malodorous as thebrining steps are repeated, because of the accumulationof debris of raw materials and growth of micro-organisms. As shown in Table 1, total organic carbon(TOC) and the turbidity of the brining wastewater usedfive times were much higher than those of fresh briningwater. Therefore, the major components to be removedprior to the reuse of brining wastewater were thought tobe the dissolved and suspended organic matters such asthe debris of vegetables generated and microorganismsgrown and their metabolites produced in the brining pro-cess. In this context, the brining wastewater reuse systemwas designed to effectively remove such organic matters.On the other hand, inorganic constituents in the briningwastewater, such as calcium and magnesium, wereintended to be kept as much as possible even after thetreatment as these are known to play a key role ininducing the delicate taste ofKimchi.

2.2. Chemical precipitation

As a first step, chemical precipitation was applied toremove the dissolved and colloidal materials thataccumulated during the brining steps and thus to relievethe load on the following purification step, microfil-tration. Chemical precipitation was carried out in a 250-l precipitation tank as well as a 1-l jar. It consisted ofthe injection of 2 M NaOH solution into the brining was-tewater, 3 min of fast mixing, 27 min of slow mixingand 1-h sedimentation steps. Changes in TOC, sus-pended solid (SS), turbidity, retention of soluble mag-nesium and sediment volume were analyzed after eachexperiment.

2.3. Operation of microfiltration

After chemical precipitation, the supernatant was fedto the microfiltration system. The microfiltration system(Memcor 1M1, Memtec, Australia) was used to removecake-forming and pore-plugging foulants by high-press-

Table 1Water quality of fresh brining water and brining wastewater

Parameter Fresh brining Briningwater wastewater

Total dissolved solid (ppm) 95 700 112 000Suspended solid (ppm) negligible 160Total organic carbon (ppm) 10 340–640Turbidity (NTU) 6 120Ca2+ (ppm) 160 142Mg2+ (ppm) 1860 1880

ure air backwashing. The membrane was a poly-propylene hollow-fiber membrane (active area, 1 m2). Inoperating the MF unit, both high-pressure air (600 kPa)and low-pressure air (120 kPa) were applied to controlthe pneumatic valves in the machine for backwashingand to remove the water filled in the lumen at back-washing, respectively. To supply high-pressure clean airto the MF unit, an air compressor with a 75-l air tankwas used, to which two air filters of 40µm and 5µmwere connected in series.

The crossflow microfiltration (CMF) was operated ina total recycle run by recycling permeate, retentate andbackwashed solution to the feed tank and the back-washing was conducted periodically during operationwhen the trans-membrane pressure increased by 6–7kPa.

As it is possible to operate the MF unit in both dead-end and crossflow filtration, the optimal operation con-ditions were determined by comparing permeate quality,flux, backwashing interval and energy consumptionunder the various operation conditions in this study. Forthe direct comparison of performance between dead-endand crossflow filtration, the initial condition was fixedat 1100 l/h of retentate flow rate and 180 l/h of permeateflow rate with pure water.

2.4. Analysis

The analytical methods fromStandard methods[3]were adopted for the measurement of total organic car-bon (TOC), suspended solid (SS) and turbidity. Analysesof inorganic ions such as Mg2+ and Ca2+ were performedby the direct injection of aqueous samples to an ion chro-matograph (Dionex 4500I, USA) equipped with a sensi-tive conductivity detector.

3. Results and discussion

3.1. Chemical precipitation of brining wastewater

Given that the brining wastewater inherently con-tained lots of soluble magnesium, chemical precipitationto induce the precipitation of magnesium hydroxide wasattempted by raising the pH of the wastewater throughthe addition of sodium hydroxide. It was known that theprecipitation of magnesium hydroxide during lime-softening processes was effective for the removal ofviruses [4], color [5], organic matters [6,7], and so on.The optimal dosage of sodium hydroxide was estab-lished by the jar-test experiment. Taking into accountthe characteristics of the food product, the applicationof polymeric or inorganic coagulants was discounted.

The results of the jar-test experiment are summarizedin Fig. 3. Removal of TOC, SS, turbidity and the amountof sediment volume increased as the NaOH dose


Fig. 3. Jar-test results: removal of (a) total organic carbon (TOC), (b) turbidity, (c) suspended solid (SS) and (d) SS after elimination of dissolvedsalts, (e) recovery of magnesium, and (f) sediment volume with respect to NaOH dose.

increased. But the recovery of soluble magnesium,which was known to be a critical component in the delic-ate taste ofKimchi and thus desired to get back as muchas possible, did not change vary much. The optimal dos-age of NaOH was estimated to be about 14 ml of 2 MNaOH per liter of the brining wastewater, although it

was quite difficult to determine the precise optimal dos-age.

3.2. Microfiltration of brining wastewater afterchemical precipitation

Microfiltration of the effluent from the precipitationtank was necessary to further purify the chemically


treated wastewater to be suitable for the reuse at the brin-ing step. As the pH of the effluent was 10, it should alsobe neutralized for reuse. However, the selection of theneutralization point in this recycling system, e.g. beforeor after MF, would be closely related to the MF perform-ance. In this context, two kinds of feeds with pH=10(non-neutralized) and pH=7 (neutralized) were appliedto MF, respectively, and comparison was made withrespect to flux, backwashing interval and permeate qual-ity. Fig. 4 shows the results of the experiments with thetwo feeds in both dead-end and crossflow filtrationmodes.

With the dead-end filtration the flux was about twotimes greater but the backwashing interval was less thanthose with the crossflow filtration regardless of the feedpH. The difference in flux between dead-end andcrossflow filtration was attributed to the difference intrans-membrane pressure. As the pump in the CMF unitwas a centrifugal type, the retentate flow rate was con-trolled only with the back-pressure valve. Therefore, asthe retentate flow rate (or feeding rate in the dead-endmode) decreased, the trans-membrane pressure increasedand thus higher flux was obtained. However, the shorterbackwashing interval was observed in the dead-endmode due to the thicker cake layer formed on the mem-brane. In summary, the lower the retentate flow rate, thehigher the trans-membrane pressure, the higher the fluxbut the shorter the backwashing interval.

The quality of permeate was monitored and is shownin Table 2. The suspended solid was removed com-pletely and turbidity removal was over 99.8%, whileTOC removal was around 30%. The permeate TOC andturbidity were almost similar in the two operationmodes, but the dead-end mode was a little better thanthe crossflow one in terms of turbidity.

Fig. 4. Comparison of flux according to the operation mode: (a) non-neutralized feed (pH=10); (b) neutralized feed (pH=7).

Table 2Comparison of permeate quality according to the operation mode

Item TOC (ppm) Turbidity Suspended(NTU) solid (SS)

Feed 629.4 10.0 37.5Permeate Dead-end 418.1 0.08 Not detected

Crossflow 419.5 0.22 Not detected

As the magnesium concentration in brining water isknown to affect the taste of the final product (Kimchi),the lower the magnesium removal, the better the briningwastewater reuse system. Through the hybridMF/chemical precipitation, the loss of magnesium wasonly 10–15% resulting in no significant difference in thetaste ofKimchi.

3.3. Particle size distributions in various MF feeds

The SS concentrations of non-neutralized and neu-tralized feed are 37.5 and 27.5 ppm while turbidities are10.0 and 9.1 NTU, respectively. Therefore, it wasexpected that the non-neutralized feed would result ingreater membrane fouling than the neutralized feed. Asseen in the backwashing intervals (Fig. 4), however, theneutralized feed showed worse filtration performancethan the non-neutralized feed in both MF modes. Areasonable interpretation of these unexpected results wasmade by analyzing particle size distributions of the twofeeds, presented in Fig. 5.

As shown in Fig. 5, most of the particles of size over10 µm in raw brining wastewater were removed throughthe chemical precipitation process. After the chemicalprecipitation, the neutralization of the supernatant still


Fig. 5. Particle size distributions of various feeds for microfiltration.

caused the fraction of larger particles (30–40µm) todecrease while the fraction of smaller particles (around1 µm) increased. This is probably because the suspendedmagnesium hydroxide crystals that had not settled in theprecipitation tank were redissolved after being neu-tralized. During this process, the adsorbed organic mat-ters on the magnesium hydroxide crystals would also bedesorbed. It is well known that smaller particles causemore serious membrane fouling than larger particles inthe membrane process according to the Carman–Koz-eny equation:

J5e3d2

pDP180(1−e)3h

(1)

In this equation, the flux (J) is proportional to the squareof the particle size (dp). Therefore, the reason why theneutralized feed demanded more frequent backwashingto maintain a certain range of trans-membrane pressurecompared with the non-neutralized one is that neutraliz-ation made the particles smaller in the MF feed.

3.4. Comparison of energy consumption

As described previously, the dead-end filtration modehad the higher flux but the shorter backwashing intervalthan the crossflow mode. As an alternative criteria forselecting the better MF mode, the energy supplied tothe pump and compressor to produce the unit volume ofpermeate was evaluated according to the MF mode andcrossflow rate.

From the basic Newton’s law, the power (Pp) thatpushes the incompressible fluid from the pump can beexpressed by multiplying the feeding flow rate (v) by thepressure drop (DP), because water can be considered asan incompressible fluid. In this study, the feeding flowrate and the trans-membrane pressure changed as the fluxdeclined and thus the power changed with the operationtime. Therefore, the energy used can be obtained by inte-grating the power with respect to time and the amount

of total permeate can also be obtained by integrating theflux (J) with respect to time. Considering the timerequired for backwashing (tb) conducted during oper-ation, the net time fraction, (tt2tb)/tt, wherett is the totaloperation time, was applied to Eq. (2). And then theamount of energy (Ep) consumed by the pump to pro-duce the unit volume of permeate can be evaluated byEq. (3), where 0.3 indicates the energy efficiency ofthe pump.

Pp(kW)5v (m3/s)3DP (kPa) (2)

Ep (J/m3)5

Ev(t)×DP(t) dt × Stt−tbttD

0.3 × EJ(t) dt × Stt−tbttD

(3)

In an air compressor, a piston compresses the air ina closed space instantaneously, so it could be consideredto be an adiabatic compression. The power (Pc) of thepiston action is given in Eq. (4)[8] and the energy con-sumption (Ec) by the air compressor to produce the unitvolume of permeate is given in Eq. (5), where the energyyield, defined by the fraction of energy supplied to theair compressor to the actual piston action, is 0.66 (valuesupplied by the manufacturer of the air compressor).

Pc (kW)5k

k−13

ZRT1

98063FSP2

P1D(k−1)/k

21G (4)

whereP1 is the inlet pressure,P2 the outlet pressure,Zthe mass flow rate of air,T1 the inlet air temperature,Rthe ideal gas constant, andk the constant reflecting thedifference from ideality of the gas (for airk is 1.395).

Ec (J/m3)5

kk−1

×ZRT1

9806× FSP2

P1D(k−1)/k

−1G × tc

0.66× EJ(t) dt × Stt−tbttD

(5)

wheretc is the real operation time of the air compressor.All of the results involved in calculating the energy

consumption are summarized in Table 3.

Table 3Energy consumption of MF unit (in kJ/m3) to produce the unit volumeof permeate

Operation mode Pump (Ep) Air compressor Total(Ec)

Dead-end 115.3 1274.0 1389.3Crossflow (0.6 m3/h) 348.5 1289.8 1638.3Crossflow (1.1 m3/h) 624.4 1241.4 1865.4Crossflow (1.3 m3/h) 730.3 1246.6 1946.9


Table 4Quality test ofBak-kimchibrined with regenerated brining watera

Number of Before fermentation After fermentationbrining

Taste Odor Taste Odor

1st NR NR NR NR2nd NR NR NR NR3rd NR NR NR NR4th NR NR NR NR

a NR=no response by connoisseur, which means no differences intaste and odor betweenBak-kimchibrined in fresh brining water andbrined in the recycled brining wastewater.

The energy consumption for pumping varied consider-ably with the operation conditions, but the energy sup-plied to the air compressor was almost constant regard-less of operational conditions and made up 60–90% ofthe total energy consumption in this study. The totalenergy consumed to produce the unit volume of per-meate in the dead-end filtration was less than that in thecrossflow filtration but this difference was brought aboutentirely by the energy consumption for the pump. If theenergy supplied to the air compressor were reduced com-paratively, the difference in the total energy consumptionwould be much larger according to the operation con-ditions.

3.5. Quality test of Bak-Kimchi prepared withrecycled brining wastewater

The effect of the recycling system on the productquality was tested by brining raw materials with therecycled brining wastewater and manufacturingBak-kimchi, a kind of Kimchi whose taste and odor wereknown to be the most sensitive to the quality of a briningsolution. Taste and odor tests by professional con-noisseurs ofKimchi proved the recycling system wassuccessful as shown in Table 4.

4. Conclusions

In this study, the hybrid MF/chemical precipitationwas developed for the brining wastewater recycle in afood (Kimchi) industry. The following conclusions couldbe drawn:

1. The addition of NaOH into the brining wastewatergenerated magnesium hydroxide precipitates whicheffectively removed the soluble organics and colloidalparticles, but the recovery of valuable magnesium

inducing the peculiar taste ofKimchi reached morethan 85%.

2. In MF of the effluent from the precipitation tank, thenon-neutralized feed was better than the neutralizedone in both dead-end and crossflow modes in termsof backwashing interval. This was attributed to theshift of the larger particles to the small ones due tothe redissolution of magnesium hydroxide precipitatesduring neutralization. The dead-end mode wassuperior to the crossflow one with respect to flux andenergy consumption.

3. The quality test of the product (Kimchi) prepared withthe recycled brining wastewater confirmed that thehybrid MF/chemical precipitation system was so suc-cessful that it could add another example to the world-wide cleaner production cases.

Acknowledgements

This work was supported by the Ministry of theEnvironment, South Korea. The authors wish to thankthe Doosan Food Industry, South Korea, for supplyingraw brining wastewater and the quality test forKimchi.The authors also thank J.S. Yeom and H.S. Lee for theirassistance in the analysis of samples at the Institute ofEnvironmental Science and Engineering, SeoulNational University.

References

[1] Amundsen A. Joint management of energy and environment. Cle-aner production, environmental management systems and energymanagement facilitating sustainable industrial development in thefood processing industry. PhD thesis within the international PhDprogram in cleaner production, cleaner products, industrial ecologyand sustainability, Centre for Environmental Studies, ErasmusUniversity, Rotterdam, The Netherlands, and Ostfold ResearchFoundation, Frednkstad, Norway, 1999.

[2] Yi H. Comparison of filtration characteristics and efficiencybetween dead-end and crossflow microfiltration. Master Thesis,Department of Chemical Technology, Seoul National University,South Korea, 1998 [in Korean].

[3] APHA, AWWA, and WEF. Standard methods for the examinationof water and wastewater. 19th ed. Washington (DC): APHA, 1995.

[4] Wentworth DF, Thorup RT, Sproul OJ. Poliovirus inactivation inwater-softening precipitation processes. J AWWA 1968;60:939–46.

[5] Rush RJ, Oldham WK. Color removal in Kraft Mill wastewaterusing recycled magnesium. J WPCF 1978;50:875–95.

[6] Thompson CG, Singley JE, Black AP. Magnesium carbonate—arecycled coagulant. J AWWA 1972;64:11–9.

[7] Weber WJ, Jodellah AM. Removing humic substances by chemicaltreatment and adsorption. J AWWA 1968;60:939–46.

[8] Perry RH, Green D. Perry’s chemical engineers’ handbook. 6thed. McGraw-Hill, 1984.

Documents

Cleaner production option in a food (Kimchi) industry