ICEF9-2004 International Conference Engineering and Food

PURIFICATION OF FOOD STREAMS BY COMBINING ION EXCHANGE AND MEMBRANES TECHNOLOGIES: APPLICATION OF DECALCIFICATION IN THE WHEY INDUSTRY

Marc-André THEOLEYRE, Francis GULA APPLEXION, [email protected] ; [email protected]

Abstract

The presence of divalent ionic species in many food streams are a drawback for further processing steps such as membrane filtration, crystallisation and chromatographic ion exclusion. Cationic ion exchange resins operating in Na or K form have been widely used for eliminating the divalent ions for which they present a high affinity. However the presence of complexing agent reduces the decalcification yield. Our study has demonstrated that a combination of a strong base resin and a strong acid cation resin regenerated in series with a monovalent salt solution can improve the decalcification process in whey products. Used prior to the nanofiltration step, the new decalcification process permits to produce a highly demineralised whey, up to 70% whereas without such a pre-treatment only 30% or 45% with diafiltration are obtained. Furthermore additional benefits are anticipated either in the production of lactose after crystallisation or by downsizing the demineralisation unit.

1. Introduction

The dairy industry produces large amount of whey, the bulk of it coming from the cheese or casein making industries, according to the process different types of whey are produced. In order to be used in the food industry (lactose production, WPC, demineralised powders,…) they have to be further processed. The main available techniques are ultrafiltration, reverse osmosis, nanofiltration, electrodyalis, ion exchange or a combination of several of them. Nowadays nanofiltration is a common practice as a pre concentration step before any further processing. Nevertheless the presence of calcium in the feed stock remains detrimental to the process performances, therefore we have studied the possibility to implement a decalcification system.

2. Description of the current status

The whey streams obtained from the cheese or casein production have a complex composition, the main components being proteins, lactose and ashes, the later being mostly calcium, potassium, chlorides, phosphates and lactic acid and a dry solid contents of 5 to 6% (table 1 & 2).

Table 1: Typical whey composition Table 2: Typical mineral composition (mg/l) Sweet whey Acid whey Sweet whey Acid whey pH > 6 4.6 Calcium 550 1020 Dry solid (g/l) 65 65 Magnesium 60 90 NPN (g/l) 2.4 2.0 Sodium 470 470 Lactose (g/l) 47 48.6 Potassium 1440 1440 Lactic acid 1 5 Chlorides 1100 4200 Vitamins (mg/l) 7.5 7 Phosphates 470 640 Proteins (g/l) 6 6 Ashes (g/l) 5 7

Nanofiltration is widely used for the pre concentration of whey whatever are the next processing steps such as lactose or demineralised whey production. Nanofiltration also provides a partial demineralisation of the feed whey, the demineralisation rate without any special pre-treatment is about 30%, using diafiltration we can achieve up to 50% but the presence of divalent cations prevents the nanofiltration process to reach higher values. The table 3 illustrates the high retention rate of divalent species, which are retained by the NF membranes whereas monovalent ions are passing through easily. When lactose is manufactured, high calcium contents in the retentate cause precipitation problems during the concentration and the crystallisation of the whey. Calcium presence is also detrimental to the lactose quality.

1/6 ICEF9-2004 International Conference Engineering and Food

Table 3: Retention rate for various salts Table 4:Typical IX sequences Muchetti & al(2000) Step Quantity Type NF 50 ; 25°C ; 15 bar TMP Production Feed Sweetening off 1 to 2 BV Sweet water Salts Retention Backwashing 1 to 2 BV Water -1 NaCl (0.1 mol.l ) 0.25 Regeneration 1 to 2 BV Acid, salt or caustic -1 CaCl2 (0.1 mol.l ) 0.60 Displacement 1 to 2 BV water -1 MgSO4 (0.1 mol.l ) 0.95 Fast rinse 2 to 5 BV water -1 Na2SO4 (0.1 mol.l ) 0.99 Sweetening on 1 to 2 BV Sweet water

The standard process used for the production of highly demineralised whey typically consists in a nanofiltration unit followed by a demineralisation unit. An ion exchange resin cycle includes several steps (table 4) each of them consuming important quantities of water or regenerant therefore it is essential to optimise all these flows in order to improve the economical aspect of the demineralisation process.

3. The scope of our work

The purpose of this work is to evaluate the implementation of a whey softening process operating in a chemical free mode, the salts from the whey itself being used as regenerant. Such a process will considerably improve lactose crystallisation and will allow to achieve high demineralisation rate of whey on improving nanofiltration performances through a combination of IX and NF techniques while optimising regenerant consumption.

Decalcification with ion exchange resin: This is a well known process, the cationic resin exchange its monovalent cations against the divalent species for which it has a higher affinity. The exchange can be illustrated with the following equations:

Exchange cycle: 2+ + 2(RSO3-Na) + Ca → (RSO3) 2-Ca + 2 Na Regeneration cycle: + 2+ (RSO3) 2-Ca + 2 Na → 2 (RSO3-Na) + Ca

Ca . Na 2 Ca []resin []feed The affinity coefficient KNa = 2 depends on the type of ion exchange resin, typical [][Ca feed. Na]resin Na values in an aqueous solution for a standard gel type cation resin are the following K H = 1.56 Ca Mg K ; K H = 4.06 ; K H = 2.59 , K H = 1.72 . These affinity coefficients illustrate the higher selectivity for divalent species compared to monovalent for a cationic resin.

However this process requires an excess of monovalent ions for reversing the reaction. In a standard decalcification process the NaCl regeneration rate required is about 200 to 300 %.

Moreover the divalent ions may form complexes with phosphates, organic acids,.. Thus they cannot be fully eliminated by the resin. This results in a very low operating capacity, such a drawback can be partly solved by using weak acid or chelating resins which presents a higher affinity for divalent. However weak acid resins can exchange ions only when the feed solution has a pH value higher than the one of the resin (typical pK of WAC 5.0 to 6.0) which is not the case of some acid whey; chelating resins might be envisaged however both types require concentrated acid for regeneration moreover they may present some regulatory issues.

In order to achieve a higher divalent ions removal which are incompletely eliminated through the previous options we have used a strong base resin operating in chloride form (figure 1) in front of the cationic unit. The strong base resins have the property to exchange all anionic species including those affecting the divalent ions removal. The strong base resins show the following selectivity pattern:

Citrate>tartrate>phosphate>bisulfate>nitrate>nitrite>chloride>hydroxide

2/6 ICEF9-2004 International Conference Engineering and Food

Therefore the troublesome organic acids will be preferentially fixed, including phosphates using the following design

Monovalent salts

SBA SAC (Cl) (Na Feed or K) whey

To NF

Figure 1: Improved softening process

4. Experimental results

The testing results have been summarized in table 5, we can notice that this resin combination considerably improve divalent removal (92%) and cycle length while applying a standard, in series NaCl regeneration procedure. With the same regeneration rate, the capacity was much higher than with the conventional single step process. We have now to consider an optimised regeneration process resulting in a minimum or almost no chemical consumptions. In order to regenerate in series these units, we need a monovalent salt solution with a certain concentration; this can be obtained from various sources depending on the factory design such as • Mother liquor from the lactose crystallisation unit • Raffinate from the lactose recovery through ion exclusion chromatographic plant • Demineralisation plant effluent after neutralisation • NF permeate

As an example, mother liquor has been tested as regenerant stream. The efficiency was close to salt solution, similar softening effect with a similar regeneration rate.

Table 5: Comparison in between the two decalcification processes for sweet whey Design SAC SBA + SAC SBA + SAC Regenerant Salt Salt Mother liquors Cycle (BV) 26 35 25 % Ca + Mg removal 77 92 88 Operating capacity (eq/l) 0.50 0.80 0.58 Ionic load (eq/l) 0.65 0.88 0.65 Regeneration level (eq/l) 2.4 2.4 1.7 Regeneration rate (%) 480 300 290

Following main conclusions can be drawn from the table 5: • The standard softening processing gives poor results: low capacity, high regeneration rate. The presence of phosphates and some organic acids (citric, lactic) decreases the removal of divalent species by complexing them. • By using a SBA resin in front of the SAC unit, we have been able to considerably improve the divalent ions removal, higher operating capacity has been achieved, lower leakage and lower regeneration rate. The regeneration rate has been reduced to such a level that mother liquors from the crystallisation unit contain a sufficient amount of monovalent salts for regenerating the SAC. Thereby, no additional chemicals are required.

At that stage, we have looked at the impact of the permutation of divalent/monovalent species on the demineralisation rate of the downstream nanofiltration unit, the results are illustrated in the table 6.

3/6 ICEF9-2004 International Conference Engineering and Food

Table 6: Impact of the pre-treatment on the NF performances for two types of raw material Concentration VCF : 3, Diafiltration : 3 l water/l retentate Acid whey Sweet whey D.S. Inlet salts Demineralisat° D.S. Inlet salts Demineralisat° % (eq/Kg DS) rate (%) % (eq/Kg DS) rate (%) Feed 6.7 1.78 4.2 1.66 NF Concentration 18.0 1.40 21.4 11.6 1.09 34.4 Diafiltration 17.0 1.16 34.7 11.0 0.73 56.2 SAC+NF Concentration 16.0 0.84 52.9 - - - Diafiltration 14.0 0.40 77.4 - - - SBA+SAC+NF Concentration 15.0 0.68 61.5 11.4 0.91 45.2 Diafiltration 14.0 0.15 91.8 10.8 0.48 71.0

In that experiment, the feed whey solution has been concentrated 3 times on the NF system and a diafiltration sequence has been added using 3 litre of water per litre of retentate; the test was run in a batch mode. Surprisingly no significant effect on the membranes flux has been noticed in any experiment, however a drastic difference has been observed regarding the demineralisation rate. The main monovalent and divalent ionic species have been measured by capillary electrophoresis, the salt concentration is an average. The results are detailed in table 6 from which hereafter comments can be made: • The demineralisation rate noticed with acid whey is on the low side compared to other published data for NF alone, the one with sweet whey is more in line with what is commonly observed. • For both types of whey, the new decalcification system provides a tremendous improvement of the overall demineralisation rate (up to 70% or 90% with diafiltration).

5. Complete process integration

The integration of the decalcification can be implemented without increasing the reagent consumption on coupling these unit operations either in the production of highly demineralised whey (figure 3) or lactose (figure 4).

Whey

Regenerant Decalcificat° waste

REVERSE DECALCIFICATION OSMOSIS

NANOFILTRATION

Water HCl NaOH

DEMINERALISATION DEMINERALISATION ANIONIC CATIONIC

Demineralised whey

Figure 3: Integrated effluent free process design for the production of fully demineralised whey

4/6 ICEF9-2004 International Conference Engineering and Food

Whey

Regenerant Decalcificat° waste

DECALCIFICATION

CONCENTRATION

Mother liquor CRYSTALLISATION

Lactose

Figure 4: Integrated effluent free process design for lactose production

6. Conclusions

The combination of an improved decalcification process consisting in coupling a strong base resin with a strong acid cation resin regenerated in series with a monovalent salt solution, allows to achieve a high softening rate event with sweet whey. This new process prevents the scaling in downstream concentration steps such as reverse osmosis or evaporation. Prior to the nanofiltration unit it permits to produce a highly demineralised whey up to 70% whereas without such a pre-treatment only 30% or 45% with diafiltration are obtained. The demineralisation rate can further improve (90%) through a demineralisation plant which is now of a much smaller size therefore consuming far less regenerant. It is possible to use a new decalcification system in the dairy industry with the following benefits • No additional reagent consumption on coupling the unit operations • A smaller deionisation system resulting in a lower reagent consumption • An improvement of the nanofiltration performances • An improvement of the lactose crystallisation unit performances • The possibility to achieve higher lactose recovery through a ion exclusion process

ABBREVIATIONS

IX Ion exchange resins NF Nanofiltration SAC Strong Acid Cation DS Dry Solid SBA Strong Base Anion VCF Volumetric Concentration Factor WAC Weak Acid Cation BV Bed Volumes

5/6 ICEF9-2004 International Conference Engineering and Food

BIBLIOGRAPHIE

AFNOR (1999). Le lait, pp 135-165 et ; 195-219 Balannec B., Gésian-Guiziou G., Rabiller-Baudry M., Boyaval E. & Chaufer B. (2001), Séparation- concentration des eaux de procédés de l’industrie laitière et production d’eau réutilisable par nanofiltration ou osmose inverse. Industries Alimentaires et Agricoles, mars 2002, pp 12-15. Barrantes L.D. & Morr C.V. (1997), Partial deacidification and demineralisation of cottage cheese whey by nanofiltration. Journal of Food Science, volume 62, n°2, pp 338-341. Bird J. (1996), The application of membrane systems in the dairy industry. Journal of the society of dairy technology, volume 49, n°1, pp 16-23. Cartier S.,Theoleyre M.A., Decloux M. (1997), Treatment of sugar decolorizing resin regeneration waste using nanofiltration, Desalination, Volume 113, pp 7-17. Gregory A.G. (1987), Desalination of sweet-type whey salt drippings for whey solids recovery. Bulletin of the International Dairy Federation, n°212, pp 38-49. Güngerich C. & Hutson G. (1996), Demineralisation of whey and whey Products. Bulletin of the International Dairy Federation, n°311, pp 11-13. Hervé D. (1974), Whey Treatment by ion-exchange resins.. Process Biochemistry, Febuary/March 1974, pp 16-17. Horst (Van der) H.C., Timmer J.M.K., Robbertsen T. & Leenders J. (1996), Nanofiltration for concentration and demineralisation of cheese whey : a mass transport model. Bulletin of the International Dairy Federation, n°311, pp 9-11. Jeantet R, Schuck P., Famelart M.H. & Maubois J.L. (1996), Intérêt de la nanofiltration dans la production de poudres de lactosérum déminéralisées. Le lait, Volume 76, pp 283-301. Jelen P. (1991), Nanofiltration – A new membrane processing application for demineralisation in the dairy industry. Journal of Canadian Institute for Food Science and Technology, Volume 24, n°5, pp 200-202. Kelly P.M., Horton B.S. & Burling H. (1992), Partial Demineralisation of whey by nanofiltration. International Dairy Federation Special Issue, pp 130-140. Kelly J. & Kelly P.M. (1996), Demineralisation and delactosation of dairy product, Bulletin of the International Dairy Federation, n°311, pp 8-9. Mikkonen H., Helakorpi P., Myllykoski L. & Leiski R.L. (2001), Effect of nanofiltration on lactose crystallisation. Milchwissenschaft, Volume 56, n°6, pp 307-310. Mucchetti G., Zardi G., Orlandini F. & Gostoli C. (2000), The pre-concentration of milk by nanofiltration in the production of Quarg-type fresh cheeses. Le lait, Volume 80, n°1, pp 43-50. Rousset F., Reboux P., Hudson T. (1997), Nanofiltration and ion-exchange for the demineralisation of whey. International whey conference 1997 Theoleyre M.A. (2000), Couplage résines échangeuses d’ions procédés membranaires. Intégration des membranes dans les procédés, Volume 14-2000, pp 83-90.

264, avenue de la Mauldre . 78681 EPONE Cédex . FRANCE . : [33] 01 30 90 50 00  : [33] 01 30 91 05 31 E-mail : [email protected] Web site : www.applexion.com

6/6