BACKGROUND OF THE INVENTION Techniques for the production of fermented beverages such as wine and beer particularly on a commercial scale are well studied and well established. After fermentation the fermented liquid may have a hazy or cloudy appearance resulting from precipitation of particulate matter derived from yeast cells and formed by proteins or polyphenols. This cloud or haze detracts from the appearance of the product and therefore results in a product that is less attractive to consumers. Consequently procedures for removing the haze forming particles from such beverages are well established.

In addition to fermented beverages other beverages such as fruit juices and the like can also have a cloudy appearance that can be removed using the established clarification processes.

Typically a first step in the clarification process involves'fining'the beverage by adding an agent (fining agent) that is able to flocculate the haze deposit so that it subsequently settles and can be removed by filtration, centrifuging or decantation. A wide variety of fining agents are known and examples of inorganic fining agents include aluminium modified silica sols and gelatin (US 4,027, 046 to Bohm et al.) and magnesium silicates (see for example US 4,508, 742 to McLaughlin et al.). Examples of organic fining agents include gelatin.

After the process of fining it is normally found that the beverage still contains a detectable haze, particularly when the beverage is cooled ('chill haze'), and it is at this stage that in commercial processes the beverage is'polished'by filtering through a filter medium. The beverage that is polished in this way is typically clear and suitable for packaging.

The choice of the filter medium for use in polishing the beverage is critical to the product. The filter medium must remove the majority of the haze forming particles

from the beverage and at the same time it must allow flow of beverage through the medium (flux) that is sufficient for commercial filtration and production. Further the filter medium must show selectivity in the components it removes from the beverage so that it does not effect sensory, organoleptic and physical attributes such as taste, colour, odour for example.

In each of the brewing and wine industries an overwhelmingly accepted filter medium is diatomaceous earth (DE) which is also known as Kieselguhr and diatomite. This medium is selected because of its favourable properties in removing haze forming particles in addition to it not apparently affecting the sensory or organoleptic properties of the beverage.

However, DE has been implicated as a human carcinogen and it has been listed as a Human Carcinogen Category 1, -International Agency for the Research on Cancer (IARC) 1997/1998. As a result of these health and safety concerns there is a need for an alternative filter medium that has at least some of the favourable filtering properties of DE. It is desirable, because of DE is predominantly used as a filter medium that plant and processes for filtering are set up for its use, that any alternative can be readily used with that plant and process with minimal adaptation.

OBJECT OF THE INVENTION The object of this invention is to provide a method of filtering beverages that obviates or alleviates any one of the above problems, or at least to provide the public with a useful choice.

SUMMARY OF THE INVENTION In a first aspect the invention could be said to reside in a method for reducing the amount of haze in a beverage, the method including the step of filtering the beverage through a bed of ground zeolite to thereby provide a clarified beverage suitable for consumption.

The beverage is preferably derived from fruits or vegetables such as beer, wine, fruit juice, vinegar and the like. In the following discussion reference to beer or wine should not be taken to limit the invention thereto but rather these beverages are used for illustrative purposes principally because they are the most commercially significant beverages requiring filtration.

Preferably the zeolite is selected from the list of known zeolites including Zeolite-A, Zeolite-X and Zeolite-Y. Most preferably the zeolite has a structural composition of [ (A102) i2 (Si02) i2L that is, Zeolite A.

The zeolites are characterised in that they are aluminosilicates having a framework of interlocking SiO4 and A104 tetrahedra and the ratio (Si + Al)/O is 1/2. The interlocking tetrahedra provide zeolites with a structure having vacant cages that allow space for inclusion of captions such as Na, K, Ba and Ca as well as molecules such as water, ammonia, carbonate ions and nitrate ions.

It will be appreciated that other minerals that have cage-like structures and/or properties similar to zeolites include phosphates such as kehoeite, pahasapaite and tiptopite; silicates such as hsianghualite, lovdarite, viseite, aprtheite, prehnite, roggianite, apophyllite, gyrolite, maricopaite, okenite, tacharanite and tobermorite. Because of the structural similarity and or the similar physical properties some of these materials have with zeolites, it may be expected that one or more of these minerals, in a suitably ground form, may also be used in the method of the present invention. The suitability or otherwise of any of these minerals can be tested using the techniques and protocols set out herein.

Preferably the zeolite filter medium is ground and graded and has a mean particle size or a particle size range of between 5-300, um, more preferably either 63-125jim or 125- 250, um, and most preferably 125-250, um. Larger zeolite particles can be conveniently ground using any suitable technique including grinding in a rod mill or other suitable mill. The crushed zeolite may then be graded using appropriately sized sieves.

Preferably the zeolite has a surface area of 600-700 m2/g.

In one preferred form of the invention the filtration step includes the step of forming a bed of zeolite filter medium and eluting the beverage through the bed. The beverage is preferably eluted through the bed under a positive driving pressure. The pressure may be in the range 20-650 kPa, and more preferably in the range 50 to 200 kPa. In one preferred form of the invention the driving pressure is 180 kPa.

A bed of filter medium may be formed by first forming a slurry of filter medium in beverage and then transferring the slurry into a filtration chamber.

The method of the present invention may be used to polish beverages such as wine and beer. Thus in beer production the ruh may be passed through the filter medium, preferably at about 0°C, to produce a clarified beer. Alternatively in winemaking the base wine may be passed through the filter medium to produce a haze stabilised wine.

The zeolite may be either natural or synthetic.

In a second aspect the invention could be said to reside in a ground zeolite capable of being used as a filter medium for reducing the amount of haze in a beverage to thereby provide a clarified beverage suitable for consumption.

Preferably the zeolite filter medium is ground and graded and has a mean particle size or a particle size range of between 5-30or, more preferably either 63-125fiv or 125- 250jim, and most preferably 125-250jim.

The zeolite may be any suitable zeolite or zeolite-like mineral as described earlier. In one preferred form the zeolite is zeolite-A. The zeolite may be either natural or synthetic.

In a third aspect the invention could be said to reside in a beverage clarified using the method of the first aspect of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS For a better understanding the invention will now be described with reference to an illustrated embodiment. The drawings describe an illustrated embodiment wherein, Figure 1 is a schematic diagram of a pilot filtration plant, Figure 2 is a schematic diagram of a filter vessel in which (1) is a sintered steel plate, (2) is a stainless steel impact plate, (3) is polycarbonate tubing, (4) is an upper stainless steel support, (5) is a lower stainless steel support, (6) is an 0-ring, (7) is a bolt, (8) is a nut, (9) is silk-cloth, and (10) is half inch tubing,

Figure 3 is a plot of flux vs time showing results of experiments with 18.83 g filter media and 180 kPa pressure gradient, four different media were used plots represented by shaded diamonds small grade zeolite A (lightest grey) large grade zeolight A (light grey) DE (darker grey) silica (darkest grey), Figure 4 is a plot of flux vs time showing results of experiments with 3.63 g filter media and 180 kPa pressure gradient, four different media were used plots represented by shaded diamonds small grade zeolite A (lightest grey) large grade zeolight A (light grey) DE (darker grey) silica (darkest grey), Figure 5 is a plot of volume and pressure vs time showing results of experiments filtering tap water with 35 g DE, Figure 6 is a plot of volume and pressure vs time showing results of experiments filtering yeast solution with 35 g of zeolite-A, Figure 7 is a plot of hazemeter readings expressed as EBC units (mean of three readings), Figure 8 is a plot of absorbance readings taken at 430nm (mean of three readings), Figure 9 is a plot of pH values of filtrates from various media as measured at a filtrate temperature of 20°C, Figure 10 is a histogram showing results of the descriptive comparison sensory test, Figure 11 is a plot of the inverse of turbidity as a function of pressure driving force for each filtrate from each filter media, Figure 12 is a histogram showing average pH of filtrate for each filter media, and Figure 13 is a histogram showing average concentration of sodium in the filtrate for each media.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS Example 1-Gene7^al Filt7-ation Procedu7e For experimental filtration of beer and wine a pilot plant was based around an egg pump in which pressure of a gas in a leak-proof vessel is increased above the liquid which forces the liquid out and through related pipe-work.

A food grade nitrogen gas was selected for the experimental studies. All wetted surfaces were food-grade 316 stainless steel.

A schematic flow diagram of the pilot plant is presented in Figure 1, in which: V1 is a food grade nitrogen supply valve PI indicates pressure of food grade nitrogen supply from the gas cylinder V2 is a pressure regulator to control the pressure supplied to the pilot plant downstream P2 indicates pressure of nitrogen supply after going through pressure regulator V3 is a control regulator to set pressure of nitrogen supply to the pressure vessel-V3 keeps the pressure constant even if the upstream pressure fluctuates V4 is a three-way valve that allows manual venting of the process line and pressure vessel P3 indicates the pressure in the pressure vessel V5 is set to vent at 6.0 bar and prevents the pressure in the vessel from becoming too great as to be unsafe V6 is an emergency shut off that can be used to instantaneously stop the flow from the filter vessel P4 indicates pressure just above the filter cake.

The pilot plant was designed such that when there was an excess pressure from the nitrogen tank, the tank could be vented from the line to the atmosphere manually using V3 and thereby preventing any damage to V4. The pressure regulator could be set for accurate regulation (between 0 to 1000 kPa). The pressure was monitored using the pressure gauge fitted to P3. A pressure relief valve (V5) was fitted to the lid of the pressure vessel to prevent over-pressure and was set at 620 kPa. A ball valve (V6) was installed as an emergency shut off.

Details of construction of the filter vessel are given as Figure 2. The filter vessel consisted of a piece of clear polycarbonate tubing clamped together with two stainless

steel flanges. The filter media was supported within the polycarbonate tubing by a sintered-stainless-steel plate. A pressure gauge (P4) is fitted above the filter vessel to measure the pressure drop across the filter bed.

The diatomaceous earth (DE), silica and zeolite-A filter media were assumed to be sterile. The filter sand was not regarded as sterile. The sand medium and all surfaces of the pilot plant were sterilised with a 70% v/v ethanol solution prior to filtration. The pilot plant was sterilised in situ using a commercial sodium metabisulphite solution (or a 70% v/v ethanol solution) at start up. This included all downstream equipment (filtrate hose, lid and sample container).

Experimental beer filtration studies were carried out in situ in Coopers Brewery Ltd, Leabrook, SA 5068. The quality of filtrate samples was evaluated and directly compared against routine commercial beer produced by conventional DE process methods.

Masses of either 35 or 65 g of filter medium were used in preliminary studies carried out in the laboratory and masses of 3.63, 11.23 and 18.83 g, respectively, when ill situ at Cooper's Brewery Ltd. These masses simulated the depth of the filter bed used in the commercial production of beer.

Example 2-Filtration of Beer and Comparison of Filter Media Preliminary trials with the pilot plant were carried out using different media with tap water, home-brewed beer and a purpose-made, beer simulant (as a yeast solution). DE as Celite 5033, pumice, cotton wool, filter sand, two size grades of zeolite, perlite and silica were trialled.

Zeolite-A is available (from Dri-Packs Pty Ltd, NSW) in the form of beads of 3 to 5 mm diameter. These were ground to appropriate size for filtration. Thus the zeolite-A was ground to both a size range of 63-125Rm (small diameter) and 125-250Fm (large diameter) using a small rod mill with stainless steel rods. The desired particle sizes were obtained with a continuous vibrating stack of screens.

The pilot plant was initially checked for possible faults by running tap water at a pressure between 206-620 kPa (30-90 psig). Home-brewed beers were filtered using DE (Celite 503) and a small grade zeolite-A with a particle size of 63-125jim.

The simulated beer was used to assess each of the filter media. All filtrates that were used for microbial and sensory analyses were standardised. This was done by using a fixed bed mass of 18.83 g of medium and a fixed pressure driving force of 180 kPa.

Investigatory samples of beer filtrate from the sintered plate only in place in the pilot plant (ie. no filter medium) highlighted that no detectable haze was removed by the sintered material.

Flux-time experiments were conducted using three fixed pressure drops (70,125, 180 kPa) and three filter bed masses (3.63, 11.23 and 18.83 g). Filter media were SuperCel and FilterCelTM (Cooper's Brewery Ltd DE Mix), the two grades of zeolite-A (large grade and small grade), silica and filter sand. Cooper's Brewery DB beer was the feed material.

Initially, to simulate the brewery practice used for all commercial-scale DE filter cake, a precoat, a precoat plus one batch of body-feed (i. e. additional DE), and a precoat plus two batches of body-feed respectively, was trialled in the pilot plant. The precoat procedure involves the preparation of a beer-DE slurry that is applied as a thin layer to the filter support and left to"dry"for a short period of time prior to filtration of the main body of beer. This assists establishment of a stable filter cake.

Commercial beer samples were filtered in trials in situ at Cooper's Brewery Ltd using each of the five selected media (silica, filter sand, DE and the two grades of zeolite-A) with three selected pressure gradients. These were, respectively, 70,125 and 180 kPa.

Three filter beds of each medium were used. The mass of each of these was, respectively, 3.63, 1 1. 23 and 18.83 g.

Figures 3 and 4 show sample results of the nine trials that were carried out for each of the five media. These are for a pressure gradient of 180 kPa and 3.63 g of filter media and 180 kPa and 18.83 g of each filter media, namely, zeolite-A small grade, zeolite-A large grade, DE and silica. Respectively these gave fluxes of : 22, 290 and 390 mL m~ 2su for a bed mass of 18.83g and a pressure driving force of 180 kPa (Figure 3).

The deeper bed of media at the pressure driving force of 180 kPa gave the best flux-time result for each of the five media. This combination had also resulted in the best results for microbial analyses of the filtrate. A pressure gradient of 180 kPa is about 20 kPa greater than the pressure gradient used generally in commercial filtration of beer haze with DE. It is nevertheless a pressure gradient that could readily be used routinely with existing commercial equipment and preparation protocols, which provides an obvious advantage when looking for alternatives to DE.

2.1 Microbiological Analysis of Filtrates Homebrewed beers, tap water and prepared yeast solution (beer simulant) were filtered in the pilot plant using zeolite-A and DE. The filtrates were analysed for viable, and total, cell count. Total cell count was assessed using a haemocytometer, and the viable cell counting by the Spread Plate Method (Meynell, G. G. and Meynell, E. 1970 Theory and Practice in Experimental Bacteriology. University Press, Cambridge, UK Pp. 23 ff. ) on Savouraud's agar media.

Figures 5 and 6 summarise results of the tap water and yeast solution trials for both DE and small grade zeolite-A. A mass of 35 g of each medium was used. This gave an approximate bed depth of, respectively, 2.5 cm and 1.5 cm. The data are plotted as filtrate volume (mL) versus time of filtration (s).

Two commercially sourced home-brewed beers, Black Rock Lager and Dark Ale, were filtered to remove haze constituents. Black Rock Lager was filtered using DE. Dark Ale was filtered using both DE and small grade zeolite-A (63-125jim particle diameter) filter media.

The Black Rock Lager filtrate obtained with DE was bright-clear. Total solids were reduced by 28% (Table 1).

Table 1 Results of filtration of home brewed Black Rock Lager beer using DE Filter Control Filtrate media weight (g) na 35.14 filter pressure (psig) na 30 filtration time (min. s) na 2.00 filtration volume (L) na 1.3 beaker mass (g) 33.24 9.61 sample volume (mL) 44.25 34.5 final mass (g) 76. 81 43.94 sample mass (g) 43.57 34.33 sample density (kg m-3) 985 995 evap. beaker/solids (g) 34. 9 10.8 solid content (g) 1.67 1.19 % solid content (wt %) 3.8 3.5 initial solid content (g) 1.67 initial solid content (wt %) 3. 8 initial solid concentration (kg m-3) 37.6 final solid content (g) 1.19 final solid content (wt %) 3.5 final solid concentration (kg m-3) 34.5 solids removed (g) 0.47 % reduction of initial solids 8. 2 A direct comparison of the filtering capabilities of DE and zeolite-A was made with the Dark-Ale beer. Both filtrates showed a satisfactory clear beer. Zeolite-A was more effective in reducing the amount of solids in the filtrate as compared to DE, reducing the total solids by, respectively, 9% and 3.6% (Table 2 and Table 3). Microbial analyses of both control and filtrate (in triplicate) samples showed the viable yeast cell numbers for both filtered beers was reduced (Table 4). The beer samples filtered using DE were reduced from viable numbers of 106 cells mL-l to 105 cells mL-l for both types of beer. For the zeolite-A-filtered beer there was no growth evident from plating the filtrate. This indicates a total removal of all viable cells of yeast.

Table 2 Results of filtration of home brewed Dark Ale beer using DE Filter Control Filtrate media weight (g) na 35.14 filter pressure (psig) na 30 filtration time (min. s) na 2.20 filtration volume (L) na 1.0 beaker mass (g) 88.66 9.66 sample volume (mL) 31.50 33.5 final mass (g) 124.57 47.75 sample mass (g) 35.91 38.09 sample density (kg m-3) 1140 1137 evap. beaker/solids (g) 1.11 10.73 solid content (g) 3.1 1.07 % solid content (wt %) 2. 8 initial solid content (g) 1.11 initial solid content (wt %) 3.1 initial solid concentration (kg m-3) 35.2 final solid content (g) 1.07 final solid content (wt %) 2. 8 final solid concentration (kg m-3) 31.9 solids removed (g) 0.04 % reduction of initial solids 9. 4

Table 3 Results of filtration of home brewed Dark Ale beer using small grade zeolite-A Filter Control Filtrate media weight (g) na 65.02 filter pressure (psig) na 60 then 90 filtration time (min. s) na 19.00 filtration volume (L) na 0.5 beaker mass (g) 88.66 9.58 sample volume (mL) 31.50 36.75 final mass (g) 124.57 47.04 sample mass (g) 35.91 37.46 sample density (kg m-3) 1140 1019 evap. beaker/solids (g) 89.77 10.59 solid content (g) 1.11 1.01 % solid content (wt %) 3.1 2.7 initial solid content (g) 1. 11 initial solid content (wt %) 3.1 initial solid concentration (kg m-3) 35.2 final solid content (g) 1.01 final solid content (wt %) 2. 7 final solid concentration (kg m-3) 27.5 solids removed (g) 0.10 % reduction of initial solids 21. 9 Table 4 Results of microbiological analysis of home brewed beers Sample/Filter Medium Meana Standard viable cell Deviation count (cell mL-l) (cell mL- Control: Black Rock Lager 4.82 x 106 2.29 x 105 Black Rock Lager beer filtered with (35g) DE 1.89 x 106 3.64 x 105 Control: Dark Ale 6.00 x 106 6.16 x 105 Dark Ale filtered with (35g) DE 8. 10 x 105 7.31 x 104 Dark Ale filtered (65g) small grade zeolite-A 0 a Mean of three replicates

To evaluate a range of filter media, a yeast solution was prepared as a test liquid and beer simulant. This liquid was filtered at 206.84 kPa (30 psi) using either 30 g or 35 g of each of the eight filter media. Media included: cotton wool, pumice, perlite, silica, filter sand, DE and the two grades of zeolite-A (small and large).

Results (Table 5) showed that DE and small grade zeolite-A were the most effective in filtering out the yeast cells. Cellulose (as cotton wool), pumice and perlite were rejected as unsuitable for further experimental trials in the filtration of yeast cells (haze) from beer because these did not remove an adequate amount of the haze in these preliminary trials.

Further, Table 5 shows that the standard deviation on three replicate filtrations for these three media gave a very large standard deviation of nearly an equal order of magnitude as the mean value. The implication is that the mean pore size varied greatly despite careful experimental technique with each preparation of the filtration bed from these three media.

Table S Results of microbiological analysis of filtrates of yeast solution Media Total Standard viable yeast Deviation cell counta (cell mL-1) (cell mL-1) cellulose (as cotton wool) 9.51 x 106 6.77 x 106 perlite 7.07 x 106 5.26 x 106 pumice 3.55 x 106 2.16 x 106 filter sand 7.43 x 105 2.91 x 104 precipitated silica 1.57 x 105 6.12 x 104 zeolite-A large grade (125-250 um) 5.07 x 104 2.65 x 104 DE (as Celite 503) 0 0 zeolite-A large grade (63-125 Fm) 0 0 a Mean of three replicates

2.2 Physical and Chemical Analyses of Filtrates Containers for collection of the filtrate were sterilised using ethanol solution (70 % v/v), purged with nitrogen to (minimise contact with oxygen) and sealed prior to filtrate collection. Filtrate was collected (about 2.5 L each trial) and stored in a cold room at a temperature of 2 to 4 °C prior to analyses.

For each sample three properties haze, colour and pH were measured: The haze level of samples was measured using a VOS 4000 hazemeter. EBC (European Brewing Convention) units were registered in a digital read-out indicating the ratio of scattered and transmitted light intensities. A haze reading of <1 EBC is commercially considered a bright (i. e. desirable) beer (Gan et al 1997 Transactions of the Iiistitlition of Chemical Engineers, PartA, Chemical Engineering Research and Design. 75,3-7).

Filtrates were kept at 2 to 4°C for a period of about 2 to 3 weeks until analysis. The filtrates were handled aseptically at all times.

Hazemeter readings on the filtrates showed that DE, small grade zeolite-A and the brewery Seitz filter gave acceptable haze levels of below 1 EBC unit. In most commercial beers, those filtrates with 0.8 EBC units or less are regarded as acceptable (Gan et al, ibid.).

Figure 7 summarises the average hazemeter reading (on three replicate filtrates) for each of the five media trialled and presents a comparison with the brewery's commercial Seitz filter (which itself uses DE). For pilot plant DE filtrates the mean haze reading is 0.6 EBC and those for the small grade zeolite-A of 0.8 EBC. Silica filtrates had a mean haze reading of just greater than 1 EBC, filter sand 5.2 EBC and the large grade zeolite-A a mean of 3.2 EBC. The commercial Seitz filtrates had a mean of 0.6 EBC.

The pilot plant DE and small grade zeolite-A therefore gave very similar haze reducing capability as the commercial equipment of the brewery's Seitz filter.

Samples for the colour test were filtered using a standard industry glass filter paper (Whatman GF/C). Absorbance readings were taken at 430 nm using a Varian DMS 200 UV Spectrophotometer.

The spectrophotometric analyses of filtrates is summarised as Figure 8. The mean absorbance reading on three replicates (produced from trials with a pressure gradient of 180 kPa and a filter bed mass of 18.83 g) for each filter medium is presented. The colour of commercially produced beer filtrates (from the Seitz filter) gave an absorbance reading of 0.32.

Figure 8 highlights the fact that filtrates from the DE filter bed of the pilot plant had an almost identical mean absorbance reading (0.32) as the commercial"control"of the Seitz filter. Large grade zeolite-A and filter sand resulted in filtrates with an absorbance reading of 0.34 which compares favourably with DE and the commercial Seitz filter.

The mean absorbance reading of filtrates from the small grade zeolite-A was 0.44, a value that is significantly greater than all other filtrates.

These spectrophotometric readings imply that a commercially unacceptable increase in colour is attached to filtrates using small grade zeolite-A. Interestingly, the colour of the filtrates from silica as the filter medium (with an absorbance reading of 0.29) was actually lighter than those of the Seitz filter.

Sample pH was measured at a sample temperature of 20 °C using a standard pH probe.

A summary of pH values of the resulting filtrate from each of the five filter media is presented in Figure 9. The figure shows that the pH value of the filtrate from small grade zeolite-A as filter medium increased from pH = 4 (ie. the Seitz filter control value) to a pH value of 6.

Figure 9 shows that large grade zeolite-A also caused an increase in pH value (from about 3.9 to 4.3) of the filtrate but resulted in a value within the range suitable for commercial beer product. The pH value of the filtrate from both filter sand and silica is seen from Figure 4.8 to be equal to that from the Seitz filter control value of pH = 4 units.

The resulting increase in pH value of the beer filtrates from small grade zeolite-A and large grade zeolite-A may be accounted for by an increase in sodium ion concentration of the filtrate. The sodium ions are therefore leached from the zeolite-A media during filtration.

This increased pH, should it be desirable, might be remedied by pre leaching the zeolite to reduce the sodium ions that are present.

2.3 Sensory Analyses of Filtrates Two methods of sensory analysis were used, the Triangular Method and the Descriptive Comparison. The Triangular Method uses three samples presented simultaneously and requires the subject to choose the"odd"sample. A"no difference" reply is not recorded. This therefore forces a choice from the subject even when the results are not clear. Beer filtrates from small grade zeolite-A were assessed against the beer filtered in the Seitz filter (ie. control) using the Triangular Method. Twelve experienced assessors (noses) determined if there was a difference apparent in the two beers.

The Descriptive Comparison method evaluates the beer filtrates by describing aroma, colour, clarity, taste and drinkability and overall impression. Filtrates are presented all at once to assessors. Instructions are as simple (as is possible) and require the assessors to rate the intensity of each characteristic on a scale from 0 to 10,0 being "poor"and 10 being"excellent".

The Triangular Method of Analysis revealed brewery industry noses could differentiate between beer filtrates from DE and those from small grade zeolite-A as filter medium.

The results from the Descriptive Comparison more clearly differentiated filtrates from the filter media.

In the Triangular Method test, from the sixteen assessors, ten were able to distinguish the"odd"filtrate from the three given samples. This number of correct replies is greater than the minimum correct reply required to establish a significant difference between the two types of beer filtrates at 5% level of significance.

Filtrates of the five selected filter media and that from the Seitz filter were also evaluated using the Descriptive Comparison sensory test. The code for each of the filtrates as presented to the assessors and filtrate identity are listed in Table 6 and the mean rating for each of the characteristic attributes of the filtrates is presented as Table 7. Sixteen noses evaluated the beer filtrates.

Table 6 Codes and description used for the Descriptive Comparison Sample Number Description 146 silica 552 De 442 filter sand 579 small grade zeolite-A 857 Seitz filter 361 large grade zeolite-A Table 7 Mean rating for Descriptive Comparison Simple Descriptive Comparison for the Evaluation of Beer Filtrates Mean Ratinga Sample Aroma Colour Clarity Taste Drinkability/ Code Overall Impression 146 4.33 6.29 7.29 4.00 4.00 552 5.60 6.47 7.47 5.29 5.69 442 4.75 6. 80 7.07 5.80 5.73 579 4.13 4.73 6.67 3.89 4.00 857 4.56 6.40 7.07 4.89 5.27 361 4.60 6.40 7.60 5.33 5.50 a Mean rating on 16 brewery assessors (noses) The mean rating values of this analysis is presented in a histogram as illustrated in Figure 10.

The DE-filtered beer is the highest rated filtrate for aroma with a mean rating of 5.60 followed in descending order by filtrates of filter sand, large grade zeolite-A, Seitz filter, silica and the least rated is the filtrate of small grade zeolite-A with mean rating of 4.13.

The best colour rating was that of the filtrate of filter sand with mean rating of colour = 6.80. This is followed by filtrates of DE, then equal mean ratings for Seitz filter and large grade zeolite-A filtrates (colour = 6.40), followed by silica and lastly the small grade zeolite-A with a mean rating of colour = 4.73. These results are supported by the

spectrophotometric analyses of the filtrates, where the absorbance of the filtrate of silica is lower than that of the other filtrates except that of the filtrate of small grade zeolite-A which is higher by about 0.12 from the other absorbance readings.

Filtrate of large grade zeolite-A has the highest mean rating for clarity of 7.60. This is followed in descending order by filtrates of DE, silica, equally rated Seitz filter and filter sand and lastly small grade zeolite-A with mean rating of 6.67. These results are not in agreement with hazemeter readings obtained in the laboratory. The filtrates of filter sand and large grade zeolite-A have high haze level contents and the small grade zeolite-A with EBC units within the acceptable level but visual analysis of these filtrates gave different results as evident from the sensory ratings.

The highest rated filtrate for its taste is that of the filter sand (mean rating of 5. 80) followed in descending order by large grade zeolite-A, DE, Seitz filter, silica and lastly with the filtrate of small grade zeolite-A with mean rating of 3.89.

Among the filtrates analysed, the most preferred for its drinkability and overall impression is the one filtered with filter sand (442) with mean rating of 5.73 and the least preferred are the small grade zeolite-A (146) and silica (146) with equal mean ratings of 4.00. The other filtrates rated in ascending order as Seitz filter (control), large grade zeolite-A and DE.

From the ranking of each characteristic according to the average intensity of the perception of the assessors (Table 6), it can be seen that the filtrate of filter sand has the best attribute overall, with high attribute ratings except for its clarity. Large grade zeolite-A is comparable to the existing medium (DE) with the alternative medium being more preferred in clarity and taste.

2.4 Ion-Exchange Beer was filtered using zeolite-A as filter media. Samples volumes were collected at intervals of time. These were then tested for pH change. Collection of samples continued until the pH of the filtrate appeared to be stabilised. Filter beds (18.83 g) of both the small and large grade zeolite-A with a pressure driving force of 180 kPa were experimentally investigated. Four (4) replicates were used and the pH of the filtrate monitored for between 8 and 16 h of continuous filtration to determine if all sodium could be exhausted from the medium.

The assumption was made that alkaline (sodium) species leaching out of the zeolite structure gave rise to the increase in pH.

During beer fermentation, the pH of the beer is reduced as a result of the increased production of the positively charged non-microbiological particles (NMPs) (Leather, Dale and Morson 1997).

With filtration however, these positively charged NMPs are removed from the beer by substitution. The cations located in the pores of the zeolite filter medium, in this case sodium cations, are substituted with the NMPs and the sodium cations are washed away to become part of the filtrate.

One reason why larger numbers of sodium ions leach from an equivalent mass of 18.83 g of the small grade (63 to 125 gm particle diameter) zeolite-A filter cake (resulting in a pH increase of 2 pH units of the filtrate), in comparison with the smaller number of sodium ions that leach from the large grade (125 to 250, um particle diameter) zeolite-A filter cake (resulting in an increase of less than 0.5 pH units) could be related to a difference in residence time of the beer in the two filter cakes.

The length of the filter path for passage of beer in the small grade zeolite-A would presumably be significantly greater than with the large grade material. There is therefore an overall larger surface area of filter medium in contact with the beer with the small grade material together with a greater residence time of the beer compared with the large grade material. The release of sodium ions therefore appears to have both a mechanical and a chemical basis in giving rise to increasing the pH of commercial beer filtrates.

One approach to this as yet unresolved problem might be to prevent the sodium cations leaving the filter medium with the use of a carefully selected chelating agent. This agent might be added during preparation of the wetted slurry of the medium. The desired outcome is that the sodium will be trapped to the chelating agent whilst the positively charged particles of the beer haze are attached within the zeolite-A framework. It should be reiterated that the increase in pH of the beer filtrates from the large grade zeolite-A filter cake was less than 0.5 pH units. Therefore it is envisaged that a judiciously selected particle size for zeolite-A filter medium will limit this in the first instance.

Example 3-Filtration of Wine For filtration studies white wine was chosen over red as white wines require filtration because visible changes such as cloudiness are more readily apparent. A base white wine that had not undergone cold stabilisation or polishing was selected.

Samples were filtered as described in Example 2 using the pilot plant with DE (Grade 100), Zeolite-A large grade (125-225, um particle size; Substitute A), commercial (Zeolite-A as Silosive A3TM [WR Grace Australia Pty Ltd, Melbourne], 5 urn particle size) and small grade (62-125 Rm particle size; Substitute C), cotton wool, pumice and filter sand were tested as filter media with appropriate controls. The systematic coding of samples is shown in Table 8.

Table 8 Systematic and confidential codes of samples Filter Media Bed pressure drop Sample (bar) DE Grade 100 1 6117 Substitute A 1 4129 (large grade Zeolite-A) Cotton wool 1 3137 Pumice 1 9141 Filter sand 1 5154 Substitute 1 6168 (5 um zeolite-A) Control (paper only) 1 7171 Control (no filtering) 0 8183 DE Grade 100 2 6217 Substitute A 2 4229 (large grade Zeolite-A) Cotton wool 2 3237 Pumice 2 9241 Filter sand 2 5254 Substitute 2 6268 (5, um zeolite-A) Control (paper only) 2 7271 Control (no filtering) 0 8283 DE Grade 100 4 6417 Substitute A 4 4429 (large grade Zeolite-A) Cotton wool 4 3437 Pumice 4 9441 Filter sand 4 5454 Substitute 4 6468 (5 µm zeolite-A) Control (paper only) 4 7471 Control (no filtering) 0 8483

Sample filtrates were collected as described in Example 2 and were analysed for: pH, free S02, fixed S02, alcohol, Cu, Fe, K, Na, Ca, heat stability and turbidity. The results are presented in Figures 11,12 and 13.

The filtrates also underwent oenological analysis by an expert panel from Southcorp Wines Pty Ltd and the results are presented in Table 9.

Table 9 Results from oenological panel tests Sample Description Result number 3137 bumt/caramelised, flat/no fruit not acceptable 3237 as per 3137 not acceptable 3437 as per 3137 not acceptable 4129 earthy/musty taint, flat/no fruit not acceptable 4229 as per 4129 not acceptable 4429 as per 4129 not acceptable 5154 fruit developed (not desirable), grassy/sulphidic taint almost accept (slight) 5254 as per 5154 almost accept 5454 as per 5154 almost accept 6117 flat, lacks fruit, slightly earthy not acceptable 6168 some burnt character (as per 3137) not acceptable 6217 earthy character not acceptable 6268 burnt character, developed (as per 3137) not acceptable 6417 slightly developed, grassy/sulphidic alright 6168 similar to 6417 alright 7171 aromatic, although slightly subdued fruit acceptable 7271 as per 7171 acceptable 7471 as per 7171 acceptable 8183 slightly flat (subdued fruit) good 8283 as per 8183 good 8483 lifted fruit best sample 9141 slightly dull fruit, burntlcaramelised (not as bad as 3137) acceptable 9241 as per 9141 acceptable 9441 as per 9141 acceptable

Measurement of the value of flux for each filter medium showed that substitute A was equivalent to DE, celite 503 and lower than DE 100 Grade. Cotton wool, washed filter sand and pumice although giving an overall larger flux, did not produce an acceptable polished wine.

The oenological analysis showed that the six media were ranked in the following order: pumice > washed filter sand > substitute B = DE > substitute A = cottonwool.

Results of the chemical and physical analytical tests showed that: 'the concentration levels of both free SO2 and fixed SO2 were not affected by any of the filter media Alcohol levels remained unaffected in all filtrates None of the filter media had any significant effect on copper levels in the filtrate With DE as the filter medium, the level of iron increased from an average of 1.4 mg L-1 to about 2. 3 mg L-1 and iron levels remained unaffected in the other filter media Substitutes A and B raised the pH from 3.2 to 4.5 Substitutes A and B raised the sodium level of the filtrate from 100 mg L-1 to 900 mg L-1 and from 100 mg L-1 to 400 mg L-1, respectively 'Substitute B increased the amount of potassium in the filtrate from 600 mg L- to 900 mg L-1 Substitute A decreased the amount of calcium from 90 mg L-1 to 60 mg Li whilst Substitute B increased calcium levels from 90 mg L-1 to 120 mg L-1.

It will be appreciated that in addition to the Zeolites tested to date, it is envisaged that alternative zeolites may also be suitable for clarification of wine, beer and other beverages. Testing of alternative zeolites can be carried out using the tests described in the preceding Examples.