GB2355989A - Continuous filtration method and apparatus - Google Patents

Abstract

The improved method and apparatus provides for the use of at least two filters being arranged in series and wherein each filter has different properties and therefore a different role in the filtration process. The employment of such a method provides a continuous filtration system for filtering fatty acid esters out of reduced strength spirit without the need to significantly chill the filtrate prior to filtration. Such a method and apparatus can preferably be employed as part of distillation process, particularly for whisky.

Description

0 2355989 1 1 "'Filter" 2 3 The present invention relates to an improved

method and 4 apparatus for the continuous filtration of specific 5 substances during the alcohol distillation process. 6 7 Whisky and similarly produced alcoholic beverages are 8 filtered in order to remove the haze which is formed 9 when water is added to cask strength spirit to reduce 10 it to bottling strength. When the water is added to the 11 cask strength spirit the fatty acid esters come out of 12 solution and make the whisky appear cloudy. 13 14 The whisky is chill filtered to ensure that the haze 15 will not form in the bottle if the whisky temperature 16 is reduced for example in a cold climate. Whisky 17 filtration uses traditional technology. The general 18 method uses a plate and frame filter press with 19 cellulose acetate sheets. When the water has been added 20 to the whisky the reduced spirit is left for at least 2

1 three hours and often longer before filtration, this 2 being necessary for the filtration to be effective 3 using this method. The whisky is then chill filtered 4 between -100C and +5'C depending on the production 5 process employed by the whisky producer. 6 7 This chilling stage prior to filtration uses a lot of 8 energy and is therefore very costly. The chilling also 9 helps the fatty acid particles to agglomerate which 10 makes it easier for the filters to remove them from 11 solution. When starting the filtration, the whisky must 12 be recycled for at least 15 minutes to ensure that 13 particles of floc which come out of the filters are 14 removed and do not enter the bottling vat. This 15 recycling is also necessary to build up a primary 16 filter cake on the filter. This is a layer of fatty 17 acid ester particles which will in themselves act as 18 filter and help to remove smaller particles which might 19 otherwise get through the filter media. 20 21 This filtration process is a batch process as the rate 22 of removal of the particles is not constant with time. 23 The turbidity of the whisky is a measurement of how 24 clear the product is, and is measured in ppm (parts per 25 million). Throughout a batch the whisky turbidity will 26 increase and by the end of the batch the turbidity of 27 the whisky will be higher than the specification. 28 However as this is a batch process, the turbidity in 29 the batch will be within specification. This increase 30 in turbidity is due to "breakthrough" from the filters. 31

3 1 Breakthrough occurs when the sheets become increasingly 2 loaded with particles. It is then more difficult to 3 force the whisky through the sheets causing an increase 4 in the pressure drop across the filter sheets. As the 5 whisky is forced through the sheets, some of the 6 particles previously caught in the sheets breakthrough 7 the filter causing the resultant turbidity increase. 8 9 When the filter sheets are exhausted they must be 10 disposed of. Due to the nature of the sheets 11 typically approximately 1 litre of whisky per sheet is 12 retained in the filter. This is lost product which must 13 be disposed of together with the sheet since it cannot 14 be recovered.

is 16 A change in the filtration technology was considered 17 for whisky to find a more energy efficient system which 18 can filter at a higher temperature. A filter which 19 could be cleaned and therefore reused would also be -20 desirable. A reduction in the loss of product in 21 concomitant with filter replacement, would also be 22 desirable, as would the replacement of existing batch 23 processing techniques, with a filtration process which 24 could be filtered immediately after the whisky had been reduced.

26 27 It is an object of the present invention to provide a 28 continuous filtration system for filtering fatty acid 29 esters out of reduced strength spirit without the need to significantly chill the filtrate prior to 31 filtration.

32 4 1 According to the present invention there is provided a 2 method of filtration which consists of a multi stage 3 process with each stage containing a filter, with the 4 filters being arranged in series. 5 6 Further, according to the present invention there is 7 provided an apparatus for continuous filtration during 8 a distillation process wherein the apparatus comprises 9 at least two filters arranged in series wherein each 10 filter has a different property and thus a different 11 role in the filtration process.. 12 13 In one particular embodiment the apparatus comprises 14 three filters. 15 16 In an alternative embodiment of the invention, the 17 invention provides an apparatus as described above 18 which consists of only two filters arranged in series, 1-9 wherein the two filters have different filtration 20 properties. 21 22 Preferably the method of the invention will involve a 23 three stage filtration process. 24 25 Preferably in such a three stage filtration process, 26 each filter will be of a different type, with 27 preferably the first filter of the series being a guard 28 filter, a second filter being a prefilter and the third 29 filter being a membrane filter. 30 31 Also preferably, there may be two filters in series, 32 whereby the guard filter provided in the three filter 1 system is not present, due to it being located at an 2 alternative stage in the filtration process. 3 Preferably the final, or membrane filter, of the series 4 will be composed of a material which is hydrophillic. 5 6 More preferably, the final membrane filter will be 7 composed of polyvinylidene fluoride (PVDF). 8 9 Preferably the filtration will be carried out with the 10 filtrate at a temperature of between 50C and 250C. 11 12 Most preferably, the filtration will be carried out of 13 the filtrate at a temperature of between 80C and 200C. 14 15 Preferably filtration will be carried out at a defined 16 pressure range, which will have been optimised to the 17 filter size and type as used. 18 19 Preferred embodiments of the filtration device and 20 process are described below. 21 22 The preferred embodiment of the present invention is a 23 three filter train through which the whisky flows in 24 series. Typically the invention consists of an initial 25 guard (or nominal rating) filter followed by a pre26 filter (combination of surface and depth) and a final 27 membrane filter. 28 29 The system performance is sensitive to several 30 variables: the type of filters to be used, the material 31 used in the filter construction, the relative micron 32 rating through the filter train, the upstream liquid 6 1 (whisky) temperature and the pressure drop across each 2 filter. 3 4 The filtration media can vary for all three of the 5 filters used. There are a large number of different 6 membranes and filter media commercially available in 7 cartridge form. Also the micron rating of the media 8 could vary especially for the guard and the pre-filter 9 and absolute rating filters could be used rather than 10 nominal rating filters. If the whisky is left to sit 11 after reduction, as is the case now a larger micron 12 rating may also be possible for the final filter. 13 14 A two- filter system could be used rather than a three15 filter system if there is a guard filter present 16 elsewhere in the system. It is also possible that just 17 a final membrane filter may be used although it is 18 unlikely that this system would be successful for large 19 batches. 20 21 It is also possible that membrane filter bags may be 22 used instead of filter cartridges. The temperature used 23 for the system could vary considerably as at the 24 present time the temperatures used by different 25 companies are different. 26 27 This technique can also benefit whisky production in 28 other ways such as in the removal of sugary products 29 thereby prolonging the lifetime of filters. 30 31 The filtration system consists of a heat exchanger to 32 chill the whisky to the desired temperature and a 7 1 three-filter train. The filters are connected in series 2 and the whisky flows through each in turn. The first 3 filter housing contains guard filter cartridges these 4 are 1 Rm nominal rating filters. The purpose of these 5 filters is to remove any large particles for example 6 char from the casks. The second filter housing contains 7 pre-filter cartridges and these are 0.2/0.45 gm nominal 8 rating combined filters. The purpose of these filters 9 is to remove larger particles and to protect the 10 membrane filter as much as possible. The final filter 11 housing contains 0.22 gm absolute rating membrane 12 filter cartridges. These are the polishing filters to 13 remove the small particles and will reduce the fatty 14 acid ester levels to an acceptable level for bottling. 15 16 Suitable filters are known in the field. Examples of 17 typical supplies include Millipore and Pall. 18 19 The temperature must be controlled closely as the 20 temperature has a large impact on the quality of the 21 final product. 22 23 There should be an upstream and a down stream pressure 24 measurement for each of the filters. The pressure drop 25 across each filter should be maintained below the set 26 level. Excessive pressure will cause breakthrough from 27 the filters. Maintaining the pressure drop causes the 28 flow to fall below an acceptable level then this is a 29 sign that the filters should be cleaned. It is possible 30 that his could be used as a real time diagnosis of 31 filter status. 32

8 1 The batch size through the filters will be dependent on 2 the type of whisky or other similar product, which is 3 being filtered and should be set for individual 4 products. To prolong the lifetime of the filters, they 5 should be cleaned before exhaustion as running heavily 6 blocked filters causes excessive wear on the filters 7 especially the membrane. 8 9 The invention is demonstrated with reference to the 10 accompanying figures wherein; 11 12 Figure 1 illustrates fatty acid ester levels for 13 membrane filtration, and 14 Figure 2 illustrates liquid turbidity as a 16 function of temperature.

17 18 In whisky filtration, the pre-filtration (or upstream) 19 whisky temperature is known to be a critical parameter.

Tolerances of the order of 1-2'C have been shown to 21 result in changes to system effectiveness as shown in 22 Figure 1.

23 24 Studies have shown a performance sensitivity to choice of filter material. Specifically, the preferred choice 26 of material is that which is hydrophilic. A typical 27 material is polyvinylidene fluoride (PVDF).

28 Alternatives such as cellulose acetate have been shown 29 to work.

31 Filter train effectiveness is measured in a variety of 32 ways as outlined in more detail below.

9 1 The liquid turbidity as a function of temperature has 2 been measured and is shown in Figure 2. other checks 3 such as Arganoleptic testing, Chill haze stability, gas 4 and liquid chromatography to respectively measure the 5 ester and higher alcohol levels in the whisky are tests 6 which have been undertaken and which quantify the 7 improvements of the filter train. 8 9 Effective filtration using a two stage process has been 10 shown on blended whisky. Also, using the two stage 11 process, the whisky can be filtered immediately after 12 reduction at ambient temperature - without any need to 13 reduce the temperature of the whisky. 14 15 Experiment 1: 411 Filters 16 17 To determine the best membrane for the filtration and 18 the temperature which the filtration process was to 19 take place the filtration trials were carried out with 20 4" capsule filters. These filters are totally self 21 contained with the pleated filter media inside a 22 plastic capsule. The capsule has an inlet and outlet 23 hose barb connection and vents to allow purging of air 24 from the system. These filters therefore do not require 25 a filter housing. The chilling was provided by a 26 compressor which gave close temperature control to + 1 27 OC. The 3 stage filtration process was set up by joining 28 the 3 filters in series. 29 30 Three different types of membrane were tested, these 31 being; the PVDF membrane from Millipore, the cellulose 32 acetate membrane from Sartorius and the PES membrane 1 from PPD. In the initial 2" filter media disc trials, 2 the Millipore membrane had performed best in the 3 sensory analysis and so this was the membrane which was 4 chosen initially to do the temperature trials on. 5 6 The aim of the trials was to create end products using 7 membrane filtration, which were similar to the current 8 products which are filtered using the traditional plate 9 and frame filter press. This product also had to have 10 chill haze stability. These success criteria are more 11 vague than they appear due to the complexity of the 12 products. The three main compounds which are 13 responsible for chill haze instability are ethyl 14 laurate, ethyl palmitate and ethyl palmitoleate. Some 15 of the products were analysed for these esters during 16 the trials. 17 18 Another way to check for chill haze instability is to 19 put the products in the fridge and then check the 20 turbidity, this was also carried out. There is an old 21 method which used to be used by the Scotch Whisky 22 Research Institute (SWRI) (a NAMAS accredited 23 laboratory) to test for likelihood of chill haze 24 instability called the Invergordon Chill Haze test and 25 we also used this method. The problem with this method 26 is that it is for blended whisky and so maximum levels 27 are provided for blended whisky but not for Malt 28 whisky. However we looked at several past bottling 29 samples of 10 year old Malt whisky and took maximum 30 values from that. 31 1 The first trials to be carried out were the temperature 2 trials. The whisky was reduced and then filtered 3 immediately at four different temperatures. These 4 temperatures were 50C, 100C, 150C and 200C. Samples were 5 taken during these trials and then the tests were 6 carried out on these samples. The results can be found 7 in table 1 and these are graphically represented in 8 figure 1 and figure 2. 9 10 As can be seen from the results shown in figure 1, 11 there is a direct correlation between ester removal and 12 the temperature of filtration. 13 14 Figure 2 shows that there is also a correlation between 15 temperature and turbidity. From these results the 16 temperature was then chosen for the rest of the trials 17 as 10"C. 18 19 More trials were then carried out with the 4" capsules 20 to try to determine batch size. The results from these 21 trials were erratic and difficult to explain. When the 22 whisky is at full strength there is more ethyl 23 palmitate present than ethyl palmitoleate, when the 24 whisky is filtered this is reversed. In these sets of 25 results there was more ethyl palmitate present than 26 ethyl palmitoleate which would appear to indicate that 27 the whisky had not in fact been filtered. This happened 28 not only for the Millipore filters but also for the 29 Sartorius filters which were also trialed at this 30 point. The conditions were the same for all trials, 31 however the end product was not the same. As this 32 system was small it was not solid piped with stainless 12 1 steel pipework, braided plastic hose was used. This 2 hose was connected at the time of the temperature 3 trials. However for the next sets of trials the braided 4 hose had been left with some alcohol in it and we 5 believe that the hose reacted with the alcohol and 6 these leachates contaminated the membranes and led to 7 the filtration not being effective. 8 9 The contaminants may have partially blinded the 10 membrane and this would have made filtration more 11 difficult and caused an increase in pressure drop 12 across the membrane and the particles may have been 13 forced through. It was also difficult to control the 14 pressure in this scale of experiment due to the size of 15 the pump and the pressure was therefore higher than it 16 should have been. This however confirms that control of 17 the pressure in the final system will be very 18 important. This trial provided data of what the whisky 19 would be like if the filtration was not working. At 20 this stage we also tried other membranes. The cellulose 21 acetate membrane also appeared to work however the 22 contamination also affected it. The Millipore membrane 23 was selected due to the level of the technical support 24 which the company provided, and also that the membrane 25 had in the past produced whisky which had good sensory 26 results. However other membranes would also have been 27 an option and may be in the future. 28 29 Experiment 2: 1011 Filter Cartridge Trials 30 31 Due to the problems with the smaller scale trials it 32 was decided at this point to move to 10" cartridges 13 1 which provide five times the filtration surface area. 2 These trials were carried out using existing solid 3 pipework and an existing heat exchanger in the 4 reduction area. The heat exchanger did not provide such 5 close temperature control as the rented chiller, 6 however, once the glycol was adjusted to the correct 7 rate the temperature was constant. The trials also 8 allowed a higher flow rate because of the surface area. 9 Experiment 3: Varying Whisky Type Used 11 12 1. 10 Year Old Malt Whisky 13 14 Trials at this stage were only carried out with Millipore filters, however different whiskies were 16 filtered to see the effects of the filtration process.

17 The three types chosen for the trials were a 10 year 18 old Malt whisky, a blended whisky and Port Wood Finish 19 whisky. The Port Wood Finish was chosen as to show that the new filters did not strip too much colour out of 21 this whisky. These samples were analysed as before and 22 also some sensory work was carried out on them.

23 24 The whisky was filtered first through the guard filter (Lifegard), then through the prefilter (Milligard) and 26 finally through the membrane filter (Durapore). The 27 batch size varied but was approximately 900 litres for 28 the tests. 100 litres for this test was supposed to 29 represent approximately 2000 litres on a full scale process. This was from calculations made at lab scale, 31 however the flow rates from these filters would 32 determine the end full scale filter size. Samples were 14 1 taken after each filter at 100 litre intervals and 2 these were tested. The results from the 10 year old 3 Malt trials are shown in table 2. 4 The turbidity of the samples was taken and recorded at 5 room temperature, after being in the fridge for 2 days, 6 at 30C. The Invergordon chill haze stability value was 7 also measured and these values are shown in table 3. 8 9 At all stages the results were compared to a current 10 product sample and a judgement was made as to whether 11 the membrane filters had produced a similar product to 12 the current filtration method. From all the figures 13 from this batch of trials it is likely that the chill 14 haze stability would become a problem after 500 litres. 15 Therefore the filters appear to be exhausted after 500 16 litres which would equate to a 10 000 litre batch size. 17 This is smaller than was expected however the batch 18 size would probably be larger than this in reality if 19 the pressure was controlled more closely. However, if 20 we are running a 2 filter system then this batch size 21 would be fine as it would allow time for one filter 22 train to be cleaned while the other was being used. 23 24 Sensory work was carried out on this product also and 25 the difference was not found to be significant, the 2G tests were repeated to ensure that the difference was 27 not significant. The results from this sensory analysis 28 are as follows. 29 30 Triangle Test 1 31 32 Number of correct responses 13 is 1 Number of incorrect responses 9 2 3 Number of correct responses 10 4 Number of incorrect responses 12 6 Triangle Test 2 7 8 Number of correct responses 4 9 Number of incorrect responses 18 11 Number of correct responses 9 12 Number of incorrect responses 13 13 14 The number of correct responses required for significance is 12 therefore the first time that test 1 16 was run it was significant, however the second time it 17 was not significant. Test 1 was run on current product 18 against the product taken from the membrane filtration 19 process after 100 litres. This sample was taken when the temperature of the filtration process was too low 21 and this may explain the difference as the ester levels 22 in this sample were also low. The results from the 23 taste test were as follows 24 Taste Test A 26 27 Number preferring current 10 Year Old Malt 28 3 29 Number preferring membrane filtered 10 Year Old Malt 1 31 Number with no preference 32 2 16 1 Taste Test B 2 Number preferring current 10 Year Old Malt 3 5 4 Number preferring membrane filtered 10 Year Old Malt 2 6 Number with no preference 7 3 8 9 2. Blended Whisky 11 The next stage was to try the filtration process with 12 blended whisky. The whisky tested was H4325, this is a 13 common blend. Similar tests were carried out however 14 this product was not analysed for the long chain fatty acid esters as we had in the Malt whisky. The results 16 from the analysis can be seen below in table 4.

17 18 Sensory work was also carried out on the blended whisky 19 and the results can be found below.

21 Triangle Test 22 23 Number of correct responses 4 24 Number of incorrect responses 12 26 Again, the number of correct responses required for 27 significance is 10 therefore the difference has been 28 shown to be not significant. A taste test was also 29 carried out on this and the results are shown below. 30 31 Taste Test 1 Number preferring current blended whisky 2 3 3 Number preferring membrane filtered blended whisky 4 2 Number with no preference 6 1 7 8 3. Port Wood Finish Whisky 9 Trials were then carried out on Port Wood Finish 11 Whisky. As previously mentioned, the main reason for 12 this was to ascertain that the new filtration process 13 would not strip the pink blush from the whisky. These 14 trials were also very successful in that the membrane filters successfully reduced the turbidity of the 16 whisky to within acceptable levels and at the same time 17 did not strip the pink blush from the whisky. The chill 18 haze stability can be seen in table 5 however because 19 the trials were carried out from a single cask, no bottling samples were available and therefore no 21 sensory comparison could be carried out.

22 23 Also as the colour was important we measure the Dr 24 Lange colour and the Corning colour. The Corning colour is measured at 430 nm and 490nm and these two figures 26 are then subtracted. There is a specification for each

27 wavelength as well as the difference and these are 28 shown in table 6.

29 The results from the filtration trials are shown in 31 table 7.

32 18 1 The colour difference is within specification however 2 the actual colour measurements are above specification. 3 This is because the casks was above specification and 4 this would be blended away. However the results show 5 that the filtration process is not stripping the colour G out of the port wood finish whisky. 7 8 Sizing of Filter Train 9 10 For the sizing of the final filters the flow rate and 11 the upstream pressure of the filters was measured. The 12 upstream pressure gives a good indication of how much 13 resistance is being caused to flow by the membrane. As 14 the whisky is filtered the membrane becomes more loaded 15 with the esters and therefore it will take more 1G pressure to force the whisky through the filter. The 17 best measure of how clean the membrane is, is the 18 pressure drop across the filter. This is the difference 19 between the upstream and downstream pressure of the.20 filter and this will measure how difficult it is to get 21 the whisky through the membrane. This will be an 22 important measure on the larger scale as when the 23 pressure drop gets above a certain level there is the 24 possibility of break through on the filter. At this 25 point the esters which have been removed may be forced 26 into and through the membrane. This will obviously 27 affect the turbidity and the taste of the whisky. The 28 point at which break through occurs will help to 29 determine the batch size of the filters. This will be 30 estimated from the Invergordon chill haze turbidities 31 and from ester levels. During the trials the pressure 32 and the flow rate of the whisky was measured at

19 1 intervals and these results can he seen in table 8 2 below. 3 4 As can be seen from the results the pressure steadily 5 increases as the loading on the membrane increases. The 6 area of pressure drop across the filter is an important 7 one for this process and there must be trade off 8 between flow rate and upstream pressure. Further 9 research will show what our maximum process pressure 10 should be and what the preferred operating pressure 11 will be. This will be different for each of the filters 12 and should be decided at the pilot plant scale of the 13 filtration process. It may be possible to keep the 14 pressure drop more constant by introducing a back 15 pulsing system which will send a pulse of air through 16 the membrane when the pressure drop reaches a certain 17 level and this may clean off some of the surface debris 18 and allow the whisky to flow more freely. 19 20 The filter train was sized according to the flow rate 21 which is required and the flow rates which we achieved 22 through one 10" filter. The flow rate taken is not the 23 flow rate at the start of the run, it is the flow rate 24 which we can achieve at the end of a run. It is assumed 25 that the filter can be cleaned to at least 80% of it's 26 original performance and the sizing is based on this. 27 28 The housings are all oversized as 10" and 20" 29 cartridges can both also be used in the 30" filter 30 housing to reduce cost and surface area and also some 31 of the inlets can be blanked off to reduce the surface 32 area. This sizing gives greater flexibility. Also the 1 housings come in certain standard sizes and it is more 2 cost effective to buy the standard sizes than to get 3 specific sizes made to order. 4 5 Cleaning Of Filters 6 7 The cleaning of the filter cartridges was the next 8 challenge, as, if the filters cannot be cleaned up 9 properly then they may only be used once and not 10 changed less frequently. The first method of cleaning 11 which was tried was hot water, both forward and 12 backward flush. However the esters are not water 13 soluble and this did not work. After consulting a 14 textbook and the SWRI for help in this matter it was 15 decided that there were two possibilities. one was 16 detergent and the other was neutral spirit. Although 17 there was a limited amount of neutral spirit available 18 there was not enough to flush the filters, therefore 19 pieces of the membrane cut from an exhausted 4" capsule 20 were soaked in four different solutions, these were: 21 22 1.Hot water 23 2. Hot water and detergent 24 3. 50% neutral spirit 2S 4. 96% neutral spirit 26 27 The membranes were soaked overnight and then removed 28 from the solutions. After inspecting all the samples 29 with a scanning electron microscope (SEM) it could be 30 seen that all except the hot water cleaned the filters. 31 21 1 Next the 1011 cartridges were soaked in the neutral 2 spirit overnight, and this did remove some of the 3 esters however for effective cleaning a dynamic flush 4 is required. Using neutral spirit will create a number 5 of other problems such as accounting to Customs & 6 Excise for the spirit and if 96% spirit is to be used 7 then the area will need to be a zone 1 area and this 8 increases equipment and instrumentation costs and will 9 cut down on the choice of sensors which we can use. 10 Also there is the added cost of disposal and the 11 neutral spirit will require to be cleaned up if it is 12 to be reused. It therefore does not seem to be a very 13 practical solution. 14 15 The best solution would appear to be detergent and warm 16 water, with a final rinse with demin water. A company 17 called Henkel have been contacted about possible 18 suitable detergents from their range and their membrane 19 cleaning expert is presently looking into this. The 20 detergent solution would then go to the foul drain on 21 site. This would therefore appear to be the most cost 22 effective and easiest option. 23 24 More trials will therefore be carried out on the 10" 25 filter cartridges totry and clean the filters 26 effectively. This is very important to ensure that an 27 economically viable lifetime can be obtained from the 28 filters. The filters should be able to be cleaned to 29 obtain at least 80% of their original performance. The 30 lifetime of the filters is also dependent on the 31 filters not being run at extreme pressures and it is 32 better to clean the filters more regularly as running 22 1 them when they are heavily loaded with filtrate causes 2 stress on the filtration media. 3 4 Summary 5 6 1.The use of membrane filtration has a number of

7 advantages over the plate and frame filter press: 8 Firstly the whisky can be filtered at 6'C higher than 9 currently. Secondly the cartridges are cleanable and 10 therefore reusable and would have to be changed 11 probably every 3-4 months instead of after every 12 batch as is currently the case. The whisky can be 13 filtered immediately after reduction using the 14 cartridges and no recycling is necessary, as there is 15 no shedding of floc from the filters. After use the 16 cartridges can be air purged and this reduces the 17 hold up volume and makes the losses in product 18 considerably less than the current technology. The 19 cartridges can be used as part of a continuous 20 process. The process can be used in number of 21 different drinks industries where removal of chill 22 haze is a problem e.g. American, Canadian, Irish and 23 Japanese whiskeys, in brandies in wines and in 24 certain beers. 25 26 One of the key benefits of the multistage filter is 27 the fact that filtration can take place without the 28 need to chill the whisky. The commercial benefit of 29 the invention-being seen in the reduced energy costs 30 for the producer.

23 Table 1: Temperature Trial Analysis Results Current Millipore @ Millipore @ Millipore @ Millipore Product 5C 10C 15C 20C Ethyl Laurate (ug/ml) 9.3 7.6 11.9 16.1 20.8 Ethyl Palmitate (ug/ml) 1.5 0 1.4 1.7 2.3 Ethyl Palmitoleate (ug/ml) 2.1 0 1.6 2.7 3.6 Total Esters (ug/ml) 12.9 7.6 14.9 20.5 26.7 Turbidity (room temp), ppm 0.6 0.4 0.35 0.55 0.65 Turbidity (from fridge), ppm 2 0.8 1.2 5.6 12 Invergordon Turbidity, ppm 8.7 6.4 7.1 15.6 30.6 24 Lifegard Lifegard Lifegard Milligard 100 Milligard Milligard 1 OOL 500 L 900 L L 500 L 900 L Ethyl Laurate (ug/ml) 16.8 18.7 23.3 16.1 15.2 17.6 Ethyl Palmitate (ug/ml) 13 8.6 17.8 1.7 4.3 7.9 Ethyl Palmitoleate (ug/ml) 10.7 7.3 13.7 1.9 4.5 6.7 Total Esters (ug/ml) 40.5 34.6 54.8 19.7 24 32.2 Turbidity (room temp), ppm 12.8 0.75 0.95 0.25 0.35 0.55 Invergordon Turbidity, ppm 58 36 38 16.6 36 31.5 Table 2a. Analysis results: 10 Year Old Malt with 10" Cartridge Trials.

Current Durapore Durapore Durapore Durapore Product 100 300 600 900 Ethyl Laurate (ug/ml) 7.6 12.9 14.1 14.3 13.9 Ethyl Palmitate (ug/ml) 1.5 1.7 1.6 3 2.5 Ethyl Palmitoleate (ug/ml) 1.5 2.3 2.3 3.2 3.1 Total Esters (ug/ml) 10.6 16.9 1.8 20.5 19.5 Turbidity (room temp), ppm 0.45 0.35 0.2 0.25 0.3 Invergordon Turbidity, ppm 7.4 8.9 9.8 17.4 19.4 Table 2b. Analysis results: 10 Year Old Malt with 10" Cartridge Trials.

26 Turbidity at Turbidity in Turbidity at Invergordon Room temp Fridge 2 days 3 OC Chill Haze Durapore 100L 0.55 0.55 2.3 8.9 Durapore 200L 0.6 0.6 2.3 4.4 Durapore 300L 0.65 0.65 2.95 9.8 Durapore 400L 0.75 0.75 3.1 9.2 Durapore 500L 0.85 0.85 3.4 15.8 Durapore 600L 1.6 1.6 4.8 17.4 Durapore 700L 1.7 1.7 5.2 17.4 Durapore 800L 1.45 1.95 6.2 19.4 Current Product 0.7 0.7 2.85 7.4 Table 3. 10 Year Old Malt Chill Haze Stability Results 27 Turbidity at Turbidity at Invergordon Room temp 3oC Chill Haze Durapore 100L 0.8 2.5 4.4 Durapore 200L 0.8 2 3.8 Durapore 300L 0.8 2.2 3.4 Durapore 400L 0.8 2.6 4 Durapore 500L 0.8 2.35 4.15 Durapore 600L 0.8 2.4 4.1 Durapore 700L 1.15 2.75 3.8 Durapore 800L 1.15 1.8 3.55 Current Product 0.95 2.4 3.65 Table 4. Blended Whisky Chill Haze Stability Results 28 Turbidity at Invergordon Turbidity in Room temp Chill Haze Fridge 2 days Durapore 100L 0.75 46 5.8 Durapore 200L 0.7 32.5 2.35 Durapore 300L 0.75 37.5 2.8 Durapore 400L 0.8 40.5 2.8 Durapore 500L 0.75 42.5 2.45 Durapore 600L 0.7 38.5 2.35 Durapore 650L 0.7 44 2.1 Current 1.05 64 13.4 Product Table 5. Chill Haze Stability Results for Port Wood Finish Whisky 29 Minimum Target Maximum DrLange 12 12.5 13.5 430nm 0.2 0.23 0.26 490nm 0.23 0.25 0.27 430-490 -0.04 -0.02 0.03 Table 6. Colour Specifications for Port Wood Finish Whisky

Litres 430nm 490nm 430-490 DrLange 29 29 0 12.9 30 30 0 13 300 28 28 0 13 400 30 30 0 13 500 29 28 0.01 13 600 30 31 -0.01 13 650 26 27 -0.01 13 Table 7. Colour Measurements For PWF From Membrane Filtration Trials 0 31 Year Old Malt Pressure (bar) Litres Lifegard Milligard Durapore 0.1 0.5 0.8 0.1 0.6 1 300 0.1 0.7 1.2 400 0.3 0.75 1.25 500 0.4 0.8 1.25 600 0.5 0.8 1.4 700 0.5 0.8 1.4 800 0.55 0.8 1.5 900 0.6 0.85 1.5 Table 8a. Upstream Pressure of Filters, 10" Trials.

32 Blended Whisky Pressure (bar) Litres Lifegard Milligard Durapore 0.8 0.8 0.4 0.85 0.9 300 0.8 0.95 400 0.45 0.7 1 500 0.85 1.05 600 0.6 1 1.1 700 0.9 1.2 800 0.8 0.95 1.2 Table 8b. Upstream Pressure of Filters, 10" Trials.

33 Port Wood Finish Pressure (bar) Litres Lifegard Milligard Durapore 0.1 1.2 0.8 0.1 1.2 1 300 0.2 1 1.1 400 0.4 0.6 1.1 500 0.4 1 1.1 600 0.5 1 1.15 650 0.5 1.15 Table 8c.Upstream Pressure of Filters, 10" Trials.

34 Table 9: Blended Whisky Flow rate required 120 I/min Guard Filter Prefilter Membrane Filter Flow Rate (10") 30 I/min 10 I/min 10 I/min Filter Size 6 x 10" 5 x 30" 5 x 30" List Price of Filters E270 E1200 E1845 Housing Size 3 x 30" 7 x 30" 7 x 30" Housing Price E2000 E4000 E4000 Table 10: Malt Whisky Flow rate required 150 I/min Guard Filter Prefilter Membrane Filter Flow Rate (10") 20 I/min 8 I/min 7 I/min Filter Size 4 x 30" 8 x 30" 8 x 30" List Price of E540 F-1 920 E2952 Filters Housing Size 5 x 30" 12 x 30" 12 x 30" Housing Price 0000 E7000 U000 36

Claims (8)

1 CLAIMS 2 3 1. A continuous filtration method for filtering fatty

4 acid esters out of reduced strength spirit without 5 the need to significantly chill the filtrate prior 6 to filtration, the method consisting essentially 7 of a multi stage process with each stage 8 containing a filter, with the filters being 9 arranged in series. 10 11

2. An apparatus for continuous filtration during a 12 distillation process wherein the apparatus 13 consists essentially of at least two filters 14 arranged in series wherein each filter has a 15 different property and thus a different role in 16 the filtration process. 17 18

3. An apparatus as claimed in claim 2 which consists 19 essentially of only three filters arranged in series, wherein the three filters have different 21 filtration properties.

22 23

4. A method as claimed in claim 1 wherein the multi 24 state process is a three stage filtration process.

26

5. An apparatus as claimed in claim 3 wherein each 27 filter will be of a different type, with the first 28 filter of the series being a guard filter, the 29 secc>nd filter being a prefilter and the third 30 filter being a membrane filter. 31 37 1

6. An apparatus as claimed in claims 2, 3 or 5 2 wherein the final, or membrane filter, of the 3 series is composed of a material which is 4 hydrophillic.

6

7. An apparatus as claimed in claims 2, 3, 5 or 6 7 wherein the final membrane filter is composed of 8 polyvinylidene fluoride (PVDF).

9

8. A method as claimed in claim 1 or 4 wherein the 11 filtration is carried out with the filtrate at a 12 temperature of between 50C and 250C.