Cross-flow filtration of fruit juice 4 Pre-treatment4.1 Introduction 4.1 Introduction Process solutions using integrated hybrid membrane systems will often be the best solution to a specific industrial separation problem (Hägg, 1998). In this project two processes are being studied: Dead-end microfiltration with polymeric thread filters and Cross-flow microfiltration both with polymeric and ceramic membranes. The first process is intended to be used as a pre-treatment step and the filters have a pore size of 3, 5 or 10 mm. Aim The aim of this project is to substitute the filtration step with membrane technology combined with Filtomat thread filters. In this work the effect of some microfiltration parameters on the pressure during filtration, and the retention/transmission of proteins, sugar, turbidity causing compounds, phenols and colour is studied. Quality requirements Sour cherry and black currant juices (final product) should follow the quality parameters listed below according to Vallø Saft A/S (Denmark):
4.2 Materials and Methods 4.2.1 Sour cherry and black currant juice The experiments were performed using sour cherry and black currant juice supplied by Vallø Saft A/S (Denmark). This juice was produced according to the flow diagram shown in appendix 1 and 2, and obtained after the second and final centrifugation step (supernatant). The juice was stored at -20oC before use. Sour cherries (Prunus cerasus L.) named "Stevnsbær" were used to produce the juice. The species of black currant (Ribes nigrum) used was Ben Lemond. The berries were harvested in summer 1998 in Denmark. The characteristics of the centrifuged juices used in this work are shown in Table 4.1. Table 4. 1. Characteristics of centrifuged (twice) sour cherry and black currant juice.
Centrifuged sour cherry juice contains several different particles: phenols, proteins and different kinds of polysaccharides. Phenols and proteins can combine in complexes, which can dissipate. Phenols alone can also dissipate but only at low temperatures. Centrifuged black currant juice contains several different particles: phenols, proteins and different kinds of saccharides. Phenols and proteins can combine in complexes, which can dissipate. Phenols alone can also dissipated but only at low temperatures. 4.2.2 Filtomat thread filter Filtomat thread filters of 0.01 m2 from Filtration Ltd. (Israel) and supplied by Gustav Fagerberg A/S (Denmark) were used to run the experiments. Construction These filters or cassettes (Fig. 4.1) consist of a thread made of polyester wound round a plastic support. Two outlets (permeate) are located at the underside of the plastic support and these are connected to a plastic tube, which goes all the way through the support. This tube has pores, where the filtrate is collected and led to the permeate tank. The cassette is inserted into a closed container (filter housing).
Figure 4.1. Schematic representation of a Filtomat thread filter. 1. Plastic support, 2. Thread layer, 3. Permeate outlet. The thread can be wound with different degrees of tightness and this will result in different pore sizes. The pore sizes of the filters used in this work were 3, 5 and 10 mm. These numbers are statistical and highly dependent upon external parameters such as water turbidity (NTS), water quality (PPM) and type of suspended solids. A sample is checked from each production batch, and the size of the sample depends upon the size of the production batch (source Filtration Ltd.). 4.2.3 Experimental set-up The microfiltrations were run in a flexible filtration system assembled at the Department of Biotechnology at the Technical University of Denmark. The experimental set-up is illustrated in Figure 4.2. Figure 4.2. Schematic draw of the experimental set-up. A. Feed tank, B. Feed pump, C. Flowmeter, D. Two-way valve, E. Manometer, F. Filter housing, G. Filtomat thread filter, H. Frequency converter. Operation The feed tank and the juice were temperated before each experiment. A gear pump delivered the feed flow, which was measured by a flowmeter. The dead-end filtration can be controlled using 4 valves. Before filtration valves 1 and 4 were closed leading to the filling of the filter housing with juice. This was necessary to avoid air bubbles, which could influence the results. Valves 1 and 3 were closed during filtration. Valve 1 was open, valve 2 was closed and the gear pump stopped to stop the filtration. Manometers placed at the retentate and permeate side measured the pressure difference over the filter during filtration. 4.2.4 Equipment The following equipment was used during the experiments: Thermometer
4.2.5 General guidelines for the experiments The general aim of the experiments was to examine the potential of Filtomat thread filters and to study the effect of some microfiltration parameters on the pressure during filtration, and the retention/transmission of proteins, sugar, turbidity causing compounds, phenols and colour.
Experimental design MODDE (modelling and design) is a PC-Windows program for the generation and evaluation of statistical experimental designs (Umetri AB, 1992-1995). This statistical program was used to plan the factorial experiment and to evaluate the results in order to find out which factors had a real influence on the responses (Screening). Using factorial experiments has several advantages:
A 23 factorial experiment was designed varying the temperature, the flow,
and the pore size of the filters.
*= The high level of the flow for sour cherry juice was 70 and 80 for black currant juice. Table 4.2. Experimental design.
The filters were cleaned by flushing with water under the tap. The flow was 540 l/h and the filters were cleaned for 3 min. (1½ min. on each side). 4.2.6 Data treatment and Analysis 4.2.6.1 Data treatment All data was treated using the statistical program MODDE in order to evaluate which factors (temperature, flow and/or pore size) had an influence on the responses listed below.
4.2.6.2 Analysis 4.2.6.2.1 Pressure D TMP The pressure difference TMP was measured with the manometers described in the section on Materials and Methods, and calculated as shown in the theory section. TMP is a measurement for the membrane state. There are two phenomena, which can explain the increase of the TMP when running a filtration: a cake is being formed on top of the membrane, and / or the process fluid that is being filtered has reached a viscosity (up-concentration, temperature changes) which makes it necessary to increase the driving force in order to press the filtration medium through the membrane. TMP2 TMP was measured at the beginning (TMP1) and at the end of thefiltration (TMP2). The difference between these was chosen as a response (D TMP).The response TMP2 can give an indication of how easy it is to filter sour cherry juice under the given conditions. 4.2.6.2.2 Water permeability The water permeability (WP) gives an indication of the state of the thread filters, and it is defined as the amount of ultrafiltered water (V) passing through the surface of the filter (Afilt) in a certain time (t) and at a certain pressure difference (TMP). The water permeability is calculated by the equation: Wp (l/m2/h/bar) = V (l) * 3600 (sec/h) / Afilt (m2) / t (sec) / TMP (bar) All filters had previously been used between 3 and 5 times. The water permeability was
measured for these filters before filtration and for new filters with the same pore sizes.
4.2.6.2.3 Protein content The total protein content before and after each filtration was determined with a Macro-N (Foss Electric, Denmark). 4.2.6.2.4 Sugar content The sugar content was analysed using a Refractometer. 4.2.6.2.5 Turbidity The turbidity was analysed using a Nephla laboratory turbidity photometer conforming to DIN EN 27027 and ISO 7027. All samples were diluted to a oBrix of 3 before the measurements were performed. 4.2.6.2.6 Total Phenol content Principle Phenolic substances are oxidised by the Folin-Ciocalteu reagent, which contains a mixture of phosphotungstic acid (H3PW12O40) and phosphomolybdic acid (H3Pmo12O40). The reagent becomes partly reduced resulting in the production of the complex molybden-tungsten blue, which is measured spectrophotometrically at 765 nm (Singleton and Rossi, 1965). Equipment Perkin Elmer l 2 UV/VIS
spectrophotometer Reagents olin-Ciocalteus phenol reagent from Merck diluted 1:10 with double distilled
water Procedure 1.0 ml Folin-Ciocalteus phenol reagent diluted 1:10 with double distilled water was added to 0.2 ml sour cherry juice. 0.8 ml of 7.5 % (w/v) sodium carbonate was added to develop the colour and the mixture was mixed on a whirl mixer. The mixture was then left for 30 min. with caps on. After mixing again with a whirl mixer the absorbance was read at 765 nm, using double distilled water for background correction. If the absorbance exceeded 1, the sour cherry juice was diluted with double distilled water, and the assay was carried out again. The concentration of total phenols was calculated from the standard curve obtained by subjecting known amounts of gallic acid solutions (0, 5, 10, 20, 40, 60, 80, 100 mg/l gallic acid) to the same treatment as the sour cherry juice and blank samples. Results were expressed in gallic acid equivalents (GAE). 4.2.6.2.7 Colour The colour of the juice was measured in two different ways for sour cherry juice:
Vallø Saft method Colour measurement (Vallø Saft) Principle The anthocyanins are responsible for the red/purple colour of the juice (Grassin and Fauquembergue, 1996). The red colour is measured with a spectrophotometer at 520 nm where red coloured liquids have the maximum absorption. The value should be between 0.700 and 1.500 after diluting according to this assay for this harvest. Equipment Perkin Elmer l2 UV/VIS spectrophotometer Reagents Trisodium citrate dihydrate, pure, Merck Procedure A buffer of pH 3 was prepared by mixing 18 g of Trisodium citrate dihydrate with 55.5 g
Citric acid monohydrate in 1 litter double distilled water. After mixing, the absorbance of the diluted sample was measured against the buffer at
520 nm. pH differential method Principle Equipment Perkin Elmer l 2 UV/VIS spectrophotometer Reagents Sodium acetate trihydrate, p.a. Merck Procedure The two buffers are prepared as described below and afterwards adjusted to the exact pH: pH 1.0 buffer: 125 ml of 0.2 N Potassium chloride (136 g/l) pH 4.5 buffer: 400 ml of 1 M Sodium acetate (136 g/l) A juice sample is first diluted with the pH 1.0 buffer. The absorbance read at 510-540 nm should be lower than 1 and preferably between 0.4 and 0.6. The sample is diluted to the same degree with the pH 4.5 buffer. Scanning the two diluted samples from 350 nm to 700 nm gives two curves which maximum adsorption is between 510-540 nm. The absorbance at 700 nm should for each curve be 0 if the juice sample does not contain any haze. Subtracting the value read at 700 nm from the maximum absorbance is a correction for haze. The difference in the maximum absorbance will be proportional to the anthocyanin content, which can be calculated with the following equation based on the Lambert-Beers law: C (mg/l) = A/e L * MW *103 * Dilution Factor Where C is the concentration of the major anthocynin pigment in the fruit
The concentration of the major anthocynin pigment in a fruit is representing the concentration of the anthocynin pigment in the fruit. In sour cherry cyanidin-3-rutinoside is the major anthocynin pigment. It should be emphasised that the pH differential method is a measure of the monomeric anthocyanin pigments and results may not seem to be correlated with the colour intensity of the juice samples as they are judged visually. The colour was measure at 640 nm using a Perkin Elmer l 2 UV/VIS spectophotometer for black currant juice. All samples were diluted to a ° Brix of 3 before the measurements were performed. 4.3 Results and discussion 4.3.1 Sour cherry juice Since the area of the filters is 0.01 m2, the flux in these experiments was between 2000 l/h/m2 (low level of the flow: 20 l/h) and 7000 l/h/m2 (high level: 70 l/h). Water permeability Table 4.3 shows the results for the Water permeability measurements for all filters before filtration at a flow of 30 l/h and the standard deviations (SD). Table 4.3. Water permeability (Wp) at a flow of 30 l/h.
At this flow there were no significant differences between the Water permeability for the new Filtomat thread filters with a pore size of 5 and 10 m m. The same phenomenon was observed for the used filters with a pore size of 5 and 10 mm. Lower values were obtained for the filters with a pore size of 3 mm and it can be deduced that the TMP increases when the pore size of the filter decreases from 10 mm or 5 mm to 3mm. The new filters of all pore sizes had lower Wp values than the same filters
after they have been used 3 to 5 times. An explanation for this could be that the threads
in the used filters have loosened. This result emphasises the importance of breaking
in the filters before use to avoid unnecessary uncertainties about the results. The results of the screening experiments run with sour cherry juice are shown in Table 4.4. Table 4.4. Results for the screening experiments run with sour cherry juice
Changing the levels of the three factors had a slight effect, when considering the
values of the quality parameters for the different experiments. Effect of the factors Table 4.5 shows the effect of the factors and their interactions on the responses: - An empty box shows that the factor does not have a significant effect on the response, - * Means that the factor has a significant effect on the response when analysing the data with a 95% confidence level, - ** Means that the factor has a significant effect on the response when analysing the data with a 99% confidence level, - *** Means that the factor has a significant effect on the response when analysing the data with a 99.9% confidence level,
Table 4.5. Effect of the factors on the responses. The effect in changing each factor from the low level to the high level will in the
following be considered. The increase on turbidity when performing at higher temperatures was probably not caused by the transmission of phenols or proteins since the temperature factor had no significant effect on either of the responses. The turbidity could be caused by b -glucans released by the fungus Botrytis cinerea when processing the berries. This fungus, which is related to the contamination of red berries, secrets a beta 1,3-1,6 linked glucan. This gum dramatically reduces the filterability and the clarity of the juice (Grassin and Fauquembergue, 1996). The increase on the flow from 20 to 70 l/h only affected the response TMP2. TMP2 increased with increasing flow. D TMP and TMP2 were lower when filtering with
bigger pore sizes, indicating that more particles passed through the filter. The factor
pore size had however no significant effect on any of the other responses. The interaction between the temperature and the pore size of the filters had a significant effect on the colour measured following the method from Vallø Saft; at higher temperatures and bigger pore sizes, more colour passed through the filter. The interaction between the flow and the pore size of the filters had a significant effect on the colour measured following the method from Vallø Saft; at higher flows and bigger pore sizes, less colour passed through the filter. This result is not logical and it indicates that this method (Vallø Saft) may not be reliable. Comparison to juice filtered at Vallø Saft Table 4.6 shows the comparison of sour cherry juice filtered at Vallø Saft following the production process used in industry and juice obtained at the Department of Biotechnology (DTU) by using Filtomat thread filters (experiment 7). Experiment 7 was run at high flow, at low temperature and through the 10 mm filter. This experiment was chosen since higher fluxes result in lower costs and low temperatures reduce microbial growth during processing keeping the quality of the juice for a longer period. Table 4.6. Quality parameters for juice produced at Vallø Saft A/S and at the Department of Biotechnology.
All quality parameters for the juice filtered with Filtomat thread filters were higher
than the ones for the ultrafiltered juice from Vallø Saft. The ultrafiltered juice from Vallø Saft had been frozen before the analyses were run which could influence the results. When cherry juice is frozen, phenol-protein complexes precipitate which could explain why the colour is lower and the turbidity is higher than the quality limits set by Vallø Saft. Turbidity is not an independent parameter when measured in sour cherry juice. When the sour cherries are crushed, 25 % of the stones are mashed and chemical compounds like benzaldehyde or hydrogen cyanide are liberated. Benzaldehyde is an aromatic oil which increases the value of the turbidity. A clear filtered juice could therefore have a turbidity of 20-40 FNU (source Vallø Saft). 4.3.2 Black currant juice Since the area of the filters is 0.01 m2, the flux in these experiments was between 2000 l/h/m2 (low level of the flow: 20 l/h) and 8000 l/h/m2 (high level: 80 l/h). The results of the preface screening experiments are shown in Table 4.7. The protein content, sugar content, phenols and colour were found more or less in the same amount as in the centrifuged juice, whereas the turbidity was reduced three times by filtering the centrifuged juice. The turbidity, the sugar content and the colour for all the experiments were within the quality limit, which means that the quality of all filtered juices was excellent. It is clear that the use of Filtomat thread filters improves the quality of the juice when filtering the centrifuged black currant juice from Vallø Saft A/S. Table 4.7. Results for the preface screening experiments run with black currant juice.
Effects of the factors Table 4.8. Effect of the factors on the responses
The effect in changing each factor from the low level to the high level will be
considered in the following. When the filtration flow was increased from 20 l/h to 80 l/h, the final TMP increased as well as the turbidity in the filtered juice. Probably some particles are pressed through the filter when the flow is increased. The increase on the turbidity when running at higher temperatures and flows was probably not caused by the transmission of phenols and proteins, since these factors have no significant effect on either of the responses. When Filtomat thread filters with a pore size of 10 mm were used, more particles could pass through and that explains why the turbidity and the phenol content were higher. The interaction between the temperature and the pore size of the filters had a significant effect on the final TMP (TMP2) and the transmission of the colour. Looking at the results related to the chemical composition of the juice (Table 4.7: protein, sugar, phenol and colour), no real differences are observed when changing the level of the factors. When the turbidity values are considered, no real differences are observed either, even though they statistically differ from each other. All the values are also under 5 FNU, which means that the quality of the juice is excellent. The opposite phenomenon is observed for the response D TMP. The four experiments carried out at a high flow (80 l/h) resulted in higher D TMP, but flow was not found to have a significant effect on D TMP. This result could indicate a correlation between flow, TMP and fouling. Comparison to juice filtered at Vallø Saft Table 4.9. Quality parameters for juice produced at Vallø Saft A/S and at the Department of Biotechnology.
Protein, sugar, phenol and colour were found in the same range for both juices. Turbidity was slightly lower for the juice produced at the Department of Biotechnology. The results show that a microfiltration after this pre-treatment should not be necessary. 4.4 Conclusions The conclusions drawn from these screening experiments are:
4.5 Filtration of large amounts of juice prior to centrifugation 4.5.1 Objectives Aim The general aim of these experiments was to investigate if it was possible to replace the vacuum filter with a Filtomat filter or a Cross filter when processing black currant and cherry juice. The juice used was taken directly after the pressing step or after the clarification step. 4.5.2 Materials and Methods The experimental set-up was the same as the one used in the previous experiment with black currant and cherry juice. The black currant and cherry juice were filtered either through a Filtomat filter (20 mm) 0,01 m2 or a Cross filter (25 mm) 0,02262 m2. The temperature was 22oC, and the flow was kept constant at 15 l/h using either a flowmeter or a frequency converter. 4.5.3 Results All results are presented in Table 4.10 and 4.11. Filtrating pasteurised cherry juice after the pressing step or cherry juice after the clarification step was not possible. This was due to the large particles. Filters with a bigger pore size should be used either in a single formation or in series. Table 4.10. Results for sour cherry juice.
Filtrating black currant juice after the clarification step was possible with the filters used, but not profitable. Filters with a bigger pore size should be used either in a single formation or in series. Table 4.11. Results for black currant juice.
1 It was not possible to measure the flow because of the large particles. The flow was controlled using a frequency converter. The experiments was terminated at a TMP > 0,75 bar.2 The experiment was terminated at TMP > 0,75 bar. 3 Since the juice easily went though the Filtomat filter (20 mm) and Cross filter (25 mm), Filtomat filter with a smaller pore size (3 mm) was used. When filtrating black currant juice after the pressing step, Filtomat filters with a smaller pore size (3 mm) was used since the juice easily went though the Filtomat filter (20 mm) and Cross filter (25 mm). The filtration of black currant juice after the pressing step was further investigated using a response surface model (CCF). The volume chosen for each of these experiments was 20 l, the filter
used was a Filtomat filter 0,01 m2 (3 mm), there was
no circulation and samples were collected after 2, 8, 16 and 20 l. The two factors
investigated in this experiment and their responses were:
The experimental design is shown in Table 4.12. Table 4.12. Experimental design.
A new filter was used for each experiment, and the water permeability was tested prior to used, to ensure that this variable did not have an effect on the filtration performance (Appendix 4). Effect of the factors The effect of the factors on the responses is shown in Table 4.13. Table 4.13. The effect of the factors on the responses after
filtrating
The effect of the factors on the responses should have been evaluated after filtrating 20 l. However, it was impossible to hold the temperature at 4oC during the whole experiment. Therefore, the effect was evaluated after filtrating 16 l. Appendix 11 shows that the temperature raised rapidly after approx. 16 l had been filtrated. This is also seen in Appendix 5. Changing the temperature from the low to the high level had a significant influence on the TMP (-**), FNU (***) and Phenol (*), whereas changing the flow from the low to the high level had a significant influence on TMP (-**) and FNU (***). For TMP and FNU, the effect of the factors was more significant when filtrating larger amounts of juice. For example, flow and temperature did not have a significant influence on the FNU after 2 l had been filtrated, after 8 l and 16 l the factors had a significant influence on level 95 % and 99,9% respectively, for the turbidity (FNU). This phenomenon was not observed for the responses protein and phenol. During filtration an inner cake was created on the Filtomat filters, and this improved the filtration. The content of phenol, protein and FNU in the permeate decreased with increased amount of juice filtrated. An example of this is shown in table 4.14. The rest of the results are found in appendix 12, where 1,2, and 3 stands for 2 l, 8 l, and 16 l filtrated juice. Table 4.14. Effect of the inner filter cake.
The change of TMP as a function of the filtrated volume for the different flows (15 l/t, 30 l/t, 45 l/t) and temperatures (4 oC, 13 oC, 22 oC) is shown in Appendix 5, 6 and 7. The change of TMP as a function of the time for the different flows (15 l/t, 30 l/t, 45 l/t) and temperatures (4oC, 13 oC, 22 oC) is shown in appendix 8, 9 and 10. The diagrams indicate, that the lower the flow, the higher the TMP was. The TMP also increased faster at the lower flow especially at low temperatures. An explanation for this phenomenon is that at higher flows, the membrane is more opened, and larger particles can therefore be pressed though. The inner cake will not be created in the same extent at high flows, and the resistance in the filter will therefore be smaller. The diagrams also indicate, that at a higher temperature the resistance in the filter is smaller, and the reason for this is that the solubility of the particles has changed with the temperature. The experimental results back up these theories (appendix 12). 4.5.4 Conclusion It is not possible to filtrate pasteurised cherry juice after the pressing step or
cherry juice after the clarification step through Filtomat filters with a pore size of 20
mm and through a Cross filter with a pore size of 25 mm. It is not profitable to filtrate black currant juice after the clarification step
through a Filtomat filter with a pore size of 20 mm and through
a Cross filter with a pore size of 25 mm. Table 4.15. Quality comparison between vacuum filter and Filtomat filter.
|