Air-suspension particle coating is a process by which thin coatings are applied to powder particles. The coatings can be formulated to act as permeable barriers to increase powder shelf-life or to impart controlled release character. The ultimate objective of a coating operation is to produce individual particles, each with a well-controlled, even coating. This project was focused on the air-suspension coating of fine powders of ~100 µm in diameter for the dairy industry. Despite the widespread use of the technology in the pharmaceutical industry, its use in the food industry has been limited. Little is known about the fundamental mechanisms, and so published work to date is product and equipment specific and is statistical in the way the experimental design and analysis has been approached. This 'black box' approach is time consuming and costly. Better methods based on an understanding of the physical and chemical mechanisms are needed to deal with the numerous products and constantly changing formulations typical of the dairy industry. This thesis proposes a new approach to air-suspension particle coating research. The basis of this 'micro-level process approach', is to deconvolute the complex coating process into smaller manageable parts based on classical physical phenomena for which descriptions already exist. The thesis identifies and develops an understanding of the key micro-level processes controlling coated product quality and process performance. Four were selected for further study: drying, droplet impact and spreading, and stickiness which encompasses the two key micro-level processes of droplet impact and adherence and inter-particle agglomeration. They were studied separately to deconvolute the variable effects and interactions. Kinetic data were collected for the drying droplets containing maltodextrins, whey protein isolate and gum arabic. A mathematical model, based on 'ideal shrinkage' was developed to predict the drying kinetics of single droplets with particular interest in the development of the surface glass transition temperature. The model accurately predicted the kinetics until significant morphological changes occurred in the droplet. To better predict the kinetics late in the drying process, the droplet radius was set to be constant at a time based on the surface proximity to the surface glass transition temperature (critical X concept). This was done to arrest droplet shrinkage in line with experimental observations and to more accurately depict the drying of high molecular weight, amorphous glass forming polymers. After this point, a new flexible calculation scheme was used to better predict the variation in internal droplet structure as either a dense, 'collapsed shell' structure or a 'dense skin-porous crumb' structure. Further study should focus on the surface and internal droplet structure (porosity and mechanical integrity) development during drying, particularly the conditions leading to the arresting of the droplet radius and the subsequent rate of skin thickness progression. The critical X concept was used to make industrial-scale predictions of the optimum drying conditions that ensure maximum droplet impact and adherence efficiency and minimum inter-particle agglomeration in a Würster-style coating operation. This enabled the prediction of two key design parameters, the nozzle distance from the powder impact point and the Würster insert height. The span in design parameters showed that there is significant opportunity for design optimisation based on the critical X concept. A probe tack test was used to map the level of stickiness of droplets of different coating materials as they dried. As skin formation progressed, the stickiness passed through a maximum, in most cases to arrive at a point at which the droplet was no longer sticky at all (non-adhesive state). The maximum point of stickiness represents the ideal state to ensure successful droplet-substrate impact and adherence. The minimum point of stickiness represents the ideal state to prevent unwanted inter-particle agglomeration. The time interval between the onset of stickiness and the non-adhesive state was particularly dependent on the addition of plasticisers, but also on the formulation and the drying air conditions. Future work should look to establish a possible relationship between the surface glass transition temperature and the probe tack test stickiness measurements. The impact and spreading of droplets containing maltodextrin DE5 on to solid anhydrous milkfat was studied using a high speed video camera. It was found that the final spread diameter was able to be fixed close to the maximum spread diameter by using surfactants, thus avoiding significant recoil. Because existing literature focuses on predicting the maximum spread diameter, this work defines a need for adequate prediction methods for the final spread diameter, as this is the significant parameter in coating applications. Formulation and operating guidelines were established to independently optimise each micro-level process. These were used in a series of population based coating experiments in a pilot-scale Würster coater. This study highlighted the limited flexibility of the standard 'off-the-shelf' Würster coating apparatus for the coating of fine sized dairy powders. Because of this, the validation of the guidelines were inconclusive and optimisation could not be carried out. Further validation work is required on a custom-built apparatus for dairy powders. This work has advanced the fundamental knowledge of the coating process and is independent of material, equipment and scale. This knowledge, based on physical and chemical mechanisms, can be used to develop coating formulations and identify optimum process conditions for successful coating in less time and at less expense than is current practice. The next step is to put the guidelines into practice and craft the engineering of a continuous coating apparatus for dairy powder applications.

Content removed due to copyright restrictions: Anderson, R.C., Wilkinson, B. & Yu, P.L. (2004). Ovine antimicrobial peptides: new products from an age-old industry. Australian Journal of Agricultural Research, 55(1),69-75. Anderson, R.C. & Yu, P.L. (2003). Isolation characterisation of proline/arginine-rich cathelicidin peptides form ovine neutrophils. biochemical and biophysical research communications 312(4), 1139-1146. Anderson, R. C., Hancock, R.E.W. & Yu, P.L. (2004). Antimicrobial activity and bacterial membrane interaction of ovine-derived cathelicidins. Antimicrobial Agents and Chemotherapy, 48(2), 673-676. Anderson, R.C., Haverkamp R. & Yu, P.L. (2004). Investigation of morphological changes to S. aureus induced by ovine-derived antimicrobial peptides using TEM and AFM. FEMS Microbiology Letters, 240(1), 105-110. Anderson, R.C. & Yu, P.L. (2005). Factors affecting the antimicrobial activity of ovine-derived cathelicidins against E. coli 0157:H7. International Journal of Antimicrobial Agents 25, 205-210 / The aim of the research presented in this thesis was to investigate the properties of the antimicrobial peptides found in ovine blood, in order to assess their potential as a high-value product. Due to the large number of lambs and sheep that are slaughtered New Zealand (approximately 25 million lamb and 5 million sheep per year), there are considerable volumes of ovine blood available for processing (approximately 40 million litres per year). Currently this blood is dried and sold as a low value product. The first objective of this research was to purify and characterise the antimicrobial peptides isolated from ovine neutrophils. A number of proline/arginine-rich peptides, as well as two small fragments of larger proteins, that displayed antimicrobial activity were identified. The second objective of this research was to investigate the mechanism of action of ovine antimicrobial peptides. For this investigation, three ovine peptides, α-helical SMAP29 and proline/arginine-rich OaBac5mini and OaBac7.5mini, were synthesised. Of these, SMAP29 was the most potent. The three peptides all bound Gram-negative bacterial LPS and caused the outer membrane to be permeabilised. SMAP29 caused significant depolarisation of the cytoplasmic membrane that led to cell lysis. However, the other two peptides only caused slight depolarisation of the cytoplasmic membrane, which indicates that they probably passed through the membrane to interact with the inner cellular contents. The third objective of this research was to investigate the morphological changes to bacterial cells induced by the ovine antimicrobial peptides. Transmission electron microscopy and atomic force microscopy confirmed that SMAP29 caused significant damage to the membranes of bacterial cells and induced cell lysis; whereas, OaBac5mini caused minor alterations to the bacterial membranes but did not induce cell lysis. The fourth objective of this research was to determine the effect of the environmental conditions on the activity of the peptides. The peptides were very stable over a range of pH values and when heated to temperatures up to 80°C. The activity of the peptides decreased slightly in the presence of monovalent cations and was inhibited by the presence of divalent cations. The peptides were significantly more active in combination than individually, and they were strongly synergistic with polymyxin B, a peptide antibiotic. The final objective of this research was to develop a pilot-scale extraction process for the isolation of antimicrobial peptides from ovine blood. The laboratory-scale process was simplified and adapted to design a process that could be used industrially. The crude pilot-plant extract was active against a broad-range of food pathogens and disease causing organisms. The antimicrobial peptides found in ovine blood have the potential to be used as biopreservatives for chilled lamb products, or in a topical cream for cuts and grazes; therefore it is recommended that further research is carried out to investigate the above applications and. if successful, the feasibility of commercialising the technology.

Pages A58-A66 are missing from original but content appears complete. / This project investigates the time-dependent rheological behaviour of fresh and reconstituted milk concentrates. New experimental protocols, including sampling and measurement techniques, as well as equipment calibration and data analysis procedures were developed for both the industrial surveys and controlled rheology experiments. The controlled rheology experiments were mainly carried out on reconstituted milk concentrates to minimise the variation in composition of fresh milk. A new recombination rig was built which could minmise the age thickening process by mixing at 35°C and recirculating at 40,000 s-1 to break down the structure completely. This is the essence of this project, where age thickening is studied from a starting point of a filly broken down structure in contrast to past research. Using this method, the replicate milk concentrate samples had reproducible rheological behaviour, with a maximum reproducible error of 10%. Age thickening involves two stages, a slow initial increase in apparent viscosity with storage time, followed by a sudden sharp rise which marks the onset of gelation. The age thickening behaviour of milk concentrates is dependent on the processing variables prior to rheological measurement. These include solids content, shear rate and temperature during recombination, shear rate and residence time in the plate heat exchanger, and most importantly the raw material. The viscosity at the gelling point is an important characteristic of the age thickening process, and seems to depend mainly on the powder used, rather than the process treatments applied. Industrial surveys exhibited similar trends, even under varying conditions that could not be completely controlled. It is proposed that two types of age thickening phenomena can be distinguished: type I occurs below the temperature at minimum viscosity (65°C in this case), where weak interactions take place between the casein micelles; type II occurs above the temperature at minimum viscosity, where additional stronger covalent bonds are formed, primarily due to the denaturation of whey proteins. No mathematical model for the time-dependent rheology was developed. However, some important issues that must be taken into account during modelling were discussed. The results showed that the age thickening process is more complex than had previously been envisaged. The knowledge of the interactions between the operating conditions, rheology of fresh concentrates and powder properties should be invaluable in the improvement of plant efficiency and quality control.

Fouling of heat treatment equipment in the dairy processing industry is an expensive and persistent problem. The objective of this work was to develop a better understanding of the mechanisms of dairy fouling in heat exchangers and identify methods to control this build-up. This was part of a larger project investigating the interaction between spore-forming thermophilic bacilli (thermophiles) contamination and fouling deposits on internal surfaces of equipment. Two systems were developed to monitor the onset and build-up of fouling on the internal surfaces of two research heat exchangers. The first used a commercial sensor to measure the local heat flux and the temperature on the hot side of a plate type heat exchanger. The heat transfer coefficient was calculated and normalised with its value at the start of the run to reflect the contribution of fouling deposits to the thermal resistance, thus giving a real-time estimate of the rate of fouling. The second system used an energy balance over a tubular type heat exchanger and measured inlet and outlet temperatures to estimate the overall heat transfer coefficient thus giving a global measurement of fouling over the tubular heat exchanger. In both systems the plot of normalised heat transfer coefficient over time often stayed constant over an induction period, which was followed by a falling period indicative of growth in the fouling layer thickness and/or mass. Each system was validated by comparing the final value of the normalised heat transfer coefficient with direct measurements of fouling made at the end of a run namely: fouling deposit height for the local measurement and fouling deposit mass for the global measurement. The normalised heat transfer coefficient reported by each system correlated well with the corresponding direct measurement of the fouling layer. An important factor identified in this study was the effect of air bubble nucleation on fouling deposits. It was shown that bubbles that formed on the heated surface greatly reduced the length of the induction period to a matter of seconds rather than hours, as found in previous studies of fouling in the absence of surface bubbles. The rate of fouling was also enhanced while the bubbles remained at the surface. The structure of bubble type fouling layers was linked to the behaviour of the bubbles at the heated surface. Visual observations of these bubbles showed evidence of growth, vibration and coalescence during their period of attachment to the heated surface. Deposits from bubble type fouling consisted of all solid components found in the original milk solution, except lactose, in approximately the same ratio. By contrast fouling deposits reported in the literature with systems operating under the traditional protein denaturation mechanism were reported to consist mainly of whey proteins. Bubble induced fouling can be limited in a number of ways, the most effective being to maintain a high operating pressure in the equipment to ensure nucleation does not occur. Experiments conducted in this study showed that a pressure of 130 kPa.g was sufficient to suppress all bubble nucleation at the heated surface at a temperature of 90°C. Another method identified was the use of high linear fluid velocities to entrain any surface bubbles into the processing stream immediately upon nucleation. Linear velocities above 1.0 m/s were shown to achieve this goal in the miniature plate heat exchanger tested. However, this method is only partially successful because the local linear velocity varies with position in heat exchange equipment of complex geometries and can drop below the mainstream average velocity causing surface bubbles to form, especially in recirculation regions behind flow obstacles. A more reliable method, in situations where high operating pressures could not be used, involved conditioning the heated surface with a thin protein layer during the first few minutes of a run. Conditioning the surface resulted in bubble suppression even at high temperatures and low pressures, thus greatly extending the length of the induction period. Trials performed in this study showed that the addition of a proteolytic enzyme produced by psychrotrophic microbes greatly increased fouling. The enzyme destabilised the caseins which could attach directly to the heat exchange surface independently from the bubble fouling mechanism. Thus the quality of the milk is another important factor to consider. However, the addition of enzymes produced by thermophilic bacilli isolated from milk powder plants did not increase fouling. A theory describing the air bubble induced fouling mechanism is presented along with recommendations on how to reduce this fouling contamination in processing equipment.

Cheese is one of the most well known food products in the world dating back to the 8th century B.C. There are more than 2000 varieties of cheese that are manufactured all over the world. Feta cheese is a soft white cheese with a salty and slightly acidic taste, which has originated from Greece. Most of the feta cheese manufactured in Greece is consumed locally, the migration of greeks to other parts of the world led to a demand for feta cheese outside of Greece. The spreading of the popularity of feta cheese to other ethnic groups in different parts of the world resulted in the high demand for feta cheese worldwide. The modern and most efficient method of feta cheese production involves a membrane filtration method, known as ultrafiltration. The ultrafiltration process utilises pressure as a driving force to concentrate milk by removal of water and small dissolved molecules. Hollow fibre and spiral wound ultrafiltration membranes are the two types of membranes that are commonly used for cheese production. An extensive amount of research exists on the implementation of ultrafiltration to improve the efficiency of the cheese making process and the performance of the membranes. However, limited research has been conducted on the comparison of the hollow fibre and spiral wound membrane performance in the cheese making process. The objective of the research was to determine if the hollow fibre membranes used at Puhoi Valley Cheese can be replaced with spiral wound membranes without compromising the quality of cheese produced. In order to achieve the objective, feta cheese was produced using hollow fibre and spiral wound ultrafiltration pilot plants. The operating performances of the hollow fibre and spiral wound membrane units were compared. To ensure that the quality of cheese is maintained, the cheese manufactured on the pilot plant units was analysed in terms of composition, microbiology, texture and sensory properties. The cheese made using the hollow fibre membrane pilot plant was compared with the reference sample from Puhoi Valley Cheese as they use hollow fibre membranes to produce feta cheese. The cheese made from the spiral wound membrane unit was also compared to that made by the hollow fibre membrane pilot plant unit. The operating parameters such as the inlet and outlet pressure, pressure difference along the membrane, transmembrane pressure, flow rate, recycle rate (bleed off rate), temperature and the run time were recorded. The operating parameters of the hollow fibre and spiral wound runs were compared with the data from Puhoi Valley Cheese. The quality of cheese made on the hollow fibre and spiral wound pilot plant units were evaluated in terms of composition, texture, microbiology and sensory properties. The composition was defined by the fat, protein, total solids and salt contents. The fat content was determined by utilising the modified Schmid-Bondzynski-Ratzlaff method, protein by the Kjeldahl method, total solids by using the air drying oven and salt percentage by the volhard method. The texture of the cheese was determined by the fracturability and hardness from the compression curve generated using the single bite compression test. The microbiological testing was performed according to New Zealand testing methods for E.Coli, Staphylococcus aureus, coliforms and yeast and mould. The difference from the control method was utilised for sensory evaluation. The acid degree value method was used to determine the lipase activity in feta cheese. It was found from the composition, texture and sensory analysis that the cheese from the hollow fibre pilot plant was different from the cheese manufactured at Puhoi Valley Cheeses (PVC). The spiral wound cheeses were also found to be different to PVC cheese, however the spiral wound cheeses and the pilot plant hollow fibre cheese were the same. The differences between both the pilot plant cheeses and PVC cheese were in terms of the fat, salt, moisture contents and the lipase activity in the cheeses. The fat content in the hollow fibre and spiral wound pilot plant cheeses are lower in comparison to the PVC cheese. This difference in fat content is considered to be due to the difference in the fat to protein ratio of the milk concentrated on the pilot plant and the PVC ultrafiltration system. The lower fat content resulted in firmer cheese than PVC due to more cross linking between the protein strands in cheese. The salt content in the cheeses made using the hollow fibre and spiral wound pilot plants was lower than Puhoi Valley Cheese. This is considered to be due to the low ratio of brine volume to cheese volume used for salting the cheese. The salt content of brine decreases during brining; hence a low ratio of brine volume to cheese volume causes a significant decrease in brine concentration. The decrease in brine concentration decreases the salt intake of the cheese. As salt diffuses in the moisture diffuses out, lower salt content results in higher moisture content in the cheese. As mentioned, the moisture content of the hollow fibre pilot plant cheese was higher than the PVC cheese. The moisture content is inversely proportional to the total solids, hence higher moisture in pilot plant cheeses implies lower total solids than the PVC cheese. The lipase activity results showed that the hollow fibre and spiral wound pilot plant cheeses had higher lipase activity than the Puhoi valley cheese. The differences in lipase activity of the pilot plant cheeses and Puhoi Valley cheese were considered to be due to the incomplete inactivation of lipase present in milk during pasteurisation. The results from texture and sensory evaluation support the above mentioned differences. The microbiology results for all pilot plant cheeses were within the trigger limits set by Puhoi valley cheeses. The results from monitoring the operating parameters of both the pilot plant data show that the permeate flux decreases while the total solids in milk increase with time, which was also observed from the Puhoi Valley Cheese data. However, the rate of decrease of the permeate flux and the increase of the total solids in milk are dependent on the membrane area, feed volume, transmembrane pressure, pressure drop across the membrane and the flow characteristics. The rate of decrease in permeate flux and the rate of increase in the total solids of the hollow fibre runs and spiral wound runs are slightly different. The difference is due to the availability of larger membrane surface area and processing of larger feed volume of milk in the spiral wound runs. The transmembrane pressure and the pressure drop across the membrane were maintained as close as possible to Puhoi Valley Cheese. In conclusion, spiral wound membranes can be used to achieve the desired total solids concentration and successfully make the same feta cheese as the hollow fibre pilot plant. In order to make the same quality of feta cheese as Puhoi Valley Cheese using the spiral wound membrane pilot plant, the same composition of milk used for concentration at Puhoi Valley Cheese needs to be used on the spiral wound pilot plant unit. It is recommended that Puhoi Valley Cheeses should be replaced with spiral wound membranes if they are more economical in terms of cost than the hollow fibre membranes.

Fouling of heat treatment equipment in the dairy processing industry is an expensive and persistent problem. The objective of this work was to develop a better understanding of the mechanisms of dairy fouling in heat exchangers and identify methods to control this build-up. This was part of a larger project investigating the interaction between spore-forming thermophilic bacilli (thermophiles) contamination and fouling deposits on internal surfaces of equipment. Two systems were developed to monitor the onset and build-up of fouling on the internal surfaces of two research heat exchangers. The first used a commercial sensor to measure the local heat flux and the temperature on the hot side of a plate type heat exchanger. The heat transfer coefficient was calculated and normalised with its value at the start of the run to reflect the contribution of fouling deposits to the thermal resistance, thus giving a real-time estimate of the rate of fouling. The second system used an energy balance over a tubular type heat exchanger and measured inlet and outlet temperatures to estimate the overall heat transfer coefficient thus giving a global measurement of fouling over the tubular heat exchanger. In both systems the plot of normalised heat transfer coefficient over time often stayed constant over an induction period, which was followed by a falling period indicative of growth in the fouling layer thickness and/or mass. Each system was validated by comparing the final value of the normalised heat transfer coefficient with direct measurements of fouling made at the end of a run namely: fouling deposit height for the local measurement and fouling deposit mass for the global measurement. The normalised heat transfer coefficient reported by each system correlated well with the corresponding direct measurement of the fouling layer. An important factor identified in this study was the effect of air bubble nucleation on fouling deposits. It was shown that bubbles that formed on the heated surface greatly reduced the length of the induction period to a matter of seconds rather than hours, as found in previous studies of fouling in the absence of surface bubbles. The rate of fouling was also enhanced while the bubbles remained at the surface. The structure of bubble type fouling layers was linked to the behaviour of the bubbles at the heated surface. Visual observations of these bubbles showed evidence of growth, vibration and coalescence during their period of attachment to the heated surface. Deposits from bubble type fouling consisted of all solid components found in the original milk solution, except lactose, in approximately the same ratio. By contrast fouling deposits reported in the literature with systems operating under the traditional protein denaturation mechanism were reported to consist mainly of whey proteins. Bubble induced fouling can be limited in a number of ways, the most effective being to maintain a high operating pressure in the equipment to ensure nucleation does not occur. Experiments conducted in this study showed that a pressure of 130 kPa.g was sufficient to suppress all bubble nucleation at the heated surface at a temperature of 90°C. Another method identified was the use of high linear fluid velocities to entrain any surface bubbles into the processing stream immediately upon nucleation. Linear velocities above 1.0 m/s were shown to achieve this goal in the miniature plate heat exchanger tested. However, this method is only partially successful because the local linear velocity varies with position in heat exchange equipment of complex geometries and can drop below the mainstream average velocity causing surface bubbles to form, especially in recirculation regions behind flow obstacles. A more reliable method, in situations where high operating pressures could not be used, involved conditioning the heated surface with a thin protein layer during the first few minutes of a run. Conditioning the surface resulted in bubble suppression even at high temperatures and low pressures, thus greatly extending the length of the induction period. Trials performed in this study showed that the addition of a proteolytic enzyme produced by psychrotrophic microbes greatly increased fouling. The enzyme destabilised the caseins which could attach directly to the heat exchange surface independently from the bubble fouling mechanism. Thus the quality of the milk is another important factor to consider. However, the addition of enzymes produced by thermophilic bacilli isolated from milk powder plants did not increase fouling. A theory describing the air bubble induced fouling mechanism is presented along with recommendations on how to reduce this fouling contamination in processing equipment.

Cheese is one of the most well known food products in the world dating back to the 8th century B.C. There are more than 2000 varieties of cheese that are manufactured all over the world. Feta cheese is a soft white cheese with a salty and slightly acidic taste, which has originated from Greece. Most of the feta cheese manufactured in Greece is consumed locally, the migration of greeks to other parts of the world led to a demand for feta cheese outside of Greece. The spreading of the popularity of feta cheese to other ethnic groups in different parts of the world resulted in the high demand for feta cheese worldwide. The modern and most efficient method of feta cheese production involves a membrane filtration method, known as ultrafiltration. The ultrafiltration process utilises pressure as a driving force to concentrate milk by removal of water and small dissolved molecules. Hollow fibre and spiral wound ultrafiltration membranes are the two types of membranes that are commonly used for cheese production. An extensive amount of research exists on the implementation of ultrafiltration to improve the efficiency of the cheese making process and the performance of the membranes. However, limited research has been conducted on the comparison of the hollow fibre and spiral wound membrane performance in the cheese making process. The objective of the research was to determine if the hollow fibre membranes used at Puhoi Valley Cheese can be replaced with spiral wound membranes without compromising the quality of cheese produced. In order to achieve the objective, feta cheese was produced using hollow fibre and spiral wound ultrafiltration pilot plants. The operating performances of the hollow fibre and spiral wound membrane units were compared. To ensure that the quality of cheese is maintained, the cheese manufactured on the pilot plant units was analysed in terms of composition, microbiology, texture and sensory properties. The cheese made using the hollow fibre membrane pilot plant was compared with the reference sample from Puhoi Valley Cheese as they use hollow fibre membranes to produce feta cheese. The cheese made from the spiral wound membrane unit was also compared to that made by the hollow fibre membrane pilot plant unit. The operating parameters such as the inlet and outlet pressure, pressure difference along the membrane, transmembrane pressure, flow rate, recycle rate (bleed off rate), temperature and the run time were recorded. The operating parameters of the hollow fibre and spiral wound runs were compared with the data from Puhoi Valley Cheese. The quality of cheese made on the hollow fibre and spiral wound pilot plant units were evaluated in terms of composition, texture, microbiology and sensory properties. The composition was defined by the fat, protein, total solids and salt contents. The fat content was determined by utilising the modified Schmid-Bondzynski-Ratzlaff method, protein by the Kjeldahl method, total solids by using the air drying oven and salt percentage by the volhard method. The texture of the cheese was determined by the fracturability and hardness from the compression curve generated using the single bite compression test. The microbiological testing was performed according to New Zealand testing methods for E.Coli, Staphylococcus aureus, coliforms and yeast and mould. The difference from the control method was utilised for sensory evaluation. The acid degree value method was used to determine the lipase activity in feta cheese. It was found from the composition, texture and sensory analysis that the cheese from the hollow fibre pilot plant was different from the cheese manufactured at Puhoi Valley Cheeses (PVC). The spiral wound cheeses were also found to be different to PVC cheese, however the spiral wound cheeses and the pilot plant hollow fibre cheese were the same. The differences between both the pilot plant cheeses and PVC cheese were in terms of the fat, salt, moisture contents and the lipase activity in the cheeses. The fat content in the hollow fibre and spiral wound pilot plant cheeses are lower in comparison to the PVC cheese. This difference in fat content is considered to be due to the difference in the fat to protein ratio of the milk concentrated on the pilot plant and the PVC ultrafiltration system. The lower fat content resulted in firmer cheese than PVC due to more cross linking between the protein strands in cheese. The salt content in the cheeses made using the hollow fibre and spiral wound pilot plants was lower than Puhoi Valley Cheese. This is considered to be due to the low ratio of brine volume to cheese volume used for salting the cheese. The salt content of brine decreases during brining; hence a low ratio of brine volume to cheese volume causes a significant decrease in brine concentration. The decrease in brine concentration decreases the salt intake of the cheese. As salt diffuses in the moisture diffuses out, lower salt content results in higher moisture content in the cheese. As mentioned, the moisture content of the hollow fibre pilot plant cheese was higher than the PVC cheese. The moisture content is inversely proportional to the total solids, hence higher moisture in pilot plant cheeses implies lower total solids than the PVC cheese. The lipase activity results showed that the hollow fibre and spiral wound pilot plant cheeses had higher lipase activity than the Puhoi valley cheese. The differences in lipase activity of the pilot plant cheeses and Puhoi Valley cheese were considered to be due to the incomplete inactivation of lipase present in milk during pasteurisation. The results from texture and sensory evaluation support the above mentioned differences. The microbiology results for all pilot plant cheeses were within the trigger limits set by Puhoi valley cheeses. The results from monitoring the operating parameters of both the pilot plant data show that the permeate flux decreases while the total solids in milk increase with time, which was also observed from the Puhoi Valley Cheese data. However, the rate of decrease of the permeate flux and the increase of the total solids in milk are dependent on the membrane area, feed volume, transmembrane pressure, pressure drop across the membrane and the flow characteristics. The rate of decrease in permeate flux and the rate of increase in the total solids of the hollow fibre runs and spiral wound runs are slightly different. The difference is due to the availability of larger membrane surface area and processing of larger feed volume of milk in the spiral wound runs. The transmembrane pressure and the pressure drop across the membrane were maintained as close as possible to Puhoi Valley Cheese. In conclusion, spiral wound membranes can be used to achieve the desired total solids concentration and successfully make the same feta cheese as the hollow fibre pilot plant. In order to make the same quality of feta cheese as Puhoi Valley Cheese using the spiral wound membrane pilot plant, the same composition of milk used for concentration at Puhoi Valley Cheese needs to be used on the spiral wound pilot plant unit. It is recommended that Puhoi Valley Cheeses should be replaced with spiral wound membranes if they are more economical in terms of cost than the hollow fibre membranes.

The present study was aimed at understanding the high pressure (HP) processing-induced interactions of milk proteins in whey protein concentrate (WPC) solutions, in skim milk and in pure protein systems. The changes in milk proteins induced by heat treatments in the same systems under selected conditions were also evaluated. The main approach taken was to elucidate changes in the whey proteins in heat- and pressure-treated samples from common aliquots, under identical conditions, using various one-dimensional (1D) and two-dimensional (2D) polyacrylamide gel electrophoresis (PAGE) techniques in the absence or presence of a disulphide bond reducing agent. In some instances, the samples were also analysed using small deformation rheology, size exclusion chromatography (SEC) and transmission electron microscopy (TEM). The results of the present study indicated that, in general terms, heat treatment and HP treatment had common effects, i.e. denaturation and subsequent aggregation of whey proteins. Both heat treatment and HP treatment generated disulphide-bonded and hydrophobically bonded aggregates of whey proteins. However, the sensitivities of each of the whey proteins to heat treatment [immunoglobulin (Ig) > lactoferrin (LF) > bovine serum albumin (BSA) > β-laetoglobulin B (β-LG B) > β-LG A > α-lactalbumin (α-LA)] and pressure treatment (β-LG B > β-LG A > IgG > LF > BSA > α-LA) were considerably different. Also, HP treatment generated a comparatively greater proportion of smaller aggregates than did heat treatment. The effects of protein concentration, intensity of pressure treatment, holding time and pressurising temperature on whey protein aggregation in WPC solutions were investigated. The rate of aggregation of whey proteins increased with an increase in the concentration of protein in the WPC solution and the pressurising temperature. The combination of low protein concentration, mild pressure treatment (200 MPa) and low pressurising temperature (20°C) led to minimal loss of native-like and SDS-monomeric β-LG, whereas the combination of high protein concentration, severe pressure treatment (600 MPa) and higher pressuring temperature (40°C and higher) led to significant loss of both native-like and SDS-monomeric β-LG. The sensitivity of pressure-resistant whey proteins, such as α-LA and BSA, to the aggregation was significantly increased at pressurising temperatures of 40°C and higher. Self-supporting gels were formed when 8 or 12% (w/v) WPC solutions were pressure treated at 600-800 MPa. 20°C. Detailed analysis of the behaviour of the proteins during the formation of these gels revealed a novel pathway, suggesting that intermolecular disulphide bond formation occurred at high pressure but that hydrophobic association became important after the HP treatment. In the later part of the study, heat- and HP-induced interactions of caseins and whey proteins were studied in a more complex system, i.e. skim milk. With the application of modified PAGE techniques, it was possible to show that the high molecular weight disulphide-bonded aggregates that were formed by HP treatment of skim milk contained disulphide-linked complexes consisting of αS2-casein (αS2-CN) as well as κ-CN, β-LG and other whey proteins. The results showed that the effects of heat treatment and HP on the interactions of the caseins and whey proteins in milk were significantly different. The accessibility of αS2-CN and the formation of complexes involving αS2-CN, κ-CN and whey proteins in the HP-treated milk, as demonstrated using the modified 2D PAGE technique, and as explained by possible proposed reactions of the caseins and whey proteins in pressure-treated milk, was an important finding of the present study. Finally, a study on the effects of HP treatment in model systems using pure proteins in solution, both singly or in binary and ternary combinations, generated very useful information and clarified the role of each protein in pressure-induced aggregation and interactions of milk proteins in complex systems such as WPC and milk. It was found that the reactions of β-LG were not significantly affected by other proteins such as α-LA or BSA, but that the presence of β-LG in the system catalysed the reactions of other proteins such as α-LA or BSA.

The aim of this study was to explore the characteristics of a non-gelling exopolysaccharide (EPS) obtained from Lactobacillus delbrueckii subsp. bulgaricus NCFB 2483 and a gelling EPS obtained from Sphingomonas elodea ATCC 31461 (31461). The EPSs were isolated from the two bacterial strains grown in milk permeate-based media. They were purified and then characterised using light scattering and viscometric techniques. A greater emphasis of this research was placed on 2483 EPS since its physical characteristics have not been reported to date. In the case of 31461 EPS. a model for gelation of the sodium gellan was proposed based on rheological and light scattering measurements. The rheological properties of the two EPSs were also compared with several commercial polysaccharides. Microscopy examination of 2483 EPS was carried out using confocal laser scanning microscopy (CLSM) and scanning electron microscopy (SEM). In CLSM, the lectin SBA (from Glycine max) Alexa Fluor 488 conjugate was used to stain the EPS since it has affinity for galactopyranosyl residues present in 2483 EPS. The CLSM micrographs showed a random distribution of EPS aggregates in the culture medium. At high magnification, the SEM micrographs showed web-like EPS structures. These structures were formed during the critical point drying process, when the EPS, which filled the interstices and channels of the protein aggregates, dehydrated. The 2483 EPS aggregates were found to be stable at neutral or low pH (~3.9) but were disrupted at high pH (pH 8-10). Procedures commonly used to quantify EPS from culture medium were found to be unreliable. In the development of an improved EPS assay, each of the processing steps was examined. Key improvements included the use of Flavourzyme for protein hydrolysis; optimising ethanol concentration to prevent lactose crystallisation yet allowing complete EPS precipitation; and a suitable centrifugation regime to minimise EPS loss. The improved EPS assay gave reproducible results (5% coefficient of variation). The isolation of 2483 EPS from milk media proved to be a difficult task because of interference from non-EPS components. An effective and simple approach allowing maximum EPS recovery involved the use of a hydrolysed milk medium which was ultrafiltered (UF) to remove molecular species larger than 2.5 x 105 Da. The UF permeate was suitable for the growth of 2483 with an EPS yield of ~400mg/L. Two EPS fractions (namely a soluble and an insoluble fraction) were isolated by ethanol precipitation and the soluble 'ropy' fraction was further purified to achieve ~98% purity. The elemental analysis of the purified fraction revealed the presence of nitrogen (~2.7% w/w). This could be due to the interaction of some peptides (from the growth medium) with the EPS. The polysaccharide composition of the soluble EPS fraction comprised of galactose, glucose, rhamnose and mannose residues (5:1:0.6:0.5). Traces of glucosamine were also found in the fraction. The purified fraction of 2483 EPS was characterised. Using a capillary viscometer, an intrinsic viscosity of ~2013mL/g was determined. The flow curves of the 2483 EPS solutions obtained using a rotational viscometer showed shear-thinning behaviour and an exponent value of ~0.76 (based on the Cross-type model) is typical of random coil polymers. The concentration dependence of the viscosity plot produced gradients of ~1.1 in the dilute domain and ~3.3 in the semi-dilute to concentrated domain. The coil overlap parameters at three concentration domains (c*[ŋ],ccr[ŋ] and c**[ŋ]) were 0.55, 2.86 and 5.67 respectively. The molecular parameters of the 2483 EPS were found via static light scattering measurements to have a weight-average molar mass (Mw.) of ~2 x 106 Da, a z-average root-mean-square radius ((r2g)z1/2) of ~165nm and a low polydispersity index (Mw/Mn ~1.15). The plot of Mw versus (r2g)z1/2 gave a gradient of approximately 0.5, which also suggested that the EPS polymer adopted a random coil conformation. The second part of the research involved gellan gum. Two gellan samples were studied. The first gellan sample was obtained from the fermentation of Sphingomonas elodea ATCC 31461 using milk permeate-based medium (31461). The second sample was a commercial high acyl gellan (LT100). Both gellan samples were converted to their sodium forms (Na-31461 and Na-LT100 gellan) using cation exchange resin and purified. The Na-gellan samples were highly sensitive to changes in Na+ concentrations. From oscillatory measurements, it was found that the complex moduli of the two Na-gellan samples superimposed closely at a specific Na+ concentration. The model for the conformational changes of Na-gellan molecules from a solution to a gel was proposed based on rheological and light scattering data. At very low Na+ concentrations (<19mM, in the case of Na-LT100). Na-gellan molecules were single-stranded (Mw ~2.5 x 10 5 Da) and adopted random coil conformation (exponent value based on the Cross-type model of ~0.76). At a slightly higher Na+ concentration (~19-24mM), Na-gellan molecules formed double-helices which led to a two-fold increase in molecular weight (M w ~5.2 x 105 Da). The double-stranded molecule appeared to be stiffer (exponent value of the Cross-type model ~0.82) and the mechanical spectra (G',G”) demonstrated 'weak ge' characteristics. A further increase in the Na+ concentration (>24mM) resulted in the formation of a gel network. The study also found that at low Na+concentration, both single-stranded and double-stranded Na-gellan molecules had a tendency to form aggregates under zero-shear conditions. The interactions involved in these aggregates were considered weak and transient, according to the Cox-Merz plot and light scattering data.

The technique of micro-oxygenation involves the deliberate addition of continuous, metered amounts of oxygen into a vessel of bulk wine during the maturation period (between the end of fermentation and bottling). The aim of the process is to improve the sensory properties of red wine, particularly the mouthfeel characteristics associated with the various polyphenol constituents. The success of the process appears to depend strongly on the ability to control the rate of oxygen dosage. The effect of dosage rate on the chemical and corresponding sensory changes of a red wine is the central theme of this thesis. A method of dosing oxygen (at typical micro-oxygenation rates) into small volumes of wine (<100 litres) was developed using a dense polymer membrane diffuser. It was clearly demonstrated that wine could be reliably oxygenated at very low rates using a coiled length of FEP as the diffuser material. Oxygen dosage was regulated by adjusting the oxygen pressure inside the tube. The advantage with a dense polymer diffuser is that no bubbles are generated and the oxygenation efficiency is 100%. The diffuser was fully modeled and characterised for use in the laboratory scale trials detailed in Chapters Four and Six. The small scale oxygenation equipment was used to conduct a fully replicated experiment to investigate the evolution of a Cabernet Sauvignon wine under four oxygenation treatments at dosage rates of 0, 10, 23 and 36 mg/L/mth. The total period of the trial was 105 days. HPLC analysis indicated that the rate change of low molecular weight polyphenols is directly related to the oxygen dosage rate. The concentration of the majority of the identifiable monomers, most notably the anthocyanins decreased throughout the course of the trial. The rate of decrease was directly related to oxygen dosage rate. Thiolysis results showed an increase in mDP for all treatments over the course of the trial until day 77 when they were observed to decrease for all treatments. The decrease in mDP coincided with an addition of SO2 which was investigated in a subsequent trial. Spectrophotometric results indicated that the rate of formation of non-bleachable pigments was directly related to the rate of oxygen dosage with significant differences between the high rates (23 and 36 mg/L/mth) and the low rates (0 and 10 mg/L/mth). The trend for all treatments was for increased levels of stable pigments. The sensory results show that the measured organoleptic temporal development exhibits a similar oscillatory behaviour compared to the anecdotally derived curve presented in figure 1-2. The distinction between the respective phases described in section 1.1.1 was, however less clear. The most significant factor in the model weighting was mouthfeel and astringency which correlates with the observed changes occurring in the wine polypenols during maturation. Overall the laboratory scale trial showed that the chemical polyphenol development was directly related to the oxygen dosage rate. The sensory evolution also appeared to be accelerated with higher oxygen dosage rates, although the oscillatory nature of the sensory response given a single linear input indicates a complex underlying mechanism driving the changes. The effect of SO2 on the development of wine polyphenols with and without oxygen was also investigated. The presence of SO2 was found to have a significant effect on both mDP and the concentration of non-bleachable pigments. mDP was observed to decrease over the six week trial period irrespective of whether oxygen had been added or not. The mDP for the treatments without SO2 increased steadily over the course of the trial. Similarly the formation of non-bleachable pigments was suppressed and even retarded with SO2 present whereas for the treatments without SO2 a steady increase was observed. The implication of these results is that SO2 may have a much larger effect on tannin development than oxygen. The use of electrochemical micro-oxidation (or ELMOX) was examined ostensibly to determine proof of concept and also compare the performance of glassy carbon and titanium as electrode materials against traditional micro-oxygenation. Notable transformations occurred with titanium showing higher levels of ethanal than the other treatments both chemically and by sensory measure. A greater rate of stable pigment formation was also observed for the titanium compared to the other treatments. The respective dosage rates for the glassy carbon ELMOX and traditional micro-oxygenation treatments were too low to be able to discriminate any significant differences compared to the control wine. / AGMARDT Doctoral Scholarship