Development of a rapid liquid freezer : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Food Technology at Massey University, Manawatū, New Zealand

Thumbnail Image
Open Access Location
Journal Title
Journal ISSN
Volume Title
Massey University
The Author
Small sheep dairy farms often make insufficient volumes of milk for economic daily collection and are limited by transport distances to processors. A method of long-term on-farm storage of milk would enable the industry to grow. Freezing would allow extended milk storage on farms. But existing methods of freezing for on-farm applications have shortcomings around materials handling, labour requirements and product quality. The project reported in this thesis aimed to develop the engineering science behind an economically viable freezing method that would improve on current methods. The first period of this project focused on two freezer designs which were thought to be promising: Rolling Droplet Freezing (RDF) and Falling-Film Flake freezing (FFF). RDF was selected as the initial focus of the research program and consisted of a system where droplets of milk would roll down an angled super-hydrophobic surface against a cold air flow and freeze. RDF was abandoned due to concerns about construction costs and operating reliability. In a condensing atmosphere, droplets rolling on superhydrophobic surfaces occasionally transitioned from a Cassie-Baxter wetting state to a Wenzel wetting state, which caused the droplets to stick. FFF was then developed further. A pilot scale unit was designed and constructed, and preliminary pilot-scale trials that were conducted with pure water and ovine milk reconstituted from powder. The partition coefficient of FFF was measured as 0.946 at an operating temperature of -30°C. At higher operating temperatures the partition coefficient was reduced. Detaching frozen solids by applying a burst of heat to the freezer/ice interface was studied and this method of detachment was successful with pure water, but ineffective for ovine milk. The development of FFF was put on hold with the conception and development of the continuous tubular freezer. Ice formed in a solution can show morphologies ranging from highly dendritic structures with entrapped solutes, which are homogenous on a gross scale, to large crystals of pure ice with solutes rejected and compressed into inter-crystalline spaces. To investigate which sheep milk components influence ice morphology at various freezing rates, whole milk was separated into skim milk and into a casein-free serum phase. A simulated sheep milk ultrafiltrate was also prepared. The morphology of the ice/sample interface was observed in a custom-built microscope stage at freezing front velocities from <0.5 μms⁻¹ to 50 µms⁻¹ with a spatial temperature gradient of 35-38 Kcm⁻¹. The morphology arising from extremely rapid freezing front velocities was investigated by supercooling slides on a temperature-controlled stage and observing the nucleation and recalescence of the samples. The morphology of ice at the interface changed from a planar to columnar and then to dendritic as freezing front velocity increased, with the transitions from one morphology to another occurring at lower speeds in more complicated solutions. A map of freezing front behaviours was developed. The transition between interface morphologies was at different velocities and transition differed based upon the interface velocity. At lower interface velocities a columnar interface grew directly from a planar starting condition. At higher velocities an intermediate dendritic zone formed, which then settled into a columnar interface. The ice formed by rapid freezing from subcooled solutions was highly dendritic, with ice growth rates of approximately 21,000 μms⁻¹, which was close to the diffusion-limited ice growth rate in water of similar degrees of supercooling. The morphology of frozen ovine milk was also studied by Cryogenic Scanning Electron Microscopy (Cryo-SEM): Milk was frozen by three different methods-slow quiescent freezing (SF), rapid directional freezing (DF), and droplet freezing in LN₂. Ice crystals rejected unfrozen solids into the region between crystals in all samples, including those frozen by immersion into liquid nitrogen. There was a distinct difference in morphology between the SF and DF samples, with the bands of unfrozen solids being significantly smaller in DF samples, and the long axes of ice crystals were aligned with the direction of heat flow. SF samples lacked any particular ice growth direction, and ice crystals were orders of magnitude larger. Lactose crystallisation was observed in some SF samples but was not observed in any DF samples. Fat globules were engulfed in ice crystals in DF samples, but rejected in SF samples. To study the effects of frozen storage temperature and time, samples of raw ovine milk were stored frozen at -10°C, -18°C and -28°C to -30°C for up to 8 weeks. Further samples were stored below -20°C for 6 months. After thawing at 20°C, samples were tested for a range of properties and serum samples were collected by separating the fat phase and micellar casein phase by centrifugation. A gel was observed in milk stored at -10°C for 4 weeks and 8 weeks but was not observed in milk stored at lower temperatures. The gel dispersed under heating and homogenisation. There was no change observed in the pH, or serum protein level of thawed samples after frozen storage at any temperature. The whiteness of the milk decreased during frozen storage and the yellowness increased. Both of these changes were reversed on homogenisation. The serum Ca²⁺ levels in milk stored at -10°C and -18°C dropped over the storage period, while no trend was seen in milk stored below -28°C, indicating that the migration of Ca²⁺ may play a role in the formation of gels after frozen storage. Milk that had been stored below -20°C for 6 months had a similar viscosity and appearance to fresh milk. A possible mechanism for the formation of gels at -10°C, but not -18°C or -28°C lies in the altered solute environment, and the physical agglomeration of milk components in the spaces between ice crystals, driving the gelation of closely packed casein micelles, with Ca²⁺ stabilising this network. It is well established in literature that the viscosity of an unfrozen phase increases by several orders of magnitudes as it decreases in temperature and approaches a glassy state. This increased viscosity reduces protein mobility and solute diffusion, which reduces the rate of gel formation. The tendency for frozen milk particles to bind together during frozen storage was evaluated. Frozen pellets of whole ovine milk were stored under weights at -10°C and -18°C and pellets of frozen concentrated milk stored at -18°C and -28°C. Ovine milk pellets bound together at -10°C but not -18°C, while concentrated milk bound together at -18°C, but not -28°C. This can be linked to the volume and viscosity of the unfrozen phase in these samples. Differential scanning calorimetry was used to determine the fraction of freezable water frozen at any temperature. The melting onset temperature was observed, and this was used to determine the solids content maximally freeze concentrated solution (𝑋𝑠(𝑇𝑚)). 𝑋𝑠(𝑇𝑚)=0.875 for whole ovine milk 𝑋𝑠(𝑇𝑚)=0.85 for skim milk, and 𝑋𝑠(𝑇𝑚)=0.81 for ovine milk serum. This was also determined for whole ovine milk by the magnitude of the overall latent heat release during melting, which gave a value for whole milk of 𝑋𝑠(𝑇𝑚)=0.85±0.016. A partial phase diagram for ovine milk was generated from the data collected. The insights generated from observing both the dendritic morphology of high velocity ice fronts and progressive freezing behaviour led to conceptualising a novel tubular freezer, subsequently constructed. It was hypothesised that reducing the volume or area of ice in contact with the freezer wall, due to the inclusion of unfrozen product, could reduce the adhesion strength between a frozen product and the freezer wall. By controlling the outlet temperature, the volume fraction of unfrozen product could be controlled. The adhesion strength could thereby be controlled, and a set of operating conditions could be found that would allow a mostly frozen product to be extruded as a solid from a cooled tube by a high-pressure pump. This was tested on a benchtop scale (up to 5mL/minute, with a freezer internal diameter of 4.2mm and cooled length of 500mm), with ovine milk, fruit juice, fruit pulp, concentrated coffee, bovine cream and concentrated milks. The system successfully froze all samples. The operating pressure was found to increase with increased frozen fraction, and therefore with decreased operating temperature. The ice morphology of milk and juice frozen by this equipment was imaged by cryo-SEM and by optical microscopy. The ice crystals were radially aligned, increasing in size closer to the centre of the frozen product plug, which was expected due to the heat flows and the relationship between freezing front velocity and feature sizing. This positive preliminary result led to the construction of a larger scale prototype unit which consisted of a spiral tube with a length of 5000 mm, and an internal diameter of 10 mm. This was used successfully for a product flowrate of approximately 6 kghr⁻¹.
Four Figures were removed for copyright reasons: 1-4 (=Waschies et al., 2011 Fig 8), 1-6 (=Bhatnagar et al., 2007 Fig 1), 1-7 (=Lu et al., 2015 Figs 5 & 7), 1-14 (=Rao & Hartel, 2006, Fig 1a).
Sheep milk, Storage, Refrigeration and refrigerating machinery, Farms, Small, Dairy farming, New Zealand