Progressive freeze concentration

Dewatering and drying are common process operations in the food and biobased industry. To minimise the energy required for drying, products are often concentrated before drying. This usually increases the solid concentration from below 10 % to around 40–70 % solids, depending on among others the viscosity of the specific formulation. The most widely applied concentration method is evaporation, which is a fast process but is energy intensive and operates at elevated temperatures. This makes evaporation not suitable for food products that suffer from thermal degradation. For these products, alternative mild concentration methods can provide a better product quality.One of these methods is freeze concentration. When a solution is frozen, the ice crystals exclude the solute from their crystal matrix. If the ice is removed, a concentrated solution is obtained. Since freeze concentration makes use of cooling instead of heating, it is suitable for products containing volatile or thermally sensitive components. The process can be configurated in multiple ways. The focus of this thesis is on progressive freeze concentration, during which ice crystals are grown on the wall of a heat exchanger. The advantages of the technique are that the ice crystals can be separated easily from the concentrated solution and that it is well suited for extensive heat integration. While freeze concentration in theory should create pure ice, this requires unproductive and slow ice growth rates. At faster ice growth, part of the solute will be included in the ice fraction and be lost after separation. The present work therefore aimed to understand why solute inclusion occurs in freeze concentration and to apply this to a pilot-scale progressive freeze concentrator. For this, we used lab-scale experiments to investigate the influence of cooling, mixing, and solutes on solute inclusion. The data generated in these experiments were combined with a modelling approach to understand the mechanism of solute inclusion during progressive freeze concentration.In chapter 2 freeze concentration of model solutions of sucrose and maltodextrin is evaluated in a laboratory-scale freeze concentration system. From this initial study we concluded that, as expected, an increase in the ice growth rate would lead to higher inclusions at a constant heat exchanger temperature. However, when the initial temperature of the heat exchanger was higher and only gradually lowered, the inclusion was less without reducing the total amount of ice formed. A constant heat exchanger temperature in the beginning imposes a high degree of super-cooling, which leads to an irregular ice front creating pockets of concentrated solution that become encapsulated in the ice. When the temperature difference between the heat exchanger and the freezing point is reduced, the degree is much lower and a smooth ice front is formed that excludes the solutes more effectively. The solute inclusion can be further reduced with stronger agitation, which aids in minimising concentration polarization.Chapter 3 reports on experiments and modelling with different products, including proteins. Specifically, solutions of soy protein and whey protein were evaluated using the cooling and mixing conditions established in the previous chapter. The solute inclusion was modelled using an existing theory using empirical parameters, which was linked to the heat and mass balances describing the ice growth rate. At low ice growth rates, we found good agreement between the model and our data, but at high ice growth rates the model underestimated the level of inclusions. The conclusion is that one of the assumptions in the model, pertaining to the ice growth being slow and planar, does not hold during dendritic ice formation at higher ice growth rates.Whey protein was selected as a model protein in chapter 4. Since most products in the food industry do not consist of only protein, we investigated the concentration of mixtures of whey protein, salt, and/or sucrose. The addition of both sodium chloride and sucrose increased the inclusion of that solute, but also that of the protein. It was hypothesised that in the case of sodium chloride the solute inclusion is caused via localised super-cooling in the concentration polarization layer. In the zone where the super-cooling occurs, ice crystals form within the layer, which then incorporate domains of concentrated solution into the ice layer. The freezing point depression by sucrose is much less extreme (per unit of weight), so we here expect that besides the effect of the localised super-cooling, the inclusions are caused by the very strong increase of the viscosity and reduction of the diffusivity of the concentrated solution at low temperatures. When both sodium chloride and sucrose are added, the effect on inclusion is not additive but concurrent, i.e., the addition of both did not cause more inclusions than does adding either of the two.With these observations a theoretical model for the prediction of solute inclusion was developed in chapter 5, using the state diagram. This model predicts the existence of two ice growth regimes. The first regime is at low ice growth rates, and with high-molecular weight solutes that gives negligible freezing point depression. Here, the compositions within the boundary layer stay above the freezing line, and therefore the ice growth is stable, and no inclusion is expected in this regime. The second regime is at larger ice growth rates, and with smaller solutes that does give appreciable freezing point depression. Here, the compositions in the boundary layer cross the freezing line, which then leads to freezing within the boundary layer, converting this into a freezing zone, and leading to inclusion. The model predicts the inclusion of low-molecular weight components and the negligible inclusion of macromolecules, such as whey proteins.Finally, in chapter 6 progressive freeze concentration is evaluated in a pilot-scale progressive freeze concentrator and compared into alternative concentration methods. In the pilot-scale unit brine, sucrose, and whey protein solutions were concentrated. Overall, for all the solutions it was observed that the product losses were higher than on lab scale, mainly due to scaling issues in the separation of the concentrate and ice fractions. To reduce the losses the ice fraction should be treated to recover the product. The energy requirements were found to be approximately 150 kJ/kg ice removed from the solution. This would make progressive freeze concentration competitive with alternative concentration methods.