Fine milling and air classification (AC) is an attractive method to concentrate protein from pea and faba bean, but many more starch-rich pulses can be fractionated into enriched ingredients. However, switching raw materials may require costly trial-and-error optimization of separation settings. This work aims to predictively model protein separation from starch-rich pulse flours, as function of classifier settings. This was possible by defining populations of microstructures created upon milling, and by using mass balances to deconvolute the particle size distribution (PSD) of the flour into individual PSDs of each microstructure, in a self-consistent approach. The model was successfully developed on data of adzuki bean flour, for which milling and air classification was not reported before, and could also adequately deconvolute flours of mung bean, faba bean, and yellow pea. A particular experimental finding for adzuki bean was efficient protein separation (>90 % enrichment, >60 % protein yield) in a single classifier milling step, omitting the need for an additional air classification step. Next, a cut size model was used to quantify particle size recovery from flours as function of air classifier settings. Coupling this cut size model to the compositionally deconvoluted flours accurately predicted protein separation in the fine fraction over a range of classifier wheel speeds and air flow rates. Conversely, it is also possible to predict e.g. starch separation in the coarse fraction. The model may replace trial-and-error tests regarding optimization of classifier settings, providing both academic and industrial relevance.

Maize grain processing usually starts with degerming to lower its oil content. The created maize germ-enriched fraction (MGF) represents up to 15 % of the kernel mass. Oil recovered from the MGF is valuable, but the non-oil fraction is underutilized. We separated MGF (14 % protein, 40 % starch, 18 % oil; db) into enriched fractions via dry processing, and reflect on the energy efficiency in combination with additional aqueous extraction. Dry fractionation requires mechanical pressing and oil extraction with hexane or food-safe 2-methyltetrahydrofuran, which yields cakes having residual oil contents of 0.4 w/dw% or 3.1 w/dw%, respectively. Both cakes could be milled, and air classification enriched protein in the fine fraction from 19 % to 27 % by deflecting larger starch granules. The protein yield relative to MGF was 15 %, mainly due to milling losses. CLSM revealed the protein to be in a fibrous matrix, hence it was not enriched further by electrostatic separation. Due to the moderate protein contents and yields, aqueous protein extraction was more energy efficient from de-oiled flour than from fine fractions. Extracts from hexane-processed fractions foamed better than whey protein at 0.5 mg protein/mL. Processing with 2-methyltetrahydrofuran did not enable stable foaming, which requires optimization towards lower residual oil contents.

Grain legumes such as pea, faba bean, lupin and soybean are an important protein source for the production of plant-based foods and thus facilitate the protein transition. For many food applications, the proteins are first isolated using conventional wet methods that are resource intensive. Dry fractionation processes are therefore developed to facilitate a more sustainable protein transition. This review discusses the status of dry fractionation of grain legumes to produce protein-rich ingredients for food production and how the use of these dry-enriched ingredients could be further enhanced. Dry fractionation includes dry milling and dry separation technologies which are first briefly described. There are different strategies to further improve the separation, which include pre-treatments and improving powder bulk behaviour. Pre- and post-treatments not only improve the functional properties of dry-enriched protein ingredients but also enhance the nutritional value of the ingredients and minimize off-flavours. Opportunities still exist to further optimise dry fractionation techniques and pre-treatments to increase the purity and yield. Finally, the use of dry-enriched fractions should be accelerated by development of 1) functionality-driven ingredient formulation strategies and 2) new physical post-modifications and food fermentation strategies to enhance functionality, nutritional value and taste of the ingredients to prepare attractive food products.

Customized textures can be achieved using extrusion-based 3D food printing by varying composition and processing parameters, e.g., printing designs and post-processing conditions. This study aims to design textural properties of customized pea-based snacks by investigating the effects of composition and multiple processing parameters on fracture behavior. Macronutrient composition of printing materials (i.e. starch-to-protein ratio), geometric design, and post-processing conditions (i.e. steaming or baking time) were systematically varied. The fracture behavior of freshly printed and post processed structures were analyzed. Samples showed elastic deformation and plastic deformation after steaming and baking, respectively. This difference in fracture behavior could be linked to microstructural changes indicated by confocal laser scanning microscopic imaging. Starch-to-protein ratio of the printing material and the geometric design also influenced the fracture behavior, but to a relatively minor extent. Moreover, fracture stress and Young's modulus were strongly influenced by the dry matter content of the samples. Statistical analysis using ANOVA was performed to establish the significance of the effect of composition and processing parameters on Young's modulus and dry matter content of samples. Based on this understanding, we propose different strategies to effectively design textural properties of snacks using 3D printing. By considering multiple factors, this study provides valuable insights into designing healthy snacks with customized textures using 3D food printing.

Edible films made of cold-water fish gelatin suffer from suboptimal mechanical strength, limiting their use for sustainable packaging applications. In this study, the use of 3D-printable embedded patterns is investigated as reinforcement for gelatin cast films, aiming for cast films with customizable and enhanced mechanical strength. Various combinations of patterns (i.e. lines, grids, and triangles) and inks (varying concentrations of sodium caseinate, sodium alginate, and cellulose fibers) were 3D-printed. The 3D-printed structures were embedded into cast films and the mechanical properties were subsequently tested. Our results show that films with embedded patterns had an overall higher tensile strength at different stretching directions (i.e. parallel, perpendicular, and diagonal), compared to the plain gelatin films. The strongest mechanical anisotropy was found using the line pattern and the grid showed anisotropy in the parallel and perpendicular direction, highlighting the influence of printing path. Microscopical analysis revealed that embedded patterns affected the fracture mechanics of films. Interestingly, cellulose fibers showed alignment in the printed filaments along the printing direction, which contributes to an increased tensile strength of films after drying. Thus, by using printable embedded pattern design, edible films with customized mechanical performance can be made, contributing to future developments of sustainable packaging solutions.

Air classified chickpea protein concentrates often lack the separation efficiency of other starch-rich pulses. We subjected chickpea flour to de-oiling or to blending with different types and concentrations of silica flow aids to reduce its cohesion and to increase its dispersibility in air. Air classification enriched protein to a content of 45%. Initial protein yields below 25% increased to 70% upon extensive de-oiling, and this de-oiled ingredient containing 51% protein was enriched to 56% by electrostatic separation. Alternatively, nanosized flow aids doubled the initial protein yield. SEM-EDS image analysis revealed that higher silica concentrations increased the accumulation of nanoparticles onto flour particles, which lowered the flour's basic flow energy and moderately increased its dispersibility. This reduced fouling during air classification, but the protein yield levelled off with increasing silica concentrations because the microstructure was not disentangled further. Silica addition during milling rendered post-milling blending unnecessary, without compromising the protein yield.

Upon spray drying of dairy powders, the distribution of components within the droplets can become nonuniform. In this study, the influence of the composition on the drying behaviour, morphological development and surface composition were studied using single droplet drying. The addition of fat was found to result in an earlier locking point (1.1 instead of 1.8 s). The fat content did not influence the morphological development, but the protein and lactose ratios did influence whether the dried particle was smooth or wrinkled. Confocal Raman microscopy revealed that the initial fat, lactose and protein contents influenced the dry surface composition. For example, fat was found on the surface of emulsions that were either rich in fat or in whey protein. Single droplets of multicomponent dairy systems were dried at spray drying like temperature-time trajectories. Differences in the initial composition of the dried systems result in different morphologies and dried surface compositions. image

The positive effect of protein drying aids that reduce the stickiness of sugar-rich products on the spray drying yield is known. However, as agglomeration also depends on stickiness, one can expect an effect of these drying aids on interparticle collisions. Therefore, the effect of protein addition to maltodextrin systems on agglomeration and yield in spray drying was investigated using single droplet drying and pilot-scale spray drying experiments. Single droplet drying was used to compare the sticking regimes of whey and pea proteins at different concentrations. Adding protein increased the chance of creating agglomerates upon forced collisions between a drying droplet and a glass bead. This was reflected in an increased fraction of agglomerated particles in pilotscale experiments with injection of fines. Pea protein decreased the fraction of non-agglomerated primary particles from 38% to 22% when the protein content in the liquid feed was 50 g/kg dry basis. In pilot-scale spray drying, protein addition improved the yield. Industrially, these results can be used to steer formulations regarding agglomeration and yield.

Personalized foods with varying macronutrient compositions can be created by 3D food printing to fulfill the dietary requirements of individual consumers. In this work, we aim to quantitatively study the influence of varying multiple macronutrient concentrations on printability of food inks, by applying a systematic approach. Pea-based food inks consisting of insoluble pea fibre, pea protein, and native pea starch were formulated and water was added following these ingredients' respective water holding capacities. Printability was quantified in terms of extrudability (force required to extrude material out of a cartridge), buildability (flow point to deter-mine shape stability after deposition on the printing platform), and printing precision (surface defect index (SDI) of the printed object). This approach helps to efficiently define a printable landscape of pea-based inks with a large variation in macronutrient composition. Our results show that increasing protein concentration resulted in increased extrusion force and flow point, while the opposite effect was found for fibre. Of all the tested pea-based formulations, 71% could be printed to a height of 50 mm with high printing precision (SDI<0.14). The presented systematic approach provides a solid basis for rapid development of printable plant-based inks while avoiding a trial-and-error approach to optimize inks with highly variable macronutrient composition.

Cereal processing industry removes the fibrous tissues from kernels via abrasive milling, but this tends to remove part of the valuable endosperm components as well. Therefore, endosperm recovery from such abraded barley malt material (53% endosperm and 47% husk) was evaluated for three dry separation technologies. Electrostatic separation recovered 25% of the endosperm at 85% purity and this recovery increased to 39% in a second run, which indicated potential to further improve the setup. Increasingly finer sieves removed up to 40% of the husk with little endosperm loss, but further husk removal up to 95% by the finest sieve reduced the endosperm yield to 54% due to decreasing differences in smallest diameter between the endosperm and husk particles. Air classification outperformed sieving by yielding 71% of the endosperm while removing 95% of the husk, and further, less selective air classification steps could yield up to 94% of the endosperm while still removing 59% of the husk. Moreover, such additional air classification steps currently recovered residual endosperm particles more selectively than electrostatic separation after an initial air classification. Overall, air classification after abrasive milling increased the removal of insoluble matter from malted barley kernels. The loss of soluble endosperm components remained similar to the loss as observed in a single abrasive milling step.

Crucial for achieving premium agglomerated powder products through spray drying is that the primary particle morphology and the degree of agglomeration can be controlled. However, the costs of trial runs and the wide range of products properties complicate achieving this control of particle structure during spray drying. Single droplet drying approaches are employed to study the development of the primary particle morphology and the collision behavior of droplets under well-defined conditions. These studies shed light on the underlying mech-anisms but have been related less often to realistic drying and especially agglomeration. This review focuses on the potential and limitations of single droplet drying approaches to unravel the evolution of primary particle morphology and nozzle-zone agglomeration phenomena during spray drying. We discuss advances in single droplet drying approaches and how to combine these with pilot-scale spray drying to obtain the desired particle structures.

The indirect quantification of apparent wall slip of highly concentrated suspensions in pressure driven flows is commonly performed using a Mooney analysis. To prevent poor fits and physically impossible results, several modifications to the original analysis have been proposed in the past. The modifications are mostly empirical and, to date, there is no best practice. In this contribution, the origin of the failing original analysis is shown and the accuracy of several modified analyses is compared. Measurements are performed on a high-pressure capillary rheometer using dies with a smooth and rough internal surface and suspensions with liquid phases showing different shear rate dependencies, i.e. having flow indices between 0.20–1.0. For both types of die, a radial dependency is observed, which is related to shear-induced migration of the solids and macromolecules in the suspension. The original Mooney analysis cannot describe the changes in the local rheology and physically impossible results are a direct consequence. To include the radial dependency in the Mooney analysis, a best-fit approach is advised, until the underlying physics of shear-induced migration are better understood.

The rheological characterization of concentrated suspensions is complicated by the heterogeneous nature of their flow. In this contribution, the shear viscosity and wall slip velocity are quantified for highly concentrated suspensions (solid volume fractions of 0.55–0.60, D 4,3 ~ 5 µm). The shear viscosity was determined using a high-pressure capillary rheometer equipped with a 3D-printed die that has a grooved surface of the internal flow channel. The wall slip velocity was then calculated from the difference between the apparent shear rates through a rough and smooth die, at identical wall shear stress. The influence of liquid phase rheology on the wall slip velocity was investigated by using different thickeners, resulting in different degrees of shear rate dependency, i.e. the flow indices varied between 0.20 and 1.00. The wall slip velocity scaled with the flow index of the liquid phase at a solid volume fraction of 0.60 and showed increasingly large deviations with decreasing solid volume fraction. It is hypothesized that these deviations are related to shear-induced migration of solids and macromolecules due to the large shear stress and shear rate gradients.

Modelling the macroscopic rheology of non-Brownian suspensions is complicated by the non-linear behaviour that originates from the interaction between solid particles and the liquid phase. In this contribution, a model is presented that describes suspension rheology as a function of solid volume fraction and shear rate dependency of both the liquid phase, as well as the suspension as a whole. It is experimentally validated using rotational rheometry ( $$\varphi$$ ≤ 0.40) and capillary rheometry (0.55 ≤ $$\varphi$$  ≤ 0.60) at shear rates > 50 s−1. A modified Krieger-Dougherty relation was used to describe the influence of solid volume fraction on the consistency coefficient, $$K$$ , and was fitted to suspensions with a shear thinning liquid phase, i.e. having a flow index, $$n$$ , of 0.50. With the calculated fit parameters, it was possible to predict the consistency coefficients of suspensions with a large variation in the shear rate dependency of the liquid phase ( $$n$$ = 0.20–1.00). With increasing solid volume fraction, the flow indices of the suspensions were found to decrease for Newtonian and mildly shear thinning liquid phases ( $$n$$ ≥0.50), whereas they were found to increase for strongly shear thinning liquid phases ( $$n$$ ≤0.27). It is hypothesized that this is related to interparticle friction and the relative contribution of friction forces to the viscosity of the suspension. The proposed model is a step towards the prediction of the flow curves of concentrated suspensions with non-Newtonian liquid phases at high shear rates.

The efficient development of extrusion-based 3D-printing requires flexibility in both formulation- and process design. This task requires a fundamental understanding of the influence of material rheological properties on the extrusion process. Within this review, a qualitative toolbox for food extrusion is presented which provides guidelines for the formulation and engineering of extrusion processes in general and 3D-printing in particular. The toolbox is based on current knowledge of highly viscous food systems and the influence of individual components on the overall rheology. It includes the efficiency of particle packing, microstructure and the influence of shear rate, as well as the formation of self-supporting structures by gelation of the liquid phase and crowding of particles. Physical laws and semi-empirical equations are discussed to describe the rheology and relate relevant theory to the extrusion process. Practical information is presented, including examples of extrusion and 3D-printing of food and non-food systems. The qualitative extrusion toolbox provides a general framework for the emerging field of extrusion-based 3D-printing of food products. It can be used to identify which specific material and process parameters can be changed and how they may be altered to optimize the 3D-printing process. The general framework will assist researchers, as well as industry.

Wall slip quantification using the classical Mooney slip analysis has produced physically unreasonable results for many complex fluids. Over the past decades, the assumption that the slip velocity is solely dependent on the wall shear stress has therefore been questioned. In this contribution, the influence of the radius of cylindrical dies on the wall slip velocity of highly concentrated non-Brownian suspensions in a high-pressure capillary rheometer is re-examined, by varying the rheology of the liquid phase. Water- and oil-based suspensions (solid volume fraction similar to 0.60, D-4,D-3 similar to 5 mu m) are made using liquid phases that have different flow indices (n = 0.22 - 1.00) and a variation in their thickener concentration (25 g/L - 350 g/L). All classical Mooney plots showed negative y-intercepts and the fit worsened with decreasing flow index. A modification to the Mooney analysis is proposed that includes a geometrical dependency of the slip velocity that scales with the flow index of the liquid phase. Proposed modified Mooney plots not only show positive y-intercepts, but also show a good fit (R-2 > 0.99) over the entire range of shear rates (10 - 640 s(-1)) and corresponding wall shear stresses. The geometrical dependency is thought to arise from the shear stress gradient within the extrusion die.