Impact of fluidized bed granulation on structure and functional properties of the agglomerates based on the durum wheat semolina

- The granulation step determines the production yield and the final characteristics 11 of the agglomerated couscous grains of durum wheat. The objective of the present work was 12 to explore the capability of the fluidised bed technology to produce agglomerates of durum 13 wheat semolina. The impacts of different processing conditions have been investigated on the 14 structure and functional properties of the agglomerates. The size, shape, water content, 15 compactness, and mechanical strength of the granules were measured. The fluidized bed 16 agglomeration process has been found to produce agglomerates of durum wheat with 17 different attributes compared to those produced by granulation using the low shear mixers. 18 The results were discussed in regard to the hydro-textural approach, in order to get a better 19 understanding of the mechanisms and relationships between process, structure, and 20 properties. Two major agglomeration mechanisms contribute to the growth of the wet 21 agglomerates: a fractal-structuring process followed by a phenomenon of densification. By 22 studying the evolution of the compactness, diameter and water content, it was demonstrated that inter granular arrangements led to an expansion followed by a densification of the wet agglomerates. A relationship was proposed to describe the growth using a fluidized bed of the wet agglomerates of durum wheat semolina.

were carried out. Semolina were characterized using standardized methods (Table 1). The  were significantly higher than the true water content measured in the bed at the end of the 128 process (Fig. 2). An almost constant value of the drying rate (11.0 ±0.9 g of water / min) was 129 determined whatever the process conditions. Immediately after the end of the water addition 130 stage, the wet agglomerates were collected. Trials were conducted in triplicate.

228
The physicochemical characteristics of the wet agglomerates were measured for each class of 229 agglomerates as a function to their diameter (Fig. 4). The water content of the wet

234
The increase in the median diameter of the wet agglomerates is also associated to changes in  The results also demonstrated a positive relationship between the values of sphericity and the 239 diameter of the wet agglomerates (Fig. 4c). The large agglomerates were more spherical than  The impact of the water content of the bed was studied by applying different processing times 250 for water spraying. The physicochemical characteristics of the wet agglomerates were 251 measured for each class of wet agglomerates according to their diameter, and plotted for the 252 different measured true water content from 0.19 to 0.48 g/g dry matter ( Fig. 5-7). Increasing 253 the water content did not significantly change the shape of the curve of the size distribution, increased with values of water content higher than 0.37 g/g dry matter.

262
The relationships between the diameter and measured water contents for the different wet phenomena were observed whatever the amount of added water added during processing.

278
Whatever the amount of added water, the large agglomerates were more spherical than the 279 small ones (Fig. 7c). An increase in the amount of water did not have a significantly impact   Table 2).

315
The steam cooking stage and the drying stage did not change the shape of the distribution 316 curves of the diameters of the agglomerates of semolina (Fig. 10). The size distribution of the 317 wet agglomerates was almost fixed by the wet granulation stage (Table 2)   The drying stage induced an increase in the compactness (0.49) and a decrease in the water 339 content (0.14) of the agglomerates ( Table 2). The dewatering generates slight mechanisms of 340 shrinking, which induce a decrease of the internal porosity and diameter of the agglomerates.

341
The drying stage contributes to the texturation of the grains by the densification of the 342 structure. 343 We observed that swelling and shrinking mechanisms generated by the successive steam

361
The mechanical properties of the agglomerates were measured by the mean force required to 362 compress them (Table 3) positive slope at d 50 ≥ 1600 µm ( Fig. 4b and 7b). For d 50 ≥ 1600 µm, the value of the compactness slightly increased with the diameter of the 406 agglomerates ( Fig. 4b and 7b). This increase in compactness could be associated to the values 407 of D f close to 3. We could then suppose that some phenomena of densification of the grains 408 occurred for the larger structures (d 50 ≥ 1600 µm) due to collisions with the wall and/or 409 between the agglomerates themselves in the fluidized bed velocity field.

422
The hydrotextural diagram is limited in its upper part by the saturation curve, which 423 represents the maximum water content that a grain of a given constant compactness can 424 contain (Fig. 12). The saturation curve is calculated using Eq. (4): Whatever the process conditions, the values of water content and compactness measured for 429 the wet agglomerates were located under the saturation curve (Fig. 12). The wet agglomerates -15 -between the particles that was observed by a variation of the compactness (Fig. 12) and the 435 saturation degree.

436
The saturation degree increased (from 0.20 to 0.44) with an increase in the water content: 437 experimental values became closer to the saturation curve (Fig. 12). It was observed that the 438 growth could be fitted by a sigmoidal curve, comparable to this purposed by Ruiz et al.

503
The agglomeration process using a fluidized bed has been found to produce agglomerates of 504 durum wheat with different attributes compared to those produced by granulation using low 505 shear mixers. The dried agglomerated grains obtained using the fluidized bed were less 506 compact than the traditional couscous grains manufactured using low shear mixers. Fluidized The authors would like to thank the Agence Nationale de la Recherche (ANR ALID 2013) 529 for its financial support through the program "Dur Dur".         Values are means (± standard deviation).

714
Values in column with the same letter were not significantly different (P<0.05).  Values are means (± standard deviation).

722
Values in column with the same letter were not significantly different (P<0.05).