Influence of processing conditions on apparent viscosity and system parameters during extrusion of distiller's dried grains‐based snacks

Abstract A combination of different levels of distillers dried grains processed for food application (FDDG), garbanzo flour and corn grits were chosen as a source of high‐protein and high‐fiber extruded snacks. A four‐factor central composite rotatable design was adopted to study the effect of FDDG level, moisture content of blends, extrusion temperature, and screw speed on the apparent viscosity, mass flow rate or MFR, torque, and specific mechanical energy or SME during the extrusion process. With increase in the extrusion temperature from 100 to 140°C, apparent viscosity, specific mechanical energy, and torque value decreased. Increase in FDDG level resulted in increase in apparent viscosity, SME and torque. FDDG had no significant effect (p > .5) on mass flow rate. SME also increased with increase in the screw speed which could be due to the higher shear rates at higher screw speeds. Screw speed and moisture content had significant negative effect (p < .05) on the torque. The apparent viscosity of dough inside the extruder and the system parameters were affected by the processing conditions. This study will be useful for control of extrusion process of blends containing these ingredients for the development of high‐protein high‐fiber extruded snacks.


| INTRODUCTION
Extrusion cooking is a popular food processing technique that has been extensively used to produce protein-rich and fiber-rich products.
Extrusion process is a combination of cooking, mixing and forming, resulting in good quality direct expanded product. It is widely used in industries owing to its characteristic high throughput and automatic control (Singh & Muthukumarappan, 2014b). During extrusion, the interactive effects of temperature, shearing forces, and moisture content of the blend transform the feed ingredients at macroscopic and microscopic levels leading to structural changes of protein and starch (Brown, Fallahi, Muthukumarappan, Singha, & Sindelar, 2015). Starch is gelatinized, protein is denatured and enzymes, microbes and many anti-nutritional factors are inactivated.
The extent of mixing, shearing and compressing of the materials, and the rate of heating inside the extruder and die, depends on the raw materials and process conditions used (Singh & Muthukumarappan, 2015). The extrusion process also depends on the pressure developed inside the die and the degree to which the screw is filled (Singh & Muthukumarappan, 2014a). Hence, it is important to understand the rheological changes encountered by an ingredient inside the barrel.
Each material that flows inside the extruder has its own properties and behaves differently. The behavior can be quantified by determining mass flow and energy responses. Apparent viscosity is one of the most important rheological properties which has a direct impact on the quality of a final product. A continuous monitoring system is often utilized to measure the apparent viscosity during the extrusion process (Chen, Jao, Larkin, & Goldstein, 1978;Lam & Flores, 2003).
It is a good indicator of dough's behavior and the changes during processing. Protein denaturation and polysaccharide gel formation can affect viscosity during extrusion (Bhattacharya & Hanna, 1986).
Separate attachments have been employed to measure the dough rheology of wheat, corn, and soybeans using straight tube viscometers (Harper, Rhodes, & Wanniger, 1971), cylindrical dies of different lengths (Harmann & Harper, 1974), capillary die rheometers (Singh & Muthukumarappan, 2016a,b), and viscoamylographs (Remsen & Clark, 1978) attached to food extruders. Not only the feed material but the quality of extrudates depends on the type of extruder used, choice of screw configuration, moisture content of the feed, temperature profile in the barrel and die, and screw speed (Singha & Muthukumarappan, 2016).
Dry-milling process involved in corn ethanol production, produces distillers dried grains (DDG) and distillers dried solubles (DDS).
Distillers dried grains with solubles (DDGS) is produced after mixing and drying these two co-products (Singh, 2016). DDGS contain high levels of protein since most of the starch is removed (Rosentrater & Krishnan, 2006). It is usually used as cattle feed. However, few studies have been reported on its application in human food (Rasco & McBurney, 1989;Rosentrater & Krishnan, 2006;Wu, Youngs, Warner, & Bookwalter, 1987). The growing interest in the health benefit of protein and fiber justifies exploring the use of DDGS as a protein and fiber supplement in food products. DDGS supplementation will improve the nutritive value of food products by enriching their protein and fiber content, and expand the use of the co-product from alcohol fermentation (Tsen, Eyestone, & Weber, 1982). DDG and DDGS have been extensively used as a protein source for the development of pet food and aquafeed. Few studies have also been reported on the addition of DDG in snack foods. Reddy and Stoker (1993) added DDGS in wheat flour for the preparation of noodles and baked foods.
Cereal or grains are the primary source of most extruded snack foods because of their high expansion properties. However, they tend to be low in protein and essential nutrients. Usually, cereals lack lysine as the essential amino acid but have sufficient sulfur containing amino acids. Legumes on the other hand, are rich in lysine and deficient in sulfur containing amino acids. When combined together, the proteins of cereals and legumes complement one another to produce a protein of a better quality. Corn in different form has been widely used as raw material for extrusion. It is ideal for extrusion since it has a high starch content, which facilitates the expansion process. It is gluten-free and contains protein, fiber, vitamins, and unsaturated fatty acids.
Compared to wheat, oat, and rice, corn has higher phytochemical content such as phenolic compounds which have anticarcinogenic effects (Jideani et al., 2014). Pulse crops (garbanzo, lentils, dry beans, lupin, and various types of beans) are excellent source of protein, complex carbohydrates, fibers, essential vitamins and minerals. Additionally, they are low in fat and sodium content, no cholesterol, and high in phenolics and bioactive compounds (Roy, Boye, & Simpson, 2010).
Inclusion of pulses and legumes increases the protein content and improves potential nutritional content owing to increase in protein digestibility. This was reported by Tiwari, Brennan, Jaganmohan, Surabi, and Alagusundaram (2011) while studying the addition of pigeon pea to wheat flour-based biscuits and, de la Hera, Ruiz-París, Oliete, and Gómez (2012) while studying the effects of addition of legume flour to traditional cereal-based flours. Madhumitha and Prabhasankar (2011) reported that there was an improvement in the nutritional value of pasta by adding black gram flour and mentioned that the processing of food material increases the value and shelf life of the product.
Extrusion treatment of lentil flours has also been linked with increase in some bioactive components (Morales et al., 2015).
Little information is available on the effect of extrusion on system parameters using garbanzo as one of the ingredients (Meng et al., 2010).
Blending of DDGS and garbanzo flour will serve as a good raw material for gluten-free healthy alternative snacks for a healthy conscious population. For our study, DDGS was processed for food application and was named FDDG. We blended garbanzo flour, FDDG and corn grits at different levels. Our objective was to study the effect of the feed moisture content, screw speed, and barrel and die temperature, on the apparent viscosity and the system parameters (specific mechanical energy, torque, and mass flow rate) during extrusion cooking.

| Raw materials and blend preparation
Distiller's dried grains with solubles (DDGS) was obtained from Glacial Lakes Energy LLC, Watertown, SD. It was then washed, freeze dried, steam/pressure sterilized, oven toasted, and ground to make a wholesome food-grade ingredient. The processing of DDGS for food application was done following method described by Rosentrater and Krishnan (2006). The DDGS processed specifically for food application studies is referred to as FDDG henceforth. The initial moisture content of FDDG was 0.70%. The proximate composition of FDDG was: 35.12% protein, 0.53% fat, 1.24% ash, 35.00% dietary fiber, and 13.03% nitrogen-free extract (dry basis). The FDDG was stored at −20°C until further use. Garbanzo flour (GF) was purchased from a local store in, Brookings, SD. The initial moisture content of GF was 10.08%. The proximate composition of the GF was: 22.42% protein, 5.94% fat, 2.70% ash, 6.56% fiber, and 54.63% nitrogenfree extract (dry basis). Corn grits (CG) was obtained from Bob's Red Mill (Milwaukie, OR). The initial moisture content of CG was 11.46%.
The different ingredients, that is, FDDG, GF and CG were mixed into five different compositions (Blend I-V) as shown in Table 1. Water was added to the blends to make 14%-20% (wet basis) final moisture depending on the experimental runs ( Table 2). The ingredients were mixed in a laboratory scale mixer (KitchenAid Professional 5 Plus, Troy, Ohio, USA) for 10 min. For moisture stabilization, the blends were stored overnight at ambient temperature. The moisture content of the prepared blends was determined using the method 44-19 (AACC, 2000). The proximate composition of blends is shown in Table 1.

| Extrusion processing
Post conditioning, the blends were randomly extruded using singlescrew laboratory extruder (Brabender Intelli-Torque Plasti-Corder ® , South Hackensack, NJ) having barrel inner diameter of 19.18 mm. A screw compression ratio of 1.5:1 was used in the experiments. A pictorial representation of the single screw extruder is shown in Figure 1 ( Singh & Muthukumarappan, 2014b). Pressure at the die, and net torque exerted on the extruder drive (N-m) were measured. Stock thermocouples (model 05-00-317, C. W. Brabender) were inserted into the barrel and die to measure the dough temperature. Extrudate samples were collected every 30 s to determine the mass flow rate (g/s) by the method described by Rosentrater, Richard, Bern, and Flores (2005). SME (W-h/kg) consumption was calculated (Lam & Flores, 2003) as: where Ω is the net torque exerted on the extruder drive (N-m), ω is the angular velocity of the screw (rad/s) and MFR is the mass flow rate of dough (mass throughput, g/s).
The apparent viscosity of the dough in the extruder was calculated by approximating extruder behavior as that of a coaxial viscometer but corrected for the tapered screw geometry (Figure 2) of the extruder barrel (Konkoly, 1997;Lam, 1996;Rogers, 1970). As discussed by Lam and Flores (2003), the shear stress (τ s ) at the screw surface (N/ m 2 ) and the shear rate (̇γ s , 1/s) were calculated from the following equations: where r corr is the radius correction due to the screw's frustum geometry r eff is the effective radius (m), Ω is the net torque exerted on the screw (N m), L s is the screw length in the axial direction (m), ω is the angular velocity of the screw (rad/s), C ss is an empirical correction factor for shear stress (which is 10321.5 for this study), ̇γ s is the shear rate at the screw surface (1/s), r b is the inner barrel radius (m), and C sr is the empirical correction factor for shear rate (which is 3.48 for this study). The calibration value for this extruder has been calculated from the calculation reported elsewhere (Lam & Flores, 2003). The apparent viscosity was calculated by taking the ratio of Equations (2) and (3). where η app is the apparent viscosity of the dough in the extruder (Pa.s). (

| Experimental design and statistical analysis
Experiments were conducted using the central composite rotatable design which was developed using Design-Expert 8.0.7.1 (Statease, Minneapolis, MN, USA). Four numerical independent variables namely FDDG (X 1 ), temperature (X 2 ), screw speed (X 3 ), and moisture content (X 4 ) each at five levels were taken as shown in Table 2 The experimental design and the codes for the processing variables are shown in Table 3.
Mass flow rate (Y MFR ), specific mechanical energy (Y SME ), apparent viscosity of dough (Y AV ), and torque (Y Tor ) were taken as the responses.
Second-order polynomial regression models were established for the dependent variables to fit experimental data for each response.
where y i is the predicted response; b 0 is the interception coeffi-

| Effect of processing conditions on apparent viscosity
The fitted model shown in Table 4 had a significant coefficient of determination (R 2 ) of 0.82. The second order model ( increased the protein content. The rise in FDDG content changed the overall chemical composition and the dough functionality which may have resulted in the higher apparent viscosity. Bhattacharya and Hanna (1986) also reported an increase in viscosity when the percentage of soy protein content in the blend was increased during extrusion of corn-soy mix. (6)

| Effect of processing conditions on mass flow rate
The drag flow developed by screw rotation inside extruder and the pressure developed due to constriction at the die influences the mass flow rate (Ludewig, 1989). Multiple regression equation for MFR (Y MFR ) in terms of coded variables is shown in Table 4. ANOVA for the model of MFR as fitted (Table 5) shows that the model was significant (p < .05), whereas lack of fit was not significant (p > .05).
The response surface regression model on MFR yielded a good fit

Coded variable
Actual variable x 1 x 2 x 3 x 4 X 1 (%) X 2 (°C) X 3 (rpm) X 4 (% wb)   F I G U R E 3 Response surface plots for apparent viscosity as a function of (a) Screw speed and FDDG at 120°C temperature and 17% moisture content; (b) Temperature and FFDG at 150 rpm screw speed and 17% moisture content F I G U R E 4 Response surface for mass flow rate as a function of screw speed and moisture content at 120°C temperature and 10% FDDG level reason, higher screw speeds means higher mass flow rate and there is greater ability for the material to move along the extruder barrel (Harper, 1981). Increasing the moisture content from 14% to 20% also significantly (p < .05) increased the MFR. High moisture content aides in gelatinization of the dough and thus apparent viscosity decreases which has been observed in this study. This explains the increase in MFR with increase in moisture content of the dough. No significant effect (p>.05) of FDDG on the MFR was observed.

| Effect of processing conditions on specific mechanical energy
A multiple linear regression equation of SME (Y SME ) in terms of coded levels is shown in Table 4. Linear terms of FDDG (x 1 ), temperature (x 2 ), screw speed (x 3 ), and moisture content (x 4 ) had significant effects (p < .05) on SME. Temperature had a significant negative quadratic effect (p < .05) on SME. SME increased with increase in temperature, while excessive increase in temperature resulted in decrease in SME. The fitted quadratic model had a coefficient of determination (R 2 ) of 0.89. The model (Table 6) for SME was significant (p < .05), whereas lack of fit was not significant (p > .05). we have observed previously that viscosity increased with increase in FDDG. This indicates that higher energy and pressure is required during the extrusion of the blends containing higher percent of FDDG. The high fiber content has a tendency to bind more water, resulting in reduced availability of water for starch (Mir, Bosco, Shah, Santhalakshmy, & Mir, 2015). SME increased initially with increase in temperature but further increase in temperature resulted in decrease in SME. High temperatures are normally associated with a decrease in the melt viscosity inside the extruder, which in turn reduces the energy input of the extruder. According to Ludewig (1989), with increase in screw speed, the SME generally increases.
This is because the magnitude of change in energy input to the screw is typically greater than the decrease in torque associated with the decrease in apparent viscosity due to shear thinning behavior of the non-Newtonian materials. The high SME at high screw speed and low barrel temperature was also observed by Meng et al. (2010)

| Effect of processing conditions on torque
The multiple regression equation for torque (Y Tor ) in terms of coded variable is given in Table 4. The torque was influenced significantly (p < .05) by negative linear effects of FDDG (x 1 ), temperature (x 2 ), screw speed (x 3 ), and moisture content (x 4 ) suggesting that increase in the levels of these variables resulted in decrease in torque.
Temperature also had significant negative quadratic effect (p < .05) on torque indicating excessive increase in temperature reduced the torque of the extruder. Interaction effects of FDDG and screw speed (x 1 x 3 ) and temperature and moisture content (x 2 x 4 ) were also observed. The responses were analyzed using ANOVA and the data are presented in Table 6. Examination of the model shows a good fit with R 2 equal to 0.95 for the torque. The linear model was significant (p < .05), whereas lack of fit was not significant (p > .05) for torque.
Torque increases with increase in FDDG level (Figure 6a) which also means that it decreases with increase in garbanzo flour. This is F I G U R E 6 Response surface plots for torque as a function of (a) FDDG and temperature at 17% moisture content and 150 rpm screw speed; (b) Screw speed and temperature at 17% moisture content and 10% FDDG level; (c) Moisture content and screw speed at 10% FDDG level and 120°C temperature in agreement with findings of Bhattacharya and Prakash (1994). The torque during extrusion ranged between 9 and 19 N.m and high torque was associated with low screw speed. The effect of screw speed on torque also depends on the level of temperature. At high extrusion temperature, the response surface plot (Figure 6b) shows that torque value is almost constant with change in screw speed but at low extrusion temperature torque increased with decrease in screw speed. Filli et al. (2012) also reported decrease in torque with increase in screw speed and feed moisture during single screw extrusion of milletsoybean mixture. Since the blends showed shear thinning behavior inside the extruder, the net torque required by the screw to convey the dough through the extruder decreased significantly. According to Guha, Ali, and Bhattacharya (1997), the decrease in the magnitude of torque with increase in screw speed can be explained by the reduced degree of fill in the extruder. Decrease in torque with increasing moisture content (Figure 6c) suggests more water is available for starch gelatinization resulting in reduction in apparent viscosity. Melt viscosity is low at high moisture contents and hence less torque will be required to work the material in the screw channels. Reduction in torque can be attributed to reduced friction in the extruder because of increase in feed moisture. This indicated that increasing moisture content or screw speed reduces the difficulty of processing. Similar results were found by Onwulata, Mulvaney, and Hsieh (1994) during twin screw extrusion of corn meal and by Chang and El-Dash (2003) during extrusion of cassava.

| CONCLUSIONS
The apparent viscosity, MFR, torque, and SME were shown to be significantly influenced by the extruder operating conditions. FDDG level and extrusion temperature significantly affected the apparent viscosity, torque and SME. Apparent viscosity increased with the increase in FDDG content. Higher feed moisture and higher extrusion temperature reduced the viscosity. With increase in the screw speed and feed moisture content the MFR also increased. Increasing the moisture of the blends and the extrusion temperature resulted in a decrease in torque. High SME was observed at high screw speed and low extrusion temperature.