Process optimization of extrusion variables and its effect on properties of extruded cocoyam (Xanthosoma sagittifolium) noodles

Abstract The current industrial demand for starchy foods has been dominated by other roots and tubers, while cocoyam, despite being rich in fiber, minerals, and vitamins has remained under exploited. In this study, the effect of feed moisture content (FMC), screw speed (SS) and barrel temperature (BT) on the quality characteristics of cocoyam noodles (proximate, thermo‐physical, physicochemical, texture, color, extrudate properties, and sensory characteristics) were investigated using central composite design (CCD) of response surface methodology (RSM). Flour was produced from fresh tubers of cocoyam (Xanthosoma sagittifolium) and subsequently processed into noodles using a twin screw extruder. Results showed that the proximate compositions, thermo‐physical, physicochemical properties, and color of the cocoyam noodles were significantly (p < 0.05) influenced by the extrusion process variables. The texture and extrudate properties of cocoyam noodles were equally significantly (p < 0.05) different. The experimental data obtained and predicted values of the response models were comparable, with statistical indices [absolute average deviation (AAD, 0–0.23), bias factor (B f, 1–1.08), and accuracy factor (A f, 1–1.23)] indicating the validity of the derived models. The optimal extrusion processing conditions for quality cocoyam noodles were FMC, SS, and BT of 47.5%, 700 rpm and 50°C, respectively, as cocoyam noodles obtained at these conditions had comparable properties and were most preferred and accepted by the sensory panelists.


| Experimental design and process optimization
Response surface methodology (RSM) was used to build up a mathematical model and assisted in qualitatively interpreting and describing the relationships between the three independent extrusion variables feed moisture content (X 1 ), screw speed rate (X 2 ), and barrel temperature (X 3 ). The three-factor design gave a total of 20 experiments as presented in Table 1. The responses investigated in this study were color, texture, thermo-physical, extrudate, and physiochemical properties. The regression model describing the relationship between the independent variables in terms of their linear, quadratic, and interaction effects is expressed by the second-order empirical polynomial equation as presented in Equation (1). where Y is the predicted response, β 0 is a constant, FMC is the feed moisture content, SS is the screw speed, BT barrel temperature, β 1 -β 3 , β 11 -β 33 and β 12 -β 23 are regression coefficients for intercept, linear, and quadratic effects, respectively.

| Extrusion process of cocoyam noodles
The noodles were produced using the modified method of Sobukola, Babajide, and Ogunsade (2013) in a laboratory scale twin screw extruder. The fabricated extruder had a barrel diameter, nominal screw length, restriction die and power of 65.2 mm, 1898 mm, 2 mm, and 5 hp, respectively (Sobowale, Ayodeji, & Adebiyi, 2017). Dough was prepared by mixing 100% cocoyam flour with a predetermined amount of water, to bring the moisture level to the different desired experimental moisture contents ( Table 1). The extruder was subsequently operated using the feed moisture content, screw speed, and barrel temperature combination obtained from the CCD experimental design (Table 1). After extrusion, the extrudates were cut into smaller pieces of 2.5 mm height each, cooled to 25°C, and packaged for subsequent analysis.

| Proximate analysis
Proximate composition of noodles (moisture, protein, fat, crude fiber, ash, and carbohydrate) was determined using standard analytical methods of AOAC (2006).

| Density (ρ)
Five grams of each sample was weighed and put into 100 ml measuring cylinder containing 50 ml water (as floatation liquid) and the density determined using simple floatation principles (Sobowale, Awonorin, Shittu, & Ajisegiri, 2014). The density was derived from the mass of sample divided by volume occupied.

| Specific heat capacity (C p )
The specific heat capacity was determined using two lagged copper calorimeters (Hussain & Rahman, 1999;Sobowale et al., 2014). The specific heat capacity was determined as follows: where M p , M w , and M c are the mass of sample, water, and calorimeter, respectively; C w and C c are the specific heat capacity of water and calorimeter, respectively; G w and G p are the slope of cooling curve for water and sample, respectively.

| Thermal diffusivity (α)
The methods of Tong, Sheen, Shah, Huang, and Lund (1993) and Sobowale, Awonorin, et al. (2017) were adopted using a probe connected by K-thermocouple wires to an Alda AVD 890C + digital multimeter. The temperature history of the sample was determined by inserting of the probe into the center (radial axis of the sample).

| Thermal conductivity (K s )
The thermal conductivity was estimated from the corresponding thermal diffusivity value and other thermo-physical properties such as specific heat (C p ) and bulk density (ρ) (Rapusas & Driscoll, 1995;Sobowale, Awonorin, et al. 2017). The thermal conductivity was then calculated using the expression in Equation (3).
where K s , α, ρ, and C p are thermal conductivity, thermal diffusivity, bulk density, and specific heat capacity, respectively.

| Determination of physicochemical properties
2.6.1 | Swelling capacity, solubility index, and water absorption capacity Swelling capacity was determined using the method of Olatidoye and Sobowale (2011), while solubility index was determined using the method described by Singh, Raina, Bawas, and Saxena (2005).
Water absorption capacity was determined using the method of Olatidoye and Sobowale (2011). (2)

| Amylose and amylopectin content
The amylose and amylopectin contents of the cocoyam flour were determined using the iodine calorimetric method (Udachan, Sahoo, & Hend, 2012).

| Color analysis
Image acquisition was carried out using a color digital camera (Nikon Cool Pix l21, Nikon Corp., Tokyo, Japan) connected to a computer USB interface and mounted on a stand inside a large box impervious to light with black inner surfaces (Yam & Papadakis, 2004

| Texture analysis (Hardness)
The hardness of the cocoyam noodle was determined using a texture analyzer (TA-XT2i, stable micro system, Haslemere, UK) following the procedure of Da Silva and Moreira (2008), which consists of a three-point bending test. Samples were placed on a metal support at a distance of 90 mm apart, and the force required to break the extrudates was determined by using a steel blade of 3 mm to snap the samples at a speed of 10 mm/s. The force (N) at the fracture point was used as the resistance to breakage.

| Cooking time
Using the methods of Sanni, Bamgbose, and Sanni (2004), the cocoyam noodle was cooked by immersion in boiling water and (5) a * = a × 240 255 − 120 TA B L E 3 Thermo-physical and extrudate properties of cocoyam noodles thereafter allowed to stay for few minutes. The different time taken for each of the samples to cook was recorded.

| Residence time
Residence time (RT) was determined during extrusion using the method of Iwe, Vanzuilichem, and Ngoddy (2001). A print of red food color was introduced at the feeding port, and the time taken for the color to first show up at the die orifice was recorded as the residence time.

| Mass flow rate
The mass flow rate (MFR) was determined when steady-state operation conditions were reached as indicated by constant torque at the

| Sensory evaluation of extruded cocoyam noodles
Prior to the sensory evaluation test, ethical clearance was obtained and informed consent of the sensory panelists were sort and gotten. Sensory evaluation of the extrudates was performed using 20 panelists. Each panelist was requested to assess each coded sample and to record the degree of difference using a 9-point Hedonic scale, based on appearance, taste, chewiness, glossiness, firmness, and overall acceptability.

| Statistical analysis
All analyses were carried out in triplicate and average of the trip- using Equations (7-9). The coefficient of determination (R 2 ) was also generated to compare the experimental and predicted values given by the models.

| Proximate composition
The moisture, ash content, fat content, fiber, protein, and carbohydrate content of the cocoyam noodles as presented in Table 2  indicative of a long shelf life when stored (Kure, Bahago, & Daneil, 1998;Olatidoye & Sobowale, 2011). There were significance differences in the ash content of the samples which was observed to be relatively high when the FMC was at 47.5%, BT 65°C and 700 rpm.
There were also significance differences (p < 0.05) in the fat con-

| Thermo-physical and extrudate properties
The thermo-physical properties of the cocoyam noodles as influenced by the extrusion process variables are shown in Table 3.
Bulk density gives an indication of the heaviness of the noodle samples and relative volume of packaging material required (Butt & Batool, 2010 Sanni, 2001). The obtained specific heat capacities of the cocoyam noodles were relatively high compared to the studies of Singh and Heldman (1993), Rapusas and Driscoll (1995) and Sobowale et al. (2014). Generally, the specific heat of a food may be influenced by the product properties (moisture content, temperature and pressure) and when high, the rate of energy conduction across/within the food sample is faster, and vice versa. However, in processing of pasta products, higher values of specific heat capacity usually lead to more energy transfer and improved heat transfer rate of the food sample (Baik & Mittal, 2003;Cengel, 1998).
The thermal diffusivity values obtained were ˂1, correlating well with other studies (Baik & Mittal, 2003;Nwanekezi & Ukagu, 1999;Sobowale et al., 2014). More so, the greater the density, the greater the contact between barrel surfaces, hence a corresponding higher thermal conductivity. With reference to the regression coefficient of the thermo-physical properties (Table 4), only the interactive effect of feed moisture content and screw speed rate (X 1 X 2 ) had a significant effect (p < 0.05) on bulk density of the cocoyam noodles. Further representation of the model on the surface plot ( Figure 1a) showed decrease in bulk density with increase in the FMC and SS rate. Other thermo-physical parameters, specific heat capacity and thermal conductivity concurrently decreased with increase in SS rate and BT. On the other hand, increase in FMC and SS rate resulted in an increase in the thermal conductivity ( Figure 1b-d).
The extrudate properties of cocoyam noodles as influenced by extrusion process variables are presented in reported that RT is a function of moisture content, feed rate, screw speed, barrel temperature, and screw geometry (Anuonye, Badifu, Inyang, & Akpapunam, 2007). Equally of significant importance is the specific mechanical energy which has been reported to influence product formation, geometry and expansion of the extrudate (Iwe et al., 2001). The ER of the extrudate sample increased with decrease in moisture content of the feeds. This is due to the fact that low moisture feeds exhibit more drag and therefore exert more pressure at the die, resulting into greater expansion at the exit of the die than for high moisture feeds (Arora, Zhao, & Camire, 1993;Bhattacharya, Kodiak, & Choudhury, 1994;Oluwole, 2008). Moisture is a major plasticizer in flours, which enables them to undergo glass transition during the extrusion process, facilitating matrix deformation and expansion. This study confirmed that SS generally has a positive effect (Figure 1f) on the expansion of the extrudates due to the increase in shear, subsequently leading to a decrease in melt viscosity induced by high SS (Ali, Hanna, & Chinnaswamy, 1996;Kokini, Chang, & Lai, 1992;Sobowale et al., 2016).  (Figure 1g). Higher SSs typically produce higher MFR due to the increased capability of the extruder screw to convey material through the extruder barrel. The relatively long CT observed in almost all the cocoyam noodle samples, thus suggests that the noodles during cooking will take up water more slowly. Accordingly, the regression coefficient of the model describing CT (Table 4)

| Color and texture
The color [lightness (L*), redness (a*), and yellowness (b*)] attributes of the cocoyam noodles are shown in Table 5. The color attributes  (Table 4), indicated that the positive quadratic effect of SS (X 2 2 ) and negative quadratic interactive effects of FMC and BT (X 1 X 3 ) significantly (p < 0.05) influenced the L* and b* of the cocoyam noodles. A similar trend was also demonstrated on the surface plots ( Figure 1I and 1K in the SS and FMC subsequently led to a decrease in the L* and increase in the b* of the noodles. Texture is an important and desirable attribute of extruded products. The textural values of the cocoyam noodles ranged from 11.93 to 16.17 N. These values were observed to significantly (p ˂ 0.05) differ, as a function of FMC, SS and BT (Table 5). More significantly, higher temperature extrusion usually results into a product with more air cells, lighter (reduced wall thickness), and softer extrudates (Joshi et al., 2014). Coefficient of the regression models of the texture with respect to their linear, quadratic, or interaction effects did not show significant (p < 0.05) effect, but the influences of these process variables observed are shown on the surface plots (Figure 1l).

| Physicochemical properties
The physicochemical properties of cocoyam noodles as influenced by FMC, BT, and SS are presented in Table 6. The WAC of the noodles sample varied between 0.87 and 1.84%, this trend in values could be attributed to the different extruding conditions.
According to Niba, Bokanga, Jackson, Schlimme, and Li (2001), WAC depends on the availability of hydrophilic groups that bind water molecules and has been used to estimate the suitability, bulkiness, and consistency of extrudates (Oluwole, 2008). SC of the sample generally indicates the level of starch content and the extent of gelatinization of inherent starch. Insufficient water uptake which is directly related to swelling usually results in noodles with hard and coarse texture, but excess water uptake has been linked to noodles that are too soft (Petitot, Boyer, Minier, & Micard, 2010).
(10) The computed coefficients of determination (R 2 ) were greater than 0.85, implying a better consonance between the experimental and predicted values (Tables 4 and 7). Previous studies have affirmed that good fit of empirical model and experimental data is depicted by  were most preferred, while sample obtained from FMC of 52.5%, TA B L E 6 (Continued) SS of 800 rpm, and BT of 45°C had the lowest score (4.25) and

| Sensory properties
were least preferred.
Raw food materials undergo physical and chemical modifications such as gelatinization, breakdown of starch, denaturation of proteins, and interactions between them resulting from high temperatures and pressures with combined of shearing effect during extrusion. These changes affected the sensory properties such as appearance, taste, chewiness, glossiness, firmness and overall acceptability of the extruded products. These were reflected in the significant differences (p ˂ 0.05) in all the sensory score obtained for the cocoyam noodles.
Accordingly, these sensory properties are important for extruded food products being developed as new entrants into the market.

ACK N OWLED G M ENT
Funding provided by the Nigerian government through the Tertiary Education Trust Fund (TETFUND) of the Federal Ministry of Education, is duly acknowledged.

CO N FLI C T O F I NTE R E S T
The authors declare that we do not have any conflict of interest.

E TH I C A L R E V I E W
This study was approved by the Polytechnic ethical committee.

I N FO R M E D CO N S E NT
Written informed consent was obtained all study participants.