**2. Results**

#### *2.1. Diagnostic Checking of the Models*

The ANOVA results for the model fitting presented in Table 1 show that all of the mathematic models chosen were significant and predicted accurately the responses, the *F*-value being significant (*p* < 0.01) in all cases and *R <sup>2</sup>* values being more than 0.76.

**Table 1.** ANOVA results for the fitted models for different characteristics of flour, dough and pasta.


ηmax—peak viscosity, G\*—complex modulus, Co—cohesiveness, C\*—chroma, F—fracturability, Ch—chewiness, CL—cooking loss, RS—resistant starch, TPC—total polyphenols content, TDF—total dietary fiber.

Flour peak viscosity (ηmax), dough cohesiveness (Co) and boiled pasta chewiness (Ch) data were fitted to the quadratic model, which described 83%, 76% and 78% of the data variation, respectively. Dough complex modulus (G\*), dry pasta chroma (C\*), fracturability (F), cooking loss (CL) and polyphenolic content (TPC) results were fitted to the quartic model, with 91%, 91%, 98%, 97% and 96%, respectively, of the data variation being explained. The cubic model explained 98% of data variation for resistant starch content (RS), while the results for the dietary fiber content (TDF) were fitted to the fifth model, which explained 99% of the variation.

#### *2.2. Effects of GP on Flour and Pasta Characteristics*

A Supplementary Materials section containing the graphics for the variation of the responses with GP level is provided.

Peak viscosity increased with GP level increase (Figure S1), the biggest significant (*p* < 0.01) positive influence being obtained for the linear term Equation (1).

$$
\eta\_{\text{max}} (\text{Pa} \cdot \text{s}) = 0.42 + 0.14 \text{A}^{\*\*} - 0.01 \text{A}^2 \tag{1}
$$

where ηmax—peak viscosity, A—GP level, \*\* significant at *p* < 0.01.

An increase of G\* as the GP level was higher was observed (Figure S2a), the linear and quadratic terms having significant influence (*p* < 0.05) on the response Equation (2).

$$\mathbf{G^\ast (Pa)} = 96083.29 + 23997.76 \mathbf{A^{\*\*}} + 38352.65 \mathbf{A^{2\*}} + 808.26 \mathbf{A^3} - 27020.94 \mathbf{A^4} \tag{2}$$

where G\*—complex modulus, A—GP level, \*\* significant at *p* < 0.01, \* significant at *p* < 0.05.

More cohesive dough was obtained as the GP addition level was higher (Figure S2b), the linear term having the greatest significant (*p* < 0.01) influence Equation (3). The negative effects of GP, which are a fiber-rich ingredient, were minimized because small particle size (<180 µm) was used.

$$\text{Co} = 0.40 + 0.02 \text{A}^{\*\*} + 0.01 \text{A}^2 \tag{3}$$

where Co—cohesiveness, A—GP level, \*\* significant at *p* < 0.01.

Pasta color was affected by GP incorporation, a significant (*p* < 0.01) decrease of C\* being observed with the addition level increase (Figure S3a). The biggest negative influence was obtained for the linear term of the factor Equation (4).

$$\mathbf{C}^\* = \mathbf{23.08} - 2.72\mathbf{A}^{\*\*} + 1.42\mathbf{A}^2 - 0.35\mathbf{A}^3 + 0.03\mathbf{A}^4 \tag{4}$$

where C\*—chroma, A—GP level, \*\* significant at *p* < 0.01.

Dry pasta fracturability expressed as the maximum force needed to break the sample can be an indicator of pasta resistance to transport and manipulation. The addition of GP caused an increase of F with the level increase (Figure S3b). The linear and cubic terms presented significant positive influence (*p* < 0.05), while the quadratic term had a negative effect on the response Equation (5).

$$\mathbf{F(g)} = 3158.72 + 1837.63 \mathbf{A^{\*\*}} - 1052.11 \mathbf{A^{2\*\*}} + 256.32 \mathbf{A^{3\*}} - 19.84 \mathbf{A^{4\*}} \tag{5}$$

where F—fracturability, A—GP level, \*\* significant at *p* < 0.01, \* significant at *p* < 0.05.

GP addition caused a CL rise with the level increase (Figure S4a), the linear, quadratic and quartic terms presenting significant (*p* < 0.05) influences based on Equation (6). An acceptable cooking loss value should be less than 12% [5].

$$\text{CL}(\%) = 0.66 + 6.33 \text{A}^{\*\*} - 3.02 \text{A}^{2\*} + 0.61 \text{A}^3 - 0.04 \text{A}^{4\*} \tag{6}$$

where CL—cooking loss, A—GP level, \*\* significant at *p* < 0.01, \* significant at *p* < 0.05.

Pasta chewiness is expressed as the energy required to chop the sample until it is ready to swallow [29]. GP level increase caused a proportional decrease of pasta chewiness (Figure S4b), the highest influence being observed for the linear term Equation (7).

$$\text{Ch} = 3696.18 - 287.20 \text{A}^{\*\*} + 79.49 \text{A}^2 \tag{7}$$

where Ch—chewiness, A—GP level, \*\* significant at *p* < 0.01.

Resistant starch content was significantly influenced (*p* < 0.01) by the linear, quadratic and cubic terms in Equation (8). A rise in RS with GP addition level increase was observed (Figure S5a).

$$\text{RS}(\%) = 4.49 + 0.34 \text{A}^{\*\*} - 0.43 \text{A}^{2\*\*} + 0.50 \text{A}^{3\*\*} \tag{8}$$

where RS—resistant starch content, A—GP level, \*\* significant at *p* < 0.01.

A significant positive influence (*p* < 0.01) was obtained for the linear term of GP level, while the quadratic and quartic terms exhibited a negative and significant (*p* < 0.05) effect on TPC response (Equation (9). TPC showed higher values with GP level increase (Figure S5b), as a result of polyphenols present in the added ingredient.

$$\text{TPC}(\%) = 29.25 + 117.50 \text{A}^{\*\*} - 55.362 \text{A}^{2\*} + 11.17 \text{A}^3 - 0.77 \text{A}^{4\*} \tag{9}$$

where TPC—total polyphenols content, A—GP level, \*\* significant at *p* < 0.01, \* significant at *p* < 0.05.

GP are a rich source of soluble and insoluble dietary fibers, their incorporation in wheat pasta determining an increase of TDF with the level increase (Figure S5c). Significant positive influences (*p* < 0.01) were obtained for the linear, cubic and fifth terms, while the quadratic and quartic terms presented negative effects on the response (Equation (10)).

$$\text{TDF}(\%) = -4.58 + 9.17 \text{A}^{\*\*} - 6.31 \text{A}^{2\*\*} + 2.01 \text{A}^{3\*\*} - 0.30 \text{A}^{4\*\*} + 0.02 \text{A}^{5\*\*} \tag{10}$$

where TDF—total dietary fiber content, A—GP level, \*\* significant at *p* < 0.01.

#### *2.3. Optimization of GP Level and Models Validation*

To obtain the maximum nutritional benefits with minimum quality characteristics' impairment, the optimization of GP level as a function of the considered responses showed that wheat flour can be supplemented with 4.62% GP (Table 2), with a desirability of 0.57.


**Table 2.** Confirmation of the optimized parameters and control sample characteristics.

OGP—optimal formulation of wheat flour with grape peels, A—GP (grape peels) level, ηmax—peak viscosity, G\*—complex modulus, Co—cohesiveness, C\*—chroma, F—fracturability, Ch—chewiness, CL—cooking loss, RS—resistant starch, TPC—total polyphenols content, TDF—total dietary fiber, means in the same row followed by different letters (x–y for differences among predicted and observed values, a–b for differences between OGP and control) are significantly different (*p* < 0.05), \* relative deviation = [(experimental value − predicted value)/experimental value] × 100.

> For the model's validation, a pasta sample was made using the optimal level of GP that resulted after optimization. The responses were checked in triplicate and the experimental values were less than 5% different from the predicted ones (Table 2), except for chewiness, which was lower by 6.86% than the predicted value. Compared to the control, significantly (*p* < 0.01) higher G\*, ηmax, dough Co, F, CL, RS, TPC and TDF contents were obtained, while *C\** and boiled pasta Ch were smaller (Table 2). Consequently, the nutritional and functional values of the optimized pasta were enhanced compared to the control and the quality parameters were kept. Even if higher CL was obtained (6.81%), the value was less than 12%, the limit recommended for acceptable pasta. The OGP sample presented a more cohesive, elastic and viscous dough, which was probably related to the higher resistance to break (F) of pasta, which was desirable.

### *2.4. Determination of Control and Optimal Product Properties*
