4.3.8. Pasta Cooking Behavior

The loss of solids CL (%) during pasta cooking was determined gravimetrically by evaporation of the water that resulted after boiling 10 g of pasta in 200 mL of water for the optimum cooking time previously established [66].

#### 4.3.9. Boiled Pasta Texture

Pasta chewiness Ch (J) was determined by double cycle compression on one piece of pasta by using a Perten TVT-6700 device (Perten Instruments, Sweden) equipped with a 35 mm cylinder probe, at 50% of the sample height, a test speed of 5.0 mm/s and a trigger force of 20 g [62].

4.3.10. Rapid Digestible Starch (RDS), Slowly Digestible Starch (SDS) and Resistant Starch (RS) Contents

The international AOAC 2017.16 method was used for RDS, SDS and RS determination from boiled pasta, by using Megazyme kit. After 20 min (for RDS), 120 min (for SDS) or 240 min (for RS) of sample digestion with α-amylase and amyloglucosidase the reaction was stopped and the mix was digested again with amyloglucosidase. The resulting glucose was determined by using GOPOD reagent and reading of the absorbance at 510 nm. The results were reported as percent to dry matter.

#### *4.4. Optimization of Grape Peels Level and Models Validation*

GP level optimization was performed by using the trial version of Design Expert software (Stat-Ease, Inc., Minneapolis, USA). For this purpose, the multiple response optimization and the desirability function were used and the goals were selected as follows: G\*, Co, C\*, F, TPC, TDF and RS were maximized, CL and Ch were minimized, and ηmax\* was kept in range. The experimental design matrix containing mean values of three replications of the responses is presented in Table 4.



**Table 4.** The effects of GP on the responses used in the experimental design.

For model validation, pasta was made using the optimal level of GP obtained and the response values were checked. The real values of the optimum sample characteristics were compared to the control made of untreated wheat flour. For the evaluation of differences among the predicted and the experimental results of the optimal solution, and among the optimal and control sample, *t* tests for two samples (*p* < 0.05) were performed.
