Parametric optimality of tray dried Musa balbisiana Colla blossom

Abstract

Musa balbisiana Colla blossom has enriched applications as a key constituent of dried vegetable formulations. With restricted prior art, the article addresses the optimality of tray drying characteristics of the blossom from both statistical design and drying kinetics perspective. The process variables in due course of optimization refer to moisture content, antioxidant activity and vitamin C for variation in drying time and temperature. Model fitness, analysis of variance based analysis and numerical optimization were considered during the statistical design of experiments. Drying kinetics involved fitness studies of alternate models, moisture diffusivity and process variable characteristics. Thereby, the sensitivity of both approaches to obtain optimal parameters associated with tray dried product have been targeted for a comparative assessment.

Introduction

With growing demand and market for functional and nutritionally superior food products such as ready to eat and cook products, there has been a strong emphasis on nutritionally rich food products deduced from leafy and non-leafy vegetables. In these products, antioxidant and vitamin C are a few important components that are of primary concern. Apart from quick rehabilitation of intensive care unit (ICU) patients (Hemilä and Chalker 2019) and good skin health (Pullar et al. 2017), the primary emphasis to have adequate vitamin C content is with respect to its role to enhance haemoglobin in anaemic peoples. On the other hand, antioxidants boost the immune system, neutralizes free radicals, fight against various diseases such as cardiovascular diseases, cancer etc.

Kolphul (Musa balbisiana Colla blossom), a by-product of banana cultivation is such an underutilized vegetable with high protein, fat, vitamin C, mineral, phytochemicals and polyphenol content. It is used as a vegetable in curry, boiled or fried form across North-East India. It is advantageous from the perspective of its rich fibre content that reduces the risk of cardiovascular diseases and the lowering of blood cholesterol (Wickramarachchi and Ranamukhaarachchi 2005). Traditionally, the banana blossom is used to treat ulcer, constipation and bronchitis (Kumar et al. 2012).

With moderate infrastructure in North-East (NE) India, a low cost, scalable, energy saving and effective drying technique is most relevant for processing of various horticultural produces. Among various alternative drying processes, intermittent airflow assisted tray drying is effective for the effective moisture removal as well as nutritional retention. In a recent work, our research group (Mondal et al. 2019) inferred upon the efficacy of tray drying in comparison with oven drying to achieve better retention of nutritional characteristics. Similarly, among oven, shade and sun drying, Satwase et al. (2013) inferred better retention of nutritional constituents using tray drying process for mango kernels. Further, it is important to note that the intermittent air flow operation enables significant reduction in electrical energy consumption due to the stationary phase of the air blower. In summary, the tray drying characteristics of banana blossom are anticipated to serve as a useful guideline for setting up relevant food processing operations in NE India and promote rural entrepreneurship.

Drying process causes physical, chemical, physiological and thermal changes in the product. Therefore, it is very important to understand the variations in product characteristics in drying. Drying involves removal of moisture along with other volatile components in due course of drying. This influences the value-added product constitution. Therefore, it is very important to understand the effect of drying on the optimality of product from process parametric perspective. Drying process characteristics of horticultural produce is an important area of research to facilitate the development of ready to consume food products. Since such products are developed as a combination of several vegetables, the compatibility of dried constituents is to be determined by evaluating the drying characteristics. The drying characteristics typically involve two major approaches. Firstly, using trial and error based approach involving drying kinetics, moisture diffusivity, heat sensitive responses such as vitamin C and antioxidant activity variations are targeted. This is very important to achieve a blend of two or more vegetables in formulating a product mix such as mix vegetable soup formulations and achieve superior product formulation. Secondly, response surface methodology (RSM) approach enables a deeper insight into the combination of dependent and independent variables of the product-process system.

A critical analysis of the available state of the art is indicative towards limited investigations of banana blossom drying characteristics. Mathematical modelling of thin layer drying of chopped and buttermilk soaked banana blossom was addressed (John et al. 2014). The authors evaluated moisture kinetics as well as fat, ash, protein, crude fibre, total polyphenols and total flavonoids for 1 m/s air velocity and 40–60 °C drying temperature case. For chopped and buttermilk soaked banana blossom, the best product was inferred at 60 °C and 195 min of drying time. Similarly, drying kinetics studies on squash (Potosí-Calvache et al. 2017), pumpkin (Guiné et al. 2011), turnip (Gharehbeglou et al. 2014) and unripe banana (Olawoye et al. 2017) are available in the literature. On the other hand, the RSM based optimization of hot air drying was carried out for cauliflower (Gupta et al. 2013), Okara (Wang et al. 2016) and red currant (Šumić et al. 2016) but not banana blossom. However, for the banana blossom, few literature addressed relevant insights into the nutritional characteristics of dried banana blossom. Notable among these are studies addressing anti-oxidative and anti-hyperglycemic potential of oven dried banana blossom (Marikkar et al. 2016), biological activity of shade dried banana blossom (Divya et al. 2016), proximate characteristics of oven dried banana blossom (Awedem et al. 2015) and proximate, vitamin C and antioxidant activity characteristics of tray dried banana blossom (Mondal et al. 2019). All these investigations were addressed for a specific combination of drying temperature and time. Thus, the available literature does not indicate an integrated study of kinetics and RSM based optimization of intermittent airflow assisted tray drying process associated with banana blossom or its food product. Secondly, while moisture kinetics evaluation has been conducted for banana blossom (John et al. 2014), there is no affirmity towards key ingredients such as vitamin C and antioxidants. Moreover, chopped and buttermilk soaked blossom is not promising as it leads to loss of nutrients and reduces desired attributes of value-added products. Thirdly, trial and error based optimization addressed in the literature did not consider nutritionally important parameters such as vitamin C and antioxidant activity.

Considering the lacunae, this work targets the combinational optimization of process parameters based on kinetics and RSM studies. Kinetic evaluation involves evaluation of drying kinetics for moisture, variation of vitamin C and antioxidant activity with drying temperature and time. This enables to deduce useful insights into the compatibility of banana blossom with other non-leafy vegetables and its coherence towards product formulation. On the other hand, RSM based optimization provides the best process variables and does not require a careful selection of independent variables as considered in trial and error based approach. Such a combinational methodology would ensure deeper insights into the complexities associated with the drying process. Further, such an approach can generalize influences towards drying based design of food products. Finally, for best product-process conditions, proximate parameters were evaluated to assist furthering efforts towards high quality banana blossom food formulation.

Materials and methods

Raw materials and sample preparation

Fresh Kolphul samples were procured from Shingimari area, Kamrup, Assam, India (26.22ºN and 91.62ºE) and were kept in tightly knotted polyethylene pouches to prevent degradation. The samples were washed in tap water to remove dirt and other unwanted materials adhering to them. Thereafter, the samples were thoroughly rinsed to remove excess water. The successive blossom was taken after discarding first three bunches of the outer blossom of the banana flower. Each such blossom sample was cut longitudinally into two equal halves at its base and spread on the tray for drying in a tray dryer. The average dimensions (length × width) of the raw banana blossom sample were 8.6 cm × 0.5 cm. The average blossom thickness was 1 mm.

Chemicals

L-ascorbic and Bradford reagent were purchased from Sigma Aldrich. Oxalic acid dehydrate, sodium hydroxide pellets, dextrose, absolute methanol, sulphuric acid (98%), petroleum ether and perchloric acid were obtained from Merck India while sodium bicarbonate was provided by Rankem. Bovine Serum Albumin, 2, 6-dichlorophenol indophenol sodium salt (DCPIP) and 2,2-Diphenyl-1-Picrylhydrazyl (DPPH) extrapure were procured from SRL Pvt. Ltd., India.

Drying method

Drying experiments were performed in a tray dryer (model: assembled, make: International Commercial Traders, Kolkata, India. Equipment specification: Digital temperature controller 25–300 °C; Blower operated with air velocity of 4.5 m/s, on/off mode 20/20 s; Overall dimensions—24″ × 24″ × 24″; Tray dimensions—18″ × 18″ × 2″; Insulation thickness 2″; 4 trays; 3 kw single phase heating load) under intermittent airflow condition (4.5 m/s air velocity, 20 s run mode and 20 s stationary mode of pulse airflow). The sample was kept in a single layer on the tray. Kinetic studies were carried out in the temperature range of 40–80 °C for constant weight. For RSM based study, drying temperature and time combinations were based on RSM based rotatable central composite design (CCD). Triplicate experiments were conducted for each combination of drying temperature and time. Thereafter, average values of the variables have been evaluated along with their standard deviations. A pictorial representation of Kolphul drying process has been shown in Fig. 1.

Fig. 1.

Pictorial representation of Kolphul drying process

Drying characteristics curve

Drying characteristics curves were obtained by monitoring the time-dependent variation in sample weight in due course of drying until the constant weight of the sample was obtained. The drying was carried at 5 different temperatures (40, 50, 60, 70 and 80 °C). Thereafter, graphs were plotted between moisture ratios (MR) and time, where MR was determined using the expression:

$$MR=\frac{M_t-M_e}{M_0-M_e}$$ 1

where M₀, M_t and M_e correspond to initial moisture content (%, dry basis), moisture content (%, dry basis) at any time and equilibrium moisture content (%, dry basis) respectively.

The time-dependent variation of antioxidant activity and vitamin C were analysed with a plot of these variables with drying time for 50, 60 and 70 °C cases. The samples were analysed with 2 h duration for these variables until a constant weight was achieved.

Fitness of drying curve models

A set of empirical and non-empirical models summarized in Table 1a were used to describe the drying characteristics of the sample (Nema et al. 2013; Inyang et al. 2018). The best fit model to represent the pertinent drying kinetics of the sample was identified based on the highest R² value and lowest residual sum of squares (RSS) and chi square values. Alternate drying model expressions have been considered to identify the best fit model to represent measured drying characteristics. It is well known that the drying kinetics curve is a complex function of various factors such as drying temperature, drying time, sample thickness, drying method, humidity, initial moisture of sample and material characteristics. Therefore, these complex interactions demand the due consideration of several alternate drying models to represent measured drying kinetics data. Thereby, the pertinent mass transport behaviour of the system can be understood from the empirical analysis. Three types of drying models exist to represent measured drying kinetics data. Among these, semi-theoretical (Newton’s law, Page, Two term, Exponential, Henderson and Pebis model) and empirical (Wang and Singh, Singh et al., Verma and Diffusion model) models are widely used to represent drying characteristics of food materials. These models consider external transport resistances between surrounding environment and material for moisture transport. Further, for the evaluation of pertinent drying behaviour, these modelling approaches mostly rely much upon the measured experimental data. This is not the case for several theoretical models that regard internal resistance to moisture transport as a key assumption. The validity of such assumption is very weak, given the hypothesis that external resistance for moisture transport is significantly high in comparison with the internal moisture transport resistance (Onwude et al. 2016).

Table 1.

a Fitness parameters of alternate drying models for the representation of drying kinetics data of tray dried Kolphul samples. b A comparative summary of moisture diffisuvities of Kolphul samples

(a)
Model	Model equation	Parameters	40 °C	50 °C	60 °C	70 °C	80 °C
Newton	$MR=exp(-kt)$	k	0.00279	0.00595	0.0078	0.0128	0.01671
		R2	0.9519	0.97302	0.96943	0.98169	0.98974
		RSS	0.30984	0.11421	0.10793	0.04402	0.01718
		χ2	0.00492	0.00254	0.00284	0.00157	9.54 × 10−4
Page	$MR=exp(-k{t}^n)$	k	1.55 × 10−4	6.59 × 10−4	9.21 × 10−4	0.00288	0.0079
		n	1.47907	1.41483	1.42098	1.32531	1.17218
		R2	0.99571	0.99879	0.9956	0.99702	0.9952
		RSS	0.02718	0.005	0.01513	0.0069	0.00758
		χ2	4.38 × 10−4	1.14 × 10−4	4.09 × 10−4	2.55 × 10−4	4.46 × 10−4
Henderson and Pabis	$MR=aexp(-kt)$	a	1.12321	1.10945	1.08957	1.06456	1.02856
		k	0.00311	0.00653	0.0084	0.01353	0.01714
		R2	0.96653	0.98227	0.97518	0.98453	0.98998
		RSS	0.21215	0.07336	0.08531	0.03586	0.01583
		χ2	0.00342	0.00167	0.00231	0.00133	9.31 × 10−4
Logarithmic	$MR=aexp(-kt)+c$	a	1.33934	1.13475	1.11971	1.08122	1.05453
		c	− 0.28889	− 0.05022	− 0.05599	− 0.03039	− 0.04365
		k	0.00183	0.0057	0.00722	0.01239	0.01511
		R2	0.99354	0.98755	0.98324	0.98809	0.99487
		RSS	0.04032	0.05034	0.05606	0.02659	0.00763
		χ2	6.61 × 10−4	0.00117	0.00156	0.00102	4.77 × 10−4
Wang and Singh	$MR=M_0+at+b{t}^2$	M0	1.03697	0.98717	0.96201	0.89567	0.91872
		a	− 0.00215	− 0.00397	− 0.00482	− 0.00651	− 0.0094
		b	1.10 × 10−6	3.88 × 10−6	5.74 × 10−6	1.10 × 10−5	2.31 × 10−5
		R2	0.99783	0.98527	0.98423	0.95896	0.9848
		RSS	0.01351	0.05957	0.05273	0.09159	0.0226
		χ2	2.21 × 10−4	0.00139	0.00146	0.00352	0.00141
Singh et al. (2014)	$MR=exp(-kt)-akt$	k	0.00195	0.00544	0.00717	0.01229	0.01006
		a	0.12065	0.01675	0.01507	0.00646	0.01572
		R2	0.98988	0.98038	0.97783	0.98461	0.99401
		RSS	0.06414	0.08121	0.0762	0.03567	0.00947
		χ2	0.00103	0.00185	0.00206	0.00132	5.57 × 10−4
da Silva et al. (2011)	$MR=exp(-at+b{t}^{1/2})$	a	0.00425	0.00886	0.01128	0.01773	0.02026
		b	− 0.0284	− 0.03966	− 0.04231	− 0.04662	− 0.02967
		R2	0.98188	0.99332	0.98631	0.99215	0.99289
		RSS	0.11484	0.02764	0.04706	0.0182	0.01124
		χ2	0.00185	6.28 × 10−4	0.00127	6.74 × 10−4	6.61 × 10−4

(b)
Sample	Drying method	Temperature (°C)	Diffusivity (m2/s)	Activation energy (kJ/mol)	Reference
Kolphul (Musa balbisiana Colla)	Intermittent airflow assisted tray drying	40	9.635 × 10−12		This work
		50	1.555 × 10−11		Do
		60	2.519 × 10−11	37.38	Do
		70	3.347 × 10−11		Do
		80	5.004 × 10−11		Do
Buttermilk soaked chopped banana blossom	Tray drying	40–60	5.45–8.09 × 10−9	50.06	John et al. (2014)

Determination of moisture diffusivity and activation energy

The moisture transport phenomenon during drying can be represented using Fick’s second of diffusion:

$$\frac{\delta M}{\delta t}=D_{\mathit{diff}}\frac{{\delta }^{2}M}{\delta x^2}$$ 2

where M, D_diff, and x correspond to the moisture content, diffusivity and thickness of sample respectively.

The analytical solution of the Fick’s second law of diffusion can be obtained by presuming the validity of assumptions such as diffusion mode of transport, constant temperature and negligible shrinkage during drying process. Mathematically, these assumptions translate into the following boundary conditions: M = M₀, 0 ≤ X < L at t = 0, $\frac{\partial M}{\partial X}=0$ at X =0 and M = 0 at X = L. The solution to the above equation given by Crank equation (Eq. 3) was used to determine the moisture diffusivity.

$$MR=\frac{8}{{\pi }^2}exp\left[-\frac{{\pi }^{2}Dt}{4L^2}\right]$$ 3

where D and t denote the effective moisture diffusivity and drying time respectively.

Thus, a graph of lnMR versus t shall provide a straight line plot with slope and intercept as $-\frac{{\pi }^{2}D}{4L^2}$ and $ln\frac{8}{{\pi }^2}$ respectively. Thus, moisture diffusivity can be determined with substitution of value of slope.

The activation energy corresponding to moisture diffusivity can be determined from the Arrhenius equation expressed as:

$$D=D_{0}exp\left(-\frac{E_a}{\mathit{RT}}\right)$$ 4

where E_a, R and T refer to the activation energy, gas constant and temperature respectively.

Experimental design and statistical analysis

The experimental design adopted to understand the effect of drying on the responses (moisture content, vitamin C and antioxidant activity) was based on the rotatable central composite design of RSM. Based on the preliminary investigation, the range of drying temperature (A) and time (B) were chosen as 50–70 °C and 390–720 min respectively. Table 2a summarizes all the combinations of A and B according to rotatable CCD. The CCD based RSM design provides 14 combinations of experiments with 6 sets at the central point. The following polynomial equation was used to define the variation of response variables as a function of temperature (A) and time (B)

$$Y=A_0+C_{1}A+C_{2}B+C_{3}AB+C_4{A}^2+C_5{B}^2+\epsilon$$ 5

where Y is the response, A and B are drying temperature and time respectively in coded forms, C₁ and C₂ are the coefficients of linear terms, C₃ is the coefficient of interaction term, C₄ and C₅ denote coefficient of quadratic terms and $\epsilon$ represents the error.

Table 2.

a Central composite design based experimental data of tray dried Kolphul samples. b Analysis of variance (ANOVA) data of moisture, vitamin C and antioxidant activity responses

(a)
Run	Temp (°C)	Time (Min)	Moisture (%)	Vit C (mg/100 g)	Antioxidant activity (%)
1	45.86	555	36.36 ± 1.07	43.52 ± 0.82	56.47 ± 0.68
2	60	555	8.22 ± 0.80	86.24 ± 1.59	87.47 ± 1.56
3	60	555	7.68 ± 0.80	87.61 ± 1.59	86.24 ± 1.56
4	60	555	8.62 ± 0.80	86.26 ± 1.59	88.12 ± 1.56
5	60	788.35	5.11 ± 0.37	78.92 ± 0.96	76.55 ± 3.11
6	70	390	5.95 ± 0.24	76.34 ± 0.82	80.67 ± 1.84
7	60	555	7.11 ± 0.80	84.41 ± 1.59	88.38 ± 1.56
8	50	390	46.71 ± 2.15	32.54 ± 1.86	43.49 ± 0.63
9	60	321.65	25.47 ± 0.71	58.7 ± 1.33	60.02 ± 2.04
10	60	555	6.54 ± 0.80	88.74 ± 1.59	90.45 ± 1.56
11	60	555	8.36 ± 0.80	85.08 ± 1.59	89.92 ± 1.56
12	70	720	4.48 ± 0.08	59.44 ± 2.18	64.77 ± 2.38
13	74.14	555	3.27 ± 0.14	67.61 ± 1.68	72.51 ± 0.74
14	50	720	13.49 ± 0.40	81.07 ± 2.46	85.52 ± 1.81

(b)
	Moisture					Vitamin C					Antioxidant activity
Source	Sum of Squares	df	Mean Square	F Value	p value Prob > F	Sum of Squares	df	Mean Square	F Value	p value Prob > F	Sum of Squares	df	Mean Square	F Value	p value Prob > F
Model	2274.54	5	454.91	418.21	< 0.0001	4089.50	5	817.90	207.16	< 0.0001	2852.92	5	570.58	163.88	< 0.0001	Significant
A-Temp	1165.63	1	1165.63	1071.6	< 0.0001	395.34	1	395.34	100.13	< 0.0001	191.24	1	191.24	54.92	< 0.0001
B-Time	503.77	1	503.77	463.13	< 0.0001	453.39	1	453.39	114.83	< 0.0001	306.37	1	306.37	87.99	< 0.0001
AB	252.02	1	252.02	231.69	< 0.0001	1070.27	1	1070.27	271.08	< 0.0001	838.97	1	838.97	240.96	< 0.0001
A2	270.85	1	270.85	249.00	< 0.0001	1745.09	1	1745.09	442.00	< 0.0001	962.02	1	962.021	276.30	< 0.0001
B2	106.28	1	106.28	97.71	< 0.0001	565.38	1	565.38	143.20	< 0.0001	668.74	1	668.74	192.07	< 0.0001
Residual	8.70	8	1.09			31.59	8	3.95			27.85	8	3.48
Lack of Fit	5.47	3	1.82	2.83	0.1462	18.90	3	6.30	2.48	0.1756	15.74	3	5.25	2.16	0.2106	Not significant
Pure Error	3.23	5	0.65			12.69	5	2.54			12.12	5	2.42
Cor Total	2283.24	13				4121.09	13				2880.77	13

Statistical analysis was conducted with Design Expert 7.0 software. The 3-D response behaviour was plotted to gain insights with respect to response variations with temperature and time. Each response was modelled as a function of independent variables (A and B) using multiple regression analysis. Significant terms with their coefficients were obtained from the ANOVA table. F and p values show the significance of the terms. Low F value for lack of fit infers upon the fitness of the model. Adequacy of the model was given by R², adjusted R² and predicted R² values. Adequate precision > 4 is desirable for the model.

Optimization of process variables

Drying process parameters were optimized for the simultaneous minimization of moisture content and maximization of both antioxidant activity and vitamin C content. Experimental optimum values were compared with the RSM optimized values. RSM based optimization was carried out by numerical optimization of the data set. Further, the final optimum sets of parameters were confirmed experimentally.

Validation of optimum values

The optimum values obtained from the previous analysis were also verified by conducting experiments at the optimum independent variable values. Runs in triplicate were carried to compare experimental and predicted values. Standard errors were calculated for each response to evaluate the relative variation between two sets of values and thereby infer upon the validity of the optimized values obtained from the RSM study.

Analysis

Moisture content

The procedure followed to determine the moisture content of the sample involved oven drying of a known amount of sample at 105 °C overnight (12 h) (AOAC 2010; Mondal et al. 2019) and measuring its dry weight after cooling it in a desiccator. Based on the initial and final weight of the sample, the following expression was used to determine the moisture content (MC) per 100 g of the sample.

$$MC(\%)=\frac{W_i-W_f}{W_i}\times 100$$ 6

where W_i and W_f refer to initial weight and weight after drying in the oven respectively.

Vitamin C

2, 6-dichlorophenol indophenol (DCPIP) dye titration method was adopted to determine the vitamin C content of the sample (Sadasivam and Manickam 1992; Anjali et al. 2012; Ravula et al. 2017; Mondal et al. 2019). Firstly, the method involved homogenization of 100 mg of sample with 20 mL 4% oxalic acid in a mortar-pestle for extraction. Thereafter, 5 mL of filtered sample was mixed with 10 mL of 4% oxalic acid for titration with the DCPIP dye. Similarly, 100 ppm standard ascorbic acid was prepared for the blank run. The dye was prepared by mixing 42 mg sodium bicarbonate and 52 mg dye powder in 200 mL distilled water. The titration was carried out for both blank and extract samples with the dye solution. The end point of titration was confirmed by observing a momentary pink colour in the solution. Thereby, the vitamin C content of the sample was evaluated using the expression:

$$VitaminC(\mathit{mg}/100g)=\frac{\text{.5}\times V_2\times 20\times 100}{V_1\times 5\times W_i}$$ 7

where V₁, V₂ and W_i correspond to dye solution consumed for blank and sample and sample weight respectively.

Antioxidant activity

Percentage DPPH scavenging capacity of the samples as summarized in the literature (Barimah et al. 2017; Mondal et al. 2019; Sana et al. 2014; Sochor et al. 2010) was followed to determine the antioxidant activity of the sample. Firstly, the procedure involved homogenization of 10 mg sample with absolute methanol. After adjusting the volume of the homogenate to a concentration of 500 µg/mL, the sample extract was kept in a sonication bath (Elmasonic S 30 H, Elma) for the 30 min for extraction. The extract was filtered and the filtrate was taken for further analysis. 1 mL of filtrate was pipetted in test tube and mixed with 3 mL 0.002% methanolic DPPH solution. The control sample was prepared by mixing 3 mL of 0.002% methanolic DPPH in 1 mL absolute methanol. Both control and analysed sample were then kept in a dark environment for 30 min. Finally, the absorbance of both samples were measured with an absolute methanol as the blank using UV visible spectrophotometer (UV-2600, Shimadzu, Singapore) at 517 nm. Using these measured values, antioxidant activity (AA) of the analysed sample was evaluated using the expression:

$$AA=\frac{A_c-A_s}{A_c}\times 100$$ 8

where A_c and A_s refer to the absorbance of control and sample respectively.

Proximate analysis

The ash and fat content of the sample was obtained by AOAC (2010) method (AOAC 2010; Mondal et al. 2019). The carbohydrate content was evaluated using Clegg anthrone method elaborated in the literature (Okonwu and Enyinnaya 2016; Mondal et al. 2019). Soluble protein was evaluated using the Bradford method (Bradford 1976; Mondal et al. 2019). The hot sulfuric acid-sodium hydroxide treatment based method was followed to determine the crude fibre content of the sample (Sadasivamand Manickam 1992; Mondal et al. 2019).

Results and discussion

Drying characteristics curve

Characteristics curves for intermittent tray drying of Kolphul exhibited similar trends as those pertinent for most food materials. Figure 2a depicts the variations of moisture ratio (MR) with drying time at 40, 50, 60, 70 and 80 °C. For all cases, a rapid lowering of moisture ratios was prevalent initially followed by a slow reduction to eventually reach equilibrium at the end. Moisture ratios rapidly reduced from 1 to 0.168, 0.116, 0.117, 0.128 and 0.171 for 40, 50, 60, 70 and 80 °C for a corresponding drying time of 555, 300, 225, 135 and 105 min respectively. This is due to greater moisture removal from the product with enhanced moisture diffusivity at a higher temperature. Also, this effect was dominant at higher temperatures. The reduction in moisture removal at later drying time condition was due to a reduction in moisture transport from inside to the outer surface of the sample. Such trends were pertinent from 555 to 795, 300–570, 225–375, 135–255 and 105–165 min of drying time for the drying temperatures of 40, 50, 60, 70 and 80 °C respectively. Further, due to higher moisture diffusivity, equilibrium condition was achieved earlier at higher drying temperatures.

Fig. 2.

a Influence of time and temperature on the moisture content of tray dried Kolphul samples. b ln MR versus time plot for tray dried kolphul samples dried at various temperatures. c Graphical representation of moisture diffusivity data with Arhenius expression

Fitness of drying models

Table 1a summarizes the results associated with the fitness of alternate drying models to represent the drying characteristic curves. Among all models, the Page model was the best fit model with high R², low RSS and reduced chi squared values for all cases. However for 40 °C case, Wang and Singh model was also a good fit. The associated model parameters for different models are presented in Table 1a. Among alternate models except at 40 °C, Wang and Singh model possessed poor fitness indices (relatively lower R², higher RSS and reduced chi squared values). There was a positive effect of temperature on the drying rate constant. Compared to time, the temperature effect was significant for moisture removal from the product and hence drying rate constant enhanced with temperature.

Moisture diffusivity and activation energy

Figure 2b illustrates the plot of ln MR versus drying time. Table 1b summarizes the moisture diffusivity values evaluated from the slope of the straight line plots at various temperatures. The moisture diffusivities varied from 9.64 × 10⁻¹² to 5 × 10⁻¹¹m²/s for a variation in temperature from 40 to 80 °C. This is in agreement with the corresponding value range (10⁻⁸–10⁻¹² m²/s) for food materials (Zogzas et al. 1996). Since higher temperature facilitates better moisture removal, higher moisture diffusivity is obtained at a higher temperature. The Arrhenius plot for these diffusivity values is presented in Fig. 2c using which the activation energy of the drying process for moisture removal was determined to be 37.38 kJ/mol. This is in agreement with the value range (12–110 kJ/mol) for various food materials (Mwithiga and Olwal 2005). A similar value of 50.06 kJ/mol was obtained for chopped and buttermilk soaked Kolphul subjected to hot air drying (John et al. 2014).

Vitamin C characteristics

Figure 3a presents the time-dependent vitamin C characteristic curve at various temperatures. The drying time significantly influenced the vitamin C content of the sample. At 50, 60 and 70 °C drying temperatures, the vitamin C content raised from 4.53 to 82.23, 4.53 to 85.24 and 4.53 to 82.26 mg/100 g for a corresponding variation in drying time from 0 to 12, 0 to 10 h and 0 to 6 h respectively. This is due to accelerated moisture removal that enhances dry matter in the sample. At drying temperatures of 60 and 70 °C, the variable values followed an opposite trend from 10 to 12 and 6 to 12 h of drying time to reach the lower value of 81.46 and 60.25 mg/100 g respectively. This is due to the long drying time that deteriorated vitamin C content. Vitamin C constituents in the vegetable sample are well known to be heat sensitive. Hence, with prolonged exposure to heat, they are very likely to get degraded (Santos and Silva 2008). This is not the case for shorter exposure to heat. For the case, the detrimental effect of heat is not predominant in comparison with the enhanced dry matter content of the sample achieved due to effective removal of the moisture. The reduction in vitamin C is possibly due to the oxidation of active compounds to dehydro-ascorbic acid which hydrolyzed to 2, 3-diketogolunic acid and further oxidation and polymerization of these components to various inactive compounds (Thankitsunthorn et al. 2009).

Fig. 3.

a Time dependent variation of Vitamin C for tray dried Kolphul samples dreid at various temperatures. b Time dependent variation of Antioxidant activity for tray dried Kolphul samples dreid at various temperatures

Antioxidant activity characteristics

Antioxidant activity of the sample followed a mixed pattern with drying temperature variation. Initially, the antioxidant activity enhanced followed by a reducing trend (Fig. 3b). At 60 and 70 °C, the antioxidant activity varied from 2.27 to 86.82% and 2.27 to 85.42% for drying time variation from 0 to 10 h and 0 to 6 h respectively. Thereafter, the variable reduced from 86.82 to 78.28% and 85.42 to 65.36% for drying time variation from 10 to 12 h and 6 to 12 h respectively. At 50 °C, antioxidant activity increased from 2.27 to 84.79% for drying time variation from 0 to 12 h. At higher drying temperature, the antioxidant activity trend followed a reduced profile than those prevalent at a lower temperature. This is due to the heat sensitivity of antioxidant constituents for prolong exposure to heat. At lower temperature, long drying time was required to overcome the enhanced trend of the variable.

Drying kinetics based optimality and literature comparison

Based on trial and error (The trial and error based evaluation involved manual variation of drying temperature and time. This was achieved through the fixed choice of the intervals associated to these independent variables. For the chosen combinations of drying temperature and time, responses have been evaluated for their optimality. Accordingly, an optimal set of independent and dependent variables has been identified) based evaluation, the maximum vitamin C and antioxidant were achieved at 60 °C and 10 h of drying. The responses were obtained as 85.24 mg/100 g vitamin C and 86.82% antioxidant activity. These values are comparable to our recent work on tray dried banana blossom dried at 60 and 10 h of drying time (Mondal et al. 2019). However, corresponding moisture content was relatively high (6.96%). On the other hand, for minimum moisture content (3.27% at 74.14 °C and 555 min), other responses such as vitamin C and antioxidant activity values were low (67.61 mg/100 g and 72.51% respectively). Therefore, statistical optimization approach was applied to identify a better combination of process variables.

Experimental datasheet

The experimental data obtained at various combinations of drying temperature and time with CCD based RSM is presented in Table 2a. The response values were represented as the mean of triplicate run with standard deviation that affirms the reproducibility of the data.

Effect of drying temperature on the responses

Moisture content

As depicted from Table 2a, the drying temperature had a negative effect on the moisture content of the sample. At 390 min, the moisture content reduced from 46.71 to 5.95% for an increase in drying temperature from 50 to 70 °C. This further varied from 13.49 to 4.48% at higher drying time of 720 min and similar temperature increment. However, at the central point of drying time, the moisture content followed decreasing order from 36.36 to 3.27% for a temperature variation of 45.86 to 74.14 °C.

Vitamin C

The vitamin C content of the sample was found to enhance with drying temperature. At 390 min drying time, it varied from 32.54 to 76.34 mg/100 g for a drying temperature variation from 50 to 70 °C. At 720 min drying time, it correspondingly varied from 81.07 to 59.44 mg/100 g. However, at the central drying time (330 min), the variable increased from 43.52 to 86.39 mg/100 g for a variation in temperature from 45.86 to 60 °C and reduced to 67.61 mg/100 g at 74.14 °C.

Antioxidant activity

The antioxidant activity trend was similar to the vitamin C content trend for the sample. At 390 min drying time, the antioxidant activity varied from 43.49 to 80.67% for a temperature variation of 50–70 °C. However, at higher drying time of 720 min, the corresponding value varied from 85.52 to 64.77%. At the central drying time, the trend was increasing from 45.86 to 60 °C (56.47–88.43%), followed with a reduced trend to reach 72.51% at 74.14 °C.

Effect of drying time on the responses

Moisture content

The drying time was found to significantly influence the moisture content variation in the samples with a negative effect. For a drying time variation from 390 to 720 min, the moisture content varied from 46.71 to 13.49% and 5.95 to 4.48% at 50 and 70 °C respectively. Similarly, the trend was decreasing (25.47–5.11%) for a drying time variation from 321.65 to 788.35 min at 60 °C.

Vitamin C

The vitamin C content of the sample followed an increasing trend with increasing drying time. For the time range of 390–720 min, vitamin C increased from 32.54 to 81.07 mg/100 g at 50 °C. However, it correspondingly reduced from 76.34 to 59.44 mg/100 g at 70 °C. On the other hand, for a variation of drying time from 321.65 to 555 min at 60 °C, vitamin C content of the sample enhanced from 58.7 to 86.39 mg/100 g and thereafter reduced to 78.92 mg/100 g at higher drying time of 788.35 min.

Antioxidant activity

The antioxidant activity variation was similar to the vitamin C content trend. At 50 °C, the antioxidant activity value varied from 43.49 to 85.52% for drying time variation from 390 to 720 min. However, the corresponding trend was contrary at 70 °C with the variable reducing from 80.67 to 64.77%. This was due to the detrimental effect of higher temperature and time on antioxidant constituents. At 60 °C, the antioxidant activity enhanced from 60.02 to 88.43% and thereafter reduced to 76.55% for drying time range of 321.65–555–788.35 min.

Optimization based on RSM experimental data and literature comparison

Based on the RSM based experimental investigation, the optimum responses were 7.76% moisture content, 86.39 mg/100 g vitamin C and 88.43% antioxidant activity for 60 °C for 555 min. The data in this case are in agreement with our recent study on tray dried Kolphul at constant drying temperature (60 °C) and time (10 h) (Mondal et al. 2019).

Model fitness and analysis of variance

Table 2b summarizes the ANOVA data respectively for all responses. Among various alternate models, the quadratic model was the best fitmodel for all the responses.

The significant terms in the quadratic model for moisture content were A, B, AB, A² and B². All these possessed high F values and p < 0.0001. The lack of fit for the case was insignificant with low F value (2.83) and high p value (0.1462). R², adj R² and pred R² values for the moisture response model were 0.9962, 0.9938 and 0.9809 respectively. All these adequately confirm good fitness of the quadratic model. The model expression to represent the moisture content variation of the sample with drying time and temperature is as follows:

$$Y_1=7.75-12.07A-7.94B+7.94AB+6.06A^2+3.79B^2$$ 9

The best fit quadratic model for the vitamin C response variable affirms A, B, AB, A² and B² as the significant terms with high F values and p < 0.0001. With F = 2.48 and p = 0.1756, the lack of fit was noted to be not significant. The R² value for vitamin C response was 0.9923. Adj R² (0.9875) were in good agreement with pred R² (0.9630). The PRESS and coefficient of variance (C.V.) values were 152.66 and 2.74 which are in the desirable range for the model. The following model equation represents the RSM based quantification of vitamin C response:

$$Y_2=86.39+7.03A+7.53B-16.36AB-15.37A^2-8.75B^2$$ 10

Similarly, significant terms for the quadratic model of antioxidant activity are A, B, AB, A² and B² with high F values and p < 0.0001. The response model refers to insignificant lack of fit (F = 2.16 and p = 0.2106). R², adj R² and pred R² were 0.9903, 0.9843 and 0.9551 respectively. Adequate precision value (38.06) > 4 is desirable for the model. All these confirm good fitness of the quadratic model expressed as:

$$Y_3=88.43+4.89A+6.19B-14.48AB-11.41A^2-9.52B^2$$ 11

Response surface analysis

All response variables were found to be significantly influenced with variation in drying temperature and time, their square terms and interaction terms (p < 0.0001). Figure 4a–c depict the response plot of moisture content, vitamin C and antioxidant activity respectively for the sample. Both temperature and time had a negative non-linear influence on moisture content. The moisture content varied negatively with drying time and temperature. This is due to higher combinations of moisture diffusivity and heat transfer at higher combinations of drying time and temperature. The square of temperature and time and interaction of these as well had a significant effect (p < 0.0001) on the moisture content. The effect was positive for square and interaction terms. These response surfaces are in agreement with hot air drying of mango seed kernel (Ekorong et al. 2015) and torch ginger (Juhari et al. 2012).

Fig. 4.

3-D response surface plots of a Moisture b Vitamin C and c Antioxidant activity

On the contrary, vitamin C variation is positively influenced by temperature and drying time. However, this was up to a certain limit, after which it decreased. The vitamin C content enhancement was attributed to enhanced dry matter content per gram of sample and reduced trend thereafter due to the detrimental effect of higher temperature and drying time. The maximum vitamin C content (86.39 mg/100 g) was found at 60 °C and 555 min. Corresponding minimum values (32.54 mg/100 g) was obtained at 50 °C for 390 min. The effect of quadratic and interaction terms was significant and negative (p < 0.0001).

Similarly, antioxidant activity increased with temperature and time up to a certain range and thereafter reduced with a further increment of the variables. The antioxidant activity varied 43.49–88.43% for parametric combinations of 50 °C, 390 min and 60 °C, 555 min respectively. Both square and interaction terms had a negative influence on the dependent variable (p < 0.0001).

RSM optimization and validation of process parameters

Optimization of process variables was carried out using numerical optimization tool of Design Expert 7.0 software. The optimization was based on simultaneous maximization of vitamin C and antioxidant activity and minimization of moisture content of the sample. Such studies indicated the optimal process variables to be 61.82 °C, 611.52 min, moisture content (3.98%), vitamin C (87.69 mg/100 g) and antioxidant activity (89.04%). The standard errors between the RSM optimized responses and the experimental values at RSM optimum temperature and time were used to validate the optimised parameters generated by RSM and hence the model developed. With standard error values of 0.2–0.8, the values obtained from experiments are close to those obtained from RSM optimization. This affirmed the adequacy and validation of the values and RSM based optimality of tray drying process parameters associated with the chosen non-leafy vegetable.

Optimality of process variables based on drying kinetics and RSM

Table 3a summarizes the comparison of alternate optimum values based on kinetics, RSM non-optimized, RSM optimum and literature data. The kinetics, RSM experimental and RSM optimized based optimum values of process variables corresponds to (60 °C, 600 min, 6.96% (moisture) 85.24 mg/100 g (vitamin C) and 86.82% (antioxidant activity)), (60 °C, 555 min, 7.76% (moisture) 86.39 mg/100 g (vitamin C) and 88.43% (antioxidant activity)) and (61.82 °C, 611.52 min, 3.98% (moisture) 87.69 mg/100 g (vitamin C) and 89.04% (antioxidant activity)) respectively. Similar results were also obtained for our recent investigation of banana blossom using tray drying (Mondal et al. 2019). The corresponding values at constant temperature (60 °C) and time (10 h) refers to 7.34% moisture content, 84.51 mg/100 g vitamin C and 86.38% antioxidant activity. Among various sets of optimum values, RSM based optimum process parameters were the best. For chopped and buttermilk soaked banana blossom, moisture content of 9% was reported at 60 °C and 195 min of drying in tray dryer (John et al. 2014).

Table 3.

a Experimental, predicted and literature values of the best sets of independent and response variables. b Comparative summary of proximate characteristics of tray dried Kolphul samples and literature

(a)
Sample		Drying method	Optimized temp (°C)	Optimized time (Min)	Moisture (%)	Vit C (mg/100 g)	Antioxidant activity (%)	References
Kolphul	RSM optimized values	Tray drying with intermittent airflow	61.82	611.52	3.98	87.69	89.04	This work
	RSM non- optimized values	Do	60	555	7.76	86.39	88.43	Do
	Kinetics best	Do	60	600	6.96	85.24	86.82	Do
	Experimental at RSM optimized conditions	Do	61.82	611.52	4.72 ± 0.12	87.28 ± 2.26	90.16 ± 1.74	Do
Kolphul	Literature	Do	60	600	7.34	84.51	86.38	Mondal et al. (2019)

(b)
Sample	Drying methods	Optimal drying conditions	Optimal drying characteristics						References
			Moisture (%)	Carbohydrate (%)	Soluble protein (%)	Fat (%)	Fibre (%)	Ash (%)
Kolphul	Tray drying	61.82 °C for 611.52 min	3.98	50.94	8.83	9.04	13.26	11.82	This work
Kolphul	do	60 °C for 600 min	7.34	50.52	8.26	8.57	12.35	10.84	Mondal et al. (2019)

Proximate analysis at optimum drying condition

Among evaluated parameters, carbohydrate content (50.94%) was found to be the maximum while the lowest value was obtained for soluble protein (8.83%). Corresponding values for crude fibre, ash and fat were 13.26, 11.82 and 9.04% respectively. Literature comparison of proximate parameters of Kolphulis presented in Table 3b. With a good proportion of carbohydrate, ash, crude fibre and fat, it can serve as an ingredient to develop ready to sip and eat formulations.

Conclusion

Critical insight into the study conducted leads to the following important inferences. Firstly, Kinetic and RSM based investigation provide a better understanding of the complexity of the process, process variables interaction and coherence between them. It is useful to identify the compatible vegetables for product mix formulation. Secondly, among kinetics, RSM non optimized and RSM optimized values, the later provides the best result. With high antioxidant activity and vitamin C content, Kolphul may serve as a major ingredient in antioxidant and vitamin C enriched food formulations. Lastly, Kolphul is found to be a rich source of carbohydrate, crude fibre, ash and fat content. With all important characterization, this study provides an important guideline to formulate ready to sip or eat food products and promote food processing sectors in NE India with indigenous vegetable resources. Future experimental investigations need to consider sample colour as an important response variable to augment upon its optimality. Along with this key response variable, other physical and chemical characteristics of the dried banana blossom could be also considered in near future.

Compliance with ethical standards

Conflict of interest

The author declares that they have no conflict of interest.