*2.3. Experimental Parameters*

To reduce the number and the scope of experiments and to discover the relationships between the factors a ffecting the drying process, the statistical Design of Experiments (DOE) method was applied [60]. The most significant factors influencing the drying process identified as controllable input factors were the drying time, solar radiation intensity, and slice thickness. The e ffect of air flow rate was not investigated in this preliminary feasibility study, but was selected according to the available axial fan performance and cost, fan characteristic curve, and the total pressure drop in the solar drying system (42.1 m<sup>3</sup>/h, corresponds to roughly half the maximum fan throughput). Other controllable input factors such as mean wind velocity of 1.8 m/s, the angle of incidence of solar radiation of 90◦, and the diameter of pineapple slices were always the same as well. Uncontrollable input factors were temperature and relative humidity of the ambient air and the moisture content in the fresh pineapple. The moisture content in the dried pineapple was the output used for validating the solar drying process. The controllable input factors were varied in this experimental design at two levels, low (−) and high (+), as shown in Table 1.

**Table 1.** Two-level factorial design of controllable input factors and their associated levels.


As for the drying time, this was varied with respect to the minimum and maximum usable daily sunshine duration in Togo in the months of June to October and November to May (the primary pineapple processing periods in this country). The selected solar radiation intensities corresponded to the respective prevailing minimum and maximum values. All the relevant climate data were taken from the software Meteonorm 7 (Meteotest AG, Bern, Switzerland) [61]. A full factorial design with *k* factors attaining two levels was chosen [62]. The number of experiments was therefore given by *n* = <sup>2</sup>*k*, that is, for *k* = 3 the experimental design comprised eight tests as shown in Table 2. This experimental design was carried out in a random order twice to determine the influence of input factors on the output more accurately and to mitigate the e ffect of scattering.

**Table 2.** Setting of input factors according to the full factorial design.


The moisture content in the dried pineapple was not influenced by only the individual input factors, A, B, and C. Therefore, the e ffects of all possible interactions of two factors, and of all three factors mentioned above, also had to be investigated using analysis of variance. For an experiment design comprising 2*k* experiments, 2*k* − 1 e ffects could be identified with positive or negative signs as

presented in Table 3. The signs of individual input factors were equal to their levels from Table 2. In each row, the signs of interactions, AB, AC, BC, and ABC, correspond to the products of signs of the respective input factors. The effects can then be estimated as follows:

$$\text{effect} = \frac{2}{n} \sum\_{i=1}^{n} (\text{sign} \cdot \overline{y}\_i)\_\prime \tag{1}$$

where *i* denotes the test number and *yi*is the mean of the respective experimental results.


**Table 3.** Design matrix and signs for seven effects in the 23 full factorial design.
