5. Simulation results and discussion
5.1 Experimental plan
For the performance evaluation of GA and FGA approaches, a scenario of ten farms is
considered. The minimal acceptance quality of products is set to nine out of ten[3].
Changes in the quality of different products are assumed to be between the quality
change of baseline products – sweet pepper and cucumber which are selected because
the recommended temperature range from 7.2 to 10°C of these two products are the
same (see Labuza, 1982 for details); therefore the temperature of truck can be set to
meet the constraint (2). The length of storage period of sweet pepper and cucumber are
2-3 weeks and 10-14 days, respectively. As mentioned above, the storage period is
semi-log related to absolute temperature (°F), and the quality change is reciprocal to the
storage period. We assume the quality is between 0 and 10 and the temperature setting
of the truck could be adjusted to a step of 0.1 with precision[4]. The distances between
farms and between farms and retailer are generated from a uniform distribution with
parameters (1, 50) and the initial crossover rate and initial mutation rate are assumed to
be 0.6 and 0.05, respectively. In order to focus on the change of food quality, we assume
constraint (4) is fulfilled and one unit of the product is picked up from each farm, other
parameters are set as follows: