Goal programming approaches for coordinating flock collection in the poultry industry

We present deterministic and stochastic goal programming models for coordinating of flock collection activities in a poultry processing company. The aim is to develop weekly schedules that balance three goals: ensuring a steady workload in the processing facility, reducing processing defects due to weight differences, and fulfilling production targets of farmers, while satisfying logistical constraints. Two deterministic goal programming models are proposed: a weighted model that considers the collective interests of farmers, and a min-max model that prevents large deviations from the production target for any individual farmer. Furthermore, two-stage stochastic programming models are developed, in which forecasts of the average flock weights are uncertain. The proposed approaches are applied to a real case study in Nova Scotia (Canada). Numerical results show that, compared to the weighted models, the min-max models considerably reduce the maximum expected deviation from optimality without significantly increasing the gross deviation from production targets. Furthermore, the stochastic models led to substantial improvements over the deterministic ones, thus justifying the transition to a two-stage planning procedure. The proposed stochastic min-max model was also shown to outperform the current manual approach in terms of reducing the average weight spread between flocks collected on the same day.