A mathematical model for rose production planning with variable cycles and growing degree-day curves: A case study in Ecuador

The floriculture sector is one of the main export industries in Ecuador, with roses as its leading product. Despite its economic importance, production planning in commercial rose farms is still largely based on empirical practices. In particular, defining cutting schedules is important because these decisions must simultaneously satisfy biological, technical, and market constraints. The interaction of these factors often generates significant overproduction (“dumps”) and unmet demand (backorders), directly affecting profitability. Although mathematical optimization has strong potential in this context, its application to flower and rose production has been scarcely explored. This paper introduces a new mixed-integer linear programming (MILP) model to optimize weekly cutting schedules over a 52-week planning horizon. Scientifically, the work contributes to a novel class of production–harvesting problems in which harvesting decisions endogenously determine future production by activating new growth cycles. This differs from traditional agricultural planning models, where production and harvesting are treated as sequential and weakly connected processes. The proposed model incorporates main characteristics of rose production, including growing degree days (GDD) curves, stem activation dynamics, cycle start and termination decisions, capacity constraints, minimum sales commitments, and perishability restrictions. The objective is to maximize annual profit. The model was validated using real operational data from a commercial flower farm in Ecuador. Results show that the optimized schedules substantially improve profitability while reducing both overproduction and backorders. Computational experiments on a large-scale instance with >385,000 decision variables and 100,000 constraints demonstrate the applicability of the approach to real-world production environments.