Localisation and Circularity in Apple Supply Chains: An Algorithmic Exploration

Localisation and circularity in perishable food supply chains are essential for sustainability. Poor allocation of time-sensitive food leads to waste, higher transport emissions, and unnecessary long-distance sourcing. Algorithms used in digital trading platforms and allocation systems can help address these problems by improving how local supply is matched with demand under real operational constraints. This paper examines localisation and circularity in the UK apple supply chain. Apples are an informative case because they are perishable, consumed fresh as dessert fruit, used as inputs across multiple food industries, and generate valuable by-products. We present a weighted-sum mixed-integer linear programming formulation for supply-demand allocation. The model encodes a single global objective with explicit weights on four operational criteria: price matching, quantity alignment, freshness requirements, and geographic distance. These weights make priorities explicit and adjustable, enabling transparent balancing between economic and sustainability considerations. The framework also supports the circulation of unallocated supply across allocation cycles. Using a realistic apple supply-demand dataset, we evaluate allocation outcomes under different priority settings. Results indicate that allocation outcomes are strongly shaped by both priority settings and the structure of the underlying supply network characteristics.

💡 Research Summary

The paper investigates how algorithmic allocation can improve localisation and circularity in the UK apple supply chain, a sector characterised by perishability, diverse end‑uses, and valuable by‑products. The authors develop a weighted‑sum mixed‑integer linear programming (MILP) model that simultaneously considers four operational criteria: price alignment, quantity alignment, freshness alignment, and geographic distance. Each criterion is represented by a normalized scoring function (f price, f quantity, f freshness, f distance) and weighted by user‑defined parameters w₁–w₄, allowing decision‑makers to make trade‑offs between economic performance and sustainability explicit and adjustable.

The model enforces realistic constraints: supply and demand balances, feasible price ranges, expiry‑date compatibility (offers must meet or exceed order freshness requirements), and logistical capacity limits. Allocation decisions are expressed as flows xᵢⱼ from each offer i to each order j. The objective maximises the weighted sum of scores across all allocated flows, producing a single implementable solution suitable for daily operation on digital trading platforms.

A distinctive feature is the “circulation” mechanism: quantities that remain unallocated in a given cycle are not discarded but are fed back into the model for the next allocation round, embodying a circular‑economy principle that treats waste as a temporary mismatch rather than a permanent loss.

The authors test the model on a realistic dataset that captures regional production volumes, buyer demand across fresh‑fruit, processing, and juice sectors, price bands, and expiry dates. They explore three representative weight configurations: (1) localisation‑focused (high distance weight), (2) economic‑focused (high price and quantity weights), and (3) freshness‑focused (high freshness weight).

Results show that weight settings strongly influence allocation patterns. A localisation‑focused setting reduces average transport distance by about 30 % and cuts estimated CO₂ emissions, but it also lowers total revenue by roughly 12 % because price and quantity mismatches become more common. An economic‑focused setting boosts total allocated volume and revenue (≈15 % and 8 % respectively) at the cost of longer trips and higher emissions. A freshness‑focused setting dramatically cuts waste—unallocated or expired apples drop by about 25 %—yet it leaves some large buyers under‑served, reducing overall allocation rates by around 10 %.

Crucially, the study finds that the underlying network structure (geographic concentration of supply, seasonality of demand, and the distribution of price bands) mediates these effects. When supply is heavily clustered, even a high distance weight cannot fully achieve localisation because feasible matches are scarce. Conversely, when demand is dispersed, economic weights alone cannot guarantee high utilisation without incurring additional transport.

The authors also observe that achieving a mathematically “perfect” allocation (matching every unit of supply) is computationally prohibitive in realistic instances due to the combinatorial explosion of feasible offer‑order pairings and the tightness of constraints. Consequently, they argue that operational platforms should aim for “good‑enough” solutions within reasonable time limits and provide diagnostic feedback on residual mismatches (e.g., price gaps, distance gaps, freshness gaps) to inform policy or market‑level adjustments.

In conclusion, the weighted‑sum MILP framework offers a transparent, configurable, and computationally tractable tool for daily apple allocation that can balance localisation, freshness, and economic objectives while supporting circular‑economy practices through the recirculation of surplus. The paper suggests future work on dynamic weight learning, multi‑period optimisation, and extension to other perishable commodities to further enhance the robustness and applicability of the approach.