Home/Case Studies/US Distribution — 1,400 Stores, 14 DCs
Logistics · 2023

US Distribution — 1,400 Stores, 14 DCs

A US national retailer operating 1,400 stores from 14 distribution centres was paying more in transportation than its network design justified — and was over-exposed to single-DC disruptions.

US Distribution — 1,400 stores across 14 DCs, with K-means cluster colouring and proposed new DC sites
Approach

ARTEMIS Logistics: load-consolidation analysis across DC pairs surfaced ~30% transportation savings opportunities. K-means clustering on store demand identified new DC siting candidates at Mansfield, Holyoke, Waco, Vancouver WA, Indio, and Grand Junction.

Result

A roadmap that quantified both immediate consolidation savings and the network-redesign capex/opex tradeoff for new DC siting — with disruption-resilience modeled for each option.

Key facts
  • ~30% transportation savings identified
  • K-means DC siting analysis
  • 6 new DC candidates surfaced
  • Disruption-resilience modeled per option
Worked example

Optimal DC placement — Apple Valley vs. Palm Springs

Step 6 of the K-means workflow re-solves each cluster centroid against its served stores. In the Southern California cluster, the legacy DC in Apple Valley is geometrically off-centre — the mathematically optimal centroid sits near Palm Springs, which would shorten weighted store-miles for the cluster. The model surfaces the candidate; the business decision (real-estate cost, moving cost, recovery timeline on mileage savings) then determines whether the move is worth executing.

Cluster optimisation output for the Southern California DC — store-to-DC lines from the existing Apple Valley site (red) versus the mathematically optimal Palm Springs centroid (yellow), overlaid on a regional basemap covering Victorville, San Bernardino, Riverside, Palm Springs and Joshua Tree National Park
K-means Step 6 — store-to-DC lines from the legacy Apple Valley DC (red) vs. the optimised Palm Springs centroid (yellow) for the Southern California cluster.

See another case study →

All case studies