Evaluating Automated Guided Vehicle (AGV) Use in Distribution Centers

Walmart

Overview

Distribution centers (DCs) serve the need of retailers by fulfilling their orders. Walmart was investigating the role of automated guided vehicles (AGVs) to improve the efficiency of its DCs and reduce reliance on manual labor.  We evaluated different methods of using automated guided vehicles (AGVs) in Walmart distribution centers and recommended the most cost-effective option.
agv-distribution-case-study-img

Business Problem

Walmart, which is the largest retailer in North America, needs to improve distribution center productivity and reduce the cost of operations. DCs form the backbone of the distribution of goods from manufacturers to retailers. In a DC, employees known as pickers fulfill store orders by traversing through the aisles of inventory, while obtaining requested items and placing them on a forklift. In this project, Walmart asked us to determine how AGVs could improve productivity and reduce manpower costs for pickers. Our team created a simulation and decision support solution for the following AGV methods in a DC:
  • Single-tethered picking: Each picker was assigned to an AGV instead of a forklift truck.
  • Double-tethered picking: Each picker was assigned to two AGVs to fulfill two orders simultaneously.
  • Zone-to-zone picking: We divided all aisles in the DC into multiple zones and posted a picker in each zone. While AGVs moved from aisle to aisle on their own, pickers loaded AGVs in their zones.

Our Approach

After we spent several days in a Walmart DC to understand and document its unique operations and processes, we developed and implemented a simulation system model to recreate the DC’s current operations virtually. We validated the system by comparing the model key performance indicators (KPIs) with the actual KPIs, and we established that both were very close.

Next, we developed simulation systems to model single-tethered, double-tethered and zone-to-zone picking options, using smart algorithms to route pickers and AGVs. Finally, we packaged these simulation algorithms into a decision support system to create multiple scenarios. Users could view each scenario’s simulation through the visual animation of pickers and AGVs, and then analyze any scenario’s results via reports, charts, and graphs.

Key Benefits

  1. Our simulation systems proved that they were excellent representations of real-world scenarios. The proof: Our simulation KPIs were within two percent of the DC’s actual KPIs.
  2. From a cost point of view, we decided that single-tethered picking by AGVs was the best method. Contrary to our initial intuition, zone-to-zone picking wasn’t as effective, since pickers spent too much time waiting for AGVs.

Results

Walmart decided to implement single-tethered picking for one DC and ordered AGVs for that DC. Our efforts helped the largest retailer in North America improve productivity and reduce operational costs.