Problem:

The modern purchasing algorithm at Target is an automated algorithm that places orders to vendors with the details of the Target warehouses (distribution centres – DCs) where these items need to be shipped. Since the DCs serve multiple stores and vendors ship multiple items across categories, a purchase order can be so big that it doesn’t fit in a single truck or so small that it doesn’t make sense to commission an entire truck for the same.
This is where the Truck Load Optimization (TLO) algorithm steps in and recommends purchase orders that are in full truck loads. This algorithm also recommends truck numbers for existing purchase orders such they can be clubbed together to fill a truck.

Impact / Value:

Target commissions thousands of trucks each week to bring items from vendors to DCs. Hence, an algorithm to order in full truck loads drives economies of scale and saves fuel cost which leads to a sustainable purchasing.

Solution:

A three phased approach involving Bin Packing, Simulations and Binary Integer Programming

Outline of Talk:

1. Brief overview of the Target Supply Chain network
2. Problem statement of Purchasing Algorithm and Truck Load Optimization (How TLO is enhances the purchasing algorithm)
3. Ideal Formulation as a Linear Integer Programming and challenges in solving
4. Three Phased Solution:
   a. Handling Capacity Constraints
   b. Handling Thresholds
   c. Handling Utilization Goals
5. Impact of TLO ( Future planning of trucks, Ordering Cost Reduction, Sustainable Purchasing )
6. Opportunities ( Input / Truck size recommendations, 3D packing based on stackability / crushability)