Convect AI Flow Platform Logo

Model Details

Model Details

Overview

The Delivery Schedule Optimization application uses a time-expanded min-cost flow model to optimize warehouse-to-customer deliveries. This document explains the key concepts and constraints of the model.

Model Structure

Time Expansion

The model creates a time-expanded network where:

  • Each warehouse has a node for each day in the planning horizon
  • Each customer has nodes for their valid shipping windows
  • Arcs represent possible shipments between warehouses and customers

Network Components

  1. Warehouse Nodes

    • One node per warehouse per day
    • Capacity constraints on daily shipments
    • Connected to source node with total capacity
  2. Customer Nodes

    • One node per customer per shipping window
    • Demand constraints based on order volumes
    • Connected to sink node with total demand
  3. Shipping Arcs

    • Connect warehouse nodes to customer nodes
    • Costs based on transportation policies
    • Minimum flow constraints for consolidation
    • Capacity constraints based on available shipping capacity

Constraints

1. Capacity Constraints

  • Warehouse daily capacity limits
  • Minimum shipment sizes for consolidation
  • Maximum flow on shipping arcs

2. Time Constraints

  • Urgent orders must be shipped within 3 days
  • Batch orders must be shipped in next calendar week
  • Custom shipping windows based on order type

3. Flow Constraints

  • Flow conservation at each node
  • Minimum flow requirements for consolidation
  • Maximum flow limits based on capacity

4. Cost Structure

  • Tiered shipping rates based on volume
  • Different costs for different transportation modes
  • Consolidation discounts

Optimization Objective

The model minimizes total transportation cost while satisfying all constraints:

  • Minimize: Σ (flow on arc × cost per unit)
  • Subject to:
    • Capacity constraints
    • Time window constraints
    • Flow conservation
    • Minimum shipment sizes

Solution Process

  1. Graph Construction

    • Create time-expanded network
    • Add source and sink nodes
    • Set up arcs with costs and capacities
  2. Flow Optimization

    • Solve min-cost flow problem
    • Determine optimal flows on each arc
    • Calculate total cost
  3. Solution Extraction

    • Convert flows to shipment assignments
    • Calculate per-order costs
    • Generate output reports

Performance Considerations

Model Size

  • Number of nodes: O(W×D + C×W)
    • W: number of warehouses
    • D: number of days
    • C: number of customers
  • Number of arcs: O(W×C×D)

Solution Time

  • Typically solves in seconds to minutes
  • Depends on:
    • Number of orders
    • Planning horizon length
    • Number of warehouses and customers
    • Complexity of shipping rules

Limitations

  1. Deterministic model (no uncertainty)
  2. Linear cost structure
  3. No consideration of:
    • Loading/unloading times
    • Vehicle capacity
    • Driver schedules
    • Weather conditions