15.2 Distribution Strategies, Cross-Docking, and Vehicle Routing

Distribution management focuses on the tactical and operational execution of moving finished goods from manufacturing plants and distribution hubs to retail shelves or customer doorsteps. Industrial engineers select distribution strategies to minimize handling touches, maximize freight fill rates, and establish optimal vehicle delivery routes. Advanced techniques such as cross-docking and mathematical vehicle routing algorithms directly determine the velocity, cost, and reliability of the logistics network.

---

1. Distribution Network Strategies and Topologies

The configuration of physical distribution channels dictates how inventory is positioned and how transport is scheduled.

Core Distribution Network Topologies
 (a) Direct Shipping            (b) Direct Shipping with Milk Runs
     [Plant]                         [Plant]
     ┌──┴──┐                         ┌──┴────────────────┐
     ▼     ▼                         ▼                   ▼
   [R1]   [R2]                   [Stop 1] ──> [Stop 2] ──> [Stop 3]

 (c) Intermediate DC with Storage (d) Cross-Docking Facility
     [Plants A, B, C]                [Suppliers A, B, C]
            │                               │ (Inbound Staging)
            ▼                               ▼
     [Storage DC]                    [Cross-Dock Hub] (Dwell < 24 hrs, No Storage)
     (Pallet Racks, Picking)                │ (Sortation & Direct Transfer)
            │                               ▼
            ▼                        [Outbound Retail Trailers]
     [Retail Stores]                 [Stores 1, 2, 3]

Comparison of Distribution Strategies

---

2. Cross-Docking Operations and Architectural Requirements

Cross-docking is a logistics technique where freight is transferred directly from inbound receiving trailers to outbound delivery trailers with minimal or zero intermediate storage dwell time (typically under 24 hours, and virtually never exceeding 48 hours).

Cross-Dock Facility Material Flow (I-Shape Terminal Layout)
               Inbound Dock Doors (Receiving Side)
           ▼         ▼         ▼         ▼         ▼
     ┌───────────────────────────────────────────────────┐
     │  [ Inbound Staging & Deconsolidation ]            │
     │                     │                             │
     │                     ▼                             │
     │  [ Automated Barcode / RFID Sortation Conveyor ]  │
     │                     │                             │
     │                     ▼                             │
     │  [ Outbound Staging & Direct Trailer Loading ]    │
     └───────────────────────────────────────────────────┘
           ▲         ▲         ▲         ▲         ▲
               Outbound Dock Doors (Shipping Side)

Operational Configurations: Pre-Distribution vs. Post-Distribution

1. Pre-Distribution Cross-Docking: The supplier pre-sorts, packs, and labels individual cartons or pallets with the ultimate retail store destination before the shipment leaves the factory. The cross-dock personnel or sortation conveyors simply read the destination barcode and route the container directly to the corresponding outbound trailer.
2. Post-Distribution Cross-Docking: The supplier ships bulk pallets of single SKUs to the cross-dock. Upon arrival, the facility's Warehouse Management System (WMS) dynamically allocates the palletized goods to open store orders, cross-dock operators break down pallets, and cartons are sorted and labeled for outbound delivery trailers.

Critical Technical Requirements for Cross-Docking

• Electronic Data Interchange (EDI 856 / ASN): Inbound trailers must transmit an Advanced Shipping Notice (ASN) prior to arrival, itemizing pallet and carton contents down to the serial shipping container code (SSCC).
• Automated Identification: Every container must carry a standardized machine-readable identifier (GS1-128 barcode or EPC RFID tag) to enable high-speed overhead scanning on sortation conveyors.
• Synchronized Carrier Dock Scheduling: Inbound delivery windows and outbound linehaul departures must be tightly synchronized. If an inbound trailer is 2 hours late, downstream outbound trucks face departure delays or move partially empty.
• Facility Architectural Layout: Modern cross-dock terminals utilize an elongated I-shape rectangular footprint. Mathematically, an I-shape minimizes the average travel distance between inbound dock doors on one side and outbound dock doors on the opposing side. For massive terminals with more than 150–200 dock doors, T-shaped or X-shaped footprints are deployed to prevent forklift congestion.

Economic Advantages

Traditional warehousing allocates 55% to 65% of its operating budget to put-away, storage replenishment, and manual order picking. Cross-docking bypasses all three functions, driving warehouse handling costs down to basic unloading, sorting, and direct loading.

---

3. Network Routing: The Traveling Salesperson Problem (TSP)

When a single vehicle must visit a set of geographically dispersed customer locations and return to its home depot, the routing optimization is modeled as a Traveling Salesperson Problem (TSP).

Mathematical Formulation

Given a set of nodes $V = {0, 1, 2, \dots, n}$, where Node 0 is the central depot and nodes ${1, \dots, n}$ represent customer stops, with travel cost/distance $c_{ij}$ between nodes $i$ and $j$:

$\min \sum_{i=0}^n \sum_{j=0}^n c_{ij} x_{ij}$

$\text{Subject to: } \sum_{j=0, j \ne i}^n x_{ij} = 1 \quad \forall i \in V \quad \text{(Leave each node exactly once)}$

$\sum_{i=0, i \ne j}^n x_{ij} = 1 \quad \forall j \in V \quad \text{(Enter each node exactly once)}$

$u_i - u_j + n x_{ij} \le n - 1 \quad \forall 1 \le i \ne j \le n \quad \text{(Subtour Elimination Constraints)}$

$x_{ij} \in \{0, 1\}, \quad u_i \ge 0$

• The Miller-Tucker-Zemlin (MTZ) subtour elimination constraints prevent the formation of isolated cycles that do not pass through the central depot.
• Complexity: The TSP is NP-hard. An exact exhaustive search requires evaluating $\frac{(n-1)!}{2}$ possible tours for a symmetric network, making heuristic approaches mandatory for practical engineering.

The Nearest Neighbor (NN) Heuristic

The Nearest Neighbor method is a greedy construction heuristic:

1. Start at the central depot (Node 0).
2. Identify the unvisited customer node closest to the current location.
3. Travel to that node, add it to the tour, and flag it as visited.
4. Repeat Steps 2 and 3 until all $n$ customer nodes have been visited.
5. Travel from the final customer node directly back to the central depot.

• Limitation: Because NN makes locally greedy choices early in the tour, it frequently leaves isolated, distant nodes for the final steps, forcing a catastrophic and expensive final return leg.

---

4. The Capacitated Vehicle Routing Problem (CVRP) and Clarke-Wright Savings Method

When a fleet of vehicles with finite payload capacities ($Q$) must serve customer demands ($d_i$) from a central depot, the problem expands into the Capacitated Vehicle Routing Problem (CVRP).

Clarke-Wright Savings Algorithm Derivation

Developed by Geoff Clarke and John Wright, this heuristic constructs multi-stop vehicle routes by iteratively merging radial round-trips based on distance savings.

Clarke-Wright Savings Concept
 Baseline: Separate Radial Routes (Total = 2*c(0,i) + 2*c(0,j))   Merged Route: Single Tour (Total = c(0,i) + c(i,j) + c(j,0))
             [Depot 0]                                                          [Depot 0]
             ╱   ▲   ╲   ▲                                                      ╱       ▲
            ╱   ╱     ╲   ╲                                                    ╱         ╲
           ▼   ╱       ▼   ╲                                                  ▼           ╲
       [Cust i]         [Cust j]                                          [Cust i] ──────> [Cust j]

1. Baseline Inefficient Solution: Each customer $i$ is served by an independent out-and-back radial round trip from the depot (Node 0): $\text{Distance for Customer } i = c(0, i) + c(i, 0) = 2 c(0, i)$ $\text{Total Baseline Distance} = \sum_{i=1}^n 2 c(0, i)$

2. Merged Route: If customer $i$ and customer $j$ are combined onto a single vehicle route ($0 \to i \to j \to 0$): $\text{Merged Distance} = c(0, i) + c(i, j) + c(j, 0)$

3. Savings Formulation: The distance saved ($s(i, j)$) by linking customer $i$ and customer $j$ directly rather than returning to the depot between stops is: $s(i, j) = [2 c(0, i) + 2 c(0, j)] - [c(0, i) + c(i, j) + c(j, 0)]$ $s(i, j) = c(0, i) + c(0, j) - c(i, j)$

Algorithmic Execution Steps

1. Calculate the savings $s(i, j) = c(0, i) + c(0, j) - c(i, j)$ for all distinct pairs of customers $i$ and $j$ ($i \ne j$).
2. Rank all candidate pairs $(i, j)$ in strictly descending order of savings.
3. Starting from the top of the savings list, merge customer $i$ and customer $j$ onto a common route provided:
   • Both $i$ and $j$ are currently adjacent to the depot (neither node is already interior to an established multi-stop route).
   • Customers $i$ and $j$ do not already belong to the same route (prevents forming closed loops that bypass the depot).
   • The sum of customer demands on the proposed merged route does not exceed vehicle capacity $Q$: $\sum_{k \in \text{Route}} d_k \le Q$
4. Repeat Step 3 until the savings list is exhausted or no further feasible capacity merges exist.

---

5. Freight Consolidation Economics

Freight consolidation combines multiple small consignments destined for a common geographic zone into a single high-volume linehaul transport.

Break-Bulk Terminal Mechanics

• Inbound Gathering: Local pickup trucks gather small LTL shipments from multiple regional manufacturing plants and bring them to an origination break-bulk terminal.
• Consolidation: Shipments sharing a distant regional destination are loaded into a single 53-ft trailer to achieve a high weight/cube utilization (approaching $45,000\text{ lbs}$).
• Linehaul: The trailer executes a long-distance linehaul under full truckload economics.
• Break-Bulk Sort & Delivery: At the destination break-bulk hub, the trailer is unloaded, and freight is distributed among local delivery trucks for final delivery.

Freight Tariff Structures and the Weight-Bumping Rule

Motor freight carriers publish sliding-scale tariff rates per hundredweight ($/cwt$). As shipment weight crosses defined weight brackets, the rate per hundredweight decreases:

Standard LTL Weight Brackets:  L5C (< 500 lbs)  -->  M5C (500-999 lbs)  -->  M1M (1,000-1,999 lbs)
                               -->  M2M (2,000-4,999 lbs)  -->  M5M (5,000-9,999 lbs)  -->  TL (Truckload)

• The Bumping / Phantom Weight Rule: Carriers permit shippers to "bump" a shipment's billed weight to the minimum weight of the next higher bracket if doing so produces a lower total freight charge: $\text{Billed Charge} = \min\left(W_{\text{actual}} \times R_{\text{current}}, \; W_{\text{bracket, min}} \times R_{\text{higher}}\right)$

---

6. Step-by-Step Worked Engineering Calculations

Worked Example 15.2.1: Nearest Neighbor Heuristic for TSP

Problem: A distributor must route a delivery van from its central depot (Node 0) to four customer retail locations (Nodes 1, 2, 3, and 4) and return to the depot. The symmetric travel distance matrix (in miles) is provided below.

1. Construct the complete delivery tour using the greedy Nearest Neighbor heuristic starting at Depot 0.
2. Compute the total traveled mileage of the Nearest Neighbor tour.
3. Compare this result with an alternative sequence $0 \to 3 \to 2 \to 4 \to 1 \to 0$ to assess optimality.

Solution:

Step 1: Execute Nearest Neighbor Step-by-Step

• Start at Node 0:
  • Distances to unvisited nodes: $c(0, 1) = 12$, $c(0, 2) = 25$, $c(0, 3) = 18$, $c(0, 4) = 30$.
  • Minimum distance is $c(0, 1) = 12$. Move to Node 1.
  • Visited: ${0, 1}$. Current path: $0 \to 1$.
• From Node 1:
  • Distances to unvisited nodes: $c(1, 2) = 15$, $c(1, 3) = 22$, $c(1, 4) = 28$.
  • Minimum distance is $c(1, 2) = 15$. Move to Node 2.
  • Visited: ${0, 1, 2}$. Current path: $0 \to 1 \to 2$.
• From Node 2:
  • Distances to unvisited nodes: $c(2, 3) = 10$, $c(2, 4) = 14$.
  • Minimum distance is $c(2, 3) = 10$. Move to Node 3.
  • Visited: ${0, 1, 2, 3}$. Current path: $0 \to 1 \to 2 \to 3$.
• From Node 3:
  • Only unvisited node remaining is Node 4. $c(3, 4) = 16$. Move to Node 4.
  • Visited: ${0, 1, 2, 3, 4}$. Current path: $0 \to 1 \to 2 \to 3 \to 4$.
• Return to Depot:
  • Travel from Node 4 back to Node 0. $c(4, 0) = 30$.
  • Final Nearest Neighbor Tour: $0 \to 1 \to 2 \to 3 \to 4 \to 0$.

Step 2: Calculate Total Mileage $\text{Total Distance} = c(0, 1) + c(1, 2) + c(2, 3) + c(3, 4) + c(4, 0)$ $\text{Total Distance} = 12 + 15 + 10 + 16 + 30 = 83\text{ miles}$

Step 3: Evaluate Alternative Tour ($0 \to 3 \to 2 \to 4 \to 1 \to 0$) $\text{Alternative Distance} = c(0, 3) + c(3, 2) + c(2, 4) + c(4, 1) + c(1, 0)$ $\text{Alternative Distance} = 18 + 10 + 14 + 28 + 12 = 82\text{ miles}$

• Engineering Takeaway: The alternative tour ($82\text{ miles}$) is shorter than the Nearest Neighbor tour ($83\text{ miles}$). This demonstrates that the greedy Nearest Neighbor heuristic is not guaranteed to find the global optimum because it was forced to make an expensive 30-mile final return leg ($4 \to 0$).

---

Worked Example 15.2.2: Clarke-Wright Savings Method for CVRP

Problem: A regional parts depot (Node 0) operates delivery trucks with a payload capacity of $Q = 1,000\text{ lbs}$. Three commercial repair shops must be serviced:

• Customer 1: Demand $d_1 = 400\text{ lbs}$
• Customer 2: Demand $d_2 = 500\text{ lbs}$
• Customer 3: Demand $d_3 = 350\text{ lbs}$

The travel distances (in miles) from the depot and between customers are:

• Depot distances: $c(0, 1) = 20$, $c(0, 2) = 30$, $c(0, 3) = 25$
• Inter-customer distances: $c(1, 2) = 22$, $c(1, 3) = 28$, $c(2, 3) = 32$

1. Calculate the baseline total mileage if every customer is served by an independent out-and-back radial trip.
2. Compute the Clarke-Wright savings value $s(i, j)$ for all pairs.
3. Determine the final vehicle routes and verify capacity feasibility.
4. Compute the final total system mileage and net miles saved.

Solution:

Step 1: Baseline Radial Solution $\text{Baseline Distance} = 2 c(0, 1) + 2 c(0, 2) + 2 c(0, 3) = 2(20) + 2(30) + 2(25) = 40 + 60 + 50 = 150\text{ miles}$

Step 2: Compute Pairwise Savings Using $s(i, j) = c(0, i) + c(0, j) - c(i, j)$

• Pair (1, 2): $s(1, 2) = c(0, 1) + c(0, 2) - c(1, 2) = 20 + 30 - 22 = 28\text{ miles}$
• Pair (2, 3): $s(2, 3) = c(0, 2) + c(0, 3) - c(2, 3) = 30 + 25 - 32 = 23\text{ miles}$
• Pair (1, 3): $s(1, 3) = c(0, 1) + c(0, 3) - c(1, 3) = 20 + 25 - 28 = 17\text{ miles}$

Ranked Savings Table:

1. Pair (1, 2): $s(1, 2) = 28\text{ miles}$
2. Pair (2, 3): $s(2, 3) = 23\text{ miles}$
3. Pair (1, 3): $s(1, 3) = 17\text{ miles}$

Step 3: Route Merging with Capacity Verification

• Evaluate Rank 1 (Pair 1, 2):
  • Combined demand: $d_1 + d_2 = 400 + 500 = 900\text{ lbs}$.
  • Capacity check: $900\text{ lbs} \le Q = 1,000\text{ lbs}$. Feasible!
  • Form Route 1: $0 \to 1 \to 2 \to 0$ (or $0 \to 2 \to 1 \to 0$). Nodes 1 and 2 are now linked.
• Evaluate Rank 2 (Pair 2, 3):
  • Node 2 is already linked to Node 1. Merging Node 3 with Node 2 creates a multi-stop route $0 \to 1 \to 2 \to 3 \to 0$.
  • Combined demand: $d_1 + d_2 + d_3 = 400 + 500 + 350 = 1,250\text{ lbs}$.
  • Capacity check: $1,250\text{ lbs} > Q = 1,000\text{ lbs}$. Infeasible! Vehicle capacity violated; reject this merge.
• Evaluate Rank 3 (Pair 1, 3):
  • Merging Node 3 with Node 1 also requires serving all three customers on one vehicle ($1,250\text{ lbs} > 1,000\text{ lbs}$). Infeasible! Reject.
• Node 3 must remain on its own dedicated radial route: $0 \to 3 \to 0$.

Step 4: Final Routing Solution and Net Mileage

• Route 1: $0 \to 1 \to 2 \to 0$ $\text{Distance}_1 = c(0, 1) + c(1, 2) + c(2, 0) = 20 + 22 + 30 = 72\text{ miles}$ $\text{Payload}_1 = 400 + 500 = 900\text{ lbs} \quad (90\%\text{ capacity utilization})$
• Route 2: $0 \to 3 \to 0$ $\text{Distance}_2 = 2 c(0, 3) = 2 \times 25 = 50\text{ miles}$ $\text{Payload}_2 = 350\text{ lbs}$
• Total System Mileage: $\text{Total Miles} = 72 + 50 = 122\text{ miles}$
• Net Savings: $\text{Savings} = \text{Baseline} - \text{Total} = 150 - 122 = 28\text{ miles}$ (Note: Net savings exactly matches $s(1, 2) = 28\text{ miles}$).

---

7. NCEES Reference Handbook Tips & Realistic Exam Traps

• The Clarke-Wright Savings Formula Trap: The formula is $s(i, j) = c(0, i) + c(0, j) - c(i, j)$. A common exam blunder is adding the inter-customer distance ($c(0, i) + c(0, j) + c(i, j)$) or forgetting that the savings represents what is eliminated when returning to the depot is bypassed.
• Capacity Constraint Violations: In any CVRP problem, always verify cumulative demand before executing a merge. FE questions frequently present a customer pair with the highest mathematical savings that exceeds the truck payload limit. If capacity is exceeded, the merge is invalid and must be skipped.
• Subtour Traps in Heuristic Merges: In Clarke-Wright, you can only link customer $i$ and customer $j$ if both nodes are currently directly connected to the depot. If customer $i$ is already an interior stop on an existing route (e.g., $0 \to k \to i \to m \to 0$), customer $i$ cannot be linked to customer $j$ without disconnecting an existing customer.
• Cross-Docking Dwell Time Definition: On the exam, if a question describes a distribution center where inventory is stored in multi-level pallet racks for several weeks before being picked to fulfill retail orders, that facility is a traditional warehouse, NOT a cross-dock. Cross-docking involves minimal dwell (< 24 hours) and zero rack storage.