Transportation Planning, Routing and Scheduling

Introduction and Context

Transportation decisions determine how products move through a network of suppliers, plants, distribution centres, cross-docks, customers and return points. In logistics, transportation is not merely a movement activity. It is a coordinated planning problem involving shipment consolidation, fleet capacity, route design, dispatch timing, loading, backhauls and customer-service commitments.

At postgraduate level, transportation planning should be viewed as a hierarchy of decisions. Strategic decisions determine network structure, fleet ownership and modal policy. Tactical decisions determine consolidation rules, carrier allocation, fleet sizing and delivery territories. Operational decisions determine daily routes, vehicle loading, dispatch sequences and response to disruptions.

The central managerial challenge is to minimize total logistics cost while satisfying capacity, time, regulatory and service constraints. Cost minimization alone may produce late deliveries, excessive handling, poor asset utilization or unacceptable customer experience. Effective planning therefore uses a total-cost and service-based perspective.

Learning Objectives

By the end of this chapter, the learner should be able to:

1. Explain the hierarchy of transportation planning decisions.
2. Evaluate shipment consolidation and fleet-capacity requirements.
3. Formulate basic vehicle-routing and scheduling models.
4. Apply constructive and improvement heuristics to practical routing problems.
5. Design load-building, backhauling and cross-docking processes.
6. Assess last-mile delivery alternatives under service constraints.
7. Interpret optimization results critically and manage implementation risks.

1. The Transportation Planning Hierarchy

Transportation planning operates across three linked horizons.

These levels are interdependent. A strategically centralized network may reduce inventory and facility cost but increase transport distance. A large owned fleet may provide control but create fixed costs and empty capacity. Operational routing cannot fully correct a poorly designed network.

A useful objective function is total logistics cost:

Here, C_fixed is vehicle, facility or contract cost; C_distance is distance- or fuel-related cost; C_time is driver and waiting cost; C_handling is loading, unloading and transfer cost; C_inventory is the cost of goods in transit and safety stock; C_penalty represents late, failed or non-compliant delivery; and C_externality represents environmental and social costs. The equation is a cost-accounting framework rather than a universal optimization formula. Its terms must be measured consistently over the same planning horizon.

2. Shipment Consolidation

Consolidation combines several smaller shipments into a larger movement. It may occur spatially, temporally or both. Examples include combining orders for customers in one geographic area, holding freight until a full truckload is available, or consolidating inbound supplier deliveries at a consolidation centre.

The benefits include lower cost per unit, fewer vehicle trips, reduced congestion and improved equipment utilization. The disadvantages include waiting time, additional handling, increased inventory in transit and a possible reduction in delivery flexibility.

A simple consolidation decision compares the saving from combining shipments with the cost of waiting and handling. If two shipments have stand-alone transport costs C₁ and C₂, while their consolidated cost is C₁₂, the direct transport saving is:

Consolidation is attractive only if the saving exceeds incremental waiting, handling, inventory and service-risk costs. The formula assumes comparable shipment requirements and excludes costs that must be added separately.

Consolidation policies include:

• Quantity-based consolidation: wait until a minimum weight, volume or pallet quantity is reached.
• Time-based consolidation: collect shipments during a fixed window and dispatch at scheduled times.
• Geographic consolidation: combine deliveries or pickups within a territory.
• Cross-company consolidation: combine compatible freight from several shippers.
• Hub-and-spoke consolidation: move freight through a central hub before regional distribution.

A postgraduate planner must distinguish physical capacity from operational compatibility. Two loads may fit in a vehicle but still be incompatible because of temperature, hazardous-material, security, product-segregation or customer-priority requirements.

3. Fleet Capacity and Vehicle Utilization

Fleet planning must consider weight, cube, pallet positions, axle limits, legal restrictions, driver hours and delivery sequence. The binding constraint is the first constraint that prevents further loading.

For a planning period, a basic capacity requirement can be estimated as:

where D is required transport demand in compatible capacity units, Q is usable capacity per vehicle, and N is the minimum number of vehicle-load equivalents. The ceiling function indicates that a fraction of a vehicle still requires capacity. This is a lower bound, not necessarily the actual fleet requirement, because route geography, time windows and vehicle availability may require additional vehicles.

If demand is measured in tonnes, usable vehicle capacity is:

where Q_rated is rated payload and α is the permitted utilization factor, with 0 < α ≤ 1. The factor may reflect safety margins, uneven demand, loading inefficiency and legal restrictions. Payload and volume should not be combined as if they were interchangeable.

Vehicle utilization can be reported as:

and

where W denotes weight and V denotes volume. A route can have high weight utilization but poor cube utilization, or the reverse. A useful dashboard therefore reports both.

Worked Example: Capacity Screening

A depot must deliver 92 tonnes during a planning day. Each available vehicle has a rated payload of 12 tonnes, but the operating policy permits 90% utilization. Usable capacity is:

Q_usable = 12 × 0.90 = 10.8 tonnes.

The minimum vehicle-load equivalents are:

N = ⌈92 ÷ 10.8⌉ = ⌈8.52⌉ = 9.

Thus, at least nine vehicle-load equivalents are required. The result does not guarantee feasibility: time windows, route distance, driver regulations and loading time may require more than nine vehicles or a different dispatch plan.

4. Vehicle Routing Problems

The vehicle-routing problem (VRP) determines which customers each vehicle serves and the sequence in which they are visited. A standard capacitated VRP generally assumes a depot, a set of customers, known demand, vehicle capacity and travel costs. Extensions include time windows, multiple depots, heterogeneous fleets, pickup-and-delivery precedence, split deliveries, stochastic demand and electric-vehicle charging.

Let customers be indexed by i,j ∈ V, with depot 0. Let c_ij be travel cost from node i to node j, and x_ij equal 1 if that arc is used and 0 otherwise. A basic routing objective is:

Subject to each customer being entered and left once:

for every customer. Capacity feasibility requires the total demand assigned to a route not to exceed vehicle capacity:

where q_i is customer demand, Q is vehicle capacity and R is the customer set on one route. Additional constraints are required to prevent disconnected subtours and to ensure depot departure and return. The formulation above is a conceptual core, not a complete mixed-integer programming model.

Time Windows and Scheduling

A customer may require service between earliest time a_i and latest time b_i. If t_i is service-start time, then:

For travel time τ_ij and service duration s_i, sequence feasibility requires:

when customer j follows i. Waiting is permitted if the vehicle arrives before a_j. If late service is permitted, lateness [[MATH{L_i}]] may be penalized using L_i = max(0, t_i − b_i). Hard and soft time windows must be distinguished: a hard window prohibits violation, whereas a soft window permits violation at a stated penalty.

Scheduling also includes driver work limits, legal breaks, loading appointments, yard congestion and customer unloading capacity. A mathematically short route may be operationally infeasible if it creates excessive waiting or violates working-time rules.

5. Exact Optimization and Heuristics

Exact mixed-integer optimization can produce an optimal solution for small or moderately sized instances, but computational effort grows rapidly as customers and constraints increase. Large operational problems commonly use heuristics or metaheuristics.

Constructive heuristics

• Nearest neighbour: repeatedly visit the closest feasible unserved customer. It is fast but can create poor final legs.
• Savings method: begin with separate depot-to-customer routes and combine routes when the saving is attractive.
• Sweep method: sort customers by polar angle around the depot, group them into capacity-feasible routes and sequence each group.
• Insertion heuristic: insert an unserved customer into the position producing the smallest feasible incremental cost.

For customers i and j, the classical distance saving from merging separate routes is:

where 0 denotes the depot. A positive saving means that joining the customers may reduce distance, but the merge remains subject to capacity, time-window and route-compatibility constraints.

Improvement and metaheuristic methods

Local search improves an existing solution through moves such as 2-opt, relocation and exchange. In 2-opt, two route arcs are removed and replaced by two alternative arcs; the move is accepted if it reduces cost and preserves feasibility. Tabu search temporarily prohibits recently reversed moves to avoid cycling. Simulated annealing occasionally accepts an inferior move to escape local optima. Genetic algorithms evolve a population of encoded routes. Large-neighbourhood search destroys part of a solution and reconstructs it intelligently.

No heuristic is universally best. Evaluation should consider solution quality, computation time, stability, explainability and ease of operational implementation. A planner should compare heuristic results with lower bounds, historical performance and manually validated feasibility.

6. Dispatching, Load Building and Backhauling

Dispatching converts a plan into executable vehicle movements. It assigns loads, drivers and equipment; sets departure times; communicates instructions; and monitors exceptions. A sound dispatch process includes a frozen planning horizon, a cut-off time for new orders and explicit rules for urgent insertions.

Load building

Load building is a three-dimensional and sequence-dependent problem. Important rules include:

1. Respect weight, cube, pallet and axle limits.
2. Load deliveries in reverse stop sequence where practical.
3. Keep incompatible products segregated.
4. Protect fragile or temperature-sensitive goods.
5. Place frequently accessed items where unloading is safe and efficient.
6. Maintain traceability through scans, seals and load manifests.

A load that is physically feasible may still be operationally poor if the driver must unload several pallets to access an earlier delivery. Loading quality should therefore be assessed by both utilization and handling effort.

Backhauling

Backhauling uses a vehicle’s return journey to collect freight. It reduces empty kilometres and spreads fixed vehicle cost across more productive work. Backhaul feasibility depends on pickup time, origin, destination, vehicle capacity, product compatibility and the need to return equipment to the depot.

Empty-kilometre ratio is commonly measured as:

where K_empty is distance travelled without revenue or planned load and K_total is total distance. The metric must be defined consistently; some organizations exclude repositioning mandated by a contract. Reducing E is desirable, but not at the expense of late primary deliveries or excessive waiting.

7. Cross-Docking and Multimodal Coordination

Cross-docking transfers inbound freight to outbound vehicles with little or no storage. It is effective when demand is predictable, inbound and outbound schedules are synchronized, product identification is reliable and handling infrastructure is available.

A cross-dock process normally includes appointment scheduling, receiving, verification, sortation, staging, outbound loading and dispatch confirmation. The principal risks are schedule propagation, congestion, mis-sorting and the absence of buffer stock when an inbound vehicle is late.

In multimodal transportation, planning must coordinate road, rail, sea, air and inland waterways. Transfer decisions must include not only line-haul cost but also terminal handling, transfer time, reliability, documentation, security and emissions. A lower-cost mode may be unsuitable when it increases safety stock or threatens a critical service commitment.

8. Last-Mile Delivery

Last-mile delivery is often the most expensive and least predictable part of the distribution process because of dispersed demand, congestion, failed deliveries, narrow time windows and customer-specific requirements.

Common operating models include direct van delivery, parcel-carrier networks, parcel lockers, pickup points, crowdsourced delivery and micro-fulfilment. The appropriate model depends on order density, product characteristics, promised delivery speed, urban restrictions and customer willingness to collect.

Last-mile performance should be measured using multiple indicators:

The objective is not simply maximum stops per hour. Excessive route density can produce unsafe driving, failed deliveries and employee fatigue.

9. Illustrative Case: NorthStar Grocery Distribution

Background

NorthStar, an illustrative regional grocery distributor, serves 18 stores from one depot. It operates refrigerated vehicles with 10-tonne usable payload, has store delivery windows between 06:00 and 11:00, and receives supplier backhauls from four farms after store deliveries.

Problem

The company historically assigned routes by territory. Vehicles averaged 72% payload utilization, but empty kilometres were 31%, late deliveries reached 9%, and drivers frequently had to rearrange loads at stores. Management wanted to reduce cost without weakening the delivery promise.

Analysis

An order and route audit identified four causes:

• territorial routes ignored actual daily demand;
• high-priority stores were mixed with flexible-window stores;
• farm pickups were planned after routes rather than during route construction;
• load sequencing was performed manually without a stop-by-stop manifest.

The planning team separated hard and soft time windows, classified products by temperature and handling compatibility, and calculated feasible vehicle capacity after reserving space for planned backhauls. A savings-based initial solution was then improved using route relocation and 2-opt moves. The final routes were reviewed by dispatchers for road restrictions, unloading time and driver-work compliance.

Approach and solution

NorthStar introduced a rolling daily process:

1. Freeze confirmed orders at 18:00.
2. Build temperature-compatible loads using usable weight and cube limits.
3. Construct routes using time-window and capacity feasibility.
4. Insert farm pickups only when the vehicle retains capacity and the pickup does not threaten store windows.
5. Generate a reverse-stop loading manifest.
6. Dispatch with GPS tracking and exception alerts.
7. Re-optimize only defined exceptions, such as vehicle breakdown or a critical order.

Outcome

In the first eight-week pilot, the company reported an increase in average payload utilization to 84%, a reduction in empty kilometres to 19%, and improvement in on-time delivery to 96%. These figures are illustrative management results, not independently audited evidence. The pilot also revealed that a small number of routes became longer in distance but more reliable because waiting and reloading were reduced.

Learning points

The case demonstrates that routing software does not substitute for operational data or managerial judgment. Backhauls must be integrated before route construction, not appended afterward. Load sequencing affects route duration, and service reliability may justify a route with slightly higher distance. Finally, performance improvement requires a controlled implementation process, not only a mathematical solution.

10. Technology, AI and Governance

Transportation management systems integrate orders, maps, fleet availability, carrier rates, telematics and proof of delivery. Geographical information systems support network visualization and travel-time estimation. Application programming interfaces can connect marketplaces, carriers and customer-notification platforms.

AI and machine learning can forecast travel time, predict failed delivery, estimate arrival time, detect anomalous fuel use and recommend dynamic re-routing. However, predictive accuracy depends on representative data. Algorithms may reproduce historical bias, prioritize profitable customers unfairly or make decisions that dispatchers cannot explain. Human override, audit trails, cybersecurity and data-quality controls are therefore essential.

Dynamic routing should not be allowed to compromise safety, legal driving limits, cold-chain requirements or contractual service obligations. Personal data such as customer addresses and driver locations must be protected through access control, retention policies and secure transmission.

Environmental governance includes emissions reporting, alternative-fuel selection, urban access restrictions and route choices that reduce unnecessary kilometres. Carbon reduction should be measured using a stated boundary and consistent activity data; otherwise apparent improvements may result from accounting changes rather than operational change.

11. Limitations and Best Practices

Transportation models simplify reality. Travel times vary, demand is uncertain, loading times are not constant, and drivers possess local knowledge that may not appear in the data. A mathematically optimal plan based on inaccurate inputs is not an optimal operational plan.

Best practices include:

• validate master data for addresses, demand, service times and vehicle capacities;
• separate hard constraints from negotiable preferences;
• model weight, cube and compatibility independently;
• include loading, unloading, waiting and break times;
• use scenario analysis for demand and travel-time uncertainty;
• compare algorithmic output with dispatcher review;
• monitor actual versus planned route duration;
• define escalation and re-planning rules;
• evaluate total cost, service, safety and sustainability together;
• pilot changes before scaling them across the network.

Chapter Summary

Transportation planning links strategic network design with tactical fleet and consolidation decisions and operational routing, loading and dispatching. Consolidation can reduce transport cost but may create waiting, handling and service costs. Fleet capacity must be evaluated using weight, cube, legal and operational constraints. Vehicle-routing models assign customers and sequences while respecting depot, capacity and time-window requirements. Exact optimization is valuable for smaller problems, whereas savings, sweep, insertion, local search and metaheuristic methods are widely used for larger operational instances.

Effective dispatching requires feasible loads, sequence-aware loading and clear exception rules. Backhauling reduces empty movement when integrated with primary delivery planning. Cross-docking depends on synchronized schedules and reliable information. Last-mile decisions must balance cost, service, productivity, safety and sustainability. Technology and AI improve visibility and prediction, but governance, data quality, explainability and human oversight remain essential.

Key Terms

Transportation planning: Coordinating movement decisions across strategic, tactical and operational horizons.

Shipment consolidation: Combining multiple shipments into a larger movement.

Usable capacity: Capacity available after applying safety, legal and operating limits.

Vehicle utilization: The proportion of vehicle weight or volume capacity used.

Vehicle-routing problem: The problem of assigning and sequencing customer visits by vehicles.

Time window: Permitted interval for beginning or completing service.

Hard constraint: A condition that cannot be violated.

Soft constraint: A preference that may be violated at a penalty.

Savings heuristic: A route-construction method based on cost reduction from merging routes.

Local search: Iterative improvement through small changes to a current solution.

Metaheuristic: A general search strategy designed to explore difficult optimization problems.

Dispatching: Converting a transportation plan into executable vehicle movements.

Load building: Selecting and positioning freight within vehicle limitations.

Backhaul: Return movement that carries freight after the primary delivery journey.

Empty kilometres: Distance travelled without a productive load.

Cross-docking: Transferring inbound freight to outbound transport with minimal storage.

Last-mile delivery: Final movement from a distribution point to the customer.

Transportation management system: Software coordinating transport orders, planning, execution and visibility.

Route feasibility: Compliance of a route with all capacity, time, legal and operational constraints.

Review Questions

5 Short Answer Questions

1. Distinguish strategic, tactical and operational transportation planning.
2. Why should weight and cube utilization be reported separately?
3. What is the purpose of a time window in vehicle routing?
4. State two benefits and two risks of shipment consolidation.
5. Explain why a backhaul should be considered during route construction.

5 Long Answer Questions

1. Explain the structure and practical limitations of a capacitated vehicle-routing model.
2. Compare exact optimization, constructive heuristics and metaheuristics for transportation planning.
3. Discuss how load sequencing affects vehicle productivity and customer service.
4. Analyze the operational conditions required for successful cross-docking.
5. Evaluate the role of transportation management systems and AI in routing governance.

3 Analytical / Application Questions

1. A depot has 76 tonnes of compatible demand, vehicles with 11 tonnes rated payload and an allowed utilization factor of 0.85. Calculate the minimum vehicle-load equivalents and explain why the actual number of routes may be higher.
2. Design a routing policy for a distributor with urgent customers, flexible-window customers and planned supplier pickups. Identify the hard constraints, soft constraints and exception rules.
3. A company reduces empty kilometres but experiences a decline in on-time delivery. Diagnose the likely trade-off and propose a balanced performance-measurement framework.

Multiple Choice Questions with Answers

Question 1

Which planning decision is primarily tactical?

A. Selecting the location of a new distribution centre

B. Determining daily stop sequence for a vehicle

C. Setting fleet size and carrier contracts for the coming quarter

D. Recording proof of delivery

Correct Answer: C. Setting fleet size and carrier contracts for the coming quarter

Question 2

A vehicle has a rated payload of 12 tonnes and an operating utilization factor of 0.90. What is its usable payload?

A. 10.8 tonnes

B. 11.1 tonnes

C. 12.9 tonnes

D. 13.3 tonnes

Correct Answer: A. 10.8 tonnes

Question 3

What does a hard time window imply?

A. The window may be violated without consequence

B. The window cannot be violated in a feasible solution

C. Only the vehicle capacity constraint applies

D. The customer can always be served by a backhaul

Correct Answer: B. The window cannot be violated in a feasible solution

Question 4

In the classical savings expression S_ij = c_0i + c_0j − c_ij, what does a positive value indicate?

A. The two customers must be assigned to different depots

B. Merging routes may reduce travel cost

C. The vehicle has exceeded its payload

D. The route has violated a time window

Correct Answer: B. Merging routes may reduce travel cost

Question 5

Which practice most directly supports sequence-efficient vehicle loading?

A. Loading the first delivery deepest in every case

B. Ignoring stop order until the vehicle reaches the first customer

C. Using a reverse-stop loading manifest where practical

D. Mixing incompatible products to maximize cube utilization

Correct Answer: C. Using a reverse-stop loading manifest where practical

Question 6

What is the main purpose of a backhaul?

A. To increase empty return distance

B. To use return capacity for productive freight

C. To eliminate all customer time windows

D. To replace vehicle loading controls

Correct Answer: B. To use return capacity for productive freight

Question 7

Cross-docking is most effective when:

A. Inbound and outbound schedules are synchronized

B. All freight requires long-term storage

C. Demand and arrival times are completely unknown

D. Product identification is unnecessary

Correct Answer: A. Inbound and outbound schedules are synchronized

Question 8

Which metric measures the percentage of total distance travelled empty?

A. U_weight = W_loaded/W_capacity

B. E = K_empty/K_total × 100%

C. N = ⌈D/Q⌉

D. S_ij = c_0i+c_0j−c_ij

Correct Answer: B. E = K_empty/K_total × 100%

Question 9

Why can a mathematically optimal route be operationally unsuitable?

A. Optimization never uses cost data

B. It may rely on inaccurate travel, demand or service-time inputs

C. All optimal routes necessarily violate capacity

D. Optimization cannot represent a depot

Correct Answer: B. It may rely on inaccurate travel, demand or service-time inputs

Question 10

Which governance practice is especially important when AI recommends dynamic routes?

A. Remove all human override

B. Avoid maintaining audit trails

C. Protect data and preserve explainability and safety controls

D. Permit the algorithm to ignore legal driver limits

Correct Answer: C. Protect data and preserve explainability and safety controls