Freight, Urban, Last-Mile and Multimodal Applications

Introduction and context

Route planning becomes substantially more difficult when transport operations are embedded in different physical, commercial, and regulatory environments. A line-haul carrier may plan a small number of high-volume movements between terminals; a parcel operator may assign thousands of time-sensitive deliveries; a home-delivery firm must coordinate customer availability, vehicle capacity, and promised time windows; and a multimodal freight planner must synchronize different networks, transfer facilities, and documents.

The central managerial challenge is not simply to find the shortest route. It is to design a feasible service system that balances cost, reliability, asset utilization, emissions, safety, customer experience, and resilience. This chapter applies vehicle-routing and scheduling concepts to line-haul freight, parcel delivery, home delivery, service routing, reverse logistics, intermodal freight, cross-docking, urban consolidation, and crowdsourced delivery.

Learning objectives

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

1. distinguish major freight and delivery-routing applications;
2. formulate operational objectives and constraints for each application;
3. adapt vehicle-routing models to time windows, service operations, returns, and multimodal transfers;
4. evaluate cross-docking, urban consolidation, and crowdsourced delivery decisions;
5. interpret cost, service, sustainability, and risk trade-offs;
6. assess technology, governance, and implementation issues in applied routing systems.

1. A unifying view of applied route planning

A routing application can be described through six questions:

A useful abstraction is a directed network G = (N, A), where N is the set of nodes and A is the set of feasible arcs. Let c_ij be the generalized cost of travelling from node i to node j, t_ij the travel time, and q_i the quantity associated with customer i. Generalized cost may include fuel, labour, tolls, vehicle depreciation, expected delay, emissions cost, and service penalties.

The classical capacitated vehicle-routing problem (CVRP) uses binary variable x_ijk, equal to 1 when vehicle k travels directly from i to j, and 0 otherwise. If d_i is demand at customer i and Q_k is vehicle capacity, a basic capacity condition is:

Here, y_ik indicates whether vehicle k serves customer i, and node 0 is the depot. A common objective is:

The objective is applicable when costs are additive and known or represented by expected values. Real operations normally require additional constraints: every customer must be served once, routes must start and end at permitted depots, flow must be conserved, subtours must be prevented, and route duration must respect labour or regulatory limits.

For a customer with time window [a_i, b_i], service may begin at s_i only if:

If vehicle k reaches i after travelling from j, then, with service duration h_i and sufficiently large constant M:

This constraint is meaningful only when x_jik = 1; M must be large enough to deactivate the condition otherwise, but not unnecessarily large because excessive values can weaken mixed-integer optimization.

In practice, planners often use a hierarchical objective: first protect safety and regulatory feasibility, then meet contractual service requirements, and only then minimize cost or emissions. A weighted objective can be written as:

The coefficients α, β, γ, and δ express managerial priorities and must be calibrated carefully. Combining unlike measures requires normalization or monetary valuation; otherwise the weights have no interpretable scale.

2. Line-haul transport

Line-haul transport connects terminals, plants, ports, distribution centres, and major customers over relatively long distances. Compared with local delivery, it usually involves fewer stops, larger vehicles, planned departure times, and stronger economies of scale. The planning problem is often a combination of:

• shipment consolidation;
• origin–destination assignment;
• terminal selection;
• vehicle and driver scheduling;
• empty repositioning;
• appointment and dock coordination;
• disruption recovery.

A line-haul route may be a direct movement, a milk run serving several facilities, or a scheduled network with intermediate terminals. The best route is not necessarily the geographically shortest one. A slightly longer route may offer better road quality, lower tolls, fewer border delays, or improved backhaul availability.

Consolidation and empty mileage

Consolidation reduces cost per unit by combining compatible shipments. However, waiting for additional freight may increase inventory and delivery lead time. A planner therefore compares the marginal saving from consolidation with the cost of delay. Empty repositioning is especially important when freight flows are imbalanced. A vehicle returning empty creates cost without generating revenue, although repositioning may be necessary to serve future demand.

A basic utilization measure is:

For volume-constrained freight, volumetric utilization is more informative:

Weight and volume utilization should not be averaged indiscriminately; a vehicle can be weight-constrained on one route and cube-constrained on another.

Scheduling and reliability

Line-haul schedules include departure, arrival, loading, unloading, and buffer times. If a trip has planned duration T_p and uncertainty represented by standard deviation σ_T, a planner may set a buffer based on a selected reliability target. The exact buffer depends on the assumed distribution and service target; it should not be added mechanically. Historical travel-time percentiles are often more reliable than a normal-distribution assumption, especially when congestion and border delays produce skewed data.

3. Parcel delivery and the last mile

Parcel delivery involves a high number of small consignments, heterogeneous addresses, failed-delivery risk, and frequent changes after dispatch. The last mile is the segment from a local depot, store, or microhub to the recipient. It is often expensive because stops are dispersed, vehicles spend time parking and walking, and deliveries must be attempted individually.

Important decisions include depot-to-zone assignment, route sequencing, vehicle type, delivery promises, parcel locker placement, driver shift design, and customer communication. The cost of a stop is not only driving distance. It includes curb search, building access, elevator time, handover, proof of delivery, and possible reattempts.

A practical parcel-routing objective may include:

The stop cost is often estimated from observed service times rather than assumed to be constant. Apartment buildings, secure campuses, and rural properties may have very different service profiles.

Failed delivery and alternative delivery points

If a delivery attempt succeeds with probability p, and the cost of a successful first attempt is C_1 while a failed attempt incurs C_f and a later successful attempt incurs C_2, the expected cost of an initial attempt can be represented as:

The formula assumes two outcomes and known conditional costs. It can be extended to multiple attempts, lockers, collection points, or customer rescheduling. Delivery windows, locker availability, and recipient preferences can alter p and therefore change the economically preferred service design.

Parcel networks increasingly use lockers, pickup points, electric cargo bikes, and microhubs. These options reduce failed deliveries and vehicle access problems, but create handling transfers, capacity constraints, security requirements, and additional customer travel. Their value must therefore be assessed across the complete network rather than at the vehicle-route level alone.

4. Home delivery and attended services

Home delivery differs from parcel delivery when goods are bulky, valuable, fragile, temperature-sensitive, or installed in the home. Examples include furniture, appliances, groceries, and medical equipment. These operations may require two-person crews, specialized vehicles, assembly, removal of packaging, or customer signatures.

The planning problem is a vehicle-routing problem with rich service times and time windows. A customer may require delivery within [09:00, 11:00], a two-person crew, and a vehicle with sufficient dimensions. Feasibility therefore depends on skills and equipment as well as distance.

Customer time windows can improve service but often increase route fragmentation. Narrow windows reduce sequencing flexibility and may require additional vehicles. A sound practice is to offer differentiated windows: low-price flexible slots, premium narrow slots, or scheduled collection alternatives. This converts some operational complexity into a customer-choice and pricing decision.

For home delivery, planners should measure:

• on-time-in-full performance;
• first-attempt success;
• customer waiting time;
• damage and installation quality;
• route productivity;
• vehicle and crew utilization;
• cost per stop and cost per order;
• emissions per order.

A low cost per stop is not sufficient if damage, lateness, or repeat visits are high.

5. Service routing and technician scheduling

Service routing moves people rather than products, or combines travel with an on-site task. Examples include field maintenance, healthcare visits, meter reading, emergency repair, and inspection. The problem is commonly called a technician-routing and scheduling problem.

Distinctive constraints include skill compatibility, appointment windows, task precedence, spare-parts availability, overtime, and emergency insertion. A technician may be unable to perform a task without a certification or a particular tool. A route that is geographically efficient can therefore be infeasible.

Let z_ik equal 1 if technician k performs task i. A skill-compatibility condition can be represented as:

where e_ik = 1 if technician k is qualified and equipped for task i, and 0 otherwise. This is a feasibility condition, not a measure of performance.

Service planners also face dynamic arrivals. When an emergency request appears, the system may reoptimize only the affected routes rather than solve the entire problem from scratch. Dispatchers should preserve commitments already communicated to customers and record the operational reason for changes. In healthcare and public services, equity and urgency may take priority over travel-cost minimization.

6. Reverse logistics

Reverse logistics includes returns, repairs, reusable packaging, product recalls, recycling, and waste collection. The flow is often less predictable than forward distribution, and returned items may have different destinations depending on inspection results.

A return may be routed to resale, repair, refurbishment, recycling, liquidation, or disposal. Thus, routing is linked to disposition decisions. A vehicle may collect returns during a delivery tour, but this creates capacity, handling, contamination, and time-window constraints. For example, food waste, chemicals, and clean reusable packaging cannot necessarily share the same vehicle or compartment.

Reverse-routing decisions should evaluate total recovery value:

The expression is useful for comparing alternatives but must include quality uncertainty and environmental compliance costs where relevant. A return with positive material value may still be uneconomic if collection is highly dispersed.

7. Intermodal and multimodal freight

Intermodal transport uses two or more modes, such as road–rail, road–sea, or road–air, while maintaining the shipment or loading unit through transfers. Multimodal planning coordinates the entire movement, often under a single service arrangement, whether or not cargo is transferred between loading units.

The planner must select routes across a time-dependent network. An arc may represent a road movement, a train departure, a vessel sailing, a terminal handling operation, or a customs process. Each arc has cost, duration, capacity, reliability, and eligibility conditions.

A generalized path cost can be represented as:

Here c_a is the cost of arc a, C_transfer is handling and interchange cost, C_delay is the expected cost of lateness, and C_risk represents damage, loss, or disruption exposure. The path is feasible only if transfer arrival precedes the next departure and equipment, documentation, and capacity requirements are met.

Synchronization and terminal operations

A low-cost rail option is not attractive if a missed connection causes a 24-hour delay. Planners therefore evaluate schedule reliability, connection buffers, terminal congestion, and recovery alternatives. The relevant comparison is door-to-door performance, not mode-specific line-haul cost.

Modal choice also depends on shipment characteristics. Rail and water can offer lower energy use and emissions per tonne-kilometre, but may have longer transit times and less flexibility. Air offers speed at high cost and emissions. Road often provides the best door access and schedule flexibility. These are tendencies, not universal rules; actual performance depends on distance, network density, equipment, and load factor.

8. Cross-docking and urban consolidation

Cross-docking transfers inbound goods directly or nearly directly to outbound vehicles, minimizing storage. It is effective when demand, timing, labeling, and destination information are sufficiently reliable. The core planning problem is synchronization: inbound trailers must arrive early enough for sorting, but excessive early arrival creates congestion and staging inventory.

A cross-dock process includes receiving, scanning, sorting, temporary staging, loading, and dispatch. Operational controls include appointment scheduling, door assignment, wave planning, barcode or RFID accuracy, and exception management. Cross-docking is not automatically beneficial. It can increase handling and vulnerability when inbound variability is high or outbound destinations change frequently.

Urban consolidation centres combine shipments from several suppliers before delivery into a dense urban area. They can reduce the number of large vehicles entering restricted zones and support electric vehicles or cargo bikes. However, they add an upstream handling stage, require land and labour, and may shift rather than eliminate traffic. Their evaluation should consider total vehicle-kilometres, congestion, emissions, noise, delivery reliability, and financial viability.

9. Crowdsourced delivery

Crowdsourced delivery uses independent individuals or flexible contractors who accept delivery tasks, often through a digital platform. It can provide surge capacity and rapid service without owning a full fleet. It is particularly relevant to meals, groceries, urgent parcels, and peer-to-peer deliveries.

The platform must match tasks to couriers while considering location, capacity, acceptance probability, promised time, and legal restrictions. Unlike a fixed fleet, available capacity is endogenous: compensation, weather, competing platforms, and local events affect participation. Therefore, deterministic route plans can fail when couriers decline tasks or disappear from the market.

A responsible design addresses worker classification, pay transparency, insurance, safety, data protection, algorithmic fairness, and customer accountability. Surge pricing may balance demand and supply but can create inequitable access or unacceptable earnings volatility. Service quality metrics should distinguish platform-controlled failures from courier-controlled failures.

10. A detailed applied case: NorthCity omnichannel distribution

Background

NorthCity Retail, an illustrative regional retailer, operates one distribution centre, three urban stores, and home delivery in a dense metropolitan area. It receives 1,200 online orders per day. Forty percent of orders are small parcels, 45% are grocery or household baskets, and 15% are bulky items. The city imposes a daytime restriction on heavy vehicles in the central district.

Problem

The retailer uses separate routes for store replenishment, parcel delivery, and returns. Vehicles frequently enter the city partially loaded. Home-delivery failure is 12%, and returns are collected by a dedicated van that is full only near the end of its shift. Management wants to reduce cost without lowering the promised delivery rate of 95%.

Analysis

The first diagnostic separates demand by service requirement. Bulky orders need larger vehicles and sometimes two-person crews; groceries require temperature control; parcels can use lockers or cargo bikes; returns may be collected on delivery routes if capacity and hygiene rules permit. A single undifferentiated route model would incorrectly combine incompatible tasks.

The team maps order density, delivery windows, parking difficulty, failed-delivery probability, and urban access restrictions. It identifies a suitable microhub near the central district. Store replenishment is scheduled before the restricted period, while parcels and small baskets are transferred to electric cargo bikes and vans at the microhub.

Approach and solution

The planning system uses a rolling horizon. Each morning it creates a baseline plan; during the day it updates routes when orders are cancelled, traffic changes, or a vehicle becomes unavailable. The objective uses a high penalty for missed promised windows, a lower penalty for additional distance, and a separate cost for failed delivery.

Returns are offered during selected delivery windows. The system accepts a return on a route only when projected vehicle capacity, handling time, and product-compatibility conditions remain feasible. Customers in apartment districts are offered lockers or pickup points, increasing first-attempt success. Bulky deliveries remain with specialized crews.

Outcome

In the illustrative scenario, the redesign reduces overlapping urban vehicle movements, increases parcel first-attempt success, and improves load factors. It does not eliminate cost: the microhub introduces rent and handling labour, and electric cargo bikes require charging and secure storage. The business case is therefore strongest in dense zones with high order concentration, access restrictions, and sufficient daily volume.

Learning points

1. Segmentation is a prerequisite for valid routing.
2. Last-mile performance depends heavily on stop service time and failed delivery, not only kilometres.
3. Consolidation can reduce urban traffic while adding handling cost.
4. Returns should be designed with forward delivery rather than treated as an unrelated activity.
5. Rolling-horizon planning is more suitable than a fixed daily plan when conditions change.
6. Performance must be assessed at network level using cost, service, capacity, and environmental measures.

11. Technology, analytics, and implementation

Modern systems combine transport-management software, geographic information systems, telematics, barcode scanning, customer applications, and optimization engines. Artificial intelligence can support travel-time prediction, demand forecasting, address correction, stop-time estimation, and anomaly detection. It does not remove the need for operational judgment: a prediction can be accurate on average but harmful if it systematically underestimates delays in particular neighbourhoods.

Implementation should proceed through data governance and pilot testing. Essential data include accurate addresses, road restrictions, vehicle dimensions, customer windows, service durations, skills, depot calendars, and historical exceptions. Planners should monitor data drift, version model changes, and preserve an override process for safety-critical decisions.

A useful deployment sequence is:

1. establish clean master data and service definitions;
2. define hard constraints and non-negotiable regulatory rules;
3. select objective measures and baseline performance;
4. test algorithms against historical and extreme scenarios;
5. run a controlled pilot;
6. compare predicted and actual performance;
7. train dispatchers and drivers;
8. scale gradually with audit controls.

12. Challenges, ethics, and governance

Applied routing is constrained by more than mathematics. Common limitations include inaccurate travel times, incomplete demand, unstable customer availability, poor address quality, and hidden operational rules. Optimization can also produce socially undesirable outcomes if its objective excludes worker welfare, accessibility, emissions, or neighbourhood impacts.

Relevant governance issues include driver hours, vehicle and load safety, dangerous-goods rules, customs compliance, urban access permits, privacy, employment law, and accessibility. Location and delivery data can reveal household behaviour; collection and retention should therefore follow data-minimization and security principles.

Algorithmic systems should be explainable enough for dispatchers and affected workers to understand important decisions. Performance should be checked for geographic or demographic bias, including systematic service exclusion or excessive delivery burdens in particular areas. Sustainability claims should use consistent boundaries: emissions per parcel, per tonne-kilometre, or per successful delivery can produce different conclusions.

Best-practice framework

A robust applied-routing programme follows these principles:

• design around the complete process, not only vehicle travel;
• separate hard feasibility constraints from soft preferences;
• segment freight by service, compatibility, and urgency;
• use realistic service-time and failure data;
• plan buffers from empirical reliability evidence;
• evaluate consolidation with its added handling and inventory effects;
• use rolling-horizon reoptimization for dynamic operations;
• include reverse flows and empty movements;
• measure total cost, service, emissions, safety, and equity;
• retain human accountability for exceptions and high-impact decisions;
• pilot before scaling and compare results with a credible baseline.

Chapter Summary

Freight and delivery applications require different adaptations of route planning. Line-haul planning emphasizes consolidation, schedules, terminal coordination, and empty repositioning. Parcel and last-mile operations are shaped by stop times, access constraints, delivery failure, lockers, microhubs, and alternative vehicles. Home delivery adds bulky goods, crews, installation, and customer time windows. Service routing combines travel with skills, tools, task duration, and emergency insertion. Reverse logistics links collection with inspection, disposition, and recovery value.

Intermodal planning treats modes and terminals as a connected, time-dependent network in which transfer reliability can be as important as line-haul price. Cross-docking and urban consolidation can reduce storage or urban traffic but introduce synchronization and handling costs. Crowdsourced delivery provides flexibility but raises uncertainty, labour, safety, privacy, and accountability concerns. Across all applications, successful systems combine optimization with reliable data, operational governance, human oversight, and network-level evaluation.

Key Terms

Line-haul transport: Long-distance movement between major terminals or facilities.

Last mile: The final delivery segment to the recipient or collection point.

Capacitated vehicle-routing problem: Routing problem in which vehicle load cannot exceed capacity.

Time window: Earliest and latest permissible service times.

Stop time: Time spent serving a customer beyond travel time.

First-attempt success: Completion of delivery during the initial attempt.

Home delivery: Delivery to a residence, often with specialized service requirements.

Service routing: Routing that combines travel with on-site tasks.

Skill compatibility: Requirement that a worker possess specified qualifications or equipment.

Reverse logistics: Movement and management of returns, repairs, recovery, or disposal.

Intermodal transport: Use of multiple transport modes, commonly with a standardized loading unit.

Cross-docking: Transfer from inbound to outbound vehicles with minimal storage.

Urban consolidation centre: Facility that combines shipments for coordinated urban delivery.

Microhub: Small local transfer or staging facility close to a delivery zone.

Empty mileage: Vehicle travel without revenue-generating freight.

Generalized cost: Monetary or weighted representation of financial and operational impacts.

Rolling-horizon planning: Repeated replanning as new information becomes available.

Crowdsourced delivery: Delivery capacity supplied by independent or platform-connected couriers.

Review Questions

5 Short Answer Questions

1. Why is the shortest route not necessarily the best line-haul route?
2. Identify three factors that make last-mile delivery costly.
3. What is the purpose of a time-window constraint?
4. Why must reverse logistics include disposition decisions?
5. State two advantages and two limitations of cross-docking.

5 Long Answer Questions

1. Explain how a parcel-delivery routing model differs from a home-delivery model.
2. Discuss the principal synchronization problems in intermodal freight planning.
3. Explain how skills, tools, and emergency requests affect service routing.
4. Evaluate the strategic role of urban consolidation centres.
5. Discuss governance requirements for AI-supported delivery routing.

3 Analytical / Application Questions

1. A carrier serves dense apartments and dispersed suburban homes. Design a segmented last-mile operating model and identify the performance measures needed to evaluate it.
2. A retailer is considering collecting returns on forward-delivery routes. Specify the feasibility constraints and explain when dedicated reverse routes would remain preferable.
3. Compare road-only and road–rail alternatives for a shipment with a strict delivery promise. Identify the cost, reliability, transfer, emissions, and disruption variables that should enter the decision.

Multiple Choice Questions with Answers

Question 1

Which feature most strongly distinguishes line-haul planning from typical last-mile planning?

A. A small number of high-volume movements between terminals

B. A requirement to visit every household in a neighbourhood

C. A dominant emphasis on customer handover time at each stop

D. The absence of scheduled departure times

Correct Answer: A. A small number of high-volume movements between terminals

Question 2

If vehicle k serves customers whose demands total 8,000 kg and its capacity is 10,000 kg, what is its weight utilization?

A. 20%

B. 50%

C. 80%

D. 125%

Correct Answer: C. 80%

Question 3

Which condition is represented by a customer time window [a_i, b_i]?

A. Vehicle capacity must equal customer demand

B. Service must begin between the earliest and latest permitted times

C. Travel time must be zero inside the window

D. The customer must be visited twice

Correct Answer: B. Service must begin between the earliest and latest permitted times

Question 4

Which factor is especially important in home delivery of bulky goods?

A. Only parcel barcode density

B. Crew and equipment requirements

C. Elimination of all service time variation

D. Absence of customer appointments

Correct Answer: B. Crew and equipment requirements

Question 5

What is a central purpose of cross-docking?

A. To maximize long-term storage inventory

B. To transfer inbound goods to outbound vehicles with minimal storage

C. To eliminate the need for shipment information

D. To ensure every shipment uses air transport

Correct Answer: B. To transfer inbound goods to outbound vehicles with minimal storage

Question 6

In intermodal planning, why are transfer buffers important?

A. They guarantee that all modes have identical costs

B. They protect connections against arrival and handling variability

C. They eliminate terminal capacity constraints

D. They make documentation unnecessary

Correct Answer: B. They protect connections against arrival and handling variability

Question 7

Which statement best describes reverse logistics?

A. It concerns only empty vehicle repositioning

B. It includes returns, repair, recovery, recycling, and disposal flows

C. It excludes inspection and disposition decisions

D. It always costs less than forward distribution

Correct Answer: B. It includes returns, repair, recovery, recycling, and disposal flows

Question 8

Which constraint is specific to technician-routing applications?

A. A technician must be qualified for the assigned task

B. Every customer must receive an identical product

C. All routes must use rail

D. Vehicles cannot have scheduled working hours

Correct Answer: A. A technician must be qualified for the assigned task

Question 9

A failed first delivery has probability 0.20, and a successful first delivery has probability 0.80. If the first-attempt cost is 4 units, while failure and subsequent completion cost 14 units, what is expected cost?

A. 6 units

B. 8 units

C. 10 units

D. 14 units

Correct Answer: B. 8 units

Question 10

Which is a major governance concern in crowdsourced delivery?

A. The complete disappearance of demand uncertainty

B. Worker classification, safety, pay transparency, and accountability

C. Guaranteed availability of unlimited couriers

D. The impossibility of using digital platforms

Correct Answer: B. Worker classification, safety, pay transparency, and accountability