Transport Demand, Capacity and Network Planning

Introduction

Fleet operations are shaped by a fundamental managerial problem: transportation demand varies across customers, locations, time periods and service commitments, while vehicle, driver, terminal and road capacity are finite. Effective planning therefore requires more than assigning vehicles to orders. It requires the integration of demand forecasting, capacity design, load consolidation, geographic segmentation and network configuration.

A transport network may use direct delivery, milk runs, regional depots, cross-docks, hubs or combinations of these structures. The appropriate design depends on shipment density, delivery urgency, geography, product characteristics, operating cost, service requirements and uncertainty. Poor planning produces empty kilometres, low vehicle utilisation, congestion, missed delivery windows and excessive fixed investment. Good planning creates sufficient capacity while preserving flexibility and resilience.

This chapter develops a managerial and analytical framework for forecasting shipment demand, balancing capacity, consolidating loads and designing depots, territories, lanes and hub-and-spoke or direct-delivery networks.

Learning Objectives

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

1. Distinguish transport demand from derived demand and explain its principal drivers.
2. Construct and evaluate shipment-demand forecasts under uncertainty.
3. Calculate capacity, utilisation, payload efficiency and required fleet size.
4. Explain consolidation decisions and identify their cost and service trade-offs.
5. Design territories, lanes and depot locations using operational criteria.
6. Compare direct-delivery, hub-and-spoke and hybrid networks.
7. Apply scenario analysis, safety capacity and resilience principles to network planning.
8. Evaluate the managerial, technological, ethical and environmental implications of network choices.

1. Transport Demand as Derived Demand

Transport demand is the quantity of movement requested between origins and destinations over a specified period, expressed in shipments, tonnes, pallets, cubic metres, vehicle trips or revenue units. It is usually derived demand: firms do not seek transport for its own sake but because goods must be positioned to support production, trade, retailing or consumption.

Demand has four essential dimensions:

• Volume: quantity moved, such as tonnes or pallets.
• Spatial distribution: origins, destinations and lane flows.
• Temporal distribution: hourly, daily, weekly and seasonal patterns.
• Service quality: required speed, reliability, temperature control, visibility and delivery-window compliance.

A forecast that predicts only total tonnes is insufficient for fleet planning. Two networks with identical annual tonnage may need very different fleets if one has concentrated full-truckload flows and the other has dispersed, time-sensitive small shipments.

Important demand drivers include industrial output, sales forecasts, population and income, e-commerce activity, inventory policy, product mix, fuel prices, trade restrictions, weather, promotions and customer service policies. Demand is also affected by logistics decisions. For example, a high-frequency replenishment policy raises transport activity even when annual product consumption is unchanged.

Demand observations and units

A transport-planning database should normally contain order date, promised date, origin, destination, commodity, weight, volume, handling-unit count, service class, vehicle requirements and actual delivery outcome. Forecasts should be prepared at a level that is operationally useful. Forecasting national annual tonnage is appropriate for strategic fleet investment, whereas forecasting daily pallet demand by depot and delivery territory is needed for dispatch planning.

2. Forecasting Shipment Demand

2.1 Forecasting hierarchy

A useful hierarchy is:

1. Strategic forecast: one to five years; supports depot, fleet and contract decisions.
2. Tactical forecast: one to twelve months; supports capacity procurement, staffing and lane planning.
3. Operational forecast: hours to several weeks; supports vehicle assignment, routing and dock scheduling.

Forecasts should be reconciled across levels. The sum of depot or lane forecasts should be consistent with the regional and corporate forecast unless a documented adjustment is made.

2.2 Basic forecasting methods

The moving average for the next period using the most recent n observations is:

where D is observed demand and F is forecast demand. It is simple and useful when demand is relatively stable, but it reacts slowly to turning points.

The weighted moving average is:

where wᵢ are non-negative weights. Larger weights can be assigned to recent observations.

Simple exponential smoothing is:

The smoothing constant α controls responsiveness. A higher α reacts more strongly to recent changes. This method is appropriate for level demand without a systematic trend or seasonality.

For seasonal demand, a decomposition approach may represent demand as:

where T is the underlying trend, S is the seasonal factor and E is an irregular factor. An additive form may be more appropriate when seasonal variation is approximately constant in absolute units. The model must be selected from data rather than assumed universally.

Causal models may relate shipments to explanatory variables such as sales, production, price or economic activity. A simple linear specification is:

where β coefficients measure estimated relationships and ε represents unexplained variation. Correlation alone does not establish causation; model validation and business interpretation are essential.

2.3 Forecast error and uncertainty

Forecast error for period t can be defined as:

Mean absolute deviation is:

Root mean squared error is:

MAD is expressed in demand units and is relatively interpretable. RMSE penalises large errors more heavily. Percentage measures such as MAPE can become unstable when actual demand is zero or very small; therefore, they should not be used without examining the data.

Forecasts should be accompanied by ranges or scenarios. A point forecast is not a guarantee. For capacity planning, the relevant question is often: “What capacity is required to achieve a specified service probability?”

3. Capacity Planning and Fleet Sizing

3.1 Types of capacity

Transport capacity includes more than vehicle payload. It consists of:

• payload mass, measured in tonnes;
• loading volume, measured in cubic metres or pallet positions;
• vehicle and driver availability;
• operating hours and legal driving limits;
• loading-dock and terminal throughput;
• road and lane capacity;
• specialist equipment, such as refrigeration or lifting devices.

The effective capacity of a vehicle is constrained by the tightest relevant limit. A truck may reach its cubic-volume limit before its legal mass limit, particularly when carrying low-density products.

3.2 Utilisation measures

Payload utilisation is:

Volume utilisation is:

Distance utilisation, often called load-factor utilisation, is:

These measures must not be confused. A vehicle can have high payload utilisation but poor distance utilisation if it returns empty. A network should therefore track both loaded capacity and empty repositioning.

3.3 Required fleet capacity

If planning demand is expressed in tonnes per period, a simplified fleet requirement is:

where N is required vehicle equivalents, D is demand per period, q is rated payload per vehicle, r is the number of productive trips per vehicle per period, and u is the planned utilisation factor. The result should be rounded upward when whole vehicles are required.

This formula assumes comparable trips, stable payload requirements and a utilisation factor that captures realistic inefficiency. It is not suitable by itself where route length, delivery windows, driver regulations, vehicle classes or multi-compartment requirements vary substantially.

Worked example. A depot must move 2,400 tonnes per month. Each vehicle can carry 10 tonnes and complete 18 productive trips per month. Planned payload utilisation is 80%.

Effective monthly capacity per vehicle is:

Thus:

The depot requires 17 vehicle equivalents before considering maintenance spares, demand peaks or specialist vehicles. If a 10% operational reserve is required, the planning fleet is approximately 18.7, which should normally be rounded to 19 vehicles, subject to the organisation’s reserve policy.

3.4 Capacity buffers and safety capacity

A capacity buffer protects service when demand, travel time or vehicle availability is uncertain. A simple planning expression is:

The buffer rate should be based on historical variability, supplier reliability, labour conditions, weather and the cost of failure. Excessive buffers create idle assets; inadequate buffers create premium outsourcing, late deliveries and lost sales.

Where demand uncertainty is modelled statistically, safety capacity may be expressed as:

where SC is safety capacity, z is the selected service-factor quantile, and σ_L is the standard deviation of demand over the relevant planning lead time. This approximation assumes a reasonably stable distribution and a clearly defined service target. It should not be applied mechanically to highly intermittent or non-normal demand.

4. Load Consolidation and Shipment Design

Consolidation combines compatible shipments so that transport capacity is used more efficiently. It may involve combining orders for one customer, multiple customers on a route, or multiple suppliers moving toward a common destination.

The principal economic effect is the spreading of fixed trip costs over more freight. If a vehicle trip costs C, the average transport cost per shipment generally falls as compatible shipment quantity increases, although handling and waiting costs may rise.

Consolidation decisions should evaluate:

• compatibility of commodities and packaging;
• delivery-window overlap;
• route geometry and detour distance;
• product security and damage risk;
• temperature or hazardous-material restrictions;
• inventory and customer waiting costs;
• terminal handling and transshipment cost;
• vehicle payload and volume constraints.

A useful decision rule is to consolidate when the savings in transport and handling exceed the additional inventory, delay and service-failure costs. The decision is therefore not simply “fill the truck.” A full vehicle that arrives late may be less valuable than a partially filled vehicle that protects a critical delivery commitment.

Consolidation may be:

• Temporal: hold orders until a scheduled departure.
• Spatial: combine shipments sharing an origin, destination or corridor.
• Functional: use a cross-dock to transfer inbound shipments directly to outbound routes.
• Contractual: combine freight from different shippers through a logistics service provider.

The trade-off can be represented conceptually as:

The expression is a cost identity for decision analysis, not a universal optimisation formula. Each component must be measured consistently over the same time period and scope.

5. Territories, Lanes and Depot Planning

5.1 Territory design

A territory is a defined geographic or customer-service area assigned to a depot, fleet, sales region or operating team. Good territory design balances workload, travel time, geographic coherence and service requirements.

Important design criteria include:

• comparable demand and stop density;
• maximum route duration;
• road accessibility and congestion;
• customer time windows;
• driver familiarity and labour rules;
• emergency coverage and backup options;
• fairness of workload between territories.

Territories should not be designed solely by drawing equal-area polygons. Demand is usually spatially uneven, and a small urban territory may generate more work than a large rural territory.

5.2 Lane planning

A lane is a recurring origin–destination movement. Lane analysis should identify shipment volume, frequency, equipment type, directionality, transit time, variability, cost, empty-return rate and strategic importance.

High-density, stable lanes may justify dedicated vehicles or contracted capacity. Irregular lanes may be served through a spot market, consolidation service or shared network. Directional imbalance is important: a lane with strong outbound demand but little return freight may appear profitable when only outbound cost is examined but generate substantial empty kilometres.

5.3 Depot location and allocation

Depot planning balances transport cost, facility cost, labour availability, land cost, customer proximity, risk and future expansion. A simplified centre-of-gravity estimate for a single facility is:

where (xᵢ, yᵢ) are customer or demand-point coordinates and wᵢ is a weight such as shipment volume. This is a screening method, not a final location decision. It ignores road networks, congestion, land constraints, taxes, zoning, service windows and facility economies of scale.

More advanced models use facility-location optimisation. A common capacitated formulation minimises fixed facility costs and allocation costs subject to demand satisfaction and facility capacity. Such models can support strategic analysis, but the output depends on data quality and assumptions about future demand.

6. Network Configuration Choices

6.1 Direct delivery

In a direct-delivery network, freight moves from origin to destination without an intermediate hub. Advantages include fewer handling events, short transit time, lower damage exposure and simple shipment visibility. Direct delivery is attractive when shipment volumes are large, destinations are concentrated or service is urgent.

Its disadvantages include low vehicle utilisation for dispersed demand, many origin–destination movements and higher coordination complexity when volumes are small.

6.2 Hub-and-spoke

In a hub-and-spoke network, shipments move from origins to a consolidation hub and then to destinations. The hub enables pooling and high-frequency line-haul movements. It can reduce the number of direct links and improve vehicle utilisation.

However, hubs introduce handling, sorting, transfer cost, schedule dependence and a potential single point of failure. They are most effective when there is sufficient flow density and predictable timing.

6.3 Hybrid networks

Most practical networks are hybrid. Large customers may receive direct full-load deliveries, while small customers are served through regional hubs or cross-docks. A hybrid design allows differentiated service by lane, product and customer value.

Network selection should compare total cost and service, not merely the number of kilometres. Relevant measures include total transport cost, cost per shipment, average and percentile transit time, on-time performance, damage rate, empty kilometres, CO₂ emissions, resilience and capital investment.

Figure: Direct delivery uses origin–destination links; hub-and-spoke routes flows through a consolidation point.

7. Planning Under Uncertainty and Resilience

Network plans should be tested against plausible scenarios rather than a single forecast. Useful scenarios include demand growth, customer loss, fuel-price increases, driver shortages, severe weather, facility closure, border delay and supplier disruption.

Three complementary approaches are useful:

• Sensitivity analysis: change one assumption at a time to identify influential variables.
• Scenario analysis: evaluate coherent combinations of changes.
• Stochastic or robust optimisation: incorporate probability distributions or seek solutions that perform acceptably across uncertain conditions.

Resilience measures include alternative depots, cross-trained staff, backup carriers, flexible contracts, pre-approved diversion routes, spare vehicles and distributed inventory. Resilience has a cost, so management should compare the expected cost of disruption with the cost of preparedness.

A network can be efficient but fragile. For example, a single highly automated hub may minimise normal operating cost while creating severe consequences if it fails. The appropriate design depends on the organisation’s risk tolerance and the criticality of the goods.

8. Technology, Data and Artificial Intelligence

Transport-management systems integrate orders, forecasts, fleet availability, carrier capacity, geospatial data and delivery status. Geographic information systems support territory and depot analysis. Telematics provide actual travel time, vehicle location, fuel use and driving behaviour.

Machine-learning methods can improve demand forecasts where large, high-quality datasets contain nonlinear relationships, promotions or complex seasonality. They should not replace managerial judgement. Forecast monitoring must detect data drift, structural market changes and biased historical patterns. Explainability is especially important when algorithmic recommendations affect carrier allocation, driver workload or customer service.

Digital twins and simulation can test alternative depot locations, dispatch rules and disruption scenarios before implementation. Data governance should define ownership, access rights, retention, cybersecurity and quality controls. Personally identifiable driver or customer data must be handled lawfully and proportionately.

9. Governance, Ethics and Sustainability

Network planning has consequences beyond cost. Territory design can affect driver fatigue, workload equity and exposure to unsafe routes. Capacity plans must comply with driving-hour regulations, vehicle weight limits, dangerous-goods rules, employment law and environmental requirements.

Ethical planning avoids shifting unreasonable risk to subcontractors or using unrealistic delivery promises to conceal inadequate capacity. Procurement decisions should consider labour standards, safety records and environmental performance, not only price.

Sustainability analysis should include vehicle manufacture, fuel or electricity, empty kilometres, depot energy and construction, packaging and failed-delivery activity. A simplified emissions estimate is:

where distance must be measured in compatible units and the emission factor may be expressed as kilograms of CO₂-equivalent per vehicle-kilometre or per tonne-kilometre. The boundary and factor source must be documented. Consolidation can reduce emissions per shipment, but additional handling or detours may offset the benefit.

Applied Case: NorthRiver Consumer Products

Background

NorthRiver Consumer Products, an illustrative regional distributor, supplies 420 retail outlets from three factories. Its annual demand is growing, but delivery performance has fallen. The company operates a central depot, sends many partially filled vehicles directly to stores and frequently hires premium same-day transport.

Problem

Analysis shows that demand is geographically dispersed, with high store density in two metropolitan areas and lower density in rural zones. Average payload utilisation is 61%, while 28% of vehicle-kilometres are empty or repositioning kilometres. Forecasts are prepared only at annual regional level, so weekly peaks are not visible to fleet planners.

Analysis

The planning team disaggregates demand by depot, territory, weekday and product family. It identifies three stable high-volume lanes suitable for contracted full-load service. Small metropolitan orders are suitable for cross-dock consolidation, while rural stores require direct delivery because detours through a hub would increase transit time substantially.

A twelve-week rolling forecast is introduced using recent demand, promotional calendars and seasonal factors. Forecast error is measured with MAD, and peak capacity is planned using a documented service buffer. The team also maps delivery windows and uses road travel-time data rather than straight-line distance for territory design.

Approach and solution

NorthRiver adopts a hybrid network:

1. Direct full-load service is retained for major retail customers and stable factory-to-region lanes.
2. A metropolitan cross-dock consolidates small shipments into scheduled delivery waves.
3. Rural territories remain primarily direct-delivery territories.
4. A contracted-carrier panel provides flexible peak capacity, with minimum safety and service standards.
5. Two vehicle equivalents are retained as a controlled reserve rather than relying entirely on emergency hiring.
6. Forecasts are reviewed weekly, while depot and fleet decisions are reviewed quarterly.

Outcome

In the illustrative evaluation, average payload utilisation rises from 61% to 78%, empty kilometres fall, and premium same-day purchases decline. The cross-dock increases handling cost, but the reduction in fragmented trips more than offsets it. Service performance improves because delivery waves are aligned with store windows. The company does not close its central depot immediately; instead, it stages the change and monitors demand stability before making a capital decision.

Learning points

The case demonstrates that network improvement is not equivalent to centralisation. The best design differentiates flows. Forecast granularity, lane economics, time windows and resilience must be considered together. A hybrid network can outperform both an all-direct and an all-hub design when demand characteristics vary across customers and geography.

Challenges and Limitations

Forecasts are vulnerable to structural breaks, promotions, new customers and poor master data. Capacity formulas can conceal bottlenecks when trips differ in distance, service time or legal constraints. Utilisation percentages may look favourable while service quality deteriorates. Location models may produce mathematically attractive sites that are infeasible because of zoning, labour or road access. Hub networks may reduce line-haul cost while increasing handling and disruption exposure.

Managers should therefore combine quantitative models with operational validation, stakeholder consultation and pilot implementation. Assumptions, data limitations and sensitivity results should be documented so that decisions can be audited and revised.

Best-Practice Framework

A robust planning process follows these steps:

1. Define service objectives, planning horizon and decision scope.
2. Clean and segment shipment data by lane, territory, product and time.
3. Develop reconciled forecasts with uncertainty ranges.
4. Measure vehicle, driver, terminal and lane capacity separately.
5. Identify consolidation opportunities and service constraints.
6. Generate alternative territory, depot and network designs.
7. Evaluate cost, service, risk, emissions and capital requirements.
8. Stress-test alternatives under disruption scenarios.
9. Pilot the preferred design and monitor operational outcomes.
10. Establish a review cycle for forecast error, utilisation and network performance.

Chapter Summary

Transport demand is derived from economic and supply-chain activity and must be understood by volume, geography, time and service requirement. Forecasting methods range from moving averages and exponential smoothing to seasonal and causal models; all require error measurement and judgement. Fleet capacity is constrained by payload, volume, drivers, hours, equipment and terminals. Utilisation must be measured across payload, volume and distance dimensions.

Load consolidation lowers fragmented transport cost when compatibility and service conditions permit, but it can increase handling, waiting and inventory costs. Territory, lane and depot planning should reflect demand density, travel time, time windows, infrastructure and resilience rather than geographic area alone. Direct, hub-and-spoke and hybrid networks each have suitable applications. Under uncertainty, scenario analysis, safety capacity and alternative resources are essential. Technology can improve visibility and forecasting, but governance, privacy, safety, labour standards and environmental impacts must remain central to professional decision-making.

Key Terms

Transport demand: The required movement of goods between locations over time.

Derived demand: Demand for transport created by another economic or supply-chain activity.

Forecast hierarchy: Strategic, tactical and operational forecasts linked across planning levels.

Moving average: A forecast based on the average of recent observations.

Exponential smoothing: A forecasting method that weights recent observations more heavily.

Forecast error: The difference between actual and forecast demand.

Capacity: The maximum transport work possible under specified resource and operating constraints.

Payload utilisation: Actual payload divided by rated payload.

Distance utilisation: Loaded vehicle-kilometres divided by total vehicle-kilometres.

Safety capacity: Additional capacity held to protect a selected service level.

Load consolidation: Combining compatible shipments into a larger movement.

Territory: A defined geographic or customer-service area assigned for operations.

Lane: A recurring origin–destination movement.

Depot: A facility used for storage, dispatch, transshipment or fleet support.

Cross-dock: A facility where inbound goods are rapidly transferred to outbound vehicles with limited storage.

Hub-and-spoke network: A network that routes flows through a central or regional consolidation point.

Direct delivery: Movement from origin to destination without an intermediate hub.

Empty kilometres: Vehicle-kilometres travelled without revenue or productive freight.

Network resilience: The ability to maintain or recover service after disruption.

Demand variability: Fluctuation in shipment requirements across time or locations.

Review Questions

Short Answer Questions

1. Why is transport demand described as derived demand?
2. Distinguish payload utilisation from distance utilisation.
3. State two advantages and two disadvantages of hub-and-spoke networks.
4. What information should a lane analysis contain?
5. Why should forecast error be monitored alongside a point forecast?

Long Answer Questions

1. Explain how demand forecasting, fleet sizing and capacity buffers should be integrated in tactical transport planning.
2. Compare direct delivery, hub-and-spoke and hybrid networks using cost, service and resilience criteria.
3. Discuss the operational and strategic considerations involved in designing delivery territories.
4. Explain the economic and service trade-offs involved in load consolidation.
5. Evaluate the role and limitations of technology and artificial intelligence in transport network planning.

Analytical / Application Questions

1. A depot must move 3,600 tonnes per month. Each truck carries 12 tonnes, completes 15 productive trips per month and is planned to operate at 75% payload utilisation. Calculate the required vehicle equivalents and explain why the result may need adjustment.
2. A firm is considering a hub for dispersed small shipments. Identify the data required to compare the hub option with direct delivery, and propose at least three disruption scenarios for evaluation.
3. A network has high outbound vehicle utilisation but frequent empty returns. Diagnose likely causes and recommend a lane, territory or consolidation intervention, explaining the measures you would monitor.

Multiple Choice Questions with Answers

Question 1

Transport demand is called derived demand because it is primarily created by:

A. The desire to own vehicles

B. The need to support production, trade or consumption

C. Government vehicle registration

D. Driver availability

Correct Answer: B. The need to support production, trade or consumption

Question 2

A vehicle has a rated payload of 10 tonnes and carries 8 tonnes. Its payload utilisation is:

A. 20%

B. 80%

C. 100%

D. 125%

Correct Answer: B. 80%

Question 3

In simple exponential smoothing, a higher α generally makes the forecast:

A. Less responsive to recent demand

B. More responsive to recent demand

C. Independent of actual demand

D. Equal to the annual average

Correct Answer: B. More responsive to recent demand

Question 4

Which network normally includes an intermediate consolidation point?

A. Direct-delivery network

B. Hub-and-spoke network

C. Single-customer route

D. Empty repositioning route

Correct Answer: B. Hub-and-spoke network

Question 5

A depot must move 1,440 tonnes monthly. Each vehicle provides 120 tonnes of effective monthly capacity. Required vehicle equivalents are:

A. 10

B. 12

C. 14

D. 17

Correct Answer: B. 12

Question 6

Which measure captures the proportion of vehicle-kilometres travelled with freight?

A. Payload utilisation

B. Volume utilisation

C. Distance utilisation

D. Driver utilisation

Correct Answer: C. Distance utilisation

Question 7

A centre-of-gravity location calculation primarily uses:

A. Weighted demand-point coordinates

B. Driver age and tenure

C. Vehicle colour and type

D. Customer satisfaction scores only

Correct Answer: A. Weighted demand-point coordinates

Question 8

A major risk of excessive load consolidation is:

A. Lower vehicle utilisation in every case

B. Increased waiting or service delay

C. Elimination of handling activity

D. Guaranteed lower inventory cost

Correct Answer: B. Increased waiting or service delay

Question 9

Which is an appropriate use of scenario analysis?

A. Testing only the most likely forecast

B. Evaluating coherent combinations of demand and disruption changes

C. Replacing all operational data with intuition

D. Ignoring uncertainty in capital decisions

Correct Answer: B. Evaluating coherent combinations of demand and disruption changes

Question 10

The main purpose of safety capacity is to:

A. Guarantee zero operating cost

B. Protect service against demand or operating uncertainty

C. Eliminate all subcontracting

D. Maximise empty kilometres

Correct Answer: B. Protect service against demand or operating uncertainty