Transportation Demand and Service Planning

Introduction

Transportation demand and service planning convert uncertain customer requirements into operational decisions about shipment frequency, routing, capacity, lead time and service reliability. In materials management, transportation is not merely a movement activity. It is a coordinating mechanism linking purchasing, production, inventory, warehousing, distribution and customer service.

A transportation plan must answer five related questions:

1. How much freight will be required, where and when?
2. What characteristics distinguish one shipment from another?
3. What service level does each customer or material category require?
4. What transport capacity and frequency are economically justified?
5. How should cost, responsiveness, resilience and environmental performance be balanced?

Demand planning is difficult because freight demand is derived from business activity. It depends on sales, production schedules, sourcing decisions, inventory policies, promotions, seasonality, network design and disruptions. Consequently, a sound plan combines quantitative forecasting with managerial judgement and continuous performance monitoring.

Learning Objectives

After studying this chapter, the learner should be able to:

• explain the relationship between material flows, freight demand and transportation supply;
• forecast transportation demand using appropriate time-series and causal methods;
• segment shipments according to service, cost and physical characteristics;
• translate service requirements into frequency, lead-time, capacity and reliability decisions;
• calculate basic capacity, safety-capacity and shipment-consolidation measures;
• evaluate trade-offs among transport cost, inventory, responsiveness and risk; and
• design a practical transportation planning process supported by digital data and governance.

1. Transportation Demand as a Derived Demand

Transportation demand is the quantity of transport service required to move goods between origins and destinations over a specified period. Unlike demand for a final consumer product, it is usually a derived demand: it arises because materials must be purchased, transformed, stored or delivered.

The fundamental freight-demand unit should be selected carefully. Possible units include tonnes, pallets, cases, cubic metres, vehicleloads, shipments, shipment-kilometres and tonne-kilometres. These measures are not interchangeable. A high-value, low-density product may require considerable vehicle space but little weight capacity. Conversely, bulk minerals may exhaust weight capacity before vehicle volume is filled.

For shipment i, freight work may be represented as:

where W_i is transport work in tonne-kilometres, q_i is shipment mass in tonnes and d_i is distance in kilometres. This measure assumes a direct movement and does not include empty repositioning, detours or handling. Network-level transport work should account for actual route distance and, where relevant, empty running.

Transportation demand has four dimensions:

• Volume: quantity to be moved.
• Time: required dispatch, arrival and delivery windows.
• Spatial pattern: origins, destinations and network lanes.
• Service attributes: reliability, visibility, security, temperature control and flexibility.

A monthly volume forecast alone is therefore inadequate. A planner must also forecast lane-level demand, shipment-size distribution, daily or weekly peaks and the service class attached to each flow.

2. Demand Planning Architecture

A useful planning architecture separates strategic, tactical and operational horizons.

The planning cycle generally follows this sequence:

1. collect demand, inventory and order data;
2. clean and reconcile the data;
3. forecast volume by lane and time period;
4. classify shipments and service requirements;
5. convert volume into loads, equipment and capacity requirements;
6. design frequency, consolidation and routing policies;
7. test costs, service and risk under alternative scenarios;
8. execute through transport-management systems; and
9. compare actual outcomes with the plan and revise assumptions.

A key principle is that forecasts should be generated at the level at which decisions are made. A national forecast may support strategic fleet planning, but it cannot by itself determine the number of daily vehicles required on a particular lane.

3. Forecasting Freight Demand

3.1 Data requirements and data quality

Important data fields include order date, requested delivery date, actual dispatch and receipt dates, origin, destination, product, quantity, weight, cube, packaging unit, transport mode, carrier, cost, service class and exception reason. Data should be checked for duplicated orders, cancelled shipments, missing distances, inconsistent units, transshipments counted twice and abnormal movements caused by one-time projects.

Forecast accuracy should be assessed against a clearly defined baseline and horizon. A forecast of monthly tonnes may be accurate while the daily vehicle requirement remains poor because shipment timing is highly variable.

3.2 Time-series methods

A naïve forecast uses the most recent observation:

where F_{t+1} is the forecast for the next period and A_t is actual demand in the current period. It is useful as a benchmark, especially for stable demand.

A moving average of order n is:

All observations in this basic form receive equal weight. It smooths random variation but can respond slowly to a trend.

An exponential-smoothing forecast is:

where α is the smoothing constant, 0 < α ≤ 1. A larger α reacts more rapidly to recent changes; a smaller α produces greater smoothing. This method assumes that the underlying process can be represented reasonably by a changing level and that errors are not strongly autocorrelated.

When systematic trend exists, a trend-adjusted method such as Holt’s method may be used. With level l_t and trend b_t:

Here α and β are smoothing constants, h is the forecast horizon, and the model assumes a locally linear trend. Seasonal freight should use seasonal indices or a seasonal forecasting model rather than treating recurring peaks as random noise.

3.3 Causal and collaborative forecasting

Freight demand may be explained by production plans, sales forecasts, supplier schedules, promotions, construction programmes, economic indicators or commodity prices. A simple linear relationship can be written as:

where Y_t is freight demand, X_t is an explanatory variable, a is the intercept, b is the estimated effect of X_t, and ε_t is the unexplained error. Causal models are useful when the explanatory variable is available before the transport decision and the relationship is stable enough to support planning.

Collaborative planning improves visibility when manufacturers, suppliers, carriers and customers share forecasts, commitments and exceptions. However, collaboration requires common definitions, agreed ownership of data and protection of commercially sensitive information.

3.4 Forecast-error measures

For actual demand A_t and forecast F_t, the error is:

Mean absolute deviation is:

It expresses average error in the original demand unit. Mean absolute percentage error is:

MAPE is unsuitable when actual demand is zero or very close to zero. Weighted measures or scale-dependent measures should then be considered. Forecast accuracy must not be confused with forecast bias. A planner should also monitor mean error:

A persistent positive or negative ME indicates systematic under- or over-forecasting.

Worked Example 1: Exponential smoothing

Actual demand in period 1 is 120 loads and the period-1 forecast was 100 loads. Let α = 0.30. The forecast for period 2 is:

The result is in loads because both input terms are in loads. If period 2 actual demand is 130 loads, the period-3 forecast is:

Operationally, 113.2 is a planning expectation; vehicle dispatch decisions may require rounding, aggregation or a capacity buffer rather than automatic rounding to 113 vehicles.

4. Shipment Segmentation

A single service policy for all freight is usually inefficient. Segmentation groups shipments with similar economic, physical and service characteristics so that differentiated policies can be applied.

4.1 Useful segmentation dimensions

Value and criticality. High-value or production-critical materials may justify premium transport, tighter monitoring and redundant capacity. A low-value consumable may be consolidated.

Demand volume and variability. High-volume stable flows support scheduled milk runs, rail or contracted full-truckload capacity. Low-volume intermittent flows may require consolidation or less frequent service.

Physical characteristics. Weight, cube, density, hazardous classification, temperature sensitivity, fragility and handling requirements affect equipment and mode selection.

Time sensitivity. Some shipments have a hard delivery deadline; others have an acceptable delivery window. The distinction between customer-requested date and economically necessary date is important.

Geography and lane structure. Dense, repetitive lanes can support dedicated services, whereas dispersed origins and destinations often require networks, hubs or brokerage.

Risk and security. Controlled, dangerous or theft-prone goods require qualified carriers, chain-of-custody procedures and sometimes secure routing.

An ABC analysis can prioritize managerial attention by annual freight spend or contribution value, but ABC alone does not capture criticality or service risk. A two-dimensional classification combining value and criticality is often more useful. For example, a low-cost component that stops a production line should not be assigned a low-priority transport policy merely because its purchase price is small.

4.2 Segment-to-policy translation

Segmentation should be reviewed periodically. A product may move from routine to critical during a shortage, product launch or maintenance shutdown.

5. Translating Service Requirements into Planning Decisions

Customer service is multidimensional. Common dimensions include on-time delivery, complete delivery, order-cycle time, lead-time consistency, damage rate, shipment visibility and recovery performance.

A useful service specification states:

• the promised delivery window;
• the order cut-off time;
• required frequency;
• acceptable transit-time range;
• completeness and condition requirements;
• tracking and proof-of-delivery expectations; and
• escalation and recovery rules.

5.1 Lead time and variability

Transportation lead time can be separated conceptually into order processing, waiting, loading, line-haul, border or terminal delay, unloading and final delivery. Mean lead time alone is insufficient; variability determines the risk of late delivery and the inventory required to protect service.

For a continuous-review inventory system, transportation lead-time demand is often represented as:

where ROP is reorder point, μ_L is expected demand during lead time, σ_L is the standard deviation of demand during lead time, z is the standard-normal factor associated with the target cycle-service probability. This formula assumes that lead-time demand is approximately normally distributed and that the selected service definition is appropriate. It should not be applied uncritically to intermittent or highly skewed demand.

If demand per period has mean μ_D and standard deviation σ_D, lead time has mean μ_L and standard deviation σ_L, and demand and lead time are independent, then:

The first expression has units of demand; the second has units of demand as well. The variance expression incorporates both demand and lead-time uncertainty. Correlation between demand and lead time would require an additional covariance term.

5.2 Frequency decisions

Frequency affects responsiveness, consolidation and inventory. More frequent departures reduce waiting time and can lower cycle stock, but they may increase cost per unit because loads are smaller.

For a regular service with inter-departure interval T, average waiting time for randomly arriving freight is approximately T ÷ 2, assuming arrivals are uniformly distributed and departures are reliable. This is a planning approximation, not a guarantee. If departures are daily, the average waiting time attributable solely to the schedule is approximately half a day.

The economic decision must include transport cost, handling, inventory carrying cost, stockout risk and customer consequences. A low-frequency service may appear inexpensive while generating excessive inventory or emergency shipments.

5.3 Capacity planning

Required capacity should reflect forecast volume, peak demand, equipment constraints and uncertainty. A simple planning calculation is:

where C_req is required available capacity in the planning period, D_peak is peak demand in compatible capacity units, and u is the planned utilization fraction, 0 < u ≤ 1. The formula assumes capacity units are comparable and that utilization is defined consistently. It should not be interpreted as permission to plan every vehicle at 100% utilization; loading, route geometry, legal weight limits and service recovery require practical slack.

If each vehicle provides usable capacity Q and the planning period contains N dispatch opportunities, the number of vehicles or trips required is approximately:

The result must be rounded upward when fractional trips cannot be operated, then checked against driver hours, loading time, legal limits and return availability.

Worked Example 2: Capacity with a service buffer

A distribution centre expects peak weekly demand of 420 pallets. Each vehicle can legally and practically carry 28 pallets. Management plans at 85% utilization to protect service.

Effective planned capacity per trip is:

Required trips are:

Therefore, at least 18 trips are required, subject to schedule feasibility. Planning at legal capacity alone would give 420 ÷ 28 = 15 trips, which understates the operational requirement because it leaves no allowance for uneven loading, variability or recovery.

6. Consolidation, Shipment Size and Mode Choice

Consolidation combines smaller shipments into larger movements. It can reduce cost and emissions per unit, but it may increase waiting, handling and damage exposure. The decision should compare the marginal transport saving with the cost of delay and additional inventory.

Mode choice is not a simple ranking of road, rail, water or air. It depends on total logistics cost and service requirements. Relevant criteria include line-haul cost, accessorial charges, transit time, reliability, network coverage, shipment size, handling risk, emissions, regulatory restrictions and resilience.

A total-cost perspective may be represented conceptually as:

where TLC is total logistics cost, T is transport cost, I is inventory and pipeline cost, H is handling and terminal cost, R is expected disruption or failure cost, and E is relevant environmental or compliance cost. This is a decision framework rather than a universal accounting identity; each component must be defined consistently and double counting avoided.

7. Network and Service Design

Transportation service planning operates within a network of suppliers, plants, warehouses, cross-docks, customers, hubs and carriers. Direct shipment usually reduces handling and transit stages but may produce low load factors. Hub-and-spoke designs can consolidate flows but add transfer time and dependency on hub performance.

Important design choices include:

• direct versus consolidated movement;
• private fleet versus common carrier;
• dedicated versus shared capacity;
• fixed timetable versus dynamic dispatch;
• regional distribution centres versus national distribution;
• primary carrier plus backup carrier;
• normal routing versus contingency routing.

Service design should use scenario analysis. At minimum, planners should test normal demand, peak demand, supplier delay, vehicle unavailability, weather disruption, border delay and sudden demand decline. A plan that is least expensive under average conditions may be unacceptable when the cost of failure is high.

8. Technology, Analytics and Artificial Intelligence

Transportation-management systems integrate orders, rates, carrier capacity, routing, appointments, tracking and freight audit. Telematics and electronic proof of delivery improve event visibility. Control towers provide exception-based monitoring across multiple carriers and modes.

Analytical applications include demand sensing, estimated-time-of-arrival prediction, dynamic consolidation, vehicle-capacity matching and predictive maintenance. Artificial intelligence can identify nonlinear patterns and recommend actions, but it does not remove the need for data governance. Models can reproduce historical bias, fail during structural change or optimize a narrow cost measure at the expense of service and safety.

Good governance requires:

• documented data ownership and definitions;
• validation against operational baselines;
• monitoring for forecast drift and bias;
• human approval for safety-critical or high-impact decisions;
• audit trails for recommendations and overrides; and
• protection of commercially sensitive and personal data.

9. Performance Measurement and Control

A balanced transportation dashboard should combine demand, cost, service, asset and risk indicators.

On-time performance must specify the agreed time window. A delivery one minute before the window and one hour late should not be treated as equivalent. Similarly, a high vehicle fill rate can hide poor customer service if loads wait too long for consolidation.

10. Illustrative Case: Regional Components Distribution

Background

A fictional manufacturer supplies components from one plant to 18 industrial customers. Demand is stable for several customers but highly variable for maintenance customers. The existing policy dispatches a truck whenever an order is ready. This has produced many partially filled vehicles, premium freight during shortages and inconsistent delivery times.

Problem

Management wants to reduce transportation cost without lowering the 96% on-time-in-full target for production-critical customers. Available data show that 62% of shipments are less than half a vehicleload, while three regional lanes account for 71% of total volume.

Analysis

The planning team first cleanses shipment records and separates regular production demand from emergency maintenance orders. It then classifies customers by criticality and delivery-window flexibility. The three dense lanes have predictable weekly demand and are suitable for scheduled consolidation. Maintenance orders are low in volume but high in urgency and are retained in an expedited service class.

Forecasts are produced weekly at lane level using exponential smoothing, with production schedules supplied by major customers as causal information. Peak weekly pallet demand is estimated at 420 pallets. Vehicles have a practical capacity of 28 pallets, but the operating policy uses 85% planned utilization, requiring 18 weekly trips as calculated earlier.

Approach and solution

The firm implements:

1. fixed departure days on the three dense lanes;
2. a standard economy service with a two-day delivery window;
3. an expedited service for critical maintenance material;
4. reserved backup capacity equal to two trips per week;
5. supplier and customer appointment discipline;
6. electronic proof of delivery and exception alerts; and
7. a monthly review of forecast bias, on-time-in-full performance and premium freight.

The company does not force all customers into the consolidated schedule. Production-critical customers receive reserved capacity, while flexible customers accept a wider window in exchange for lower cost.

Outcome

In the first three months, average load factor improves, emergency shipments decline and delivery reliability increases on regular lanes. The cost reduction is smaller than the initial estimate because the firm maintains backup capacity and pays for monitoring. This is intentional: resilience and service are treated as economic requirements rather than unnecessary overhead.

Learning points

The case demonstrates that transportation savings arise from segmentation and synchronization, not simply from demanding higher carrier utilization. It also shows why forecast granularity, peak capacity and service differentiation must be considered together. A single average-volume plan would have concealed the urgent maintenance segment and produced unacceptable service risk.

11. Challenges, Ethics and Regulation

Planning decisions may create ethical and regulatory consequences. Carriers must comply with rules concerning driver hours, vehicle weights, dangerous goods, customs, worker safety, accessibility and environmental reporting. Cost reduction must never be achieved by encouraging unsafe loading, excessive working hours or concealment of emissions and incidents.

Procurement teams should avoid unfairly transferring forecast uncertainty to small carriers through punitive contracts or unrealistic appointment requirements. Tendering should consider financial stability, safety records, labour practices and data security, not only price.

Privacy is relevant when telematics, driver records or customer-location data are collected. Data should be limited to legitimate purposes, protected against unauthorized access and retained according to applicable law and policy.

12. Best-Practice Planning Framework

A mature organization should:

• forecast at lane, time and service-class level;
• maintain a documented hierarchy of forecast ownership;
• distinguish baseline demand from projects, promotions and emergencies;
• segment by criticality as well as value and volume;
• translate service promises into measurable operating specifications;
• plan peak capacity and recovery capacity separately;
• use total logistics cost rather than transport rate alone;
• validate models with operational personnel;
• monitor bias, variability and exceptions continuously; and
• review policies after structural changes, disruptions or network redesign.

The central managerial discipline is to make trade-offs explicit. Faster transport may reduce pipeline inventory but increase monetary and environmental cost. Consolidation may reduce unit cost but increase waiting. High utilization may improve productivity but reduce resilience. Effective planning does not eliminate these tensions; it measures and manages them.

Chapter Summary

Transportation demand is a derived demand shaped by material flows, production, inventory and customer requirements. Effective planning forecasts not only total volume but also lane, timing, shipment size and service characteristics. Naïve forecasting, moving averages, exponential smoothing and causal models are useful when their assumptions are understood and their errors are monitored.

Shipment segmentation enables differentiated policies based on criticality, volume, variability, physical characteristics, geography, urgency and risk. Service requirements must be translated into delivery windows, frequency, lead time, reliability, capacity and recovery rules. Capacity planning should include peak demand and practical utilization rather than relying on nominal vehicle capacity.

Consolidation, mode choice and network design require a total logistics-cost perspective that includes transport, inventory, handling, failure and environmental consequences. Technology and AI can improve visibility and prediction, but governance, human oversight and data quality remain essential. The professional planner therefore balances cost, service, resilience, safety, regulatory compliance and sustainability through a continuous plan–execute–measure–learn cycle.

Key Terms

Derived demand: Demand for transport created by the need to move materials or products.

Freight work: Transport activity commonly measured in tonne-kilometres.

Lane: A defined origin–destination movement or corridor.

Forecast bias: Persistent tendency to over-forecast or under-forecast.

Exponential smoothing: Forecasting method that weights recent observations more heavily.

Lead time: Elapsed time between initiating and completing a material movement or order cycle.

Service window: Agreed interval within which delivery is expected.

Criticality: Consequence of failure or delay for operations or customers.

Consolidation: Combining smaller shipments into a larger movement.

Utilization: Proportion of available capacity used under a defined measure.

Safety capacity: Capacity reserved to absorb variability or disruption.

On-time-in-full: Delivery meeting both timing and quantity requirements.

Premium freight: Higher-cost transport used to recover urgency or failure.

Empty running: Vehicle movement without revenue-generating freight.

Total logistics cost: Combined cost of transport, inventory, handling, risk and relevant external effects.

Control tower: Integrated capability for monitoring and managing network events.

Demand variability: Fluctuation in demand around its expected level.

Service segmentation: Assignment of different transport policies to distinct shipment groups.

Review Questions

5 Short Answer Questions

1. Why is transportation demand described as derived demand?
2. Distinguish forecast accuracy from forecast bias.
3. Identify four dimensions that should be included in a freight forecast.
4. Why may ABC analysis be insufficient for transportation segmentation?
5. State two reasons why planning vehicles at 100% capacity can be imprudent.

5 Long Answer Questions

1. Explain how strategic, tactical and operational transportation planning differ, and show how information flows between them.
2. Compare moving averages, exponential smoothing and causal forecasting for freight-demand planning.
3. Develop a shipment-segmentation framework for a manufacturer serving both routine and emergency customers.
4. Discuss the trade-offs among frequency, consolidation, inventory and customer service.
5. Evaluate the role of transportation-management systems and AI, including their governance risks.

3 Analytical / Application Questions

1. A lane has peak demand of 560 pallets per week. Each vehicle has a practical capacity of 32 pallets and the firm plans at 80% utilization. Calculate the minimum weekly trips and explain what operational checks are still required.
2. A company is considering reducing departures from five per week to two per week. Analyse the likely effects on average waiting time, consolidation, inventory, premium freight and service risk.
3. Design a service policy for three shipment segments: high-value urgent parts, stable full-load materials and low-value flexible supplies. Justify frequency, mode, capacity reservation and performance measures for each segment.

Multiple Choice Questions with Answers

Question 1

Which statement best describes transportation demand?

A. It is independent of production and inventory decisions

B. It is usually derived from the need to move materials and products

C. It is measured only in vehicle numbers

D. It exists only after a carrier accepts a tender

Correct Answer: B. It is usually derived from the need to move materials and products

Question 2

With exponential smoothing, α = 0.30, actual demand = 120 loads and prior forecast = 100 loads. What is the next forecast?

A. 103 loads

B. 106 loads

C. 110 loads

D. 120 loads

Correct Answer: B. 106 loads

Question 3

Which measure is expressed in the original demand unit?

A. MAD

B. MAPE

C. Forecast percentage bias only

D. Seasonal index

Correct Answer: A. MAD

Question 4

Why should shipment criticality be considered alongside shipment value?

A. Low-value items can never require urgent service

B. Critical low-value components can cause major operational losses if delayed

C. Value and criticality are always identical

D. Criticality determines only vehicle colour

Correct Answer: B. Critical low-value components can cause major operational losses if delayed

Question 5

A vehicle can carry 28 pallets, but planned utilization is 85%. What is its planned effective capacity?

A. 20.8 pallets

B. 23.8 pallets

C. 28.85 pallets

D. 32.9 pallets

Correct Answer: B. 23.8 pallets

Question 6

If departures are reliable and evenly spaced every two days, the approximate average schedule-related waiting time for randomly arriving freight is:

A. One hour

B. Half a day

C. One day

D. Two days

Correct Answer: C. One day

Question 7

Which condition is required for the stated lead-time-demand variance formula to apply directly?

A. Demand and lead time must be independent

B. Demand must always be zero

C. Lead time must be exactly one period

D. All shipments must be urgent

Correct Answer: A. Demand and lead time must be independent

Question 8

Which shipment is the strongest candidate for a premium service?

A. A low-value, non-urgent replenishment with a wide delivery window

B. A production-critical component whose delay stops a plant

C. A stable full-load movement with fixed weekly demand

D. A routine shipment already waiting for consolidation

Correct Answer: B. A production-critical component whose delay stops a plant

Question 9

What is a principal risk of maximizing vehicle utilization without service safeguards?

A. It guarantees lower total logistics cost

B. It eliminates the need for forecasts

C. It can reduce resilience and increase late-delivery risk

D. It makes capacity planning unnecessary

Correct Answer: C. It can reduce resilience and increase late-delivery risk

Question 10

Which measure combines transport cost with inventory, handling and failure considerations?

A. Vehicle fill rate

B. Total logistics cost

C. Transit distance alone

D. Number of carriers

Correct Answer: B. Total logistics cost