Demand, Capacity and Logistics Planning

Introduction

Logistics systems must match customer demand with the people, equipment, space, transport, inventory and information needed to fulfil it. This matching is difficult because demand is uncertain, capacity is not always flexible, and decisions made by one supply-chain member affect others. A retailer’s promotion may increase orders to a wholesaler; the wholesaler may order even more from a manufacturer; and the manufacturer may interpret the increase as a permanent change in market demand. This amplification is known as the bullwhip effect.

Demand, capacity and logistics planning therefore connect commercial expectations with operational resources. Forecasting estimates likely demand; sales and operations planning (S&OP) reconciles demand with supply; aggregate planning converts broad plans into workforce, inventory and production decisions; and capacity analysis tests whether facilities and transport resources can deliver the plan. Effective planning does not eliminate uncertainty. It makes assumptions explicit, prepares alternatives and coordinates decisions across the supply chain.

Learning Objectives

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

1. Explain the role and limitations of demand forecasting in logistics.
2. Calculate and interpret moving-average, weighted-average and exponential-smoothing forecasts.
3. Describe the S&OP and aggregate-planning processes.
4. Distinguish design, effective and actual capacity and calculate capacity utilisation.
5. Explain demand variability, safety capacity and the bullwhip effect.
6. Recommend practical methods for coordinating logistics resources.

1. Demand in the Logistics System

Demand is the quantity of a product or service that customers require during a specified period. It may be measured as units, pallets, orders, tonnes, vehicle loads or monetary value. Logistics planners should distinguish several forms of demand:

• Independent demand originates from external customers and is not directly determined by another item’s requirements.
• Dependent demand is derived from the requirement for another item, such as packaging needed for finished goods.
• Derived demand results from a related activity, such as transport demand created by sales volume.
• Observed demand is recorded sales or shipment data; it may differ from true demand when stockouts, lost sales or rationing occur.

A demand pattern can contain level, trend, seasonality, cyclical movement and random variation. Level is the underlying average; trend is a persistent increase or decrease; seasonality is a repeating pattern within a known period; cycles are longer, less regular fluctuations; and random variation is unpredictable noise. A planner should diagnose these components before selecting a forecasting method.

Forecasts are not promises. They are structured estimates used to make decisions. A forecast is useful only when its time horizon, aggregation level, data quality and purpose are clear. A short-term warehouse labour forecast may use daily order history, whereas a facility-location decision may use annual demand by region over several years.

2. Forecasting Principles and Methods

A sound forecasting process normally includes defining the forecasting object, collecting and cleaning data, examining the pattern, selecting a method, generating the forecast, measuring error and revising the method when conditions change. Forecasts should be prepared at the level at which decisions are made, while avoiding unnecessary detail that produces noise.

Naïve forecasting

The simplest method sets the next period’s forecast equal to the most recent actual demand:

Here, F_t+1 is the forecast for period t+1, and A_t is actual demand in period t. The method is inexpensive and often useful as a benchmark. It is unsuitable when demand has a strong trend or changing seasonality.

Moving average

A moving average uses the most recent n observations:

The observations must use the same unit and time interval. A larger n smooths random variation but responds more slowly to changes. A smaller n responds faster but is more sensitive to noise.

Worked example. Weekly demand was 420, 450 and 390 units in the last three weeks. A three-week moving-average forecast is:

F₄ = (420 + 450 + 390) ÷ 3 = 420 units.

This forecast assumes that recent observations are equally relevant and that no important structural change has occurred.

Weighted moving average

A weighted moving average gives different importance to recent observations:

where each weight w_i is dimensionless, non-negative, and the weights sum to 1. If the latest three demands are 420, 450 and 390, with weights 0.5, 0.3 and 0.2 respectively, then:

F₄ = (0.5 × 420) + (0.3 × 450) + (0.2 × 390) = 423 units.

Weights should reflect evidence about responsiveness, not personal preference alone.

Exponential smoothing

Simple exponential smoothing updates the previous forecast using the latest forecast error:

where 0 ≤ α ≤ 1 is the smoothing constant, A_t is actual demand, and F_t is the previous forecast. A high α reacts quickly; a low value produces greater smoothing. Simple exponential smoothing is appropriate mainly for level data without pronounced trend or seasonality. Trend and seasonal patterns require extensions such as Holt or Holt–Winters methods.

Example. If A_t = 500, F_t = 460 and α = 0.30, then:

F_t+1 = (0.30 × 500) + (0.70 × 460) = 472 units.

Forecast-error measures

Let the forecast error be:

A positive error means actual demand exceeded the forecast. Common measures are:

MAD is mean absolute deviation and retains the demand unit. MSE penalises large errors more heavily and has squared units. MAPE is a percentage, but it becomes unstable or undefined when actual demand is zero or very close to zero. These measures compare methods; they do not by themselves prove that a forecast is operationally adequate. A small average error can conceal serious errors during peak periods.

Forecast bias should also be monitored. Persistent positive errors indicate under-forecasting; persistent negative errors indicate over-forecasting. Managers should investigate bias caused by promotions, stockouts, data exclusions, incentives or systematic judgement.

3. Sales and Operations Planning

S&OP is a cross-functional process that aligns market demand and operational supply over a medium-term horizon, commonly several months to approximately two years. It is not merely a forecasting meeting. It is a decision process involving sales, marketing, operations, procurement, logistics, finance and senior management.

A practical S&OP cycle contains five activities:

1. Data gathering: update sales, inventory, orders, capacity, supplier and cost information.
2. Demand review: create an unconstrained demand plan, including promotions, customer intelligence and statistical forecasts.
3. Supply review: test production, labour, warehouse, transport, supplier and inventory feasibility.
4. Reconciliation: evaluate alternatives, trade-offs, risks and financial consequences.
5. Executive decision: approve one plan, record assumptions and assign accountability.

The demand plan should initially show what the market may require, not what current capacity can fulfil. The supply review then identifies gaps. Typical responses include changing production timing, using overtime, employing a third party, building inventory before a peak, changing product mix, negotiating supplier flexibility or revising customer commitments.

S&OP works best when data definitions are shared. “Demand” should specify whether it means orders, shipments, consumption or unconstrained market potential. Plans should be expressed in common units, with monetary reconciliation where appropriate. A frozen short-term period can reduce disruptive changes, while a flexible longer-term horizon permits scenario planning.

4. Aggregate Planning

Aggregate planning determines broad resource levels rather than detailed daily schedules. It balances demand and supply over a medium-term horizon by deciding combinations of workforce, inventory, overtime, subcontracting, backorders and capacity changes.

Two classic strategies illustrate the trade-off:

• Level strategy: maintain relatively stable production and workforce, using inventory or backorders to absorb demand variation.
• Chase strategy: vary production or workforce to follow demand more closely, reducing inventory but increasing hiring, training, overtime or adjustment costs.

Most organisations use a mixed strategy. A planner may retain a core workforce, use overtime for moderate peaks, build inventory ahead of predictable seasonality and subcontract exceptional volume.

An aggregate plan should consider total relevant cost, service and feasibility. A simplified planning cost can be represented as:

This is a planning identity, not a universal optimisation formula. Each component must be defined consistently and measured over the same horizon. A low-cost plan is unacceptable if it violates safety, labour agreements, storage limits, supplier constraints or customer-service requirements.

Aggregate planning example

A distributor expects monthly demand of 1,000 units in month 1 and 1,400 units in month 2. It begins with 200 units of inventory and can process 1,000 units per month in regular time. It may use overtime for up to 300 units in month 2. A level regular-time plan produces 1,000 units each month:

• End of month 1 inventory = 200 + 1,000 − 1,000 = 200 units.
• End of month 2 inventory before overtime = 200 + 1,000 − 1,400 = −200 units.

Thus, regular time alone creates a 200-unit shortage in month 2. Using 200 overtime units avoids the shortage and leaves zero ending inventory. The decision should then compare overtime cost with alternatives such as prebuilding 200 units in month 1, subcontracting, or accepting a controlled backorder. The calculation demonstrates why aggregate planning must examine inventory flow, not only total capacity over the entire horizon.

5. Capacity Analysis

Capacity is the maximum output a resource can produce under specified conditions and during a stated period. It must always be described with its unit, time basis and operating assumptions.

• Design capacity is the theoretical maximum under ideal conditions.
• Effective capacity is the realistic maximum after allowing for maintenance, breaks, changeovers, product mix, quality requirements and other normal restrictions.
• Actual output is what was actually produced or processed.

Two useful ratios are:

These ratios answer different questions. Utilisation compares performance with the ideal design; efficiency compares it with realistic capability. Neither ratio alone measures customer service or profitability.

Example. A cross-dock is designed for 10,000 cartons per day. Normal staffing, handling and maintenance produce an effective capacity of 8,000 cartons per day. Actual throughput is 7,200 cartons. Therefore:

• Utilisation = 7,200 ÷ 10,000 × 100% = 72%.
• Efficiency = 7,200 ÷ 8,000 × 100% = 90%.

The operation appears to use only 72% of design capacity, but it is using 90% of realistic capacity. Further volume may require overtime, layout changes or an additional shift.

Capacity must be analysed at the constraint, or bottleneck. A system’s throughput cannot sustainably exceed the capacity of its most restrictive required resource. Bottlenecks may be a packing station, loading door, driver availability, supplier, customs process, information system or storage location. Improving a non-bottleneck may create local efficiency without increasing total flow.

Capacity planning should distinguish:

• Capacity cushion: capacity deliberately held above expected demand to handle uncertainty.
• Capacity flexibility: the speed and cost with which capacity can be changed.
• Capacity timing: whether resources become available before, at, or after demand occurs.

A useful planning sequence is to forecast demand, convert it into workload, compare workload with effective capacity, identify bottlenecks, develop alternatives and test the plan under peak and disruption scenarios.

6. Demand Variability, Safety Capacity and Inventory

Variability arises from changing customer demand, order batching, promotions, lead-time uncertainty, supplier unreliability, transport disruption and inaccurate data. Variability consumes capacity because resources must handle peaks, waiting, rework and unplanned changes.

If demand during a replenishment lead time is approximately normally distributed, a commonly used safety-stock relationship is:

where SS is safety stock in units, z is the standard-normal service factor, and σ_LT is the standard deviation of demand during lead time. This formula assumes that lead time is fixed, demand variation is represented adequately by a normal approximation, and the chosen service level is meaningful. If both demand and lead time vary independently, a common approximation is:

where L is average lead time in periods, σ_d is standard deviation of demand per period, d is average demand per period, and σ_L is standard deviation of lead time in periods. The units are consistent: both terms under the square root have units of squared demand.

These formulas are approximations, not guarantees. Intermittent demand, correlation between demand and lead time, promotions and non-normal distributions may require simulation or empirical percentiles. Safety stock also cannot compensate indefinitely for poor forecasting or unreliable suppliers.

7. The Bullwhip Effect

The bullwhip effect is the tendency for order variability to increase as information moves upstream from customers to retailers, wholesalers, manufacturers and suppliers. End-customer consumption may be relatively stable while upstream orders fluctuate sharply.

Major causes include:

• forecasting based on orders rather than point-of-sale consumption;
• order batching to reduce transaction or transport cost;
• temporary price discounts and forward buying;
• shortage gaming, where customers inflate orders during constrained supply;
• long and variable replenishment lead times;
• poor information sharing and lack of trust;
• local performance measures that encourage conflicting decisions.

The consequences include excess inventory, stockouts, unstable production schedules, overtime, transport inefficiency, obsolescence and poor supplier relationships.

Countermeasures include sharing point-of-sale data, shortening lead times, ordering more frequently in economical quantities, stabilising prices, allocating scarce products transparently, using vendor-managed inventory where appropriate and measuring total supply-chain performance. Collaborative planning, forecasting and replenishment (CPFR) can align forecasts, promotions and replenishment decisions, but it requires compatible systems, governance and willingness to share reliable information.

A crucial distinction is between demand variability and order variability. Better visibility does not necessarily remove genuine market variation; it reduces the additional variation created by local decisions and distorted information.

8. Integrated Logistics Planning Framework

The following logic integrates the chapter’s concepts:

Figure: A coordinated planning cycle linking demand information, forecasting, S&OP, capacity and feedback.

The cycle begins with clean demand data and a documented forecast. S&OP converts the forecast into a cross-functional plan. Capacity analysis tests feasibility and exposes bottlenecks. Execution produces actual results, which feed back into forecast-error analysis, capacity assumptions and future decisions.

9. Technology, Data and Governance

Enterprise resource planning systems, warehouse-management systems, transport-management systems and demand-planning platforms can improve visibility and automate calculations. Point-of-sale feeds, barcode scans, telematics and electronic data interchange reduce latency and manual error. Artificial-intelligence and machine-learning tools may identify nonlinear patterns or combine large numbers of explanatory variables, but they do not replace process knowledge.

Technology should be governed through data ownership, version control, access rights, audit trails and clear exception management. A sophisticated model trained on stockout-distorted sales data may reproduce the distortion. Automated forecasts can also create false confidence when market conditions change. Human review is especially important for promotions, product launches, disasters and regulatory changes.

Ethically and professionally, planners should avoid manipulating forecasts to meet performance targets, concealing capacity constraints or allocating scarce supply unfairly. Forecasts may influence labour scheduling and supplier commitments, so organisations should consider privacy, discrimination, worker safety and transparent decision rules.

10. Illustrative Case: FreshBite Distribution

Background

FreshBite, a fictional regional grocery distributor, supplies 80 stores. Demand for chilled ready meals is stable on ordinary weeks but rises sharply before weekends and during promotional campaigns. Stores send replenishment orders each day, while the distribution centre schedules labour weekly.

Problem

The distributor’s central forecast is based on store orders. During a promotion, stores order extra cases early and continue ordering defensively because they fear shortages. The distribution centre interprets the orders as permanent growth. It hires temporary labour, uses emergency transport and accumulates excess stock after the promotion. Despite this excess, several stores experience stockouts during the first promotional days.

Analysis

The planning team compares point-of-sale sales, store orders, inventory records and promotion calendars. It finds that store orders vary much more than consumer consumption. The warehouse’s design capacity is 12,000 cases per day, effective capacity is 9,600 cases, and promotional throughput reaches 9,200 cases. Efficiency is therefore approximately 95.8% (9,200 ÷ 9,600 × 100), leaving little protection against late trucks or absenteeism.

Approach and solution

FreshBite introduces a joint promotion forecast using point-of-sale data and agreed store assumptions. It creates a frozen order window for the first two days of each promotion, replenishes stores using agreed minimum and maximum levels, and prepositions selected high-volume items. The S&OP meeting reviews the promotion six weeks ahead, checks supplier and transport capacity, and approves a temporary labour plan. A dashboard tracks forecast bias, service level, warehouse efficiency and waste rather than orders alone.

Outcome

In this illustrative case, order spikes become smaller and the warehouse plans labour earlier. Inventory is concentrated on products with reliable promotional demand rather than spread across the entire range. The approach does not eliminate uncertainty, but it reduces information distortion and makes capacity decisions more deliberate.

Learning points

1. Orders are not always a reliable proxy for consumption.
2. Promotions must be represented in both demand and capacity plans.
3. Service, waste and total cost should accompany utilisation measures.
4. Coordination requires shared data, agreed decision rules and cross-functional accountability.

Challenges and Limitations

Forecasting cannot reliably predict unprecedented events. Historical data may be incomplete, censored by stockouts or changed by assortment and channel shifts. Aggregate plans can hide important differences between products, customers and locations. Capacity ratios may reward output even when quality or safety deteriorates. Safety-stock formulas depend on assumptions that may not hold in intermittent or highly seasonal demand. S&OP can become ineffective if meetings are dominated by one function, if decisions are not implemented, or if incentives reward local optimisation.

Managers should therefore combine quantitative methods with operational judgement, scenario analysis and frequent review. A good plan is not the one that appears most precise; it is the one whose assumptions, risks, triggers and responsibilities are visible.

Best-Practice Checklist

• Define demand consistently as consumption, orders, shipments or market demand.
• Use a simple benchmark before adopting a complex forecasting model.
• Measure error and bias by product, location and time period.
• Separate unconstrained demand from feasible supply.
• Review promotions, launches and disruptions explicitly.
• Analyse effective capacity and bottlenecks, not design capacity alone.
• Protect a capacity cushion where variability and service consequences justify it.
• Share timely point-of-sale and inventory information.
• Align incentives with total supply-chain performance.
• Document assumptions, decisions, exceptions and follow-up actions.

Chapter Summary

Demand, capacity and logistics planning coordinate uncertain customer requirements with limited operational resources. Forecasts describe likely future demand but must be evaluated for error, bias, horizon and data quality. Naïve, moving-average, weighted-average and exponential-smoothing methods are useful foundations when their assumptions are understood. S&OP integrates demand, supply and financial considerations through a recurring cross-functional decision process. Aggregate planning balances workforce, inventory, overtime, subcontracting and service over a medium-term horizon.

Capacity analysis distinguishes design, effective and actual capacity and directs attention to bottlenecks, flexibility and capacity cushions. Demand variability creates the need for inventory, flexible resources and robust scenarios. The bullwhip effect arises when order variability becomes amplified upstream through forecasting from orders, batching, promotions, shortage gaming and poor information sharing. Point-of-sale visibility, shorter lead times, stable commercial policies and collaborative planning can reduce this amplification. Technology supports planning, but governance, professional judgement and ethical decision-making remain essential.

Key Terms

Aggregate planning: Medium-term planning of broad resources, output and inventory.

Bullwhip effect: Amplification of demand or order variability upstream in a supply chain.

Capacity cushion: Deliberate capacity held above expected demand for uncertainty or peaks.

Capacity flexibility: Ability to change capacity in volume, timing or mix.

Design capacity: Theoretical maximum output under ideal conditions.

Dependent demand: Demand derived from the requirement for another item.

Exponential smoothing: Forecasting method that updates a forecast using recent actual demand and a smoothing constant.

Effective capacity: Realistic capacity after normal operating restrictions.

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

Independent demand: Demand originating from external customers.

Level strategy: Aggregate plan maintaining relatively stable production or workforce.

Moving average: Forecast based on the arithmetic average of recent observations.

Naïve forecast: Forecast that uses the most recent actual value as the next prediction.

Point-of-sale data: Information recorded when a customer purchase occurs.

Safety stock: Inventory held to protect against demand or lead-time uncertainty.

Sales and operations planning: Cross-functional process aligning demand, supply and financial plans.

Service level: Target probability or measure of meeting customer demand, defined by the organisation.

Utilisation: Actual output expressed as a percentage of design capacity.

Weighted moving average: Forecast that assigns different weights to recent observations.

Review Questions

5 Short Answer Questions

1. Distinguish independent demand from dependent demand.
2. What is the purpose of S&OP?
3. Define design capacity and effective capacity.
4. State two causes of the bullwhip effect.
5. Why can MAPE be problematic when actual demand is zero?

5 Long Answer Questions

1. Explain the main stages of an S&OP cycle and the contribution of different functions.
2. Compare level, chase and mixed aggregate-planning strategies.
3. Explain how bottlenecks influence logistics capacity decisions.
4. Describe the assumptions and limitations of simple exponential smoothing.
5. Discuss how information sharing can reduce the bullwhip effect.

3 Analytical / Application Questions

1. A warehouse has a design capacity of 15,000 cartons per day, effective capacity of 12,000 cartons, and actual throughput of 10,200 cartons. Calculate utilisation and efficiency, then explain what each result implies.
2. Weekly demand for the last three weeks is 700, 760 and 740 units. Calculate the three-week moving-average forecast. If weights of 0.5, 0.3 and 0.2 are applied from most recent to oldest, calculate the weighted forecast and explain which method reacts more strongly to the latest observation.
3. A retailer’s point-of-sale demand is stable, but its orders to a manufacturer fluctuate sharply during discounts. Diagnose the likely causes and recommend three coordinated interventions involving forecasting, commercial policy and replenishment.

Multiple Choice Questions with Answers

Question 1

Which statement best describes a design capacity?

A. Actual output after disruptions

B. Theoretical maximum output under ideal conditions

C. Output required to achieve a target service level

D. Average output after normal maintenance losses

Correct Answer: B. Theoretical maximum output under ideal conditions

Question 2

A three-period moving average uses demands of 120, 150 and 180 units. What is the forecast for the next period?

A. 130 units

B. 140 units

C. 150 units

D. 160 units

Correct Answer: C. 150 units

Question 3

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

A. More responsive to recent actual demand

B. Independent of recent actual demand

C. Less responsive to recent actual demand

D. Valid only for seasonal demand

Correct Answer: A. More responsive to recent actual demand

Question 4

Which activity belongs primarily to the supply review in S&OP?

A. Estimating competitor advertising

B. Testing warehouse and supplier feasibility

C. Recording only historical sales

D. Selecting a product brand name

Correct Answer: B. Testing warehouse and supplier feasibility

Question 5

A facility has actual output of 7,200 units, design capacity of 10,000 and effective capacity of 8,000. What is its efficiency?

A. 72%

B. 80%

C. 90%

D. 125%

Correct Answer: C. 90%

Question 6

Which situation is most likely to create a bullwhip effect?

A. Sharing point-of-sale data across partners

B. Ordering in small frequent quantities based on consumption

C. Forward buying during temporary price discounts

D. Reducing replenishment lead-time variability

Correct Answer: C. Forward buying during temporary price discounts

Question 7

What does a positive forecast error e_t = A_t − F_t indicate?

A. Forecast demand exceeded actual demand

B. Actual demand exceeded forecast demand

C. Actual demand was zero

D. The forecast was necessarily unbiased

Correct Answer: B. Actual demand exceeded forecast demand

Question 8

Which is a limitation of MAPE?

A. It has squared demand units

B. It cannot compare forecasts in percentage terms

C. It becomes unstable when actual demand is zero or near zero

D. It always penalises large errors more than MSE

Correct Answer: C. It becomes unstable when actual demand is zero or near zero

Question 9

A capacity cushion is best described as:

A. Capacity deliberately held above expected demand

B. Inventory that has passed its expiry date

C. Theoretical output under ideal conditions

D. A reduction in effective capacity caused by maintenance

Correct Answer: A. Capacity deliberately held above expected demand

Question 10

Which intervention most directly reduces order amplification caused by poor visibility?

A. Increasing batch sizes

B. Sharing reliable point-of-sale and inventory data

C. Extending replenishment lead times

D. Allowing each member to use unrelated forecasts

Correct Answer: B. Sharing reliable point-of-sale and inventory data