Demand, Capacity and Planning Data

Introduction

Aggregate production planning converts uncertain market requirements and confirmed customer commitments into a feasible medium-term plan for products, resources, facilities and time periods. It occupies the link between strategic decisions—such as facility location, technology and workforce structure—and detailed scheduling decisions concerning individual jobs and machine sequences.

The quality of an aggregate plan depends heavily on the quality of its planning data. A forecast expressed in sales units cannot be used directly if the production system measures capacity in labour-hours, machine-hours or standard hours. Similarly, demand data may contain duplicated orders, cancelled requirements, seasonal effects, new-product uncertainty or inconsistent units of measure. The planner must therefore establish a common, reconciled data structure before evaluating alternatives such as level production, chase production, overtime, subcontracting or inventory accumulation.

This chapter explains how to translate forecasts and orders into aggregate demand, measure available and effective capacity, assess productivity and utilization, identify resource constraints, and prepare consistent planning data by product family, period and facility.

Learning Objectives

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

1. Distinguish forecasts, orders, independent demand and dependent demand.
2. Convert product-level requirements into product-family and aggregate demand.
3. Construct a time-phased demand statement using consistent units and assumptions.
4. Calculate capacity, productivity, utilization and efficiency.
5. Identify bottlenecks and resource constraints across facilities and periods.
6. Prepare a validated planning-data table for aggregate production planning.
7. Critically evaluate data quality, uncertainty, governance and the use of digital tools.

1. The Planning Data Architecture

Aggregate planning normally covers a medium-term horizon, often three to eighteen months, divided into months or weeks. The appropriate horizon depends on the time needed to hire labour, procure materials, alter capacity and respond to customers. The planning level is deliberately aggregated: planners work with product families, labour categories, facilities and broad capacity measures rather than individual operations.

A useful planning-data architecture contains five linked layers:

The data must be time-phased. A total annual forecast is insufficient when demand and capacity vary by month. It must also be dimensionally consistent. For example, finished units, kilograms, standard labour-hours and machine-hours are different measures and cannot be added without a valid conversion relationship.

Important terminology

Forecast is an estimate of future demand based on historical data, market information, judgement or a combination of these. Customer order is a confirmed requirement, although it may still be subject to cancellation or modification. Backlog is accepted demand not yet fulfilled. Product family is a group of items sharing similar production processes, resources or planning characteristics. Planning period is the time bucket used for analysis. Resource requirement is the capacity consumed by a planned quantity. Available capacity is the capacity that can potentially be supplied; effective capacity is the usable capacity after normal losses and policy restrictions.

2. From Market Information to Aggregate Demand

2.1 Forecasts and orders are different signals

Forecasts represent anticipated demand, whereas orders represent commitments already visible in the order system. Simply adding the two can double-count demand if orders were used to create the forecast. A common planning convention is:

where t identifies the planning period. This “forecast consumption” logic assumes that actual orders consume the forecast rather than being added to it. If orders exceed the forecast, the excess becomes visible demand; if orders are below the forecast, the remaining forecast represents expected future orders.

This rule is not universal. A business may maintain separate order and forecast streams when the forecast excludes existing contracts, or when statistical forecasts are explicitly calculated for uncommitted demand. The data owner must document the selected convention.

2.2 Demand reconciliation

A practical reconciliation sequence is:

1. Extract open orders, backlog, quotations and statistical forecasts.
2. Remove cancelled, duplicated and obsolete records.
3. Map each item to a product family and facility.
4. Assign each requirement to the period in which production or delivery is expected.
5. Apply the approved forecast-consumption rule.
6. Add known promotions, launches, service parts and contractual requirements.
7. Separate normal demand from exceptional or one-time demand.
8. Record uncertainty, confidence ranges and responsible data owners.

Demand should generally be represented in the unit that drives production. A beverage producer may plan in cases, a steel mill in tonnes, and an electronics assembler in standard hours or units. If products have different sizes or resource contents, physical units alone are misleading.

2.3 Converting products into family demand

Let q_i,t be the quantity of item i required in period t, and let a_i,f equal 1 when item i belongs to family f and 0 otherwise. Family demand is:

This simple aggregation is appropriate when all items in a family are measured in comparable units. When products consume different amounts of a constrained resource, a weighted measure is more informative:

where r_i is the resource requirement per unit, expressed, for example, in standard labour-hours per unit. The resulting R_f,t is in standard hours and can be compared with available hours.

The conversion coefficients must be maintained carefully. A product redesign, changed staffing method, new machine or improved yield can make historical standards obsolete.

2.4 Net production requirement

Inventory and scheduled receipts affect what must be produced. A basic inventory balance is:

where I_t is ending inventory, P_t production during period t, S_t scheduled receipts from external or previously released supply, and D_t demand. Rearranging gives the production requirement when a target ending inventory I_t^* is specified:

The equation assumes that all quantities are measured in the same equivalent units and that inventory is usable for the relevant demand. If the result is negative, production may be zero, with the surplus absorbed by inventory, cancellation of receipts or a revised target. Safety stock, quality holds, perishability and allocation rules may require additional terms or constraints.

Worked Example 1: Reconciling demand and production

A product family has the following data for April: forecast 4,800 units, confirmed orders 5,100 units, opening inventory 900 units, scheduled receipts 400 units and a desired ending inventory of 700 units. Orders and forecast are assumed to be overlapping signals.

Using the forecast-consumption rule:

Required production is:

Thus, the plan must provide 4,500 units of production, assuming all opening inventory and scheduled receipts are available and acceptable. The planned ending inventory check is 900 + 4,500 + 400 − 5,100 = 700 units.

3. Capacity: Potential, Effective and Demonstrated

Capacity is not a single number. Its meaning depends on the time basis, resource, product mix and operating assumptions.

Design capacity is the theoretical maximum under specified design conditions. Rated capacity is the output implied by a standard rate. Available capacity is the time technically scheduled for production. Effective capacity deducts planned losses such as breaks, preventive maintenance, meetings, setup and expected downtime. Demonstrated capacity is output historically achieved under comparable conditions. For medium-term planning, effective or demonstrated capacity is usually more credible than design capacity.

3.1 Capacity in hours

For a resource with N identical machines, H scheduled hours per machine per period, s shifts or operating blocks represented in the schedule, and an availability factor a, available productive hours may be estimated as:

The variables must not be counted twice. If H already represents total scheduled hours across all shifts, s should not be included again. The availability factor is dimensionless and should be based on a documented definition.

If products require r_i hours per unit, the capacity constraint for a plan is:

where x_i is planned output of item i and C_effective is effective resource capacity in hours. The left and right sides both have units of hours.

3.2 Capacity in units

When a resource makes one family at a stable rate, capacity can be stated in units:

where u is the accepted output rate in units per hour. For a mixed product family, a single units-per-hour figure can be deceptive. Standard hours or equivalent units should then be used.

3.3 Yield and quality losses

If gross production is subject to a good-output yield y, where 0 < y ≤ 1, good units are:

To obtain [[MATH{Q}]] good units, the required gross production is:

This applies when defective output cannot be economically reworked or returned to the process. Yield should not be applied again if the capacity rate already represents good units per hour.

3.4 Overtime, subcontracting and temporary capacity

Capacity additions should be represented separately from regular capacity because they have different costs, quality risks and lead-time implications. A planning model may distinguish regular hours, overtime hours, subcontracted units and temporary labour. Overtime is not automatically available: legal limits, fatigue, skill requirements, maintenance effects and employee agreements constrain it.

4. Productivity, Utilization and Efficiency

These terms are related but not interchangeable.

Productivity relates output to input. For a single input:

Examples include units per labour-hour, tonnes per machine-hour and value added per employee. Multifactor productivity may be written as:

The inputs must be expressed in commensurable units, often monetary units, and the selected price base must be stated. Productivity comparisons can be invalid when product mix, quality, outsourcing or accounting conventions differ.

Utilization measures the extent to which available capacity is used:

Some organizations use available or effective capacity in the denominator. Therefore, the denominator must always be named. Using design capacity shows how intensively the installed system is used; using effective capacity shows how much scheduled capability is consumed.

Efficiency compares actual performance with a standard or effective expectation:

A facility can have low design-capacity utilization but high effective-capacity efficiency if the system is deliberately scheduled below its theoretical maximum. Likewise, high utilization can coexist with poor service if variability causes queues and missed due dates.

Worked Example 2: Capacity and performance

A work centre has 8 machines. Each is scheduled for 160 hours in a month. Planned availability after breaks and preventive maintenance is 85%, and the standard good-output rate is 2.5 units per machine-hour.

Available productive hours are:

Good-unit capacity is:

If actual good output is 2,448 units, efficiency relative to effective capacity is:

If design capacity is based on 8 × 160 × 2.5 = 3,200 units, design-capacity utilization is 2,448 ÷ 3,200 × 100% = 76.5%. The two percentages answer different questions and should not be reported as though they were equivalent.

5. Resource Constraints and Bottleneck Analysis

A bottleneck is a resource whose available capacity limits the output of the system or prevents achievement of the plan. It may be a machine, skilled labour group, supplier, tool, inspection laboratory, warehouse dock or regulatory approval.

A practical constraint analysis involves:

1. Listing every relevant resource by facility and period.
2. Converting planned demand into resource requirements.
3. Comparing requirements with effective capacity.
4. Ranking shortages by business impact and substitutability.
5. Checking whether the apparent bottleneck shifts when the product mix changes.
6. Testing feasible responses: mix changes, overtime, routing alternatives, subcontracting, inventory, maintenance changes or demand negotiation.

For resource [[MATH{k}]], the load ratio is:

A ratio above 1 indicates a mathematical overload under the stated assumptions. A ratio near 1 may still be risky because forecasts, downtime and processing times are uncertain. Capacity buffers are therefore managerial safeguards, not merely unused capacity.

The bottleneck should be assessed at system level. Increasing a non-bottleneck resource does not increase throughput if the limiting resource remains unchanged. Conversely, a resource that appears unconstrained in aggregate may become critical in a particular week or shift.

6. Preparing a Consistent Planning Dataset

A planning dataset should have one clear record for each combination of product family, period and facility, supported by documented master data. A useful structure is shown below.

Data validation controls

Completeness: every family, period and facility has a value or an explicit null reason. Consistency: units, calendars, product codes and facility codes are standardized. Reconciliation: inventory balances, family totals and facility totals agree with source systems. Validity: quantities, rates and factors fall within reasonable ranges. Timeliness: the data cut-off date is recorded. Traceability: each value has a source, owner, version and change history.

A planning workbook or system should distinguish actuals, forecasts, assumptions and decisions. Overwriting an original forecast with a planner adjustment destroys auditability. Instead, retain the original value and store the adjustment with its reason, author and date.

Data hierarchy and aggregation

Aggregation should preserve the information needed for decisions. Excessive aggregation hides bottlenecks and product-mix effects; excessive detail creates noise and computational burden. A family is appropriate when its members have similar demand behaviour, routings, resource consumption, inventory policies and substitution possibilities. Products with different constraints should not be grouped merely because they share a marketing label.

7. A Detailed Illustrative Case: Northstar Pumps

The following is a fictional illustrative case.

Background

Northstar Pumps operates two facilities. Plant A machines housings and Plant B performs final assembly and testing. The company plans monthly production for three families: Standard, High-Pressure and Service Replacement. The aggregate plan for May must satisfy customer requirements while maintaining a minimum ending inventory of 200 equivalent units.

Problem

The sales file contains a forecast of 5,000 equivalent units and confirmed orders of 5,400. The planning team initially added these figures, proposing demand of 10,400 units. This would have created an apparent capacity crisis. In addition, Plant B reported capacity in assembly-hours while the sales file used units.

Analysis

Because orders were already included in the forecast process, the approved rule was to use the larger of forecast and orders. Therefore:

Opening usable inventory is 700 units and scheduled receipts are 300 units. With target ending inventory of 200 units:

Plant B requires 0.50 assembly-hours per equivalent unit. Its effective May capacity is 2,100 assembly-hours:

The load ratio is 2,300 ÷ 2,100 = 1.095, or 109.5%. Plant B is overloaded by 200 hours. Plant A has sufficient machining capacity, so adding machining hours would not solve the system constraint.

Approach and solution

The team evaluated four responses: 200 hours of overtime, subcontracting 400 equivalent units, delaying low-priority service replacements, and reducing the target inventory. The selected plan used 120 hours of qualified overtime and subcontracted 160 equivalent units to an approved supplier. The remaining 80-hour gap was removed by improving the test-cell changeover plan. The supplier was subject to incoming inspection because its historical yield was lower than Plant B’s.

The revised effective capacity was:

The required 2,300 hours could therefore be covered, leaving an 80-hour planning buffer. The subcontract quantity was expressed in equivalent units and converted using the same standard; this avoided treating supplier units as free capacity without considering their effect on quality and logistics.

Outcome

The final plan met the May demand and inventory target, maintained a documented buffer, and identified Plant B—not Plant A—as the limiting resource. Management also revised its master data to distinguish forecast, orders and forecast-consumed demand. The case illustrates that data reconciliation can remove artificial shortages, while capacity conversion can reveal genuine constraints.

Learning points

1. Forecasts and orders must not be added automatically.
2. Inventory and scheduled receipts change production requirements.
3. Capacity must be expressed in the resource actually constrained.
4. A system bottleneck may differ from the facility with the largest nominal workload.
5. Overtime and subcontracting require quality, legal and supplier-capability checks.

8. Technology, Governance and Contemporary Practice

Enterprise resource planning systems, advanced planning systems and manufacturing execution systems can connect demand, inventory, orders, routings and capacity. They improve speed and traceability but do not guarantee correct plans. Incorrect calendars, obsolete standards or duplicated master data can produce precise but invalid recommendations.

Machine-learning forecasts can detect nonlinear patterns and incorporate external variables, but forecast accuracy does not remove the need for managerial review. A planner should examine forecast bias, structural breaks, promotions, new products and explainability. AI-generated recommendations should be governed through version control, approval thresholds, access controls and audit logs. Sensitive customer, employee and supplier data require appropriate privacy and cybersecurity protection.

Contemporary planning increasingly uses scenario analysis and digital twins. A scenario may vary demand, yield, labour availability, supplier reliability or overtime limits. The objective is not to predict one certain future but to identify robust decisions and early-warning indicators. Sustainability should also be included where material: energy availability, emissions constraints, water use, waste, ethical labour conditions and responsible subcontracting can affect feasible capacity.

9. Limitations and Best Practices

Planning data are estimates, not physical facts. Forecast error, uncertain yields, absenteeism, supplier disruption and changing customer priorities make capacity statements conditional. Aggregate models may conceal sequence-dependent setups, learning effects, skill bottlenecks and distribution constraints. A feasible monthly plan may still be impossible to schedule by day.

Best practices include:

• define the demand-consumption convention before aggregation;
• use product families based on operational similarity, not convenience;
• maintain separate units for finished goods, standard hours and monetary values;
• use effective or demonstrated capacity rather than theoretical capacity alone;
• document calendars, yields, standards, assumptions and data cut-off dates;
• test bottlenecks by period, facility and product mix;
• reconcile inventory and capacity equations before optimization;
• retain original data and record planner overrides;
• include uncertainty ranges and capacity buffers;
• validate aggregate plans against detailed scheduling and material feasibility;
• review ethical, legal, safety, environmental and supplier implications.

Chapter Summary

Aggregate production planning begins with reliable, reconciled demand and resource data. Forecasts and orders must be combined according to an explicit convention to avoid double counting. Product-level requirements are aggregated into families or converted into standard resource hours using documented coefficients. Inventory balances determine production requirements after considering opening stock, scheduled receipts and target ending inventory.

Capacity may be expressed in hours or units, but the measure must match the constrained resource. Design, available, effective and demonstrated capacity have different meanings. Productivity relates output to input; utilization compares output with a stated capacity base; and efficiency compares actual output with effective or standard performance. Bottleneck analysis compares resource requirements with effective capacity by period and facility.

A professional planning dataset is complete, consistent, traceable, timely and auditable. Digital systems and AI can improve forecasting, scenario analysis and data integration, but governance and human validation remain essential. The central managerial principle is that an aggregate plan is only as credible as the assumptions, conversions and constraints embedded in its planning data.

Key Terms

Forecast: An estimate of future demand.

Customer order: A confirmed or accepted customer requirement.

Backlog: Accepted demand not yet fulfilled.

Product family: A group of products with similar planning or production characteristics.

Gross demand: Demand after applying the approved forecast-and-order reconciliation rule.

Net production requirement: Quantity that must be produced after considering inventory and scheduled receipts.

Opening inventory: Usable stock available at the beginning of a planning period.

Scheduled receipt: Supply expected from a prior release or external source.

Design capacity: Theoretical maximum output under specified design conditions.

Effective capacity: Usable capacity after normal operating losses and restrictions.

Demonstrated capacity: Historically achieved capacity under comparable conditions.

Productivity: Output divided by input.

Utilization: Actual output relative to a stated capacity base.

Efficiency: Actual performance relative to effective or standard performance.

Bottleneck: A resource that limits system output or plan feasibility.

Load ratio: Required resource capacity divided by available effective capacity.

Standard hour: A normalized measure of resource time required for a product or activity.

Yield: Proportion of gross output that becomes acceptable output.

Data lineage: The documented origin, transformation and ownership of a data value.

Capacity buffer: Deliberately uncommitted capacity retained to absorb uncertainty.

Review Questions

Short Answer Questions

1. Why should forecasts and confirmed orders not automatically be added together?
2. Distinguish design capacity from effective capacity.
3. What is the purpose of converting product quantities into standard hours?
4. Define productivity and utilization.
5. Identify three controls that improve planning-data quality.

Long Answer Questions

1. Explain the process of converting product-level forecasts and orders into aggregate family demand.
2. Discuss the differences among available, effective and demonstrated capacity, including their managerial uses.
3. Explain how inventory balances influence aggregate production requirements.
4. Compare productivity, utilization and efficiency and show why their denominators matter.
5. Describe a systematic method for identifying and responding to bottlenecks.

Analytical / Application Questions

1. A family has a forecast of 8,000 units, orders of 7,200 units, opening inventory of 1,100 units, scheduled receipts of 500 units and a target ending inventory of 900 units. Using the forecast-consumption rule, calculate gross demand and required production. State your assumptions.
2. A facility has 10 machines, each scheduled for 150 hours, with an effective availability factor of 80%. If the product mix requires 0.6 machine-hours per unit, calculate effective unit capacity and assess whether a plan for 1,950 units is feasible.
3. Two facilities have sufficient aggregate annual capacity, but one facility is overloaded in two months. Explain why annual capacity may conceal a bottleneck and recommend data and planning actions to address it.

Multiple Choice Questions with Answers

Question 1

A forecast is 4,800 units and confirmed orders are 5,100 units. If orders and forecast overlap, what is gross demand under the forecast-consumption rule?

A. 300 units

B. 4,800 units

C. 5,100 units

D. 9,900 units

Correct Answer: C. 5,100 units

Question 2

Opening inventory is 900 units, demand is 5,100 units, scheduled receipts are 400 units, and target ending inventory is 700 units. What production is required?

A. 4,100 units

B. 4,500 units

C. 5,400 units

D. 6,700 units

Correct Answer: B. 4,500 units

Question 3

Which capacity measure deducts planned losses such as breaks, maintenance and expected downtime?

A. Design capacity

B. Effective capacity

C. Theoretical demand

D. Nominal product capacity

Correct Answer: B. Effective capacity

Question 4

A work centre has effective capacity of 2,000 hours and requires 2,200 hours for the proposed plan. What is its load ratio?

A. 0.91

B. 1.00

C. 1.10

D. 2.20

Correct Answer: C. 1.10

Question 5

Which measure is defined as output divided by input?

A. Productivity

B. Utilization

C. Load ratio

D. Yield loss

Correct Answer: A. Productivity

Question 6

A process has 1,088 effective machine-hours and produces 2.5 good units per machine-hour. What is its effective good-unit capacity?

A. 435.2 units

B. 1,090 units

C. 2,720 units

D. 3,200 units

Correct Answer: C. 2,720 units

Question 7

If a product requires 0.4 standard labour-hours per unit and planned output is 3,000 units, what labour requirement is generated?

A. 750 hours

B. 1,200 hours

C. 3,000 hours

D. 7,500 hours

Correct Answer: B. 1,200 hours

Question 8

What is the principal reason for aggregating products into families in aggregate planning?

A. To eliminate all forecast uncertainty

B. To plan at a manageable level while preserving relevant resource and demand characteristics

C. To ensure every product has identical processing time

D. To replace detailed scheduling permanently

Correct Answer: B. To plan at a manageable level while preserving relevant resource and demand characteristics

Question 9

A process has a yield of 0.95. Approximately how much gross production is needed to obtain 950 good units?

A. 902.5 units

B. 950 units

C. 1,000 units

D. 1,050 units

Correct Answer: C. 1,000 units

Question 10

Which practice best supports auditability of planning data?

A. Overwriting the original forecast with every planner adjustment

B. Storing only the final approved plan

C. Recording source, version, owner and reason for each adjustment

D. Combining actuals, assumptions and decisions in one unlabelled field

Correct Answer: C. Recording source, version, owner and reason for each adjustment