Supply Chain Network Design and Capacity Strategy

Introduction

Supply chain network design determines the long-term structure through which products, information, cash and services move from suppliers to customers. It includes decisions about the number, location, ownership, role and capacity of facilities; the allocation of demand to facilities; transportation modes; sourcing arrangements; and the timing and scale of expansion. Unlike day-to-day logistics decisions, network decisions are capital-intensive, difficult to reverse and consequential for cost, customer service, resilience, taxation and sustainability.

A strategically sound network must therefore answer several interconnected questions:

• Where should plants, warehouses, distribution centres and fulfilment facilities be located?
• How much capacity should each facility provide, and when should it be expanded?
• Which activities should be performed internally and which should be outsourced?
• Should the network be centralized, regionalized, nearshored or reshored?
• Where should inventory be positioned, and which activities should be postponed until demand is clearer?
• How should the design respond to disruption, regulation, tax differences and uncertain demand?

Network design is not simply a cost-minimization exercise. A low-cost network may expose the firm to excessive geopolitical, climate, labour, cyber, supplier or transportation risk. The objective is to identify a portfolio of feasible designs that balances total landed cost, service, risk, flexibility, capital intensity and strategic control.

Learning Objectives

After studying this chapter, students should be able to:

1. Explain the strategic role and major decisions of supply chain network design.
2. Compare centralized, regional, nearshore, reshored and hybrid network structures.
3. Apply facility-location, capacity, make-or-buy and postponement logic.
4. Formulate and interpret basic network optimization models.
5. Evaluate alternatives using cost, service, risk, tax and scenario analysis.
6. Recommend a network configuration while recognizing governance, implementation and sustainability constraints.

1. The Strategic Logic of Network Design

A supply chain network consists of nodes and links. Nodes include suppliers, plants, cross-docks, warehouses, retailers, service centres and customers. Links represent transport, information, financial and contractual relationships. The network may be represented as a directed graph, with facilities as vertices and flows as arcs.

Network design has four characteristics that distinguish it from operational planning:

Long time horizon. Facilities and contracts may remain in place for 10–30 years. The decision must therefore consider expected future demand, not only current volume.

High fixed cost. Land, buildings, automation, tooling, information systems and regulatory approvals create costs that do not vary directly with short-term throughput.

Interdependence. A plant location affects transportation, inventory, tax, labour, supplier access, customer lead time and risk simultaneously.

Irreversibility. A poor location cannot usually be corrected quickly. Exit costs, stranded assets and reputational effects may be substantial.

A useful strategic principle is to design for the total system rather than optimize individual functions. For example, locating a plant in a low-wage country may reduce conversion cost but increase ocean freight, pipeline inventory, quality costs, customs exposure and response time.

Network design decisions by level

The levels must be consistent. A strategic design with insufficient warehouse capacity cannot be repaired through better daily scheduling alone.

2. Facility Location and Configuration

2.1 Location criteria

Facility location should be assessed through a multi-criteria lens. Important factors include:

• proximity to demand and suppliers;
• transport infrastructure, port and airport access;
• labour availability, productivity, wages and industrial relations;
• land, utilities, construction and operating costs;
• energy availability and reliability;
• political, legal and macroeconomic stability;
• customs, tariffs, duties and tax treatment;
• environmental constraints and climate exposure;
• proximity to technical skills, universities and service providers;
• security, cyber exposure and business continuity;
• cultural compatibility and management feasibility.

A location should not be selected solely because it has the lowest nominal labour cost. The relevant comparison is usually total landed cost and risk-adjusted economic value.

2.2 Centralization and regionalization

Centralization consolidates inventory, production or distribution in fewer facilities. It can reduce duplicated fixed costs and exploit economies of scale. It may also reduce safety stock because demand variability can be pooled. However, it increases average delivery distance and may create a single point of failure.

Regionalization places facilities closer to important customer or market regions. It generally improves delivery speed, responsiveness and resilience to border disruption, but it duplicates assets and may reduce utilization.

The choice is often product-specific. Stable, predictable products with high scale economies may suit centralization. Highly customized, urgent or regulation-sensitive products may require regional capacity.

2.3 Postponement

Postponement delays product differentiation, packaging, configuration or final assembly until demand information becomes more reliable. A firm may produce a common base product centrally and complete localization in regional facilities.

Postponement is attractive when:

• demand for final variants is uncertain;
• the common platform has significant scale economies;
• final customization is technically feasible late in the process;
• customers value variety and responsiveness;
• the cost of late customization is lower than the cost of holding many finished variants.

Postponement is not universally beneficial. It may require modular product architecture, flexible equipment, skilled labour and reliable information systems. It can also increase final-stage cost or reduce manufacturing efficiency.

3. Capacity Strategy

Capacity is the maximum feasible output of a resource over a defined period under specified operating conditions. Capacity decisions include scale, timing, flexibility and location.

3.1 Capacity concepts

Design capacity is the theoretical maximum output. Effective capacity is the output achievable under normal product mix, maintenance, labour and quality conditions. Actual output is what the facility produces. Capacity utilization is commonly expressed as:

where u is utilization, Q is actual output per period and C_e is effective capacity per period. Both numerator and denominator must use the same units and time period. Utilization above 100% indicates that the stated effective-capacity estimate is inconsistent with actual operations or that overtime, temporary labour or other contingent capacity is being used.

A firm must distinguish utilization from productivity. A plant can have high utilization but poor productivity if excessive labour, scrap or overtime is required.

3.2 Capacity cushion

Capacity cushion is the proportion of capacity intentionally held above expected demand. A simple expression is:

where c is the capacity cushion, C_e is effective capacity and D_p is planned peak-period demand. The measure is useful only when demand and capacity are measured consistently. A positive cushion supports resilience and growth; too much cushion increases unit cost and asset underutilization.

Capacity strategy can be classified as:

• Lead strategy: add capacity before demand materializes.
• Lag strategy: add capacity after demand is demonstrated.
• Match strategy: add capacity in smaller increments as demand develops.
• Flexible or contingent strategy: combine core capacity with overtime, subcontracting, temporary facilities or flexible technology.

Lead capacity improves service and protects market share but carries demand risk. Lag capacity limits idle assets but may cause lost sales and congestion. Match strategies balance the two but require disciplined forecasting and investment timing.

3.3 Economies and diseconomies of scale

Average fixed cost falls as volume increases because fixed cost is spread over more units. However, congestion, coordination complexity, labour scarcity, transport distance and management difficulty may eventually produce diseconomies of scale. The optimal facility size is therefore not necessarily the largest technically possible facility.

Capacity should also be evaluated by bottleneck. A plant with 100,000 units of assembly capacity may effectively have only 70,000 units if a testing or packaging operation is limited to 70,000 units.

4. Quantitative Network Models

4.1 Total landed cost

A basic comparison of network alternatives can use total landed cost:

where:

• TLC = total landed cost per planning period;
• F = facility fixed and operating cost;
• P = purchasing or production cost;
• T = transportation cost;
• I = inventory carrying cost;
• H = handling and warehousing cost;
• D = duties, tariffs and customs costs;
• R = expected disruption, quality and recovery cost;
• X = other relevant costs, such as taxes, compliance and disposal.

The terms must be defined consistently. For example, taxes that are recoverable should not be treated as economic cost, while non-recoverable taxes should be included. Expected disruption cost is not a guaranteed cash expense; it is a probability-weighted estimate.

4.2 Facility-location model

A simplified mixed-integer facility-location model selects facilities and allocates flows:

subject to:

Here, f_j is the fixed cost of opening facility j, y_j is 1 if facility j is opened and 0 otherwise, c_ij is the variable cost of serving customer i from facility j, x_ij is the quantity shipped, d_i is demand of customer i, and K_j is facility capacity. The model assumes known demand, linear variable costs, divisible flow and feasible transportation. Real models may add multiple products, transport modes, service constraints, supplier limits, taxes, emissions and disruption scenarios.

Worked example: two possible warehouses

A company serves two markets. Market A demands 600 units and Market B demands 400 units per week. Warehouse North has weekly fixed cost of £4,000 and capacity of 700 units. Warehouse South has weekly fixed cost of £3,000 and capacity of 600 units. Variable serving costs are:

If only North opens, capacity is 700 units, below total demand of 1,000; the design is infeasible. If only South opens, it is also infeasible. If both open, the lowest-cost feasible allocation sends 600 units of A to North and 400 units of B to South. Total weekly cost is:

£4,000 + £3,000 + (600 × £5) + (400 × £4) = £12,800.

This result illustrates that fixed-cost minimization alone is inadequate. Both facilities are required because of capacity, even though one facility might appear cheaper for a particular market.

4.3 Break-even analysis

When comparing two alternatives, the break-even volume is useful if costs are linear:

where F₁ and F₂ are fixed costs, v₁ and v₂ are variable costs per unit, and Q^* is the volume at which total costs are equal. The formula applies only when the alternatives have comparable output, constant unit variable costs and no capacity discontinuities. If capacity limits differ, the break-even result must be checked against feasible ranges.

5. Make-or-Buy and Ownership Strategy

Make-or-buy decisions determine whether an activity is performed internally or purchased from an external provider. Relevant considerations include:

• avoidable internal costs;
• supplier price and switching costs;
• tooling, technology and capital requirements;
• quality and process capability;
• intellectual property and data security;
• supply continuity and supplier financial health;
• volume flexibility;
• learning, innovation and strategic control;
• labour, tax, environmental and regulatory obligations.

The correct financial comparison is not internal full cost versus supplier price. It is usually avoidable internal cost plus opportunity cost compared with the external total cost. Sunk costs should not influence a future decision unless they affect future cash flows.

Outsourcing can convert fixed cost into variable cost and provide specialist capability. It can also increase dependence, reduce process visibility and expose the firm to supplier opportunism or capacity competition. A hybrid approach may retain strategically important processes while outsourcing standardized or capacity-variable work.

6. Global, Regional, Nearshore and Reshored Networks

Global sourcing and production

Global networks may exploit lower costs, specialized capabilities, scale and market access. Their disadvantages include long lead times, inventory exposure, customs complexity, geopolitical risk, currency volatility, compliance requirements and reduced visibility.

Nearshoring

Nearshoring relocates selected activities to a geographically proximate country or region. It typically reduces transit time and time-zone separation relative to distant offshore sourcing while retaining some cost advantages. Nearshoring is most compelling when responsiveness, engineering collaboration or regional market access is important.

Reshoring

Reshoring returns activities to the firm’s home country. It may improve control, employment, intellectual-property protection, resilience and customer responsiveness. It may also increase wages, land cost and regulatory expense. Reshoring should therefore be justified through total system economics and strategic risk, not political slogans alone.

Tax and regulatory considerations

Network analysis should distinguish accounting profit from economic value. Transfer pricing, customs valuation, withholding taxes, tax holidays, value-added taxes, permanent-establishment rules and rules of origin can materially change location economics. Tax assumptions must be legally reviewed and stress-tested because incentives may expire or be conditional on employment and investment commitments.

A network should never be designed to evade tax, labour, environmental or customs obligations. Governance requires transparent documentation of assumptions, arm’s-length pricing, accurate product classification and auditable intercompany transactions.

7. Risk, Resilience and Scenario-Based Design

Risk is the possibility that uncertain events cause adverse deviation from objectives. Common network risks include supplier failure, port closure, extreme weather, cyberattack, epidemic, sanctions, political instability, demand collapse and utility interruption.

A robust design can include multiple suppliers, geographic diversification, alternative transport modes, strategic inventory, modular products, flexible capacity and qualified emergency providers. Resilience is not identical to redundancy. Redundancy duplicates resources; resilience also includes visibility, response capability, recovery speed and adaptation.

Expected-value analysis

If scenario s occurs with probability p_s and produces cost C_s, expected cost is:

with ∑_s p_s = 1. Expected value supports comparison but does not capture the full effect of rare catastrophic events, risk aversion or liquidity constraints. Managers should therefore examine worst-case impact, recovery time and service consequences in addition to expected cost.

A scenario model may compare a single global plant, two regional plants and a hybrid arrangement under normal demand, demand surge, port closure and geopolitical disruption. The preferred option may not have the lowest normal-state cost; it may create superior risk-adjusted value.

Service constraints

Cost-minimizing models may select an unacceptable design unless service is explicitly represented. Constraints may require that:

• a specified percentage of demand is delivered within a promised time;
• maximum customer distance or transport time is respected;
• critical customers have dual sourcing;
• recovery time after disruption remains below a threshold.

Service can also be included through penalty costs for late delivery, lost sales or customer dissatisfaction. Penalty estimates should be evidence-based rather than arbitrary.

8. A Structured Network-Design Process

A professional network study generally follows these stages:

1. Define the decision scope. Establish products, markets, time horizon, facilities, constraints and decision rights.
2. Build a fact base. Validate demand, cost, lead time, capacity, service, tax, emissions and risk data.
3. Segment products and customers. Avoid applying one network logic to products with different value, volatility or urgency.
4. Develop strategic alternatives. Include centralized, regional, hybrid, nearshore, reshored and postponement options.
5. Model total economics. Include fixed, variable, transportation, inventory, tax, duty, quality, carbon and risk costs.
6. Test scenarios. Vary demand, exchange rates, tariffs, disruption probability, fuel prices and capacity availability.
7. Assess qualitative factors. Consider management complexity, workforce, innovation, compliance and stakeholder effects.
8. Select a portfolio and implementation path. Use staged investment, pilots and decision gates where uncertainty is high.
9. Monitor performance. Track cost, service, utilization, resilience, emissions and assumption changes.

A model is a decision-support instrument, not a substitute for managerial judgment. Its output should be challenged through sensitivity analysis and independent validation.

9. Illustrative Case: Orion Appliances

The following is a fictional illustrative case.

Background

Orion Appliances sells small kitchen appliances in North America and Europe. It currently manufactures all products in one Asian plant and ships finished goods by ocean freight. Annual demand is 1.2 million units, with high seasonal variation. The network has low conversion cost but experiences long replenishment lead times, port disruption exposure and excess inventory of finished variants.

Problem

The board is considering three alternatives:

• retain the current global plant;
• add a European regional assembly and postponement centre;
• establish a North American nearshore plant while retaining the Asian plant for base production.

The decision must consider cost, service, capacity, tax, disruption and sustainability.

Analysis

The supply-chain team segments products into two groups. Stable, high-volume products account for 70% of demand and can remain globally produced. Custom-colour and region-specific products account for 30% and have volatile demand. The team estimates that late packaging and final assembly can be performed in regional centres.

The current design has the lowest normal conversion cost but requires finished-goods inventory for each variant. The European postponement option reduces variant inventory and improves response time. The North American nearshore option improves service and provides a second production source, but its labour and facility costs are higher.

The team constructs a mixed-integer model with facility fixed costs, production costs, transport, inventory, duties, carbon cost and capacity. Four scenarios are tested: baseline demand, 20% demand growth, a 30-day Asian port closure and a tariff increase on imported finished goods.

Approach and solution

The recommended design is hybrid. Stable common platforms remain in Asia, while final configuration is postponed in Europe and North America. A limited nearshore production line is established for critical products rather than duplicating the full Asian plant. The company also qualifies a second component supplier and maintains contractual surge capacity.

The recommendation is not based on the lowest baseline cost. It is based on an acceptable cost premium in exchange for shorter lead times, lower variant inventory, improved tariff flexibility and reduced dependence on one finished-goods source. The implementation is staged: first deploy postponement in Europe, then add North American capacity after service and demand assumptions are validated.

Outcome and learning points

The case demonstrates five principles:

1. Product segmentation prevents overdesigning the entire network for the most volatile items.
2. Postponement can reduce finished-goods variety without requiring full regional manufacturing.
3. Capacity diversification may be valuable even when it raises normal-state cost.
4. Scenario analysis reveals options that appear unattractive under a single forecast.
5. Staged investment reduces the risk of committing irreversible capital too early.

10. Technology, Data and AI in Network Design

Digital twins can represent facilities, flows, inventory, capacity and scenarios in an integrated model. Geographic information systems support spatial analysis of customers, suppliers, infrastructure, hazards and travel time. Optimization software solves location, allocation and capacity models; simulation tests variability and congestion; control towers improve visibility after implementation.

AI and machine learning can support demand forecasting, disruption detection, lead-time estimation and scenario generation. However, AI does not eliminate the need for causal reasoning, data governance or managerial accountability. Historical data may encode past network bias, while rare disruptions are underrepresented in training data. Models should therefore be explainable enough for governance, tested for sensitivity and supplemented with expert-designed stress scenarios.

11. Limitations, Ethics and Sustainability

Network models simplify reality. Common limitations include inaccurate demand forecasts, omitted transition costs, assumed linear rates, unavailable risk data, double-counted savings and failure to represent human behaviour. Results should be presented as ranges or scenarios when uncertainty is material.

Ethical network design includes responsible labour standards, safe working conditions, fair supplier treatment, privacy protection, anti-corruption controls and respect for indigenous or community rights. A location that is financially attractive because it externalizes pollution or exploits vulnerable labour is not strategically responsible.

Sustainability analysis should consider transport emissions, energy mix, facility construction, packaging, waste and product end-of-life. Carbon reduction should be measured across the relevant life cycle and not achieved by shifting emissions to an unmeasured tier. Trade-offs may arise: regional facilities can reduce transport distance but increase building energy use and reduce production scale efficiency.

Best-Practice Principles

• Optimize total landed cost, not isolated purchase or labour cost.
• Use demand and product segmentation.
• Make capacity flexible where uncertainty is high.
• Include service, resilience, tax, duties and sustainability explicitly.
• Separate sunk costs from relevant future costs.
• Test normal, adverse and extreme scenarios.
• Validate data and document assumptions.
• Use postponement when modularity and demand uncertainty justify it.
• Prefer staged investments when decisions are difficult to reverse.
• Establish governance for model ownership, review and accountability.

Chapter Summary

Supply chain network design establishes the long-term structure of facilities, flows, capacity and ownership. Effective design balances total landed cost with service, risk, flexibility, tax, sustainability and strategic control. Facility location requires consideration of demand, suppliers, transport, labour, infrastructure, regulation and disruption exposure. Centralization provides scale and inventory pooling, whereas regionalization improves responsiveness and resilience. Capacity strategy may lead, lag or match demand, and must account for effective capacity, bottlenecks and capacity cushion.

Make-or-buy decisions should compare relevant avoidable costs and strategic consequences. Nearshoring and reshoring can improve responsiveness and control but may increase operating cost. Postponement delays differentiation and can reduce variant inventory. Facility-location and mixed-integer models provide structured analysis, but scenario testing and managerial judgment remain essential. A robust network is not necessarily the cheapest under normal conditions; it is one that delivers acceptable economic and strategic performance across plausible futures.

Key Terms

Facility location: Selection of geographic sites for supply-chain activities.

Network topology: The structure of nodes and links through which flows move.

Centralization: Concentration of activities or inventory in relatively few facilities.

Regionalization: Placement of facilities close to major geographic markets.

Capacity: Maximum feasible output of a resource over a defined period.

Design capacity: Theoretical maximum output under ideal conditions.

Effective capacity: Output achievable under normal operating conditions.

Capacity cushion: Capacity deliberately held above planned demand.

Capacity utilization: Actual output divided by effective capacity.

Postponement: Delaying differentiation or customization until demand is clearer.

Make-or-buy: Decision to perform an activity internally or purchase it externally.

Nearshoring: Relocating an activity to a geographically proximate country or region.

Reshoring: Returning an activity to the firm’s home country.

Total landed cost: Complete economic cost of supplying a product to its destination.

Scenario analysis: Evaluation of decisions under alternative future conditions.

Resilience: Ability to prepare for, absorb, recover from and adapt to disruption.

Capacity flexibility: Ability to vary output, resources or configuration as conditions change.

Mixed-integer model: Optimization model containing continuous and discrete decision variables.

Inventory pooling: Reduction in aggregate safety stock achieved by serving demand from a shared location.

Risk-adjusted value: Economic evaluation incorporating uncertainty and potential adverse outcomes.

Review Questions

Short Answer Questions

1. Why are network-design decisions more difficult to reverse than operational logistics decisions?
2. Distinguish design capacity, effective capacity and actual output.
3. What are two advantages and two disadvantages of centralization?
4. Define postponement and identify one condition under which it is attractive.
5. Why should make-or-buy analysis use avoidable cost rather than full historical cost?

Long Answer Questions

1. Explain how a firm should evaluate a facility location using cost, service, risk, tax and sustainability criteria.
2. Compare lead, lag, match and flexible capacity strategies, including the conditions favouring each.
3. Discuss the strategic differences between global sourcing, nearshoring and reshoring.
4. Explain the structure, assumptions and limitations of a basic mixed-integer facility-location model.
5. Describe a complete professional process for designing and implementing a supply chain network.

Analytical / Application Questions

1. A company is considering one centralized warehouse or two regional warehouses. Design a comparison framework that includes fixed cost, transport, inventory pooling, service and disruption risk. State what data would be required.
2. A product has uncertain regional demand and can be customized either at the factory or near customers. Recommend whether postponement should be investigated, and identify the operational and technology prerequisites.
3. Construct three disruption scenarios for a globally sourced product and explain how expected cost, service impact, recovery time and strategic resilience should influence the network recommendation.

Multiple Choice Questions with Answers

Question 1

Which characteristic most distinguishes strategic network design from operational logistics planning?

A. It involves only daily shipment scheduling.

B. It is generally capital-intensive and difficult to reverse.

C. It excludes customer-service considerations.

D. It always minimizes transportation distance.

Correct Answer: B. It is generally capital-intensive and difficult to reverse.

Question 2

A facility has effective capacity of 10,000 units per month and actual output of 8,000 units. Its utilization is: u = Q/C_e

A. 20%

B. 40%

C. 80%

D. 125%

Correct Answer: C. 80%

Question 3

Which situation most strongly supports postponement?

A. Stable demand for one standard product

B. Highly uncertain demand for multiple final variants

C. A process that cannot be modularized

D. Very long customization time after customer order

Correct Answer: B. Highly uncertain demand for multiple final variants

Question 4

In a basic facility-location model, the binary variable y_j usually indicates whether:

A. Customer j places an order

B. A shipment is late

C. Facility j is opened

D. Demand is forecast accurately

Correct Answer: C. Facility j is opened

Question 5

What is a principal advantage of centralization?

A. Guaranteed elimination of disruption risk

B. Inventory pooling and scale economies

C. Shorter delivery distance to every customer

D. No requirement for capacity planning

Correct Answer: B. Inventory pooling and scale economies

Question 6

Which cost should normally be excluded from a make-or-buy comparison if it cannot be changed by the decision?

A. Avoidable internal labour cost

B. External supplier price

C. Relevant opportunity cost

D. Sunk cost

Correct Answer: D. Sunk cost

Question 7

If a disruption scenario has probability 0.2 and cost £500,000, its contribution to expected cost is:

A. £100,000

B. £250,000

C. £500,000

D. £2,500,000

Correct Answer: A. £100,000

Question 8

Nearshoring generally refers to:

A. Moving all production to the lowest-wage country

B. Moving activities to a geographically proximate country or region

C. Returning production only to the home country

D. Eliminating external suppliers

Correct Answer: B. Moving activities to a geographically proximate country or region

Question 9

Which statement about a low-cost network is most accurate?

A. It is automatically the most resilient network.

B. It should be evaluated using total landed cost and risk, not nominal labour cost alone.

C. It never requires scenario analysis.

D. It always has the shortest customer lead time.

Correct Answer: B. It should be evaluated using total landed cost and risk, not nominal labour cost alone.

Question 10

A capacity cushion is best described as:

A. Actual output divided by transport cost

B. Planned demand divided by design capacity

C. Capacity held above planned demand

D. Inventory held at a customer location

Correct Answer: C. Capacity held above planned demand