Network Design, Economics and Performance Management

Introduction

Cold-chain network design concerns the configuration and governance of facilities, transport links, inventory positions, suppliers, technologies and service commitments used to preserve temperature-sensitive products. Unlike conventional logistics, cold chains manage a coupled physical and commercial system: product quality deteriorates with time and temperature, capacity is often constrained by refrigeration and handling requirements, and regulatory failures may make inventory unsaleable even when it is physically present.

Strategic decisions therefore include the location and number of distribution centres, sourcing and supplier portfolios, make-or-buy choices, temperature-controlled capacity, inventory positioning, transport modes and customer service models. These decisions should be evaluated using total cost to serve, lifecycle economics, risk-adjusted performance and the required service level—not purchase price alone.

Learning Objectives

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

1. Design and compare cold-chain network configurations.
2. Apply total-cost-to-serve and lifecycle-costing logic.
3. Evaluate sourcing, outsourcing, capacity and service-model alternatives.
4. Develop contracts and service-level agreements (SLAs) for perishable products.
5. Construct balanced performance systems covering quality, cost, service, waste and compliance.
6. Critically assess uncertainty, resilience, sustainability and ethical governance in network decisions.

1. The Strategic Logic of a Cold-Chain Network

A network is a set of nodes and links. Nodes may include farms, processors, pack houses, cross-docks, refrigerated warehouses, pharmacies, hospitals, retailers and reverse-logistics points. Links include road, rail, air and maritime transport, supported by information, energy and regulatory controls.

The central design trade-off is aggregation versus responsiveness. A small number of large facilities can reduce fixed cost and improve equipment utilization, but may increase transport distance, lead time and disruption exposure. A larger number of regional facilities can improve freshness and customer responsiveness, but duplicates refrigeration, quality systems, labour and inventory.

Cold-chain design must also recognize product heterogeneity. Frozen seafood, fresh produce, vaccines and chilled meat differ in temperature range, shelf life, regulatory status, packaging, demand uncertainty and tolerance for delay. A single network may therefore require product-family segmentation rather than one universal design.

Important decision dimensions include:

2. Facility Location and Network Configuration

2.1 Location criteria

Location analysis should combine quantitative and qualitative factors:

• demand density and geographic dispersion;
• travel time, road reliability and access to airports or ports;
• electricity reliability, backup generation and water availability;
• land, labour, taxes and construction cost;
• proximity to suppliers and customers;
• food, pharmaceutical and environmental regulations;
• climate hazards, flood exposure and political risk;
• availability of qualified refrigeration technicians;
• opportunities for renewable energy and heat recovery.

A facility that is geographically close but exposed to outages may provide poorer effective service than a more distant facility with reliable infrastructure. Thus, nominal distance should be supplemented by time reliability and risk-adjusted cost.

2.2 Cost-to-serve formulation

For a candidate network, annual total cost can be represented as:

where:

• F = annualized facility and equipment cost;
• H = handling and operating cost;
• T = transportation cost;
• I = inventory carrying cost;
• Q = quality, monitoring and compliance cost;
• W = waste, spoilage and markdown cost;
• R = expected disruption, recall and failure cost;
• C = customer-service and administrative cost.

All terms must be measured over the same period and in the same currency. If alternatives provide different service levels, cost comparison is incomplete unless service differences are explicitly valued or constrained.

A useful unit measure is:

where CTS_u is cost to serve per saleable unit and U is the number of units delivered in acceptable condition. Using saleable units rather than shipped units prevents low-waste and high-waste networks from appearing equivalent.

2.3 Facility-location models

A simplified capacitated facility-location model selects facilities and assigns demand to them. Let y_j equal 1 if facility j is opened, and 0 otherwise. Let x_ij be the quantity shipped from facility j to customer zone i. A representative objective is:

Here f_j is fixed annual cost, c_ij is variable transport and handling cost per unit, h_j is inventory cost, I_j is average inventory, p_i is the penalty or economic value of shortage at customer zone i, and s_i is shortage quantity.

Typical constraints are:

where d_i is demand and K_j is facility capacity. The model assumes that demand, costs and capacity are adequately represented for the planning period. In practice, scenario or robust formulations are needed for uncertain demand, travel time, energy price and disruption.

The model should not be treated as an automatic answer. Its value depends on the quality of input data, the definition of capacity, the treatment of perishability and the realism of service constraints.

3. Lifecycle Economics and Investment Decisions

3.1 Lifecycle cost

A refrigerated facility may have a low construction cost but high energy consumption, maintenance requirements and failure risk. Lifecycle costing captures costs over the asset’s economic life:

where C₀ is initial capital cost, C_t operating cost in year t, M_t maintenance cost, E_t energy cost, Q_t quality and compliance cost, R_t risk-related cost, S_t savings or residual operating benefits, V_n residual value at year n, and r the discount rate.

The equation assumes cash flows are known or represented by expected values and occur at year-end. If costs occur monthly or risk varies materially, a more granular discounted-cash-flow model is appropriate. Nominal cash flows should be discounted using a nominal rate; real cash flows should use a real rate.

For uncertain outcomes, expected cost may be estimated as:

where P_k is the probability of scenario k and C_k is its cost. Expected value alone can conceal catastrophic low-probability outcomes, so downside analysis, sensitivity analysis and risk limits are also required.

3.2 Make, buy, lease or outsource

Outsourcing may provide specialist refrigeration expertise, geographic reach and variable capacity. It may also create dependency, weaker visibility, data ownership concerns and reduced control over handling practices. The correct comparison is between relevant avoidable costs and the full economic cost of the alternative.

A provider should be assessed on:

• validated temperature capability;
• quality-management maturity and audit history;
• financial stability and business continuity;
• information-system integration;
• claims and insurance arrangements;
• labour practices and regulatory compliance;
• scalability and multi-client conflict risk;
• ability to manage recalls, excursions and emergencies.

A common error is to compare a provider’s transport rate with an internal budget that excludes supervision, equipment depreciation, insurance, failed deliveries and quality personnel. Conversely, outsourcing may appear expensive if internal assets are treated as free. Both alternatives require comparable cost boundaries.

4. Capacity, Inventory and Service Models

4.1 Capacity planning

Required capacity is not simply average demand. A practical capacity assessment considers peak demand, inbound and outbound synchronization, dwell time, pallet configuration, temperature zones, sanitation downtime, maintenance downtime and emergency reserve.

If annual throughput is Q units, operating days are D, peak factor is α, and planned utilization is u, average required daily capacity is:

The formula assumes α scales average demand to the relevant peak and that u is below 1.0 to preserve operational flexibility. It does not replace detailed modelling of arrival patterns, dock capacity or refrigeration load.

Excess utilization can cause queueing, temperature exposure and missed dispatch windows. Capacity should therefore be measured separately for storage positions, dock doors, blast-freezing capability, picking labour, vehicle availability and qualified inspection capacity.

4.2 Inventory and perishability

The economic order quantity formula is often unsuitable for short-life products because it assumes replenishment, demand and holding conditions that may not hold. Nevertheless, the classic expression is useful as a conceptual benchmark:

where D is annual demand, S is ordering cost per order and H is annual holding cost per unit. It assumes stable demand, constant unit cost, instantaneous replenishment and no spoilage. Perishable inventory requires modifications for shelf life, age distribution, service risk and markdown or disposal cost.

FIFO is insufficient where remaining shelf life differs significantly. FEFO—first-expired, first-out—prioritizes the product with the earliest expiry date. Its effectiveness depends on reliable lot, expiry and temperature history data.

4.3 Service-model choices

Service models may range from centralized next-day delivery to regional same-day replenishment, vendor-managed inventory, scheduled milk runs or direct store delivery. They should be evaluated against a defined customer promise, such as delivery within a time window, minimum remaining shelf life, order completeness and temperature compliance.

A useful service measure is fill rate:

This should be supplemented by on-time-in-full (OTIF), because a shipment delivered late may still be counted as complete. For perishables, remaining shelf life at receipt and temperature excursion rate are often more meaningful than fill rate alone.

5. Contracts and Service-Level Agreements

A cold-chain contract should allocate responsibilities, incentives and evidence requirements. Essential provisions include:

1. product specifications and temperature ranges;
2. packaging, loading, pre-cooling and calibration responsibilities;
3. custody-transfer points and acceptance criteria;
4. permitted excursion duration and escalation thresholds;
5. data ownership, sensor access and retention period;
6. delivery windows, shelf-life guarantees and order accuracy;
7. claims, recall, insurance and liability rules;
8. audit rights and corrective-action procedures;
9. business-continuity and contingency obligations;
10. pricing indexation, fuel and energy adjustment mechanisms;
11. ethical, labour, environmental and anti-corruption requirements;
12. termination, transition and data-portability provisions.

SLA metrics must be operationally controllable and objectively measured. A penalty for temperature excursions is weak if sensor placement, calibration, download responsibility and causation are undefined. Contracts should distinguish provider-caused failures from customer-caused loading delays, force majeure and equipment faults that were reported but not repaired.

Incentive design should avoid encouraging unsafe behaviour. For example, a carrier rewarded only for punctuality may reduce pre-trip checks or accept loads before adequate pre-cooling. Balanced incentives should combine punctuality with temperature integrity, documentation, safe work and product acceptance.

6. Performance Management and Balanced Scorecards

A performance system translates strategy into measurable behaviour. A balanced cold-chain scorecard should contain at least five perspectives.

A KPI must have a definition, numerator, denominator, owner, data source, frequency, target and escalation rule. For example:

The denominator must be specified; otherwise one site may report waste against receipts while another reports it against sales. KPI gaming is reduced through independent data validation, periodic audits and reconciliation with financial and inventory records.

A composite score can be used cautiously:

where z_m is a normalized score for metric m and w_m is its weight. Normalization must preserve direction: for a “lower is better” measure, a lower value should receive a higher score. Composite scores should never conceal a critical failure; a serious food-safety or pharmaceutical-compliance breach should trigger a separate override or stop rule.

Leading indicators include preventive-maintenance completion, calibration status, training completion and near misses. Lagging indicators include spoilage, complaints, recalls and rejected deliveries. Mature governance uses both.

7. Technology, Data and AI

IoT sensors, telematics, warehouse-management systems and electronic proof of delivery can create an event history across custody transfers. Useful data include temperature, humidity where relevant, door opening, location, shock, dwell time, vehicle status, lot number and expiry date.

AI can support demand forecasting, dynamic route planning, anomaly detection and predictive maintenance. However, a prediction is not a control. Models require representative data, validated thresholds, explainable exception logic and human accountability. Sensor drift, missing data, biased historical demand and cyberattack can produce false assurance. Access control, encryption, retention policies and incident response are therefore part of cold-chain governance.

8. Applied Case: NorthStar Fresh (Illustrative)

Background

NorthStar Fresh, an illustrative regional distributor of chilled prepared meals, serves 420 retail outlets from one national distribution centre. Demand has grown, but spoilage is 7.2%, OTIF is 88%, and summer electricity interruptions have caused repeated temperature excursions.

Problem

Management is considering three alternatives: retain the single centre, add two regional cross-docks, or outsource regional warehousing to a specialist 3PL. The board requires at least 96% OTIF, spoilage below 3.5% and documented temperature traceability.

Analysis

The team first defined cost to serve using facility, transport, inventory, energy, quality, waste and disruption costs. It then segmented stores by geography and demand volatility, mapped realistic travel time rather than straight-line distance, and tested normal, peak and disruption scenarios. A lifecycle review showed that the 3PL option had lower initial capital cost but higher contract and coordination cost. The regional cross-dock option reduced delivery distance but added handling events and required disciplined FEFO execution.

The team also discovered that the existing OTIF measure counted deliveries within the day but not within the agreed two-hour window. After redefining the metric, reported performance fell further, but the measure became commercially meaningful.

Approach and solution

NorthStar selected a hybrid design: retain the central facility for production-linked inventory, add one regional cross-dock near the densest demand cluster, and contract overflow refrigerated storage with a qualified 3PL. The contract required calibrated sensors, event-level data sharing, a minimum remaining shelf-life guarantee and joint root-cause reviews.

The company introduced FEFO scanning, pre-cooling verification, route risk alerts and a scorecard weighted toward temperature integrity and waste rather than delivery speed alone. Capacity was reserved for seasonal peaks, and a backup-power project was justified through lifecycle and expected-loss analysis rather than capital cost alone.

Outcome

In the first year of the redesigned operating model, the target case projected OTIF above 96%, spoilage below 3.5% and improved resilience. The principal lesson was not that decentralization is always superior. It was that network configuration, measurement definitions, contract design, data quality and operating discipline must be designed as one system.

9. Challenges, Ethics and Contemporary Issues

Cold-chain decisions have distributional consequences. A service model that prioritizes high-margin urban customers may reduce access for rural or vulnerable populations. Procurement decisions should consider food security, equitable access, worker safety and responsible labour practices.

Environmental performance requires more than minimizing kilometres. Refrigerant leakage, electricity source, packaging, rejected product, vehicle utilization and facility insulation all matter. A shorter route using inefficient equipment may have greater emissions than a longer route using consolidated, low-emission transport. Carbon claims should use transparent boundaries and avoid double counting.

Regulatory obligations vary by product and jurisdiction, but common principles include traceability, validated temperature control, sanitation, calibration, documented corrective action and recall capability. Firms should maintain a legal register and distinguish regulatory minimums from voluntary best practice.

The main limitations of quantitative network models are parameter uncertainty, oversimplified service assumptions, omitted behavioural responses and dependence on historical data. Robust practice combines optimization with site validation, scenario analysis, pilot implementation and post-decision review. Strategic flexibility—such as modular capacity, dual sourcing and interoperable data—may have value even when its immediate cost is higher.

Best-Practice Decision Sequence

1. Define product, shelf-life, regulatory and customer-service requirements.
2. Establish a common cost boundary and measure saleable output.
3. Segment demand, products and risk rather than averaging them indiscriminately.
4. Develop feasible network alternatives, including ownership and outsourcing choices.
5. Test capacity, quality, cost and resilience under multiple scenarios.
6. Evaluate lifecycle cost and downside risk, not only initial investment.
7. Translate the selected design into contracts, SLAs, data standards and escalation rules.
8. Implement a balanced scorecard with auditable definitions.
9. Pilot, measure and revise assumptions after operational learning.

Chapter Summary

Cold-chain network design is a strategic integration problem involving facility location, sourcing, capacity, inventory, transport, service and governance. The appropriate objective is total cost to serve, measured against saleable output and explicit service requirements. Facility-location models can structure alternatives, but their results depend on realistic capacities, time reliability, perishability and risk assumptions.

Lifecycle costing prevents low-capital alternatives from appearing artificially attractive when energy, maintenance, quality and disruption costs are ignored. Make-or-buy and outsourcing decisions require comparable cost boundaries and strong controls over custody, data and accountability. Capacity planning must address peaks, downtime and multiple bottlenecks, while FEFO and shelf-life measures are essential for perishable inventory.

Contracts and SLAs should specify temperature, shelf life, evidence, liability, corrective action and continuity. Balanced scorecards should cover quality, cost, service, waste, compliance and resilience, combining leading and lagging indicators. Technology and AI can improve visibility and prediction, but cannot replace validated controls, cybersecurity or managerial accountability. Finally, ethical, environmental and regulatory considerations must be incorporated into network economics rather than treated as afterthoughts.

Key Terms

Total cost to serve: The complete economic cost of delivering acceptable product and service to a customer.

Lifecycle costing: Evaluation of acquisition, operating, maintenance, risk and residual costs over an asset’s life.

Facility-location model: An analytical model selecting facility sites and assigning demand subject to cost and capacity constraints.

Capacitated facility location: A location model that limits assigned flow by facility capacity.

Saleable unit: A unit delivered in acceptable quality and condition for sale or use.

FEFO: First-expired, first-out inventory rotation.

OTIF: On-time-in-full delivery performance.

Service-level agreement: A documented agreement defining measurable service obligations and remedies.

Temperature excursion: Exposure outside the specified temperature range or duration.

Cross-dock: A facility where goods are rapidly transferred without prolonged storage.

Risk-adjusted cost: Cost incorporating expected or scenario-based consequences of failure and disruption.

Capacity utilization: Used capacity divided by available practical capacity.

Lifecycle cost: Discounted total cost over an asset’s economic life.

Fill rate: Proportion of ordered quantity delivered in full.

Balanced scorecard: A performance framework combining multiple strategic perspectives.

Leading indicator: A measure that signals future performance or emerging risk.

Lagging indicator: A measure of an outcome that has already occurred.

Traceability: Ability to track product identity, history, location and custody.

Resilience: Capability to absorb disruption, recover and continue serving customers.

Review Questions

Short Answer Questions

1. Why can a decentralized cold-chain network improve service but increase cost?
2. Define total cost to serve and explain why saleable units are useful as its denominator.
3. What is the difference between FIFO and FEFO?
4. Identify four essential provisions of a cold-chain SLA.
5. Distinguish a leading KPI from a lagging KPI and give one example of each.

Long Answer Questions

1. Explain how facility location, inventory, transportation and service-level decisions interact in a perishable distribution network.
2. Discuss the advantages and limitations of using a capacitated facility-location model for cold-chain design.
3. Explain how lifecycle costing can change a refrigerated-facility investment decision.
4. Evaluate the strategic and operational risks of outsourcing cold storage and distribution to a 3PL.
5. Design a balanced cold-chain scorecard covering quality, cost, service, waste and compliance.

Analytical / Application Questions

1. A distributor must choose between one centralized facility and two regional facilities. Develop a decision framework that includes fixed cost, transport, inventory, waste, service and disruption risk. State the data required and the scenarios you would test.
2. A carrier reports 98% on-time delivery, but customer complaints about warm product are increasing. Diagnose how the KPI system and SLA may be defective, and propose revised measures and controls.
3. A company is considering an energy-efficient refrigeration system with higher capital cost but lower energy and spoilage costs. Explain how to evaluate the investment using lifecycle costing, sensitivity analysis and non-financial criteria.

Multiple Choice Questions with Answers

Question 1

Which cost category captures the economic impact of spoiled or downgraded product in a total-cost-to-serve analysis?

A. Facility cost

B. Waste cost

C. Administrative cost

D. Residual value

Correct Answer: B. Waste cost

Question 2

In a capacitated facility-location model, what does a binary variable yⱼ commonly indicate?

A. The quantity shipped to customer i

B. The probability of disruption

C. Whether facility j is opened

D. The average inventory age

Correct Answer: C. Whether facility j is opened

Question 3

Why is FEFO generally more appropriate than FIFO for many perishable products?

A. It always minimizes transport cost

B. It prioritizes the earliest expiry date

C. It eliminates the need for traceability

D. It guarantees zero spoilage

Correct Answer: B. It prioritizes the earliest expiry date

Question 4

If annual demand is 100,000 units, peak factor α is 1.2, operating days D are 250 and planned utilization u is 0.8, what daily capacity is required using K_day = αQ/(Du)?

A. 480 units

B. 500 units

C. 600 units

D. 750 units

Correct Answer: C. 600 units

Question 5

Which item is most appropriately classified as a leading indicator?

A. Recall cost

B. Customer complaint rate

C. Preventive-maintenance completion

D. Spoilage already recorded

Correct Answer: C. Preventive-maintenance completion

Question 6

What is the principal purpose of an SLA in cold-chain logistics?

A. To replace all regulatory requirements

B. To define measurable service obligations and responsibilities

C. To guarantee demand forecasts are correct

D. To eliminate the need for audits

Correct Answer: B. To define measurable service obligations and responsibilities

Question 7

In lifecycle costing, why must the discount rate be consistent with the type of cash flow?

A. To make all costs zero

B. To ensure nominal and real values are compared consistently

C. To remove maintenance costs

D. To avoid measuring residual value

Correct Answer: B. To ensure nominal and real values are compared consistently

Question 8

Which measure best complements OTIF for a temperature-sensitive product?

A. Remaining shelf life at receipt

B. Number of suppliers only

C. Warehouse floor area only

D. Vehicle purchase price only

Correct Answer: A. Remaining shelf life at receipt

Question 9

What is a major limitation of using expected cost alone for disruption analysis?

A. It cannot include any probabilities

B. It may conceal severe low-probability outcomes

C. It always overstates facility cost

D. It measures only service quality

Correct Answer: B. It may conceal severe low-probability outcomes

Question 10

Which practice most directly improves traceability?

A. Removing lot identifiers

B. Using event-level product and custody records

C. Measuring delivery only once per year

D. Counting all deliveries as on time

Correct Answer: B. Using event-level product and custody records