Operations, Economics and Supply-Chain Decision Making

Introduction

Additive manufacturing (AM) is often introduced as a production technology that builds parts layer by layer from digital data. In industrial practice, however, technical feasibility is only one part of the decision. A part may be printable but economically unattractive, operationally unreliable, or strategically unsuitable for the supply chain. Sound AM decision making therefore combines process engineering, cost accounting, capacity analysis, quality management, procurement, logistics, and risk assessment.

The central managerial question is not simply “Can this part be printed?” It is “Under which conditions should this part be printed, by which technology, at what location, in what quantity, and with what level of quality assurance?” This chapter develops methods for answering that question.

Learning Objectives

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

1. Construct a total-cost model for an AM production order.
2. Estimate machine, material, labor, energy, post-processing, quality, and overhead costs.
3. Apply make-or-buy and break-even analysis.
4. Evaluate machine capacity, utilization, bottlenecks, and production schedules.
5. Select an AM technology using technical, economic, and strategic criteria.
6. Assess distributed manufacturing, digital inventories, cybersecurity, and supply-chain resilience.
7. Identify uncertainty, operational risk, ethical obligations, and governance requirements in AM.

1. The AM Operations System

An AM operation is a system rather than an isolated machine. Its principal stages are:

1. demand definition and part qualification;
2. digital design, data preparation, and build planning;
3. material preparation and machine setup;
4. build production;
5. depowdering, support removal, cleaning, heat treatment, machining, or surface finishing;
6. inspection and documentation;
7. packaging, delivery, and feedback into design or process control.

The effective production lead time is therefore greater than the machine build time. It may be represented as:

Here, each term is a time component, measured in hours or days. The equation is a planning identity, not a universal physical law. It assumes that the listed components are defined consistently and that overlapping activities are either excluded or explicitly modeled. For example, inspection may occur during post-processing; if so, double counting must be avoided.

AM operations also have distinctive characteristics:

• High geometric freedom: complex geometry may be produced without dedicated tooling.
• Digital dependence: build files, machine parameters, material certificates, and inspection records are production assets.
• Variable economics: cost is influenced by part volume, build height, nesting, support requirements, batch size, and post-processing.
• Process sensitivity: powder condition, resin age, thermal history, orientation, and machine calibration affect quality.
• Potential postponement: production can occur close to the point and time of demand.

2. Building a Total-Cost Model

A useful cost model separates costs that vary with production from costs that exist because capacity has been installed. For a production order containing n parts, an approximate total cost is:

where:

• C_mat = material cost;
• C_machine = machine ownership or usage cost;
• C_labour = direct labor cost;
• C_energy = electricity and other utilities;
• C_post = post-processing cost;
• C_quality = inspection, testing, and quality-system cost;
• C_setup = setup and changeover cost;
• C_overhead = allocated indirect cost;
• C_logistics = packaging, transport, and handling cost.

If the order contains n conforming parts, the average unit cost is:

This average is meaningful only when n represents accepted parts. If a yield of y is expected, with 0 < y ≤ 1, the required production quantity is approximately:

This simple expression assumes independent, stable yield and does not capture correlated failures such as a failed build plate. In practice, batch-level failure risk should be modeled separately.

2.1 Machine Cost

Machine cost can be estimated using an hourly rate. A basic annualized rate is:

where:

• C_purchase is purchase price;
• V_residual is expected residual value;
• CRF is the capital-recovery factor;
• C_annual,maint is annual maintenance cost;
• C_annual,facility is annual facility-related cost;
• H_productive is productive machine hours per year.

For an interest rate i and economic life N years, the capital-recovery factor is:

The formula assumes a constant discount rate, end-of-period cash flows, and a uniform annual-equivalent treatment. If tax, inflation, depreciation rules, or financing structure are important, a discounted-cash-flow model is preferable.

Machine cost for a job is then:

where t_machine includes build, setup, unloading, and machine occupancy time, depending on the organization’s costing policy.

2.2 Material Cost

Material cost is not necessarily equal to the mass of material in the finished part. It may include supports, runners, failed builds, powder refresh, waste, and handling. A practical estimate is:

where m_feed is purchased material consumed, in kilograms, and p_mat is price per kilogram. For powder-bed processes, “consumed” must be defined carefully: surrounding powder may be reusable, but it may require sieving, blending, testing, or a mandated refresh ratio.

Material selection must also consider certification, storage, shelf life, moisture sensitivity, traceability, and safety. A lower purchase price can be misleading if it increases failed-build probability or post-processing effort.

2.3 Labor and Energy

Direct labor cost is commonly estimated as:

where h_j is labor time for activity j and r_j is the fully burdened hourly rate for the worker performing it. Activities include programming, loading, monitoring, depowdering, finishing, inspection, and documentation.

Energy cost is:

where E is energy consumed in kilowatt-hours and p_E is the electricity price per kilowatt-hour. If average power is P kilowatts and operating time is t hours:

The estimate should include auxiliary equipment such as heaters, inert-gas systems, compressors, ventilation, washing stations, and curing units when they are material to the decision. Energy is often a smaller cost than labor or machine ownership, but it becomes strategically important when electricity prices, carbon reporting, or facility constraints dominate.

2.4 Post-Processing and Quality Cost

Post-processing may include support removal, washing, blasting, tumbling, heat treatment, hot-isostatic pressing, machining, coating, dyeing, curing, and dimensional correction. Its cost should be estimated as a separate routing rather than hidden inside a generic overhead rate.

Quality cost includes prevention, appraisal, and failure costs:

• Prevention: qualification, training, calibration, process development, and controlled procedures.
• Appraisal: inspection, testing, scanning, and certificate generation.
• Internal failure: scrap, rework, repeated builds, and investigation.
• External failure: warranty replacement, recall, field repair, liability, and reputational damage.

For safety-critical components, quality cost cannot be minimized by reducing inspection alone. The required level of assurance is determined by risk, regulation, customer requirements, and the consequences of failure.

3. Worked Cost Example

An illustrative service bureau must produce 40 polymer parts. The following assumptions apply:

• material used: 3.2 kg at 48 currency units/kg;
• machine occupancy: 18 h at 32 currency units/h;
• direct labor: 10 h at 26 currency units/h;
• energy: 42 kWh at 0.20 currency units/kWh;
• post-processing: 40 parts at 4.50 currency units/part;
• inspection: 40 parts at 1.80 currency units/part;
• setup and file preparation: 95 currency units;
• packaging and logistics: 60 currency units;
• expected good-part yield: 95%.

The direct order cost before yield adjustment is:

• material: 3.2 × 48 = 153.60;
• machine: 18 × 32 = 576.00;
• labor: 10 × 26 = 260.00;
• energy: 42 × 0.20 = 8.40;
• post-processing: 40 × 4.50 = 180.00;
• inspection: 40 × 1.80 = 72.00;
• setup: 95.00;
• logistics: 60.00.

Thus:

Expected accepted output is:

Therefore, the expected cost per accepted part is approximately:

This calculation does not include a separate charge for failed-part disposal or the statistical uncertainty of the 95% yield assumption. If a failed build wastes the entire batch, a more conservative risk model is required.

4. Make-or-Buy and Break-Even Analysis

A make-or-buy decision compares the relevant costs and strategic consequences of internal production with external supply. Only avoidable internal costs should normally be compared with the supplier’s price. Allocated fixed overhead that will remain after outsourcing is not an avoidable saving.

Let internal variable cost per unit be C_make,var, avoidable fixed cost be F_make, supplier price per unit be C_buy, and purchased quantity be Q. Internal production is financially preferable when:

The break-even quantity is:

This formula applies when C_buy > C_make,var and when fixed and variable costs are reasonably stable. If the denominator is zero or negative, internal production does not achieve a financial break-even under the stated assumptions.

Example

A company can print a component internally with variable cost of 22 units per part and avoidable annual fixed cost of 18,000 units. A qualified supplier charges 40 units per part. Then:

Above 1,000 parts, internal production has the lower relevant cost, assuming equivalent quality, delivery, and risk. Below 1,000 parts, buying is financially preferable. However, supplier qualification, intellectual-property protection, lead time, minimum order quantities, and resilience may change the decision.

A complete make-or-buy evaluation should include a weighted strategic score covering quality capability, delivery reliability, capacity flexibility, data protection, regulatory compliance, sustainability, and supplier concentration.

5. Capacity Planning and Scheduling

AM capacity is constrained by more than nominal build volume. Important resources include machine hours, build-envelope volume, operator availability, post-processing stations, inspection equipment, qualified materials, and engineering personnel.

A simple utilization measure is:

where D is required workload and A is available capacity, expressed in common units such as machine-hours. If U > 1, the plan is infeasible unless capacity is added, demand is postponed, work is outsourced, or productivity improves.

Available machine-hours may be estimated as:

where N_m is the number of machines, H_shift is scheduled hours per machine, S is the number of shifts, and U_planned is the planned productive fraction after maintenance, setup, and expected downtime. The final factor should not be confused with actual utilization; it is a planning allowance.

Bottleneck analysis is essential. A printer with spare time does not create useful capacity if depowdering, heat treatment, machining, or inspection is overloaded. The effective capacity of the system is governed by the most constrained required resource.

Batching can reduce setup and handling costs, but excessive batching increases queue time and inventory. Scheduling should therefore balance:

• due dates and customer priority;
• build compatibility and nesting efficiency;
• material and machine qualification;
• changeover time;
• post-processing availability;
• maintenance windows;
• emergency and spare-capacity requirements.

6. Technology Selection

Technology selection should begin with requirements rather than machine preference. Key criteria include material properties, dimensional accuracy, surface finish, build volume, production rate, minimum feature size, anisotropy, certification, labor content, total cost, and environmental conditions.

A useful multi-criteria decision model is:

where S_k is the score of technology or supplier k, w_i is the weight of criterion i, and x_ik is the normalized performance of option k on criterion i. The weights should sum to one. Scores must be normalized consistently; otherwise, a criterion measured in large numerical units can dominate unfairly.

Weighted scoring is a decision aid, not proof of optimality. Sensitivity analysis should test whether a small change in weights reverses the ranking. Dominated alternatives, regulatory constraints, and non-negotiable technical requirements should be eliminated before scoring.

7. Distributed Manufacturing and Digital Inventories

Distributed manufacturing places qualified production capability near customers, service sites, or demand centers. A digital inventory replaces some physical stock with controlled digital design files, process recipes, material specifications, and inspection requirements. Production occurs when demand is confirmed.

Potential benefits include reduced transport, shorter response time, lower finished-goods inventory, local customization, and improved continuity during regional disruption. Potential disadvantages include duplicated equipment, inconsistent process capability, cybersecurity exposure, local skill shortages, qualification costs, and difficulty maintaining identical quality across sites.

A distributed network requires:

• version-controlled design and build files;
• authorized digital rights and access control;
• site qualification and machine calibration;
• common materials and parameter windows;
• standardized inspection plans;
• traceable material and process records;
• secure data transmission and backup;
• periodic inter-site correlation testing.

A digital inventory is not merely a computer folder. It is a governed product-data system containing a released configuration, approved process parameters, revision history, permissions, and evidence that the part can be reproduced within specified limits.

8. Supply-Chain Resilience and Risk

Resilience is the ability of a supply chain to prepare for disruption, absorb its effects, recover, and adapt. AM can improve resilience by enabling local production of low-volume spares and reducing dependence on obsolete tooling. It can also introduce new vulnerabilities through reliance on specialized powders, proprietary software, a single machine supplier, or centralized digital infrastructure.

A practical risk process consists of:

1. identify hazards and failure modes;
2. estimate likelihood and consequence;
3. identify existing controls;
4. assign an owner and mitigation;
5. monitor leading indicators;
6. review residual risk after mitigation.

A basic risk-priority score may be calculated as:

where L is likelihood, C is consequence, and D is difficulty of detection, each assigned an agreed ordinal scale. This score is useful for prioritization but is not a probability and should not be treated as one. A rare catastrophic failure may deserve attention even if its numerical score is lower than a frequent minor defect.

Important AM risks include powder or resin supply interruption, machine downtime, cyberattack, unauthorized design modification, inaccurate process data, operator error, inconsistent post-processing, insufficient validation, and regulatory nonconformance. Mitigations include dual sourcing, safety stock of critical materials, preventive maintenance, qualification of alternate machines, offline recovery procedures, access control, independent inspection, and documented change management.

9. Sustainability, Ethics, and Governance

Economic analysis must not ignore external effects. Relevant considerations include energy consumption, material waste, solvent use, powder handling, worker exposure, transportation, product life, repairability, and end-of-life recycling. A lightweight printed part may reduce use-phase energy, but that benefit must be assessed against process energy and material production impacts.

Ethical and governance responsibilities include:

• protecting worker health through ventilation, personal protective equipment, and safe powder handling;
• preventing unauthorized reproduction of patented or safety-critical designs;
• preserving traceability for regulated parts;
• reporting defects rather than concealing unfavorable yield;
• validating AI-generated geometry or process recommendations before release;
• controlling access to defense, medical, and proprietary data;
• ensuring that automated decisions remain reviewable and accountable.

Artificial intelligence can assist demand forecasting, anomaly detection, parameter optimization, and predictive maintenance. It does not remove the need for physical validation. A model trained on one machine, material batch, or geometry family may not generalize to another. Data provenance, drift monitoring, explainability appropriate to the risk, and human authorization are essential.

10. Illustrative Case: Regional Production of a Service Spare

Background

A maintenance company supplies a polymer replacement cover for industrial pumps. Annual demand is 600 units, but demand is irregular. The existing supply chain purchases molded parts from a distant supplier with a six-week lead time and a minimum order of 500 units.

Problem

The company experiences obsolete inventory, emergency freight charges, and occasional shortages. An AM cell could produce the cover locally, but its machine capacity is limited and the part requires support removal and dimensional inspection.

Analysis

The internal variable cost is estimated at 31 currency units per accepted part. Avoidable annual fixed cost for the cell is 9,000 units. The supplier price is 55 units per part, excluding emergency freight. The financial break-even quantity is:

Expected demand of 600 parts exceeds this quantity. However, the analysis also examines capacity. Each batch produces 20 parts and requires 8 machine-hours. Annual machine demand is:

If the qualified cell offers 400 planned productive hours, utilization is:

The cell therefore has nominal capacity, but inspection and support removal must also be checked. The company validates the design, establishes a controlled digital inventory, qualifies two local operators, and retains the external supplier as a surge source.

Outcome

The hybrid strategy reduces routine lead time and physical inventory while preserving contingency capacity. It is superior to an exclusive make strategy because a single local machine failure would otherwise create a supply interruption. It is also superior to exclusive buying because normal demand exceeds the economic break-even quantity and emergency logistics are expensive.

Learning Points

The case demonstrates that the best decision is often a portfolio decision rather than a binary choice. Cost, capacity, lead time, resilience, data governance, and qualification must be considered together. Distributed AM is valuable when demand is uncertain, parts are difficult to stock economically, and the digital production system can be governed reliably.

11. Best-Practice Decision Framework

A disciplined AM operations decision can follow these steps:

1. Define the functional, regulatory, and demand requirements.
2. Confirm technical feasibility and identify required post-processing.
3. Map the complete process route and all constrained resources.
4. Estimate total cost using accepted output, not only nominal output.
5. Separate avoidable, fixed, sunk, and strategic costs.
6. Compare make, buy, and hybrid alternatives.
7. Test capacity at expected and peak demand.
8. Evaluate quality, cybersecurity, sustainability, and supply risks.
9. Perform sensitivity analysis on yield, utilization, material price, labor rate, and demand.
10. Pilot the decision, measure actual performance, and update the model.

The quality of the decision depends on the quality of assumptions. Cost models should be reconciled with actual machine logs, material consumption, labor records, scrap, inspection results, and delivery performance. Continuous improvement is therefore both an operational and an economic activity.

Challenges and Limitations

AM cost estimation is difficult because machine rates differ between organizations, utilization is often uncertain, and post-processing may dominate total effort. Yield estimates based on small samples can be unreliable. Allocated overhead can distort comparisons, especially when managers treat accounting allocations as avoidable cash costs. Weighted scoring models can conceal subjective judgments, while risk scores can create false precision. Finally, digital inventories and distributed production increase the number of sites and interfaces that must be controlled. These limitations require scenario analysis, pilot data, sensitivity testing, and engineering judgment.

Chapter Summary

AM operations decisions require an integrated view of production, economics, quality, and supply-chain strategy. Total cost includes material, machine, labor, energy, post-processing, quality, setup, overhead, and logistics. Yield must be incorporated because the relevant denominator is accepted output. Machine-hour rates convert ownership and facility commitments into job costs, while break-even analysis compares avoidable internal costs with supplier prices.

Capacity is constrained by machines, people, post-processing, inspection, materials, and qualified data. Technology selection should combine technical requirements with normalized economic and strategic criteria. Distributed manufacturing and digital inventories can shorten lead times and reduce physical stock, but they require strong configuration control, cybersecurity, qualification, and traceability. Resilience depends on identifying both conventional and digital vulnerabilities and providing credible mitigations. Finally, sustainability, worker safety, intellectual property, regulatory compliance, and accountable use of AI are integral to responsible AM decision making.

Key Terms

Additive manufacturing: Production of objects by depositing or consolidating material from digital geometry, generally layer by layer.

Accepted output: Parts that meet specified requirements and are released for use or sale.

Avoidable cost: A cost that disappears when a particular production or sourcing decision is changed.

Break-even quantity: The volume at which two alternatives have equal relevant cost.

Capacity utilization: Required or actual workload expressed as a fraction of available capacity.

Capital-recovery factor: A factor converting an investment into an equivalent annual cost.

Digital inventory: Controlled digital product and process information used to reproduce a part.

Distributed manufacturing: Production performed across geographically separated, qualified sites.

Direct labor: Labor time traceable to a specific job or product.

Failure cost: Cost arising from scrap, rework, warranty, recall, or other nonconformance.

Make-or-buy analysis: Comparison of internal production and external procurement alternatives.

Machine-hour rate: Cost assigned to one hour of machine occupancy or productive use.

Post-processing: Operations performed after the primary build to achieve required properties or finish.

Risk priority number: An ordinal product of likelihood, consequence, and detectability ratings.

Supply-chain resilience: Capacity to withstand, recover from, and adapt to disruption.

Total cost: The combined relevant cost of producing, processing, assuring, and delivering a product.

Utilization: The proportion of available resource capacity used by workload.

Yield: Fraction of produced units that meet requirements without unacceptable rework.

Review Questions

Short Answer Questions

1. Why should AM unit cost be calculated using accepted parts rather than only planned parts?
2. Identify four cost categories in a complete AM production model.
3. What is the difference between avoidable and allocated fixed cost?
4. State two benefits and two risks of a digital inventory.
5. Why can post-processing become the bottleneck in an AM system?

Long Answer Questions

1. Explain how machine-hour rates can be constructed and discuss the assumptions behind capital recovery.
2. Describe a systematic make-or-buy analysis for an AM component, including financial and strategic factors.
3. Explain how capacity planning should include machines, labor, post-processing, inspection, and maintenance.
4. Compare centralized and distributed AM supply networks.
5. Discuss the ethical, cybersecurity, and regulatory responsibilities associated with digital AM production.

Analytical / Application Questions

1. An internal AM route has avoidable fixed cost of 24,000 currency units per year and variable cost of 28 units per part. A supplier charges 52 units per part. Calculate the break-even quantity and explain two nonfinancial factors that could change the decision.
2. A machine cell has three machines, each scheduled for 1,600 hours annually, with a planned productive fraction of 75%. Annual demand requires 3,900 machine-hours. Calculate planned capacity and utilization, then recommend one operational action.
3. A company is considering local AM production for emergency spares. Construct a decision matrix with at least five criteria, assign defensible weights, and explain how sensitivity analysis could test the robustness of the recommendation.

Multiple Choice Questions with Answers

Question 1

Which cost is normally included in a complete AM order-cost model?

A. Only raw material cost

B. Machine, labor, energy, post-processing, and quality costs

C. Only the machine purchase price

D. Only transportation cost

Correct Answer: B. Machine, labor, energy, post-processing, and quality costs

Question 2

A job costs 1,405 currency units and produces 38 accepted parts. What is the approximate accepted-part cost?

A. 26.62 currency units

B. 32.50 currency units

C. 36.97 currency units

D. 42.15 currency units

Correct Answer: C. 36.97 currency units

Question 3

In the break-even formula Q_BE = F_make ÷ (C_buy − C_make,var), what does F_make represent?

A. Supplier price per unit

B. Avoidable fixed internal cost

C. Internal material mass

D. Expected production yield

Correct Answer: B. Avoidable fixed internal cost

Question 4

If required workload is 900 machine-hours and available capacity is 1,200 machine-hours, utilization is:

A. 25%

B. 60%

C. 75%

D. 133%

Correct Answer: C. 75%

Question 5

Which factor is most likely to create a bottleneck after printing?

A. A completed digital drawing

B. Excessive inspection or post-processing demand

C. A lower file revision number

D. Reduced packaging color choices

Correct Answer: B. Excessive inspection or post-processing demand

Question 6

What is a defining feature of a governed digital inventory?

A. Uncontrolled copies of any available CAD file

B. A released, traceable design with approved process information

C. Physical storage of finished products only

D. A supplier’s advertising catalogue

Correct Answer: B. A released, traceable design with approved process information

Question 7

The risk-priority expression RPN = L × C × D is best understood as:

A. A precise probability of failure

B. An ordinal tool for prioritizing risks

C. A machine-energy equation

D. A material-yield law

Correct Answer: B. An ordinal tool for prioritizing risks

Question 8

Why should a make-or-buy analysis exclude fixed overhead that remains after outsourcing?

A. It is not an avoidable saving

B. It is always zero

C. It is a variable supplier cost

D. It measures material yield

Correct Answer: A. It is not an avoidable saving

Question 9

Which is a principal advantage of distributed AM?

A. Guaranteed elimination of cybersecurity risk

B. Potentially shorter response time near demand centers

C. No need for site qualification

D. Identical performance without process control

Correct Answer: B. Potentially shorter response time near demand centers

Question 10

Why must AI recommendations for AM process optimization be physically validated?

A. AI cannot process any numerical data

B. A model may not generalize across machines, materials, or geometries

C. Validation increases nominal machine speed automatically

D. Validation replaces cybersecurity controls

Correct Answer: B. A model may not generalize across machines, materials, or geometries