Translating Structured Expert Judgment into Order-of-Magnitude Cost Estimates: A Framework for Early-Stage Space Mission Feasibility Assessment

Abstract

Cost-feasibility assessment remains a persistent challenge in the New Space economy, where early-stage missions must be evaluated under profound technological, operational, and financial uncertainty and where reliable historical cost data are often unavailable. Conventional parametric and regression-based cost models offer limited support in these settings, as critical architectural decisions are taken before technologies, interfaces, and operational concepts have stabilized. This study proposes a hybrid, expert-based framework for structuring early-stage cost-feasibility reasoning in frontier space missions by integrating systematic literature review, structured expert elicitation, and Partial Least Squares Structural Equation Modeling. The framework is demonstrated through a case study of lunar oxygen production via regolith sublimation. Implemented across three phases, the approach identifies and refines mission-relevant cost drivers through Delphi-based expert judgment and synthesizes these assessments into a coherent quantitative structure. The resulting model explains approximately 57% of the variance in expert cost assessments, indicating stable and interpretable patterns in expert judgment under conditions of high uncertainty. Within this structure, technological readiness and operational efficiency emerge as the most influential dimensions shaping perceptions of overall project magnitude. Rather than aiming to predict realized mission costs, the framework is designed to support relative comparison, early-stage architectural decision-making, and the disciplined organization of uncertainty. Anchored to an equipment-cost proxy and expert-derived driver weights, the framework produces an illustrative, order-of-magnitude project-scale estimate of approximately USD 30.7 billion for the reference mission configuration, consistent with expectations for large-scale space-resource initiatives. The principal contribution of the study lies in providing a transparent process for translating dispersed expert knowledge into structured cost-feasibility reasoning, with implications for early mission planning, derisking strategies, and stakeholder alignment in high-uncertainty space ventures.

Keywords

cost-feasibility expert judgment frontier space missions technological readiness uncertainty management New Space economy

INTRODUCTION: THE EVOLVING ECONOMICS OF SPACE INNOVATION

Over recent decades, the space sector has undergone a structural transition from government-led programs toward a more distributed ecosystem involving public–private collaboration and commercially oriented activities.¹ As mission concepts expand into frontier domains such as in situ resource utilization (ISRU) and space-based industrial operations,² early-stage decision-making increasingly takes place under conditions of limited empirical data and high technological uncertainty. In this context, the ability to assess cost-feasibility before system architectures and operational concepts have stabilized becomes a critical constraint on mission design and investment.

Cost Underestimation and Accountability in Space Programs

The persistent unreliability of cost estimates in the space sector, often attributed to practices that underreport actual expenditures,³ highlights the need for more robust, project-specific approaches to financial planning. Persistent underestimation undermines accountability and complicates both technical decision-making and stakeholder coordination in complex space programs.

Evidence from large-scale infrastructure projects suggests that integrating planning, construction, and early operational phases within a single contractual framework can strengthen incentives for realistic cost estimation and more comprehensive risk assessment. When the same entity is responsible for delivery and early operations, the financial consequences of underestimation become more immediate, encouraging more disciplined budgeting practices.^4,5 In public–private partnerships, such alignment reduces information asymmetries and promotes shared accountability for schedule and budget outcomes.⁶

Within the New Space economy, adopting similarly integrated approaches depends in part on the availability of cost-assessment tools that can operate under conditions of high uncertainty. Frameworks that support transparent and coherent early-stage cost reasoning can facilitate more effective risk allocation between public agencies and private operators, while aligning financial expectations with evolving technical realities.

COST ESTIMATION AND FINANCIAL MODELING IN SPACE MISSIONS: A REVIEW

Persistent cost underestimation in space programs has frequently been linked to the structural limitations of parametric regression–based models.⁷ Efforts to address these shortcomings through analogical reasoning or reliance on engineering judgment have produced only limited improvements. Over time, several alternative approaches have been proposed to enhance cost-estimation performance, but many have proven difficult to apply in practice.⁸

Drury offers a structured overview of cost-forecasting practices by distinguishing between major methodological families.⁹ One widely used approach is the engineering bottom-up method, which derives cost estimates from detailed technical assessments of input–output relationships across the production process. This method requires clearly defined operational stages to ensure accurate measurement. While it is effective for identifying variable costs, it is less reliable for capturing fixed or overhead components,¹⁰ limiting its applicability to narrowly specified or repetitive activities.

A second group of methods, often described as inspection-of-accounts approaches, constructs forecasts through detailed analysis of financial statements and classification of costs according to behavioral characteristics. Although this approach provides insight into cost drivers and managerial control, its reliance on historical accounting data and subjective judgment reduces its usefulness in unstable or rapidly evolving sectors.

Several mission-specific models illustrate these limitations. The P-Beat and Process-Based models developed by Boeing and NASA required highly detailed datasets that constrained timely decision-making. Similarly, the QUICKCOST model failed to account for contextual factors such as organizational culture, project scale, and risk tolerance, which are critical for assessing feasibility in complex space missions.^11,12

Subsequent studies have sought to address these limitations through refinements in both cost variables and modeling scope. Collopy et al. proposed replacing traditional weight- or volume-based parameters with surface area as a primary scaling variable, showing that cost follows a power relationship with an exponent of approximately 2.2.¹³ Complementary approaches expanded the analytical scope of cost estimation. For example, Peeters’s “5C approach” introduced a holistic framework that integrates parametric costing, life-cycle analysis, contractual design, risk management, and transparency as interdependent elements of cost control.¹⁴ In parallel, researchers at NASA developed iterative methods that incorporate expert judgment and cost-driver matrices to better calibrate uncertainty and reflect the complexity of novel mission concepts.^15,16

Building on these developments, much of the contemporary literature evaluates project feasibility using net present value (NPV) analysis, combining expected revenues with the time value of money to assess economic viability over the project life cycle.¹⁷ However, Keller and Collopy later observed that approximately half of all cost overruns are endogenous, arising from intrinsic project complexity rather than from shortcomings in cost models themselves. They argue that improving governance and control mechanisms may therefore be more effective than further refining predictive formulas.¹⁶ This insight is particularly relevant in the New Space economy, where rapid iteration and disruptive innovation reduce the reliability of historically calibrated models.¹⁸

In practice, profitability-oriented frameworks such as NPV often lack the level of detail required for budgeting and early-stage financial planning. In response, researchers have developed more dynamic methodologies that combine statistical techniques with structured expert input. Peterson and Probst, for example, introduced semiquantitative models based on indices and iterative feedback.¹⁹ Subsequent work applied the Delphi technique and structural equation modeling to capture complex relationships among latent cost drivers as assessed by experts.²⁰

Despite these advances, most cost-estimation frameworks, whether qualitative or quantitative, continue to depend heavily on historical benchmarks. As a result, their performance deteriorates when empirical data are limited. Dai notes that this reliance reduces their usefulness during early design phases and constrains their applicability in emerging, high-uncertainty industries such as space.²¹

Taken together, this literature suggests that the intrinsic complexity of disruptive innovation remains insufficiently addressed by traditional parametric and analogical models. Trivailo et al. argue that speculative assessments are particularly valuable in the early stages of project development, when quantitative data are scarce and engineering evaluations remain uncertain.²² In such contexts, expert judgment can provide critical insight, but only if it is supported by structured aggregation and scaling techniques to limit bias. Methods such as the Analytic Hierarchy Process²³ and the Delphi technique offer systematic approaches for ranking cost drivers and consolidating expert opinions into interpretable quantitative outputs.

However, the application of these methods in the space domain remains largely fragmented and method-driven, without situating them within an integrated framework that reflects the sequencing of early mission design, the interdependence of technological and operational choices, and the specific uncertainties characteristic of novel space systems. As a result, there remains a gap in how expert-based methods are systematically combined and operationalized to support early-stage cost-feasibility reasoning in space missions where technologies, architectures, and operational concepts are still in flux.

METHODOLOGICAL FRAMEWORK: AN EXPERT-BASED APPROACH TO COST-FEASIBILITY UNDER UNCERTAINTY

Hybrid approaches that combine qualitative expert judgment with quantitative modeling provide the conceptual foundation for the framework developed in this study. Building on this logic, the proposed model aims to improve early-stage cost estimation in frontier industries and is demonstrated through a case study that reflects the key challenges faced by emerging space missions.

Case Context: Space Mining as a Testbed for Frontier Cost Modeling

The proposed framework is applied to space mining, a domain that exemplifies the high uncertainty and complexity typical of emerging space ventures. Space mining differs fundamentally from incremental, product-oriented innovation historically dominated by large incumbents.²⁴ Instead, it represents a form of business-model disruption within the broader space economy. This dynamic is particularly visible in the ability of new entrants to pursue novel operational and commercial approaches, while established actors often face difficulties adapting to growth paths that require strategic transformation rather than incremental technological improvement.¹⁹

The strategic relevance of space mining is highlighted by Crawford,²⁵ who identifies several ways in which extraterrestrial resource extraction could reshape the space economy. These include supporting the establishment and long-term operation of scientific outposts, enabling sustained exploration through locally sourced resources, and laying the foundations for commercial expansion beyond the satellite sector.

From an economic perspective, a substantial body of literature has sought to estimate the potential profitability of space mining, building on early contributions such as Lewis’s Mining the Sky, which argued that significant resource value may exist beyond Earth.²⁶ However, much of this work emphasizes potential returns and macroeconomic opportunity, while giving comparatively less attention to questions of cost-feasibility and early-stage financial viability.²⁷

Research Design: A Multiphase Methodology

Building on this rationale, the study adopts a structured, multiphase research design to assess the cost-feasibility of lunar oxygen mining under conditions of high uncertainty.

The analysis is conducted on a reference architecture representing a lunar regolith-based oxygen production system. The study focuses on the operational segment of the mission, including regolith excavation, material handling, oxygen extraction, and the supporting subsystems required to sustain these processes on the lunar surface. The cost drivers considered in the analysis, detailed in the appendix, reflect key performance and design characteristics such as system mass, energy usage, process efficiency, and mission duration.

Elements beyond this operational scope are treated as exogenous and are not explicitly modeled. These include, for example, the development of enabling technologies (e.g., nuclear power systems), launch system design, and downstream infrastructure such as storage, distribution, and in situ logistics. The study, therefore, evaluates cost-feasibility within a bounded mission segment, rather than across the full life cycle of a complete lunar industrial system.

The approach integrates qualitative expert judgment with quantitative modeling to account for both technological and financial considerations. Organized into three phases, the methodology combines structured expert elicitation with statistical analysis, enabling insights from different sources to be systematically compared, refined, and consolidated into a coherent cost-assessment framework.²⁸

Phase 1—Literature-based scoping of cost drivers

The first phase consists of a systematic review of technical and economic literature aimed at defining the mission typology and identifying an initial set of cost drivers. This phase serves as the foundational scoping step of the framework, establishing the assumptions and boundaries that guide subsequent expert elicitation and modeling.

In the context of lunar oxygen mining, recent studies emphasize ISRU as a key enabler for reducing mission costs and supporting sustained lunar operations. The literature identifies lunar ice and regolith as primary resource candidates, with several techno-economic assessments focusing on oxygen extraction through regolith processing.^29–32 In particular, models developed by the Colorado School of Mines highlight sublimation followed by electrolysis as a viable alternative to conventional excavation and hauling approaches,³³ a conclusion further supported by Lefeber.³⁴

Drawing on this body of work, Phase 1 consolidates insights from mission design, resource extraction, and cost-modeling studies to define the contextual assumptions and an initial set of cost drivers used in the subsequent phases of the analysis. In this case, the selection of a sublimation–electrolysis processing architecture informs assumptions regarding material throughput, energy requirements, and system configuration, which are translated into preliminary cost drivers capturing key system characteristics such as equipment mass, process efficiency, and mission duration.

From an engineering and process-design perspective, the reviewed literature provides the technical foundation for deriving analogical parameters. Prior studies examine a range of excavation and material-handling architectures for lunar regolith, including bucket-based and continuous excavation systems,^35–39 as well as several oxygen-extraction pathways such as carbothermal reduction and electrolysis-based processes.^40–42 Cross these analyses, recurring operational constraints, most notably waste generation and dust dispersion, are consistently identified as major risks to equipment performance and long-term system reliability.⁴³

Key physical parameters are grounded in established measurements of lunar regolith properties, including bulk density⁴⁴ and oxygen content.⁴⁵ These characteristics inform feasible assumptions regarding material throughput, energy requirements, and process efficiency.

Drawing on this body of work, current techno-economic assessments converge on oxygen production rates on the order of several tens of kilograms per hour as technically achievable under projected power constraints.⁴⁶ At the system level, complementary studies anticipate substantial propellant demand on the lunar surface.⁴⁷

Collectively, these findings support the use of regolith-based oxygen production as a realistic reference case for early-stage cost assessment and parameter scaling within the framework developed here.^48,49

In parallel, a broader body of cost-modeling and economic literature informs the financial structure of the framework. Early terrestrial cost-estimation methods^50,51 introduce parameter weighting and scoring approaches that were later adapted for space applications through readiness-based and capability-oriented cost coefficients. Subsequent studies extend these concepts by linking technical performance assumptions to assessments of economic feasibility under uncertainty.^52,53

Later research incorporates NPV analysis and related profitability metrics^{26,28,54–56} to evaluate long-term project viability. Complementary modeling efforts further suggest that public–private partnership configurations can improve mission outcomes under uncertainty,⁵⁷ reinforcing the value of shared cost-accounting frameworks. Comparable orbital and in-space resource studies support the scale assumptions adopted here, indicating that sustained operations over multiyear horizons can achieve material production volumes consistent with early-stage demand projections.^42,58,59

Together, these studies provide a coherent set of financial parameters and performance indicators across both technical and economic dimensions. This synthesis yields standardized variables with indicative value ranges, which serve as baseline inputs for expert assessment and validation in Phase 2.

Phase 2—Expert judgment and Delphi-based validation

The second phase consists of two rounds of expert elicitation designed to validate the cost drivers identified in Phase 1 and to strengthen the analysis in the absence of reliable historical data. Consistent with the approach outlined by Trivailo, expert judgment is used as a structured substitute for empirical observation in this early-stage context. The first round adopts a Delphi-based design, which is well suited to environments characterized by high uncertainty and limited theoretical consensus.⁶⁰

The Delphi process applied in this study departs from the traditional iterative format. Rather than consulting the same experts across multiple rounds, a two-stage design is employed in which distinct expert subgroups are engaged once. The first stage is exploratory and focuses on qualitatively confirming the relevance of the cost drivers identified in Phase 1, as well as identifying any overlooked variables. The second stage is investigative, aiming to validate and quantify these drivers in a structured manner. A larger expert sample is allocated to the second stage, allowing divergent views to be examined and a broader range of perspectives to be incorporated.

Methodologically, the approach draws on Dalkey’s formulation of the Delphi technique, which treats individual expert responses as observations within a defined event space.⁶¹ In this configuration, aggregated responses provide a statistically interpretable representation of expert knowledge and serve as a structured substitute for empirical data when direct observations are unavailable.

Expert responses are summarized using the median as the representative measure, reflecting the skewed distributions typical of expert-generated estimates and reducing sensitivity to outliers.⁶² Response dispersion is assessed consistently to evaluate the degree of agreement among experts.⁶³

In this study, the first Delphi round involved eight experts drawn from both research institutions and private ventures active in in-space resource utilization. Their backgrounds span space mining, thermomechanics, propulsion, systems engineering, assembly–integration–testing, and mission design. The outcome of this phase is a validated and refined set of cost drivers, organized across technological, operational, and financial dimensions, which forms the basis for the subsequent quantitative assessment.

The analysis considers cost drivers that capture key aspects of lunar oxygen production, including mission readiness, system mass and complexity, operational performance, energy requirements, human involvement, and logistics associated with deploying and operating mining systems on the lunar surface. Technological drivers reflect the maturity of critical subsystems, development effort, and system reliability, including fault detection and energy-storage performance. Operational drivers address mission duration, material throughput, conversion efficiency, system complexity, and the level of human oversight required to sustain operations. Financial and contextual drivers capture payload delivery costs, in situ resource demand, and downstream processing requirements that influence overall project scale and feasibility.

The complete list of cost drivers, together with their descriptive parameters and indicative ranges used in the expert elicitation, is reported in the appendix.

The second Delphi round adopts a structured format aimed at quantifying expert assessments and consolidating consensus on the relative importance of each cost driver. In this case study, a broader and more diverse group of 50 professionals participated, including representatives from space agencies, industrial manufacturers, and consulting organizations. Most respondents were directly involved in lunar or cislunar projects.

To improve data quality and capture a broader range of perspectives, the design allows for partial repolling across iterations. This approach enables the composition of respondents to vary over time, while prioritizing the involvement of highly experienced experts in the early stages. Doing so helps refine parameter ranges and limits the influence of outlier assessments as the elicitation process progresses.^64–66

Expert evaluations are collected through structured pairwise comparisons and Likert-scale ratings, allowing respondents to express both the direction and relative strength of their judgments.^67,68 Responses are then aggregated using standard reliability and consistency checks to produce weighted scores that summarize expert assessments while reducing individual bias.^69,70 The outcome of this phase is a consolidated ranking of cost drivers, which serves as the input for the subsequent quantitative modeling stage.

Phase 3—Quantitative modeling using Partial Least Squares Structural Equation Modeling

The third phase translates the structured expert assessments obtained in Phase 2 into a quantitative representation. Given the reliance on Likert-scale data, the non-normal distribution of responses, and the lack of historical cost datasets, Partial Least Squares Structural Equation Modeling (PLS-SEM) is adopted as an appropriate technique for this exploratory setting.⁷¹ Unlike covariance-based SEM approaches, which are typically applied to hypothesis testing using established empirical data, PLS-SEM is well suited to analyzing structured expert judgment and identifying consistent patterns across complex, multidimensional inputs.⁷²

At a conceptual level, the modeling framework distinguishes between observed indicators derived from expert responses and latent constructs that group related cost drivers. The analysis focuses on how experts collectively assess the relative importance of technological, operational, and financial factors, rather than on inferring causal relationships. Iterative refinement is used to ensure internal consistency and stability in the resulting construct structure.⁶³

The structural model integrates the validated cost drivers from Phase 2 into a hierarchical representation linking technological, operational, and financial dimensions. These dimensions are aggregated into higher-level constructs, such as efficiency, complexity, and time, which together form an Aggregate Cost Driver. This aggregate construct is then used to derive a Project Cost Proxy that supports comparison across alternative mission configurations.⁷³

In the case study, equipment cost is used as the primary scaling variable for the Project Cost Proxy, reflecting differences in mission architecture and system design. This assumption is consistent with established cost-decomposition approaches in space mission analysis, where equipment-related expenditures represent a substantial share of total project cost and capture key design trade-offs.⁶¹ Other cost components that are largely external to system design are treated as exogenous parameters. On this basis, the model generates comparative cost estimates by adjusting equipment expenditure according to the weighted aggregate cost drivers derived from expert assessments.

RESULTS AND VALIDATION OF EXPERT-BASED COST-FEASIBILITY

The structural model derived from the sublimation-based mission design is evaluated to assess the internal coherence and robustness of the expert-based results. Validation focuses on confirming that the aggregated expert assessments provide a consistent and nonredundant representation of the underlying cost drivers, rather than on statistical forecasting performance.

At the measurement level, internal consistency checks are used to verify that individual cost drivers contribute distinct information and that grouped constructs reflect coherent expert judgments. These checks confirm that the indicators can be meaningfully aggregated without inflating variance.⁷⁴

At the structural level, the model summarizes how experts collectively assess the relative importance of technological, operational, and financial drivers within the overall cost framework.^75–77 The integrated set of drivers accounts for approximately 57% of the variance observed in experts’ cost estimates, indicating that the framework captures stable and interpretable patterns in expert judgment.⁷⁸ Among the assessed dimensions, technological readiness and operational efficiency emerge as the most influential contributors within the expert assessments.

Validation further examines the stability of the results across alternative model specifications and subsets of experts, confirming that no single cost driver or respondent group disproportionately influences the outcomes. In the absence of historical cost data, the validation process emphasizes internal coherence, sensitivity to expert disagreement, and consistency with established engineering and economic expectations, rather than out-of-sample predictive performance.^79–82

Detailed reliability diagnostics, consistency checks, and supplementary analyses are reported in the appendix. The results are intended to support relative comparison across alternative mission configurations and informed early-stage cost reasoning,⁸³ rather than to provide point forecasts of realized mission costs.

Finally, the PLS-SEM framework is applied to derive an indicative cost estimate for the operational mission segment of the sublimation-based mission design. The resulting aggregate cost score is translated into an order-of-magnitude monetary estimate through benchmarking against comparable large-scale terrestrial and in-orbit engineering and mining systems,^17,20,84,85 yielding a value of approximately USD 30.7 billion. This figure reflects the aggregation of expert judgments across technological, operational, and financial dimensions and is consistent with expectations for large-scale space-resource initiatives. More broadly, it illustrates how structured expert knowledge can be translated into coherent order-of-magnitude cost estimates for early-stage mission planning.

PRACTICAL IMPLICATIONS USING EXPERT JUDGMENT TO PLAN AND DE-RISK FRONTIER MISSIONS

Frontier missions in the emerging space economy, such as lunar oxygen production, operate under conditions of high uncertainty, limited operational precedent, and weak historical analogues for cost estimation. In these settings, the central challenge is to support informed decision-making at stages where key architectural choices must be made before empirical cost data are available. The practical value of the framework presented in this study lies in addressing this challenge. By structuring dispersed expert knowledge into a transparent and integrated assessment, the framework supports early-stage cost reasoning while also providing actionable signals for research prioritization, risk reduction, workstream coordination, and stakeholder alignment.⁸³

A key implication is that structured expert elicitation should be applied deliberately and early in mission planning, rather than used reactively when quantitative models prove insufficient. Frontier space missions often require early downselection among alternative architectures, process chains, and deployment strategies, well before systems have matured to the point where parametric cost models are reliable. Decisions taken at these stages tend to shape a substantial share of downstream cost and risk, as they define interfaces, mass and power budgets, operational concepts, and subsystem maturity requirements. In this context, the proposed framework functions as a planning instrument: it enables order-of-magnitude cost assessment, highlights the cost relevance of competing design choices, and helps identify which elements of a mission concept warrant early de-risking. This role is particularly relevant during pre-Phase A and early Phase A development, when architectures remain flexible, and the marginal value of improved assumptions is highest.⁸⁶

Beyond producing an indicative project-scale estimate, the framework has implications for how mission teams organize work and structure decision-making. By eliciting expert assessments across technological, operational, and financial dimensions and integrating them into a coherent representation, the approach encourages planners to treat cost-feasibility as a continuous input to mission design rather than as a late-stage validation exercise. This perspective has direct implications for the composition of mission teams and advisory bodies. Early expert elicitation benefits from intentionally broad participation across mission functions—including systems engineering, operations, power and thermal systems, resource processing, autonomy and fault management, and programmatic cost analysis—because cost-feasibility in frontier missions is shaped as much by integration and interface risks as by individual component performance.⁸⁷ By promoting early cross-functional engagement, the framework helps reduce blind spots and improve the consistency of assumptions across parallel workstreams.

A further implication concerns how uncertainty and disagreement among experts are interpreted. In conventional planning contexts, divergence in expert views is often treated as noise to be averaged out in pursuit of a single “best” estimate. In frontier missions, however, such disagreement can be informative. Dispersion in expert assessments may signal immature technologies, poorly specified interfaces, or performance parameters that remain contingent on untested assumptions. Rather than indicating methodological weakness, persistent divergence can reflect structural uncertainty embedded in the mission concept itself. The framework makes these patterns of disagreement visible and therefore actionable. By doing so, it allows mission planners to identify areas where additional evidence, targeted experimentation, or early prototyping is most valuable, and where unresolved uncertainty is likely to generate future redesign or cost escalation if left unaddressed.

This perspective translates directly into a practical mechanism for research prioritization. Areas of pronounced expert disagreement can be interpreted as learning bottlenecks: elements of the mission concept where improved empirical grounding—through targeted experimentation, prototyping, or dedicated technology demonstrations—is likely to reduce uncertainty in downstream cost reasoning. In the context of lunar oxygen mining, such bottlenecks may relate to operational durability in dust-laden environments, the scalability of regolith handling and processing throughput, the maturity of autonomous fault detection and recovery, or the end-to-end efficiency of processing chains under lunar constraints. While the framework does not resolve these uncertainties directly, it provides a structured basis for identifying them and for justifying why they warrant early attention. In doing so, it supports more disciplined investigation of critical assumptions, rather than the opportunistic expansion of research agendas.

A related implication concerns the prioritization of de-risking efforts under early-stage resource constraints. Frontier missions typically operate with limited budgets, time, and testing capacity, requiring deliberate trade-offs about where uncertainty should be reduced and where it can be provisionally accepted. By translating expert assessments into relative weightings and aggregated cost drivers, the framework helps distinguish between factors that are likely to have a material influence on overall project magnitude and those that are comparatively marginal. This distinction can inform both the sequencing and intensity of de-risking activities. High-impact drivers justify focused attention, early system-level integration, and close coordination across workstreams. Conversely, drivers that exhibit lower relative importance and higher consensus among experts can often be managed through standard engineering margins and conventional verification approaches, without absorbing disproportionate managerial effort. The implication is not that these factors are unimportant but that they are less decisive for early-stage cost-feasibility reasoning and therefore less suitable as primary levers for early program focus.

The framework also supports a temporal logic for mission sequencing and staged commitment.⁸⁸ Frontier space missions often benefit from phased architectures, in which early investments are directed toward reducing the most consequential uncertainties before large-scale capital is committed. By identifying cost drivers that combine high relative importance with substantial expert disagreement, the framework highlights where early validation activities, such as targeted technology demonstrators, subsystem prototypes, or operational trials, are likely to yield the greatest strategic value. Reducing uncertainty in these areas can enable more confident architecture selection and budgeting decisions. Conversely, when drivers exhibit higher consensus and lower influence on overall cost-feasibility, mission teams can more credibly defer major investments, standardize solutions, or accept provisional assumptions until later design stages. In this way, the framework informs not only what matters most but also when action is warranted, supporting incremental mission development strategies under uncertainty rather than all-at-once commitments.

A related implication concerns financial structuring and stakeholder alignment, particularly in New Space missions involving mixed public–private participation. Although early-stage cost estimates are inherently uncertain, achieving order-of-magnitude coherence remains critical, as it shapes which financing arrangements, partnership structures, and governance models are feasible. When project cost magnitude is poorly specified or contested, technical ambition can become misaligned with financial capacity, increasing the likelihood of later scope revisions, delays, or loss of credibility with funders.⁸⁹ By linking expert judgments about key cost drivers to a transparent, integrated project-scale estimate, the framework provides a basis for early discussions on risk-sharing arrangements, funding sequencing, and the allocation of responsibilities between public agencies and private operators.

Finally, the framework contributes to organizational coordination by establishing a shared language for cost-relevant assumptions. In multiactor mission environments, engineers, financiers, and policy stakeholders often operate with different interpretations of technical risk and uncertainty. Engineers may emphasize feasibility and performance margins, and financiers may focus on downside exposure and uncertainty ranges, while policy actors may prioritize strategic value and program legitimacy. When cost drivers remain implicit or are embedded in opaque models, these differing perspectives can lead to misalignment. A structured cost-driver framework makes assumptions explicit and comparable, enabling more effective communication across stakeholder groups. In practice, this can reduce common distortions in early-stage programs, including engineering optimism bias, excessive financial conservatism, and policy-level misinterpretation of how technical uncertainty translates into cost and schedule risk.

The practical contribution of the framework lies primarily in its process rather than in any single numerical output. The USD 30.7 billion estimate derived in this case serves as an order-of-magnitude reference, but its broader value is to demonstrate how a structured pipeline, combining literature-based scoping, expert elicitation, and quantitative synthesis, can translate dispersed expert knowledge into coherent cost-feasibility reasoning when empirical data are scarce. For this reason, the approach is transferable beyond lunar oxygen mining. Comparable conditions arise in other ISRU concepts, lunar infrastructure development, cislunar logistics, on-orbit manufacturing, and early-stage deep-space commercial systems, where technological maturity is uneven, and integration risks dominate. While the specific cost drivers and expert cohorts will vary across applications, the underlying logic of the framework remains applicable.

The framework can also be interpreted as supporting learning and iteration rather than one-off estimation. Because it is grounded in structured expert judgment, the method can be reapplied as mission concepts evolve, test results accumulate, and subsystem readiness improves. Repeated elicitation cycles allow the relative importance of cost drivers to be updated, track whether expert disagreement is converging, and provide an operational indicator of knowledge maturation over time. Declining dispersion may signal effective de-risking or improved empirical grounding, while persistent divergence may point to deeper structural uncertainty that warrants architectural revision or changes in operational concepts. In this sense, the framework functions as a sounding board for mission planning: an adaptive tool that helps teams revise assumptions, monitor learning progress, and maintain coherence between technical ambition and cost-feasibility reasoning across the mission life cycle.

Taken together, these implications suggest that structured expert judgment, when embedded in a transparent and replicable modeling framework, can materially improve early-stage planning in frontier space missions. The principal benefit is not the elimination of uncertainty but its disciplined organization into actionable signals—indicating where learning is most needed, where de-risking efforts should be concentrated, how workstreams should be sequenced, and how stakeholders can align around a coherent cost-feasibility narrative. This positions the framework as a practical complement to engineering design and program management processes in New Space contexts, where the cost of late discovery is high, and early coherence is essential for credible decision-making.

CONCLUSION AND FUTURE DIRECTIONS FOR RESEARCH

This study addresses a persistent challenge in frontier space missions: how to reason about cost-feasibility in environments characterized by limited historical data, immature technologies, and high integration uncertainty. By combining structured expert elicitation with PLS-SEM, the proposed framework offers a transparent and replicable approach for organizing expert judgment into a coherent basis for early-stage cost reasoning.

Applied to the case of lunar oxygen production via regolith sublimation, the framework reveals consistent patterns in expert assessments across technological, operational, and financial dimensions. The integrated set of cost drivers explains approximately 57% of the variance observed in experts’ cost estimates, indicating that the model captures stable and interpretable structures in expert judgment rather than isolated or inconsistent opinions.

In this context, technological readiness functions as a central organizing factor in early-stage cost reasoning rather than as a deterministic predictor. Lower levels of maturity are associated, in expert judgment, with higher capital intensity and wider uncertainty ranges, reflecting the need for iterative testing, refinement, and integration before performance assumptions can stabilize.⁹⁰

The indicative cost estimate of approximately USD 30.7 billion corresponds to a project scale that is substantially higher than the largest terrestrial mining developments, such as the Simandou iron ore project in Guinea, estimated at approximately USD 20–23 billion,⁸⁵ while remaining below major space programs such as Artemis, which has reached in 2025 approximately USD 90 billion in cumulative investment.⁹¹

The primary value of the approach, however, lies in its process logic. By making cost-relevant assumptions explicit, highlighting areas of expert convergence and divergence, and linking relative driver importance to early architectural choices, the framework supports informed decision-making in settings where uncertainty cannot be eliminated but can be structured and managed.

More broadly, the study demonstrates how structured expert judgment can inform not only approximate cost magnitude but also research prioritization, de-risking strategies, and stakeholder coordination in complex, multiactor space programs. The hybrid methodology complements traditional engineering and financial analyses by offering a systematic way to translate dispersed domain knowledge into actionable planning signals, without relying on strong assumptions about data availability or model completeness.

Future research could extend this work along several dimensions. Expanding the expert sample and repeating elicitation rounds over time would allow the framework to be applied longitudinally, supporting the tracking of learning and convergence as technologies mature. Such extensions could also enable the use of PLS with Maximum Likelihood estimation as a complementary validation technique.⁹² Alternative survey designs and aggregation techniques could be explored to test the sensitivity of results to different elicitation formats, such as a six-grade scale, which removes central values,⁶⁸ while comparative applications across multiple mission concepts would further assess the framework’s transferability. Taken together, the results suggest that disciplined organization of expert judgment, rather than reliance on point forecasts or overly detailed early models, can materially improve cost-feasibility reasoning in frontier space ventures. In environments where the cost of late discovery is high and early coherence is essential, the framework provides a pragmatic foundation for aligning technical ambition, financial capacity, and strategic decision-making under uncertainty.

Footnotes

AUTHOR’S CONTRIBUTIONS

G.M. is the sole author of this article and is responsible for the study conception, literature review, analysis, drafting, and final approval of the submitted version.

AUTHOR DISCLOSURE STATEMENT

The author declares no potential conflicts of interest. This work was conducted independently and does not represent the views or official position of any institution or employer.

FUNDING INFORMATION

The author received no financial support for the research, authorship, or publication of this article.

Supplemental Material

References

1. Deudney

. Dark skies: space expansionism, planetary geopolitics and the ends of humanity. Oxford University Press; 2020.

2. Bordacchini

. Space policies, Issues and Trends in 2017–2018. European Space Policy Institute; 2018; Report 65.

3. Prince

. Why NASA’s management doesn’t believe the cost estimates. Eng Manag J 2002.

4. Bloomfield

, Ahern

. Long-term infrastructure partnerships: Contracting risks and risk-reduction strategies. State Local Government Rev 2011;43(1):49–59; doi: 10.1177/0160323X11400435

5. Flyvbjerg

. Policy and planning for large-infrastructure projects: Problems, causes, cures. Environ Plann B Plann Des 2007;34(4):578–597; doi: 10.1068/b32111

6. Makovsek

. Private sector participation in infrastructure: Can the price of risk transfer be efficient? Int Transport Forum Discuss Paper 2016; doi: 10.13140/RG.2.1.1214.8083

7. Keller

, Collopy

, Componation

. What is wrong with space system cost models? A survey and assessment of cost estimating approaches. Acta Astronaut 2014;93:345–351.

8. Smart

. Process based cost modeling. In: International Society of Parametric Analysts. Science Application International Corporation (SAIC); 2006.

9. Drury

. Management and cost accounting, 10th ed. Annabel Ainscow: Andover, Hampshire; 2018.

10.

10. Roy

. Cost engineering: why, what and how? Decision Engineering Report Series. Cranfield University; 2003.

11.

11. Hamaker

. Do more details imply better accuracy in cost estimates? Parametric World 2010;13:22.

12.

12. Wertz

. Reducing space mission cost. Microcosm Press: El Segundo; 1996.

13.

13. Collopy

. The challenge of modeling cost – steps toward a solution. In: 1st International Conference on Innovation and Integration in Aerospace Sciences. Queen’s University Belfast. CEIAT: Northern Ireland, UK; 2005.

14.

14. Peeters

, Madauss

. A proposed strategy against cost overruns in the space sector: The 5C approach. Space Policy 2008;24(2):80–89.

15.

15. Peterson

. Rapid cost assessment of space mission concepts through application of complexity indices. Jet Propulsion Laboratory; 2005.

16.

16. Vergaaij

, McInnes

, Ceriotti

. Economic assessment of high-thrust and solar-sail propulsion for near-earth asteroid mining. Space Research 2021;67(9):3045–3058.

17.

17. Sonter

. The technical and economic feasibility of mining the near-earth asteroids. Acta Astronaut 1997;41(4–10):637–647.

18.

18. Rush

, Roy

. Expert judgement in cost estimating: Modelling the reasoning process. Concurrent Eng 2001;9(4):271–284.

19.

19. Rosenbloom

. Technological discontinuities, organizational capabilities, and strategic commitments. Industrial Corporate Change. Oxford Academic 1994;3(3):655–685.

20.

20. Probst

, Nitzl

, Kraus

, Förstner

. Cost estimation of an asteroid mining mission using partial least squares structural equation modelling (PLS-SEM). Acta Astronaut 2020;167:440–454.

21.

21. Dai

. Product cost estimation: Technique classification and methodology review. J Manuf Sci Eng 2006;128:563–573.

22.

22. Trivailo

, Sippel

, Şekercioğlu

. Review of hardware cost estimation methods, models and tools applied to early phases of space mission planning. Progress in Aerospace Sciences 2012;53:1–17.

23.

23. Saaty

. The analytic hierarchy process: planning, priority setting, resource. McGraw-Hill: New York; 1980.

24.

24. Christensen

. The innovator’s dilemma. Harvard Business School Press; Boston; 1997.

25.

25. Crawford

. The long-term scientific benefits of a space economy. Space Policy 2016;37:58–61.

26.

26. Lewis

. Mining the sky: untold riches from the asteroids, comets and planets. Addison-Wesley: Reading, MA; 1996.

27.

27. Diederen

. Global resource depletion: metal minerals scarcity and the elements of hope. Peak Summit: Alcatraz, Italy; 2009.

28.

28. Greene

, Caracelli

, Graham

. Toward a conceptual framework for mixed-method evaluation designs. Educ Eval Policy Anal 1989;11(3):255–274.

29.

29. Pelech

, Roesler

, Saydam

. Technical evaluation of off-Earth ice mining scenarios through an opportunity cost approach. Acta Astronaut 2019;162:388–404.

30.

30. Berger

. NASA’s full Artemis plan revealed: 37 launches and a lunar outpost. Ars Technica; 2019.

31.

31.ASIME. Abstract book. In: Asteroid Science Intersections with In-Space Mine Engineering; Outcome from the ASIME 2018: Asteroid Intersections with Mine Engineering. ASIME: Luxembourg; 2018.

32.

32. McLeod

, Krekeler

. Sources of extraterrestrial rare Earth elements: To the moon and beyond. Resources 2017;6(3):40.

33.

33. Sowers

. Commercial lunar propellant architecture: A collaborative study of lunar propellant production. REACH 2018;13:100026.

34.

34. Lefeber

. Relaunching the Moon Agreement. AILA 2016;41(1):41–48.

35.

35. Hill

. Lunar simulants as feedstocks for ISRU processing: Mineralogy and chemistry. In: Earth & Space 2006: Engineering, Construction, and Operations in Challenging Environments. 2006; pp. 1–8.

36.

36. Mueller

. A review of lunar regolith excavation robotic device prototypes. In: Proceedings of the AIAA SPACE 2011 Conference & Exposition. AIAA; 2011; p. 7234.

37.

37. Agui

. Granular flow and dynamics of lunar simulants in excavating implements. In: Earth Space: Engineering, Science, Construction, and Operations in Challenging Environments. 2010; pp. 84–94.

38.

38. Skonieczny

. Considering the effects of gravity when developing and field testing planetary excavator robots. In: Field and Service Robotics: Results of the 10th International Conference. ( Wettergreen

, Barefoot

., eds.) Springer; 2016, pp. 299–594.

39.

39. Just

. Parametric review of existing regolith excavation techniques for lunar in-situ resource utilisation and recommendations for future excavation experiments. Planet Space Science 2020;180

40.

40. Gustafson

, White

, Fidler

. Oxygen production via carbothermal reduction of lunar regolith. SAE Int J Aerosp 2009;04(1):311–316.

41.

41. Schwandt

, Hamilton

, Fray

, Crawford

. The production of oxygen and metal from lunar regolith. Planet Space Science 2012;74(1):49–56.

42.

42. Jones

. Should oxygen, hydrogen, and water on the Moon be provided by Earth supply, life support recycling, or regolith mining? In: 50th International Conference on Environmental Systems. NASA Ames Research Center: Moffett Field, CA; 2021.

43.

43. Park

, Liu

, Kihm

, Taylor

. Characterization of lunar dust for toxicological studies. I: Particle size distribution. J Aerosp Eng 2008;21(4):266–271.

44.

44. Schreiner

, Dominguez

, Sibille

, Hoffman

. Thermophysical property models for lunar regolith. Advances in Space Research 2016;57(5):1209–1222.

45.

45. Anand

, Crawford

, Balat-Pichelin

, et al. A brief review of chemical and mineralogical resources on the Moon and likely initial in situ resource utilization (ISRU) applications. Planet Space Sci 2012;74(1):42–48.

46.

46. Cilliers

. Estimating the scale of space resource utilisation (SRU) operations to satisfy lunar oxygen demand. Planet Space Sci 2020;180.

47.

47. Shishko

. Business case analysis of lunar thermal mining. In: NASA Cost and Schedule Symposium – Johnson Space Center. California Institute of Technology; 2019.

48.

48. Colvin

, Crane

, Lal

. Assessing the economics of asteroid-derived water for propellant. Acta Astronaut 2020;176:298–305.

49.

49. Ehresmann

. Lunar based massdriver application. Acta Astronaut 2017;134:189–196.

50.

50. Nicholas

. Selection procedure. In: Mining engineering handbook. ( Nicholas

., ed.) Hartman H SME: New York; 1993; pp. 2090–2105.

51.

51. Miller-Tait

. UBC mining method selection. In: Mine Planning and Equipment Selection Symposium. MPES; 1995; pp. 163–168.

52.

52. Sarsfield

. The cosmos on a shoestring: small spacecraft for space and Earth science. RAND Critical Technologies Institute; 1998.

53.

53. Pelech

, Sibille

, Dempster

, Saydam

. A framework for off-Earth mining method selection. Acta Astronaut 2021;181:552–568.

54.

54. Oxnevad

. An investment analysis model for space mining ventures. In: International Astronautical Federation Conference. AACR; 1991.

55.

55. Andrews

, Bonner

, Butterworth

, et al. Defining a successful commercial asteroid mining program. Acta Astronaut 2015;108:106–118.

56.

56. Hein

, Matheson

, Fries

. A techno-economic analysis of asteroid mining. Acta Astronaut 2020;168:104–115.

57.

57. Sommariva

. Space resource utilisation by solely private markets. Acta Astronaut 2020;170:712–718.

58.

58. Metzger

. Economic modeling of the competitiveness of lunar water over Earth-launched water, including Wright’s law, optimization of reliability, and economies of scale. University of Central Florida; 2022.

59.

59. Dinner

. NASA squeezed oxygen from mock moon dust. It could help astronauts breathe easy one day. Space.com; 2023.

60.

60. Adler

. Gazing into the oracle: the Delphi method and its application to social policy and public health. Jessica Kingsley Publishers: London; 1996.

61.

61. Dalkey

, Helmer

. An experimental application of the Delphi method to the use of experts. Manage Sci 1963;9(3):458–467.

62.

62. Aitchison

. The lognormal distribution. University of Cambridge Press: Cambridge; 1957.

63.

63. Garson

. Partial least squares: regression & structural equation models. Statistical Associates Blue Book Series: Asheboro, NC; 2016.

64.

64. Hartman

, Baldwin

. Using technology to improve Delphi method. J Comput Civ Eng 1995;9(4):244–249.

65.

65. Brockhoff

. The performance of forecasting groups in computer dialogue and face-to-face discussions. In: The Delphi method: techniques and applications. ( Linstone

, Turoff

., eds.) Addison-Wesley: London; 1975.

66.

66. Rohrbaugh

. Improving the quality of group judgment: Social judgment analysis and the Delphi technique. Organ Behav Hum Decis Process 1979;24(1):73–92.

67.

67. Likert

. Technique for the measurement of attitudes. Arch Psychol J 1932;140:1–55.

68.

68. Taherdoost

. What is the best response scale for survey and questionnaire design; review of different lengths of rating scale/attitude scale/Likert scale. Int J Acad Res Manag 2019;8(1):1–10.

69.

69. Cronbach

. Coefficient alpha and the internal structure of tests. Psychometrika 1951;16(3):297–334.

70.

70. Barua

. Methods for decision-making in survey questionnaires based on Likert scale. J Asian Sci Res 2013;3(1):35–38.

71.

71. Nitzl

. The use of partial least squares structural equation modelling (PLS-SEM) in management accounting research: Directions for future theory development. Journal of Accounting LIterature 2016;37(1):19–35.

72.

72. Dash

, Paul

. CB-SEM vs PLS-SEM methods for research in social sciences and technology forecasting. Technol Forecast Soc Change 2021;173:121092.

73.

73. Becker

J-M

, Ringle

, Sarstedt

. Estimating moderating effects in PLS-SEM and PLSc-SEM: Interaction term generation × data treatment. JASEM 2018;2(2):1–21.

74.

74. Hair

. A primer on partial least squares structural equation modeling (PLS-SEM). Sage Publications: Thousand Oaks, CA; 2017.

75.

75. Fornell

, Larcker

. Evaluating structural equation models with unobservable variables and measurement error. J Marketing Res 1981;18(1):39–50.

76.

76. Hair

, Risher

, Sarstedt

, Ringle

. When to use and how to report the results of PLS-SEM. EBR 2019;31(1):2–24.

77.

77. Sarstedt

, Hair

Jr , Cheah

J-H

, et al. How to specify, estimate, and validate higher-order constructs in PLS-SEM. Australasian Marketing J 2019;27(3):197–211.

78.

78. Moores

, Chang

JCJ

. Ethical decision making in software piracy: Initial development and test of a four-component model. MIS Quarterly 2006;30(1):167–180.

79.

79. Cohen

. Statistical power analysis for the behavioural sciences, 2nd ed. Lawrence Erlbaum Associates: Hillsdale, NJ; 1988.

80.

80. Chin

. The partial least squares approach to structural equation modeling. Modern Methods for Business Research 1998;295(2):295–336.

81.

81. Barnston

. Correspondence among the correlation, RMSE, and Heidke verification measures; refinement of the Heidke score. Wea Forecasting 1992;7(4):699–709.

82.

82. Liengaard

, Sharma

, Hult

GTM

, et al. Prediction: Coveted, yet forsaken? Introducing a cross-validated predictive ability test in partial least squares path modeling. Decision Sciences 2021;52(2):362–392.

83.

83. Sanders

, Klein

. Systems engineering framework for integrated product and industrial design including trade study optimization. Procedia Comput Sci 2012;8:413–419.

84.

84. Govreau

. 2025 global mining project spending outlook. Eng Mining J 2025;226(Fasc. 1):30–33.

85.

85. Fofana

, Sanchez

. Mining boom and economic growth: A case study of the Simandou project in Guinea using a new CGE model. In: Global Trade Analysis Project. Presented during the 28th Annual Conference on Global Economic Analysis. Kigali, Rwanda; 2025.

86.

86.NASA. Project Pre-Phase A: Concept Studies. 2023. Available from: https://www.nasa.gov/reference/3-3-project-pre-phase-a-concept-studies/

87.

87. Shokri

, Haas

, G. Haas

, Lee

. Interface-management process for managing risks in complex capital projects. J Constr Eng Manage 2016;142(2).

88.

88. Ho

, Chen

, Kim

. Value of Bootstrapping Staged Deployment of Infrastructure: Case Study in Space Infrastructure Deployment. In: International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. 2017.

89.

89.CFBL Consulting. Change and Cost Management on Infrastructure Projects. 2024. Available from: https://www.cfbusinesslinks.com/change-and-cost-management-on-infrastructure-projects/

90.

90. Bissing

. An analysis of effects of technology readiness on cost growth. Air University, Air Force Institute of Technology, Department of the Air Force: Wright-Patterson Air Force Base, OH; 2021.

91.

91.NASA. NASA’s Management of The Artemis Mission. NASA Office of Inspector General. Report No. IG-22-003. 2021.

92.

92. Sharma

. A comparison of PLS and ML bootstrapping techniques in SEM: a Monte Carlo study. In: Springer Proceedings in Mathematics & Statistics (PROMS), vol. 56. Springer; 2013.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.84 MB