Comparing Construction Carbon Estimates Across System Boundaries: A Boundary-Normalized Protocol for Materials, Maintenance, and BIM-Structured Data

1. Introduction

Carbon estimation in construction is sensitive to methodological scope. A design option can appear preferable or inferior depending on whether the estimate includes only product-stage materials, extends to transport and on-site activities, or also incorporates scheduled maintenance. The practical consequence is not merely variation in totals. Boundary choice can change the rank ordering of alternatives, especially when one option reduces initial emissions but requires more frequent intervention, or when another has a carbon-intensive production stage but a longer service interval. A rigorous comparison therefore requires a procedure that distinguishes differences caused by physical design from differences caused by accounting scope.

Evidence from pavement assessment illustrates the issue clearly. One comparative study evaluated embodied energy, embodied carbon, and cost for flexible and rigid pavement alternatives under a common functional unit, while explicitly limiting the boundary to construction and maintenance phases. It also noted that many prior studies are difficult to compare directly because they use different system boundaries, functional units, analysis periods, and data sources. That observation is directly transferable beyond pavements: any cross-study or cross-project carbon comparison can be distorted when stage inclusion rules differ. [1]

At the same time, material innovation intensifies the boundary problem. Low-carbon cementitious systems, recycled aggregates, industrial by-product substitutions, geopolymers, and smart materials can reduce embodied emissions at production stage, extend durability, or alter operational and replacement patterns. Their apparent carbon benefit depends on whether the estimate captures only initial material production or also service-life effects and secondary replacements. A boundary-aware protocol is therefore needed to compare construction carbon estimates in a way that is methodologically explicit rather than implicitly governed by whichever stages an analyst happened to include. [2] [3]

2. Research question and evidence scope

This manuscript addresses the following research question: how should construction carbon estimates be compared when competing studies, project options, or digital models adopt different system boundaries across production, transport, construction, maintenance, operation-related material effects, and end-of-life stages? The purpose is methodological rather than empirical. The article does not report new measurements, project samples, or performance results. It proposes a comparative protocol that can later be tested prospectively.

The evidence scope is a targeted narrative synthesis of supplied studies relevant to three linked issues. First, direct application evidence comes from pavement estimation work that quantifies embodied energy and embodied carbon for construction and scheduled maintenance under explicitly defined functional units and alternative material substitutions. Second, material-context evidence comes from discussions of low-carbon construction materials and smart materials, which show that decarbonization pathways can arise through lower production emissions, circular resource use, durability improvement, and reduced replacement frequency. Third, methodological foundations come from BIM-oriented studies that frame digital models as lifecycle information structures supporting sustainability assessment, coordination across design-construction-operation phases, and integration of heterogeneous project records for analysis and optimization. [1] [2] [3] [4] [5] [6]

Because the evidence base is intentionally bounded, the manuscript does not claim exhaustive coverage of all lifecycle assessment standards or all carbon-accounting traditions. Its contribution is narrower: a reproducible comparison protocol for boundary-sensitive construction carbon estimation, designed so that future users can implement it with project-level material, equipment, maintenance, and BIM data.

3. Prior findings relevant to boundary comparison

The pavement study provides the most direct precedent for boundary-controlled comparison. Its analysis used a common functional unit of a 1 km by 3.5 m single-lane pavement and included material production, transportation, construction, and scheduled maintenance activities. To compare flexible and rigid alternatives with different design periods, the study annualized total embodied energy and embodied carbon over 15 years for flexible pavements and 25 years for rigid pavements. It also showed that alternative materials such as supplementary cementitious replacement and recycled aggregates can reduce embodied energy and carbon, while stressing that local design standards, material availability, traffic conditions, and maintenance strategies strongly influence outcomes. [1]

This prior work is valuable because it demonstrates two transferable principles. First, direct comparison becomes stronger when the functional unit and stage definitions are held constant. Second, annualization can partially restore comparability when service intervals differ. The proposed protocol extends these principles by formalizing when annualization is sufficient and when boundary differences remain too large for direct ranking. [1]

The low-carbon materials context adds an important caution. Decarbonized materials are not interchangeable accounting objects. Cement replacement by industrial by-products, recycled aggregate use, alkali-activated systems, and functional composite designs may lower production-stage emissions, but questions remain about long-term service performance, industrial scalability, standardization, and carbon-accounting methods. These unresolved issues matter for boundary comparison because a material that appears favorable under a cradle-to-gate estimate may require different maintenance assumptions under a construction-plus-maintenance boundary. [2]

The smart materials literature points to another boundary condition: carbon reduction can be achieved through multiple mechanisms located at different lifecycle stages. Some materials mainly reduce embodied emissions during production, while others contribute through operational energy savings, durability extension, self-healing, circularity, or temporary carbon sequestration. The review also states that no single solution dominates all sustainability dimensions. This directly implies that system-boundary choice can shift the apparent advantage among alternatives, because different mechanisms become visible only in certain boundary definitions. [3]

BIM-oriented studies do not supply carbon factors themselves, but they are methodologically relevant because they define BIM as an information-rich lifecycle framework. One study describes 7D BIM as embedding sustainability and energy performance analysis across the building lifecycle, while another emphasizes BIM-supported sustainability assessments and lifecycle performance evaluations. A further study shows how BIM data can be integrated with heterogeneous management records and simulation models to support predictive and optimization workflows. These findings support the use of BIM not as a carbon calculator by itself, but as the data architecture for traceable stage mapping, quantity extraction, maintenance scenario definition, and comparator generation. [5] [4] [6]

4. Conceptual framework: what counts as a boundary difference

The proposed protocol defines a system boundary as the ordered set of lifecycle modules included in a carbon estimate, together with the rules used to attribute flows to those modules. For construction-focused comparison, the base module library is: P = material production; T = transport to site; C = on-site construction and equipment use; M = scheduled maintenance and replacement; Om = operation-related material effects directly attributable to the design option, excluding whole-building energy unless explicitly modeled; E = end-of-life processing, demolition, and disposal or recovery. This library is intentionally simple so that it can be applied to pavements, buildings, and other infrastructure.

Each estimate is represented by a boundary vector B = [bP, bT, bC, bM, bOm, bE], where each element equals 1 if the module is included and 0 otherwise. Boundary comparison then becomes a formal task rather than a narrative description. Two estimates may be nested, partially overlapping, or disjoint in some modules. A nested case occurs when one estimate includes all modules of another plus additional modules. A partially overlapping case occurs when each estimate includes at least one module omitted by the other. Only the first case admits straightforward incremental interpretation.

Boundary difference must be separated from parameter difference. If two estimates both include production and maintenance but use different maintenance intervals, transport distances, or substitution rates, their boundaries are the same even though their assumptions differ. The protocol therefore stores assumptions in a separate parameter set A. Boundary harmonization is not equivalent to full methodological harmonization; it is a precondition for it.

Material substitution profiles require special handling. A low-carbon alternative may change not only the production factor of a material but also service life, inspection frequency, construction method, and end-of-life behavior. For that reason, the protocol treats each substitution profile as a design scenario S linked to both quantities and stage-specific factors, rather than as a single coefficient applied to the production stage. [2] [3]

5. Proposed methodology

The protocol proceeds in eight steps. Step 1 defines the comparator set: two or more design alternatives, studies, or digital models intended for comparison. Step 2 assigns a common functional unit FU and analysis period H whenever direct harmonization is possible. Step 3 maps each estimate to the module library and records the boundary vector B. Step 4 extracts or reconstructs stage-specific carbon values by module. Step 5 standardizes quantities and maintenance events relative to FU and H. Step 6 computes direct comparability status. Step 7 generates boundary-normalized outcome measures. Step 8 documents residual incomparabilities, data pedigree, and sensitivity scenarios.

Operational definitions are as follows. Let C_i,m denote carbon assigned to comparator i in module m after source-specific stages have been mapped to the common module library. Total carbon for comparator i is TC_i = Σ_m b_i,m C_i,m. For any pair of comparators i and j, define the common-module indicator g_ij,m = 1 if b_i,m = 1 and b_j,m = 1, and 0 otherwise. Shared-stage carbon is then defined separately for each comparator on the common module set: SSC_i|j = Σ_m g_ij,m C_i,m and SSC_j|i = Σ_m g_ij,m C_j,m. This definition is symmetric in module selection but not in magnitude, because each comparator retains its own stage values on the same shared basis. For modules not shared, the exclusive-stage carbon of comparator i relative to j is XSC_i|j = Σ_m max(0, b_i,m - b_j,m) C_i,m. If a comparison horizon is required, annualized carbon is AC_i(H*) = TC_i(H*)/H*, where H* is a single harmonized comparison window chosen before calculation. H* is the contractual or analytical window used for comparison, not an implicit service life unless the protocol explicitly sets the two equal.

The operational rule for direct comparability is intentionally narrow. Two estimates i and j are directly comparable for total-carbon ranking only if: (1) FU_i = FU_j; (2) both are recalculated on the same harmonized comparison window H*; (3) after formal stage mapping, the symmetric difference of their included module sets is empty; and (4) they use materially aligned accounting conventions for the decision at hand, namely the same emission-factor convention and allocation basis, with no unresolved geographic or temporal mismatch large enough to change interpretation. Direct comparability in this protocol therefore means comparability of reported totals after boundary harmonization; it does not imply that all uncertainty has been eliminated.

When direct comparability fails, reporting follows a fixed precedence rule rather than analyst discretion. First, map source-specific stage labels to the common module library through a documented function M: L -> {P, T, C, M, Om, E, U}, where L is the set of source labels and U is a residual unmapped category. Every original stage label must be assigned to exactly one target module or to U. If estimated carbon assigned to U exceeds 5% of the source total, the pair is classified as not adequately mappable and only descriptive reporting is permitted. Second, if the common-module set is empty, no quantitative comparison is reported beyond separate totals and boundary descriptions. Third, if the pair is nested, report SSC_i|j, SSC_j|i, module-specific deltas on shared modules, and each comparator's exclusive-stage carbon relative to the other. Fourth, if the pair is partially overlapping but non-nested, report the same shared-stage quantities and deltas, while treating XSC_i|j and XSC_j|i as two separate exclusive-stage terms rather than as a single boundary-gap quantity. Fifth, report annualized shared-stage carbon only when both comparators can be expressed on the same H* using technically plausible maintenance schedules. Under this rule, totals from non-identical boundaries are never ranked by themselves; the primary comparison output is shared-stage carbon, with exclusive-stage terms presented separately.

Carbon by module is calculated from quantities and factors. For a material k used in module m, C_i,m,k = Q_i,m,k × EF_i,m,k, where Q is the quantity associated with FU and EF is the emission factor under the stated data source and processing assumptions. For equipment- or activity-based contributions, the same structure applies, with Q expressed as fuel use, operating time, or transport work. Maintenance-stage carbon is the sum across scheduled interventions r within horizon H: C_i,M = sum over r of I(t_r less than or equal to H) × C_i,M,r, where I is an indicator function equal to 1 when the intervention occurs within the horizon and 0 otherwise. [1]

A BIM-structured implementation is recommended when project data exist digitally. Quantity extraction can populate Q values from model elements; schedule-linked maintenance assumptions can define intervention times; and stage tags can be stored as attributes attached to assemblies or activities. The rationale is supported by work describing BIM as a lifecycle framework spanning design, construction, operation, cost, and sustainability dimensions, and by studies emphasizing BIM-enabled sustainability assessment and integration of heterogeneous data for analysis and optimization. [5] [4] [6]

6. Variables, assumptions, and scenario structure

The minimum variable set contains: functional unit FU; harmonized comparison window H*; boundary vector B; original stage-label set L; mapping function M; material quantities Q by module; emission factors EF by material or activity; transport distance and mode; equipment productivity or fuel use where available; maintenance regime R = {type, frequency, timing, quantity}; substitution profile S; accounting convention K covering allocation and factor basis; and data pedigree D describing source type, temporal relevance, and regional relevance. Optional variables include uncertainty ranges, salvage assumptions, waste rates, and operational effects directly induced by the material choice. If a study also reports a nominal design life or service life, that term should be recorded separately from H* and not substituted for it unless the protocol explicitly harmonizes them.

The protocol adopts four assumptions that must be stated explicitly. Assumption 1: stage mapping is sufficiently complete that at least 95% of estimated carbon within the declared boundary can be assigned to P, T, C, M, Om, or E; any remainder must be placed in the residual category U and reported. Assumption 2: functional equivalence holds within the chosen FU; compared alternatives satisfy the same structural or service requirement. Assumption 3: maintenance scenarios are technically plausible for each material system and are specified prospectively rather than tuned to favor an option. Assumption 4: emission factors, quantities, and allocation rules refer to the same physical definitions and are regionally acceptable for the decision context. None of these assumptions is guaranteed merely by the presence of a BIM model.

Scenario structure should distinguish at least three scenario families. Baseline scenarios use conventional materials and prevailing maintenance schedules. Substitution scenarios modify one or more material inputs, such as supplementary cementitious replacement or recycled aggregates, while preserving required performance. Boundary-extension scenarios keep design assumptions fixed but expand or contract included modules. This separation is necessary because a design effect and a boundary effect are analytically different. The pavement study, for example, changed both material substitution levels and pavement-type designs while holding the construction-plus-maintenance boundary consistent; the proposed protocol adds a second analytical layer by recalculating how those same alternatives would compare under narrower or broader boundaries. [1]

A single default governance rule is preferable to an open menu of thresholds. The default rule in this protocol is binary: if direct comparability is not achieved, no overall total-carbon ranking is issued. Instead, the report must present shared-stage results first and exclusive-stage terms second. As a sensitivity exercise for future applications, analysts may preregister an auxiliary threshold based on the ratio max(XSC_i|j, XSC_j|i)/max(SSC_i|j, SSC_j|i); for example, they may examine whether conclusions change when this ratio is constrained below 0.10, 0.20, or 0.30. Such thresholds are decision rules for robustness checking, not empirical laws established by this manuscript.

7. Comparator design and relation to prior methods

The proposed protocol differs from the pavement method in purpose, not in contradiction. The pavement study already improved comparability by fixing a functional unit and common construction-maintenance boundary, and by annualizing results across different design periods. The present protocol generalizes that logic to cases where comparators do not start with aligned boundaries. It therefore supplements direct estimation with a comparison layer that can determine whether total values may be ranked, must be annualized, or should be decomposed into shared and non-shared modules. [1]

Compared with broad discussions of low-carbon material development, the protocol adds an operational accounting structure. Material studies identify major decarbonization pathways and note unresolved issues in durability, industrial scaling, and carbon-accounting standardization. The protocol translates those concerns into explicit variables: substitution profile, maintenance regime, and data pedigree. This matters because an attractive cradle-to-gate material is not necessarily attractive when maintenance-stage consequences are included, and vice versa. [2]

Compared with smart-material sustainability reviews, the protocol is narrower and more formal. The review reports that different smart material families contribute to carbon reduction at different lifecycle stages and that no single solution dominates all sustainability dimensions. The protocol converts that insight into a boundary rule: stage-specific benefits must not be compared as if they were system-wide benefits unless the compared boundaries expose the same lifecycle mechanisms. A self-healing material, for example, should not be judged only on production-stage carbon if its main claimed benefit is reduced replacement frequency. [3]

Compared with BIM adoption and BIM-integration studies, the protocol uses BIM instrumentally rather than behaviorally. The cited BIM literature shows that BIM supports lifecycle information management, sustainability analysis, and integration of heterogeneous records, but it also shows that organizational readiness, management support, and implementation barriers influence actual deployment. Accordingly, the protocol assumes BIM can improve traceability only when project organizations can populate and govern the required data fields across lifecycle stages. In settings without that capacity, a non-BIM tabular implementation remains possible but may have lower completeness and auditability. [4] [5] [6]

8. Prospective validation design

Because this article is a protocol, validation is prospective. The first validation objective is internal consistency: when two estimates are already boundary-identical, the protocol should reproduce ordinary direct comparison without altering rankings except through explicit annualization choices. The second objective is discrimination validity: when estimates differ only by added modules, the protocol should isolate the incremental carbon attributable to the added boundary rather than embedding it in an undifferentiated total. The third objective is decision robustness: under plausible alternative maintenance assumptions, the protocol should reveal whether ranking changes are driven by physical scenario differences or by boundary expansion.

A suitable test design would use paired case sets. One set would contain alternatives with matched functional units and matched construction-maintenance boundaries, analogous to the pavement example. Another set would intentionally vary boundaries across product-only, construction-only, construction-plus-maintenance, and extended lifecycle scenarios while holding quantities as constant as possible. For each pair, analysts would record the original labels, the mapped module set, whether direct comparability was achieved, whether decomposition was invoked, and whether rank order differed between shared-stage and total-stage results. Comparability classification should then be checked against a preregistered adjudication rubric rather than informal expert opinion alone. [1]

Evaluation metrics should include comparability classification agreement against adjudication, proportion of comparisons requiring decomposition, size of exclusive-stage carbon relative to shared-stage carbon, ranking stability across nested boundaries, and data completeness by module. The adjudication procedure should be specified in advance: at least two qualified assessors independently review each pair using the protocol's formal conditions on FU, H*, mapped module sets, accounting convention, and residual unmapped share; disagreements are resolved by a third assessor using the same rubric, and both initial and resolved decisions are retained in the validation record. In BIM-enabled implementations, additional process metrics may include the proportion of quantities auto-extracted from the model versus manually entered, and the traceability of maintenance events to model objects or schedules. No target numerical benchmarks are asserted here; they should be set in future empirical studies. [5] [6]

Sensitivity analysis should vary at least horizon length, maintenance frequency, substitution rates, and emission-factor sources. The pavement evidence suggests that results are strongly affected by regional factors, design standards, traffic conditions, and maintenance strategies. A validation study should therefore test whether the protocol continues to distinguish boundary effects from parameter effects when those contextual variables change. [1]

9. Failure cases and limits of generalization

The protocol can fail when functional equivalence is weak. If two alternatives do not deliver the same service level, carbon comparison per nominal functional unit may be misleading even if boundaries match. A second failure case arises when maintenance behavior is speculative, especially for emerging low-carbon or smart materials whose long-term degradation mechanisms are not yet well established. In such cases, decomposition can still be informative, but annualized or maintenance-inclusive rankings should be treated as provisional. [2] [3]

A third failure case concerns data architecture. BIM can support lifecycle traceability, but the cited literature also indicates that readiness, management support, and implementation barriers affect whether digital workflows are actually reliable across design, construction, and operation phases. If maintenance records, substitution details, or construction equipment data are missing, the protocol may classify boundaries correctly while still producing uncertain stage totals. [4] [5]

Generalization is limited in three ways. First, the module library is simplified for construction-focused comparison and does not substitute for a full lifecycle standard. Second, the protocol is strongest for alternatives whose principal differences can be represented through quantities, factors, and scheduled interventions; it is weaker for highly dynamic use-phase interactions unless those interactions are explicitly modeled. Third, the narrative evidence base includes pavements, low-carbon materials, smart materials, and BIM workflows, so transfer to other asset classes is reasoned but not yet empirically demonstrated.

10. Conclusion

This manuscript proposes a conceptual and procedural basis for comparing construction carbon estimates across different system boundaries. Its bounded contribution is a boundary-normalized protocol that separates direct comparison from decomposed comparison, defines explicit variables and module mappings, and states an operational rule for when total carbon values may be ranked without qualification. The framework also links material substitution, maintenance scheduling, and BIM-structured data management within a single comparison procedure.

The synthesis of prior work supports three conclusions. First, direct comparison is strongest when functional unit, boundary, and analysis period are harmonized, as shown in comparative pavement assessment. Second, low-carbon and smart material options can shift carbon burdens across lifecycle stages, so boundary choice can alter apparent superiority. Third, BIM offers a practical data structure for lifecycle stage mapping and traceability, but its usefulness depends on organizational and informational readiness. [1] [2] [3] [5] [4] [6]

What remains untested is empirical performance of the protocol in real project datasets: its reliability under incomplete records, its effect on ranking stability, and its sensitivity to uncertain maintenance assumptions and regional emission factors. Those questions require prospective case-based validation. Until such testing is completed, the protocol should be used as a disciplined method for transparent comparison rather than as proof that any specific material or design is categorically lower carbon under all boundaries.

References

1. Pradeep, B.; Lavanya, N.; Archana, M. R.; Basutkar, S. M.; Anjaneyappa, V. (2026). Energy and carbon estimation in construction and maintenance of pavements using conventional and alternative materials: a comparative analysis. Scientific Reports. 16(1), 25753. doi:10.1038/s41598-026-55867-1.
2. Zhang, Junfei (2026). Low-Carbon Construction and Building Materials. Materials. 19(9), 1726. doi:10.3390/ma19091726.
3. Martínez Ángeles, Hugo; Navarro Rubio, Cesar Augusto; Ríos Moreno, José Gabriel; Gonzalez-Garcia, Ivan; Reyes Araiza, José Luis; Garduño Aparicio, Mariano; Chavero-Navarrete, Ernesto; Trejo Perea, Mario (2026). Smart Materials Employed in the Construction Industry: A Systematic Review of Types, Properties, Applications, and Sustainability Performance. Materials. 19(12), 2676. doi:10.3390/ma19122676.
4. Alrasheed, Khaled A.; Alotaibi, Abdulaziz; Baarimah, Abdullah O.; Alawag, Aawag Moshen; Almujibah, Hamad; Ramu, Madhusudhan Bangalore (2026). BIM-Supported green building adoption in the construction industry. PLOS One. 21(9), e0356894. doi:10.1371/journal.pone.0356894.
5. Alsehaimi, Abdullah; Ghani, Muhammad Usman; Baarimah, Abdullah O.; Alawag, Aawag Moshen; Ramu, Madhusudhan Bangalore (2026). Mitigating obstructions to attain successful application and implementation of building information modeling (BIM) in residential construction projects’ lifecycle. Scientific Reports. 16(1), 12389. doi:10.1038/s41598-026-43261-w.
6. Li, Ruixin; Almenweer, Reem A.; Isleem, Haytham F.; Metwally, Ahmed Sayed M. (2026). Optimizing construction risk management through BIM integration and strategic planning. Scientific Reports. 16(1), 28592. doi:10.1038/s41598-026-64364-4.