Metrics for building component disassembly potential: a practical framework

1.1 Background and current challenges

Due to its intensive and ever-increasing consumption of raw materials (ECORYS 2014) and energy (IEA 2024), the construction sector has become a critical area for environmental policies in the European Union (EU). With the development of energy efficiency regulations (EU 2024), attention is shifting toward the rapid reduction of product stage greenhouse gas (GHG) emissions. For example, whole-life GHG assessment of most new buildings will be mandatory in EU member states beginning in 2028, and GHG limits from 2030 onwards (EU 2024).

The interdependency between material use and GHG emissions has been clearly demonstrated (Röck et al. 2020). The focus on urgently needed GHG emission reductions is increasingly directed toward the reduction of product stage emissions, for compelling reasons: the consumption of construction materials is currently growing faster than the global population (Schiavina et al. 2022), which leads to increased GHG emissions (Zhong et al. 2021), despite the Intergovernmental Panel on Climate Change’s (IPCC) recommendations that construction sector emissions should be reduced by 80–90% by 2050 (IPCC 2022).

Significant environmental benefits associated with the intact disassembly and reuse of existing building products are identified in the literature (Eberhardt et al. 2021; Gallego-Schmid et al. 2020; Material Economics 2018) and should be considered in the early design phase (Graubner & Reiche 2001). Quantitative assessment methods are developed to provide systematic, evidence-based tools for evaluating how well construction projects align with set objectives. For instance, life-cycle assessment (LCA) has been widely adopted and become part of mainstream building design processes. In contrast, the assessment methods for circularity are in their infancy, yet are being researched intensively.

1.2 Current methods for assessing disassembly potential (DP)

An overview of the existing assessment methods was conducted through a database search in Scopus, complemented by Google Scholar with search terms ‘disassembly AND assessment AND construction OR buildings’. Key parameters of each method were mapped, including the number of steps, categories of input and data (Table 1).

One set of assessment approaches includes general guidelines, checklists and scoring systems, notably defined in ISO 20887 (2020), which establishes key principles. The EU’s Level(s) framework—developed for assessing and reporting on the sustainability performance of buildings—addresses the assessment of design for disassembly (DfD) potential by categorising the construction process into three stages based on the design phase, and by offering a checklist to facilitate disassembly (EC 2021). These criteria lay the groundwork for conceptual definitions, such as the classification of connection types, but are not designed for quantitative assessment of DP.

Early work by Durmisevic et al. (2003) introduced a methodology for whole-building reversibility assessment. While time-consuming with five steps and 10 criteria, the method’s factors within the disassembly criteria have been adopted in other DP assessment methods. The method by Verberne (2016) refined by van Vliet (2018) assesses the circularity level of building materials through three scales that form the steps: materials, systems and buildings. Cottafava & Ritzen (2021) offer both a full and a simplified method, tested with case studies. The Dutch Green Building Council (DGBC) method, developed by Alba Concepts (van Vliet et al. 2021), continues this development with a streamlined approach, assessing DP based on the connection type, connection accessibility, independency and geometry of the product edge, also considering the environmental impact of products. Recognised as comprehensive, the DGBC method is widely established (Attia et al. 2024; van der Zwaag et al. 2023).

Several authors propose integrating assessment with building information models (BIMs) and LCA methods. Luthin et al. (2023) present an LCA framework incorporating technical and social aspects. Kim & Kim (2023) suggest a method using BIM for automated DfD performance evaluation, combining mathematical evaluation, CO₂ emissions and financial aspects. Akinade et al. (2015) develop a BIM-based deconstructability assessment score for evaluating deconstruction performance during design. A similar method is introduced by Akanbi et al. (2018) for a BIM-based whole-life performance estimator. van der Zwaag (2023) put forth a BIM-integrated method for indicative early design phase and a detailed later stage DP assessment. According to the literature, the assignment of disassembly parameters is difficult, particularly at early design phases. The detailed disassembly parameters are intended to be derived from a data platform. However, it remains unclear how data regarding DGBC method indicators, such as the geometry of the product edge or independency, can be extracted from the database.

1.3 Research gaps and questions

Although the principles for DfD are mostly similar in the examined literature, numerous differing DP assessment methods have been developed. Unestablished and unstandardised circularity evaluation methods may hinder DP assessment during design (Corona et al. 2019; Rahla et al. 2019). Existing methodologies may not have been sufficiently tested with industry actors (e.g. demolition companies) who possess hands-on building disassembly knowledge (Attia et al. 2024). Although assessment methods exist, a lesser emphasis has been placed on critically examining their applicability for building design practice.

The comparison of steps and amount indicators in the most established line of development shows a trend towards simplification (Table 1). While the literature may not explicitly state the reasons for this shift, it seems likely that the intent is to reduce complexity and labour intensity, thereby enhancing the practicality of DP assessments. While BIM integration offers automation potential, current proposals have remained mostly theoretical. The available data are typically situational, which means that the amount and level of detail in the source, such as the BIM model, varies depending on the design phase and project scope. It is important to note that while advanced BIM-integrated LCA methods offer significant benefits, designers still need to perform a substantial amount of manual work to assess DP effectively.

Various methods have been proposed to assess the circularity of buildings. Some focus on a holistic overall circularity assessment, considering multiple indicators beyond materials, such as adaptability, economic factors and LCA (Luthin et al. 2023). Others specifically evaluate DP (van Vliet et al. 2021). However, there is a lack of DP assessment methods suitable for supporting practical design decisions made during the early design phase.

The aim of this paper is to develop a robust, simple and reliable DP assessment method that connects (1) to the practical design process and (2) can be utilised with BIM data from the early phases to the detailed design phase. The research questions are as follows:

• How can a building’s DP be evaluated during the early building design phases with the situationally available data used for LCA?

• How can quantitative values for building component DP be assessed in a sufficiently reliable and practical manner to improve existing LCA calculations, particularly for end-of-life (EoL) scenarios?

2.1 Research design

The research was conducted within the Finnish context. The research questions were addressed, and a streamlined DP assessment method was developed through two approaches (Figure 1). First, to assess the practicality of existing DP assessment methods, a review of current methodologies was conducted. The applicability of these methods was evaluated by analysing the BIM data and construction drawings of three cases study buildings, focusing on data quality and granularity. A streamlined assessment method was developed that has fewer and clearer parameters. Its practicality was then tested with the three case study buildings.

Figure 1

Simultaneously, a secondary research path was created to evaluate the reliability of DP assessments. A mapping review and categorisation of 263 openly available Finnish structural joint details was conducted, from which a set of 11 representative type joints was selected as the basis for expert interviews. Utilising a two-round Delphi method, 12 experts were interviewed in order to assess the DP of the type joints.

The reliability of second-round expert assessment results was evaluated using the intraclass correlation coefficient (ICC). Each type of joint was evaluated with both the proposed method and the existing DGBC assessment method. To assess the reliability of the proposed method and compare it with the DGBC method, Pearson and Spearman correlations were calculated against normalised expert interview results. The differences in correlation strength were tested by using three methods for dependent correlations: Steiger’s (1980) z-test, Meng et al.’s (1992) z-test and Zou’s (2007) confidence interval (CI) approach. Additionally, agreement between the methods and expert assessments was evaluated using a Bland & Altman (1986) analysis, which reported mean bias, standard deviation of differences (SD_Diff), CIs for the bias and limits of agreement (LoA). By using normalised interview results, all analyses ensured comparability.

The research design assessed the practicality and applicability of the proposed method within current building design processes, incorporating emerging DfD knowledge from the field.

2.2 Feasibility of the dp assessment methods

To evaluate the feasibility of current DP assessment methods (Table 1) within early design-phase Nordic building practices, three representative Finnish case study buildings were chosen (Table 2). The selection criteria included: (1) the availability of BIM data and (2) the diversity of typologies and construction types. While the sample does not encompass all building typologies, it includes three structurally distinct cases, allowing for a diverse validation of a realised building projects’ suitability for DP assessment. The BIMs used in this study reflect the final handover documentation typically produced by Finnish architects and structural engineers, representing the current standard in Finnish building projects.

Table 2

Case study building information and evaluated data.

	LITTLE FINLANDIA, ‘FIN’	POSTIVARIKKO, ‘POS’	VÄRE MODULE J, ‘VAR’
Image of BIM data
Location	Helsinki	Helsinki	Espoo
Construction year	2022	2023	2018
Design service life (years)	30	50	50
Floors	1	1	5
Heated net floor area (m2)	2,622	1,205	1,417
Building type	Multifunctional building	Assembly building	Educational building
Construction type	Volumetric unit	Post and beam	Prefabricated element
Primary material	Wood	Wood	Concrete
Utilised BIM data	Contractor final BIM model, 20 January 2021	Architect + Construction engineer final BIM model, 7 April 2022	Architect + Construction engineer final BIM model, 24 January 2019
11 Site elements	–	–	–
121 Foundations	–	–	–
122 Ground floors	○	○	○
123 Structural frame	○	○	○
1233 Columns	○	○	●
1234 Beams	○	○	●
1235 Intermediate floors	○	○	●
124 Facades	○	○	○
125 External decks	○	×	×
126 Roofs	○	○	○
131 Internal dividers	○	○	○
132 Space surfaces	○	○	○
133 Internal fixtures	–	–	–
134 Other internal space elements	×	×	×
135 Box units	×	×	×
2 Services elements	○	○	○

[i] Note: ● = Building information model (BIM) data quality allows for individual product-based disassembly potential (DP) assessment; ○ = BIM data quality allows for quantity take-off-based DP assessment; ‘–’ = BIM data do not allow sufficient information for either type of DP assessment; ‘×’ = BIM has no relevant data.

For all three projects studied, BIM data were insufficient to assess the DP of individual building products, apart from load-bearing steel beams, concrete columns and hollow-core slabs in Väre Module J (VAR) (Table 2). This limitation poses significant challenges for the applicability of all reviewed DP assessment methods (Table 1), which depend on precise product-based data, such as the independence of elements or the geometry of the product edge. Although complementary detail information was generally available through structural engineering drawings, evaluating this information across multiple sources—including element graphs, large-scale BIMs and detailed construction drawings—would require an extensive amount of additional work for DP assessment, even after final project materials have been completed. Furthermore, these documents are typically unavailable during early design phases, when preliminary BIM data are the only available source. Therefore, the provided BIM data pertain to material take-off granularity, necessitating a streamlined method for feasible DP assessment.

2.3 Mapping and analysis of typical joint details

For a representative selection of typical building structures, a mapping review of openly available Finnish load-bearing structural type connections was concluded using the authors’ university library and modern search engines, with full and partial search terms liitos (in Finnish, joint), periaateliitos (joint principle), vakioliitos (standardised joint), detalji (detail), vakiodetalji (standardised detail), periaatedetalji (detail principle) and mallidetalji (model detail). The review focused on load-bearing concrete, wood and steel connections, given that the primary load-bearing material for 98% of newly constructed Finnish buildings in 2008 was either concrete, wood or steel (Hytönen & Seppänen 2009). The load-bearing structures constitute the largest overground structural cause of product-phase GHG emissions (Häkkinen & Kuittinen 2020), and hence it was decided to focus on them. The territorial scope was limited to the Nordic countries, and the results were supplemented by findings collected by the authors during active architectural practice. A total of 263 structural joints were mapped and categorised (Table 3).

These connections were then classified according to their primary bearing material, accessibility, connection category and exact connection type (see Appendix 2 in the supplemental data online). Three primary materials and main construction systems were identified (Table 4). The selection of type joints to represent these categories and their potential variation in other EU countries was discussed with a consulting structural engineer. Despite variation in practice due to regulations and best practices, such as connectors (e.g. steel parts, screws/bolts) and different thicknesses of layers (e.g. for acoustics), a range of solutions was included to account for these variations within the selected joints. Based on the scoping review, 11 representative type joint details were chosen for interviews (Table 4; and see the supplemental data online).

All the type joint details depicted a structural connection between intermediate floor and load-bearing vertical structures, effectively covering a substantial number of structural connections multi-storey building load-bearing structures. As such, the number of the type joints was a compromise between the interview lengths and capturing a sufficiently broad sample from the existing building practice with potentially different DfD characteristics.

2.4 Expert interviews and interrater reliability analysis

The experts were interviewed during the development of the proposed assessment method (see Section 2.5). The rationale for assessing the mapping review’s representative type joints and their DP through expert interviews was based on the challenge of most existing DP assessment methods being overly theoretical for current BIM conventions (see Sections 1.2 and 2.2) and lacking field experience (Attia et al. 2024). To enhance reliability, a two-round Delphi method (Brady 2015) was employed to facilitate a structured, iterative interview process, enabling the experts to refine their assessments.

A total of 12 experts participated, comprising eight structural engineers and four deconstruction contractors. Informed consent was obtained from all participants prior to the interviews. Two rounds of interviews were conducted: the first round gathered initial insights and numerical assessments of type joints, while the second round allowed the experts to reconsider and adjust their evaluations based on the first-round results. Interviews were conducted online, recorded and transcribed automatically. The experts assessed each structural connection type joint on a scale from 1 to 5, covering three aspects: (1) accessibility; (2) detachability; and (3) the condition of structural components post-deconstruction. For more information on the participants, questions and structure and for an explanation of the numerical scale, see Appendix 4 in the supplemental data online.

To evaluate the reliability and replicability of the interview-based assessment, the ICC for the second-round results was calculated. A two-way random-effects model, absolute agreement, multiple raters/measurements ICC (2,k) was employed, adhering to the latest interrater reliability analysis guidelines (Koo & Li 2016).

2.5 Proposed method based on material quantities

A streamlined method for assessing buildings’ DP was proposed based on the findings of Sections 1.2. and 2.2. The proposed method was designed to comply with the studied project’s BIM data (Table 2) focusing primarily on resources and their distribution across assemblies. In this study, the term ‘resource’ means a building material or component delivered to construction site (e.g. a construction material or a window), while an ‘assembly’ is an entity comprising of multiple resources (e.g. an exterior wall or a modular element).

The proposed method focuses on evaluating the DP in relation to resource mass or global warming potential (GWP). The relation to mass reveals the utilisation potential of main material flows in the building and informs about the material efficiency in circular construction. The relation to GWP enables designers to ensure that the most GHG-intensive materials have a technical potential for reuse, hence reducing the emissions of the next building. Both indicators are beneficial because mass and GWP of building products or assemblies do not always correlate.

In assessing each resource’s DP, the evaluation focuses on two key parameters consistent with existing methods (Table 1):

• accessibility of the resource (e.g. an obstructing concrete cast layer)

• connection type of the resource (e.g. a screw).

These parameters are quantified by numerical values: accessibility C_a and connection type C_t both range from 0 < C ≤ 1, with assigned values set out in Appendix 7 in the supplemental data online. When multiple parameter options exist, selection follows these criteria:

• remove the minimum number of layers necessary to access the connection

• open the minimum number of connections required to dismantle the resource intact

• apply the most conservative accessibility value for multiple obstructing layers (e.g. closest to zero).

An example of a material list and evaluation of a type joint is provided in Figure 2 and Table 5.

Figure 2

2.6 Method application to type joints and correlation

The type joints (Table 4) were assessed using both an existing and a recommended (Attia et al. 2024) DGBC method (van Vliet et al. 2021) and the proposed new method described in Section 2.5. To evaluate the reliability of the proposed method and compare it with the DGBC method, Pearson and Spearman correlations were calculated against normalised expert interview results. Differences in correlation strength were tested using three methods for dependent correlations: Steiger’s (1980) z-test, Meng et al.’s (1992) z-test and Zou’s (2007) CI approach. Agreement between methods and expert assessments was further assessed using a Bland & Altman (1986) analysis, reporting mean bias, SD_Diff, CIs for the bias, and LoA. All analyses were based on normalised interview results to ensure comparability.

2.7 Method application to case study buildings

To test the proposed method’s practicality, a DP assessment was concluded on the three case study projects (Table 2). Material take-off quantities were calculated from the latest combined BIM data for each project. All BIM assemblies were cross-examined with construction engineer drawings to ensure accuracy in material identification. The materials and their respective service lives were matched to entries in the national generic emissions database CO2data.fi (Finnish Environment Institute 2024), intended for use in Finnish national building permit applications. Material weights were calculated with the circular assessment method (One Click LCA 2024) in September 2024. Replacements in module B4 were included in the material take-off mass. The carbon footprints were calculated using the Finnish national low-carbon assessment method (One Click LCA 2021) in January 2025, incorporating modules A1–A5, B4 and C1–C4. Calculation periods were based on the service lives provided in Table 2. Evaluation of Postivarikko (POS) and VAR assessed BIM material take-off data at the granularity level of a single construction type’s single material quantity. Little Finlandia (FIN) material take-off data C_a and C_t were evaluated as whole spatial elements following the principles clarified in Section 2.5.

3.1 Expert interview and interrater reliability analysis results

Numerical responses (grades 1–5) from expert interviews are illustrated in Figure 3. The least accessible connections, based on average scores, were observed in type joints 2–4 where connectors are embedded in concrete casts, and in type joint 9, where the connection is sealed by a massive brick wall. The lowest deconstructability scores, on average, were assigned to type joints 3, 5 and 8, which primarily consist of joints filled with concrete cast. In contrast, the highest deconstructability scores were given to type joints 6, 10 and 11, which feature dry connections. The underlying reasons for discrepancies are analysed in Section 3.4.

Figure 3

Second-round result ICCs and their 95% CIs were calculated using R, based on a mean-rating (k = 2), absolute-agreement, two-way random effects model. The resulting values (Table 6) range from 0.79 to 0.97, indicating reliability from ‘good’ to ‘excellent’. Thus, the interview results can be utilised in assessing the reliability of the DP assessment methods.

3.2 Method correlation with expert interviews and accuracy

The DP scores of the type joints were calculated using both the proposed and the DGBC (van Vliet et al. 2021) methods. These values, against normalised interview DP results, are plotted in Figure 4.

Figure 4

To evaluate the reliability of the proposed method and compare it against the DGBC method, statistical tests for differences between correlated correlation coefficients were conducted. As illustrated in Table 7, both Pearson and Spearman correlations indicated a very strong positive association between the proposed method and expert interview results, whereas the DGBC method showed only a strong association.

To determine whether this difference in correlation strength was statistically significant, Steiger’s (1980) z-test, Meng et al.’s (1992) z-test and Zou’s (2007) CI approach were applied. All three methods confirmed a significant difference between the proposed and the DGBC (van Vliet et al. 2021) methods in terms of alignment with expert assessments (Table 8). Steiger’s test yielded z = 2.045 (p = 0.041), and Meng et al.’s test yielded a comparable result (z = 2.051, p = 0.040), with a 95% CI for the difference of [0.040, 1.742]. Similarly, Zou’s method produced a 95% CI = [0.013, 0.826], which does not include zero.

These results confirm that the proposed method aligns significantly more closely with expert evaluations than the DGBC method, reinforcing the proposed method’s high reliability. Moreover, the proposed method achieves this performance while remaining more efficient and computationally accessible.

To evaluate further the agreement between the assessment methods and expert interview results, a Bland & Altman (1986) analysis was conducted (Figure 5). The proposed method demonstrated a small mean bias of –0.12 with SD_Diff = 0.134, and 95% LoA ranging from –0.383 to 0.142. The 95% CI for the bias (–0.21 to –0.03) did not include zero, indicating a slight but consistent underestimation relative to the expert scores.

Figure 5

In comparison, the DGBC method exhibited a slightly larger mean bias of –0.165 (SD_Diff = 0.114), with narrower LoA (–0.389 to 0.059) and a 95% CI for the bias ranging from –0.242 to –0.088. This also indicates a systematic underestimation, with a slightly tighter agreement range but greater deviation from expert evaluations.

Overall, while both methods slightly underestimate the expert ratings, the proposed method shows a smaller bias and broader but more symmetric agreement limits, supporting its use as a reliable and consistent approximation of expert judgment.

3.3 Case study building evaluations

The proposed method was applied to assess the DPM_total and DPGWP_total of the three case study buildings (Table 2). The results align with expected disassembly feasibility, demonstrating consistency between the method’s outputs and anticipated real-world performance. The findings were categorised according to the Construction 2000 building part classification, providing a structured analysis of reversibility across different building components.

Case study FIN—constructed from spatial elements—received the highest relative DPM_total at 72%, whereas the wooden column-beam constructed case POS reached 33% and the partially in-situ cast concrete structure of case VAR had a lower DPM_total of 15% (Figure 6).

Figure 6

When assessing the relative DPGWP_total, case VAR demonstrated relatively high results at 45% compared with its DPM_total, which was 15%. This discrepancy can be partly attributed to the extensive use of carbon-intensive yet relatively light glass facades and partition systems in VAR (Figure 7).

Figure 7

3.4 Discrepancies

4.1 Reliability and applicability of the proposed method

The second-round interview intraclass reliability results indicated good to excellent reliability (Table 6), enabling reliable comparison with DP assessment methods. The proposed method and the DGBC method were evaluated through multiple statistical comparisons with expert interviews. Additionally, the practicality of the proposed method was tested using three case studies. The strong positive correlation with expert assessments and consistent case study results suggest that the proposed method is both reliable and user-friendly in practical DP assessment scenarios. Notably, the streamlined method facilitates DP assessments in both early and developed design phases, supporting the shift towards more DfD-oriented workflows in building design practice.

Completing a building’s whole-life LCA requires an inventory of components defined according to the LCA method or regulation. Major decisions affecting a building’s GWP typically occur in early design phases, as described by Häkkinen et al. (2015). However, for a DP assessment, further maturity is needed in understanding product connections, and access-obstructing layers. This implies that optimisation for DfD should be considered equally early, alongside with requirements for legal GHG limits. As connection types and their accessibility can be related to other technical requirements (e.g. fire safety or acoustic performance may result in covering a joint with gypsum board), it would be important to introduce the DP along other essential technical requirements.

4.2 Limitations and uncertainties

Evaluating the reliability of the proposed method will require more case studies with varied building typologies, also from different locations. In addition, non-load-bearing building parts and building service systems should be tested following a similar process of assessments and interviews. The presented three case studies align with the limited expert interview sample size of 12, but a larger dataset could reinforce or refine the method and numerical values listed in Appendix 7 in the supplemental data online.

The proposed formula assesses DP, but smaller sample sizes may not accurately reflect the DP of individual resources. Larger variances in smaller sample sizes could lead to all or none of a project’s resources being disassembled intact, contrary to the implied DP_r value.

The method was excessively tested by the research group to eliminate vague terminology and unclear connection type and accessibility options. More assessments could reveal the need for additional options to address less typical or more complex structures and testing with a larger group could propose unclarity in accessibility or connection type options. The interviews were conducted with load-bearing-type joints due to the load-bearing structures’ significance as the source of whole-life carbon emissions. However, more type joints for non-load-bearing structures would be needed to address the method’s reliability in holistic building component assessment.

The scope of the proposed method is strictly on the assessed resource’s DP. It relies on existing LCA tools to obtain the resource’s GWP and estimated service life. This approach ensures each method calculates relevant parameters only once, making the method complementary but not a standalone holistic tool.

4.3 Further research

The proposed method was primarily tested within the Finnish context, using most common Finnish type joints based on available data (see Section 2.3). Within this context, the method demonstrated a very strong positive correlation with expert evaluations at the assembly level (see Section 3.2). However, further research is needed to assess the accuracy of DPGWP_r and DPM_r across different contexts, including predominant structural systems in other Nordic countries and Europe. Additionally, validating the accuracy of DPGWP_total and DPM_total requires comparing the method’s results with demolition inventories from completed intact disassembly projects across Europe.

Challenges related to implementing existing evaluation methods with Finnish BIM data detail level were identified in Section 2.2. The development of the proposed method aimed to streamline assessment compared with current methods (Table 1). The goal was to create a reliable, concise approach for DP assessment. Broader implementation in practical design workflows is necessary to fully assess its impact on DfD-oriented design during early design phases.

Service life calculations for replacement scenarios (module B4) included in M_r and GWP_r were based on Finland’s national database (Finnish Environment Institute 2024). These values incorporate conservative assumptions, defaulting to deep renovation after 50 years. Consequently, the proposed method’s outcomes might reflect two potentially misleading scenarios:

• Resources with high GWP during product and construction stages (modules A1–A5), short service life, and high C_a and C_t values may inflate DPGWP_total. Such resources should generally be avoided due to their high environmental impact.

• Resources with low GWP in modules A1–A5, high mass, long service life, and low C_a and C_t could decrease DPGWP_total and DPM_total. These resources should sometimes be prioritised due to their low environmental impact.

These discrepancies can be identified by cross-checking the DPGWP_total and DPM_total results with the LCA results. While such discrepancies were not observed in case study assessments, further research is required to explore correlations between high DPGWP_total and DPM_total values and low whole-life carbon emissions.