Transportation-induced damage to prefabricated volumetric building modules significantly increases modular construction project costs, yet clear transportability criteria are lacking. This study systematically examines the structural behavior of modules under transportation vibrations, identifies key damage mechanisms, and proposes a novel module damage criterion (MDC) for reliable damage prediction. The MDC was formulated based on the quantification of damage mechanisms elucidated through full-scale shaking table tests on a cold-formed steel (CFS) module. Its predictive capability was then assessed using both laboratory test data and a real-world module transportation scenario. Test results demonstrated that module damage within vehicle-module interaction systems originated from three mechanisms: (1) resonance-induced dynamic response amplification (up to tenfold); (2) road class-driven plasterboard response intensification; and (3) localized failure at plasterboard-to-steel screw connections due to stress concentration and material vulnerability. The MDC accurately predicted damage initiation in critical components of the roof-ceiling systems, with timing errors below 11 %. Therefore, the MDC effectively evaluates modules' transportability under the hazardous condition of vehicle-module resonance, supporting decision-makers and enhancing operational efficiency and reliability of modular construction.

Modular Integrated Construction (MiC) shifts construction off-site, but on-site bolting of steel module connections remains hazardous and manual. This paper presents Tower-Crane Robotic Arm Bolting System (T-CRABS), an automation system that repurposes the tower crane, a ubiquitous site asset, to deploy a suspended robotic arm. This approach overcomes the payload and access limitations of aerial or ground robots in cluttered workspaces. The contributions to construction automation are threefold: (1) a system design that enables robotic bolting operations by leveraging existing site infrastructure without occupying floor space; (2) a kinetostatic model that captures center-of-gravity shifts during arm motion, providing accurate predictions of platform tilt essential for stable bolting in construction site conditions; and (3) a thrust-based active damping controller that ensures robust pose stabilization during bolting without energy-intensive levelling. Laboratory validations confirm effective swing suppression, robust disturbance rejection with low thrust utilization, repeatable accuracy, and successful bolting. T-CRABS provides a practical path to automate critical MiC assembly, extending robotic automation from factory to construction site.

The growth of Modular Integrated Construction (MiC) has intensified the transportation of oversized goods in urban areas, posing challenges for semi-trailer trucks due to blind spots and heavy traffic. Multi-camera Surround View Monitoring (SVM) systems have the potential to provide 360 degrees visibility to eliminate blind spots and enhance transportation safety. However, existing SVM systems are primarily designed for passenger vehicles and face limitations when applied to semi-trailer trucks due to camera placement constraints and dynamic tractor-trailer rotations. This paper develops a systematic methodology for designing and implementing a novel SVM system tailored for oversized semi-trailer trucks. It features three innovative components: a Camera Placement Optimization (CPO) method to determine the optimal camera placement, an Articulation Angle Estimation (AAE) method to determine articulation angle utilizing spatial and temporal information, and a Dynamic Image Concatenation (DIC) method to generate Bird's Eye View (BEV) images. These methods were thoroughly evaluated in a simulation environment, with results showing that: 1) the CPO method is applicable to various semi-trailer trucks (e.g., drop-deck, flatbed, dry van) and can improve BEV image sharpness by 10% compared to conventional installation plans; 2) the AAE method enhances articulation angle estimation with over 50% greater accuracy and reduces fluctuations compared to existing spatial feature-matching approaches; and 3) the DIC method can generate panoramic BEV images that effectively depict the truck's surroundings and enable distance estimation to potential risks in challenging transportation scenarios.

Embodied carbon (EC) accounts for a substantial part of greenhouse gas emissions. Steel modular construction (MC) has been promoted worldwide. However, limited research has investigated the EC of steel modular high-rises. This paper aims to systematically examine the cradle-to-end-of-construction EC of steel modular high-rise residential buildings, and identify whether and how it can achieve EC reductions compared with conventional concrete construction. A multi-level spatiotemporal EC assessment model was developed to address the EC in the temporal dimension to identify EC-intensive lifecycle stages and in the spatial dimension to examine the EC in line with the building elements at material, component, module, floor, and building levels. A 17-story steel modular building was selected for case study, and its EC results were compared with those of a baseline case with the same layout design but adopting conventional cast-in-situ construction. The cradle-to-end-of-construction EC of the case building was quantified as 609 kgCO2e/m2, of which the cradle-to-site EC contributed over 90 %. Structural steel in modules were the primary EC sources at all five spatial levels. The case building slightly increased (by 6 %) the cradle-to-end-of-construction EC, but achieved 38.9 % transportation and 56.8 % construction EC reductions, respectively, which indicates that cities will benefit from using MC by shifting the carbon emission burden from local construction sites to module manufacturing locations. The developed multi-level EC model provides a novel method for future research on EC of steel modular buildings, and the findings will shape future practices of decarbonizing modular buildings in a systematic manner.

This study examined the potential of End-of-Life circular economy systems to reduce embodied carbon in steel modular buildings. A multicycle life cycle assessment framework integrating material flow analysis and time-specific impact factors was developed. A case study involving a typical steel module in Hong Kong showed that landfilling generated 154.9 t CO2 eq. over the period 2020-2070, whereas recycle-priority and reuse-priority scenarios achieved 123.1 t CO2 eq. and 47.2 t CO2 eq., respectively. Reusing the steel module up to ten times reduced embodied carbon by 4.1 t CO2 eq./m2. Component-level analysis revealed that architectural components could significantly contribute to embodied carbon reductions. However, decarbonisation of the upstream industry (steel, aluminium, and electricity) reduced the benefits of multiple reuses by 24.2 %. This paper provides a comprehensive and flexible framework for multiple lifecycle assessment and offers valuable insights into how steel modular construction can enhance decarbonisation through End-of-Life circular economy systems.

Prefabricated construction documents are characterized by domain-specific terminology, multilingual content, and rich multimodal elements including texts, tables, and images. General-purpose embedding models exhibit degraded performance in this vertical domain, and existing retrieval approaches often lack explicit cross-granularity alignment between page- and element-level objectives, failing to exploit their complementary strengths. To address these limitations in industrial informatics, this article proposes a multigranularity fusion (MGF) framework for representation learning. A multimodal and multilingual prefabricated construction knowledge base (PCKB) is first constructed from real-world prefabricated construction documents, featuring query-answer pairs explicitly linked to both page- and element-level sources. The MGF framework jointly optimizes three contrastive objectives: a page-level loss to preserve contextual coherence, an element-level loss to enhance fine-grained semantic fidelity, and an answer-guided auxiliary loss to align representations with downstream relevance. Through extensive experiments with multiple embedding models, consistent improvements in retrieval performance are observed after domain-specific training. Notably, the explicit fusion and alignment of page- and element-level objectives yield improvements, while the answer-guided loss enhances element-based retrieval and generally boosts overall performance. Finally, a retrieval module is implemented in a practical system that supports interactive switching between page- and element-level results for both PCKB and user-uploaded documents.

Modular construction has been promoted to address insufficient onsite working space issues, improve quality control, and enhance construction efficiency. However, its broader application in high-rise buildings remains constrained by insufficient understanding of the structural behavior of critical inter-module connections and challenges encountered during actual construction. This study presents the design and experimental evaluation of 20 novel noncontact lap splices using grout-filled ducts (NLSGD) for vertical connections between the module walls of high-rise modular buildings. The test results revealed that the lap length and duct length were the primary parameters governing the ultimate pullout capacity. Parametric variations in duct diameter, side cover thickness, clear spacing, and grouting material led to changes of up to 12 % in pullout strength. Moreover, the duct lengths into the module walls were recommended to be at least 20 times the diameter of the inserting rebar. The specific recommendations on other parameters were also provided in contrast with codes in different regions including Hong Kong, Europe and the US. The results demonstrate that the proposed NLSGD connection is reliable in terms of both structural performance and constructability. The research provides valuable insights into vertical connection design for high-rise modular buildings and offers practical guidance for the implementation of the presented novel connection.

Facing a pressing housing shortage globally in general, modular integrated construction (MiC) presents a pivotal, scalable solution by rapidly delivering standardized public housing units. To sustain a robust supply of public housing, it is imperative to understand the supply and demand dynamics of MiC to proactively respond to market changes. This study aimed to provide a systematic exploration of the influencing factors of the MiC supply and demand for public housing developments and determine the key ones that are most influential in shaping the MiC market development. A mixed methods research design was implemented, combining a qualitative case study and a quantitative questionnaire survey. Hong Kong was selected as the research context for its leading position in adopting MiC. Drawing on the socio-technical systems theory, the initial influencing factors were systematically identified from multi-level and multi-dimensional perspectives. A total of 14 demand-related and 9 supply-related key influencing factors (KIFs) were highlighted out of the initial ones. The KIFs of “policy on MiC promotion” and “policy on public housing supply” were found to mostly influence the demand for MiC, and the KIFs of “experience” and “MiC market demand” were revealed to determine the development pattern of the supply of MiC. The study provided critical insights for policymakers seeking to align MiC supply and demand for public housing developments. The findings should lay a solid foundation for a MiC supply-and-demand prediction model and facilitate the optimization and upgrading of MiC supply chain configurations and the ancillary industries, ultimately improving public housing supply.

The significance of reusable connections has surged recently due to increasing waste generation and the extraction of natural resources within the construction industry. The adoption of interlocking systems has become prominent in contemporary structural connections due to their ease of assembly. This paper examines the newly developed hold-down type reusable column base connection, which is based on a plate interlocking mechanism. Specifically, it investigates the effects of base plate thickness and connector thickness under monotonic lateral loading. The reusability of the newly developed connection is demonstrated through repeated loading tests, indicating that the column base connection can be dismantled and reused. The loading regimen includes two initial cycles up to the yield strength, followed by a final loading cycle leading to ultimate failure, thereby establishing the connection's reusability. The degree of semi-rigidity is assessed by comparing the connection's rotational stiffness to that of pinned and fixed connections. The experimental results in this study provided a technical base to understand the structural behaviour of the connection and, in turn, develop a theoretical approach to predict the initial elastic stiffness and yield point of the connection, the key parameters that ascertain its eligibility for reuse. The theoretically predicted values align closely with the experimental results obtained. Additionally, a design example is provided to enhance the readers’ understanding and demonstrate the practical applicability of the developed component model for the proposed interlocking-type column base connection.

Many high-rise concrete modular buildings have their precast concrete modules designed as the lost formwork for in-situ concrete casting, in which the module walls do not act as structural components to provide lateral force resistance. This type of module wall design leads to significant concrete casting works on site and thicker walls than in conventional cast-in-situ buildings, which reduces the benefits of modular construction. Using composite module walls to resist the external force will address the issues, but ensuring the integrity of the composite module walls is challenging. This study aims to develop a new composite module wall design. In this design, the two outside layers of the composite module wall by precast concrete are horizontally connected by the stirrup cages and post-inserted bars in the two boundary element zones and bar-hook connectors in the web of the wall. The composite module walls are vertically connected through non-contact laps of rebars. The finite element (FE) method was employed to investigate the lateral behavior of the composite module wall. A reliable FE model was developed and validated against experimental data of precast shear wall with non-contact laps. Utilizing the validated model, a comparative study against a conventional cast-in-situ wall and an extensive parametric study were systematically conducted. The results show that the stiffness and capacity of the composite module wall are comparable to a baseline cast-in-situ wall design. The ultimate drift of the composite module wall increased by 7% compared to that of the baseline due to the failure mode of reinforcement rupture. The parametric study indicates that the capacity of the composite module wall improved by 63%, but the ductility decreased by 22% when the axial loading ratio increased from 0.1 to 0.4. A lap length greater than 25 times the bar diameter and rough interface between the inner layer and outer layers are recommended. The proposed composite module wall design is found to overcome the constructability challenge of the inter-module connection. The revealed overall behavior and parameter influential rules of the proposed composite module wall offer important guidelines for the design and application of concrete modular construction.

The adoption of modular construction (MC) is reshaping the construction supply chain and complicating associated embodied carbon (EC) flows. While previous studies have compared EC between MC and conventional methods, they remain limited to project-level, case-by-case analyses. A critical gap exists in understanding how MC redistributes EC at the city scale and how emission responsibilities should be allocated among interconnected cities. To this end, this study quantifies the city-scale EC flows and emission responsibilities among cities across the MC supply chain. A hybrid agent-based modelling and geographic information system model was developed and validated using data from 40 representative MC projects in Hong Kong. Results reveal that while burgeoning adoption of MC in Hong Kong achieves an overall reduction of 13.58 million kgCO2e compared to conventional methods, it concurrently shifts a significant emission burden. Additionally, 8.61 million kgCO2e emissions that should have occurred at local construction sites in Hong Kong are transferred to other cities with module factories, particularly in the Greater Bay Area of China, including Jiangmen, Foshan, Huizhou, and Zhaoqing. Furthermore, Hong Kong's allocated emission responsibility shifts drastically from 100% (consumption-based) to merely 2.99% and 0.80% (territorial- and production-based), demonstrating that over 97% of its EC is imported. Overall, this study pioneers the assessment of city-scale EC flows and emission responsibility in MC supply chain, contributing to scholarly understanding of how MC adoption changes regional EC distribution and emission responsibilities. Practically, these findings provide policymakers with recommendations to revisit responsibility allocation principles and refine regional decarbonization policies.

Three-dimensional (3D) volumetric modules are critical elements to be monitored and managed at modular integrated construction (MiC) sites. Intelligent module localization with excellent performance should be beneficial but has been underexplored. This study developed a box-aware module localization (Box-Loc) model by integrating point cloud and deep learning to get module locations from 3D bounding boxes. Synthetic datasets were established combining real-life and virtual prototyping-enabled pseudo-lidar point clouds. The model adapted 3D object detection algorithms to the construction context and employed transfer learning to mitigate overfitting. The real-life case study demonstrates that the Box-Loc model outperformed previous methods, reducing the average error from tens of centimeters to a minimum of 8.46 cm in the Y coordinate and decreasing the inference time from minutes to 55.7 ms. The real-time and centimeter-level module localization should facilitate safer and more efficient site monitoring and management, thus motivating a wider adoption of MiC globally.

The significant resource consumption and environmental impacts resulting from the building construction industry highlight the urgent need for circular construction strategies, particularly through the reuse of structural components. This study presents an enhanced demountable interlocking web connection designed to improve the deformation capacity and ductility of floor diaphragm systems, thereby facilitating structural deconstruction and reuse. The proposed connection incorporates the geometric optimisation and the advanced material by using notably low-yield-point steel, to shift the failure mode from shear-dominated to bending-dominated behaviour. Monotonic shear tests on 12 specimens, complemented by finite element analysis, investigate the effects of cut spacing, steel grade, and the connector geometry on the structural performance. Results show that low-yieldpoint steel significantly enhances deformation capacity, with peak load displacements reaching 6.25 mm-approximately double those of conventional steels. The redesigned ductile connector demonstrates nearly triple the deformation capacity of prior H-shaped designs, accompanied by a gradual post-peak load degradation indicative of improved ductility. A fixed-ended beam model is proposed for predicting shear capacity, showing close agreement with experimental and numerical results. The findings offer practical design recommendations for demountable connections that support circular construction objectives by enabling safe disassembly and reuse of structural elements.

Steel modular construction technology has been increasingly adopted in high-rise buildings, where a concrete core has been typically employed to ensure the lateral stability of the overall structure. Within this type of steel modular system, the connections between steel modules and concrete core walls are critical for effective load transfer, especially under severe horizontal loads. However, up to now, only a few experimental tests and systematic numerical investigations have been conducted on such module-to-core wall (M2C) connections, limiting their application in real-life engineering cases. Therefore, this study aimed to develop an innovative M2C connection for high-rise steel modular buildings and systematically examine its tensile behaviour through experimental tests and numerical simulations. The innovative M2C connection was developed based on the principles established from examining existing connection forms. The experimental tests were performed on ten specimens, and numerical simulations were conducted on 253 models with due consideration on various critical influencing factors. The analysis results demonstrate that the failure mode of the developed connection is the fracture of welds in the T-shaped plate or the failure of the bolts, and that the steel grade, thickness of the T-shaped plate flange, and thickness of the constraint plates are the most critical influencing factors. The findings provide valuable insights into the tensile behaviours of the newly developed M2C connection, and practical guidelines for the design and application of this M2C connection in future high-rise steel modular buildings.

Anticipating energy conservation possibilities in residential, commercial, and industrial structures through machine learning is a novel and highly efficient method. ML algorithms can examine intricate data patterns to detect inefficiencies, project savings, and enhance strategies for energy utilization. This study presents new machine learning models to predict the energy-saving potential based on data from the electric consumption of commercial, residential, and industrial buildings in southern California. Various prediction models and optimization algorithms are employed, including AdaBoost Regression, Stochastic Gradient Boosting, Lasso Regression, Flower fertilization optimization algorithm, and Starfish optimization algorithm. Also, Dempster-Shafer theory is utilized to build ensemble models by integrating single models with two optimization algorithms. A multicollinearity analysis uses the variance inflation factor to detect the correlation between features. Furthermore, the SOBOL technique and the Cosine Amplitude Method are implemented for sensitivity analysis. The results demonstrate that hybrid modeling approaches, combining predictive algorithms with optimization techniques and ensemble strategies, achieve high accuracy and low error in forecasting energy savings. These models can forecast where and when savings are possible, enabling targeted interventions and integrating predictions into energy management systems for real-time monitoring and automated optimization of energy consumption.

This study aimed to develop an innovative module-to-module (M2M) connection for high-rise steel modular buildings and examine its structural behaviour under shear loads. The M2M connection was developed based on the examination of the special characteristics of connections in high-rise steel modular buildings from the perspectives of construction efficiency and structural safety. Experimental tests with eight specimens were performed to investigate the shear behaviour of the developed connection. Numerical models were then established and validated base on the test results to accurately simulate the shear behaviour of the developed connection. Subsequent parametric numerical simulations with 153 models were performed to examine the effects of critical influencing factors on the shear behaviour of the connection. The analysis results indicated that the middle sleeve in the developed M2M connection can provide effective protection for the vertical connectors from horizontal loads. Furthermore, the identified critical influencing factors, such as the steel grade of different components and axial compression ratio, can facilitate the protection of the vertical connector across all loading stages, thereby ensuring the disassembly and reassembly ability of the developed connection. The axial compression ratio had the most significant effects on the shear behaviour of the connection, as it not only increased the yield and ultimate load of the connection by over 100%, but also exhibited significant interactive effects with the thickness of the connection plate and the steel grade of the middle sleeve and the connection plate.

This work systematically investigates the mechanical behavior of a novel fiber-reinforced ultrahigh-performance lightweight concrete (UHPLWC) incorporating expanded glass aggregate (EGA). The research addresses the need for high-strength and ductile materials with reduced self-weight in modern construction. An extensive experimental program was conducted on twelve mixes, covering two distinct strength grades (L80 and L100) with densities ranging from 1809 to 2130 kg/m3, and reinforced by polypropylene fibers, steel fibers, or their hybrid combinations. The complete stress-strain response, elastic modulus, failure modes, and flexural toughness of the UHPLWC materials were investigated, in which the critical role of fibers in transforming the behavior of the material from brittle to ductile was quantified. The experimental results show that the hybrid fiber systems induced significant strain-hardening in flexure, leading to an increase in bending strength of up to 157% and a more than 15-fold enhancement in energy absorption capacity of UHPLWC. A defining trade-off was identified, where the use of EGA resulted in a lower elastic modulus (24.4-31.6 GPa) compared to conventional UHPC. Finally, the study proposes and validates a constitutive material model for UHPLWC that accurately predicts its complete compressive stress-strain relationship, capturing the distinct post-peak softening of both brittle and ductile mixes.

To reduce embodied carbon (EC) emissions from the construction industry, promoting modular buildings, especially for highly populated regions, is of crucial importance. Nonetheless, the heavyweights of concrete modules have been reported to lead to a series of logistical problems in manufacturing, transportation, and hoisting. On the other hand, the double-panel issues in modular buildings posed concerns related to inefficient material usage, thus leading to potential EC increments. This paper aims to adopt the practical and idealized parametric lightweight designs of modules to estimate the EC reductions throughout the cradle-to-end-of-construction stages. Two types of lightweight module designs, i.e., dematerialized module and lightweight concrete module, were examined in high-rise concrete modular buildings in Hong Kong, considering the structural system features. The results indicate that the utilization of lightweight concrete in non-structural components of modular buildings resulted in EC reductions of 10.1% for a typical floor. Dematerialization by 30% on modules exerts the potential to reduce 11.6% EC for modular buildings compared to regular designs. The findings presented in this paper could provide valuable reference to assess the carbon-saving potentials of adopting lightweight modules in Hong Kong and worldwide.