Integrated Digital Twin and BIM Approach to Minimize Environmental Loads for In-Situ Production and Yard-Stock Management of Precast Concrete Components

Abstract

Digital twin (DT) technology, integrated with building information modeling (BIM), enables real-time feedback and predictive analytics in construction. This study presents a BIM-enabled DT framework to optimize in situ production and yard-stock management of precast concrete (PC) components with a focus on minimizing CO₂ emissions. Using Oracle Crystal Ball, scenario-based simulations revealed up to an 8.9% reduction in environmental impact. Distinct from prior research that largely emphasized cost or off-site strategies, this study uniquely addresses on-site sustainability by embedding carbon metrics into the decision-making process. The framework was validated through a large-scale logistics warehouse project that showcased its practical utility. This research contributes a replicable method for enhancing sustainability in precast construction through digital technologies.

1. Introduction

The construction industry is undergoing a paradigm shift, driven by the emergence of digital technologies that enable the management of complexity, uncertainty, and sustainability [1,2]. As projects grow more intricate with the addition of diverse stakeholders, sustainability imperatives, and unpredictable site conditions, digital transformation has become essential. Among these advancements, digital twin (DT) technology stands out as a pivotal tool, enabling the real-time synchronization of physical assets with virtual models to support simulation-driven and predictive decision-making [3,4].

The integration of DT with building information modeling (BIM) has garnered increasing attention. BIM provides a foundational platform for data-driven project coordination, design management, and stakeholder collaboration [5,6,7,8,9]. While BIM offers a comprehensive static representation of a facility, DT technology augments this by incorporating real-time data streams to enable dynamic responses, such as anomaly detection, maintenance forecasting, and risk evaluation, within a closed-loop feedback system [10,11].

Despite these technological advances, significant research gaps remain, particularly in applying BIM–DT integration to optimize the production and yard-stock of precast concrete (PC) components. Prior research has often focused on modular production systems or factory-based simulation models, which tend to lack the flexibility required for resource-constrained, site-specific scenarios [12,13].

Moreover, the existing studies have largely overlooked real-time planning for in situ production and yard layout of PC components [14,15,16]. Efficient management of these components demands precise coordination of production, transportation, and temporary storage, all of which are susceptible to spatial limitations, scheduling conflicts, and environmental considerations [17]. A comprehensive and real-time platform is urgently needed to monitor, predict, and optimize these tasks.

In particular, the studies on in situ PC production have primarily focused on production area and layout planning [18,19,20,21,22]. However, yard-stock areas are reported to require more than five times the space of production areas [23,24,25,26], and inefficient stockyard layouts can lead to material congestion, schedule delays, and increased environmental burden [27,28,29]. Nonetheless, research on operational strategies and layout management for yard-stock areas remains limited [30,31]. In other words, the previous studies do not incorporate a simulation-based decision-making framework utilizing digital tools such as BIM, DT, or Monte Carlo methods. Moreover, there is a lack of empirical case studies involving CO₂ emission analysis based on actual field data or the application of optimized yard-stock space management.

In situ production replicates the processes of factory manufacturing, rebar placement, formwork, concrete pouring, curing, and temporary storage on the construction site [32,33,34,35]. This method has been shown to reduce environmental loads by over 14.58% and construction costs by up to 39.4% compared to factory production, without sacrificing quality [20,21,36,37]. These benefits highlight the need for intelligent, real-time scheduling and layout planning using BIM-integrated DT systems and simulation tools. Studies also show that in situ production can yield quality levels equal to or exceeding those of in-plant production, which makes it competitive in terms of convenience, quality, cost, and schedule [21,36].

Lim and Kim (2024) [21] and Lim and Kim (2020) [36] are foundational studies focusing on in situ production optimization that offer a reference point for time and CO₂ comparisons in their work. Gao (2024) [37] provides practical cases of BIM–DT integration and digital tracking applicable to yard-stock management component. Su (2025) [38] contributes an advanced multi-objective optimization framework that is methodologically aligned with the use of Monte Carlo-based simulations. Pradhan (2025) [39] highlights lifecycle and method selection criteria that enrich comparative perspectives. Cotoarbă et al. (2025) [40] offer innovative approaches to uncertainty management and DT-based collaboration that may broaden existing research scopes.

Traditional methods for managing PC production and yard layouts often result in inefficient material flows and space usage, leading to increased environmental burdens, particularly CO₂ emissions. While prior research has utilized building information modeling (BIM) and digital twin (DT) technologies, most have been limited to design coordination and visualization, lacking execution-level frameworks for real-time environmental optimization on construction sites. This study proposes a BIM–DT integrated framework that aims to reduce environmental loads, especially CO₂ emissions, by optimizing yard layouts and minimizing transportation distances between production and installation zones. Grounded in real-world construction data, the model not only demonstrates significant quantitative improvements in sustainability metrics but also proves its applicability in practical construction scenarios.

To gain advantages in in situ production, it is essential to optimize real-time production and layout planning based on site conditions, which requires the integration of BIM-based DT technology and probabilistic simulation tools. Accordingly, this study proposes a BIM-based DT framework designed to optimize the in situ production and yard-stock layout of PC components. The objectives are to minimize carbon emissions from construction through predictive modeling and simulation.

This paper emphasizes the importance of an integrated platform that simultaneously addresses layout efficiency and carbon reduction strategies as a key enabler for sustainable construction. The scope of this study is confined to steel-reinforced concrete (SRC) precast components. Accordingly, this study demonstrates a comprehensive distinction from previous studies in the following key aspects:

• Development of a BIM–DT integrated model for environmental load reduction;
• Shift in focus from cost/schedule efficiency (common in prior studies) to environmental performance enhancement;
• Validation using data from a large-scale, real-world project.

2. Literature Review

2.1. In Situ Production and Yard-Stock PC Components

In construction sites with spatial constraints, a designated area is required to facilitate the in situ production and yard-stock of PC components without hindering ongoing construction activities. The spatial requirements are determined by the size and quantity of PC members, and ensuring consistent quality demands dedicated site managers and skilled labor. Moreover, considering the advantages of in situ production, such as reductions in CO₂ emissions and construction costs, this approach may be preferentially selected by clients who seek direct control over the production process, as opposed to relying on factory-based manufacturing. The yard layout process follows the sequence of production, yard-stock, and installation [20,21,41]. As illustrated in Figure 1a, components are first produced based on zone-specific requirements identified from design drawings. Since component shapes and sizes vary, multiple mold types are required. Zones are classified by component type (e.g., beams, columns, slabs), and production is planned to meet the earliest possible installation deadline within each zone. As the production, yard-stock, and installation quantities are the same, the components to be installed are identified. The quantity is calculated to determine the target quantity for in situ production. The location of the column and beam components for each zone of in situ production is analyzed through design drawing analysis.

The number of members and area for the yard-stock in Figure 1b are calculated. And the required yard-stock area is calculated by multiplying the number of members. The available area for production and yard-stock is calculated for each process, and the yard-stock and production arrangement plans are simulated based on the reviewed area. If the yard-stock and production are not satisfied, the arrangement simulation is completed through repeated modifications (feedback routine). For yard-stock simulation, first, the crane movement path is analyzed and the possibility of utilizing the yard space is reviewed. At this time, the yard space is determined by dividing it into before and during the installation of PC members. The yard-stock utilization order is determined and the possibility of utilizing the yard-stock space according to the in situ production schedule is reviewed.

As illustrated in Figure 1c, the number of cranes required for the installation of PC components is determined based on the in situ production schedule. Crane specifications are selected by considering the radius of rotation and lifting capacity [41,42]. Based on the number of cranes, a zoning plan is established in which crane operating zones are defined to prevent spatial interference during simultaneous operations. Subsequently, the installation sequence of PC components, such as columns, beams, and slabs, is determined by analyzing crane movement paths within each zone. The installation duration per component is then calculated based on type-specific productivity metrics. A simulation of the installation schedule is carried out in weekly or daily units in accordance with the overall site construction plan. This simulation supports decision-making for optimal resource allocation and construction sequencing. In practice, PC components produced in situ are first placed in the stockyard and then installed sequentially according to the predefined order and zoning constraints.

2.2. Previous Studies

2.2.1. Integration of BIM and DT in the Construction Industry

BIM has become a foundational platform for construction information management, supporting tasks such as design coordination, scheduling, and resource allocation [10,31]. However, BIM traditionally functions as a static system, lacking the ability to reflect real-time conditions on site. To address this limitation, DT technology is increasingly being adopted to create dynamic, real-time representations of physical environments [3,4].

The integration of BIM and DT technology has significantly improved decision-making across the project lifecycle by enhancing visualization, operational control, and performance monitoring. For instance, Opoku et al. (2021) utilized DT technology to improve site visibility and adaptive decision-making [43], while Kassem et al. (2015) applied IoT-based DT frameworks for real-time equipment tracking [44]. Other studies, such as those by Xu et al. (2024), have reviewed the technical challenges and strategic benefits of integrating digital technologies in construction [45].

Importantly, recent research has expanded DT applications to include environmental impact assessment through life cycle assessment (LCA) [46]. Chen et al. (2021) [47] and Tagliabue et al. (2023) [48] demonstrate how DT technology can be linked to LCA metrics to promote low-carbon construction and compliance with environmental regulations.

For LCA, Chen et al. (2021) [47] adopt a process-based approach to quantify the embodied carbon of buildings across their entire lifecycle, encompassing cradle-to-cradle boundaries and including the product, construction, use, end-of-life, and beyond stages (i.e., reuse and recycling). In the product stage, emissions are calculated based on raw material inputs, machinery operations, and energy consumption, all of which are measured in appropriate units such as mass, volume, or area. The construction stage accounts for emissions from transportation, on-site waste, and site activities. The use stage incorporates emissions from surface-level CO₂ exchanges, routine maintenance, repairs, component replacement, and refurbishments throughout a standard reference study period of 60 years. End-of-life emissions include those arising from demolition, off-site transport, waste processing, and final disposal. The beyond stage quantifies the net benefits or burdens resulting from the substitution of primary products with recovered or recycled materials. Each lifecycle stage is analytically subdivided into specific sub-processes, and corresponding emission factors are systematically applied. By explicitly presenting detailed equations that were omitted in the previous literature, this framework provides a comprehensive and accurate methodology for life-cycle-based embodied carbon estimation.

Lu et al. (2020) further illustrated the value of integrating sensor data into BIM–DT systems to enable anomaly detection in facility monitoring [11]. Opoku et al. (2021) emphasized the use of real-time data from IoT and machine learning models to enhance the predictive capabilities of DT technology [43]. According to Xu et al. (2024), the transition from BIM to DT represents a fundamental evolution in construction workflows, one that enables system-wide intelligence and predictive control [45].

2.2.2. DT for Predictive Planning and Risk Management

DT technology’s core strength lies in its ability to simulate future scenarios and support proactive decision-making in complex environments. For example, Lu et al. (2020) developed a DT-based approach to improve coordination and productivity in on-site construction processes [11]. Zhang et al. (2024) applied DT technology for real-time environmental monitoring and integrated lifecycle carbon footprint analysis into planning workflows [49].

Case studies have shown that time and cost management using 4D and 5D BIM models has enabled stakeholders to prepare simulations and predict scenarios to achieve resource optimization [50]. It can also play a critical role in establishing predictive maintenance and energy management strategies [51]. In the port area, this process has led to strategic asset management for risk and safety management activities [52,53].

More recently, scenario-based and probabilistic simulation tools have been embedded within DT frameworks. Bakhshi et al. (2024) introduced a DT platform that incorporates simulation techniques for proactive risk mitigation and emission tracking [54]. The integration of optimization algorithms with DT technology has proven especially useful in improving the responsiveness and reliability of construction project control systems.

2.2.3. In Situ PC Production and Stockyard Management

Effective management of in situ production and temporary storage is critical in precast construction, where just-in-time assembly and spatial optimization are essential. Kosse et al. (2023) proposed a simulation-based model for optimizing yard layouts in precast construction [55], while Lim and Kim (2024) explored the integration of scheduling and spatial constraints in PC logistics planning [21].

However, many existing models remain static and are not responsive to real-time conditions. To overcome this limitation, it is essential to incorporate sensor inputs and intelligent simulation platforms within DT systems. Environmental optimization is an area that has not yet been adequately addressed in the literature.

2.2.4. Reducing Carbon Emissions Using DT

Recent advancements in DT technologies have demonstrated substantial potential in reducing carbon emissions across the entire lifecycle of built environments. Several studies have shown that integrating DT technology with real-time sensing and AI-driven analytics enables dynamic monitoring and optimization of energy usage and CO₂ emissions in both construction and operation phases [56,57]. For instance, cognitive digital twins applied to HVAC (heating, ventilation, and air conditioning) systems in smart buildings have achieved up to 30–50% carbon reduction by adapting to occupant behavior and environmental conditions [58]. Moreover, city-scale digital twin platforms, such as those implemented in Sydney and the European “Destination Earth” project, allow for urban-level simulation of emission sources and inform sustainable policy planning [59]. These findings underscore the importance of embedding DT-enabled carbon intelligence within construction workflows, from design to operation, as a pathway to achieving low-carbon and climate-resilient infrastructure.

2.2.5. Holistic Lifecycle Carbon Assessment Frameworks Using BIM

ISO 14040 [60], ISO 14044 [61] and EN 15978 [62] provide a foundational basis for life cycle assessment (LCA), encompassing goal and scope definition, inventory analysis, impact assessment, and interpretation phases [63]. These frameworks promote consistency and transparency in quantifying environmental impacts over a building’s entire lifecycle. Berges-Álvarez et al. (2024) [64] have explored the integration of LCA with building information modeling (BIM), enabling extraction of design geometry and material specifications for automated environmental calculations. Mazur and Olenchuk (2023) [65] applied BIM-driven LCA to quantify CO₂ footprint reductions across housing typologies. While BIM-based LCA workflows have demonstrated value, they are often static in nature, focusing on early design phases without capturing real-time updates or site-level data during the construction phase. This presents a key limitation in responding dynamically to carbon-intensive activities on site.

2.2.6. Precedents in DT–BIM for Environmental Performance Optimization

Badenko (2024) [66] demonstrated how coupling BIM and DT allowed for predictive energy optimization and real-time operation feedback, achieving up to 15% energy savings in building operation. Kosse (2022) [67] applied Industry 4.0-based digital twins to automate precast component production and integrate real-time data into logistics optimization. Widjaja (2025) [68] expanded this to include BIM-based reinforcement layout optimization via DT feedback loops, achieving waste reduction and enhanced coordination. Although these studies demonstrate the potential of DT–BIM in performance monitoring, they have not fully addressed environmental impact minimization, especially in terms of in situ CO₂ reduction and dynamic yard-stock scenarios. Most research focuses on energy or operational parameters without integrating material-specific carbon tracking in precast workflows.

2.2.7. PC Sustainability Strategies: In Situ vs. Off-Site Trade-Offs

Comparative analyses reveal that PC typically yields around 10% lower CO₂ emissions per cubic meter compared to cast-in-place equivalents [69]. Another study revealed that, although prefabricated systems generate the highest carbon emissions during the installation phase, they can still achieve a reduction of approximately 86 kg-CO₂ per square meter [70]. In situ production methods, depending on project-specific conditions, have been shown to realize more than a 14.3% reduction in CO₂ emissions compared to off-site (factory-based) production [36]. Producing directly in situ can further reduce carbon emissions associated with logistics by minimizing transportation distances, improving operational flexibility, and optimizing construction workflows [71]. In contrast, factory-based production excels in terms of process speed, quality control, and safety. Some case studies have demonstrated that off-site methods can achieve reductions of over 50% in waste and emissions, reduce construction time by up to one-third, and maintain defect rates below 5% [72]. Generally, off-site precast production is advantageous in terms of resource efficiency and schedule acceleration, while in situ production, when coupled with proper planning and appropriate equipment, can outperform it in terms of carbon emissions. Particularly, factors such as transportation distances, equipment efficiency, site layout, and material reuse significantly influence the sustainability outcome, which underscores the need for project-specific optimization strategies.

2.2.8. Research Gap

Despite growing interest in DT applications, there remains a lack of fully integrated BIM–DT systems specifically tailored to the complexities of PC logistics. In particular, few frameworks support simulation-based optimization of CO₂ emissions in tandem with production and yard management. This study aims to address this gap by proposing a unified platform that enables environmentally conscious decision-making for in situ PC component production.

3. Methodology

In general, DT technology is considered a virtual mirror that describes the physical characteristics of a system and transmits and receives information for control, monitoring, and decision-making processes [73]. DT technology should be connected and synchronized while running simulations of physical counterparts over time. DT technology is defined as a model that pairs physical and digital models by automatically exchanging data between them [74].

However, in this study, we apply DT technology that is different from general models by using an optimization tool. The conceptual diagram in Figure 2 shows a framework for efficiently managing the in-situ production and installation of PC structures based on DT technology, while achieving sustainability (carbon reduction) goals at the same time. The DT model consists of a bidirectional link between the actual physical site and the BIM-based DT that virtually implements it. The BIM model is updated based on real-time on-site data using Dynamo, and simulation and analysis are made possible through the BIM model.

The digital twin system is linked to optimization tools such as Oracle Crystal Ball to achieve the goals of minimizing carbon dioxide (CO₂) emissions. The optimization tool uses information derived from the BIM model and actual field data to perform sensitivity analysis and control variable settings, thereby deriving the optimal scenario. The results are fed back into the digital twin model and reflected in decision-making in the field, which continuously improves the accuracy of the model.

3.1. Project Analysis

To validate the effectiveness and practical applicability of the proposed BIM-based DT framework, a real construction project was selected as a case study. The project involves the construction of a large-scale logistics warehouse located in A City. The structure consists of four above-ground floors, with PC slabs installed from level 1 to level 4. The building’s core structure is composed of reinforced concrete (RC), and the ramp areas utilize a combination of steel and SRC systems. This study focuses primarily on the PC slab construction processes across the four floors. The positional and loading data of column and beam elements, excluding the slab, are collected.

3.2. Data Analysis and Preprocessing

In the data analysis and preprocessing phase, the detailed procedures of in situ production, storage, and installation applied on-site are analyzed, as illustrated in Figure 3. The construction process of PC buildings consists of PC component production, transportation, and installation. The production process begins with formwork preparation, including mold cleaning, mold assembly, embedded hardware installation, and rebar placement followed by inspection. Subsequently, concrete is cast after the application of release agent, which is followed by steam curing and demolding, which completes the production of the components. Once produced, components undergo lifting, cleaning, and finishing operations before being placed in the yard for storage. The installation process begins with preparation for erection, including connection with lifting equipment, upending, hoisting to the designated installation position, and temporary fixation. This is followed by verticality inspection and final alignment using a plumb-down operation. The installation planning for PC components is conducted during the preparation stage, based on the working radius and lifting capacity of the crane. The crane also influences mold allocation and schedule-related factors. Idle cranes not utilized for installation are employed for in situ production processes. Therefore, the production and installation workflows were analyzed based on actual field-applied data that reflect practical operational constraints and resource allocation.

3.3. Basic Model Creation and Data Acquisition from BIM–DT

In the basic model creation and data acquisition from BIM–DT phase, the DT model is implemented and linked with actual site information and BIM models. The model is used for simulation-based optimization. This study aims to optimize the in situ production and yard-stock management of PC components through the integration of BIM and DT technologies. While BIM captures the physical attributes and design information of each component, the DT functions as a system for simulation-based operational decision-making and process optimization. The two technologies serve complementary roles, forming a real-time or near-real-time feedback loop that enhances the sustainability and responsiveness of on-site operations. Figure 4 shows a BIM model implemented using Dynamo. The software used in this study was Autodesk Revit 2022. A Dynamo script was created to obtain real-time information, and the BIM model was structured based on the IFC schema to ensure data consistency and interoperability. The IFC class and property set assigned to the selected member in the BIM model can be retrieved.

Initially, information such as the location, geometry, quantity, type, and installation sequence of PC components is extracted from the BIM model and transformed into a data structure suitable for DT-based simulation. This data processing step is automated using Dynamo, and the refined dataset is subsequently exported in Excel format for integration with the external simulation tool, Oracle Crystal Ball. Figure 5 shows the 4D simulation of the site after the installation of PC members has started, which was built using BIM. PC members are not installed during the lead time from D + 1 to D + 86. Therefore, only the construction from D + 87 to D + 134 is shown. In D + 98, the production areas of zone A and B are determined to be 630 m² and 580 m² in real time using Dynamo, respectively, and the yard-stock areas of zone A and B are determined to be 7124 m² and 6812 m² in real time, respectively. In the case of a general in situ production system, since the erection area is added, it is difficult to produce and store PC members in the same location, so the production and yard-stock locations have to be moved. However, in this study, only the production location was moved, and the yard-stock location was set up to be the closest location to the installation location. The monthly site plans obtained using BIM are consistent with those presented in our parallel study [75].

3.4. CO₂ Emission Minimization Simulation

3.4.1. CO₂ Emission Quantification

To estimate CO₂ emissions, it is essential to first perform a detailed quantity takeoff. As shown in

Table 1. Quantity of columns and beams.

Work	Item	Unit	A Column	B Column	A Girder	B Girder	Total
Concrete work	Concrete	m3	6.444	6.745	20.957	21.238	30,457
Reinforcement work	HD10	t	0.302	0.333	0.171	0.156	729
	HD13	t	0.104	0.175	0.314	0.238	609
	HD16	t	0.050	0.112	-	-	441
	HD 19	t	-		-	-	254
	HD 22	t	1.212	1.301	0.971	1.117	2284
	HD 32	t	-	-	1.269	1.532	1110
	Steel joint	t	0.461	0.461	0.922	0.922	1797
	Sub-total	t	2.129	2.382	3.647	3.965	7070
Form work	Steel form	t	0.016	0.016	0.010	0.010	38

, the material quantities for key columns and beams were calculated. For instance, for Column A and Column B, the quantities were determined as follows: concrete volumes of 6.444 m³ and 6.745 m³, steel weights of 2.129 tons and 2.382 tons, and form steel weights of 0.016 tons, respectively. Similarly, for Beam A and Beam B, the calculated quantities were concrete volumes of 20.957 m³ and 21.238 m³, steel weights of 3.647 tons and 3.965 tons, and form steel weights of 0.010 tons. When aggregated, the total material quantities amounted to 30.457 m³ of concrete, 7.070 tons of steel, and 0.038 tons of form steel.

In the case project, steel molds were utilized for the production process. As illustrated in Figure 6, the steel molds applied in the in situ production are identical in specification and configuration to those used in conventional in-plant production. These molds are ordered with the same technical specifications from the manufacturers that supply the precast plant. Columns are produced in sets of two units and, similar to beam production, a spacing of 0.1 m is maintained between adjacent molds to ensure sufficient working space during on-site operations.

CO₂ emissions are calculated using actual labor inputs, along with oil and electricity consumption during the production and yard-stock processes. CO₂ emission calculation formulas are provided in Equations (1) and (2). The CO₂ emission factors for steel and concrete were observed to be 3500 kg-CO₂ per ton and 140 kg-CO₂ per cubic meter, respectively. The CO₂ emission factor for labor is 0.02 kg-CO₂ per person-hour. Equation (3) presents the estimation of CO₂ emissions associated with each construction activity based on oil consumption during the construction phase. This estimation builds upon the energy consumption calculated using Equation (4). Furthermore, CO₂ emissions derived from electricity usage at the construction stage are estimated using Equation (5), which is based on the energy consumption model described in Equation (6). Based on these calculations, the total CO₂ emissions are determined as follows:

$${\mathrm{C}\mathrm{E}}_{\mathrm{c}}=\sum _{\mathrm{i}=1}^{\mathrm{n}}\left[{\mathrm{Q}}_{\mathrm{c}\mathrm{i}}\times \left({\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{l}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{o}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}\right)\right]$$

(1)

$${\mathrm{C}\mathrm{E}}_{\mathrm{b}}=\sum _{\mathrm{i}=1}^{\mathrm{m}}\left[{\mathrm{Q}}_{\mathrm{b}\mathrm{i}}\times \left({\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{l}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{o}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}+{\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}\right)\right]$$

(2)

$${\mathrm{E}}_{\mathrm{C}\mathrm{O}}=0.0017\times {\mathrm{A}}_{\mathrm{f}}+37.5$$

(3)

$${\mathrm{Q}}_{{\mathrm{C}\mathrm{O}}_2\mathrm{o}}={\mathrm{E}}_{\mathrm{c}\mathrm{o}}\times 3.06$$

(4)

$${\mathrm{E}}_{\mathrm{c}\mathrm{e}}=0.0247\times {{\mathrm{A}}_{\mathrm{f}}}^{0.79}$$

(5)

$${\mathrm{Q}}_{{\mathrm{C}\mathrm{O}}_2\mathrm{e}}={\mathrm{E}}_{\mathrm{c}\mathrm{a}}\times 1.64$$

(6)

${\mathrm{C}\mathrm{E}}_{\mathrm{t}}$: total CO₂ emission, ${\mathrm{C}\mathrm{E}}_{\mathrm{c}\mathrm{i}}$: CO₂ emission of column, ${\mathrm{C}\mathrm{E}}_{\mathrm{b}\mathrm{j}}$: CO₂ emission of beam, ${\mathrm{Q}}_{\mathrm{c}\mathrm{i}}$: ith column quantity, ${\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{l}}$: unit CO₂ emission of labor, ${\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{o}}$: unit CO₂ emission of oil, ${\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}$: unit CO₂ emission of electricity, ${\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}$: unit CO₂ emission of lighting and heating, ${\mathrm{C}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}$: unit CO₂ emission of environmental conservation, ${\mathrm{Q}}_{\mathrm{b}\mathrm{i}}$: ith beam quantity, $\mathrm{i}$: number of installed ith column (1,…, n), $\mathrm{j}$: number of installed ith beam (1,…, m), ${\mathrm{E}}_{\mathrm{C}\mathrm{O}}$: energy (oil) consumption during the construction stage (TOE), ${\mathrm{A}}_{\mathrm{f}}$: total floor area (m²), ${\mathrm{Q}}_{{\mathrm{C}\mathrm{O}}_2\mathrm{O}}$: CO₂ emission based on oil use in the construction stage (T-CO₂), ${\mathrm{E}}_{\mathrm{c}\mathrm{e}}$: power consumption in the construction stage (TOE),${\mathrm{Q}}_{{\mathrm{C}\mathrm{O}}_2\mathrm{e}}$: CO₂ emission based on power consumption in the construction stage (T-CO₂).

In the actual case project, the average unit installation times for PC components were recorded as 19 min for columns, 8 min for girders, and 10 min for slabs. The total number of components installed included 850 columns, 1371 girders, and 4032 slabs. The overall installation period, based on calendar days, amounted to 48 days, during which the crane operated with a rotational coverage of 20 m per minute. Excluding the time required for fixing the component positions, the net installation time for PC components was also calculated to be 48 days, as shown in

Table 2. Actual resource for yard-stock and erection.

Item		Unit	Quantity
Labor	Equipment operator	Day	59
	common labor	Day	144
Material	Diesel	L	12,900
Equipment	Crane (550 ton)	Day	430

. For reference, in conventional RC structures, the standard installation times are, similarly, 19 min for columns, 8 min for girders, and 10 min for slabs; however, in the case of PC structures, the installation durations for columns and girders are reduced by approximately 50%. Based on crane performance data [76], the estimated total installation time for PC components in RC construction scenarios was 172 days. Additionally, the average unit yard-stock time for PC columns and beams was recorded as 13 min, with a total yard-stock handling period of approximately 11 days, as shown in

Table 3. CO₂ emission calculation for in situ production (unit: T-CO₂).

Classification	Production	Yard-Stock and Erection
Material use	37,940	-
Labor use	243	120
Oil use	987	23,785
Electricity use	543	146
Lighting, and heating use	287	65
Environmental conservation	202	37
Total	40,002	24,153

. CO₂ emission calculation formulas for yard-stock and erection are provided in Equations (7)–(10).

$${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{t}}={\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{c}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{b}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{s}}$$

(7)

$${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{c}}=\sum _{\mathrm{i}=1}^{\mathrm{n}}\left[{\mathrm{Q}}_{\mathrm{c}\mathrm{i}}\times \left({\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{o}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}\right)\right]$$

(8)

$${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{b}}=\sum _{\mathrm{i}=1}^{\mathrm{m}}\left[{\mathrm{Q}}_{\mathrm{b}\mathrm{i}}\times \left({\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{o}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}\right)\right]$$

(9)

$${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{s}}=\sum _{\mathrm{i}=1}^{\mathrm{l}}\left[{\mathrm{Q}}_{\mathrm{s}\mathrm{i}}\times \left({\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{o}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}+{\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}\right)\right]$$

(10)

${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{t}}$: total CO₂ emission for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{c}}$: CO₂ emission of column for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{b}}$: CO₂ emission of beam for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{s}}$: CO₂ emission of slab for yard-stock and erection, ${\mathrm{Q}}_{\mathrm{c}\mathrm{i}}$: ith column quantity, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}}$: unit CO₂ emission of labor for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{o}}$: unit CO₂ emission of oil for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{l}}$: unit CO₂ emission of electricity for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{l}\mathrm{h}}$: unit CO₂ emission of lighting and heating for yard-stock and erection, ${\mathrm{C}\mathrm{E}\mathrm{Y}\mathrm{E}}_{\mathrm{u}\mathrm{e}\mathrm{c}}$: unit CO₂ emission of environmental conservation for yard-stock and erection, ${\mathrm{Q}}_{\mathrm{b}\mathrm{i}}$: ith beam quantity, ${\mathrm{Q}}_{\mathrm{s}\mathrm{i}}$: ith slab quantity, $\mathrm{i}$: number of installed ith column (1,…, n), $\mathrm{j}$: number of installed ith beam (1,…, m), $\mathrm{k}$: number of installed ith slab (1,…, l).

3.4.2. Scenario Definition and CO₂ Emission Optimization Process

Table 4. Assumptions for 4D simulation.

No.	Assumptions
1	The cost and time satisfy the client’s requirements.
2	All PC components are produced, stored, and installed in situ.
3	The PC components production and yard-stock location is near the installation location.
4	As this site has a large floor area and not many floors, a mobile crane is used.
5	The production cycle for each PC component is consistently maintained at two days.
6	For material transportation routes and each individual transport line, both transport efficiency and unit cost can be applied.

shows the assumptions for 4D simulation using BIM. The total cost and time are not considered because they satisfy the client’s requirements. All PC components are produced, stored, and installed in situ. The PC member production and yard-stock location can always be located near the installation location. As this site has a large floor area and a low number of floors, it is difficult to use a tower crane. Hence, a mobile crane is used.

The entire area and each yard-stock yard are defined along the X- and Y-axes and simplified into rectangles with parallel boundaries. For material transportation routes and each individual transport line, both transport efficiency and unit cost can be applied. The assumptions related to transportation are described in detail in the study by Lim and Kim [21]. In this paper, only the aspects directly associated with CO₂ emissions are addressed.

Table 5. Constraints for 4D simulation.

No.	Constraints
1	During the estimation of yard-stock areas, unnecessary secondary transport must be minimized to reduce energy consumption from equipment operation.
2	The use of cranes and transport equipment must be limited to the minimum level required to prevent excessive energy waste.
3	Formworks must be reused at least 40–50 times as a principle.
4	The production and yard-stock processes of all columns and beams must be included, and the installation processes of all columns, beams, and slabs shall be incorporated.

shows the constraints for 4D simulation using BIM. To minimize environmental loads, unnecessary secondary transportation should be avoided and equipment utilization should be limited to prevent excessive energy consumption. Formworks are assumed to be reused at least 40–50 times to reduce environmental impacts from material production and disposal. Furthermore, the production and yard-stock of all columns and beams, as well as the installation of all columns, beams, and slabs, must be incorporated.

The number of required molds is calculated based on production quantity, production cycle, and total production duration. Similarly, the number of cranes is determined by the unit erection time, total installation quantity, and allowable erection period. The total yard area is estimated by multiplying the number of stored components by the unit of yard area per component type. These calculations are formalized in Equations (11)–(14).

$${\mathrm{Q}}_{\mathrm{s}}={\mathrm{Q}}_{\mathrm{s}\mathrm{c}}+{\mathrm{Q}}_{\mathrm{s}\mathrm{b}}$$

(11)

Q_s: in situ production quantity, Q_sc: in situ column quantity production, Q_sb: in situ beam quantity production.

$${\mathrm{N}}_{\mathrm{m}}=\sum _{\mathrm{i}=1}^{\mathrm{n}}{\displaystyle \frac{{\mathrm{Q}}_{\mathrm{M}\mathrm{i}}\times {\mathrm{T}}_{\mathrm{P}\mathrm{C}}}{{\mathrm{T}}_{\mathrm{s}}}}$$

(12)

N_m: number of molds; Q_Mi: in situ production quantity of each mold type; T_PC: production cycle time; T_s: in situ production time; i: number of mold types (1,…, n).

$${\mathrm{N}}_{\mathrm{c}}={\displaystyle \frac{\left({{\mathrm{T}}_{\mathrm{U}\mathrm{E}}\times \mathrm{Q}}_{\mathrm{S}}\right)}{{\mathrm{T}}_{\mathrm{e}}}}$$

(13)

$$\mathrm{S}\mathrm{u}\mathrm{b}\mathrm{j}\mathrm{e}\mathrm{c}\mathrm{t} \mathrm{t}\mathrm{o} {\mathrm{N}}_{\mathrm{C}}\ge 1,\mathrm{i}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{g}\mathrm{e}\mathrm{r}$$

N_C: number of cranes; T_UE: unit erection time; Q_S: in situ production quantity; T_e: erection time.

$${\mathrm{A}}_{\mathrm{Y}\mathrm{S}}=\sum _{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{Q}}_{\mathrm{Y}\mathrm{S}\mathrm{i}}\times {\mathrm{A}}_{\mathrm{Y}\mathrm{S}\mathrm{i}})$$

(14)

A_YS: yard-stock area, Q_YSi: yard-stock quantity of each mold type, A_UYSi: unit of yard-stock area of each mold type, i: number of mold types (1,…, n).

Figure 7 illustrates the optimization process proposed in this study. Based on structural component data extracted from a Revit-based BIM–DT model, this study estimates the production quantity of PC columns and beams. Subsequently, it quantifies CO₂ emissions across three major stages: production, yard-stock, and installation. This quantification considers workload per stage, number of workers, productivity rates, and contingency factors to compute total work time. By multiplying these values by stage-specific CO₂ emission factors, the emissions for each phase are calculated. Furthermore, the model integrates Monte Carlo simulation results generated via Oracle Crystal Ball, specifically incorporating stockyard area as a constraint. The cumulative results from these analyses yield the total CO₂ emissions, enabling the identification of the optimal configuration. The algorithmic structure of this process is detailed in Algorithm A1 of Appendix A.

3.4.3. Optimization Model Design Using Oracle Crystal Ball

Lim et al. (2020) [41] explained six assumptions about the available area, selected as the main influencing factor, and derived the highest cost reduction rate among the scenarios applicable to the field. In addition, Lim et al. (2020) [42] derived the Min and Max for each factor such as in situ production quantity, lead time, number of molds, and number of cranes, and analyzed the effect of quantity on cost and CO₂. However, this study assumes that all members are produced and derives factors affecting CO₂ emission. In addition, the goal of this study is to reduce environmental loads by 10%.

This study presents a procedure for quantifying and optimizing CO₂ emissions generated during the production and yard-stock processes for precast concrete (PC) components by utilizing quantity data extracted from the BIM model in conjunction with Oracle Crystal Ball. The CO₂ emission objective function applied in this procedure is as shown in Equation (15).

$$\mathrm{M}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{z}\mathrm{e} \left({\mathrm{C}\mathrm{E}}_{\mathrm{t}}\right)={\mathrm{C}\mathrm{E}}_{\mathrm{c}}+{\mathrm{C}\mathrm{E}}_{\mathrm{b}}$$

(15)

The range of control variables is set for each scenario. The optimal values applicable to the actual site are derived from the simulation results, and these are fed back into the model to enable iterative improvement.

For efficient in situ production and yard layout, implementation of DT using BIM and optimal space planning strategies based on time series analysis are required. These values were established based on actual site conditions during the preparation stage of in situ production. The assumption conditions of the main factors for BIM implementation are as shown in

Table 6. Input values for CO₂ emission optimization.

Category			Details	Units	Distribution Types
Fixed Variables	Required Construction Duration		18	month	-
	Quantity	Column	850	ea	-
		Beam	1371	ea	-
Parameters	Planned Construction Duration		8	month	Normal Distribution
	Number of Molds	Column	32	ea	Normal Distribution
		Beam	90	ea	Normal Distribution
	Number of Cranes		3	ea	Normal Distribution
	Maximum Yard-Stock Area		15,235	m3	Normal Distribution
	CO2 emission		113,744	T-CO2	Normal Distribution

. These values were chosen considering the on-site conditions during the in situ production preparation stage. The required construction period presented by the client is 18 months, and the applied quantity is 1035 ea of columns and 1906 ea of beams.

The number of molds applied is 32 ea for columns and 90 ea for beams. Two cranes were assigned, corresponding to the number of designated work zones. The maximum yard-stock area derived by reflecting these conditions was calculated to be 15,235 m². The construction period applied on site is set at 8 months to comply with the client’s requested conditions. The factors used as parameters are quantity, number of molds, and number of cranes.

4. Result and Sensitivity Analysis

The production and yard-stock areas for a unit of area are calculated in real time, as shown in

Table 7. Daily area (unit: 1000 m²).

Daily area	D + 17	D + 34	D + 51	D + 68	D + 85	D + 102	D + 119	D + 134
Production area	12.1	12.1	12.1	12.1	12.1	0.0	0.0	0.0
Yard-stock area	2.8	5.7	8.5	11.3	13.9	8.2	4.1	0.0

, using Dynamo. The following equations, Equations (16) and (17), were applied for the area. The production area, installation area, and yard-stock areas for one member are 16.5 m² and 9.6 m², respectively. The maximum yard-stock area of 15,235 m² was derived in real time at lead-time D + 86. The reason why the highest number of yard-stock members is shown during this period is because the PC members produced during the lead-time were accumulated. And since the yard-stock members are installed, the number of PC members decreases as they are installed. The period from D + 86 to D + 134 is the time-lag, which is 3 months. That is, when construction begins, the yard-stock area gradually increases, and the largest yard-stock area is recorded just before the start of installation (D + 86). When installation begins, the yard-stock area decreases, and when the area becomes 0, the PC construction is finished. The daily areas are consistent with those presented in our parallel study [65].

$$\mathrm{A}\mathrm{p}(\mathrm{t})=1\sum \mathrm{n}\{\mathrm{A}\mathrm{i}\mid \mathrm{t}\in [\mathrm{S}\mathrm{i},\mathrm{E}\mathrm{i})\}$$

(16)

$$\mathrm{A}\mathrm{y}(\mathrm{t})=1\sum \mathrm{n}\{\mathrm{A}\mathrm{i}\mid \mathrm{t}\in [\mathrm{E}\mathrm{i},\mathrm{I}\mathrm{i})\}$$

(17)

A_p(t): In situ production area occupied at time t, A_y(t): yard-stock area occupied at time t, S_i: production start date of the ith component, E_i: production completion date of the ith component, I_i: installation date of the ith component

In the same way as schedule optimization, CO₂ emission optimization was performed through maximum repetition simulation (1,000,000 times) in the Oracle Crystal Ball Release 11.1.3.0.0 program. All possible cases of crane and CO₂ occurrence were derived to derive the Min and Max for each factor. The factors used as parameters at this time were quantity, production cycle, lead-time, number of molds, and number of cranes. As a result of deriving the Min–Max management range of the crane as shown in Figure 8, the Min was set to 2.00 ea and the Max was set to 3.270 ea. The reason why the crane does not decrease below 2.00 months is because the minimum number of cranes is required to match the process. As a result of analyzing the management range, the crane and CO₂ emissions, according to the change in each factor, were generally proportional, so they appeared in graphs of a similar shape.

Using the management scope of the crane and CO₂ emissions derived through Monte Carlo simulation, the optimal case is derived as shown in

Table 8. Optimization result for schedule.

Item		Unit	Value
CO2 emission		T-CO2	64,355
Construction time		month	6.5
Number of molds	Column	ea	30
	Beam	ea	91
Cranes		ea	2
Yard-stock area		m2	15,729

. The CO₂ emissions are derived in real time as 64,355T- CO₂, the construction time is 6.5 months, and this value is the smallest value among the derived CO₂ emission values. The number of molds is 30 columns, there are 91 beams, the number of cranes is two, and the yard-stock area is 15,729 m². Upon completion of the simulation, the optimized results are fed back into the BIM environment through Dynamo. This feedback loop enables iterative improvements between design and operation, fostering continuous optimization and enhancing decision-making accuracy in construction planning. When the derived values are applied to the DT model and explained in Figure 9, (a) is modified to (a) D + 96, (b) D + 106, (c) D + 116, and (d) D + 127. The proposed system is designed to divide the entire process into four phases, with the data being updated accordingly. This configuration ensures systematic synchronization between each stage. Through the results, it is determined that the CO₂ emissions increase as the number of cranes increases, which means that the crane is a factor that increases CO₂ emissions. In the future, the values of various factors can be determined on-site, considering the minimum CO₂ emissions, by considering various factors. The number of forms according to the yard-stock area is derived, the lead time and total construction period applied to the number of forms are derived, and the yard-stock area is calculated. CO₂ emissions were calculated by applying these results step by step.

Table 9. Sensitivity analysis.

Variable Pair	Correlation (r)	p-Value	R2
Cranes—CO2 Emission	0.47	0.000076	0.22
Molds—CO2 Emission	0.22	0.00051	0.05
Duration—CO2 Emission	0.00	0.00054	0.00
CO2 Emission—Yard-stock Area	−0.05	0.00023	0.00

shows the sensitivity analysis. The number of cranes and the CO₂ emissions exhibited the most pronounced correlation (r ≈ 0.47). This result indicates that equipment operation is directly associated with increased energy consumption and carbon emissions. It also implies that, as the number of cranes increases, additional waiting time between tasks can be required, which thereby extends idle periods. Furthermore, the project duration showed a negative correlation with the yard-stock area (r ≈ −0.12), which suggests that longer durations tend to improve yard efficiency and reduce spatial occupation. The relationships among the remaining variables were weak, and the duration and CO₂ emissions appeared largely independent.

Table 10. Descriptive statistics (unit: T-CO₂).

Item	Min	Mean	Max	Std. Dev.
CO2 Emission	55,603	58,297	62,862	1421.12

indicates that the CO₂ emissions were calculated with a minimum of 55,603 T-CO₂, a mean of 58,297 T-CO₂, a maximum of 62,862 T-CO₂, and a standard deviation of 1421.12 T-CO₂.

5. Discussion

5.1. CO₂ Emission Result Assessment

This study proposes a BIM-based DT framework for optimizing in situ production and yard-stock management of PC components. Oracle Crystal Ball simulation was used to perform probabilistic CO₂ emissions optimization. The results showed that the environmental impact can be reduced by 8.9%.

In this study, SRC precast components were applied. In previous studies, it was found that SRC precast components can reduce the construction period by at least 18.7% compared to RC structure precast components [77,78] and emit at least 11.7% less CO₂ [79,80]. It is noteworthy that, when comparing RC and SRC structures, a maximum reduction of up to 32.02% in construction duration can be achieved [78]. However, this result was derived from a study conducted on non-PC structural systems, and therefore does not directly apply to PC structures. If process optimization is applied to actual sites, it is expected that CO₂ emissions will be minimized up to 8.9%.

While this study primarily focuses on optimizing carbon emissions, it is essential for future research to recognize the potential trade-offs between carbon reduction and project schedule acceleration objectives. Shortening the construction duration may increase CO₂ emissions as more equipment and resources are required. On the other hand, reducing equipment operation time to reduce CO₂ emissions can require an extension of the construction period. Accelerating the schedule can lead to increased carbon emissions due to intensified equipment use, while minimizing emissions can require extended timelines. Therefore, a multi-objective optimization model was considered, and the most eco-friendly layouts were introduced to support informed decision-making.

5.2. Comparison with Previous Studies of This Case Site

This study conducted Monte Carlo simulations using Oracle Crystal Ball and achieved an optimized CO₂ emission reduction of 8.9%. As shown in

Table 11. CO₂ emission comparison with previous studies (unit: T-CO₂).

Classification	This Study		Lim and Kim (2020) [36]		Kim et al. (2023) [76]
	In Situ Production, Yard-Stock and Erection	CO2 Optimization	In Situ Production	In-Plant Production	Actual Erection	Simulation-Calculated Erection
Material use	37,940	37,940	32,169	32,169	-	-
Labor use	363	283	-	-	372	341
Oil use	24,772	19,371	987	987	73,004	53,965
Electricity use	689	538	543	543	427	307
Transport equipment use	-	-	-	5397	-	-
Lighting, and heating use	352	275	-	-	197	153
Environmental conservation	239	186	-	-	110	83
Total	64,355	58,597	33,699	39,095	74,110	54,850

, CO₂ emission values derived from the same case study in previous research were compared. Lim and Kim (2020) [36] estimated the CO₂ emissions for the total number of columns and girders as 33,699 T-CO₂ for in situ production and 39,095 T-CO₂ for in-plant production. The differences in CO₂ emission values between this study and Lim and Kim (2020) [36] for in situ production are attributed to the following factors: (1) Lim and Kim did not consider labor use, lighting and heating use, or environmental conservation in their calculations and, (2) while Lim and Kim (2020) [36] applied RC structural systems, this study applied an SRC system, which resulted in different material usage.

Additionally, Kim et al. (2023) [76] estimated CO₂ emissions for the erection phase using actual inputs for labor, oil, electricity, lighting and heating, and environmental conservation, yielding a total of 74,110 T-CO₂. By applying the model developed in this study, a 25.99% reduction in CO₂ emissions was achieved compared to the actual construction. In contrast, this study demonstrated a reduction of 8.9%, which appears to be smaller than that of Kim et al. (2023) [76]. However, it should be noted that Kim et al. (2023) [76] only assessed the erection phase and did not include material use, which constitutes a significant portion of CO₂ emissions. When material use is included in Kim et al. (2023) [76]’s analysis, the resulting reduction is 8.2%, which is lower than the 8.9% reduction achieved in this study.

As shown in Table 11, rebar and structural steel materials account for the highest CO₂ emissions among all components. Therefore, the most effective approach to reducing total CO₂ emissions is to minimize the material usage of rebar and steel. However, in this study, the material quantity and structural design were not considered as influencing factors. Instead, the parameters used in the simulation model included the planned construction time, number of molds, number of cranes, and maximum yard-stock area. Accordingly, in order to achieve a more substantial reduction in CO₂ emissions—particularly addressing the material use, which accounts for over 33% of total emissions—future research should focus on the development of low-carbon structural materials and design strategies that minimize rebar and steel consumption.

5.3. Mechanisms by Which DT–BIM Integration Reduced CO₂

The integrated BIM–DT system is structured around three core mechanisms for reducing carbon emissions on construction sites: precise data utilization based on full-process modeling, simulation-based optimization, and the incorporation of real-time operational information. Firstly, BIM enables detailed modeling of PC components, such as columns and beams, by defining their geometry, quantity, and material specifications. This accurate quantification helps prevent overproduction and unnecessary stockpiling, thereby reducing resource waste and, indirectly, associated carbon emissions.

Subsequently, the digital twin reflects actual site conditions and operational parameters, such as crane deployment and task durations, in real time, enabling holistic optimization of production, yard-storage, and installation workflows. Analytical tools like Oracle Crystal Ball are employed to model variables such as working time, mold count, and crane allocation using probability distributions (normal distribution) and to simulate outcomes under multiple scenarios. This process facilitates the derivation of optimal work plans that minimize fuel-intensive operations, directly contributing to CO₂ emission reduction.

The introduction of in situ production, in particular, eliminates long-distance transportation from factories and shortens the movement paths of cranes and equipment. This leads to a significant decrease in fuel consumption, one of the major sources of carbon emissions. This study’s simulation results indicate a CO₂ reduction effect of approximately 8.9%.

Additionally, the integrated framework quantitatively calculates resource consumption and CO₂ emissions at each process stage (production, yard-stock, erection) by applying LCA coefficients. These results are visualized through a dashboard interface, enabling stakeholders, such as clients and contractors, to clearly understand each process’s environmental contribution and adopt carbon-reduction strategies accordingly.

In summary, the DT–BIM integration offers a scientific mechanism for carbon mitigation on construction sites through production planning precision, simulation-based optimization, reduced material transport distances, efficient yard-space utilization, LCA-based emissions quantification, and real-time feedback systems. This framework serves not merely as a design aid but as a next-generation digital decision-making system that simultaneously addresses environmental performance and construction efficiency.

5.4. Decision-Making Suggestion

The decision-making process for minimizing environmental loads, as conducted in this study, is detailed in Figure 10. This process can be partially modified and adapted for practical implementation on construction sites, depending on the specific objectives and goals.

(1)

In the data definition and boundary setting phase, the geometry, quantity, and material properties of precast concrete (PC) elements such as columns and beams are defined and extracted from the BIM model. Concurrently, real-time field sensor data and operational information are collected through the digital twin system. This stage also involves clearly delineating the boundaries of environmental load analysis by defining the scope of the life cycle assessment (LCA), for example as cradle-to-gate or cradle-to-site.

(2)

The simulation environment setup for in situ production via the DT phase involves constructing a simulation environment that integrates BIM data with real-world site information using a DT platform. Key variables such as the production time, number of workers, productivity rates, transportation paths, and storage area dimensions are modeled. This simulation framework enables real-time tracking of resource inputs and environmental impacts across each construction phase and serves as a basis for scenario-based comparative evaluations.

(3)

In the LCA-based CO₂ emission estimation phase, CO₂ emissions are quantified for each process phase—production, yard-stock, and installation—based on its respective resource and activity inputs (e.g., materials, labor, electricity, and fuel consumption). Each phase’s contribution to total carbon emissions is analyzed to identify major emission drivers, which provides insights for targeted environmental interventions.

(4)

In the optimization scenario design phase, based on the calculated CO₂ emission data, various optimization scenarios are developed (e.g., carbon-reduction-oriented, schedule-driven, cost-saving, or multi-objective scenarios). Monte Carlo simulations are conducted using tools such as AI-based models or Oracle Crystal Ball. Input variables (e.g., working hours, manpower, storage area) are treated as probabilistic distributions—typically normal distributions—that reflect their statistical characteristics. The output includes optimized plans for in situ production, yard storage allocation, and installation sequencing.

(5)

The risk evaluation phase evaluates potential risks associated with each scenario, including spatial congestion, scheduling delays, and material interference. Through the DT, dynamic feedback mechanisms are established, which enables real-time scenario adjustments. For example, if excessive stacking occurs in a yard area, simulation reruns can automatically reallocate PC components to alternative zones or revise installation sequences.

(6)

In the final decision-making and visualization reporting phase, all analytical outcomes are visualized in 3D using the BIM environment. Dashboards provide real-time insights into CO₂ emission contributions by phase as a result of space utilization rates, process durations, and scenario comparisons. Stakeholders, including owners, contractors, and engineers, can use this information to select and implement the most sustainable and operationally efficient strategies.

5.5. Discuss Trade-Offs (Digital Overhead vs. Savings)

This study demonstrates a concrete effort to strike a balance between productivity and environmental sustainability by integrating DT and BIM systems. However, while this approach offers clear benefits in terms of reducing environmental loads, it also entails significant costs and technical complexities associated with the deployment of digital systems. In this context, this study serves as a representative case for analyzing the trade-offs between digital overhead, stemming from initial implementation costs and ongoing maintenance, and the environmental and operational benefits of reduced CO₂ emissions and optimized production planning.

Digital overhead begins to accumulate from the initial stages of system implementation. Establishing BIM data and integrating them with a digital twin environment requires advanced modeling, real-time site data synchronization, the deployment of IoT sensors, and the application of simulation-based analytical tools such as Oracle Crystal Ball. These technical requirements increase upfront costs and demand considerable time and resources to build the necessary digital infrastructure. Moreover, such systems require continuous maintenance and data synchronization during operation, including regular updates of BIM tools (e.g., RVT, IFC) and personnel support for reflecting real-time sensor data into the DT environment. These factors collectively contribute to the high managerial complexity of digital-based systems.

Despite these challenges, the integrated DT–BIM framework proposed in this study yields significant advantages. The most notable achievement is the reduction in CO₂ emissions: through simulation-based optimization of production, yard-stock, and installation processes, this study demonstrates a potential reduction of approximately 8.9% compared to conventional methods. Particularly in in situ production scenarios, material transport is minimized, and equipment movement is reduced, which leads to a notable decrease in fuel consumption—a key source of CO₂ emissions. Furthermore, the dashboard visualization of the resource input and carbon output per process stage enables both clients and contractors to gain a clear understanding of each stage’s environmental contribution, facilitating more strategic decision-making.

Additionally, this framework contributes to schedule and spatial optimization. Zoning plans that account for crane rotation radii and operating capacities, sequencing of precast component placements within the yard, and installation planning based on crane logistics are shown to improve productivity even in space-constrained urban construction sites. Scenario-based evaluation using Monte Carlo simulations enhances resilience to uncertain site variables and allows for proactive risk mitigation (e.g., material congestion, spatial limitations).

In conclusion, this study illustrates that, while the integration of digital twin and BIM systems entails certain costs and complexities, the long-term benefits, such as CO₂ emissions reduction, construction efficiency, and risk mitigation, can be substantial. Such integrated strategies are especially valuable in the current industrial landscape, where enhanced ESG (environment, social, governance) evaluation criteria, carbon taxation, and green-smart construction practices are gaining traction. Therefore, the proposed approach not only represents a technological innovation but also exemplifies a digital decision-making framework capable of driving the environmental and operational transformation of the construction industry.

5.6. Present Limitations and Future Research Directions

This study presents a practical framework aimed at reducing environmental loads, particularly carbon dioxide emissions, through the integration of DT technology and BIM in the in situ production and yard-stock management of PC components. However, the proposed system also entails several technical and operational limitations, along with areas for potential improvement:

(1) One significant challenge lies in the reliability and accuracy of the real-time data provided by the digital twin system, which is susceptible to variables such as sensor coverage, data transmission stability, and discrepancies between the actual site and the digital model. If the real-time field conditions are not accurately reflected in the BIM environment, simulation-based decision-making may result in errors. Incomplete or imprecise data inputs undermine the effectiveness of the digital twin.

(2) From the perspective of environmental impact analysis (LCA), this study primarily focuses on CO₂ emissions, excluding multidimensional environmental indicators such as total energy consumption, material lifecycle, water usage, and fine particulate matter. This narrows the scope of result interpretation and suggests a need to expand the analytical framework for more comprehensive sustainability evaluations.

(3) The decision-making model does not incorporate multi-agent simulations that account for interactions among multiple stakeholders. Complex decision structures involving resource competition and schedule conflicts among cranes, equipment, and laborers are difficult to capture using a single optimization algorithm.

To enhance the framework, future research may take the following directions:

(1) Simulation methods that employ Bayesian analysis or probabilistic LCA models should be introduced to address uncertainties inherent in construction environments. These approaches can generate more robust decisions by accounting for variable fluctuations and risk. Furthermore, the development of multi-objective optimization algorithms that balance carbon reduction, cost savings, and schedule compression is necessary. Machine learning or reinforcement learning techniques may improve prediction accuracy and automation levels.

(2) Integrating the BIM–DT system with supply chain management (SCM) or enterprise resource planning (ERP) systems could enable the prediction of material arrival times, the adoption of just-in-time (JIT) supply strategies for yard logistics, and the minimization of material waste. Another critical task is securing long-term empirical data to quantitatively validate the framework’s applicability and performance in actual projects.

(3) Expanding the set of environmental assessment indicators is essential. This includes incorporating factors such as energy consumption, on-site health and safety (e.g., noise, vibration, dust), and material recyclability. The development of a dashboard that can visualize and quantify these indicators would enhance decision-making transparency and stakeholder confidence.

(4) Ultimately, the development of internationally standardized open BIM–DT protocols is necessary to ensure interoperability across platforms and to streamline sensor metadata structures. Such standardization would improve scalability and maintenance efficiency. This evolution goes beyond mere technological advancement and contributes to establishing a strategic digital construction management system aligned with the era of carbon neutrality.

6. Conclusions

This study proposed a BIM-based DT framework to optimize in situ production, yard-stock management, and installation of PC components in large-scale construction projects. By integrating probabilistic simulation with BIM-derived data based on data that are actually applied in the field, the framework effectively addressed the core constraints of environmental impact assessment within a unified digital environment.

Using Oracle Crystal Ball, this study conducted Monte Carlo simulations to evaluate CO₂ emissions. CO₂ emissions were optimized to 64,355 T-CO₂, which highlights the significant environmental benefits achievable through intelligent layout and production planning. Notably, increased crane use was associated with a higher carbon output, which indicates that equipment operation is a key driver of environmental load.

Overall, the results validate the feasibility of predictive and simulation-driven planning in managing complex construction variables. By enabling real-time synchronization of production, yard layout, and installation processes, the proposed framework contributes to more sustainable and resilient construction strategies. This study was conducted by utilizing data that were actually applied in the field and presenting assumption conditions. The optimization model can be used to control carbon emissions, and the DT model can be used to reflect real-time feedback in the field.

From an academic perspective, this research advances the application of DT technology in construction by holistically integrating environmental performance into a single model. Practically, it provides actionable insights for project managers and policy makers seeking to implement smart, low-carbon construction practices. However, the findings are limited to a single case project and do not yet incorporate real-time sensor feedback or automated 3D yard modeling.

Future research should focus on expanding the applicability of the framework in the following ways:

• Integrating real-time sensor data for dynamic feedback control;
• Applying machine learning techniques to improve CO₂ emission forecasting;
• Enhancing safety and risk management through predictive analytics;
• Developing cloud-based visualization dashboards for intelligent site monitoring.

By addressing these directions, the framework can evolve into a fully autonomous, adaptive DT system capable of supporting the next generation of sustainable construction management.