Journal Search

Korean Journal of Construction Engineering and Management

Korean Journal of Construction Engineering and Management

ISO Journal Title : Korean J. Constr. Eng. Manag.
Open Access Journal Bimonthly
  • ISSN (Print) : 2005-6095
  • ISSN (Online) : 2465-9703
Online Submission

Current Issue : Vol.27 No.3, 2026-05

more

Proposal of a Checklist for Safety Management to Prevent Accidents at Modular Construction Sites

Younghun Jun ; Kyoontai Kim

https://dx.doi.org/10.6106/KJCEM.2026.27.3.003

There is a perception that modular construction method is safer compared to conventional construction methods that only perform work on-site. However, due to the nature of on-site tasks such as hoisting, assembling, roofing, and finishing work, workers may still be exposed to risks. Therefore, considering the potential accident risks that may occur on modular construction sites, research on preemptive disaster prevention and safety management for these sites is necessary. The purpose of this study is to develop a safety management checklist for disaster prevention at modular construction sites. The worker safety management checklist reflects empirical results from application to small and medium-sized modular construction sites in Korea. And it examines the work characteristics, key inspection items, appropriateness, usability, and effectiveness of modular construction sites. It is expected that the results of this study will contribute to research on worker safety management at modular construction sites in the future.

Full Text PDF Abstract

Development of an Intelligent Clash Information Exploration Module Based on a Semantic Knowledge Graph for BIM Semantic Retrieval

Donguk Shin ; Hyunwoo Lee ; Wonbok Lee ; Yoonjae Sung ; Bonsang Koo

https://dx.doi.org/10.6106/KJCEM.2026.27.3.012

BIM models are typically designed separately by discipline and later integrated, which often leads to numerous clashes. However, existing BIM clash detection tools are limited to physical interference detection and do not support decision-making for clash resolution. Consequently, reviewers must manually classify and analyze extensive clash results, leading to inefficiencies in the review process. To address these limitations, this study proposes an LLMChain-based intelligent clash information exploration module that integrates a semantic knowledge graph with a Large Language Model (LLM). The BIM model was converted into a semantic knowledge graph incorporating clash attributes and spatial information, and the module was designed using the LangChain framework with a three-stage structure: query interpretation, Cypher query generation, and response generation. Experimental validation showed that the proposed module achieved Executability of 0.92 and Semantic accuracy of 0.85, improving by 0.22 and 0.19, respectively, compared to the GPT-4o single-prompt model. These performance improvements demonstrate that stepwise reasoning based on the LLMChain framework and graph-based semantic exploration effectively enhance the reliability and stability of complex query processing, indicating strong potential to substantially advance the automation of BIM clash review workflows.

Full Text PDF Abstract

Evaluation of Scan-to-BIM Quality for Stairwells Using a Mobile Laser Scanner

Chan-Sol Choi ; Min-Koo Kim

https://dx.doi.org/10.6106/KJCEM.2026.27.3.026

As digital transformation accelerates in the construction industry, Scan-to-BIM technology has emerged as a key component for construction quality verification and facility maintenance. While terrestrial laser scanners (TLS) offer high precision, their long acquisition times and limited portability constrain field applicability. This study aims to address these limitations by utilizing a mobile laser scanner (MLS) capable of rapid data acquisition, and to quantitatively evaluate the Scan-to-BIM accuracy of staircase point clouds captured with MLS. The performance was compared against TLS data to assess the feasibility of MLS for practical site applications. MLS and TLS scans were conducted on an actual building staircase, and BIM models were generated from the acquired data. Quantitative analyses were performed on point density, stair width, riser height, and other key dimensional attributes. Results indicate that although MLS data acquired at low speed (0.3m/s) did not satisfy the point density criteria with an average point spacing of 4.21㎜ the average dimensional error was 4.57㎜, meeting the geometric accuracy required for LOD Level 4. To improve point density, the PU-GAN point cloud upsampling algorithm was applied, reducing the average point spacing from 4.21㎜ to 2.11㎜. In conclusion, the mobile laser scanner demonstrated its potential as a practical alternative by providing rapid data acquisition and sufficient geometric accuracy. This study contributes to validating the field applicability of MLS-based Scan-to-BIM workflows and highlights its value as an effective method for construction quality assessment.

Full Text PDF Abstract

Accident Prediction for Small-Scale Construction Sites using Administrative Data: A Cost-Sensitive DNN and Split Modeling Approach

Ji-Ho Im ; Dong-Woo Lee ; Chi-Woong Moon ; In-Taek Jeon

https://dx.doi.org/10.6106/KJCEM.2026.27.3.035

This study aims to address the data imbalance and performance degradation problems encountered in developing an AI-based construction accident risk prediction model using administrative data provided by KISCON and CSI. Existing integrated models, which were trained on the entire dataset, failed to capture the characteristics of small-scale projects, resulting in poor accident detection for sites under 500 million KRW. To overcome this limitation, this study first restored missing progress rate information by applying an S-curve regression model based on KISCON construction duration and CSI accident data, utilizing it as a key input feature. Subsequently, we proposed a separate Deep Neural Network (DNN) model specifically targeting the "under 500 million KRW" project segment. To address the severe data imbalance in small-scale projects, we implemented a cost-sensitive learning strategy by assigning higher weights to the accident class and applying decision threshold optimization. Experimental results showed that the accident detection rate (Recall) of the existing integrated model for small-scale projects was merely 1.5%. In contrast, the proposed separate model achieved a recall of 64.8%, demonstrating a significant improvement in predicting potential risks at small construction sites. These results suggest that a segregated modeling approach considering data heterogeneity by project scale is essential for building effective construction accident prevention systems.

Full Text PDF Abstract

Automated Bridge Component Extraction Method for Enhanced Bridge Scan-to-BIM

Sung-Jae Bae ; Junbeom Park ; Minji Song ; Joon-Hee Ham ; Jung-Yeol Kim

https://dx.doi.org/10.6106/KJCEM.2026.27.3.042

The absence or inaccuracy of bridge design drawings hinders effective inspection and maintenance. To overcome this, Scan-to-BIM methods have been studied, yet bridge point cloud data (PCD) often includes various background objects due to outdoor scanning. These irrelevant objects degrade the performance of semantic segmentation and are inefficient to remove manually. Learning-based approaches still struggle with low background recognition accuracy and may exclude critical components. This study proposes a clustering-based method using HDBSCAN, along with a user-interactive software tool that allows efficient background removal while preserving essential structural components. The software supports real-time weight adjustment and visualization for flexible output refinement. A case study on four real-world bridges demonstrated high performance, with precision of 0.90, recall of 0.96, mIoU of 0.87, and overall accuracy of 0.93. The proposed method is practical and well-suited for PCD preprocessing in Scan-to-BIM workflows, providing an effective solution for component extraction with minimal loss of critical data.

Full Text PDF Abstract

LLM-Based Interpretation and Refinement of Ambiguous Site Instructions - A Support System for Novice Site Managers in Apartment Finishing Works -

Ju-Yeon Park ; Min-Jung Kim ; JeeHee Lee

https://dx.doi.org/10.6106/KJCEM.2026.27.3.055

Ambiguous on-site verbal instructions in apartment finishing works can lead to misconstruction and rework, especially for novice site managers. This study proposes an LLM-based support framework that classifies such instructions into three ambiguity types (A/B/C), diagnoses missing information, and generates refined written instructions and checklists for field use. Using 54 real on-site utterances collected from apartment finishing works, 600 synthetic instructions were generated with GPT-5.1 to train a TF?IDF and logistic regression classifier. On an external set of 54 real instructions, the model achieved an accuracy of 0.65 and an F1-macro of 0.64, with relatively better performance on Type A and systematic confusion between Types B and C. A completeness analysis showed that average coverage improved from 0.35 in the original instructions to 0.88 in the refined outputs. These results suggest that LLM-based refinement can support novice site managers in interpreting and supplementing ambiguous verbal on-site instructions.

Full Text PDF Abstract

Determinants of Efficiency Gaps and Productivity Trajectories in the Specialty Construction Industry by Region

Chanwoo Lee ; Minsu Cha

https://dx.doi.org/10.6106/KJCEM.2026.27.3.064

The specialty construction industry faces an unprecedented crisis due to declining orders and rising material costs, necessitating sustainable growth strategies through productivity enhancement. This study empirically analyzed the static efficiency and dynamic productivity changes of the industry from 2020 to 2024 using Data Envelopment Analysis (DEA) and the Malmquist Productivity Index (MPI). To ensure statistical reliability, regions with severe "spatial mismatch" between headquarters and site locations were defined as outliers and controlled to mitigate data distortion. The analysis reveals that specialty construction markets in major metropolitan areas are experiencing "diseconomies of scale," characterized by Decreasing Returns to Scale (DRS). This suggests that the phenomenon stems from structural market saturation rather than individual firm inefficiency. Dynamic analysis indicates a slight decline in Total Factor Productivity; however, decomposition shows that firms have maintained stable internal management efficiency (TECI) through self-reliant efforts despite harsh conditions. Instead, the decline of external environmental factors (TCI), such as institutional shifts and economic downturns, was identified as the decisive cause of productivity stagnation. This study holds academic significance by addressing spatial data distortions to enhance the objectivity of regional productivity evaluations. From a policy and practical perspective, it substantiates that the industry's crisis stems from external constraints rather than internal capacity. Consequently, these findings provide an empirical basis for shifting the policy paradigm from simple protectionism toward structural improvements, such as ensuring appropriate construction costs and optimizing regional supply and demand.

Full Text PDF Abstract