Developing a Bridge Health Index (BHI) with a Wighted Priority Index (PI) for Maintenance Decision-Making: An Open Data-Based Approach in Korea

Abstract

Bridges are a key component of transportation infrastructure, playing a critical role in socioeconomic development and public safety. In South Korea, many bridges were constructed during the nation’s economic boom in the 1970s and 1980s, meaning that the number of aging bridges with a service life of over 30 years now accounts for approximately 17% of all bridges in the country. Unfortunately, there have been instances of bridge collapses due to the lack of a proper maintenance system for these aging structures. To address this issue, a priority evaluation model called the Bridge Health Index (BHI) was developed to assess maintenance prioritization. In this study, nine actual bridges were evaluated using the proposed model, with the aim of ensuring that the maintenance and management of bridges at risk can be conducted efficiently and reliably.

1. Introduction

Bridges, as key components of transportation infrastructure, play a vital role in supporting socioeconomic development and ensuring smooth traffic flow [1]. In South Korea, the collapse of the Jungja Bridge in Bundang in 2023 and the Dongsan Bridge in Pyeongchang in 2020 and the corrosion of the cables on the Jeongneungcheon Overpass in 2016 have underscored the importance of bridge maintenance and comprehensive safety inspections. Many Korean bridges were constructed rapidly during the economic growth of the 1970s and 1980s, and according to 2023 statistics from Korea’s Ministry of Land, Infrastructure, and Transport, 16.9% (approximately 35,000) of the 165,823 public facilities in the country have been in use for more than 30 years [2,3,4].

For bridges specifically, 17.7% (approximately 6825) of South Korea’s 39,000 bridges have been in use for over 30 years. Furthermore, according to the safety grade statistics of 2023, 21.3% (approximately 7300) of the 34,000 bridges older than 30 years are categorized into one of six safety grades (A, B, C, D, E, or unclassified). The distribution of bridges over 30 years old in each grade is as follows: A grade (13.2%), B grade (16.9%), C grade (46.8%), D grade (82%), E grade (78.9%), and unclassified (22.1%). As the safety grade decreases, the proportion of aging bridges increases [5]. This rise in the number of aging bridges points toward the structural vulnerabilities and technological limitations inherent in the designs and construction practices from decades ago. In this context, bridge maintenance and safety inspections have become increasingly important.

The lack of maintenance for aging bridges is considered a serious issue globally, raising concerns about both safety and the sustainability of infrastructure. Structural defects and the aging of public facilities, including bridges, have been steadily increasing worldwide [6]. The issues are exacerbated by the absence of proper safety inspections and maintenance systems. Historical cases of bridge collapses further underscore the importance of maintenance [7]. Notable incidents in recent decades, such as the collapse of the Seongsu Bridge in South Korea in 1995; the I-35W highway bridge in the United States in 2007, the Morandi Bridge in Genoa, Italy, in 2018; and the pedestrian suspension bridge in the city of Morbi in Gujarat, India, in 2022, have amplified concerns about bridge safety. In such situations, thorough preparedness for maintenance is essential to ensure public safety and to protect lives. At the national level, establishing a systematic and efficient maintenance system is crucial for maintaining bridge safety and guaranteeing public security.

To address the aging of bridges, South Korea must implement more effective maintenance systems and adopt appropriate investment strategies [8]. However, compared to other developed countries, South Korea allocates less funding for bridge maintenance. For instance, in 2024, the United States Department of Transportation announced that it would invest more than USD 5 billion in the repair and reinforcement of major bridges across the country [9]. In contrast, South Korea’s investment in bridge maintenance is approximately only one-third of that of other developed nations. While research is underway to evaluate the safety of bridges, limitations remain when it comes to large-scale assessments. Prominent methods for assessing bridge stability include probabilistic approaches and neural network techniques. Since these methods are generally confined to the evaluation of individual bridges, there is a need for simpler yet effective methods to ensure the overall safety of bridges [10,11,12,13,14,15,16]. Previous studies have explored the challenges of bridge maintenance decision-making from various perspectives. For instance, Sinha et al. [17] suggested a data-driven reliability analysis integrating Weigh-In-Motion (WIM) data and National Bridge Inventory (NBI) attributes via ensemble machine learning models to improve condition assessments from a probabilistic standpoint. However, this approach is primarily diagnostic and not well-suited for prioritizing maintenance across large-scale inventories. Similarly, Gusella [18] proposed a probabilistic, loss-based framework for reinforced concrete (RC) half-joints subjected to traffic loads. This framework incorporates fragility curves and economic metrics to quantify expected annual losses. Nevertheless, the model remains focused on local structural elements and is not readily applicable to network-wide asset management. In contrast, Sassu et al. [19] proposed a time and cost-effective strengthening strategy for deteriorated RC Gerber-type bridges, offering a structural intervention rather than a comprehensive prioritization model.

In contrast, the present study introduces a novel bridge maintenance prioritization framework that emphasizes both practical feasibility and nationwide scalability. By integrating nationally verified inspection indicators and adopting a dual-index framework comprising a Bridge Maintenance Priority (BMP) score and a locally sensitive Priority Index (PI) score, this study aims to bridge the gap between data-driven assessment and operational decision-making. This model prioritizes maintenance needs by considering a range of factors, including a bridge’s risk index, criticality index, comparison index, and weighted factors from the PI calculation guide provided by the Korea Infrastructure Safety Corporation (KISC). It is designed to evaluate and prioritize bridge maintenance using relevant data [20]. The model considers 12 factors, including safety grade, service life, design load, seismic design status, annual average daily traffic volume (AADT), secondary damage in the event of bridge collapse, facility type, and the number of lanes. Additionally, the PI calculation guide assigns weights based on facility types and superstructure forms, taking into account damage types, component types, performance types, and grade types [21]. Factors related to the risk score were assessed using the Analytic Hierarchy Process (AHP) [22]. To ensure the reliability of the AHP survey, statistical software such as the Statistical Package for the Social Sciences (SPSS, version 29.0) was used to calculate Cronbach’s alpha values. The weighted scores for different superstructure types were applied based on detailed safety inspection reports from the Korea Expressway Corporation. The bridge prioritization algorithm was then applied to and evaluated on nine bridges with detailed safety inspection reports.

This study expands on the authors’ previous work [23], which proposed a prioritization model using AHP-derived risk, criticality, and comparison scores. While the authors’ previous study proposed a prioritization model based on risk, criticality, and comparison scores using the AHP, the present study advances the methodology by incorporating a PI that reflects component-level defect conditions derived from detailed safety inspection data. Unlike the previous approach, which relied primarily on overall safety ratings and generalized factor weights, the proposed model introduces a dual-index framework that combines both the BMP score and the PI score. This structure enables more refined and field-responsive prioritization by capturing localized structural vulnerabilities, thereby enhancing its applicability in real-world maintenance planning and emergency decision-making. Relying solely on the BMP system presents a critical limitation, as it does not fully reflect structural-specific risk factors. To overcome this, the PI was developed as a supplementary index that integrates parameters derived from national inspection data, aiming to simplify the evaluation procedure while enhancing the reliability and transparency of the prioritization process. Furthermore, the proposed model is designed to rapidly and systematically derive maintenance priorities for large bridge inventories, ultimately supporting the efficient allocation of limited budgets and facilitating timely investment decisions.

2. The Concept of Risk

The International Organization for Standardization (ISO) defines risk as “the effect of uncertainty on objectives” [24]. Risk management for structures aims to ensure safety and minimize potential hazards through a systematic process. Regular safety inspections are particularly crucial for ensuring structural integrity. The approach adopted by the Government of New South Wales, Australia, consists of four stages, risk identification, assessment, management, and action review, as shown in Figure 1 [25].

Researchers worldwide have conducted numerous studies on structural risk management. For instance, Rashidi et al. [26] developed the Concrete Bridge Remediation Decision Support System (CBR-DSS), which prioritizes bridges in the poorest conditions and efficiently allocates resources by considering various technologies and indicators to optimize maintenance and restoration work. Elsewhere, Santarsiero et al. [27] analyzed Italy’s bridge management guidelines, highlighting differences from other globally adopted approaches. They proposed a method for the high-level assessment and prioritization of bridges using publicly available data sources, such as Google Street View (GSV) and OpenStreetMap (OSM).

3. Proposal of Bridge Maintenance Prioritization Model

We propose a new model for prioritizing bridge maintenance during the planning process. The model utilizes open-source information collected from facility management systems, road bridge and tunnel condition information systems, public data portals, and Korea’s National Land Safety Management Agency’s Investment Priority Index calculation guidelines [28]. The components of the Bridge Health Index model are illustrated in Figure 2.

As shown in Equation (1), we propose a new equation for priority determination, which is central to bridge maintenance. This equation comprehensively considers risk scores, criticality scores, comparison scores, and Priority Index weight scores to determine the maintenance priority of bridges.

$$\begin{array}{c}\mathrm{Bridge} \mathrm{Health} \mathrm{Index} \mathrm{Score}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} =(\sum \left(Risk Score\right)\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \times \sum \left(Criticality Score\right)\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \times \sum \left(Comparison Score\right))+PI Score\end{array}$$

(1)

.

The risk score is calculated as the product of risk factors and impact weights. The elements considered for selecting risk factors consist of factors related to bridge design and maintenance, excluding social, political, and historical factors. The criticality score reflects the bridge’s social impact, including the Average Annual Daily Traffic (AADT), to account for maintaining smooth mobility services for citizens. Another element evaluates the potential secondary damage in the area below the bridge in the event of a collapse, assessing the impact on rivers, roads, and commercial areas caused by collapsed structures and vehicles. The comparison score is categorized by facility type and the number of bidirectional lanes, as indicated in

Table 1. Facility classification criteria [29].

Facility Type	Description
Type 1	(1) Bridges with superstructures in the form of suspension bridges, cable-stayed bridges, arch bridges, and truss bridges.
	(2) Bridges with a main span length of 50 m or more (excluding single-span bridges).
	(3) Bridges with a total length of 500 m or more.
	(4) Cover structures with a width of 12 m or more and a total length of 500 m or more.
Type 2	(1) Single-span bridges with a span length of 50 m or more.
	(2) Bridges with a total length of 100 m or more that do not fall under Type 1 facilities.
	(3) Cover structures that do not fall under Type 1 facilities, with a width of 6 m or more and a total length of 100 m or more.
Type 3	(1) Bridges installed on a road mentioned under Article 10 of the Road Act with an extension of 20–100 m.
	(2) Bridges of least 20 m in length installed on a road other than roads mentioned under Article 10 of the Road Act.

, in accordance with the Special Act on the Safety and Maintenance of Facilities by Korea’s Ministry of Land, Infrastructure, and Transport [29].

3.1. Risk Score Evaluation Criteria

Using standard bridge data from Korea’s Ministry of Land, Infrastructure, and Transport, four main factors were employed to determine the risk score: bridge service life, design load, seismic design presence, and safety grade [2]. However, these factors have varying impacts on each bridge, making analyses and the establishment of criteria challenging. Therefore, building on previously established weighting guidelines [16], as well as values proposed in earlier studies [17], we adopted those weights and multiplied them by the relevant evaluation criteria to compute each bridge’s risk score. Specifically, the final weights were set as follows: safety grade (0.40), service life (0.30), seismic design status (0.20), and design load (0.10). These values reflect both the structural importance and the operational considerations generally recommended in risk-based maintenance practice. These weight values were initially derived through a structured Analytic Hierarchy Process (AHP) survey involving 35 civil engineering experts. While the AHP analysis yielded highly skewed values for the weight distribution safety grade (0.655), service years (0.24), seismic design (0.0651), and design load (0.0399), the relative importance ranking was preserved, and the numerical values were moderately adjusted to enhance practical applicability and interpretability. The final adjusted weights reflect a balance between theoretical rigor and field applicability, ensuring the model’s usability in large-scale infrastructure evaluations. To verify the reliability of the AHP results, Cronbach’s alpha analysis [30] was conducted using SPPS, confirming acceptable consistency among expert responses.

Among the four identified risk factors, safety grade was thus confirmed to have the greatest overall influence, followed by service life, seismic design status, and design load. The final weighting results are summarized in Figure 3, which illustrates the relative impact of each factor.

The score for each element used to evaluate the risk factors was calculated as follows. Service life and design load were classified into four categories each. Service life was divided into less than 10 years, 10–20 years, and 20–30 years and over 30 years. The design load was categorized as DB-24, DB-18, DB-13.5, and unspecified. Seismic design was categorized as either present or absent. The design vehicle load (DB) used in Korea represents the load of an imaginary semi-trailer vehicle and follows the same procedure as the Load and Resistance Factor Design (LRFD) standards of the American Association of State Highway and Transportation Officials (AASHTO). The safety grade was divided into five levels, from grade A to grade E, according to Article 29 of the Enforcement Decree of the Special Act on the Safety and Maintenance of Facilities by Korea’s Ministry of Land, Infrastructure, and Transport [29], as shown in

Table 2. Safety grades by bridge state [29].

Safety Grade	State
A	Optimal condition, without problems
B	Minor defects in sub-members; some measures are required to improve durability
C	Minor defects in main members or extensive defects in sub-members; measures are required to improve durability
D	Moderate defects in main members; action is required
E	Serious defects in main members; urgent action is required after the prohibition of their use

.

Table 3. Scores by risk factor.

Risk Factor	Content	Score	Impact Factor
Age	Age < 10	0.06	0.3
	10 ≤ Age < 20	0.19
	20 ≤ Age < 30	0.31
	30 ≤ Age	0.44
Design load	DB-24	0.06	0.1
	DB-18	0.19
	DB-13.5	0.31
	Unclear	0.44
Seismic design status	Y	0.17	0.2
	N	0.83
Safety grade	A	0.04	0.4
	B	0.12
	C	0.19
	D	0.27
	E	0.38

presents the score composition based on the risk evaluation criteria for each risk factor of the bridge.

The risk score is calculated using Equation (2).

$$\mathrm{Risk} \mathrm{Score}=\sum \left(Risk Factor\times Impact Factor\right)$$

(2)

3.2. Criticality Score Evaluation Criteria

When determining maintenance priorities for bridges, it is essential to have indicators that can quantitatively assess social value, criticality, and risk. To quantitatively determine the social importance and risk of bridges, factors such as traffic volume and the locational conditions of bridges are considered. In this study, the Annual Average Daily Traffic (AADT) was applied as a criterion for assessing the importance of bridges. The AADT is regarded as one of the key indicators for determining the efficiency and safety of road and traffic systems, providing the most accurate measure of the number of vehicles and pedestrians crossing a bridge. Additionally, the study comprehensively considered the impact of bridges on surrounding areas by using open data sources such as Google Street View (GSV) and OpenStreetMap (OSM) to evaluate secondary damage factors, including the effects on rivers, bridges, roads, and commercial areas located below the bridge in the event of a collapse.

To calculate the criticality score, the AADT was categorized into six criteria, and secondary damage factor indices were also applied, with scores assigned based on each criterion. This study adjusted the evaluation criteria and corresponding score ranges based on the perspectives of bridge management organizations.

Table 4. Scores by criticality factor.

AADT (Cars/Day)	Score
AADT < 20,000	0.03
20,000 ≤ AADT < 40,000	0.07
40,000 ≤ AADT < 60,000	0.13
60,000 ≤ AADT < 80,000	0.19
80,000 ≤ AADT < 100,000	0.26
100,000 ≤ AADT	0.32
Secondary Damage	Score
River: Impact on environment, water quality, flood risk	0.2
Road: Traffic disruption, access to businesses/residences	0.3
Bridge: Structural damage, cost of repair/replacement	0.5

presents the evaluation criteria and score allocation for each criticality factor. The criticality score is calculated using Equation (3).

Criticality Score = (AADT Score + Secondary Damage)

(3)

3.3. Comparison Score Evaluation Criteria

The comparison score, which serves as an indicator of the relative size of each bridge, is calculated by multiplying the evaluation criteria by the unique dimensions of each bridge. This approach reflects the factors that enable a meaningful comparison of bridge sizes. Two factors were set as comparison elements: facility classification and number of lanes.

First, the total number of bidirectional lanes serves as an indirect indicator of the bridge’s size and traffic volume. The more lanes a bridge has for bidirectional traffic, the higher the traffic volume it supports, thereby providing an estimate of the bridge’s traffic capacity. The number of lanes was determined using GSV and OSM.

Second, bridge facilities are categorized into Type 1, Type 2, and Type 3 according to the enforcement decree of the Special Act on the Safety and Maintenance of Facilities, as designated by Korea’s Ministry of Land, Infrastructure, and Transport. The detailed criteria for this classification are determined by national facility classification standards. According to this classification, the evaluation criteria and scores for each comparison factor are presented in

Table 5. Scores by comparison factor.

Safety Grade	Content	Score
Facility classification	Type 3	0.23
	Type 2	0.32
	Type 1	0.45
Number of lanes	Lanes < 2	0.03
	2 ≤ Lanes < 4	0.07
	4 ≤ Lanes < 6	0.13
	6 ≤ Lanes < 8	0.19
	8 ≤ Lanes < 10	0.26
	10 ≤ Lanes	0.32

.

3.4. Criteria for Evaluating Scores Based on Priority Index (PI) Weights

Through the detailed safety inspection reports of the Korea Expressway Corporation, the structural characteristics and safety status of bridges can be comprehensively assessed. These reports provide a thorough analysis of six structural categories: superstructure, substructure, bearings, and other components, including deck slabs, girders, piers, abutments, and expansion joints. This detailed assessment helps in determining the safety rating of the bridge. However, since the safety of a bridge varies based on the condition and characteristics of each of its parts, applying a precise safety rating to each superstructure type offers a more reliable approach. By individually identifying and managing the structural vulnerabilities of a bridge, specific safety issues can be detected and addressed early on. Therefore, assigning detailed safety ratings to each structural component enhances the overall safety of the bridge and improves maintenance efficiency by optimizing resource allocation. The Priority Index (PI) score is calculated using the weights for damage types (Dn), member types (Mn), performance types (Pn), and grade types (Gn), as shown in Equation (4).

$$\mathrm{Priority} \mathrm{Index}=\mathrm{Dn}\times \mathrm{Mn}\times \mathrm{Pn}\times \mathrm{Gn}$$

(4)

It is assumed that the damage-type weight is uniformly set at 1.0, because the member type condition rating is influenced by various potential defects. The weights for member types are detailed in

Table 6. General bridge (girder detachable bridge) weights.

Facility	Girder-Detached Bridge
Superstructure	Deck girder	0.15 0.25
Substructure	Abutment pier	0.13 0.13
Bearing support	Bearing support	0.15
Other	Expansion joint	0.07

and

Table 7. General bridge (integral bridge) weights.

Facility	Integral Bridge
Superstructure	Deck girder	0.15 0.25
Substructure	Abutment pier	0.13 0.13
Bearing support	Bearing support	0.15
Other	Expansion joint	0.07

, while the performance-type weights are shown in

Table 8. Weights by performance.

Safety Grade	Safety Performance
General national highway	0.68
Highway	0.68

, and the grade-type weights are provided in

Table 9. Weights by grade.

Grade	Weight by Grade
A	0.05
B	0.09
C	0.18
D	0.30
E	0.38

.

4. A Study on the Application of Nine Bridges in the Proposed Prioritization Model

To evaluate the maintenance prioritization model proposed in this study, maintenance priorities were derived for nine bridges for which detailed safety inspection reports were available. For stable research performance, bridge data management and processing were conducted using a workstation (NZXT KRAKEN 240/PRO 5955WX/WRX80E-SAGE, Daegu, Republic of Korea) at the Intelligent Construction System Core-Support Center, Keimyung University, Republic of Korea.

4.1. Select Bridges with Prioritized Models

To select the bridges for applying the maintenance prioritization model, nine bridges in the Gyeongsang region of South Korea, for which detailed safety inspection reports are openly available from Korea Expressway Corporation, were chosen. These bridges were selected based on the availability and completeness of their inspection records, as well as their diversity in terms of structural type, construction year, traffic volume, and safety condition. These selected bridges were characterized by their construction year, design load, seismic design status, integrated safety rating, Average Annual Daily Traffic (AADT), secondary damage factors, facility type, and the number of bidirectional lanes. These data are presented in

Table 10. Data on nine bridges applying the maintenance prioritization model [2].

Bridge (Bridge Category)	Year of Completion	Length (m)	Maximum Span (m)	Design Load	Seismic Design Status	Total Safety Grade	AADT	Secondary Damage	Facility Classification	Number of Lanes
A (highway)	1995	595.1	30	DB-24	N	C	54,869	Bridge	Type 1	4
B (highway)	2003	800	50	DB-24	N	B	110,851	River	Type 1	8
C (highway)	2005	840	60	DB-24	N	B	24,676	Road	Type 1	4
D (highway)	1995	400	50	DB-24	N	B	37,154	River	Type 1	4
E (highway)	2007	260.4	60	DB-24	N	A	54,319	Bridge	Type 1	4
F (highway)	2007	255.4	60	DB-24	N	A	54,319	Bridge	Type 1	1
G (highway)	2007	120.2	50	DB-24	N	A	54,319	Bridge	Type 1	1
H (highway)	2007	490.7	50	DB-24	N	B	54,319	River	Type 1	2
I (highway)	2005	500	50	DB-24	N	B	25,304	River	Type 1	2

, and the nine bridges were selected as the target for the PI application. Photographic views of these bridges are provided in Figure 4.

The construction years of the bridges range from 1995 to 2007, with some having been in use for up to 29 years. Although all bridges are designed to support the same load (DB-24), none have been subjected to seismic design considerations. The overall safety ratings include three bridges with an A grade, five with a B grade, and one with a C grade. The AADT values vary from 24,676 to 110,851. Additionally, all bridges belong to Facility Type 1, and the number of bidirectional lanes ranges from 2 to 10. The nine bridges feature various superstructure types, including steel box girder, PSC box girder, and PSC beam. The allocation of safety ratings based on structural types as determined through the detailed safety inspection reports is presented in

Table 11. Safety grade by location of the member.

Bridge Name	Facility	Deck	Girder	Abutment	Pier	Bearing Support	Expansion Joint
A	PSC beam	D	E	B	B	C	C
B	Steel box girder	B	B	B	B	B	B
C	PSC box girder	C	C	B	C	B	B
D	PSC box girder	B	B	C	C	B	C
E	Steel box girder	C	B	C	C	B	C
F	Steel box girder	B	B	C	C	B	C
G	Steel box girder	B	B	C	C	B	C
H	PSC box girder	B	B	C	C	B	C
I	Steel box girder	C	C	B	B	B	C

.

4.2. Prioritizing Maintenance of Data-Based Bridges

The proposed maintenance prioritization model was applied to bridges using data collected from open sources, and the results were derived. The calculation process for deriving the maintenance prioritization results is shown in Figure 5, and the results of ap-plying the data-based bridge maintenance prioritization model are illustrated in Figure 6, Figure 7 and Figure 8.

To evaluate the maintenance priorities of the bridges, Bridge A achieved the highest score of 0.2626414, indicating the most urgent need for maintenance. In contrast, Bridge D received a score of 0.1187838, representing the lowest priority for maintenance. Bridge A obtained the highest scores in both the Bridge Maintenance Priority (BMP) score segment (0.1246014) and the Priority Index (PI) score segment (0.13804). Additionally, it recorded the highest scores in the risk and criticality sections of the BMP score, with values of 0.341 and 0.63, respectively. The comparison score section also showed a score of 0.58, the second highest among all bridges. These evaluation results are expected to have a significant impact on bridge safety and maintenance, allowing for the efficient allocation of resources for bridge management. Furthermore, utilizing this scoring system is crucial for being able to make prompt and informed decisions when developing maintenance plans.

4.3. Data-Based Bridge Maintenance Priorities Analysis

To validate the results of the Bridge Maintenance Priority analysis, as shown in

Table 12. Risk, criticality, and comparison score results.

Bridge	Risk Score	Criticality Score	Comparison Score
A	0.341	0.63	0.58
B	0.313	0.52	0.71
C	0.277	0.37	0.58
D	0.313	0.27	0.58
E	0.245	0.63	0.58
F	0.245	0.63	0.48
G	0.245	0.63	0.48
H	0.277	0.33	0.52
I	0.277	0.27	0.52

and

Table 13. Maintenance priority selection results.

Bridge	BMP Score (Rank)	PI Score (Rank)	BHI Score (Rank)
A	0.1246 (1)	0.138 (1)	0.2626 (1)
B	0.1156 (2)	0.0539 (9)	0.1694 (3)
C	0.0594 (6)	0.0863 (2)	0.1457 (6)
D	0.0490 (7)	0.0698 (8)	0.1189 (9)
E	0.0895 (3)	0.0832 (3)	0.1728 (2)
F	0.0741 (4)	0.0741 (5)	0.1481 (4)
G	0.0741 (4)	0.0741 (5)	0.1481 (4)
H	0.0475 (8)	0.0741 (5)	0.1216 (7)
I	0.0389 (9)	0.0826 (4)	0.1215 (8)

, a methodology was applied to calculate the Bridge Health Index (BHI) score by combining the BMP and PI scores. This evaluation method comprehensively reflects the structural and operational conditions of each bridge, incorporating the factors of risk, criticality, and comparison. By utilizing detailed safety inspection reports, including the PI score, the structural characteristics and safety status of each bridge can be accurately assessed, resulting in a reliable prioritization for maintenance.

Among the eight indices in the BMP score, the years of use and overall safety grade were found to have the greatest influence on prioritization in the risk score section. As time passes, the years of use naturally lead to aging, which indicates a decline in structural safety. The accumulation of physical and chemical factors, such as material fatigue, corrosion, deterioration, and repeated loads, can cause significant maintenance issues for the bridge. Therefore, the years of use have a significant impact, and as shown in Table 3, they account for a weight of 0.3 in the risk score impact factor. Accordingly, as shown in Table 12, bridges with greater years of use received higher risk scores.

The overall safety grade index carries the highest weight of 0.4 in the risk score impact factor. It is an important indicator that assesses not only the structural condition of the bridge but also environmental and operational factors, providing a comprehensive evaluation of safety and efficiency. This allows for the identification of risk factors and supports the development of efficient maintenance plans, minimizing social and economic impacts. Therefore, the overall safety grade is a key determinant in prioritizing maintenance. Similarly to the years of use, bridges with higher scores in the overall safety grade also received higher risk scores. With regard to the design load and the presence of seismic design, the nine bridges had identical values for both factors. Consequently, these factors did not significantly impact the risk score.

The AADT, secondary damage potential, and number of lanes in the criticality and comparison sections also significantly influence a bridge’s BMP score. In the criticality section, a higher AADT indicates a higher frequency of vehicle use on the bridge, and as a result, secondary damage to the bridge cannot be overlooked. This increases the urgency for maintenance to prevent secondary damage. The number of lanes represents the traffic capacity of the bridge, and the more lanes there are, the higher the importance of the bridge. For bridges F and G, they have single-lane entrances leading to the merge points between the bridges, resulting in fewer lanes in both directions. Therefore, they received the lowest comparison score of 0.48. Additionally, since these two bridges are located in the same area, they received identical scores in other evaluation categories, leading to the same final score. The facility type did not have a significant impact, as all bridges were classified similarly. In all, bridges with high scores in risk, criticality, and comparison factors ranked higher in BMP score priority. Bridge A received the highest scores in the risk and criticality sections among the nine bridges, and the second-highest score in the comparison section, giving it the highest overall BMP score.

Relying solely on the BMP score may not fully reflect the precise condition and specific structural defects of bridges. To enhance the reliability of the prioritization process, Priority Index (PI) scores were introduced. These scores are derived from detailed safety inspection reports for each bridge’s six structural categories and incorporate the weights of the investment prioritization index guide. This method allows for a more thorough assessment of the structural condition of each bridge.

Bridge A achieved a high PI score of 0.138 among the nine bridges, indicating serious structural defects in the girder and deck slab sections, which were rated with grades E and D, respectively. This high PI score suggests that Bridge A has significant vulnerabilities in its superstructure’s safety, likely due to its longer service life compared with the other bridges. Thus, Bridge A is prioritized for urgent maintenance according to its BHI score. Conversely, Bridge B had the lowest PI score of 0.0539, mainly because its overall structural condition is good and no severe defects were found in the detailed safety inspection. While BMP scores consider a comprehensive range of factors, such as risk, criticality, and comparison, PI scores provide a precise evaluation of individual component conditions. The difference between BMP and PI scores therefore reflects the broader range of factors included in the BMP assessment.

The introduction of PI scores significantly enhances the reliability of maintenance prioritization assessments. By complementing the detailed structural and safety information that might be overlooked when only using BMP scores, the combined BHI score offers a more comprehensive evaluation. This approach allows for the clear identification of urgent maintenance needs, as demonstrated by Bridge A, and supports the safe and sustainable use of the bridge. The maintenance prioritization results are illustrated in Figure 9.

The nine bridges were analyzed based on their BMP score (social importance), PI score (structural risk), and BHI score (composite score) and subsequently categorized into three priority groups, as shown in Figure 10.

Priority 1 includes bridges requiring urgent maintenance, such as Bridge A and Bridge E. Bridge A, with the highest BHI and PI scores, is at significant structural risk and demands immediate attention. It also holds the highest BMP score, reflecting its high social importance. Bridge E, with the second-highest BHI score, shows notable structural deficiencies.

Priority 2 consists of bridges requiring preventive maintenance, including Bridges B, F, and G. These bridges have high BMP scores, indicating significant social importance, but relatively low PI scores, suggesting a lower structural risk. As such, they require medium-term preventive management. Priority 3 includes bridges with a lower need for frequent inspection, such as Bridges D, H, and I. These bridges, characterized by low BHI and PI scores, have minimal structural risk and lower social importance and can be managed effectively through periodic inspections.

The prioritization of the nine bridges serves to optimize the allocation of limited maintenance resources and establish tailored management strategies based on each bridge’s social importance and structural risk. This approach ensures that urgent maintenance needs are addressed promptly while stable bridges are managed through preventive measures or regular inspections, maximizing both efficiency and safety.

5. Conclusions

The aging of infrastructure constructed as far back as the early twentieth century has brought the need for effective bridge management to the forefront worldwide. Many nations, including South Korea, are facing issues related to the aging of major bridge structures. Bridges and public facilities built during South Korea’s economic growth periods of the 1970s and 1980s are increasingly experiencing serious aging problems. As a result, temporary maintenance and repairs are becoming insufficient, highlighting the growing need for a more effective and efficient bridge maintenance system.

A data-driven model for prioritizing bridge maintenance has been developed that not only considers the structural characteristics of bridges but also qualitative factors such as risk, criticality, and comparison factors, as well as structural-type-based weights derived from actual bridge data. Traditional methods of prioritizing maintenance, which primarily rely on national safety ratings, tend to focus on bridges with lower safety grades, often overlooking the social value and importance of those structures. The proposed model incorporates not only risk scores but also criticality-, comparison-, and superstructure-based weighted scores, reflecting the social value and significance of each bridge. This is reflected in Table 13, which displays the prioritization results for Bridge B and Bridge E. While Bridge B received the second-highest Bridge Maintenance Priority (BMP) score of 0.1156 among the nine bridges of this study, it had the lowest Priority Index (PI) score of 0.0539, leading to a change in the Bridge Health Index (BHI) score priority ranking in contrast to Bridge D, which had the third-highest BMP score.

The proposed prioritization model applies both the BMP based on qualitative assessments and the PI based on structural-type detailed safety assessments, thus achieving greater reliability and establishing a more rational assessment standard compared to the national standard that relies only on overall safety grades. In addition, by using data provided by national agencies and local governments to evaluate the risk of each bridge and derive maintenance priorities, the model is expected to be effectively applied even in situations requiring emergency repairs. However, it is important to consider additional factors, such as other facilities located on the substructure of the bridge, collision risks with vehicles due to clearance height limitations, and the impact of landslides or scour. To address these issues, it is essential for national and local governments to actively improve the accuracy and completeness of data.

Furthermore, the proposed model is designed to function as a preliminary risk assessment tool for large-scale bridge inventories where full-scale inspections or detailed structural analyses are difficult to perform. It enables the early identification of bridges requiring urgent intervention by prioritizing structures with critical defects, such as grade E components or severe structural damage, even when their BMP scores are relatively low. This aligns with national safety regulations that mandate immediate maintenance actions upon the detection of serious defects. Therefore, the model not only improves reliability and applicability in normal maintenance planning but also serves as a decision-support tool in emergency situations. Continued research and development are needed to ensure the safety of bridges and enhance public confidence in bridge safety.