Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 Falling Risk Assessment of Advanced Shoring Method Bridge Construction Projects Tung-Tsan Chen1, C.C. Wang 2 1 Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. 2 Associate Professor, Department of Civil Engineering and Engineering Management, National Quemoy University, No.1 University Rd., Kinmen County, 892, Taiwan, R.O.C., Email: tungtsan@nqu.edu.tw Associate Professor, Department of Civil Engineering and Engineering Informatics, Cheng Shiu University, Kaohsiung, 833, Taiwan, R.O.C. Email: ccw@csu.edu.tw ABSTRACT: Falling accounts for the majority of accidents in a bridge construction project. Site safety management today relies on checklist assessment, which is dependent on the ability and experience of the evaluators. In addition, the method used to carry out site safety management today could not address the most critical factors which lead to site accidents. In order to go beyond the limitations of the traditional checklist method, in this study the factors of falling in construction site are analyzed in a Fault Tree (FT), and then the FT is transformed into a Bayesian Network (BN). The BN enables us to explore the probabilities of accidents happening. When compared with current cases of bridge construction in Taiwan, it is worth noting that the result from the BN model agrees with the result from traditional checklist method. In another word, BN model verified to be a realistic and useful model to deal with site safety management problems. INTRODUCTION The construction of infrastructure in a nation directly affects every aspect of citizens’ lives. This is especially true for communication and transportation. Improving the working condition of construction site guarantees a smooth progression, avoiding the various societal cost accidents might induce. In recent years the design and construction of bridges have become more and more complex. When there are faults in the design, the quality, cost, and even the well-being of the on-site construction workers are affected negatively. In response to the above problem, it is important to find the faults in bridge construction, explore their relationships, and analyze the probability of occupational hazard. This enables concerned construction agencies to come up with ways to avoid accidents, thus reduce the chance of accidents in construction site. Since falling accounts for the most accidents in construction site, this study focuses on the application of Fault Tree Analysis and its transformation into a Bayesian Network on the factors contributing to falling. Design, Construction, and Maintenance of Bridges 14 Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. LETERATURE REVIEW In this study Fault Tree Analysis is used to determine the factors that lead to accidents and their relationships with each other, and the Bayesian Network is used to overcome the problem of insufficient data. Therefore, a review of studies concerning Fault Tree, Bayesian Network, and transformation of Fault Tree to Bayesian Network is discussed here. Fault Tree Analysis (FTA), which can effectively analyze the causal relationship of fault event and probability technology. However, if there is ambiguity in many factors (e.g., human error), it is difficult to accurately estimate the probability of occurrence of the fault event. Rodak and Sillima (2012) combined Fault Tree with probabilistic risk analysis, and applied the product to the management of underground water. Durga Rao (2009) applied Dynamic Fault Tree to simulate the complex interactions in a construction system, in order to test for its reliability. Fu-Ming Chen issued in 2005 proposed a model that combines Fault Tree Analysis, Fuzzy Set, and Analytic Hierarchy Process and assessed the risk in Braced Excavation. Yunng- Ru, Lai issued 2009 assessed the factors concerning pier degradation with the Fuzzy Fault Tree. Many researchers apply the Bayesian Network on education and examination. Mislevy, Steinberg, and Almond (2002) and Mislevy, Almond & Lukas (2003) proposed Evidence-Centered Design (ECD), which relies on Bayesian Network for its assessment and analysis. Combined Bayesian Network with experts’ opinions, acquiring the probability values in the Bayesian Network subjectively, according to the conscious of the experts, Bayesian Networks allow follow-up by high reliability numerical probability can be analyzed (Xiao etc., 2008). STATISTIC ON OCCUPATIPNL ACCIDENT IN TAIWAN In Taiwan, the field of construction has higher occupational hazard than other fields of work (Figure 1.) According to the statistics of Council of Labor Affairs, in 88 construction related cases from the past 10 years, there are 97 death, 66 severely injured, and 5 lightly injured. Among these cases falling is the most often the cause for accidents. The causes for falling could be various, such as unsafe work behavior, tight work schedule, or faulty equipment etc. Design, Construction, and Maintenance of Bridges 15 Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 16 0.24 0.22 All industry 0.223 Manufacturing 0.21 0.2 Construction 0.188 Fatalities per 1000 Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. 0.18 0.172 0.175 0.161 0.16 0.14 0.129 0.131 0.128 0.12 0.123 0.109 0.1 0.08 0.06 0.077 0.063 0.069 0.067 0.065 0.059 0.04 0.05 0.041 0.02 2000 2001 2002 2003 0.044 0.045 0.038 0.033 0.035 0.034 0.03 0.038 0.035 0.034 0.036 2004 2005 2006 2007 2008 0.036 0.028 2009 0.035 2010 Year Fig. 1. Fatalities per 1000 persons in construction industry and all industries (excluding deaths from occupational disease and traffic accidents), 2000~2010. FALL RISK ASSESSMENT BASED ON BAYESIAN NETWORK For establishing a better model of the Bayesian Network for Fall Risk Assessment, 36 experts were surveyed with a questionnaire that has 97 items. They were to grade these items according to their experience and knowledge. The result is then tested on four bridge construction project (Advance Shoring Method). Building a Fault Tree frame In the FTA, falling is placed as the top event. According to the safety management based on Domino Theory (Heinrich etc., 1980), the reasons for falling Design, Construction, and Maintenance of Bridges Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 could be categorized as place of incident and its condition, indirect causes, and basic causes. These factors and the top event (falling) are related through the use of logic gates. Through experts’ opinions and review of related studies, two major causes that lead to falling are: (1) Improper set-up of the steel frame support; (2) Improper dissemble of the steel frame support. If needed, these factors could be further divided into sub-factors. Lastly, according to the Domino Theory and Safety management (Jitwasinkul, Hadikusumo 2011; Lingard, Rowlinson 2005), the four basic causes for occupational accidents are: insufficient safety training, improper site management, improper health and safety management, and improper health and safety planning. A Fault Tree is established when the interaction of these basic causes, along with indirect causes, are defined with the help of occupational accident logs and expert opinions (As shown in Figure 1 and Figure 2). From Fault Tree Analysis transfer Bayesian Network The construction of the FT follows a top/down design. It’s first to identify a particular undesired event as a top event, and then proceed from the event to its causes until the primary events are reached. The relationships between events and causes are commonly defined and represented by means of 「AND 」or 「OR」 logic gates The transformation from FT into BN includes graphical and numerical tasks. In the graphical mapping, primary events, intermediate events, and the top event of FT are represented as root nodes, intermediate nodes, and the leaf node in the corresponding BN, respectively. The nodes of a BN are connected in the same way as corresponding components in FT. In the numerical mapping, the occurrence probabilities of the primary events are assigned to the corresponding root nodes as prior probabilities. For each intermediate node and leaf node, a conditional probability table (CPT) is developed. The CPTs are developed according to the type of gate. Finally, the Fault Tree is transformed into a Bayesian Network. In this process overlapping nodes are integrated into the same node. In addition, meaningful supplemental arrows are added between nodes based on experts’ opinion. Design, Construction, and Maintenance of Bridges 17 Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 Fig.1. Overall FT of falling accidents of advanced shoring method bridge (1) Fig.2. Overall FT of falling accidents of advanced shoring method bridge (Cont.) Design, Construction, and Maintenance of Bridges 18 Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 Code T G1 G2 G3 F1 Risk factors FTA of Falling Risk of Occupational hazards Incorrect operating of supporting steel Poor deck structure construction Improper lifting operations Poor assembly of supporting steel Code C9 C10 C11 C12 C13 F2 Improper removal supporting steel C14 F3 F4 F5 E1 E2 Poor formwork operating Improper operation of concrete pouring Improper pre-stressed operating Steel cables operator error Lifting material collision C15 B1 B2 B3 B6 D1 improper pier column top operation B7 D2 D3 D4 D5 D6 C1 C2 C3 C4 Poor installation of supporting frame Improper main truss installation Improper main truss removal Crane operator error No guide cable No safety belt Up and down channel blockage Irregularities of Working car Poor safety belt belts and rings Improper environment operation (e.g. Strong wind and rain) Safety equipment improperly Climbing in hanging objects No safety net B8 B10 B11 B12 B13 B14 B16 B17 A1 Risk factors Not guide rope Not anti-drop railing Work belt not fastened properly Workers behind lifting jack No intermediate support Incorrectly command of hanging personnel Personnel to operate error Improper control procedure Incorrect procedure Dangerous procedures or method Improper activities or posture Non-compliance with the Code of Practice Operation error Work site disorder Scattered material No safety equipment Unsafe equipment No personal protective equipment Not implement of self-management Lack of safety and health signs H / S Training and education A2 H / S Environmental maintenance A3 A4 Planning of the H / S Management of the H / S C5 C6 C7 C8 MODEL VERIFICATION The results from applying this study’s proposed method on four bridge construction projects can be seen in Table 1. In the process of this study, every construction project’s safety assessment is carried out by using safety assessment checklist. The probabilities for the four basic causes are determined subjectively. After the post probabilities of intermediate nodes are calculated, the result is plugged into ArgenaRisk. It is noted that the resultant probabilities from the Bayesian Network tightly resembles those from the realistic safety checklist. The four construction projects are assessed monthly for their safety evaluation record. The number one project in safety is indeed without flaw in its safety management. In contrast, the project that ranked last in safety management is Design, Construction, and Maintenance of Bridges 19 Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 20 projected to have a falling risk of 81.264%. Through the application and comparison of BN model on these four construction projects, it is confirmed that a BN model is applicable and reliable in carrying out safety management in bridge construction (Advance Shoring Method). Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. Table 1. Comparison between BN and real site assessment Project No. Fall risk (%) from BN Risk rank By BN Real site assessment (score) Real safety rank 1 30.069 4 89.65 4 2 46.358 2 82.34 2 3 81.264 1 76.06 1 4 45.471 3 86.62 3 CONCLUSIONS In this study a process is introduced, first as a Fault Tree Analysis that addresses the fall risk in a bridge construction project (Advance Shoring Method). Later, it is then transformed into a Bayesian Network to further explore the factors concerning falling. In the process of transforming the Fault Tree into a Bayesian Network, meaningful arrows between nodes are added based on experts’ opinions. Lastly, the logic gates in the FTA are transformed into the probabilities in a BN model, which is then tested on the four bridge construction projects. After comparing the result from the model with the result from the traditional checklist method, the BN model proves to be a reliable and accurate method in dealing with fall risk management. Therefore, a BN model as proposed in this study is able to reduce the risk of falling, given such conditions. REFERENCES Chen Fu-Min, "The risk analysis of the building retained excavation accident", master's thesis, National Kaohsiung University of Applied Sciences, Kaohsiung (2005). Design, Construction, and Maintenance of Bridges Downloaded from ascelibrary.org by Carleton University on 10/30/15. Copyright ASCE. For personal use only; all rights reserved. Design, Construction, and Maintenance of Bridges GSP 251 © ASCE 2014 Durga Rao, K., Gopika, V., Sanyasi Rao, V.V.S., Kushwaha, H.S., Verma, A.K., Srividya, A.,“Dynamic fault tree analysis using Monte Carlo simulation in probabilistic safety assessment”, Reliability Engineering and System Safety ,Vol. 94,pp.872–883 (2009). Heinrich, H.W., Petersen, D. & Roos, N. (1980), Industrial Accident Prevention, 5th ed., New York: McGraw-Hill. Jitwasinkul, B.; Hadikusumo, B.H.W. Identification of Important organizational factors influencing safety work behaviours in construction projects, Journal of Civil Engineering and Management Vol.17, No.4,pp.520-528(2011). Lai Yin-lu, "Fuzzy Fault Tree Analysis of bridge piers deterioration ", master's Thesis, National Cheng Kung University, Tainan, Taiwan (2009). Lingard, H.; Rowlinson, S.,“Occupational Health and Safety in Construction Project Management”. 1st Ed. Spon Press, London. pp.440(2005). Mislevy, R.J., Almond, R.G., Lukas, J. F, A brief introduction to evidence-centered design, pp.8-19(2003). Mislevy, R.J., Almond, R.G.,& Steinberg, L.S., Enhancing the design and delivery of assessment systems: A four-process architecture, Journal of Technology, Learning, and Assessment, Vol.1, No.5, PP.1-66(2002). Rodak, C., Silliman, S.,“Probabilistic risk analysis and fault trees: Initial discussion of application to identification of risk at a wellhead”, Advances in Water Resources ,Vol.36,pp.133–145(2012). Xiao, L., Haijun, L., Lin, L., “Building method of diagnostic model of bayesian networks based on fault tree” in Proceedings of the International Society for Optical Engineering, Beijing, pp. 71272c-1-71272c-6(2008). Design, Construction, and Maintenance of Bridges 21