Bu çalışmada bulanık kümeler yardımıyla inşaat şantiyelerine özgü bir risk analiz modeli geliştirilmiştir. Türkiye ve dünyada iş güvenliğine ilişkin istatistikler ve güncel durum gözönüne serilmiş, iş güvenliği yönetim sistemleri incelenmiş, bu sistemlerin ayrılmaz bir bileşeni haline gelen risk analizi ve tehlike değerlendirme teknikleri ayrıntılı bir şekilde ele alınmış ve inşaat sektörü için uygun bir risk analizi modeli için gerek literatür gerekse de saha araştırması yapılmıştır. Toplam 5239 iş kazası ayrıntılı bir şekilde incelenmiş, 58 şantiyedeki iş güvenliği uygulamaları sahada araştırılmış ve model bir tünel şantiyesinde uygulanmıştır. Kaza Olabilirliği parametresinin bulunmasında, iş kazası dosyaları incelenerek, her şantiyede her tipte kazanın hangi yüzde ile gerçekleştiğinden hareketle elde edilen sayısal veriler, bulanık kümeler yardımıyla sözel ifadelere çevrilmiştir. Güvenlik Düzeyi parametresi için ise her kaza tipi için alınması gereken önlemlerin ayrı ayrı sınıflandırıldığı ve uzmanlarca ikili karşılaştırmalar yoluyla farklı ağırlıklar verilen yeni bir kontrol listesi kullanılmıştır. Bu kontrol listesindeki her maddenin ağırlığını saptamak amacıyla Analitik Hiyerarşi Yöntemi kullanılmış, kontrol listesi 1-10 arası bir ölçekte kontrol yapan uzmanların, her iş güvenliği önlemine evet/hayır veya 0/1 vermesi yerine bir puan vermesine dayanmıştır. Kaza Şiddeti parametresi için ise, 1-5 arası ölçekte şiddet tanımları yapılmış, deneyimli uzmanların görüşleri ile sayısal ifadeler oluşturulmuş, sonrasında sözel ifadelere çevrilmiştir. Yöntemin üç parametresi bulanık kural tabanlı sistemin girdileri olarak kullanılmış ve her kaza tipi için çıktı parametresi Risk Düzeyi bulanık çıkarım ve harmanlama yöntemi ile bulunmuştur.

Figures of the occupational accidents and the related fatalities and injuries reached an alerting level in Turkey as well as in the world. Especially in the construction industry, the working environment not conforming to safety rules and other elements leads to thousands of fatal or non-fatal injuries in each year. One of the common characteristic of the commonly implemented management systems is emphasizing the importance of the hazard assessment and risk analysis. In this study, an approach for risk analysis for occupational safety on construction sites with a fuzzy rule-based safety analysis is recommended to deal with the uncertainty and insufficient data. By this approach, historical accident data in the industry, subjective judgments of the experts and the current safety level of a construction site can be combined by the utilization of fuzzy rule based system. In the scope of the study, records were taken from the Social Insurance Institution (SII) General Directory archives in Ankara and 4347 of them occurred on construction sites. In addition to 4347 files, 892 court expert reports which are submitted to criminal and labour courts were examined thoroughly. The likelihood of each particular cause of accident differs for different types of construction work. In the approach presented in this study, accident likelihood and fuzzy set definition of each cause of accident were defined according to the construction work. By the combination of subjective judgement and gathered data, the linguistic variables were employed to develop fuzzy membership functions for accident likelihood. The Accident Likelihood (AL) is one of the main parameters used to assess the Risk Level (RL) of a construction site. The second parameter is the Current Safety Level (CSL) and it needs to be defined as fuzzy sets. Firstly acquisition of safety measures to be taken and ranking of their importance were accomplished. A new kind of checklist was prepared to assist those employers and employees who seek to comply with the rules and regulations of the International Labour Organization and current safety legislation in Tur key. While the checklist prepared according to the boolean approach asks “is the scaffolding safe?” for example, the approach presented in this paper asks “how safe is the scaffolding?”. Using this checklist an expert can evaluate an item related to site safety, by a scale between 1 and 10. All the safety measures for each accident cause were weighted by experts. In this research, the pair wise comparison method and Analytical Hierarchy Process were utilized. To define the Consequent Severity variable, a literature research was done and it was derived that despite the fact that many different approaches exist, 1-5 or 1-10 scale are commonly utilized to assess the consequences of the occupational accidents. However, the subjective judgment of the experts is also required here because the objective data extremely rare or insufficient in Turkey. The most common nine type of occupational accidents and unclassified accidents defined as “other types” are assessed by the experts using 0-100 scale. The accident severity is defined as five different linguistic terms from minor to catastrophic and the averages of the scores given by the experts for each type of accident put forward the fuzzy definition of Consequent Severity. In practical implication of the safety measures, the fuzziness of the antecedents eliminates the need for a precise match with the inputs. All the rules that have any truth in their premises will contribute to the fuzzy risk level expression. Risk Level, combined by matching them against rules in a rule base system, was evaluated with Mamdani-type inference system and then defuzzified to assess the hazard level of the job site. In this study, the aim was to find the risk level of a construction site and five linguistic variables (very safe, safe, slightly risky, risky and extremely risky) were used to describe the performance and the risky atmosphere of the site. The output set also was defined using fuzzy Risk Level (RL) sets as in the same ways the fuzzy inputs were defined. In the last part of the study, the outcomes and contribution to the science were discussed and some recommendations were put forward. The model presented was constructed by performing a broad literature research and field study and it is thought that it will satisfy some urgent demands of the construction industry.