Review of previous applications of innovative information technologies in construction health and safety

Health and Safety (H&S) is an important aspect in the construction planning process. The statistics (U.S. Bureau of Labor Statistics 2015; EUROSTAT 2015) indicate that construction industry is a leader in some of the most undesirable statistical categories such as number of accidents at work and fatal accidents, both in absolute numbers and relative to the number of workers employed in the industry. Owing to this enormous loss of life, injuries and disabilities, which impact not only the immediate stakeholders but also the society as a whole, it is not surprising that H&S in the construction industry is of interest not just to practitioners and regulators but to researchers as well. H&S in the construction industry is a propulsive research field with a myriad of scientific papers and books published and research conducted to improve the health and lessen the probability of construction workers’ injuries and death. A prominent section of construction safety research is the application of innovative information technologies (ITs) in order to improve levels of H&S in the construction sites.

Perhaps, the first question that needs to be answered is what constitutes an innovative IT. This is not an easy task since something that was innovative 10 years ago might be extremely common today or what is widely used in one field of research might be completely new and unexplored in another research field. In the context of this research, innovative IT in construction H&S is any application of new or existing technology in a new or different way, with a goal to improve any aspect of construction health and safety.

Using innovative ITs to improve H&S is not a new occurrence in H&S-related research. Zhou et al. (2015) have identified that the first paper appeared in the late 1980s and since then a total of 63 research papers (up to 2013) dealing with innovative technology application as an intermediate solution to prevent workers from injuring themselves and/or others were published. They have also found a sharp increase in the yearly number of H&S-related publications since 2005 (shown in Figure 1), from one or two publications per year to as much as ten publications per year.

Fig. 1

The reasons for the increase in the technology-enabled approaches in the construction safety can be found from the fact that traditional methods of analysis of accidents and causation data provide albeit valuable but general information for safety planning which cannot accurately predict when and where will an accident occur in a construction site (Zhang et al. 2013). Such an increase in number of innovative technologies in H&S and later in turn reliance on them to produce results that can have direct effect on human health might have unintended consequences. According to Weick, the use of innovative technologies can increase the potential for mistakes and accidents since (Weick 1985): “reliance on a single, uncontradicted data source can give people a feeling of omniscience, but because these data are flawed in unrecognized ways they lead to nonadaptive action”. Zhou et al. (2012) agree that digital systems do not encourage the active challenging of assumptions, and checking and validating results produced by software. This presents a serious challenge to all potential applications of innovative technologies in the field of construction H&S.

The goal of this paper was to present previous applications of innovative ITs and to identify whether there exists a literature gap of unexplored or insufficiently explored topics in the H&S literature, which might be worth focusing research efforts in order to best improve the construction H&S.

To do so, after describing the methodology in Section 2, this paper examines and describes the existing efforts in applying innovative ITs in the construction H&S research (Section 3). In Section 4, this paper presents the identified research gaps and gives recommendations on addressing them. Section 5 discusses the possibilities of implementing identified research in construction H&S practice, and Section 6 concludes this paper.

BIM is most often defined as digital representation of physical and functional characteristics of a facility (NBIMS V.3 2015). Kamardeen (2010) further states that BIM “is not just a 3D virtual model but a repository of intelligent building objects with attributes that can understand the interaction between each other, and non-geometric data about the objects and the facility for decision-makings”. Although BIM models are produced using 3D modelling software (Gilkinson et al. 2015), the elements in BIM models are not just a collection of lines or polygons as they are in 2D or 3D CAD, respectively. They represent both the graphical and non-graphical aspects of a building (Eastman et al. 2008) since additional parametric information can be attributed to each of the model elements. BIM models can contain various information about the building and the building elements and actually be treated as a database (Qi et al. 2013).

Given how BIM seems to provide a solution to many of the problems faced by the construction industry and to have a large potential for application, it is not surprising that BIM is also the most often used innovative technology in the literature. In 25 studies (out of a total of 72 studies that describe the development of a construction H&S safety tool), BIM is the only innovative technology applied to H&S, and in 17 other studies, BIM is used either in conjunction with another innovative technology or as a foundation for that other technology (such as for VR). It also has the most unique applications. BIM is used for hazard identification, visualization, safety training, rule checking, JHA, design for safety suggestions and others. A special occurrence with BIM (and 3D/4D CAD, to a degree) is that it is not only an innovative technology by itself but also a platform for other technologies, for example, VR, AR and serious games.

As it can be seen from the “application” column of Table 1, the most prominent use of BIM in H&S is for hazard identification. Zhang et al. (2013) have used BIM to automatically detect falling hazards and model guard-rails and protective covers around and over the openings in the structure. This research topic continued through the works of Sulankivi et al. (2013) who have applied the developed tool on a real-world project and Melzner et al. (2013) who implemented the same tool as an add-on to the existing BIM authoring software.

Other authors have also explored the possibilities of BIM and rule checking for identifying fall hazards. Taiebat (2011) has defined a framework for hazard identification in BIM, which supports the development of a BIM-based hazard recognition tool. His research consisted of development of a design for construction worker safety tool, which efficiently makes designs for safety suggestions available to designers and constructors.

Besides fall hazards, other hazard types identifiable in BIM are congestion hazards. Choi et al. (2014) have “suggested a framework for work-space planning process in a 4D BIM environment that integrates the characteristics of activity, workspace, and construction plan”. Teo et al. (2016) have developed another hazard identification tool, called the intelligent productivity and safety system (IPASS), which allows users to analyse and monitor key aspects of the safety performance of the project before the project starts and as the project progresses.

BIM can also be used for providing designers with suggestions on how to alter their designs to make them safer to construct. One such research was presented in the papers by Qi et al. (2011, 2013) who have developed a Prevention through Design (PTD) tool that automatically checks imported BIM models and alerts users on how construction safety can be improved. Kamardeen (2010) proposed a 3D+safety model analysis approach, which would enable designers to understand the safety consequences of their design decisions by automatically detecting and flagging potential safety risks.

Temporary structures such as scaffolds, shoring, falsework, temporary stair towers and cranes all have significant impact on construction safety bur are rarely modelled in BIM. To address this issue, Kim and Cho (2015) attempted to automatically generate required temporary structures and analyse their associated safety risk. Kim and Ahn (2011) established a procedure for assessing possible hazards by visually presenting falsework objects and their locations. Their prototype is mainly focused on designing scaffolds but could be developed to include more types of temporary facilities.

A number of commercial software packages exist for leveraging potentials of BIM during the construction phase. The potential applications include using BIM for quantity take-off, scheduling, visualizing construction sequence, budgeting, logistics, resource planning and progress monitoring. Implementation of BIM during the construction phase has proven benefits in all the mentioned potential applications, so the logical step is to expand the application to include construction site safety. Most applications for including H&S in BIM are still in the research phase, either conceptual or as a prototype, and are not widely used. Existing commercial software tools, on the other hand, have only limited application in the field of construction safety.

BIM is also often used as a starting point for other innovative technologies or used in cooperation with other technologies. Table 2 lists all such research where BIM is used in combination with other technologies to improve construction H&S.

BIM can be used in combination with other technologies to enable different types of hazard identification. When used in combination with wireless sensors, BIM can be used to improve monitoring of confined spaces on construction sites. Riaz et al. (2014 , (2015) have combined BIM with a wireless sensor network to remotely monitor temperature and oxygen level values. Another combination of BIM and sensors used for identifying construction hazards was developed by Kim et al. (2016) who proposed an automated hazardous area identification model based on the deviation between the optimal and actual routes. The authors hypothesize that deviation from the optimal route suggests that a hazard is present on the route and that is why the worker deviated from the calculated optimal route.

Wang et al. (2015) have combined BIM with laser scanning and point cloud generation to identify fall and cave-in hazards. The presented method automatically identifies these two hazard types and visualizes the pit, the hazards, and the required protective safety equipment in a BIM model. Database is also one of the technologies that can be integrated with BIM for the purpose of hazard identification. In her thesis, Zhang (2014) proposes automated identification of two hazard types, fall hazards and congestion hazards. The first type is fall hazards, which were already discussed in the previous section. The second type is congestion hazards. The research proposes using Global Positioning System (GPS) to gather location data of workers conducting activities on different element types to determine the workspace’s shape and size for each of the activity for each building element. Knowing what area the workers occupy while conducting their activities can be used to model occupancy of the construction site. Connecting this information to the construction schedule can provide temporal and spatial locations of workers, which could be used to identify workplace congestion.

BIM can also be used to identify hazards through JHA. Two such sets of research were discovered whose goal was to identify hazards faced by workers on the construction site by integrating BIM/3D CAD with a database. Given that performing JHA is generally time consuming and labour intensive and that the construction schedule often changes, Zhang has developed an application to automate the JHA process by integrating the JHA database with BIM models (Zhang 2014 , Zhang et al. 2015a). This was achieved by developing the construction safety ontology whose purpose was not only to formalize current construction safety knowledge but also to support hazard identification and mitigation through BIM.

The second set of research, the construction hazard assessment with spatial and temporal exposure (CHASTE), has adopted a different conceptual approach to overcome the problems of hazard identification in construction. It is a conceptual model that enables forecasting safety risks at appropriate levels of detail and reliability in construction or in any other project-based production system with dynamic environments (Rozenfeld et al. 2009). The fundamental difference is that accidents are replaced with loss-of-control events and the potential for any worker to be exposed by them (Rozenfeld et al. 2010). Another difference from earlier research is that CHASTE explicitly accounts for the fact that construction workers are frequently endangered by workers performing other activities (Rozenfeld et al. 2009).

BIM can also be used as a foundation for other technologies: for producing 3D building models, which are then used in VR, AR, AV, or in serious game technologies. These technologies are used mostly for safety training and for visualizing construction sequence.

Safety training in general aims to improve the capability of hazard recognition. To this end, Chen et al. (2014) have developed the system for augmented virtuality environment safety (SAVES). SAVES is an application that integrates BIM models with 2D images to interactively train construction workers on hazard identification and safe working procedures (Chen et al. 2013).

Another innovative technology used for safety training is VR. Sacks et al. (2013) define VR as “a technology that uses computers, software and peripheral hardware to generate a simulated environment for its user”. They also tested the hypothesis that safety training using innovative methods (in their case VR) would be feasible, more effective and would result in higher recall in identifying and assessing construction safety risks than equivalent training using conventional safety training methods.

Safety training can also be conducted through the use of serious games. Zhang and Issa (2015) define serious games as video games that are focused on supporting activities such as education, training, health, advertising, or social change. They proposed a method for gathering behaviour of players in an emergency evacuation scenario. Lin et al. (2011) have also used serious games for safety education. Their game, safety inspector, aims to provide a comprehensive safety training environment in which students assume the roles of safety inspectors and walk through the virtual construction site to identify potential hazards.

Previous two subsections have described research in H&S, which includes the use of BIM. There are of course other innovative technologies that among others include 4D CAD, GIS, databases and knowledge-based systems. A full list of innovative technologies’ H&S-related research identified in this literature review is given in Table 3.

Technologies other than BIM can also be used for hazard identification. A technology conceptually similar to BIM but focused on the surroundings of the building element is GIS. An example of using GIS for hazard identification is presented in a paper by Bansal (2011). In the paper, he discusses “the use of GIS in the development of safety database from which safety information are retrieved and linked with the activities of the schedule or components of a building model”. The motivation of including GIS in H&S research lies in the fact that BIM and 4D CAD cannot include all factors that influence construction safety. He therefore proposes that including geospatial analysis capabilities of GIS on a single platform with a 3D model and 4D scheduling may help in effective safety planning process.

Another technology used for hazard identification, which is even more similar to BIM, is 4D CAD. Both include 3D visual representations of building elements, and construction schedules can be connected with model elements in both technology types, but elements in 4D CAD lack additional parametric information. Mallasi (2006) has used 4D CAD to identify and minimize workspace congestion between construction activities’ execution space. In his research, he developed a methodology and a tool to assist planners with the assignment of activities’ execution space, as well as identification and visualization of workspace congestion.

Benjaoran and Bhokha (2010) have also used 4D CAD as a starting technology, and they developed a rule-based system that analyses design information to automatically detect falling hazards and to indicate necessary safety measures. This research, unlike other fall hazard-related research, does not use BIM but rather 4D CAD. The model elements are traditional 3D objects combined with a construction schedule. A paper by Chan King et al. (2012) also describes a tool that utilizes 4D CAD for hazard identification. They have developed a multidimensional simulation tool called construction virtual prototyping (CVP). This tool utilizes 3D models and simulation of worker behaviour to model and visualize the construction process, thus allowing users to simulate construction processes in a 3D environment before construction takes place.

All research efforts previously described in this subsection have used a type of 3D model in their hazard identification process. Zolfagharian et al. (2014) have shown that 3D models are not necessary for hazard identification. Their research proposed an automated safety planning plug-in (ASPP), a plug-in for scheduling software applications whose objective is to mitigate the occurrence of construction accidents. The tool extracts construction activities from computer-based schedules, integrates construction schedules with construction safety databases (construction safety hazard database, risk assessment database and safety measure database) and identifies risk levels, safety regulations and hazards related to each of the activities.

A subset of hazard identification research is the sensing and warning technologies. In essence, their goal is to eliminate collision hazards and identify them before they occur. There are several different types of technologies used for identifying collision hazards. They include RFID, GPS, radar systems, sonar systems, cameras and infrared cameras. Although a large number of technologies are available, the industry is still reluctant to implement proximity warning systems due to a lack of field testing and research to determine the effectiveness of the systems, poor reliability due to possibly high false alarm rates and poor reliability and high maintenance requirements due to the harsh environment (Ruff 2007).

Teizer et al. (2010) have presented the development of such a sensing and warning system in their paper. The objective of their research was to increase situational awareness and safety in construction activities, which include construction machinery by utilizing an autonomous wireless proactive real-time warning device. The technology alerts both workers and equipment operators when the proximity between the antenna (installed on the equipment) and the receiver (carried by workers) gets too close.

A conceptually different method is presented by Zhu et al. (2016). Their research proposed the use of Kalman filters to predict movements of workers and mobile equipment on construction sites using multiple cameras as input. An advantage of this method is that it does not require installing remote sensors on either equipment or workers, which incurs additional costs and in some cases is impossible since the equipment cannot be physically tagged.

Some technologies integrate more than one application. One of such combinations is the use of innovative technologies for hazard identification and safety training. This combination is expected, since safety training most often includes training construction site workers to identify hazards.

Hadikusumo and Rowlinson (2002, 2004) have developed a design-for-safety-process (DFSP) tool. The tool aims to accomplish three tasks: capture safety knowledge from safety engineers, assist safety engineers in identifying construction hazards and train students and inexperienced safety engineers (Hadikusumo and Rowlinson 2004). DFSP tool is therefore first used as an innovative method to capture the knowledge of safety engineers in identifying safety hazards at construction sites and accident precautions. The captured knowledge is then entered in the DFSP database, which is one of the three components of the tool, the other two being virtually real construction model and VR functions (Hadikusumo and Rowlinson 2002). Integration of these components helps to identify safety hazards that are produced at the design stage by enabling the user to do a walk-through in the virtually real project. The DFSP tool can also list possible hazards and accident precautions for selected model component (Hadikusumo and Rowlinson 2002).

Virtual prototyping (VP) is another technology that can be used in hazard identification and safety training. Guo et al. (2013) define VP as a “computer aided design and manufacturing process with construction of digital product models (virtual prototypes) and realistic graphical simulations.” They have developed a conceptual framework of adopting VP technology to aid in construction safety management. The framework consists of three components: modelling and simulation, identification of unsafe factors and safety training. A real-life case study presented in the paper demonstrated that VP can provide a virtual experimentation platform for identifying safety problems in construction processes (Guo et al. 2013).

Following the presentation of previous research into innovative ITs in construction H&S, this subsection contains a summary of the research based on the number of appearances of each of the innovative technologies, each of the applications and the number of papers per journal or conference.

It is not surprising that the most often cited technology is BIM, which has 41 occurrences. It is followed by the database, which however has only eight occurrences. The third most represented technology is AR (five occurrences), closely followed by RFID, knowledge-based systems, 4D CAD and sensing technologies, all of which have four occurrences. Furthermore, ontology and wireless sensors have three occurrences, while wireless sensors, AV, GIS, risk assessment software, online tools and safety management tools have two occurrences each. All remaining technologies (laser scanning, NLP, conceptual framework, management model, model for quantitative evaluation of safety levels, prompt words, virtual prototyping, web-based system and hazard index heat map) appear only once in the literature review. All the technologies with more than one occurrence are shown in Figure 2.

Fig. 2

Similar disproportion of the number of applications is evident in the numbers of occurrence for each of the applications, which is presented in Figure 3. Hazard identification is not surprisingly the most often found technology in the literature review (30 occurrences), due to the importance of the issue and due to a specific search parameter, which among others included hazard identification. The second most represented application is design for safety suggestions (11), followed by rule checking and safety training (both with nine occurrences). The remaining applications with a significant number of occurrences are H&S issues with temporary structures (seven), JHA (six), sensing and warning (four), visualization (four), education (two) and decision support system (two). The remaining applications are only featured once in the reviewed literature. These applications include determining safety index, safety risk drivers, automated information retrieval, design evaluation, collecting data related to evacuation performance, data extraction and structuring, risk classification, monitoring and assessing construction H&S performance, preventing struck-by hazards, quantifying safety risk for activities, quantifying safety risk for building elements and safety planning.

Fig. 3

Finally, it might be interesting to see which journal or conference proceedings contained the most papers that describe the use of innovative ITs in the field of construction H&S. Most represented journal is Automation in Construction with 17 papers, followed by the Journal of Construction Engineering and Management (eight papers) and Computing in Civil Engineering conference (seven papers). Additionally, the review process identified three papers in Safety Science and two papers each in Journal of Computing in Civil Engineering and Computing in Civil and Building Engineering. A total of 21 journals or conferences appeared only once in the reviewed literature. All journals and conferences that featured more than once are presented in Figure 4.

Fig. 4