This is the authors' manuscript. The final version is available at: http://doi.org/10.1016/j.autcon.2018.01.001 An Internet of Things-enabled BIM platform for on-site assembly 1 services in prefabricated construction 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Clyde Zhengdao Li a; Fan Xue b; Xiao Li c; Jingke Hong d; Geoffrey Qiping Shen e a Assistant Professor, College of Civil Engineering, Shenzhen University, Shenzhen, China. Email: clydelee718@gmail.com b Research Assistant Professor, Department of Real Estate and Construction, Faculty of Architecture, The University of Hong Kong, Hong Kong. Email: xuef@hku.hk c Ph.D. Candidate, Department of Building and Real Estate, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong. Email: shell.x.li@connect.polyu.hk d Research professor, School of Construction Management and Real Estate, Chongqing University, Chongqing, China. Email: hongjingke@cqu.edu.cn e * Corresponding author, Chair Professor of Construction Management, Department of Building and Real Estate, Faculty of Construction and Environment, The Hong Kong Polytechnic University, Hung Hom, Hong Kong. Telephone: +852 2766 5817. Email: bsqpshen@polyu.edu.hk 24 Abstract: 25 Building Information Modelling (BIM) serves as a useful tool in facilitating the on-site 26 assembly services (OAS) of prefabricated construction for its benefits of powerful 27 management of physical and functional digital presentations. However, the benefits of 28 using BIM in the OAS of prefabricated construction cannot be cultivated with an 29 incomplete, inaccurate, and untimely data exchange and lack of real-time visibility and 30 traceability. To deal with these challenges, an Internet of Things (IoT)-enabled platform 31 is designed by integrating IoT and BIM for prefabricated public housing projects in 32 Hong Kong. The demands of the stakeholders were analysed; then smart construction 33 objects (SCOs) and smart gateway are defined and designed to collect real-time data 34 throughout the working processes of on-site assembly of prefabricated construction 35 using the radio frequency identification (RFID) technology. The captured data is 36 uploaded to cloud in real-time to process and analyse for decision support purposes for 37 the involved site managers and workers. Visibility and traceability functions are 38 developed with BIM and virtual reality (VR) technologies, through which managers 39 can supervise the construction progress and approximate cost information in a real-time 1 40 manner. The IoT-enabled platform can provide various decision support tools and 41 services to different stakeholders, for improving the efficiency and effectiveness of 42 daily operations, decision making, collaboration, and supervision throughout on-site 43 assembly processes of prefabricated construction. 44 45 Keywords: Internet of things; BIM; on-site assembly services; prefabricated 46 construction; decision support system 47 1. Introduction 48 Prefabrication has been widely adopted by Hong Kong Housing Authority (HKHA), 49 who is the main provider of public housing in Hong Kong, for its public housing 50 projects, due to its more efficient, cleaner and safer working environment, and better 51 quality (Tam et al. 2007; Hong et al. 2016; Li et al. 2016). For example, the public 52 housing project at Tuen Mun Area 54 Site 2, Phases 1 & 2, makes use of 11 types of 53 precast elements, including precast façade, semi-precast slab, volumetric precast 54 bathroom, tie beam, staircase, parapet, refuse chute, half landing, water meter room, lift 55 machine room and main roof slab. Some of them are proposed by the general contractor. 56 As a ‘sweet point’ of balancing construction cost and labour requirement (Tam et al., 57 2002), contractors adopt a 6-day cycle for the typical floor (usually 20 to 30 units) on- 58 site assembly in high-rise public housing projects since 1990s in Hong Kong. Among 59 the processes of on-site assembly, BIM serves as a useful platform for facilitating the 60 on-site assembly services (OAS) of prefabricated construction for its benefits of 61 providing collaborative working teams and decision makers with the physical and 62 functional representations of prefabricated components (Sacks et al. 2010; Frédéric et 63 al. 2014; Chen et al. 2015; Niu et al. 2016). For example, the status of prefabrication 64 components could be traced and visualized in BIM platform for supporting the progress 65 control (Ergen et al. 2007, Zhong et al. 2015). 2 66 However, the well-formatted information of prefabricated component at the right time 67 in the right location is still insufficient to further raise the efficiency of collaborative 68 working and decision making in on-site assembly services when adopting BIM in 69 prefabricated construction projects (Yin et al. 2009). For example, the location 70 information of both outdoor and indoor resources through positioning technologies 71 such as RFID (radio frequency identification), UWB (ultra-wideband), and GPS (global 72 positioning systems) have been synchronized in BIM for safety management (Fang et 73 al., 2016), while few studies integrate the accurate location information of on-delivering 74 prefabricated components into BIM platform for monitoring the right components to be 75 assembled in the correct position in a safer manner (Zhong et al. 2017). Additionally, 76 the information of changes, cost and schedule are delivered from previous processes 77 (i.e., design, manufacturing, logistics) could be updated to a centralized BIM platform 78 for sharing the information among different stakeholders (Li et al. 2017; Niu et al. 2017; 79 Issa et al. 2017). However, this information is usually re-entered incompletely, 80 inaccurately and untimely into various isolated systems (i.e., enterprise resource 81 planning (ERP)) of the different stakeholders in most of the current project practices, 82 which could not efficiently support the decision making in the OAS (Pang et al. 2015). 83 These problems can be further deteriorated in Hong Kong particularly due to the 84 numerous constraints such as limited resources and space (Wong et al. 2003; Chun et 85 al. 2009; Alavi et al. 2016).The solution for such situation is still a void to be filled. 86 To handle these challenges, an Internet of Things (IoT) enabled platform is to be 87 developed in this research by deploying BIM as the basic infrastructure underlying in 88 its system structure. This research employed a typical design science research 89 methodology (Peffers et al. 2007), which consists of six steps of problem identification 90 and motivation, definition of the objectives for a solution, design and development, 3 91 demonstration, evaluation, and communication, in the research and development. 92 Section 2 is the literature review which also identified the need of BIM and IoT-based 93 OAS management system. Section 3 describes the objectives of the BIM and IoT-based 94 OAS regarding field interviews, the design of SCOs, and the development of the OAS 95 decision support system. The demonstration of the system on a real project and the 96 evaluation are given in Section 4. Conclusions appear in Section 5. The specific 97 objectives of this research are: (1) To investigate and analyse business process and 98 requirement of on-site assembly of prefabricated construction; (2) To propose the 99 architecture design and develop the Internet of Things enabled platform; (3) to apply 100 the developed platform to practical project to test its performance and effectiveness. 101 This centralized BIM platform not only integrates the information delivered from the 102 previous stages but also synchronizes the location information of prefabricated 103 components for facilitating the real-time communication and coordination among the 104 different stakeholders for better decision making in the OAS. The innovativeness of 105 this platform, by looking at whole processes of the on-site assembly of prefabricated 106 construction, is to increase their connectedness by using BIM as an information hub to 107 connect information and communication technology (ICT) enhanced SCOs. The 108 architecture of the IoT-enabled platform has considered the business processes, the 109 stakeholders, the information flow, the visibility and traceability of the real-time data. 110 It uses the service-oriented open architecture as a key innovation to enable the platform 111 as a service. Given its potential to manage building information throughout processes 112 of OAS, the IoT-enabled platform is considered as a significant component of the 113 HKHA’s overall ICT architecture and systems, which aims to re-engineer the OAS of 114 prefabricated construction in Hong Kong for a better support of decision making. 4 115 2. Literature review 116 The advanced OAS planning and control systems initiated from the Last Planner® 117 System (LPS®) which is a production management system that applies pull and look- 118 ahead planning to remove constraints and make downstream activities ready (Ballard, 119 2000). Weekly work planning is adopted to reduce uncertainty and find relevant causes 120 for variances. LPS also uses the percentage of the plan completed (PPC) to measure 121 and monitor the process (Ballard, 2000; Kim et al., 2014). However, LPS is difficult to 122 visualize the flow of work process (Sacks et al., 2009). Building Information Modelling 123 (BIM) can be utilized to simulate and visualize the construction process with 3D 124 geometric models and ample information to facilitate communication among 125 stakeholders (Sacks et al., 2009). In addition, LPS is the weekly work planning that may 126 lead to a long response time to address daily constraints. Sacks et al. (2010) developed 127 the KanBIM concept which can manage day-to-day status feedback and support human 128 decision making or negotiation among stakeholders. As prefabricated construction 129 contains multiple phases from manufacturing, logistics to site assembly, the direct use 130 of LPS and BIM in prefabricated construction has an apparent gap related to the 131 interoperability and real-time traceability of information. Dave et al. (2016) therefore 132 developed a communication framework by adopting IoT (Internet of Things) to 133 strengthen the use of Lean Construction management and tracking technologies such as 134 RFID and GPS, which are critical components of IoT, to track the status of workers, 135 materials, and equipment in the whole process. A conventional RFID system contains 136 an antenna, a transceiver (RFID reader) and a transponder (Radio Frequency tag). The 137 antenna sets up an electromagnetic area where the tag detects the activation signal and 138 responds by transmitting the stored data from its memory through radio frequency 139 waves (Wang et al., 2016). RFID can be applied to monitor unit status during 5 140 manufacturing and site assembly stages while GPS can be adopted to locate the units 141 during logistics phase and calculate the remaining time to site. One RFID-enabled BIM 142 platform has been developed for prefabricated construction by researchers in Hong 143 Kong (Zhong et al., 2015; Li et al., 2016). The platform’s architecture has three 144 dimensions: infrastructure as a service (IaaS), platform as a service (PaaS) and software 145 as a service (SaaS). The IaaS level contains hardware and software layers. The hardware 146 layer consists of the SCOs (Niu et al., 2015) and the Gateway, while the software layer 147 involves a Gateway Operating System (GOS) to manage the SCOs. SCOs with 148 functional data and data collection devices are enabled by the RFID system and other 149 innovative technologies. RFID was firstly introduced as a sister technology to replace 150 barcode system for identifying items. By comparing it with barcode system and 151 magnetic strip system, RFID can store a relatively large number of data. This data can 152 be encrypted to increase data security. It is possible to read data from multiple tags in 153 one time thus increase the efficiency of data processing. In comparison with barcode or 154 magnetic system, no direct contact between a RFID reader and the tagged items is 155 needed as it uses radio wave for data transmission. In addition to reading data, it is 156 possible to write data back to the RFID tag, which greatly increases the interaction 157 between items, systems, and people. The GOS is developed to aggregate and pre- 158 process the massive real-time data such as Industry Foundation Classes (IFC) data 159 converted from BIM software (e.g. Revit), GPS data, RFID data (e.g. schedule, cost, 160 production attributions) and point cloud data. In addition, the PaaS level is related to 161 the data source management services (DSMS) which facilitate the heterogeneous 162 information and application systems by applying XML/JSON-based BIM model and 163 connecting the backend RFID system with BIM model. This enhances the initial BIM 164 platform to a multi-dimensional one. The SaaS level consists of three management 6 165 services (manufacturing, logistics, and on-site assembly) to enhance the information 166 sharing and communication for stakeholders’ decision-making at different stages. This 167 study details the deployment and application of the on-site assembly services to try to 168 improve the dilemmas of current project practices in Hong Kong including: (1) 169 Construction sites in Hong Kong are often compacted, with only limited space for 170 storing large and cumbersome components (Jaillon and Poon 2009). Thus, site 171 management is often on the critical path for the success or failure of a construction 172 project. Under this circumstance, a Just-In-Time (JIT) delivery and assembly are 173 desired but currently in Hong Kong, normally a site manager should reserve 174 components/materials of 1.5 stores on site as a buffer. The JIT delivery of prefabrication 175 components is yet to be harvested; (2) Verification of the components is inefficient 176 (Demiralp, Guven et al. 2012), mainly due to the wide use of paper or paint labels. 177 Workers should pay attention to the verification process sequentially, which will lead 178 to extra labor and time cost. Yet, the accuracy of the verification process is not 179 guaranteed since the paper-based documents, or even handwriting and modified labels 180 are usually ambiguous; (3) Current practice may cause safety issues. Construction 181 workers on the sites are usually busying with their operations, some of which need 182 enough space e.g. for crane towers to hoist various components to proper positions 183 (Mao, Shen et al. 2015). If the required spaces are occupied, serious safety issues may 184 be occurred; (4) If too many components are placed on a construction site, workers may 185 have difficulties to find out proper components (from a large pile of similar components) 186 for a specific trade (Shin, Chin et al. 2011). This has been reported in casino projects in 187 Macau. To identify the proper components through effective real-time information 188 collection approach is highly desired. Currently, no such platform, like IKEA’s 189 “assembly instructions”, has been developed to guide on-site assembly to make it more 7 190 efficient. This research is highly motivated to develop such platform that can inherit 191 information from prefabrication production and cross-border logistics and used it to 192 facilitate the on-site assembly process. 193 In order to delimit the bountry of application and process in this study, the scope of the 194 on-site assembly of prefabrication components phase, is described as follows: (1) this 195 phase beings when the prefabrication components arrive at the construction site and are 196 checked by the on-site foreman after being delivered by the third-party logistics 197 company; (2) the inputs are the delivery of prefabrication components and relevant 198 documentation; (3) this phase concludes when the delivered prefabrication components 199 are assembled and pass their respective inspections; and (4) the outputs are the 200 completion of the superstructure work. General steps for on-site assembly are as shown 201 in Figure 1. 8 b a. On-site storage a 202 203 d b. Lifting c. Fixation d. Jointing c e. Casting e Figure 1 General steps for on-site assembly 204 205 3. Architecture design and development of Internet of Things enabled platform 206 3.1 Analysis of business process and requirement 207 The purpose of this section is to analyse the business processes, identify business needs 208 and requirements regarding to on-site assembly. Through an interview with the 209 Qualified Engineer on 9th July 2014, this section summarizes the key information and 210 analysis results from the on-site assembly for solution design of the proposed platform. 211 The purpose of the business process analysis (BPA) is to map the processes of on-site 212 assembly of prefabricated construction and identify the requirements of major 213 stakeholders involved in these processes. These stakeholders include the client, the 214 main contractor and their sub-contractors. Relevant findings can provide useful 215 information for the system design of the IoT-enabled platform. 9 216 The surveyed Tuen Mun project (Area 54) proposes to build five 34-38 story buildings, 217 providing about 5,000 units and with the expectation of holding more than 14,000 218 people. Detailed information regarding prefabrication components to be used can be 219 seen in the Table1. Figure 2 below provides a typical layout of the use of prefabrication 220 components. 221 222 Table 1 Summary of the use of prefabrication components in the surveyed project 223 10 224 225 Figure 2 Layout of the typical floor of a typical block in the surveyed project 226 227 The related business processes are described in Figures 3. Figure 3 presents the major 228 on-site installation process of prefabrication components. All activities in the process 229 are (expected to be) carried out at Tuen Mun project site. The typical workflow of 230 erection of prefabricated elements onto a residential construction are also investigated. 231 Activities in Figure 3 are carried out within one typical floor (1/F or up). 11 Start Site Establishment Submissions before commencement of building works G/F, including precast water tanks Timber mock-up / study mock-up G/F–1/F slab, including transfer structure Escalating Tower Crane Typical erections of prefab. On one floor (Fig. 3) Y Floor No.= 12,17, 22, 27, or 32? Architectural works N N Building services installations Temporary works All floors finished? Y RC Superstructure works MR/F, LMR/F, UR/F, & TR/F, including booster bump, emergency generator rooms, lift machine rooms, water tanks External Finishing Works Completion End 232 233 234 Figure 3 Work flow of a typical residential construction with prefabricated element 12 235 As shown in Figure 3, the on-site assembly phase can generally be divided into five 236 main stages, namely site establishment, temporary works, superstructure works, 237 architectural works and building services installations. Stages 2 to 5 can be carried out 238 concurrently, which may not be on the same floor though, in the schedule. 239 Prefabrication assembly is most relevant in the third stage (i.e. superstructure works). 240 Stage 1 – Site Establishment: The objectives of site establishment are: (1) to provide 241 maximum security to the plant, materials and the installation works; (2) to protect the 242 public and the environment from the installation works; (3) to provide adequate 243 facilities to both the clients and the contractors’ staff; (4) to ensure that upon completion 244 of the project, the site is efficiently demobilized and reinstated to project stakeholders’ 245 satisfaction. Procedures of site establishment include: protection to existing structures; 246 establish boundaries; remove materials and items; establish accommodation; and 247 establish services. 248 Stage 2 – Temporary works: Temporary works in the project mainly include (1) tower 249 crane erection; (2) material hoist erection and (3) passenger hoist erection. 250 Stage 3 – Superstructure works: This stage is the main focus for the BPA on 251 prefabrication on-site assembly. Superstructure works include (1) fabrication and 252 installation of precast water tank; (2) G/F - 1/F including transfer structures; (3) 1/F – 253 37/F slab, noted that tower crane is planned to be escalated once the slabs of 12nd, 17th, 254 22nd, 27th and 32nd floor are completed; (4) installation of booster pump, emergency 255 generator rooms, lift machine room and water tanks. The process for one floor (from 256 G/F to 5/F) is significantly longer than that for a floor above 5/F, e.g., 12-70 days for a 257 floor from G/F to 5/F while a 6-day cycle for an upper floor. This is because many 258 issues may be encountered during this period, based on experience. 259 Stage 4 – Architectural works: Architectural works mainly include finishing works at 260 flats, finishing works at common areas and external finishing work. 261 Stage 5 – Building Services Installations: After the completion of superstructure 262 works and architectural works, building services-related facilities will be installed, 263 including: (1) Plumbing and drainage installation; (2) Town Gas; (3) Electrical 264 installation; (4) Lift installation and fire services installation. 13 265 Typical installation of prefabricated elements involved different participants and 266 locations. Two main locations are (a) the buffer, which is usually near the target 267 building/wing for the convenience of the tower crane, and (b) erection at construction 268 site. Usually the foremen will coordinate the scheduled actions. An operator at the 269 buffer will check the prefabricated elements after they have been unloaded. If there are 270 flaws or defects after delivery, the prefabrication manufacturer will be contacted for 271 further actions; and relevant information shall be recorded. Two groups of prefabricated 272 elements are delivered and erected in turn. One is the vertical components, which 273 include facades, toilets, partition walls, refuse chute, and water cabinets; the other is 274 the horizontal ones which include slabs and staircase. Thereafter, the prefabricated 275 elements are lifted for erection by tower cranes. In a typical erection, a prefabrication 276 element is adjusted horizontally then vertically. Reinforcement is carried out later, 277 followed by inspection. The time required to complete the installation for one typical 278 floor is six days, and this six-day cycle is widely adopted by contractors engaged in 279 public housing construction works. The findings from business process analysis 280 provide necessary information for the system design of the IoT-enabled platform in the 281 upcoming stages of this research. Based on the identified findings and observations on 282 the process flow of on-site assembly, the requirement analysis on this phase is listed in 283 Table 2. 284 Table 2 Requirements analysis of on-site assembly 285 NO Type Requirement Priority Functional Requirements System needs to keep a record of pending 1 Production prefabricated elements (with or without ID) for orders current working day, and next days for one floor Preferred (e.g., in a 6-day cycle) 2 Buffer Be aware of prefabrication are safely delivered 14 Must Have 3 4 Erection inspection Buffer Buffer & 5 Erection inspection Be aware of prefabrication are erected successfully Be aware of place where prefabricated components are held Must Have Optional When RFID tag is missing or not working, the delivery and/ or erection can be input by alternative Must ways (e.g., querying tag ID from RFID service Have provider followed by a manual input) Automatic SMS or Android/iOS notifications on 6 Messaging prefabrication delivery/ erection/ unexpected issues Optional for stakeholders 7 8 Erection Multiple scanners or floor partitioning for RFID inspection scanning Erection Random order RFID scanning within one floor after inspection inspection Erection 9 inspection & Buffer Erection 10 inspection & Buffer Batch upload of photos synchronized or synchronized with hand-held scan data upload Optional Optional Optional Able to record operators’ GPS locations of delivery & erection as EXIF in JPG images and automatically extractable as supplementary location Optional info Buffer, Erection 11 inspection, & General Electronic files (PDF) sharing of inspection reports and progress reports Preferred management Non-Functional Requirements 1 Performance Data and status are available at real-time 15 Preferred 2 Availability 3 Security 4 Availability 5 Security 6 Security Accessible through wireless/wired network out of Must office/ site Have One shared input account for one wing/building Preferred Accessible through iOS/Android smart devices (phones/tablets after Jan 2013) Binding PC/Phones’ IP/MAC addresses of stakeholders Digital/ vocal signature of inspector and/or buffer operator Preferred Optional Optional 286 287 The BPA described the processes of on-site assembly of prefabricated construction in 288 details, by focusing on major installation stages and involved stakeholders. It also 289 identified and prioritizes the requirements of major stakeholders involved in the 290 assembly activities. The findings from BPA provide the basis for the design of system 291 architecture of the IoT-enabled platform in the upcoming stages of this research. 292 3.2 Functional requirement and UI design 293 After three rounds of site visits, discussions, and meetings with managers from client 294 and the contractor for construction site, the functional requirement and UI (User 295 Interface) design are raised based on the business processes and requirements analysis 296 which come from real-life pilot companies. The purposes of the functional requirement 297 and UI design of this research include: (1) To introduce the concepts of user and system 298 requirements; (2) To describe functional and non-functional requirements; (3) To 299 explain how software requirements may be organized; (4) To present how the GUIs 300 will be designed; (5) To identify the key components of the IoT-enabled platform; (6) 301 To illustrate the specific functions to the programmers how to carry out detailed design 302 and programming; (7) To descript how the modules could assist end-users for 303 facilitating their operations and decision-making. This section provides the specific 304 requirements of the on-site assembly service including external interface requirements, 16 305 functional requirements, non-functional requirements, internal requirements, design 306 constraints, logical database requirements and other requirements. 307 3.2.1 Interface requirements 308 As shown in Table 3, there will be five groups of target human users and three groups 309 of external software users for OAS, each of which will have its own corresponding user 310 interfaces. All hardware interfaces will be those of the on-site assembly service on top 311 of which it will be running, with due attention should be paid to: (1) CPU usage; (2) 312 Memory usage; (3) Cache file creation; (4) Network communication. Besides, the 313 software interfaces include designated user Applications on Android and modern 314 browsers (e.g., Safari and Chrome) which are compatible with WebGL, HTML 5, and 315 Java Script on Windows/OS X/iOS/UNIX/Linux. Network protocols for systems to 316 communicate include HTTP (and HTTPS), SFTP (Secure File Transfer Protocol), 317 specified XML/JSON (Java Script Object Notation) data management services over 318 SSL (secure sockets layer). 319 320 Table 3 The five sets of target human users and three sets of external software users 321 for OAS NO Target human users and external Security software users level busy; Management level (Senior manager, 1 Engineer): Setting up master plan and pattern of assembly cycle, monitor the easy and quick access; medium concerns more on overall/ abstract/ overall progress and estimated spent. representation level; On-site coordinator (Foreman): 2 Characteristic busy; Confirm tasks for a flat from master plan with consideration of actual progress and existed exceptions, claim 17 medium easy access; building progress and quality centric; new and handled exceptions when necessary. Prefabrication receiver (assigned by 3 Foreman): confirm a component is low safely delivered to construction site. 4 5 6 7 Erection worker: confirm a low component is correctly erected. Inspector: confirm the quality of final assembly in the whole structure. BIM system: providing structure and shape data. medium low-level certificate; hard work; low-level certificate; technical/qualified staff; professional standards; high specific software end (Revit); RFID system: providing status data of components. 3rd-party solution; medium data may be slightly delayed (<1 day); Other services in the platform: the 8 part-time receiving; aforementioned ones and the services to be developed. high compatibility; medium small amount/ regular communication; 322 3.2.2 Functional requirements 323 Based on the analysis of business process and requirement, a total of four major services 324 are provided in the IoT-enable platform: (1) Assembly management (real-time 325 supervision) services: to provide a toolkit for contractor’s managers and engineers to 326 supervise the management of on-site assembly services, which include: visibility 327 service to integrate the project progress in charts and 3D BIM models, and components 328 in 3D BIM models; component tracing service to locate a missing component and return 329 the geolocation or place of storage; component tracking service to filter one or more 330 components in a given criterion, e.g., selecting all installed windows in a storey/floor; 331 (2) Assembly operations services: to provide a toolkit for managers and engineers who 18 332 are involved in operations of assembly at construction site, which include: planning 333 service to break down a job plan (typically floor plan) into tasks in charts; components 334 order listing helper service: to provide an information list for site coordinators and 335 production services, partially depended on component tracking service; assembly 336 scheduling service to make floor plans and daily plans by associating personnel with 337 planned tasks; component tracing/ tracking service; duplicated component tracing and 338 tracking service; (3) Assembly exception handling services: to implement part of 339 preplans for some of the unexpected cases, which include: progress exception handling 340 service to provide follow-up tools for the cases where the progress is not carried out 341 on-time; component exception handling service to record when an important (e.g., 342 RFID tagged) component encounters defects or replacement; (4) Assembly notification 343 services: for facilitating in reminding and notification for users, with different 344 reminders sent to subscribed users, such as progress summary, component list to be 345 delivered today, and summary of inspection result, which include Email notification 346 service to send Email alerts for managers and engineers who work in office, SMS 347 notification service, and mobile app notification service. 348 3.3.2.2 Functions of management tools 349 The functions of management tools for management level listed in Table 4 and Figure 350 4 below. Table 4 The functions of management tools for management level 351 Introduction Trigger Inputs Processing Toolkit for management level As specified in the Graphical User Interfaces (GUIs) in Figure 4. Pre-written graphic/text of introduction; or 3D WebGL/chart component-based graphics with selectable criteria Read the current progress, data via this software from the database; or editing the master plan of project Outputs Return and present the content on web or app Error Show information and hints on data input, user privilege, and 19 Handling software compatibility errors, or return to log in 352 b. Mockup UI for Gantt chart of progress with schedule a. Mockup UI for 3D graphics of project/building/floor progress c. Mockup UI for chart of estimated cost according to progress d. Mockup UI for stock and Geolocations of in-process prefabricated components 353 354 Figure 4 Mockup GUIs for 3D graphics of project/building/floor progress 355 3.3.2.3 Functions of operation tools 356 The functions of management tools for on-site operation level listed in Table 5 and 357 Figure 5 below. 358 Table 5 The functions of management tools for on-site operation level Introduction Toolkit mainly for on-site operation level Trigger As specified in the GUIs in Figure 5. Inputs Pre-written graphic/text of introduction; or 3D WebGL/map component-based graphics with selectable criteria 20 Read/write the necessary information (detailed floor plans, Processing component shape ad status, etc.) and process with this software from/to the database Outputs Return and present the content on web or app Error Show information and hints on data input, user privilege, and Handling software compatibility errors, or return to log in 359 b. Mockup UI for stock and geolocations of in-process prefabricated components for foreman a. Mockup UI for floor plan review for foreman c. Mockup UI for receiving prefabricated components d. Mockup UI for confirmation of inspection of installation 360 361 Figure 5 Mockup GUIs for floor plan review for a foreman 362 363 3.3.2.4 Functions of exception handling and notification tools 364 The functions of management tools for managers, engineers and on-site coordinators o 365 track exceptions and receiving progress/exceptions updates are listed in Table 6 and 366 Figure 6 below. 21 367 368 Table 6 The functions of management tools for managers, engineers and on-site 369 coordinators Introduction Trigger Inputs Processing Toolkit to track exceptions and receiving progress/exceptions updates As specified in the GUIs in Figuge 6. Selectable list of events (exceptions and progress) to notify; exceptions tracking and updating Read/write the pre-defined event information via this software from/to the database Outputs Return and present the content on web or app Error Show information and hints on data input, user privilege, and Handling software compatibility errors, or return to log in 370 371 a. Mockup UI of notifications and exceptions handling for foreman b. Mockup UI of notifications and exceptions handling for management 372 Figure 6 Mockup GUIs for notifications and exceptions handling for foremen and 373 managers 374 3.2.3 Non-functional requirements 375 Non-functional requirements may exist for the following attributes. Often these 376 requirements must be achieved at a system-wide level rather than at a unit level. The 377 requirements are stated in Table 7 in measurable terms. The deployment of the system 378 would was planned on cloud servers, thus many conventional system-level requiments 22 379 (e.g., system downtime and mean time between failure) was easily met. 380 Table 7 Non-functional requirements 381 Response time The maximum response time for the submission of any Performance request will be 1 minute. Capacity The maximum number of recognizable items is limited to 100,000 for each building. Maximum bug rate There will be a maximum of 1 bug in 1,000 lines of codes. Maximum time to repair In case of cloud outage, the site users (type 2-5 in Table 3) Reliability will store the data in the designated smartphone Application to be uploaded when the system is ready; while the service for the mangment user (type 1 in Table 3) will be down. A typical system reboot time takes 10 seconds, and a scheduled cloud maintainance can be a few hours. Back-end internal computers The system shall provide storage of all databases and cache files on a redundant computer and another cloud storage Availability located in a different continent. Operational availability The service shall provide users with a minimum operational availability of 99.9%. Security Security considerations 23 The on-site assembly service will ensure the privacy of user job status and ensure full control over job execution, so that alteration of scheduling criteria or actual resource allocation is not possible without administrator authority. Data transfer (1) The system shall use SSL in all transactions that may include confidential information. (2) The system shall automatically log out all users after a period of inactivity. (3) The system shall confirm all transactions with the user’s smartphone application or web browser. (4) The system shall not leave any cookies on the user’s computer after logging out. Data storage (1) The user’s web browser shall not display a user’s password except for user’s manual request (e.g., on a smart phone). It shall always be echoed with special characters representing typed characters. (2) The system’s back-end services shall store encrypted passwords of users instead of orignal ones. (3) The system’s back-end services shall only be accessible to authenticated administrators. (4) The system’s back-end databases shall be encrypted and accessible to authenticated administrators. Maintenance (1) The administration will not support job migration for the Maintainability purpose of decreasing resource fragmentation. (2) The on-site assembly service shall permit the upgrade of software without down time. 24 (3) The Mean Time To Fix shall not exceed one person day. Naming convention All codes prefer to the Hungarian notion. Portability Ease of moving to another system Can be used on all desktop computers and smart phones with modern borwsers 382 383 3.2.4 Logical database requirements 384 The logical database of on-site assembly service consists of 4 sets of data tables, 385 including: (1) Project and tasks; (2) Imported Data from BIM and RFID systems; (3) 386 Exceptions and handling; (4) Users and notifications. The 4 sets of data tables are 387 supporting the 4 group of services, respectively. Figure 7 shows a detailed composition 388 of the 4 sets of tables. Set a) consists of tables “project”, “masterPlan”, “floorPlan”, and 389 “dailyTypePlan”; 390 “compType”; set c) include “exFPlan” and “exComp”; set d) include “users” and 391 “notifications”. The primary keys and foreign keys can also be found in the figure. set b) includes “BIM_component”, 392 25 “RFID_record”, and masterPlan floorPlan dailyTypePlan PK id PK id PK id FK1 parentId type SN name cname projId managerID estStartDate estEndDate FK1 planId flat floor managerID estStartDate estEndDate FK1 FK2 fplanId compTypeId compNum plannedDate startDate endDate FK3 FK2 FK2 BIM_component PK FK2 FK1 compType PK id project PK name refBimId shapeData sycDate comment id name client designer manufacturer logService contractor estStartDate estEndDate users PK id FK1 title name level password projId notifications 393 PK id FK1 userId eventType internalNoti emailNoti smsNoti appNoti id bimId bimSN bimGUID bimTypeId isPrefab dplanId flat floor block wing shapeData version sycDate comment exComp exFPlan PK id FK1 fpId title content status eventDate createrId trackerId revHistory latestRev latestRevDate FK2 FK3 PK id FK1 compId title content status eventDate createrId trackerId revHistory latestRev latestRevDate FK2 FK3 RFID_record PK FK1 id rfidId rfidTagSN rfidTagGUID bimSN flat floor block wing qcGrage taggingTime qcTime manuTime delivTime recvTime assemTime inspTime comment 394 395 Figure 7 Database model diagram of the logical database of OAS information system. 396 397 The highest amount of data, which is not exceeding 10⁵ records (usually <10⁴ for a 398 building) for a project, is expected to be found in table “BIM_component” and 399 “RFID_record”. However, the attribute “shapeData” in table “BIM_component” can be 400 as large as 10⁴~10⁵ bytes. So the total physical size of the database can be up to 10¹⁰ 401 bytes (10GB) in assumed projects. The number of simultaneous users is expected to be 402 less than 10. 403 3.3 Smart construction object and smart gateway 404 SCOs are typical construction resources such as tools, machinery, materials, which are 405 converted into smart objects through binding them with different RFID devices, as 406 shown in the Figure 8. The purpose of SCOs is to create an intelligent construction 407 environment within the typical prefabrication production sites such as shop-floors, 26 408 warehouse, logistic and supply chain, and construction sites. SCOs are building blocks 409 for such intelligent environment, within which they can sense and interact with each 410 other. Thus, the processes of on-site assembly could be carried out smoothly. 411 Smart Construction Objects a. TSL 1128 Bluetooth UHF RFID Reader b. SAMSUNG Note 3 (SM-N900) c. UHF RFID Tag Platform Gateway a. Connection c. Write function b. Read function 412 413 Figure 8 smart construction objects and gateway 414 415 Typical construction resources are converted into SCOs through various tagging 416 schemes. Firstly, critical prefabrication components such as volumetric kitchens, toilets, 417 precast facades, will be tagged individually. That means item-level tagging scheme is 418 adopted because they easily influence the progress in prefabrication housing 419 construction. For non-critical materials, such as dry walls, and building blocks, tray- 420 level or batch-based tagging scheme is adopted. That means tags are attached to the 421 trays which carry multiple minor prefabrication components. In the pilot study, the 422 RFID tags, as shown in Figure 8, are Ultra High Frequency (UHF) tags protected in 423 strong Acrylonitrile Butadiene Styrene (ABS) plastic cases and validated individually 424 before planting. Each tag supported up to about 30 cm when embedding on the steel 27 425 ribbons inside concrete. A data operability of each was validated before planting. For 426 various workers, such as machine operators, vehicle drivers, logistics operators, and 427 on-site assembly workers, they are tagged with smart staff cards. These construction 428 resources attached with tags are passive SCOs. The deployment of RFID readers 429 follows a systematic approach. Once bound by RFID readers, they become active SCOs 430 that can sense and detect the passive SCOs. Both active and passive SCOs can sense 431 and interact with each other to create an intelligent construction environment. They 432 carry critical information that will be updated at different locations. 433 Gateway performs several key functions in the research. Firstly, it connects and hosts a 434 set of SCOs through wired or wireless communication standards. It not only allows 435 workers/operators to access information such as prefabrication production status, but 436 also defines, configures, and executes the corresponding prefabrication construction 437 agents through various services. Secondly, it communicates and interacts with upper- 438 level decision-making systems through providing useful and real-time information on 439 standardized format. It acts as a bridge between the frontline SCOs and upper-level 440 decision-making systems. For example, the gateway can connect and control RFID 441 readers through Bluetooth, and send data to cloud servers via 4G or WiFi. Bluetooth 442 data transfer can be carried out between the main device and other devices at any time, 443 the main device can select the slave device to access. Especially, it can be in the way to 444 change equipment between fast conversion. This greatly improves the stability of 445 Bluetooth connectivity. Thus, decisions and their executions could be seamlessly 446 synchronized in prefabrication housing production. Thirdly, it processes, caches, and 447 exchanges real-time data and events locally and temporally. To this end, complex event 448 processing technology is used to integrate the construction information into a 449 standardized scheme, which could be understood, shared and used among different EISs 450 in the construction industry. Finally, it provides a rich set of facilities for service 451 definition, configuration and execution. The concept of service-oriented agents is 452 adopted to represent the SCOs through a plug-and-play fashion. 453 The Gateway uses an operating system named GOS to achieve a flexible, modularized 454 and re-configurable framework, where applications and solutions are designed and 28 455 developed as web services. GOS aims to provide an easy-to- deploy, simple-to-use and 456 flexible-to-access solution for the construction industry. Within the GOS, multi-agent 457 based models are used to ensure the versatility and scalability of Gateway. Therefore, 458 communication and interactions between SCOs and other services is facilitated by 459 using an XML/JSON-based message exchanging protocol. 460 SCOs and Gateway can capture the real-time construction data to support the decision- 461 making in client’s enterprise information system. SCOs and Gateway can enhance the 462 data sharing within the high level decision-making entities and front-line construction 463 sites. The advanced decision-makings could real-timely be reflected in the construction 464 site, while, the real-time data such as prefabrication manufacturing progresses, 465 prefabrication transportation statuses could be fed back to stakeholders on real-time 466 basis. SCOs and Gateway can form a closed-loop information interaction throughout 467 the prefabrication housing construction. 468 3.4 Overall architecture design 469 The IoT-enabled platform of on-site assembly services comprises four key components, 470 as shown in Figure 9. They are smart construction objects, platform Gateway, decision 471 support system, and data source services. As shown in Figure 9, from the right to left, 472 SCOs are passive and active construction objects equipped with RFID devices. 473 Gateway connects, manages, and controls the SCOs through defining, configuring, and 474 executing the construction logics. Decision support system is to suit the on-site 475 assembly services in Hong Kong. To enhance the data sharing and interoperability 476 among BIM, stakeholders’ information systems, and the IoT-enabled platform, data 477 source services are designed to use XML/JSON-based data sharing mechanism for this 478 purpose. Under the architecture, the decision-making systems can use the real-time data 479 for advanced decision-makings. 29 SOA-based Data Access Service Agent-based Application Information Service Iots-enabled Platform for On-Site Assembly Services DM AIS UDDI Service Date Source Services MM Management Tools Database Operation tools Smart Construction Objects Exception handling … Decision Support System HOMES Definition Configuration BIM-centered System Execution Platform Gateway 480 481 Figure 9 Overall architecture design of the platform 482 4. Practical application of the on-site assembly platform 483 4.1 Description of case study 484 The Tuen Mun project (Area 54, TM54), initially designed by HKHA, proposes to build 485 five 34-38 storey buildings, providing about 5,000 units and with the expectation of 486 holding more than 14,000 people. The construction practice of the 8th-35th storeys of 487 Block 5 of the Tuen Mun project were provided as case study by our partners, due to 488 project period well meet our study. The period of the pilot study was initially set as 5 489 storeys of Block 5, roughly from early October 2015 to November 2015. The period 490 had later been extended to much more storeys (whole building) of Block 5 till the end 491 of this research. To collect required data, a series of on-site visits and interviews are 492 arranged and conducted toward concerned major stakeholders, including HKHA (Hong 493 Kong Housing Authority) staff members responsible for housing production in the 494 region, managers from precast manufacturers and logistics companies, engineers, and 495 on-site managers of contractors. Besides, engineers who are familiar with the processes 30 496 of on-site assembly activities are trained to operate the developed devices to run the 497 platform for improving the productivity of OAS, and the management data are 498 automatically collected and uploaded to the platform in real time manner. 499 4.2 Operational flow of the platform 500 The developed on-site assembly service facilitates various assembly operations, 501 supervisions and quality checking in the construction site. BIM is integrated into the 502 development of the service to visualize and monitor assembly progress. Several major 503 sub-services, such as on-site assets management service, real-time supervision service, 504 data capturing service and real-time feedback service are exploited to facilitate 505 assembly of precast components. The operational flow of the developed service is 506 shown in Figure 10. 507 Staff Registration Error Alert Order Management Assembly Confirmation and Quality Checking Progress Visualization Real Time Progress Monitoring \ 508 Figure 10 Working Logic of OAS 509 510 511 (1) Staff registration 512 The staff registration function offers on-site workers an efficient way of logging into 31 513 the system – by tap their staff cards instead of wasting time in typing passwords. This 514 service uses NFC (near field communication) technology (and their existing staff cards) 515 to identify corresponding workers, foremen, and on-site managers. Moreover, possible 516 violations of site safety regulations, and risks and dangerous activities, can be mitigated 517 for the operators on site. 518 (2) Order management 519 This module can be used by on-site workers and foremen responsible for the assembly 520 of precast elements and by managers who want to check detailed information on an 521 scheduled order and make necessary distributions, confirmations, and modifications. 522 This function communicates with the order databases at manufacturer and logistics 523 companies. The “Orders” module also includes two sub-modules similar to the 524 manufacturer: “Current Orders” and “Import Orders.” The “Current Orders” module 525 provides the orders overview. Users could check the general information of all imported 526 orders and monitor their real-time status using this module. They could also check order 527 details as well as remove and edit orders using this module. 528 (3) Assembly confirmation and quality checking 529 This function captures the real-time data of the precast element assembly upon site 530 installation in such a manner that allows meaningful and useful information to be 531 extracted. Once the required precast elements arrive at the site, these are assembled 532 onto the floor and are quality checked. Real-time data regarding current status is 533 captured through RFID reader by on-site foremen. This real-time information is then 534 transferred to the server for processing to facilitate and coordinate various stakeholders 535 and support their decision-making on the project, specifically when the project still has 536 some issues, such as delivery delay of precast components, assembly interruption and 537 other quality problems. 538 (4) Real-time progress monitoring 539 The cumulative quantity of precast elements erected based on real-time data collected 540 and the contractor’s master program can be compared using a line chart to identify any 541 delay in site construction progress, as shown in the Figure 11. This service provides a 542 Gantt chart or a 3D virtual reality presentation that uses RFID assembly data to reflect 32 543 the construction progresses in real-time in terms of prefabrication assembly status, 544 material consumptions and workers’ assignments. The main users are HKHA and on- 545 site supervisors responsible for controlling the construction objects and reporting to 546 various stakeholders on the progress, current challenges or barriers. 547 548 549 Figure 11 Function for real-time progress monitoring 550 551 (5) Progress visualization 552 Real-time precast construction progress is visualized using an imported BIM Model in 553 a web-based operating platform for monitoring produced elements, under transportation, 554 on-site arrival and erection, which are shown in different colors to indicate the status of 555 precast elements, as shown in the Figure 12. Easy real-time visualization is applied to 556 check against domestic floor actual site construction progress and identify any delay in 557 precast fabrication and delivery. Therefore, all involved project stakeholders could be 558 aware of the current situations and make associated decisions collaboratively. 559 33 560 Figure 12 Function for Progress visualization 561 562 563 (6) Error alert 564 This function is developed to detect the rightness of the assembly of precast elements. 565 Every precast component has a unique serial number that binds with a specific RFID 566 tag and is assembled at a specific location. Coordinates of the location where the RFID 567 tag of precast element installed read with a mobile phone based on GPS can be 568 compared to the design coordinates based on the BIM model, as shown in the Figure 569 13. The deviation in position can be shown in meters. Any deviation larger than the 570 reasonable tolerance in GPS can be identified manually as an error in precast element 571 installation. Please be noted that because the layout of one typical floor of the studied 572 building is quite large, with about 5,000 m² per floor, the minor deviation of GPS 573 position data will not affect the error alert analysis of precast facades and the real-time 574 data collection. 575 34 576 577 Figure 13 Function for precast component tracing and error alert 578 579 After testing, the main advantages of on-site assembly service can be summarized as 580 (1) Time-saving or man-hour saving, where a typical RFID reading of 23 facades for 581 two wings of typical floor takes about 16 mins. However, the current solution spends 582 more than 30 minutes, that is, about ten man-hours per month. The time can be 583 improved further if the factory performs tag checks before every delivery. Time can be 584 improved even further with an offline item cache. (2) Easy access and timely 585 communication, with real-time feedback from assembly sites, real-time tracing of the 586 construction objects, such as precast components, on-site workers and site equipment, 587 are achieved. Real-time data are also used for forming statistical reports and analysis 588 for the decision making of various involved stakeholders. 589 4.3 Facts on mobile apps and OAS web application 590 4.3.1 OAS RFID data gathering APP 591 The OAS RFID Data Gathering APP reads RFID EPC code via the SCO gateway and 592 uploads the time and location to the server accordingly. Example screenshots of the 35 593 SCO gateway can be found in Figure 8. During the pilot tests, many challenges were 594 engaged and resolved as follows: 595  Using multi-level menu to reduce ambiguity on the system UI. 596  A “waterproof” function was developed to make the smart gateway possible to operate in rains. 597 598  English/Chinese versions are switchable from the configuration panel. 599  Visual clues for scanning targets, including a list of items to read and their designed locations on a mini map. 600 601  brief introduction of each category of SCO. 602 603 Visual clues for tag position for inexperienced user, including typical locations and  An alternative confirmation function by taking photo is designed for handling 604 about 3% incorrect tags (missing or wrong) and less than 1% failed tags (unknown 605 reason). 606 Also, some challenges not addressed yet: 607  The system relies on manual collection (reading) of the data. 608  The location data (GPS) of SCOs is only available in 5 days (before setting up semi-precast slabs overhead). 609 610 611  The location data (GPS) becomes stable after 1 to 2 minutes when an operator climbs to the working roof. 612 4.3.2 WeChat OAS add-in 613 In order to extend the functions of OAS to mobile phones and tablets, a WeChat add- 614 in, or Official Account, was developed as a supplementary APP. The main features of 615 the add-in are: The four most valuable functions, including overall progress and a real- 616 time nD BIM model, production status, just-in-time logistic information, location test 617 of installation, were deployed on WeChat. 618 4.3.3 OAS web application 619 The OAS web application is the main media of use the functions of OAS. It is designed 36 620 on HTML5 for modern browsers, so the technical issues listed below are about using 621 on PC browsers. 622  The nD-BIM model related: (1) At the beginning stage, only façades were 623 displayed on the nD model: Later, the full precast model of Block 5 was provided, 624 and all reinforced concrete items were imported; (2) Model size too large (about 625 8MB for real-time frame). The size caused the slow loading: The WebGL data file 626 was redefined. Concrete belongs to the same family was referred to a data class. 627 The heavy class data file was cached as local storage of browser. In this way, the 628 model size was reduced to about 80KB; (3) 4D play-mode still too large (about 629 9MB) redefined “storey” classes (about 1MB). 630  added in the later versions; (2) Menu items were regrouped by objects or functions. 631 632 Menu: (1) The first version of menu appearance is plain style, CSS animation was  Progress curve: (1) The (Adobe Flash-based) chart is not working on iOS. It was 633 later changed to a Java script version; (2) Clues for days of delays were added. The 634 tip texts were converted from manual comments of master plan. 635  toolbar for history data and full-screen / windowed form. 636 637    A GPS data based location test was implemented for a coarse but automatic ways of location checking. 642 643 A calendar based setup GUI was developed for delivery orders (administration) management. 640 641 RC volumes, important dates and plan revisions of each storey were included in master programs (administration) management. 638 639 2D floor map / nD BIM only display latest frame: Filters were added on to the  Hong Kong holidays and special non-working days such as black rain signal were implemented for an automatic delay summary (administration) comments. 644 645  Factory supply status is now available in both chart and text summary. 646  Google Maps® was used for display of positioning and GPS. 37 647 4.4 Summary of the application 648 By the end of January 2016, the OAS recorded 667 prefabricated items (all facades), 649 from the 8th floor to the 22nd floor (14.5 storeys, 58 wings). Each item has four 650 important time and corresponding geolocations of manufacturing, delivery start, arrival 651 at site, and erection. According the data, a day of “factory supply shortage” was 652 discovered and was verified by Gammon’s independent system. Another unusual 653 installation was detected by GPS location test. The main advantages can be summarized 654 in three categories: 655  Time-saving or man-hour saving: (1) A typical RFID reading of 23 facades (2 656 wings) cost about 16 mins. In contrast, in the current practice a worker spends more 657 than 30 minutes. That is about ten man-hours per month; 658  Easy access and presentation: (1) nD model on many devices, including PC, tablet, mobile, etc; (2) Main functions are accessible on WeChat for drivers and workers 659 660  661 4.5 Scalability testing of the platfrom 662 The purpose of section is to provide information about scalability testing results coming 663 from several tests on the IoT-enabled platform for on-site assembly services of 664 prefabricated construction. Tests have been performed to evaluate software 665 performance scalability and the compatibility to extend to different projects. The results 666 in this document are then the merge of several tests which are carried out in different 667 parts of the platform representing critical phases in the prefabricated housing supply 668 chain. The purpose of this testing was to simulate predetermined scenarios that 669 represent real-world hosting: (1) Determine the impact of server configuration on 670 software performance; (2) Validate the test case scenarios and overall proposed scale 671 environment; (3) Validate extensibility of hosting different construction projects. 672 4.5.1 Overview 673 In a real construction project, the active accounts and their activities are quite limited. Coarse assembly location checking: GPS data can help detect some unusual data 38 674 But in the setting of test, we have assumed ten to twenty times of both users and 675 activities per user more than what we have measured during a 7-month pilot study. To 676 replicate a typical large scale service implementation of our platform system, a series 677 of auto tests were built as shown in Table 8. 678 Table 8 Scalability test deployment 679 Typical Enterprise Scalability Test Customer Deployment Active organizational units account 1-3 25 Activities per active account per minute 1-3 60 Client address lists 1 10 Indicators 680 681 The Test scenarios include: (1) Deploying on a shared web server hosted by university, 682 at maximal level of preset load; (2) Deploying on a dedicated web server hosted by 683 university, at maximal level of preset load; (3) Deploying on a renowned cloud server 684 hosted, at maximal level of preset load; (4) Deploying with 7 active projects. The test 685 environment is as shown in the Table 9. 686 Table 9 Test environment 687 Operating Server Nature Location www.ad.arch.hku.hk Shared server HK Linux 147.8.92.79 Dedicated server HK Linux openshift.com Cloud server USA Linux Client Nature Location Profile 39 System Client 1 Auto test software (By loadimpact.com) 25 users, 60 Brazil activities per user per minute Client 2 Auto test software Singapore ditto Client 3 Auto test software USA ditto 688 689 4.5.2 Scale testing for different scenarios 690 The 25 virtual users (VUs) were added incrementally in 5 minutes, i.e., 1 new active 691 user in 12 seconds. The activities increased from 1. During the 5-minute test, thousands 692 of URLs will be requested by the client and 200M to 1G data will be transferred as 693 well. The response time (fully load of a requested page by an activity) was measured. 694 （1）Scenario 1 - Client 1 + shared server 695 As shown in Figure 14, the load time was not stable and the data transmission was not 696 acceptable for intentional users. Though we found it was acceptable when using in HK 697 locally. 698 699 Figure 14 Scale testing result from scenario 1 700 701 （2）Scenario 2 - Client 1 + shared server 702 In this scenario, the load time was much fluent and stable as shown in the Figure 15. 40 703 704 Figure 15 Scale testing result from scenario 2 705 （3）Scenario 3 - client3 + cloud server (openshift.com) 706 In this scenario, the load time was quite high at the beginning, but soon reduced to very 707 low level about 100ms, as shown in Figure 16. This was because of the cloud server 708 unloads the system when it is idle and loads and initializes the system when there are 709 requests. In general, the cloud server is the best way of deployment of the system. And 710 once being deployed on cloud, the performance will not be a problem any more. 711 712 Figure 16 Scale testing result from scenario 3 713 （4）Scenario 4 - 7 active projects 714 7 active projects that have used the platform to manage prefabrication construction are 715 hosted to check the stability of platform. The system can work smoothly in all the 7 716 active projects, and the cloud server deployment showed the best performance in terms 717 of average response time. 718 41 719 4.6 Technology integrity and scalability 720 The proposed platform is designed and developed under service-oriented open 721 architecture to ensure seamless integration with existing systems (HKHA’s BIM and 722 Housing Construction Management Enterprise System (HOMES)), so that the 723 information among them could be shared and synchronized. The integrity and 724 scalability has been carried out through the following aspect: (1) IoT-enabled platform 725 can be easily deployed through existing commercial cloud space, such as Ali Cloud and 726 Amazon Cloud. High performance servers, smart computing resources sharing and 727 virtualization for integrity and scalability can be easily maintained through the provided 728 infrastructure. Specific options, such as public cloud or private cloud, can also be 729 chosen for special security considerations; (2) IoT-enabled platform considers standard 730 data requirement and supports formats of popular BIM systems (such as Revit). Apart 731 from the pilot research, IoT-enabled platform can therefore be conveniently applied to 732 other construction projects; (3) With the developed SCO, Gateway and GOS, IoT- 733 enabled platform supports heterogeneous smart Auto-ID devices and able to handle 734 different RFID tags (such as NFC and UHF tags); (4) Due to the policy of HKHA for 735 change request of HOMES, the interface to integrate IoT-enabled platform with 736 HOMES is not possible to be made until year 2018. However, the data source 737 interoperability services provided in the platform is initially implemented and tested to 738 create adaptive data exchanging interfaces for HOMES and other related systems. 739 Instead of direct integration with HOMES, the platform also provides set of visibility 740 and traceability tools for monitoring the project progress and cost for HKHA and other 741 stakeholders at this stage. 742 5. Conclusions 743 Over the years, HKHA has taken a leading role in developing and promoting the 744 application of ICT in general and BIM/ERP/RFID among construction stakeholders. 745 The architecture of the IoT-enabled platform has considered the business processes, the 746 stakeholders, the information flow, and the real-time information visibility and 42 747 traceability. It uses the service-oriented open architecture as a key innovation to enable 748 the platform as a service. Given its potential to manage building information throughout 749 processes of OAS, IoT-enabled platform is considered as significant part of the 750 HKHA’s overall ICT architecture, which aims to reengineering the OAS of 751 prefabricated construction in Hong Kong. 752 All the collected real-time information from RFID and GPS can be connected with BIM 753 in the developed IoT-enabled platform. Traceability and visibility of the physical 754 building information, progress, and cost are available for the stakeholder to monitor the 755 whole process and make decisions where necessary. The paper-based records can be 756 subsequently freed for many processes and only reserved for verification in key 757 processes. The usage of BIM technique can also be henceforth extended to construction 758 phase. With the developed platform, the main contractor can be benefitted from 759 knowing the real-time information of prefabrication components. The data collection 760 on site becomes effective, reliable and more value-added. Therefore, the whole on-site 761 team of the main contractor can be more resilient when facing changes, such as design 762 changes, order changes, changes due to repairing defective components, etc. The client, 763 HKHA, can be benefited from obtaining real-time information from the prefabrication 764 production to the on-site assembly. The visibility and traceability tools provide useful 765 tools for monitoring and checking the status and quality problems. The multi- 766 dimensional information of cost and progress provided by IoT-enabled platform, can 767 help the client to manage the progress and arrange payment accordingly. Historical 768 information of the stakeholder’s performance stored in the IoT-enabled platform can 769 even be used for facilitating contractor and sub-contractor selection. 770 Despite the various benefits, the limitations of the developed platform in the research 771 should be also outlined for its further development and broader application. Due to the 772 limitations of resource, this research only applies the developed platform to only one 773 practical project for testing its effectiveness. Besides, this research focus more on the 774 development of the functions related to schedule and cost management, while 775 management of safety, quality and construction environment are also important for 776 prefabricated construction project. Despite of the above limitations, the research not 43 777 only pioneers on developing a platform for on-site assembly services of prefabricated 778 construction with integration of Internet of Things and BIM from a new perspective, 779 but also serving as a solid basis for further research, which may include: improving and 780 extending the applicability of the platform to more practical project to enhance its 781 effectiveness; improving the platform by incorporating more functions related to the 782 management of safety, quality and construction environment. 783 Acknowledgments 784 The authors would like to thank the HKSAR ITC/LSCM R&D Centre for funding this 785 research through the Innovation and Technology Support Programme (Project 786 Reference: ITP/045/13LP). The Research Team is also grateful to the Hong Kong 787 Housing Authority and its contractors for supporting and participating in this research 788 via real project applications. 789 790 Reference: 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 Alavi, A. H., H. Hasni, N. Lajnef and K. Chatti (2016). "Continuous health monitoring of pavement systems using smart sensing technology." Construction and Building Materials 114: 719-736. Chen, K., W. S. Lu, Y. Peng, S. Rowlinson and G. Q. Huang (2015). "Bridging BIM and building: From a literature review to an integrated conceptual framework." International Journal of Project Management 33(6): 1405-1416. Chung, J. K., M. M. Kumaraswamy and E. Palaneeswaran (2009). "Improving megaproject briefing through enhanced collaboration with ICT." Automation in construction 18(7): 966-974. Demiralp, G., G. Guven and E. Ergen (2012). "Analyzing the benefits of RFID technology for cost sharing in construction supply chains: A case study on prefabricated precast components." Automation in Construction 24: 120-129. Ergen, E., B. Akinci and R. Sacks (2007). "Tracking and locating components in a precast storage yard utilizing radio frequency identification technology and GPS." Automation in construction 16(3): 354-367. Hong, J., G. Q. Shen, C. Mao, Z. Li and K. Li (2016). "Life-cycle energy analysis of prefabricated building components: an input–output-based hybrid model." Journal of Cleaner Production 112: 2198-2207. Hossein Alavi, A. and A. Hossein Gandomi (2011). "A robust data mining approach for formulation of geotechnical engineering systems." Engineering Computations 28(3): 44 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 242-274. Jaillon, L. and C. S. Poon (2009). "The evolution of prefabricated residential building systems in Hong Kong: A review of the public and the private sector." Automation in Construction 18(3): 239-248. Leu, S.-S. and S.-T. Hwang (2002). "GA-based resource-constrained flow-shop scheduling model for mixed precast production." Automation in Construction 11(4): 439-452. Li, C. Z., J. Hong, F. Xue, G. Q. Shen, X. Xu and M. K. Mok (2016). "Schedule risks in prefabrication housing production in Hong Kong: a social network analysis." Journal of Cleaner Production 134, Part B: 482-494. Mao, C., L. Shen, L. Luo and Z. Li (2015). Identification of Risk Factors Influencing the Implementation of Industrialized Building System in China. Proceedings of the 19th International Symposium on Advancement of Construction Management and Real Estate, Springer. Pang, L. Y., R. Y. Zhong, J. Fang and G. Q. Huang (2015). "Data-source interoperability service for heterogeneous information integration in ubiquitous enterprises." Advanced Engineering Informatics 29(3): 549-561. Peffers, Ken, Tuure Tuunanen, Marcus A. Rothenberger, and Samir Chatterjee. "A design science research methodology for information systems research." Journal of management information systems 24, no. 3 (2007): 45-77. Sacks, R., C. M. Eastman and G. Lee (2004). "Parametric 3D modeling in building construction with examples from precast concrete." Automation in Construction 13(3): 291-312. Shin, T.-H., S. Chin, S.-W. Yoon and S.-W. Kwon (2011). "A service-oriented integrated information framework for RFID/WSN-based intelligent construction supply chain management." Automation in Construction 20(6): 706-715. Tam, C. M., Deng, Z. M., & Zeng, S. X. (2002). Evaluation of construction methods and performance for high rise public housing construction in Hong Kong. Building and Environment, 37(10), 983-991. Tam, V. W., C. Tam, S. Zeng and W. C. Ng (2007). "Towards adoption of prefabrication in construction." Building and Environment 42(10): 3642-3654. Wong, R. W., J. Hao and C. M. Ho (2003). Prefabricated building construction systems adopted in Hong Kong. Proc. of the International Association for Housing Science on Word Congress of Housing: Process and Product, Montreal, Canada. Yin, S. Y., H. P. Tserng, J. Wang and S. Tsai (2009). "Developing a precast production management system using RFID technology." Automation in Construction 18(5): 677691. Zhong, R. Y., Y. Peng, J. Fang, G. Xu, F. Xue, W. Zou and G. Q. Huang (2015). "Towards Physical Internet-enabled Prefabricated Housing Construction in Hong Kong." IFAC PapersOnLine 48(3): 1079-1086. Zhong, R. Y., Y. Peng, F. Xue, J. Fang, W. Zou, H. Luo, S. Thomas Ng, W. Lu, G. Q. P. Shen and G. Q. Huang (2017). "Prefabricated construction enabled by the Internetof-Things." Automation in Construction 76: 59-70. Fang, Y., Cho, Y. K., Zhang, S., & Perez, E. (2016). Case Study of BIM and Cloud– 45 855 856 857 Enabled Real-Time RFID Indoor Localization for Construction Management Applications. Journal of Construction Engineering and Management, 142(7), 05016003. 46