BIM-based Damage Assessment and Scheduling for Post-Earthquake Building Rehabilitation

Transcription

1 BIM-based Daage Assessent and Scheduling for Post-Earthquake Building Rehabilitation C. Georgiou & S. Christodoulou Departent of Civil and Environental Engineering, University of Cyprus, Nicosia, Cyprus D. Vavatsikos National Technical University of Athens, Athens, Greece An a-priori assessent of the post-earthquake condition of structures has long been a target of the Architectural, Engineering and Construction (AEC) industry, with recent efforts by the Pacific Earthquake Engineering Research (PEER) Center highlighting such goal. The aforeentioned PEER efforts have developed an appropriate basis for seisic daage assessent, including cost estiation and scheduling for postearthquake building rehabilitation, but visualization tools are notably absent fro this fraework. The ethodology proposed herein relies on the integration of tools currently available to the AEC industry, such as Building Inforation Modeling (BIM), a fourth-generation prograing language, a relational database anageent syste and construction anageent tools, within the PEER Center fraework. The proposed process is autoated through the developent of appropriate software, providing building inforation odels with data about the building eleents daage state, 3D daage assessent visualizations, the expected rehabilitation cost and duration, for different levels of seisic intensity. 1 INTRODUCTION 1.1 Overview & Scope In recent years, the scientific counity has focused any of its efforts on understanding the behavior of structures under seisic loa notwithstanding the uncertainty and difficulty in forecasting earthquakes and their consequences. The eergence of Perforance Based Earthquake Engineering (PBEE, Cornell & Krawinkler 2000) is slowly pushing the study of ajor infrastructure projects away fro the classic design of a code-conforing building (one design fits all odels) to a structural design process that encopasses the growing need for specialized structures tailored to the requireents of each individual owner. As a result, an owner can now decide upon the level of perforance (i.e. safety) that is desired for a certain building at each possible level of earthquake shaking that the building ay experience. Instead of relying on engineering response paraeters (e.g., shear, oent, drift), the PEER PBEE fraework is based on the use of nonengineering quantities such as cost, downtie and casualties. The work presented herein ais the developent and presentation of a practical coputer-based approach for estiating and visualizing daages, costs and repair scheduling for post-earthquake building rehabilitation by use of Building Inforation Models. Within the proposed fraework (Christodoulou et al. 2010), the geoetric properties of a building s eleents (i.e. of coluns, beas, walls, doors and windows) are cobined with the structural response of the building (obtained through a series of nonlinear dynaic analyses for given ground otion sets scaled to specified levels of seisic intensity) and then visualized through its BIM. 1.2 Perforance-Based Earthquake Engineering Using BIM BIM is an innovative integrated design process involving the design, construction and anageent of digital representations of physical and functional characteristics of a facility. The technology and its several incarnations have proven to be a particularly useful counication, planning, analysis and decision-aking tool for designers, engineers and constructors. The benefits of BIM have been extensively researched and docuented in literature. PBEE allows the design of structures that can perfor to a desired perforance level under earthquake loa e.g., reaining fully operational for low intensity frequent earthquakes, sustaining low daage at less frequent events and perhaps needing heavy repairs or replaceent but aintaining struc-

2 The 3D odel (Figure 3) contains both structural (coluns, beas, slabs) and nonstructural (walls, doors, windows, furnishings, etc.) building copo- tural integrity at the rarer and ost intense shaking levels. In recent years several guidelines that recognize such needs have been published, such as the FEMA-356 and ATC-40 guidelines (FEMA 2000, ATC 1996). A proinent exaple is the PEER Center ethodology (Cornell & Krawinkler 2000) that has been developed to offer a coprehensive assessent of building perforance at any intensity level and any desired liit-state by integrating the seisic hazard and the structural analysis results with daage and loss analysis to produce realistic estiates of the rehabilitation cost and duration associated with any earthquake. The probabilistic prediction of the building perforance is expressed by Equation 1. G[ DV DM] P[ DM EDP] [ DV] P[ EDP IM ] dim dedp ddm DM EDP IM (1) d IM ( ) dim where, DV: decision variable, e.g., repair cost and duration. DM: daage easure, e.g., slight, oderate, severe daage or total loss. EDP: engineering deand paraeter e.g., plastic rotation, inter-storey drift ratio etc. IM: intensity easure e.g., spectral acceleration at the first ode S a (T 1 ). λ[ ]: ean annual frequency of exceedance of the quantity in brackets. G[ ]: copleentary cuulative distribution function (CCDF) of the quantity in brackets. P[ ]: probability density function (PDF) of the quantity in brackets. 2 BIM-BASED DAMAGE ASSESSMENT AND SCHEDULING FOR POST-EARTHQUAKE BUILDING REHABILITATION 2.1 Methodology The proposed ethod utilizes nuerical siulations and a cobination of the extracted probabilistic response distributions with the appropriate fragility curves, to evaluate all possible daage scenarios (Christodoulou et al. 2010). These nuerical calculations provide repair cost estiation per eleent, eleent group, storey and building at different intensity levels. Considering the uncertainty in the unit prices, the coplete repair cost distribution is deterined rather than just its ean value. Additionally, through the design of an autoated relational database and its interface with scheduling software, the building asseblies are classified according to their daage state in work breakdown structures (WBS) and assigned to specific repair activities of fixed productivity. As a result of the tie scheduling, the rehabilitation tie and hence loss-of-use cost is estiated. Beyond estiation, visualization of daage is paraount. Without the proper visualization capability daage or cost values often becoe eaningless. For exaple, any significant debris on a hospital corridor could easily render a nuber of roos useless or inaccessible. Such iplications are ipossible to decipher fro the output of any existing structural analysis progra, let alone explaining the to the client. Having PBEE analysis results available on the 3D level, offers the ability to engineers and clients to identify restricted areas, plan for possible routes for oving aterial and personnel and in general get a sense in real-tie of what is happening without having to physically go and inspect the building. To achieve this goal we propose cobining current PBEE practice with Increental Dynaic Analysis (IDA, Vavatsikos & Cornell 2002), together with readily available visualization techniques in order to create 5D BIM - where downtie and rehabilitation cost becoe the fourth and fifth diension, respectively. The resulting odel can becoe the ain counication tool between engineers and owners, offering a coprehensive assessent of the risk that the building owner will be asked to undertake in the afterath of an earthquake. This tool can be ipleented in any professional design office and easily for the basis for new structural analysis software. 2.2 Benchark Building The benchark building (Figures 1-2) is a twostorey reinforced concrete special-oent-resistingfrae structure designed by Haselton (2006). The building is syetric in both orthogonal directions with equally spaced four-bay fraes in each direction. The odel contains only structural asseblies i.e. coluns and beas. Thus, walls, doors and windows are added in order to achieve the best feasible building siulation in a real environent and analysis integration regarding the daages that ay occur to all the building eleents. The building s ground and first floor heights are 4.57 and 3.96, respectively and each floor total area is 355². Colun diensions are 55.9/55.9c and bea diensions are 45.7/55.9c. The slab thickness is 20c, outdoor and indoor wall thicknesses are 25c and 10c, respectively. 3 3D DAMAGE VISUALIZATIONS

3 nents. This odel is used both as a digital visualization tool and as the eleents data inventory tool as it is used to generate several ite listings for quantity-takeoff. In particular, the geoetric properties i.e., height, width, length, thickness, area, surface and volue of all coluns, beas, walls, doors and windows are exported. The geoetric properties are cobined with the structural response resulting fro a series of non-linear dynaic analyses, perfored by Haselton (2006) in OpenSEES (PEER Center 2009), for thirty-nine given ground otion sets in twenty-two specified levels of seisic intensity S a (T 1 ) fro 0 to 2.8g. axiu inter-storey drift ratio (IDR) of the storey on which they are located. Figure 3. Building 3D odel. Figure 1. Building ground-floor plan. Figure 2. Building first-floor plan. The 3D odel and the eleent properties (fro BIM) are then cobined with deduced fragility curves to investigate the building response to earthquake loa in order to siulate all possible daage scenarios. Fragility curves represent the probabilistic expression of the strength of each coponent and relate structural response to various levels of daage, producing the probability of a coponent reaching or exceeding a particular daage state. The fragility curves for coluns and beas are deterined for their axiu plastic rotation, while for walls, doors and windows are deterined based on the Once a structural assessent is ade, a daage easure (DM) per building coponent is coputed subject to the fragility curves associated with each coponent (EDP). This daage easure is discretized in n daage states (DS), depending on the coponent. The daage states vary, but typically they are classified as no daage - no action needed, slight daage - repairable (low-cost repairs), oderate daage - repairable (repairs are costeffective), severe daage - needs replaceent (repairs are not cost-effective), and collapse - total loss. Therefore, the daage state for each building assebly can be used as a general daage descriptor that can in turn be visualized by use of appropriately coloring the corresponding eleents in the 3D odel (i.e., white color for no daage, green for slight daage, yellow for oderate daage, red for severe daage and black for collapse). In the figures shown below, gray color is assigned to walls that collapse in order to distinguish the fro doors and windows. The aforeentioned 3D-visualization of the building daage state by selectively coloring BIM objects based on their daage level, offers the ability to engineers and owners to understand the actual building operability after a seisic event. By coparing building daages at various intensity levels as they are illustrated in Figures 3 to 8 for S a (T 1 ) equal to 0.3g, 0.6g, 0.9g, 1.2g, 1.8g and 2.4g, respectively, it is possible to understand the need to utilize this technology in PBEE. As shown in Figure 4, at 0.3g, ground floor coluns and beas do not suffer any daage (white color), while walls, doors and windows suffer coplete loss (black color). At 0.6g (Figure 5), coluns suffer slight daage, soe beas do not get daaged while other suffer slight daage. Walls, doors and windows collapse. At 0.9g (Figure 6), coluns and beas suffer oderate and slight daage, respectively. At 1.2g (Figure 7), coluns suffer oderate and severe daage, soe beas suffer slight daage while other suffer oderate daage. At

4 1.8g (Figure 8), all coluns and beas suffer severe and oderate daage, respectively. At 2.4g (Figure 9), coluns suffer total loss. Figure 7. Ground floor 3D daage visualization at S a (T 1 ) =1.2g. Figure 4. Ground floor 3D daage visualization at S a (T 1 ) =0.3g. Figure 8. Ground floor 3D daage visualization at S a (T 1 ) =1.8g. Figure 5. Ground floor 3D daage visualization at S a (T 1 ) =0.6g. Figure 9. Ground floor 3D daage visualization at S a (T 1 ) =2.4g. Figure 6. Ground floor 3D daage visualization at S a (T 1 ) =0.9g. Siilarly, at 0.3g first floor coluns and beas do not suffer any daage, walls suffer oderate and severe daage, while doors and windows suffer oderate daage (as shown in Figure 10). At 0.6g

5 (Figure 11), coluns and beas do not get daaged, while walls, doors and windows suffer coplete loss. At 0.9g (Figure 12), coluns suffer slight daage. beas suffer slight and oderate daage. At 2.4g (Figure 15), coluns suffer severe daage, while beas suffer oderate and severe daage. Figure 10. First floor 3D daage visualization at S a (T 1 ) =0.3g. Figure 13. First floor 3D daage visualization at S a (T 1 ) =1.2g. Figure 11. First floor 3D daage visualization at S a (T 1 ) =0.6g. Figure 14. First floor 3D daage visualization at S a (T 1 ) =1.8g. Figure 12. First floor 3D daage visualization at S a (T 1 ) =0.9g. Figure 15. First floor 3D daage visualization at S a (T 1 ) =2.4g. At 1.2g (Figure 13), soe beas do not get daaged while other suffer slight daage. At 1.8g (Figure 14), all coluns suffer oderate daage and

6 4 REHABILITATION COST The building repair cost distribution is calculated by Monte Carlo siulation (Latin Hypercube Sapling, LHS), stratified for N possible daage scenarios. The ethod is applied to K eleents and N possible daage scenarios, and is perfored for R ground otion sets in each one of the L intensity levels. At any intensity level, the expected repair cost of each coponent is the ean value of a vector containing RxN possible repair cost values, where N are the repair cost values obtained fro each EDP generated by each ground otion set, separately. Siilarly, the expected building repair cost is the ean value of the vector containing RxN possible repair cost values, where N is the su of the individual eleent repair cost values in each daage scenario. Additionally, the variance, the standard deviation, and the coefficient of variation of the building total repair cost can be calculated (Georgiou 2013). The whole procedure is perfored in Matlab using custodeveloped algoriths. In the case a ground otion causes collapse of the building, the distribution of the building repair cost is replaced by N possible values as they occur fro the building replaceent cost distribution. Additionally, by choosing the analysis to be perfored for particular eleents, it is possible to calculate the expected repair cost per floor and per eleent category. Figure 16 illustrates the building repair cost distribution per intensity level, in the for of the edian (50%) value as well as the standard deviation of the total repair cost (16% and 84% of the distribution). building eleents in appropriate work breakdown structures (WBS), classified according to their daage state. This RDBMS provides links between the BIM objects and the rest of the database tables archived in it, through the priary keys of each database table. For exaple, each eleent has a unique ID which is linked, through a apping table, to a CSI code and through that to a crew code. As a result each eleent is assigned with specific repair activities, unit cost and production rates. The physical properties of the eleent (surface, volue etc.), as they are generated fro the BIM, in conjunction with the production rates fro the CSI codes assigned to it, dictate the duration T of the corresponding daage rehabilitation activity, as shown in Equation 2. As a result of the tie scheduling perfored in MS-Project, the rehabilitation tie and hence loss-of-use cost after each earthquake scenario is estiated (Georgiou 2013). Figure 17 illustrates the distribution of the ean building repair duration per intensity level. Q tc T (2) i, c, n P c, c, where, T i,c, : duration of the activity i, expressed in work-hours. Q : total quantity of all eleents of type in ters of volue, surface or area. t c : total hours worked by crew c in a day. 8 hours work per work-day is adopted. n c, : nuber of available crews c. P c, : productivity per day of the crew c. Figure 16. Building repair cost distribution per intensity level S a (Τ 1 ). 5 REHABILITATION DURATION Figure 17. Building repair duration distribution per intensity level S a (Τ 1 ). A custo-developed relational database anageent syste (RDBMS) is then used to record the

7 6 CONCLUSIONS Described herein is a proposed integrated approach for post-earthquake building daage assessent and 3D visualizations, by eans of integrating building inforation odels with structural and construction engineering principles. The proposed ethod is a valuable tool for AEC industry participants, for it allows the autoation of structural, cost and scheduling analyses and their integration with 3D visualizations of buildings. The ethod enables a visuallydelivered holistic approach to daage assessent, cost and tie estiation for the rehabilitation of seisic daages, and offers a coprehensive assessent of the risk that the building owner will be asked to undertake in the afterath of an earthquake. 7 REFERENCES Applied Technology Council (ATC) Seisic Evaluation and Retrofit of Concrete Buildings, Report No. ATC-40, Redwood City, CA. Cornell, C.A. & Krawinkler, H Progress and Challenges in Seisic Perforance Assessent, PEER Center News, 3(2): 1-2. FEMA Prestandard and Coentary for the Seisic Rehabilitation of Buildings, Report No. FEMA-356, Federal Eergency Manageent Agency. Washington. DC. Christodoulou, S., Vavatsikos, D. & Georgiou, C A BIM- Based Fraework for Forecasting and Visualizing Seisic Daage, Cost and Tie to Repair, ework and ebusiness in Architecture, Engineering and Constructio; Proceedings of the Eighth European Conference on Product and Process Modeling, ECPPM2010, Septeber 2010, Cork, Ireland. Georgiou, C Building Inforation Models for Daage Assessent, Cost Estiating, and Scheduling for Post- Earthquake Reinforced Concrete Building Rehabilitation, Ph.D. Thesis, University of Cyprus, Nicosia. Haselton, C.B Assessing Seisic Collapse Safety of Modern Reinforced Concrete Moent Frae Buildings, Ph.D. Dissertation, Stanford University. PEER Center Open Syste for Earthquake Engineering Siulation (OpenSEES), Pacific Earthquake Engineering Research Center, University of California, Berkeley, Vavatsikos, D. & Cornell, C.A Increental Dynaic Analysis, Earthquake Engineering and Structural Dynaics, 31: