Flood Damage Assessment (FDA) is a key component in modern risk management frameworks providing an effective basis for decision making and the treatment of the risks. Current FDA methods do not consider the distinctiveness of buildings in analysis and therefore, cannot analyse them on a case-by-case basis, which is necessary for a variety of applications like engineering and design evaluation. This is mainly due to the limited input data used in these methods. The information required for such micro-level FDA analysis includes on one hand, complete building information (well-represented in BIM) and on the other hand, flood information that is commonly managed by GIS. While the independent use of BIM and GIS cannot satisfy all the information requirements for detailed FDA, their integration can potentially be used for this purpose. However, existing integration methods are application-specific and their adoption for FDA is challenging. This paper presents a method for BIM-GIS integration to support the requirements of a detailed assessment and 3D visualisation of flood damage to buildings. The data modelling cycle was used to design a new data model as a profile of GML for this purpose. The model was evaluated using a case study and found effective to satisfy the required criteria for micro-level FDA.
It has been recognised in the past decade that a management of flood risks which focuses solely on the hazard is
not effective
Efforts in the field of FDA have resulted in development of a variety of FDA methods to serve different
applications at different spatial scales (i.e. Macro, Meso and Micro) such as risk mapping, vulnerability assessment
or financial appraisal of damage. Within this context, a strong emphasis is made on buildings due to their significant
economic importance
An effective case-by-case analysis of damage to a building at micro level requires the use of two sets of
information: the flood parameters (e.g. depth and velocity) causing damage to the building components; and the
building components that resist against the flood impacts and are unique to each building
The independent adoption of BIM or GIS for a micro level FDA would be inadequate to support the information
requirements for this application. However, their combination provides the potential to accommodate these
requirements under one umbrella to serve the analysis and presentation of the damage to the building
In this paper, a method for combining the BIM and GIS at the data level is proposed to serve the information needs for a micro level assessment and 3D visualisation of potential building damage from floods. It is hypothesised that the use of this method can facilitate an information system to harness the combined strength of the BIM and 3D GIS, allowing for a case-by-case analysis of the building damage by the implementation of certain analysis types at a level of detail that the existing FDA methods are not sensitive towards. The outputs produced by such analyses can facilitate decision making by a range of stakeholders such as engineering and design firms, councils and insurance companies for improving the resilience of the community towards floods and their adverse impacts.
In the remainder of this paper, first, the research background as well as an overview of the related work for the integration of BIM with GIS is provided. Next, the design of the proposed integration method and its details are presented. Further, a demonstration of the integration using a case study is illustrated and the results are discussed. Finally, the paper is concluded and the future research directions are proposed.
BIM is a rich and intelligent digital repository of building information and uses an Object Oriented (OO)
approach to describe the characteristics (semantics and geometry) and behaviour of each building element as well as
its relationships with others
GIS, on the other hand, is a platform for managing and presenting spatially referenced information. Within this
domain, the exchange of geospatial data and the interoperability between systems are established using the
Geographic Markup Language (GML). GML is an Open Geospatial Consortium (OGC) standard data model for
defining the data types and constructs for describing the geographic features. With a more specific focus, the
heterogeneous geospatial information about urban data (e.g. buildings, transport, vegetation and water bodies) at
different levels of details is integrated within the framework of the 3D virtual city models such as CityGML
GIS and BIM originate from different domains and were developed for the specific needs of that field. Their
integration creates a seamless and scale-independent view of the world across both domains. It can benefit a variety
of applications that meeting their requirements would not be possible by independent use of BIM or GIS
Various attempts have been made for integrating BIM with GIS that can generally be classified into three groups: at application, process, and data levels.
At the application level, the integration methods use reconfiguring or rebuilding
There are a variety of methods developed to integrate BIM with GIS at the data level. Linking methods such as
ESRI ArcSDE facilitates data transfer between BIM and GIS software by an Application Programming Interface
(API) at either side. Translation/Conversion methods such as FME
In the more comprehensive and flexible integration methods, either a new data model as a "meta model" is
developed to mediate between BIM and geospatial information at higher level; or a data model at GIS or BIM side
(e.g. CityGML or IFC) is extended to incorporate the data from the other. The GeoBIM and utility network
extensions by Van Berlo and Laat (2010) and
In this paper, an integration method for integration of BIM and GIS for micro-level assessment of flood damage on building is presented. It supplies the resistance parameters of the building at high level of details, flood information as well as the other geographically extended features (e.g. elevation) in one unified data model to use for assessment and 3D visualisation of the flood damage to building.
The integration of BIM and GIS in this work is proposed at the data layer and implemented on the GIS side for utilisation of its spatial analysis tools for FDA calculations. The GML standard does not explicitly define the semantics of the geographic features (e.g. buildings or roads), and as discussed previously, there are limitations in representation of building and its components in the CityGML. For this reason, a new data model as a profile of GML is proposed to integrate the BIM information alongside the spatiotemporal dynamics of the flood information for assessing and visualisation of the damage on the building.
The proposed data model in this paper was developed based on the "data modelling cycle"
Next, the details of the above process for designing the proposed data model is explained, the outputs are presented and its benefits for use for a micro-level FDA are discussed.
Use cases are informal scenarios describing the expected behaviour of a system as a response to the needs of the stakeholder(s). Use cases are the basis of the extraction of the required functionalities and the data needs for system design.
In this study, the defined use case is related to the assessment and visualisation of the potential damage to a
building from a riverine flood. It requires (i) the assessment of the building safety (structural stability) by assessing
the damage to structural and load bearing components, (ii) the estimation of the total cost of repair/replacement of
the damaged components (excluding its content), and (iii) the visualisation of the location and mode of damage at
component level. This use case was detailed by investigating the land development process, a review of previous
publications and liaising with engineering and design firms, councils as well as referral authorities (e.g. Melbourne
Water) as potential users of the system outputs. The behaviour of buildings of different types may vary in flood
situations and the processes for damage analysis on them may be different from one to another. In this research, the
scope was limited to only one type which represents the most common Australian residential construction type: a
single-storey slab-on-ground brick veneer house
Based on the definition of the use cases in Section 3.2, the vulnerable components of the selected house type
were identified. This process involved a systematic and extensive review of the literature
Following the identification of susceptible components, further investigation was undertaken to identify and
document (a) the modes of damage for each component from floodwater impacts, (b) engineering methods for their
modelling and (c) the information inputs for these calculations. For the first two steps, Australian design and
construction standards (e.g. AS3700 - masonry building design) and relevant documents for improving the house
resilience to flood
Details Defining the spatial container for objects. It can have a corresponding element (e.g. building storey or a space in the building) that acts as the container object.
Representing the elevation of the area. It is required to be in multiple levels of details. Terrain can be either point-based or surface-based.
The flood parameters using multiple representations: (a) Spatio-temporal point distribution of depth and velocity vectors (for use in damage calculation); (b) Surface representation of flood (e.g. water level surface).
The footprint, address, height and the area.
Including storeys, walls, stairs, floors, foundation, beams, columns, roof, structural connections (e.g. wall ties), framing members, floorings, ceiling, soffit, skirtings and mouldings, doors, windows and cladding vents (e.g. airbricks).
For example, electrical objects like switches, meter boxes and outlets.
Construction materials of the building elements (single material or multiple)
Including cost of repair/replacement of building and utility components; and the building value.
According to the identified requirements, the conceptual data model illustrating the required concepts and their relationships was designed. Throughout the design process, a continuous investigation was undertaken to identify how these concepts are modelled in BIM (IFC) or GIS formats (GML and CityGML). This mapping was used to refine the design to improve interoperability and information translation between IFC or CityGML and the proposed model. The data model consists of seven packages inheriting its high level feature definitions from the GML. These packages are namely: the Core (CoreUrbanFlood), Terrain, Flood, Building, Utility, Valuation and MaterialDomain. In this paper, the UML class diagram was employed for developing and presentation of the data model. Due to the complexity of the model and limitations in space here, the attributes are not illustrated in the class diagrams. In addition, each package and their respective classes are colour-themed for the ease of read.
The "CoreUrbanFlood" package (see Figure 1) includes the necessary high-level objects for a micro-level FDA on a building. "UrbanFloodModel" concept is the highest level entity in the model defining an urban flooding scenario. It is a collection of defined materials, costs, spatial structures (explained in Section 3.3) and urban elements (_UrbanObject) representing the urban environment in a flooding situation. Urban objects can be the parcels (site), buildings, individual utility elements (_UtilityObject) or their aggregation as a system (UtilitySystem), flood representation (_FloodObject), and a simple or complex elevation model (_TerrainObject). Other urban objects like city furniture, roads, etc were beyond the scope of requirements in this research as the focus is only on damage assessment on buildings.
The proposed data model adopts a subset of the "Digital Terrain Model" thematic model in CityGML 2.0 for its Terrain package (see Figure 2). An elevation object in here can be stored by an independent subtype of the abstract concept "_TerrainObject". It can be either a surface-based object (e.g. TIN) represented by "TinTerrain" or in a multi-point form using "MassPointTerrain". On the other hand, a single Terrain can be represented by an aggregation of a number of _TerrainObjects in different representation forms and levels of details within the "Terrain" object. Each _TerrainObject has a validity extent - represented by a polygon, to define its effective scope.
The Flood package, illustrated in Figure 3, consists of the required classes to represent the flood information.
Any flood in the model (_FloodObject) is described by its metadata (FloodMetadata) that provides information
about its exceedance probability, flood duration, number of time steps, duration of each step and the units of
measurement for depth and velocity. As a subtype of the abstract _FloodObject, the "FloodBody" represents the
flood in urban area in either point coverage or GML surface forms. In GML, a coverage class (e.g.
multiPointCoverage) uses the relationship between RangeSet and DomainSet to link the geometry with its attributes
(see GML 3.2.1 specification for details). The flood extension for CityGML by
Surface representation of the flood was specifically considered for its 3D visualisation. It can either be presented by a single MultiSurface object using "RepresentedBySurface" relation for the maximum depths; or using the aggregation hierarchy of timeSeriesSurfaces → TimeStep → _floodBoundarySurface classes, for a surface representation for each time step. While "floodSurface" and "FloodGroundSurface" are used for water level surface and the surface between water and ground, the "FloodClosureSurface" is used to close the enclosure when the flood geometry is not a closed volume.
The valuation package (see Figure 4a) contains an abstract concept, the "_CostObject", that defines the value of a particular object. The "AssemblyCostObject" and "BuildingValue" realise the _CostObject for the repair/replacement value of building components or the construction cost of the building as a whole. Other information describing these classes include issuing institution, date of issue, currency type, etc. (a) (b)
The MaterialDomain package contains classes that define the construction materials of the building elements. As illustrated in Figure 4(b), this package consists of three kinds of material definition. "Material" class which defines a single material that can be directly assigned to a building component or be used as a layer in other material classes. "MaterialLayerSet" on the other hand, defines multiple materials as layers. An example here is a wall panel which includes the paint, lining and brick materials. The order of layers defines the position of the material in the object. "MaterialConstituentSet" is another material definition method which consists of one or more "MaterialConstituents" each of which defines the material of a part of component (e.g. "frame" or "glazing" in a window) indicated by its particular name.
Utility package contains the classes related to the interior or exterior utilities of the house. This package is illustrated in Figure 5. The utility objects can be defined independently or under a particular system (e.g. electrical, water, fuel or HVAC). Each realisation of the abstract "_UtilityObject" class may have a replacement value and a material object defined. In terms of the representation, utilities are commonly presented by either a GML solid or multiSurface. In this research, for the utilities, the focus was concentrated on the electrical system elements such as lights, outlets, meters, switches and distribution boards which are defined under abstract classes "_FlowTerminal" and "_FlowController". In addition, elements such as cables can be defined using the "FlowSegment" class that can be represented by 3D line segments using a GML Curve. The "_ControlElement" class, on the other hand, is an abstract class reserved for future use to represent utility components that are used to impart control over the other elements in the system. These classes can be directly mapped to the distribution elements in the IFC data model.
The building package (see Figure 6) comprises the classes that represent the building and individual or an aggregation of building components. The "Site" class represents a parcel characterised by an address and a 2D polygon which can contain one or more buildings. Each building has a value associated with it (BuildingValue) and can be represented by either its 2D footprint or the aggregation of 3D geometries of its components (e.g. stories). A building consists of at least one storey defined by "BuildingStorey" class and may contain utility objects or systems, as well as any subtype of the abstract class "_BuildingElement". Each of utility or building elements has a damage state and replacement cost attributes associated with them that can be used for cost analysis.
The building components defined in the model include slabs (either foundation slab or floor represented by classes of similar names), structural beams and columns, walls (either simple or complex wall represented by its different components using the relationship "consistsOfParts" defined at its supertype), roof (can be defined in the same way as wall), stairs (represented as a single component or by railings and stairflights), framing members (representing the structural framing of the building other than columns and beams), coverings such as ceiling, flooring, soffit, cornices and skirtings), windows (sliding doors are defined similarly), doors, airbricks (vents) or any type of void opening. In addition to these components, "BuildingElementPart" defines a class for a generic part of any other element. Explicit classes for wall parts ("WallComponentElement" such as the cladding) or covering parts (e.g. ceiling insulation and lining) using "CoveringLayerElement" are defined in the model to represent these objects.
In this model, windows and doors are defined either as single object or a combination of a lining (its frame) and a minimum of one panel that may have their own geometry, material and cost. "Space" class in this model defines those elements for representing the internal (e.g. room) or external (e.g. the backyard) spaces for the building. d M U : 6 e r u g i F
On the other hand, "Connection" class here is the supertype for all connections defining a link between a _BuildingElement (related element) to another (relating element). The "Mechanical connection", is a specialised connection type that employs an additional linking _BuildingElement for the explicit establishment of the connection. An example here is the brick cladding to framing connection using "WallTie". The connection between the relating and related elements can be further detailed by the geometry type (point, curve, surface or volume) that either of the elements or the realising element connects to them. In the wall tie example, the connection is simplified by a PointConnectionGeomtry class that has a GML point defined on related and relating elements. The connection concept can easily be mapped to its counterpart in IFC model for integration purposes.
To support extensibility, the building model defines a generic class, "BuildingElementProxy" to be used for elements that are not explicitly defined in the current version of the model. A "Slab edge" is an example of this element.
Subsequent to the design of the conceptual model discussed in Section 3.4, the logical and physical models were developed. XML file was selected in this research to implement the integrated information model. Therefore, the physical model was prepared in relation to XML schema specifications. According to the described UML packages in Section 3.4 and the designed physical model, an XML Schema was developed. It defines the structure of the XML file and rules for definition of objects in it. This schema comprises six namespaces, each of which corresponds to and implements one of the UML packages described previously in Section 3.4.
To verify the application of the designed integrated information model to serve the defined use cases in Section 3.2, and testing the hypothesis of the research, a case study was conducted in collaboration with Maribyrnong Council and Melbourne Water. In this study, damage to a selected house in Maribyrnong was evaluated and visualised.
The required data such as elevation model of the case study area, plans of the building under investigation, and the inputs for the flood simulation (e.g. river discharges) were provided by the council and Melbourne Water. The building value and the component costs, on the other hand, were obtained from the Office of the Valuer General and the Rawlinsons' Australian construction cost handbook 2014. The BIM model of the house was developed based on the provided plans and then exported to an IFC file. On the other hand, a 1-in-100 year flood (commonly used for planning) was simulated using MIKE 21 simulation package and the outputs (spatial distribution of depth and velocity) were exported to 1140 ESRI shape files, each corresponding to a time step. An in-house tool was used to extract flood parameters from these files into a single XML file. This XML file was then mapped to the flood concepts of the proposed model in Section 3.4 and the flood information, including its geometry and attributes, were stored.
In spite of the previous efforts explained in Section 2.3, there was no tool found that could provide a smooth conversion for both geometry and semantics of BIM data to GIS formats. Therefore, a semi-automatic process (illustrated in Figure 7) was designed in this research to import building information from IFC into the implemented database. In this method, the geometry and semantic information are obtained separately and then combined to be stored in the designed database. For geometry extraction, first the IFC elements are converted to ESRI geodatabase feature classes using the ArcGIS Interoperability Extension. Then each feature class is converted to CityGML using the export function in the extension. As the input objects are not known to the converter engine, the created objects in the output are in the form of "GenericCityObject" containing a multiSurface representation of the elements. On the other hand, the attributes and relationships between elements are obtained from an exported XML version of the IFC file and combined with their geometry using a unique identifier of IFC elements, the "TAG". Having the geometry and the semantics of the elements combined, they are then mapped to the proposed data model objects and stored in the database.
The flood parameters and building components were then used to evaluate the mode and level of damage to individual components. These calculations are based on the extracted engineering methods for assessing the damage (see Section 3.3) implemented in a prototype system. The cost of damage to each component was calculated based on its damage level and replacement cost. In addition, the geometry of the building components and colour coding were used to visualise the location as well as the level of damage in 3D.
The damage to the building and its components in the case study in Section 4 were assessed by use of the BIM and geospatial information together and the implementation of the functions extracted in Section 3.3. Figure 7 illustrates the study area and the visualisation of flood simulation outputs in 2D and 3D GIS.
(b) (c)
(d)
The damage analysis process showed no structural instability as the load bearing elements remained unaffected. However, the building suffered from approximately $51,000 damage to its other components (e.g. doors and flooring) from water impacts. This number is the sum of damage costs to individual elements which a subset of these costs is presented in Table 2.
From the above results, all three uses cases defined in Section 3.2 are shown to be successfully satisfied: the building structural safety was assessed and the cost of damage to individual components were estimated and visualised in 3D. The rich database based on the proposed data model in Section 3.4 was profitably used to integrate building information with flood parameters to assess the damage to the selected brick-veneer house at its component level and by taking into consideration its unique characteristics. This method provides a more detailed output illustrating the details of potential damage to a building that cannot be obtained by the current methods for FDA such as damage curves (presenting only a single number for building damage). Therefore, it can potentially overcome the limitations of current methods towards providing a better understanding of vulnerabilities in the building and facilitating an effective decision making for their treatment.
A range of stakeholders (e.g. engineering and design firms, councils, referral authorities as well as the building owners) can benefit from the outputs of the presented method. A majority of these parties are often challenged by similar questions such as "Is a particular development, proposed in an area with risk of flood, resilient to the potential risks and should be permitted for construction?" To answer this question, a tool based on the proposed integration method for BIM with GIS (similar to what was presented in this study) can facilitate engineers to test their designs against the mandated flood performance requirements from Australian Building Code Board. The alternative design options or mitigation measures may be considered for treating the vulnerabilities. In here, a costbenefit analysis of their deployment using the proposed method can facilitate the selection of the most effective solution. On the other hand, referral authorities and councils can assess the planning and construction requirements of the proposed building more effectively by taking into account the details of the risks. Additionally, a feedback from the council for this study indicated that in case of disputes between owners and council taken to Victorian Civil and Administrative Tribunal (VCAT) in regards to refusal of a particular proposal due to risks, a nonengineering language such as the presented 3D visualisation in here can be beneficial for the communication of risks to owners for their better comprehension of the basis of the decision. The preliminary feedback from some of the aforementioned parties has been positive. However, further systematic investigation is required for understanding the value of the additional detailed information presented to them in this work.
The proposed integration method in this work can complement the use of damage curves for large-scale applications to create a multi-scale framework towards a better and more comprehensive understanding and treatment of the flood risks at different levels. It can help in improving the resilience of the community towards floods and their adverse impacts. The application of the proposed method, however, is limited to one or a few buildings where data and computation resources can be feasibly provided. For larger number of buildings at municipality or city level, the demand for such high level of detail damage evaluation for decision making would be small and other existing methods (e.g. curves) can suitably provide the required decision support.
In this paper, a method for integration of BIM with GIS at the data level using a development of a new data model as a profile of GML was presented. The designed data model allows for a unified and consistent storage of the detailed representation of the building information alongside the flood parameters and other information (e.g. elevation model) in support of the micro-level FDA on buildings. The implementation of the data model and its use for assessing the damage to a selected building in a case study in Maribyrnong evidently supported the proposed hypothesis in this research: the BIM-GIS integration can facilitate a detailed assessment and 3D visualisation of damage costs to a building that is presently not supported by the type of inputs used in the current methods for FDA.
The preliminary feedback from the discussions with engineers and the council has been positive and a number of benefits are highlighted. Further research, however, is envisaged to systematically investigate the value of this extra and detailed information to the stakeholders involved in detailing the use cases in this research.
Furthermore, the data model presented in this paper was designed based on the analysis of the requirements for one particular construction type. The methodology used in this research can also be used for other building types and hazards to extend the data model for all types of buildings towards the development of a comprehensive repository of data for analysis of all types of buildings and events.