THE INTEGRATION OF BIM AND GIS IN CONSTRUCTION PROJECT – A DATA CONSISTENCY REVIEW

: From days to days, management of construction project has been improved during life-cycle project, starting from planning until maintenance. This happen cause of the advantages in implementation technology of Building Information Modeling (BIM) and Geographic Information System (GIS) in supporting construction project. Few years ago, enhancement in term of BIM and GIS that provides an additional extension for the purpose of information management is very interesting. With the advantages that been provided by BIM and GIS, information of construction project can be adapted into real situation of the construction site which be helpful during the life-cycle of building construction. BIM and GIS is a different platform which contains their own advantages that support construction project. In order to bring the most effectiveness in management of construction project, integration between BIM and GIS becomes an important task to support the design phase until operational phase which include the facility management and maintenance. Although this integration can support the building information management, the software that used in integration process is still having limitations and differences in fulfilling the needs of users. For that reasons, data consistency needs to be studied in order to develop the best practices of integration application. The purpose of this paper is to investigate the data consistency during the integration process. From the investigation, it showed that there are some data inconsistency occurs in IFC platform after conversion process. Through this paper, the comparison of the geometric and semantic data before and after translation process will be examined.


INTRODUCTION
In the concept of construction project, Building Information Modeling (BIM) can provide a database that can link to any model. Usually this database contains the building information for the life-cycle of construction process until maintenance parts (Eastman et al., 2011;Arayici and Aouad, 2010). Along with the models, BIM also offer the advance of three-dimensional (3D) viewing that can make the process of data interpretation become easier and accessible (Ahzar, 2011). Meanwhile, Geographic Information System (GIS) model focuses on acquisition, editing, manipulation, analysis, modelling, visualization, and storage (Goodchild, 2009). These data can be represented into 2D thematic map or 3D map. More works in GIS can be seen working in 3D (see Shichao et al., 2020;Keling et al., 2017;Azri et al., 2014). For the better practice in construction project, integration between BIM and GIS should be considered. Although these two platforms can support various type of data, it still difficult to transform the data from one platform to another platform.
In 1995, The International Alliance for Interoperability (IAI) was established to develop first version of vendor naturel standard, called Industry Foundation Classes (IFCs). IFCs are a collection of entities (classes), that form an information model and it was released in 1997 (Isikdag et al, 2004). According to Arayici and Aouad (2010), IFC standards have been implemented in software application to create a common data exchange platform. An IFC standard is a primary standard that used for information transfer in BIM and GIS domains (Lim et al., 2019). Practically, the end user of the application used IFC as a feature to save and retrieve data in a form that is understandable to other software applications.
Nowadays, the process of conversion data between BIM and GIS is still a problem because of the mismatch data among them (Basir, 2018;Sani, 2018;Andrew, 2020). This happen because of the limitation associated with these two platforms which involve the data exchange and knowledge. In general, dissimilarities and mismatches between BIM and GIS happen because of different users' requirements, different application, different developmental stages, different spatial scales, different coordinate system, different semantic and geometric representations, different levels of granularity, and different information storage and access methods (Liu et al, 2017).

DATA CONSISTENCY
Data consistency is one of the components in data quality for geospatial database. There is multiples interpretation of data consistency. Table 1 show the interpretation of data consistency based on different purposes.

Authors Name
Years Interpretation of Data Consistency Veregin, H. 1999 Data Consistency refers to the absence of apparent contradictions in a database. Srivastava, R.N. 2008 Data consistency can be termed as the absence of conflicts in a particular database. Wang, F. 2008 Spatial data consistency refers to the logical rules of structure and attributes for spatial data and describes the compatibility between dataset items. ISO/TC 211 standard 19113 2002 Logical consistency as " degree of adherence to logical rules of data structure, attribution and relationships (data structure can be conceptual, logical or physical)" Plumer, L and Groger, G. 1997 Data consistency refers to the lack of any logical contradiction within a model of reality. Abadi, D. 2019 Data consistency refers to the ability of a system to ensure that it complies (without fail) to a predefined set of rules, but this rules changes based on context. Shi et al. 2019 Data consistency is a data characteristic that contradictory conclusions cannot be derived from the given data. Table 1. Interpretation of data consistency.
There are many aspects of data consistency problems in spatial databases, such as the inconsistency between attribute and geometry data; the inconsistency of topological relations after geometry objects is modified (Xinyan et al., 2000(a); Xinyan et al., 2000(b)).
In this paper, we bring for you a review of data consistency during the data integration between BIM and GIS. Besides, the comparison based on the tools used during conversion process also was observed.

REVIEW ON DATA CONSISTENCY IN DATA INTEGRATION BETWEEN BIM AND GIS
In construction process, there are several phases that involved during the life-cycle which started from the planning and design (P&D), construction, operation and maintenance (O&M) and demolition (Ma and Ren, 2017). In order to organize all these phases, BIM was used to manage the data during entire lifecycle of construction process.
GIS can be defined as the information that related to the geography which provides the detail spatial data on specific location by using coordinates. By integrating these two platforms, BIM data can be representing in the real world.
Integrating BIM and GIS can be done by using three options which are (Ma and Ren, 2017): i. Extract data from BIM system into GIS system ii.
Extract data from GIS system into BIM system iii. Extract data from both systems (BIM and GIS) into another system.
For this paper, reviews on the part of data consistency from BIM system into GIS system will be discussed. Zhu et al. (2020) proposed an approach for automatically converting IFC clipping representation onto shapefile format for the use in GIS. The proposed method successfully automated the conversion of IFC clipping into shapefile and the type of half space, and increasing the boundary size will not increase the size of corresponding B-Reps for half space, but will slightly increase the producing time of half spaces and processing time of building components. The conversion of geometry data becomes easier, more efficient and more reliable way. Zhu et al. (2019) studied on an Open-Source Approach (OSA) that can retrieved the geometric information in IFC through the spatial structure IFC and converted into shapefile by developing an automatic mutipatch generation algorithm (AMG). By using this AMG, the multipatch can be generated regardless of the initial ring order and extrusion direction. With any ring (clockwise or anti-clockwise), patches can always be generated. However, the results of multipatch may not be closed and thus have limited usage. This study focus on geometry information, therefore semantic information need to be enhanced because only few attribute were extracted during the process. The comparison has been done between OSA, Data Interoperability Extension for ArcGIS (DIA) and Feature Manipulation Engine (FME). Table 2 show the comparison conversion process between OSA, DIA and FME. Adouane et al. (2019) developed a method to solve geometric and semantic problem during the process data conversion from BIM-IFC into CityGML. The main advantage of that method is to control data flow, especially for geometric processing, which is advantageous to prevent from potential artefacts. In order to reinforce the mapping performances, maximising the use of schemas that available in CityGML is one of the important elements when BIM-IFC datasets are involved. During this process, there is some inconsistency on geometric data which is results of geometrically errors. Figure 1 shows the data inconsistency on geometric data after conversion process.    The other limitation is the way of geometry is explicitly stored in the model, which consequence to artificially increase the size of the generated CityGML target instance.
Wang and Raja (2019) utilizes the standard open-sourced indoorGML data structure as the medium model to develop indoor ontology at semantic level. The proposed method can facilitate the semantic interoperability between BIM/IFC and GIS/IndoorGML. This method can enhance the seamless data exchange between BIM and GIS applications which lead to the improvement of indoor routing for the project operation stage. In information of IFC elements consists of ID, GUID, and property name and reference IFC instance IDs. For entity IfcDoor and IfcWindow, additional attributes such as OverallHeight and OverallWidth was included. Geometric information was completely translated during the translation process although only few detailed semantic information was carried forward during the translation.
Mustorpha and Mohd (2019) studied on the 3D indoor GIS requirements for space management which is data level integration, data management, 3D indoor GIS analysis and 3D space management. During this study, geometric information exported from Revit through IFC into multipatch feature using ESRI Data Interoperability Extension. There are four IFC data components; geometry, semantics, relationship classes and properties of building elements are converted into 3D indoor GIS. The IFC geometry been converted into boundary representation (b-rep), Sweep Solid (SS) and Constrictive Solid Geometry (CSG) while semantics of building elements between IFC classes could be transformed into semantics in GIS environment. Figure 3 show the geometric information exported from Autodesk Revit software through IFC into multipatch features by using ESRI Data Interoperability Extension for developing ESRI geodatabase.
(a) Export BIM model from Revit into IFC format (b) Import BIM IFC model to GIS format Figure 3. Example of BIM to GIS data conversion for 3D indoor GIS Table 3 shows several BIM elements in IFC format that can be transformed into the 3D indoor GIS. Therefore, the information from IFC data needs to be properly organized in 3D space management to keep their source correctly. Table 3. Semantic data in IFC organized in GIS environment Floros et al. (2018) investigated interoperability options between the aforementioned standards, by converting IFC models to CityGML LoD 4 Models. During the converting process, an important attribute has been missing which is texture of the surface. During modelling the model, Autodesk Revit can assign and visualized the material but during export to IFC, the texture of surface is lost. This happen because IFC is designed for Architecture, Engineering and Construction (AEC) purposes and does not support the texture visualization (De Laat and v. Berlo, 2011). However, it is possible to maintain this type of information as semantic information. Figure 3 show the IFC material of walls transferred to CityGML.
Besides that, the separation and manipulation of the entities in order to generate a LoD 4 model is more challenging compare to LoD 3 model since the interior of the building encloses more geometric and semantic information than the exterior. Moreover, the different software environments during the conversion process can benefit the generated CityGML model by fixing errors that affect the representation and implementation of the generated model such as a door that misses multiple surfaces due to the fact that the IFC properties and elements are either damaged or missing elements in the first place. Another example is the fact that IFC solids can overlap which can be a common mistake since in BIM all components are placed based on the set elevation views and an unintentional error is possible. Deng et al. (2016(a)) investigated 3D traffic noise mapping using data from BIM and GIS integration. In this paper, BIM and 3D GIS integration is used to assess the traffic noise for indoor and outdoor environment in one platform. The built platform contains 3D GIS models at high levels detail of data from BIM. With this integration, the 3D GIS model can access detailed indoor features for noise evaluation. Important parameters such as absorption coefficient and transmission loss can be extracted directly from BIM for noise calculation. The data conversion engine is developed to allow seamless data integration between BIM and GIS which mean all data in BIM model can be translated into GIS model without missing geometry data and information that needed for noise mapping. Teo and Cho (2016) proposed an indoor network model from BIM for various indoor-outdoor routes planning application (e.g., emergency response and pedestrian route planning). The advantage from this study is reuses the existing BIM data and extend it to GIS analysis. An algorithm to automatically generate 3D indoor network model from BIM call multipurpose geometric network model (MGNM) using IFC-to-MGNM strategy. MGNM refines geometry and attribute information which contains various elements such as room, door, windows, landing from BIM into GIS platform. Tashakkori et al. (2015) proposed the new 3D Indoor or Emergency Spatial Model (IESM) to assist rescuers in the planning emergency response in improving destination travel times. This IESM was developed based on the IFC representation of BIM which is contained geometric and semantic of building elements and space inside the buildings. This IESM can provide the detailed semantic and geometrical information of building, and emergency information which enables spatial queries and information retrieval about indoor environment. Figure 4 shows the example of the translation result's IFC into GIS which cover the semantic and geometric information in GIS platform. Through the new IESM, it can be used to identify which area of the building can be accessed by using a fire hose with a certain length and in particular identifying the originating point. Figure  5 show IESM can useful in identify all points that can be reached within 10 m walking distance from the originating point even considering the stairs level distances. Figure 5. Identifying reachable areas using certain length fire hose (Tashakkori et al., 20015). Amirebrahimi et al. (2016) presented a method for BIM-GIS integration to support the requirements of detailed assessment and 3D visualization of flood damage to buildings. Through this paper, a data model that was proposed allows information storage to be unified and consistent from building information along with flood parameters and other information (e.g. Height model) to support the micro-level Flood Damage Assessment (FDA) in the building. The integration of BIM-GIS can facilitate the detailed assessment and 3D visualization of the cost of building damage that are currently not supported by the input type used in current method for FDA. Besides that, 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. Table 4 illustrates the mapping between the building components of the proposed data model and the IFC4 classes.
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIV-4/W3-2020, 2020 5th International Conference on Smart City Applications, 7-8 October 2020, Virtual Safranbolu, Turkey (online)  Isikdag et al. (2008) examined the applicability of BIM (IFC in particular) in a geospatial environment to facilitate the data management in location selection and fire response management processes through the use of multi-model geospatial information. Through this study, it show that it possible to transfer (high level of geometric and semantic) information acquired from BIM into GIS environment.  There are two outcomes from this study that limit the implementation and representation of BIM in the geospatial context, as well as the section describing the design and physical implementation. This study also found that during the process of model transformation, two forms of data mismatches occur between BIM and the geospatial context. It corresponds to the geometric and semantic inconsistency.

MODELING THE BUILDING
Through this study, Autodesk Revit® 2017 was used to created 3D building model. This Autodesk Revit® is developed for the purposes of building information modeling (BIM) application. In Revit® BIM software, there are few design features that can be use such as architectural design, MEP and structural engineering and also construction. Each template of the design features is difference based on their requirement. Basically, Revit® can support multidiscipline, collaborative design process that provides a complete architectural design and documentation solution, started from the planning process until the finishing project (Revit®, 2017). Figure 7 show the BIM model in Autodesk Revit®. This BIM model contains all the information that needed in the construction project.

BIM DATA CONVERSION
Besides of Revit® can supports multidiscipline design process, the other reason for choosing this software is Revit® also provide various type of exporting format such as CAD's format (dwg, wxf and dgn), gbXML, IFC, SHP and etc.. The focusing for this paper is exporting data using CAD format (.dwg) and IFC format. IFC was created by IAI for data interoperability purposes between different parties in Architectural, Construction and Engineering (AEC) (Eastman et al., 2011;Laakso, 2012). The objective of these tests is to investigate the result of BIM-GIS integrations.

Importing Data to ArcGIS
ArcGIS is a platform for organization to create, manage, share and analyse the spatial data (ArcGIS, 2017). One of the ArcGIS an advantage is this platform can read and analyse the CAD formats that have been exported from Revit®. Also, by using FME software, we can perform BIM format data into ArcGIS.

CAD Format (DWG) to ArcGIS 10
By using ArcGIS interoperability tool, CAD format (.dwg) can be read and manage as layers in ArcGIS software. Through this tool, CAD format will be categories in several layers such as annotation, multipatch, point, polygon and polyline. But, this data conversation still not offers the solution for the data integration because it only represents the geometric data, not including the semantic data.

IFC to ArcGIS
For IFC format, FME software was used to convert the data into ArcGIS. Figure 8 shows the data in IFC format. This 3D data is originally from Revit® software. By using FME, the 3D building data will be imported as group of layers which identify each entity in the building. Each entity contains the geometric and also sematic data. Figure 9 show the results of the conversion process from BIM into GIS. ArcScene software was used to visualize the model. In this model, querying process can be performed.

ANALYZING DATA CONVERSION BETWEEN BIM, IFC AND GIS.
After completed the process of conversion data between BIM (AutoDesk Revit®), IFC and GIS data, the next step is the validation of the model in term of data consistency (geometric and semantic data). This validation was focus on comparison of geometric and semantic data in each platform.
First validation is focusing on geometric data. During the process of conversion between BIM into IFC, there are some errors occur such as the "Flat Roof" in BIM (AutoDesk Revi®t) represents "Slab" in IFC (FME) and GIS (ArcGIS). Figure 10 show the validation of data inconsistency of geometric data after conversion process. In IFC platform, "Flat Roof" becomes one of entity in "Slab" elements.
(a) Flat Roof element was developing in AutoDesk Revit®.
(b) Flat roof in roof elements in BIM becomes Slab elements after conversion BIM to IFC.
(c) Flat roof elements become Slab because of the process conversion from IFC into GIS. Figure 10. Comparison of data inconsistency on geometric data after conversion process from BIM into IFC.
Second validation is on semantic data. There are some semantic data missing during converting BIM into IFC and GIS. In Figure 11 show the validation on semantic data after conversion process.
(a) Wall attribute in BIM platform.
(b) Wall attribute in IFC platform.
(c) Wall attribute in GIS platform. Figure 11. Validation on part of data semantic missing during conversion process from BIM into IFC.
Through Figure 11, data inconsistency on part of data semantic can be seen. For example, in BIM platform, there is information about the level of the wall but during the translation process; the information is missing in IFC platform. In IFC platform, there are only several information that been carry forward from BIM such as the name of the element, object type, and ID. The rest information is missing. From Figure 11, we can see that wall attribute in GIS platform is same with wall attribute in IFC platform. Table 5 below show the possible reasons for occurrence of the data inconsistency during the test was conducted.

Reason for data inconsistency Geometric Data
Mismatch of data element (flat roof in BIM become slab in GIS platform) It happen because of the characteristic of flat roof and slab is quite similar in GIS platform.

Semantic Data
Missing of data semantic during data exchange process.
During process of exporting data in Revit into IFC, it's did not bring along the BIM information because of criteria selection. Table 5. The possible causes of the inconsistency during exporting process.
For part of visualization, BIM model in AutoDesk Revit® can view their material of the elements but in IFC, the material cannot be shown because IFC is designed for AEC purposes and does not support the texture visualization. In the term of storage, for the same model; the storage for IFC platform is biggest than BIM and GIS. Although IFC did not carry all semantic information from BIM (AutoDesk Revit®), the storage is still bigger; IFC is 8378 kb while BIM is 8060 kb.

SUMMARY
The paper provides an initial understanding of data accuracy of incorporating BIM data into GIS framework. Since BIM applications are still an ongoing research, many aspects would need to be investigated and assessed to explore for functionality possibilities of BIM data to be incorporated in GIS framework (including spatial and semantic data). Through this paper, a few issues on part of data consistency have been found especially on part of semantic data. In order to solve this issue, a few aspects such as the data format and the criteria selection for the semantic data need to be verified back. For future research, more advanced data analysis such as topological analysis (Salleh and Ujang, 2018), 3D spatial urban data management (Azri et al., 2016) and multi-criteria site selection analysis (Kaya et al., 2020;Mohd et al., 2016) can be carried out based on the BIM-GIS integration.
Isikdag, U., Underwood, J., Aouad, G., 2008. An investigation into the applicability of building information models in geospatial environment in support of site selection and fire response management processes, Journal of Advanced Engineering Informatics, 22(4), 504-519.