BIM BASED REQUEST FOR INFORMATION CLASSIFICATION AND DISTRIBUTION: TWO RESIDENTIAL TOWER CASES

Request For Information (RFI) is a communication tool to facilitate resolution of, or to clarify, construction issues. Traditionally, construction teams generate RFIs, but with BIM they are anticipated being generated before construction by designers. The aim of this study to improve the understanding of how design conflicts happen through the lenses of RFI and motivate a proactive design review approach. The study was developed as a case study of RFI distribution and classification in the design of residential towers using a Virtual Design and Construction (VDC) approach. Semi-structured interviews were conducted with the virtual construction team coordinator. The interviews were transcribed and transformed in process map, especially, explicating how Request For Information was generated and how the interaction among the participants occurs. A documental analysis was performed on coordination models characterizing the type of RFI, its location, and quantity distribution within design discipline. The design process map drawn shows a partially implemented VDC where design developed with traditional CAD tools benefits from a design review process developed with virtual prototyping and 3D coordination with BIM tools. The distribution patterns of RFI quantities denote where the greatest effort in the review was, exposing design complexity. The analysis here presented may be applied to new situations other than those studied in order to promote learning within and metrics for VDC.

Design errors are a problematic issue to the Architecture, Engineering and Construction (AEC) industry, and they can be avoided through the adoption of design managing actions (LOPEZ et al., 2010).In this context, the design coordination has, as its main expectation, the reduction of execution failures due to errors or inconsistencies in design.BIM can help AEC industry to find potential problems before construction starts (SACKS; BARAK, 2006).Zuppa, Issa and Suermann (2009) identified that BIM is frequently perceived of as a tool for visualizing, coordinating, and improving AEC work and productivity.Construction organizations need to openly acknowledge errors presence so that "learning from errors" can form an integral part of an organization's fabric (LOVE;SMITH, 2016).
Request For Information (RFI) is a communication tool to facilitate resolution of or to clarify design issues.RFIs can indicate design errors and initiate improving opportunities prior to construction.The cost per RFI may overcome US$ 1,000.00 each, based exclusively on the technical and administrative reviewing associated cost (HUGHES et al., 2013).Lopez et al. (2010), developed a study in order to understand how and why RFIs happen and which strategies can be thought to solve them.According to Lopez et al. (2010), the design error classification offers the bases to considerate the strategy adjustment to avoid and minimize errors.
Therefore, the aim of this study is to evaluate qualitatively the Request For Information in order to improve the understanding of how design conflicts happen and motivate a proactive identification approach.The study was developed as a case study of RFI distribution and classification in the design of residential towers using a Virtual Design and Construction-(VDC) approach.

Virtual Design and Construction (VDC) and Building Information Modeling (BIM)
BIM is understood as a set of policies, processes and technologies to promote a methodology to manage the design data and the construction design data, essential to the digital format in the whole building cycle (PENTTILÄ, 2006;SUCCAR, 2009).
According to Chua and Yeoh (2015), BIM provides the technology to share information and promote collaboration across organization and phases.BIM has gained increasing acceptance in the AEC industry.However, its adoption must be accompanied by transformation in processes.VDC represents this transformation.
VDC was first coined by Kam and Fischer (2004).Chua and Yeoh (2015) describe VDC "as a concept or approach to build, visualize, analyze, and evaluate project performance virtually and early before a large expenditure of time and resources is made".VDC is adopted in this study through this perspective.
Studies point how wide the BIM service delivery range is, such as, design coordination, clash detection and construction work sequencing (AZHAR et al., 2008).Clash detection using a BIM 3D coordination tools is able to identify several conflicts that may go unnoticed by professionals who perform their task manually.On the other hand, manual clash detection identifies conflicts that could not be found through software in cases were clashing objects are not modeled in the BIM (LEITE; AKINCI; GARRETT, 2009).Results show that the combination of clashes identified automatically, as well as those identified in the field captures the largest possible number of clashes.3D coordination tools require virtual prototyping, that is, BIM models.
BIM models from all disciplines can be brought together and compared, and with this, conflicts and constructability¹ problems are identified before they are detected in the field (EASTMAN et al., 2011).

Request For Information (RFI)
It is while the work takes place that several unpredicted design questions come up and, in general, Request For Information "are created by subcontractors and transmitted to the general contractor, and then to the design team for comprehensive review" (CHIN, 2009, p. 258).
Traditionally, construction teams generate RFIs, but with BIM the Requests For Information are anticipated being generated before construction by designers.NATH et al. (2015) suggested that the RFI submission and approval process may be improved by using BIM Collaboration Format (BCF), which permits the users to share the encoded messages among different BIM platforms about the issues identified in the BIM model.Therefore, reduction of the RFI numbers in the construction site, do to prior treatment, is an example that highlights the aggregated value of BIM use (LEITE et al., 2011).
Despite meaningful capacities of BIM tools, modeling and interoperability deficiencies remain without solution.Thus, it is important that BIM companies keep improving the modeling and interoperability options to facilitate the interdisciplinary collaboration in the creation, review, update and reuse of 3D model information (ARAM; EASTMAN; SACKS, 2013).

Information Delivery Manual (IDM)
Researches point the main source of wasted information is inadequate information exchange and it is necessary to fully understand the interactive nature of the process of design (AL HATTAB;HAMZEH, 2013).Current design coordination strategies neglect the role of appropriate information flow and the communication among the design participants (AL HATTAB;HAMZEH, 2015).
According to BuildingSMART (2010), the establishment of an Information Delivery Manual (IDM) aims to provide a reference embracing the Request For Information for AEC industry, identifying the processes that demand the exchange or sharing of information among the design participants.
By following the patterns of IDM methodology, maps of processes and exchange models can be created.Antunes and Scheer (2014) demonstrated how to identify which information was necessary for the execution of a structural design with IDM methodology.

Design Coordination
In building construction, the process of design coordination is more often carried out by comparing or matching technical drawings of different disciplines in a design coordination meeting (RILEY; HORMAN, 2001).Bellan and Fabricio (2010) analyzed managing actions, coordination procedures and tools, and observed that the professionals must realize great effort of abstraction to align the two-dimension technical drawings to understand and preview the 3D work reality.
Case studies showed a transition moment between a conventional process and the new process of BIM coordination, which reveals itself as a very efficient means for a design team, as well as for coordinators and contractors (FARINA; COELHO, 2015).
According to Tommelein and Gholami (2012), BIM coordination is a process of integration of interdisciplinary models involving interferences checking in order to detect and solve conflicts, mitigating future likely problem.
Although BIM is considered helpful in improving design quality by eliminating conflicts and reducing rework, BIM also has to be applied throughout the project for construction quality control and efficient information utilization (CHEN;LUO, 2014).In consensus, Leite;Akinci and Garrett, (2009) observed that, in a case study of mechanical, electrical and plumbing (MEP) design coordination, the combination of manually identified design conflicts in coordination meetings, with those automatically detected, as well as with those identified in the field, allowed the highest performance of the coordination process.
A research-action developed in a building company showed BIM clash detection carried in the construction phase, after the design phase, allowed to check architecture designs, structure and building systems and the sharing of all conflict information so the designers could adjust their 2D designs (NEIVA NETO; FARIA; BIZELLO, 2014).However, the construction model development required a robust BIM construction system component library, to be used by different parties involved in the whole construction cycle.These BIM components should be developed taking into account requirements for different BIM uses from modeling to cost estimation, 3D control and planning (NEIVA NETO; RUSCHEL, 2015).
Considering code checking as a tool to improve design quality in design coordination, Kehl e Isatto (2015), observe that the geometric precision of the model and the associated information influence the feasibility of its application.Besides, not all the rules are automatically verifiable or are worth the effort to become so, especially the qualitative ones, being necessary the manual verification (KEHL; ISATTO, 2015).

Method
This study is exploratory and descriptive.The research question is "How are Request For Information distributed referring to residential design?"To answer the research question, it was chosen the Case Study strategy, as outlined by Yin (2001), therefore the results are qualitative.
Figure 1 shows the research outline, which comprises five steps.Initially, a bibliographic review was carried out in the study theme.Then, the study delimitation and the elaboration of the protocol for data collection and analysis took place.Data were collected from documental analysis and interviews.Data analysis developed upon design process map observed and classification of RFIs Finally, the report is resumed in this article.To maximize how useful information could be to the research goals, the case study selection considered an information-oriented choice (TAKAHASHI, 2013).This directed the case study to be applied in a Brazilian design company known for its tradition, design competence and market leadership in VDC.Two multifamily residential buildings, 3D coordinated and virtually prototyped by this company, were selected as units of study.
Therefore, the data collection was based on multiple evidence sources: documental analysis and interviews.The former happened through virtual construction models composed by architectural, structural and installation models, as well as, Request For Information, that is, the coordination model.
The two residential multifamily building ventures had available 3D models based on Naviswork Freedom software with reports and issues summary of the design review process.Altogether, there were six navigation models, three for each unit of study, that is, residential multifamily buildings.Figure 2 shows an example of the Naviswork interface, which supported the analysis.On the left, it can be verified insufficient distance to connect plumbing, while on the right, there is a RFI panel that allows navigating by the model through all the occurrences reported by the virtual construction team coordinator.Each RFI was matched to the prognostic basis, proposed by Dantas Filho et al. (2016) as described on Table 1.et al., 2015, 2016).This classification scheme is detailed in Dantas Filho et al (2015).The analysis categories are based on RFI categories such as: design correction, information divergence, design omission, design verification (Tabel 1).

RFI Categories Definition 1 Design Correction
Problems associated to the solution execution presented at the design

Design Omission
The absence of necessary spefic design for some areas

Design Verification
Low complex issues that make design confusing.Opportunity for design enhancement

Information Divergence
Design mistake associated to lack of attention, in the same discipline, two or more different drawings Source: Dantas Filho et al. (2016) It is essential to highlight the adopted prognostic basis here is exploratory, qualitative and not generalized.What has been proposed so far is a picture of the variables and how they interfere in the design process.
Semi-structured interviews were conducted with the virtual construction team coordinator.The interviews were transcribed and transformed in process map, specially, explicating how Request For Information were generated and how the interaction among the designers occurs.

Case study description
The company where the case study was applied has 500.000m² area of VDC, distributed in 15 multifamily residential buildings, of 10 different incorporators situated in the northeast region of Brazil.The service delivery scope of this company excludes authoring, including virtual prototyping to support design coordination and material quantity extraction.
The units of study (Figures 3 and 4) were multifamily residential buildings each with more than 20 thousand square meters, designed by the same design team, whose characteristics are listed as follows in Table 2.The case study A was composed of one tower and the case study B of two towers.Figures 3 and 4 were taken from the Naviswork models that supported data collection.The interview analysis, with the virtual construction team coordinator of these case studies, showed the virtual prototype and 3D coordination process map in three stages (Figure 5) involving: the developer and investor, the virtual construction team, and the design team.In the first stage, architecture and structure models are prototyped and analyzed by the virtual construction team and a coordination model is exchanged with the developer and with the design team.In the second stage, installation models are prototyped, added, and analyzed by the virtual construction team and again a coordination model is exchanged with the developer and with the design team.
In the third stage, the virtual construction team updates the composed model, with updated solutions delivered by the design team considering the requested design changes.The composed model is validated and the virtual construction team delivers the final coordination model to the developer (investor) to be used in the construction site in order to promote improved practice and efficiency.
During each stage, the developers (and investor) and designers receive a coordination model in the Naviswork Freedom software-based navigation model with the composite model and the Request For Information resulted from the analysis of design coordination.In each phase, the designers are in charge of revising their corresponding design solution considering the RFI received.New design versions are created and, then, evaluated by the virtual construction team.The final report is delivered to the developer (and investor) and designers, including the unsolved RFIs or the new issues that came up because of the suggested changes.
In relation to the virtual construction team, three different roles were observed: the specialist, the modeler and the coordinator.The specialist checks if the design documents are complete and organizes the modeling process.The modeler creates the BIM models based on received CAD files, documents and modeling plan.The modelers may reports on design inconsistencies observed in the modeling process.The coordinator performs the constructability¹ analysis issuing RFIs.These RFIs can be due to: incomplete documentation, design inconsistencies observed in the modeling process and clash detection errors.This process is known as the constructability analysis flow¹.
In traditional design processes, without BIM or VDC, RFIs are generated along the building execution.Therefore, the cost of making changes dramatically increases as the project progresses (CURT, 2004).In the case studied here, the RFIs are generated before construction starts, allowing the designs to be revised as a whole.For this reason, the cost of making design changes are smaller.

RFI Distribution
432 RFIs were identified and analyzed in the case studies.Some of them were related to only one discipline, for example, either architecture or structure or installation design errors.Others were related to more than one discipline, for example, issues between architecturestructure, architecture-installation, and structureinstallation.The virtual construction team coordinator executing constructability analysis generates these issues.
When analyzing the RFIs in numbers by design types comparing both cases studies the following was observed.
Ordering RFIs in descending numbers, considering the resulting design types ordering of case study A as reference and comparing with the corresponding ordering of case study B it could be observed that different RFI distributions are presented (Figure 6).However, it is observed a decrease of RFI quantity from case A to case B in 4 out of 5 isolated design issues (Figure 7) and in 2 out of 3 interface design issues (Figure 8).However, it was possible to extract a pattern of distribution of RFIs among the number design issues related to isolate discipline, been the Hydro Sanitary Installation and Structural design the disciplines that most benefits with 3D coordination in the case study of residential towers (Figure 7).
It was also possible to extract a weaker pattern of distribution quantities of RFIs among the design issues related to the interface of disciplines, been the issues related to the interface of Architecture and Structure the design interface that most benefits with 3D coordination in the case study of residential towers (Figure 8).
It was observed a decrease of RFI quantity from case A to case B in most of design issues (Figure 7 and 8).The interviews pointed that designs were done by the same architecture, structure and installation professionals, with the case A design review realized before of case B design review.That evidences a learning process that may have contributed not only to the virtual prototyping and reviewing process of the virtual construction team, but may also have helped the design team avoid possible RFIs in their own design solution, highlighting the disciplines of Hydro Sanitary Installations and Architecture, which presented the greatest decrease of RFIs from case A to case B (Figure 7 and 8).
Figure 9 shows the percentage of RFI in relation to the building vertical location.It is clear that in both cases the lowest RFI percentages are on the first and second basement and roof levels.While the ground floor, the first floor with recreation items and the typical floor concentrate a greater amount of the RFI total.
It is verified the replication of this distribution for cases A and B. Intuitively, it is important to solve typical floor questions, that is so because, in residential building, this floor execution is going to be repeated over and over.However, the RFI distribution in these locations points to the importance of also solving the ground floor and the recreational first floor questions.Ground floor and the recreational first floor are about large areas with function diversity, to where all tower installations converge.

Design Omission
How is unevenness going to be fixed?(ramp?) Ground Floor Anticipate Slab between the V318 and V319 beams 9 Is the floor going to be over the landfill or is there going to be slab?Ground Floor How is ventilation going be in the bathrooms?10 Analyze, what is the ceiling final touch going to be like?Basement How is trash unevenness going to be fixed?Ground Floor 11 Analyze, will there be any filling?Type Define Square Frames Sauna Basement 12 How is the metal structure going to be fixed on the front?
Brickwork without structural support Ground Floor 13 Point shelter to control elevating platform Ground Floor Anticipate electric feeding for all outlets in the Basement 14 Will there be drainage?ss2

RFI Classification
This section is about the exploratory RFI analysis of the case studies.Table 3 presents the classification of observer RFIs into the categories of design: correction, omission, verification and divergence, followed by the summarized description of each issue.Examples of each type of IRF categories will exemplified in a figure and narrative.
Figure 10 refers to a design correction RFI, seen in both case studies.It can be noticed that a column is in conflict with a basement parking space.This kind of RFI can result from conflicts that can be identified either manually or using automatic clash detection.Some examples of RFIs classified as correction and present in both case studies can be seen in Table 3, lines 1-8.In Figure 11, is related to a RFI classified as design omission due to the absence of design structural definition about how the floor in the highlighted area is going to be executed.In this case, there is the absence of design definition about if there will be a concrete ballast over the ground or if there will be a floor slab.This information absence was identified by manual detection and it reveals itself as a challenge of automatic detection rule creation.
Other examples are shown in Table 3, lines 9-16.Similar results were obtained from the two cases, it can be said that a replication occurred indicating a pattern.The omission, verification and divergence RFIs can be a challenge for automatic detection rule creation, because they are not resulted by geometry clashes.Then, manual detection stands up as a method of identification.
The analyzed designs were carried out by companies that have stability and quality in the market, this way, the results can be applied in other companies as well.

Limitations of the research
The study was developed as a case study; therefore, findings cannot be generalized.However, hypotheses can be drawn.The study was applied to high-rise residential buildings.Thus in another typology, for example, horizontal housing developments, the distribution of RFIs by location may differ.
However, the realization of case study does not propose to enumerate frequencies and make a statistical generalization.But rather indicate that the theoretical proposition that RFIs can be classified contributing to understanding the problem and promote learning.

Conclusion
The qualitative research carried out in this work on Request for information shows patterns of behavior adding richness and depth the understanding the design review of residential towers.The design process map drawn with the IDM methodology shows a partially implemented VDC where design developed with traditional CAD tools benefits from a design review process developed with virtual prototyping and 3D coordination with BIM tools.
In relation to RFIs types' distributioncorrection, omission, verification and divergence -it has been observed design issues locations are recurrent between the case studies (Figure 9) and that the correction RFI type is the most frequent (Figure 14).Also it was observed that the verification RFI decreased from case study A to B giving space the divergence type of RFI (Figure 14).This may be due to the learning process promoted by the VDC design review.Design issues were classified as to their causes from literature review by providing a holistic view and identifying issues for creating prevention strategies (LOPEZ et al., 2010).Clashes were classified as to their existence for the establishment of a common language that allows the sharing of lessons learned for the advancement of practice and the development of theory (TOMMELEIN; GHOLAMI, 2012).
In terms of RFI quantities distribution among design disciplines, it was possible to extract a pattern related to isolate discipline design issues, been the Hydro Sanitary Installation and Structural the disciplines that most gained with 3D coordination in the case study of residential towers (Figure 7).A weaker pattern of quantities distribution of RFIs among the design issues between disciplines indicated that issues related to the interface of Architecture and Structure most gained with 3D coordination in the case study of residential towers (Figure 8).The patterns of RFI quantities distribution behavior shows where improving design effort must be, exposing design complexity.New concepts are needed to understand how RFIs are distributed and how they occur Case A Case B so that this enables a learning process of the parties involved.However, some missing information, such as classification as cause or existence, would contribute to the development of prevention strategies as proposed by Love and Smith (2016).
It can be concluded that RFIs from residential tower VDC designs review are susceptible to be distributed and classified based on the interface, location and typology, whether of the type of correction, omission, verification or divergence.Considering that, residential tower designs are examples of environment designs comprised of architecture, structure, and other installations designs, then, the analysis here presented apply to new situations other than those studied in order to promote the learning of and metrics for VDC.

Figure 2 -
Figure 2 -Coordination model used for the data analysis and data collection (Work A)

Figure 9 -
Figure 9 -RFI types by building location

Figure 10 -
Figure 10 -RFI Examples of Correction

Figure 12
Figure 12 evidences RFI classified as design verification, present in both cases.It is possible to observe the structural design is not wrong, but the fact of the beams being the same height, that makes the roof installation unviable without the proper final touch of the edge beam.Other examples are shown inTable 3, lines 17-24.

Table 2 -
Case study characteristics

Quantity of towers 01 02 Quantity of floors 19 19 Quantity of units 38 76 Quantity of underground floors 02 02 RFI quantity 252 180
Source: The authors

Table 3 -
Case study's evidences of RFI types

Table 3 ,
lines 25-32.Figure14presents the distribution of the RFI types identified in each category.It is observed that, in both cases, correction RFIs are more numerous.On the other hand, omission, verification and divergence RFIs, when grouped, are as numerous as the largest classification.