Integration of BIM with Facility Systems to Support HVAC Troubleshooting AECbytes Viewpoint #73 (March 12, 2015)

Semiha Ergan, Assistant Professor, New York University, Polytechnic School of Engineering
Xue Yang, BIM360 Implementation Consultant, Autodesk Inc.

How often do you feel like reporting a “too hot” or “too cold” complaint when you are at your office, in the gym, or in a social setting? We can hear you loud and clear. Corrective maintenance for “too hot/too cold” calls is still among the top complaints of building occupants. Frequently, we see facilities personnel running around and chasing ductworks and HVAC components under ceiling tiles while responding to work orders.

The main challenges in the current practice are the lack of access to the required facility/system information in a timely manner and the large search space of possible causes in a complex HVAC system. A recent project that I have led with my student, Xue Yang, addressed these challenges by leveraging BIM and other data sources for facilities information to support troubleshooting of HVAC related problems.

Challenges in the Current Practice of Corrective Maintenance

Troubleshooting HVAC related problems and pinpointing the cause for a variety of occupant complaints such as “too hot/cold, too humid, too noisy, too stuffy air,” are not easy!   It is typically the case that HVAC mechanics would need to check several HVAC system components as these components could result in the same symptom for a reported problem. It is also common to see HVAC mechanics going back and forth between FMS archives, command centers, and facility floors to get access to facility/system specific information that needs to be collected as evidence.  

Given these, it is not unexpected that HVAC mechanics make decisions with limited understanding of problems, resulting in delays and misinterpretation of root causes, unless they are quite familiar with the spaces at hand and seasoned HVAC mechanics.  We have developed an automated approach to bring efficiencies to the process of finding applicable causes of a reported problem and retrieval of required facility information for a given work order in a facility.

Overview of the Approach

 The heart of the approach is in a data schema that extends IFC and integrates as-built design, BAS (Building automation system), and CMMS (Computerized maintenance management system) data required for troubleshooting in a single repository.  The approach utilizes the data schema to run three modules. Everything starts when an IFC file is loaded that contains HVAC system information in an as-built BIM. Figure 1 shows a snapshot from the prototype implemented by the research team. We implemented the prototype using the Java programming language and interfaces with IFC Viewer to display model content. 

Figure 1. Input IFC file that includes facility and HVAC system information. (a): Input requirements. (b) and (c): Entering a work order as an input. (Image courtesy: ASCE, JCCE)

An HVAC mechanic can run the following modules to reduce the search space for the work order and identify a subset to check.

Module 1: Work Order Contextualizer

A typical work order includes a date/time stamped complaint (problem type) for a specific room (space ID). However this information is not enough for HVAC mechanics to figure out what the cause of that problem might be. A cause could be a component in an HVAC system (e.g., a fan, a heating coil), or space related (e.g., space type, as-designed number of occupants). HVAC mechanics need to know the following information:

  • What type of an HVAC system conditions that space (e.g., all air, all water, etc.),
  • How widely the problem is spread in spaces (e.g., single zone, multiple zone),
  • What control system (e.g., pneumatic, electric) runs at the backend, and,
  • Whether there is a pattern in the reported time-frame for similar work orders in that space.
Answers to these questions form the context of a work order and change the rationale of HVAC mechanics in considering or eliminating certain components as potential causes of a reported problem.  All the answers to these questions are automatically reasoned and extracted by the Work Order Contextualizer algorithm. 

The algorithm first convert’s IFC’s representation of HVAC systems into a graph-based representation, generates graphs of the HVAC system components in supply and return directions, and uses those graphs to differentiate whether a certain HVAC component in a system belongs to the central system or terminal systems (Figure 2). Such information is then used in subsequent modules to identify applicable causes for a given problem type.

Figure 2. An example graph generated for two spaces, differentiating and labeling components that belong to a terminal system (i.e., a system between conditioned spaces and a primary air distribution system; a VAV box is an example) and components that belong to a central system (i.e., primary distribution system, such as AHU). (Image courtesy ASCE, JCCE)

Module 2: Applicable Cause Finder

When an HVAC mechanic knows the context for a work order, then it is easier to eliminate components that are irrelevant to the problem reported. Module 2 utilizes a set of matrices that map a generic list of problem types to HVAC component categories and automatically identifies generic applicable causes for the work order context identified. However, the final list of causes identified will be generic and not specific to a given facility and hence further needs to be customized.

Module 3: Refiner

At this step, HVAC mechanics would know what types of components and factors could be the cause of a reported problem, but they would need to figure out whether all such components exist in the specific system they are troubleshooting and what the corresponding values are for the parameters of these components.

The algorithm utilizes the underlying data schema to integrate data from BAS (e.g., set points, current sensor readings), CMMS (e.g., what work orders were issued for the same space in the past), and building as-built design information (e.g., location of HVAC components, topological relationships of components) to generate an integrated BIM for facility use. Next, the algorithm checks all the instances of components and traces through the system to refine the applicable list of causes listed.

The resulting refined applicable causes are provided to HVAC mechanics in visual forms. Whenever an HVAC mechanic selects an identified component as a cause, corresponding information about that component is shown, including the location information in floor plan, in 3D, color coded CMMS work order data, BAS data, and control relationships in 2D schematic diagram. For example, Figure 3 shows this information for a selected supply air fan.

Figure 3. Output of the approach, which highlights a supply fan and its parameters.

Test Results and Conclusions

We wanted to understand how generically this approach can help HVAC mechanics who work for different facilities with different types of HVAC systems.  We used seven as-built BIMs that differed mainly in HVAC system and control types and we changed the work order context during tests (Figure 4).

Figure 4. Testing the approach with the models of seven different HVAC systems.

Findings showed that the approach was able to reduce the search space of applicable causes by 68% on the average and achieved high precision and recall rate regardless of the work order contexts used.

So, how does this approach help the practitioners? Well, HVAC mechanics can use the approach before they go to the field and stat their investigation. Instead of tracing duct works and running back and forth between available data sources, they can use the approach to retrieve the information they need to check, reduce the search space of checking and tracing HVAC components, and plan ahead of time for their field trip by focusing on the right set of components.

Such an approach can help HVAC mechanics to pinpoint the right set of causes for a given HVAC related problem, so that they don’t miss the root cause or waste time in tracing and locating components on the field.

Further Work

What is next? My current research group at NYU is building on this work and investigating how such an approach can be made mobile for HVAC mechanics to check the required information on the fly.

About the Authors

Dr. Semiha Ergan is an Assistant Professor at New York University, Polytechnic School of Engineering in Civil and Urban Engineering Department. She holds a Ph.D. degree from Carnegie Mellon University. She has been leading research projects in relation to facility informatics and visualization, utilization of information models and advanced visualization technologies for improvement of AEC/FM processes, where she has over 50 publications. She is the director of Facilities Informatics and Visualization Lab at NYU and an active member of the AEC industry.

Dr. Xue Yang is 360BIM consultant at Autodesk Inc. She holds a Ph.D. degree in Advanced Infrastructure Systems from Carnegie Mellon University, where she was working with Prof. Ergan. She holds a MS degree in Civil Engineering from Tongji University, China. She is an enthusiastic researcher and dedicated to seeing the wide spread use of BIM technology in the FM industry.

AECbytes content should not be reproduced on any other website, blog, print publication, or newsletter without permission.

Related Archive Articles

  • Got Macros? Scripting and Coding for BIM
  • In this article, Karen Kensek, Assistant Professor in USC's School of Architecture, advocates the writing and use of macros in AEC firms to improve the efficiency of BIM, which "out of the box" is not synchronized with the way firms work.
  • Notes on the Synthesis of BIM
  • In this Viewpoint article, architect and educator Randy Deutsch assesses the design and construction industry's current efforts to produce and analyze vast amounts of information and proposes a call for a synthesis of its technology and collaborative work processes.
  • BIM for Facilities Management
  • The current status and future outlook of the application of BIM in the field of FM (facilities management), including the FM capabilities of FM:Systems, ArchiFM, Bentley Facilities, Onuma Systems, and EcoDomus.