«Quantifying the effect of using BIM and COBie for facility management on work order processing times: a case study Sarel, Lavy1 and Nishaant, Saxena2 ...»
The work order may be a single phase or multi-phase work order depending on the complexity of the system and disciplines involved (mechanical, electrical, plumbing, architectural). The work order may either be completed by the maintenance technicians or put on hold for later completion, if new parts are to be ordered. Once the work order is completed by the technicians, they record the number of hours worked on that work order. These hours are examined at the campus level by the facility managers before updating the status as complete. Texas A&M University System has outsourced FM services for all TAMHSC campus buildings, except the McAllen campus of TAMHSC, to a single organization. As of June 2013, the outsourcing company employs maintenance technicians for all the campuses. Different technicians are employed at different campuses due to the large geographical distance between campuses. Since the maintenance activities for all TAMHSC campus buildings, except one, are carried out by a single organization, it is reasonable to analyze the performance of the campuses that use BIM for FM vs. the ones that do not use BIM for this purpose.
Data segregation For analysis purposes, the researchers selected work orders for a duration of one year, ranging from September 2013 to September 2014. Since FM services for the McAllen campus are not performed by the same company, work orders from this campus were excluded from the study.
Data fields considered for this study included the campus in which the work was performed, description of the work order, work order category (i.e., preventive, corrective, event, etc.), the actual number of hours taken to complete the work order, and work order entry and completion dates. However, work order completion date is not a reliable data field for research as the end date of a work order is recorded when the maintenance supervisor marks a work order as “complete” in the CMMS. Since the supervisor may not update the status of the work order on the actual day it was completed, the end date cannot be used as a reliable data point in this study.
Upon initial examination of the data, a total of 7,429 work orders were observed for the time period ranging from September 2013 to September 2014. There were 2,111 work orders with “zero” number of actual hours worked and as such, they were not included since they represented inaccurate data entered into the system. In addition, all work orders for the Temple campus had zero actual number of hours worked, leading the researchers to exclude that campus from the study.
Another significant observation was that there were no work orders for the Corpus Christi campus, and therefore, this facility was also excluded from the study. The campuses which were finally included in the study are: Bryan, Round Rock, College Station, Houston, Kingsville and Dallas.
This classification of work orders is illustrated in Table 1.
Data Analysis In order to study the impact of BIM on different categories, the work orders had to be classified into separate categories. After consulting with the facility manager of TAMHSC and Mr. Hyde Griffith from Broaddus & Associates, it was decided that the work orders for maintenance of MEP systems and architectural components should be analyzed separately. The researchers sought input from the facility manager of TAMHSC for segregation of work orders. Based on the work order descriptions, mechanical category was assigned to the maintenance of air handling units, exhaust fans, etc.; the electrical category was assigned to the maintenance of electrical systems, such as lighting, emergency generator, etc.; the plumbing category was assigned to maintenance of plumbing systems, such as water leaks, problems with drainage of water, plumbing fixtures, etc.;
and the architectural category was assigned to the maintenance of building components, such as doors, windows, walls, etc. The other category was used to exclude work orders not related to Building Information Modeling. These included work orders such as event setup, i.e., setting up chairs and tables for functions, telecom system complaints (like changing caller identification numbers), picking up supplies, cleaning areas in the building, moving furniture, etc. Due to time constraints, it was not possible to have all the data segregated by the facility manager of TAMHSC, hence a smaller list of “common” maintenance types was compiled by the researchers. This compact list of work orders was segregated into five categories, namely: mechanical, electrical, plumbing, architectural, and other by the facility manager of TAMHSC. This list was further used as a guide by the researchers for segregating the remaining work orders. After segregating the data, there were 954 work orders in the mechanical category, 1,084 work orders in the electrical category, 636 in the plumbing category, and 724 in the architectural category, for a total of 3,398 work orders. This segregation of work orders is illustrated in Table 2.
Table 2: Work orders segregation for the t-test analyses # Category Number of work orders (including Number of work orders (excluding combined work orders) combined work orders)
1. Mechanical 954 530
2. Electrical 1,084 646
3. Plumbing 636 550
4. Architectural 724 660 Total 3,398 2,386 Work orders for preventive and corrective maintenance were combined to conduct the statistical analysis as there was insufficient data in the preventive maintenance category to perform a separate statistical analysis. Work order times from the Bryan and Round Rock campuses were combined into the BIM user category and were compared with the work order times for the College Station, Dallas, Houston and Kingsville campuses, which were combined into the Non-BIM user category.
A statistical analysis was carried out with a 95% confidence interval on the time spent on processing work orders by using a two-sample t-test.
Results and Discussion Statistically significant differences were found to exist between the work order processing times of BIM users and Non-BIM users in the mechanical and plumbing categories, at the.05 level of significance. Results show that in both cases, BIM users have a higher mean processing time. In the electrical and architectural categories, there was no evidence of a statistically significant difference between the work order processing times of BIM users and Non-BIM users. The results of the t-test are summarized in Tables 3 through 6 and a graphical comparison of means is illustrated in Figures 1 through 4.
After observing the results of the t-test, the researchers examined the data in order to understand the reason for the results. Upon examination of work order descriptions in the data entries, the researchers found that there were no standard rules for recording work orders in all facilities.
Numerous instances were found where work orders for multiple systems/ components of a building were combined into a single work order in all four categories across all campuses. Examples of combined work orders in the architectural category included a combined work order for renovation in four different labs, modifications in four separate rooms, etc. Instances where different categories of maintenance activities were combined together into a single work order were also observed. One such work order combined carpeting, painting, replacement of modular furniture, and electrical work in the waiting area of a floor. In the mechanical category, there were instances of maintaining multiple exhaust fans, fan coil units, fume hoods, air handling units, etc. through a single work order. In the plumbing category, work orders were combined for leaks in different rooms, inspections of multiple water heaters, inspection of multiple sprinkler valves, etc. In the electrical category, there were multiple combined work orders for fixing, inspecting and testing of lights. Such work orders provide ease of recording but make data validation inconsistent because there is no way to find out how many lights were actually inspected/fixed, how complex was the repair process, and how much time was spent walking inside a building or multiple campus buildings within a campus when fixing the lights. This inconsistency can be observed clearly by examination of the combined work orders for monthly check of lighting placed in the Houston campus. The time taken to check the lighting varied considerably from 59 hours in the month of January, 2014, to just 5 hours in the month of April, 2014.
Inspection is a unique category of maintenance that, if performed individually for each component, would become cumbersome and result in inefficiency. Another factor to be considered for inspection activities is that MEP systems are complex in nature, since a system is composed of machines connected together with electrical wires, water pipes, or ducts. This increases the complexity of inspection activities as deficient performance of a system may not be due to suboptimal performance of any one machine. Hence, grouping inspection activities into a single work order makes practical sense; however, a standard procedure for recording inspection activities was not followed across all the buildings/campuses of TAMHSC to enable consistency in data validation. A standard procedure for recording work orders is necessary because the number and location of components varies from one campus to another. This led the researchers to make a critical decision about excluding combined work orders for inspection activities from the data set and repeating the statistical analysis across all four categories: mechanical, electrical, plumbing and architectural.
After excluding combined work orders, there were 530 work orders left in the mechanical category, 646 in the electrical category, 550 in the plumbing category, and 660 in the architectural category, bringing the total number of work orders down to 2386 from the earlier total of 3,398 work orders (a 30% reduction). This “re-segregation” of the data is illustrated in Table 2. The t-test results performed with this data show statistically significant differences between the work order processing times of BIM Users and Non-BIM Users in the plumbing and architectural categories, at the.05 level of significance, where BIM users have a higher mean processing time in both cases.
There is no statistically significant difference between the work order processing times of BIM Users and Non-BIM Users in the mechanical and electrical categories. The results of the t-test are summarized in Tables 3 through 6 and a graphical comparison of means is illustrated in Figures 1 through 4.
These results are significant as they seem to contradict the perception of owners and facility managers that the use of BIM may help reduce maintenance work order processing times.
However, deriving conclusions based exclusively on the results of the statistical analysis may not be appropriate. The fact that there are no standard procedures for recording work orders for all facilities (as demonstrated in the combined work orders) may have affected these findings.
Inaccurate recording of information is a hindrance in determining the efficacy of new technological systems. In order to accurately determine the difference in work order processing times, it is crucial for owners and facility managers to establish standard rules and procedures for recording work orders across all facilities.
Figure 1: Mean work order processing times and 95% Confidence Intervals (C.I.) for the mean of mechanical work orders in BIM user and Non-BIM user facilities
4 4.0 3.9 3.8
Figure 2: Mean work order processing times and 95% C.I. for the mean of electrical work orders in BIM user and Non-BIM user facilities
ConclusionsBIM has been readily adopted by the design and construction industry because there are immediate benefits in coordinating among multiple teams and time and money savings can be realized in a relatively shorter span of time. FM on the other hand is a much longer process that lasts throughout the entire life of a building and it may take longer to validate the benefits of using advanced technology for FM. The shift from drawing-based FM to BIM and COBie data-based FM is a relatively new development in the industry and TAMHSC is a pioneer in this field as it is one of the first institutions to adopt this approach. However, the FM industry is slow to adopt technology and any such adoption is fraught with multiple challenges. This research investigated a case study to understand the effects of using BIM for FM on work order processing times. To accomplish this, actual work order data from multiple TAMHSC campus buildings was collected and statistically analyzed.
Existing literature on this topic indicates that using BIM and COBie approach for FM presents significant benefits. The U.S. General Services Administration (2011) states that there should be a reduction in the time required to process work orders. To reinforce this, a study conducted by Broaddus & Associates (Beatty et al., 2013), which was based on surveys and interviews, indicated that there could be a reduction of 8.7% in time spent on the work order process. However, the findings of this current study contradict these two sources. Statistical analysis of the actual work order processing times for the work orders shows that more time was spent on processing work orders by using BIM and COBie data for FM in two categories: mechanical and plumbing. In the second test, where combined work orders were excluded from the statistical analysis, it was observed that more time was spent on processing work orders by using BIM and COBie data for FM in two categories: plumbing and architectural.