AbstractAsset performance evaluations serve as a benchmark that can be used to make asset manufacturer selection decisions, i.e., choosing the brand of manufacturer that provides the highest performance among all brand competitors. Understanding the environmental conditions to which assets are subjected provides facility managers another data point to understand asset performance. This research builds upon a previously published performance-based manufacturer selection metric by investigating the linkage between asset performance and exposure to local climate, using chiller and air handler data from 20 United States Air Force installations. The link between environmental factors such as Heating Degree Days (HDD); Cooling Degree Days (CDD); Solar Irradiance; and a variable that accounts for relative humidity, and asset performance is investigated using analysis of variance (ANOVA) testing and correlation coefficients. The results reveal that most assets, regardless of location, (1) possess a moderate to strong performance; and (2) cumulative climate exposure and asset manufacturer selection influence asset performance. This work highlights the need for facility managers to consider the influence of climate on technical performance and use it as a decision criterion in manufacturer selection.IntroductionFacility managers overseeing the operation and sustainment of built infrastructure assets are tasked to make data-driven decisions throughout the life cycle of assets, often in resource-scarce environments. These decisions begin with selecting an asset to purchase from a manufacturer for use in their facility. This decision is made with expectations about the asset’s performance and longevity. Additional consideration must be given as to the frequency and robustness of a preventative and corrective maintenance program. Throughout the asset’s life cycle, facility managers continue to make decisions up until disposal, at which time they need to replace the asset entirely. All of these decisions and associated costs can be evaluated using Total Cost of Ownership (TCO) models, which calculate all costs incurred by owners of any physical assets over the asset’s lifespan (Durán et al. 2016).TCO models have been widely reviewed in literature, and they are used extensively by facility managers to understand all costs related to owning assets. Various infrastructure system costs have been evaluated through a TCO framework, including facilities, roofing systems, stormwater systems, and pavements (Coffelt and Hendrickson 2010; Forasté et al. 2015; Grussing 2014; Rehan et al. 2018). Performing a TCO evaluation enables facility managers to understand the true cost of owning and operating an asset and provide a point of comparison if facility managers employ different asset manufacturers for use in their portfolios. Comparing different asset manufacturers allows facility managers to employ the best performing asset in their inventory, and one that provides the best return on investment when considering all costs. Most facility-based TCO models utilized in the facility management sector do not incorporate asset performance as a metric (Roda and Garetti 2014). However, some pavement management systems have been constructed to account for conflicting issues like technical and economic performance. Failing to consider asset performance leaves facility managers with an incomplete understanding of total costs for assets, e.g., the asset with the lowest purchase price may not translate to being the highest performing asset for use in a facility.Efforts in the manufacturing industry have been made to include asset performance as a TCO model factor (Roda et al. 2020). Using this research as a guide, modeling of the technical performance of built infrastructure assets to aid in manufacturer selection decisions has been completed (Brown et al. 2021). Manufacturer selection is the idea of choosing one manufacturer over another based on some number of selection criteria, which can be factored into TCO models. To calculate asset performance, a technical performance metric has been created that utilizes built infrastructure data such as asset condition, remaining asset service life, and variation in asset condition from similar assets to quantify the performance of assets.These recent contributions to the body of knowledge help develop more holistic TCO models that consider all financial aspects, from direct costs like initial procurement costs to indirect costs that may stem from performance-related criteria. Ultimately, viewing asset life cycle decisions through a technical performance lens helps facility managers understand how their assets are performing and what benefit, or lack thereof, they are receiving in terms of the successful operation of those assets. Calculation of asset performance also enables facility managers to investigate exogenous factors impacting asset performance. One of these factors might be the climate in which that asset is operated. It is well cited in literature that climate, and especially extreme climate events, impact infrastructure systems. Civil engineering infrastructure has been studied to understand the effects of climate on assets (Dowds and Aultman-Hall 2015; Liao et al. 2018; Shi et al. 2020). Climate may affect the frequency and rigor of asset maintenance, such as winter weather conditions increasing maintenance operations for pavement (Chinowsky et al. 2013; Dao et al. 2019). Climate conditions may also affect the expected life cycle of assets by increasing deterioration rates for those assets (Tari et al. 2015). As a product of climate change, changing weather conditions may also affect infrastructure systems that are vulnerable to the effects of extreme weather events (Douglas et al. 2017; Guest et al. 2020; Pregnolato et al. 2017). Literature has provided a link to the effect of climate on assets at a macro-level, but investigating specific climate zones can help answer the question: what, if any, role does climate play in the technical performance of built infrastructure assets?While previous research has investigated the role of climate on assets, this study is the first to evaluate the effects of climate on the performance of assets, completed at the manufacturer-level, and the first to propose that climate should be included as a component of a performance-based manufacturer selection process. Trends in climate and specific climate variables that literature has shown to affect asset operation are investigated to determine if climate affects asset performance and how equipment from different asset manufacturers respond to climate influences. Leveraging the authors’ previous work, United States Air Force (USAF) assets from 20 separate geographic installations, spanning three different climate zones according to the Köppen-Geiger classification, are used. Two asset types, chillers and air handlers, are investigated to detail the relationship of climate with respect to technical asset performance. Four climate variables that model the effects of asset exposure to sunlight, humidity, and temperature are used in this analysis. This research aims to evaluate the link between asset performance and climate to help facility managers make manufacturer selection decisions to employ the asset manufacturer that provides the highest performance in their facilities, tailored to their climate zone.Data and Case StudyA case study using observed, manufacturer-level infrastructure data was conducted to investigate the link between built infrastructure asset performance and climate. This case study builds upon the work of Brown et al. (2021) in the development of a performance-based metric to quantify the performance of assets and link relevant climate data to investigate any trends that may exist.This analysis used built infrastructure asset data from BUILDERTM Sustainment Management System, an industry-leading program used to track and manage infrastructure assets (BUILDER Sustainment Management System 2012). BUILDER has been adopted across the Department of Defense (DoD) and it is used in the private sector by many educational and municipal organizations (Sustainment Management System 2020). Additional information regarding the features, capabilities, and organization of BUILDER has been discussed and can be found in Bartels et al. (2020) and Grussing et al. (2016). BUILDER supplied the following data points for this analysis: 1.observed asset condition, which is a 0- to 100-point value that represents the health of an asset as observed by a trained inspector, and the assessment’s corresponding inspection date;2.the installation date of the asset, or when it was first put into service;3.the asset’s remaining service life (RSL, years) is the number of useful years left an asset has in service adjusted to account for the current degradation rate of the asset;4.the manufacturer of the asset; and5.the location of the asset (indoor versus outdoor unit).The USAF utilizes BUILDER to manage its infrastructure assets, and its data was utilized for this case study. BUILDER has many capabilities, one of which is generating predicted and expected condition values based on the current state of assets, while the expected condition is not utilized in this study, instead the observed condition is, and the RSL is a dynamic value that adjusts based on the current health of the asset and expected changes in the future. A total of 20 Air Force installations from across the contiguous United States were studied (Fig. 1). These locations were chosen to provide a sample of installations from each Köppen-Geiger climate zone to portray locations subjected to different climates. These locations represent one-third of all Air Force installations within the contiguous US, providing a good representation of the USAF’s data.Two asset types were included in this analysis, chillers and air handlers. These assets have relatively long expected service lives: 20 years for chillers and 24 years for air handlers, exhibit condition degradation throughout their lifespans, and have several major manufacturers. Additionally, both of these asset types are subjected to environmental impacts throughout their operation. The study includes package units from 20 tons to 1,500 tons for chillers and 2,000 cubic feet per minute (CFM) to 75,000 CFM for air handlers. To relate asset performance to climate variables, each asset is grouped by its manufacturer. Two chiller manufacturers are studied and hereafter labeled Manufacturer A and Manufacturer B. Three air handler manufacturers are studied, Manufacturer A, Manufacturer B, and Manufacturer C. Manufacturer A and B are the same for chillers and air handlers; that is, each produces both types of assets. Manufacturer C only produces air handlers. This naming convention is consistent with that which is proposed in previous work that provides additional detail (Brown et al. 2021). This analysis does not aim to provide definitive conclusions about which manufacturer, by name, an organization should procure. Instead, it aims to show the utility of investigating the link between manufacturer performance and climate variables. As such, manufacturer names have been omitted to avoid any endorsement of one brand over another.Climate Zone Classifications and Selection of Climatic VariablesThe Köppen-Geiger climate classification categorizes each part of the globe into climate zones based on precipitation and temperature data for the region (Peel et al. 2007). The Köppen-Geiger classification further divides each climate zone into sub-regions for more accurate grouping by like climate areas; however, the five main climate zones are utilized for this analysis. A map of the Köppen-Geiger climate zones pertinent to this study has been created (Fig. 1) utilizing open-access data of Köppen-Geiger climate zones (Beck et al. 2018). The 20 Air Force installations included in this analysis are marked on the map. Based on Köppen-Geiger classification, seven installations fall within the Arid zone, seven in the Temperate zone, and six in the Cold region.The climatic variables chosen for analysis are Heating Degree Days (HDD), Cooling Degree Days (CDD), Total Solar Irradiance, and a variable to account for the total number of days where the average relative humidity was above 55% (hereafter referred to as H-55). These variables were selected because numerous citations in the literature point to their effect on the operation of assets (Crawley 1998; Jazaeri et al. 2019; de Rubeis et al. 2020) as well as the current climatic design standards used for Heating, Ventilation, and Air Conditioning (HVAC) units (ASHRAE 2009; Roth 2017). Based on this information, it is hypothesized that they may influence the technical performance of assets. HDDs are an environmental measure of how cold the climate is for a given day below a specific threshold value; CDD is a measure of how warm the climate is for a given day above a threshold value (EIA 2020). A standard value of 65 °F is used for the threshold value for both HDD and CDD. Total Solar Irradiance is the total light intensity observed in watts per square meter between sunrise and sunset for a day. This value provides a measure of the amount of exposure assets have to the sun. The H-55 variable is a variable created to account for the total number of days where the average relative humidity was above 55%, which provides a metric for how humid the environment is that the asset is operating in, and also provides an indication for environmental conditions that drive HVAC asset operation. All weather data were sourced from AccuWeather’s propriety database. The chosen climate variables do not provide an exhaustive look at all climatic variables that may affect an asset’s performance but provide a starting point for analysis. These chosen variables target the climatic variables that influence chillers and air handlers based on how they operate and are backed up by research.Initial Data Filtering and Data Summary StatisticsThe built infrastructure data stored in BUILDER provides a wealth of information for the case study and investigation of the link between asset performance and climate. However, initial pre-processing was required to organize the data in a ready-state for analysis. Initial data filtering and processing actions included rebaselining the temporal scale of stored data from an absolute date to a relative asset age to compare all assets on a similar basis. Any assets with a change in condition greater than or equal to zero between inspections were removed to only consider assets that have not had a major repair or improvement. Any assets that had a condition less than 100 at installation were removed because assets should be in perfect condition at the installation date. This pre-processing step was meant to exclude assets that may have been incorrectly entered into the BUILDER database. Lastly, any assets with missing or incomplete data fields were removed. A complete record of an asset’s manufacturer, condition, and RSL must be available to calculate asset performance. An extensive explanation of this filtering and exclusion process has been covered in Brown et al. (2021). For this climatic influence analysis, an additional filtering criterion was applied to remove assets that had an installation year before 1985 or an installation year after 2018. This step was done to align the asset data with the temporal range of available climate data. The data sourced from AccuWeather was daily climatic data from 1985 to 2018.The initial population of asset data available from BUILDER included 8,579 unique chiller and air handler units from the 20 Air Force installations. Data filtering and exclusion criteria reduced the data population by 66% (5,705 assets were removed). This large percentage highlights the need for rigorous asset management programs that track and manage data for their built infrastructure assets. Despite the percentage of removed assets, the analysis still contains 2,874 unique assets from 1,341 facilities at the 20 Air Force installations. This filtered population of assets contains 765 chiller units and 2,109 air handlers. Of the chiller units, 33% of units are located in the Arid climate zone, 42% are located in the Temperate climate zone, and 25% are in the Cold climate zone. For the air handlers, 23% are located in the Arid climate zone, 48% are located in the Temperate climate zone, and 29% are located in the Cold climate zone.Further breakdown of the number of units of each manufacturer brand within each climate zone is detailed in Table 1. This table shows the prevalence of each manufacturer within the climate zone. Overall, there are a majority of manufacturer A branded chillers across the three climate zones. Manufacturer A is also the most prevalent brand in operation for air handlers for these Air Force installations.Table 1. Manufacturer prevalence within each climate zoneTable 1. Manufacturer prevalence within each climate zoneAsset typeClimate zoneManufacturerPrevalence of manufacturer in climate zone (%)ChillersAridManufacturer A49Manufacturer B51TemperateManufacturer A82Manufacturer B18ColdManufacturer A84Manufacturer B16Air handlersAridManufacturer A50Manufacturer B29Manufacturer C21TemperateManufacturer A63Manufacturer B21Manufacturer C16ColdManufacturer A59Manufacturer B14Manufacturer C27MethodologyAsset Performance MetricInvestigating the link between asset performance and climate requires a metric to quantify asset performance, so the authors have formulated a way to do this using available built infrastructure data. This methodology creates an age-based metric that uses asset condition and includes a measure of condition variation to compute each asset’s performance value. Eq. (1) below is the equation used to calculate asset performance (1) Performance=(wi×Conditionscaled)+(wj×RSLscaled)+(wk×[1−RMSEscaled])This equation follows a weighted sum model approach to utilize three parameters to calculate asset performance. The first parameter is Conditionscaled, which is the observed condition of the asset directly taken from the BUILDER database. The 0-100 point value for condition is scaled to a number between zero (0) and one (1) using a minimum-maximum normalization technique. The next parameter is RSLscaled, which is a measure of the remaining service life (RSL) of the asset and represents the number of years between the current age and the asset’s expected service life. RSL is updated after each asset assessment to either decrease or stay the same depending on the asset’s current deterioration rate. For example, if the asset is degrading quicker than expected, the RSL is decreased. This number is also scaled to a value between zero (0) and one (1). The final parameter of the equation is RMSEscaled, which provides a consideration for condition variation. The inclusion of this parameter compares the condition of assets at one location to the mean condition for all similar assets in the organization’s inventory. The variation is computed using root-mean-square error (RMSE), which is a formulation of distance of individual means to an overall population mean. Facility managers should value assets that behave similarly to the majority of their assets because this allows for more predictable operation and easier time planning maintenance activities. This parameter is also scaled to a number between zero (0) and one (1) and then subtracted from one. This final subtraction operation allows assets that exhibit conditions closer to the mean to have greater influence in the performance equation.Each parameter in the performance metric equation has a unique weighting factor attached, allowing decision-makers to choose which parameter is most important and should carry the highest weight. Each weighting factor must be greater than or equal to zero (wi≥0,wj≥0,wk≥0) and all factors must sum to one (wi+wj+wk=1). This analysis has equally weighted each parameter (wi=wj=wk=0.333). The final performance metric is a value between zero (0) and one (1), where one (1) indicates the highest performance when compared to like assets and zero (0) indicates the lowest performance compared to like assets. This equation provides a way to quantify asset performance based on asset condition and informed by service life and variance. A detailed description of this equation and example calculations can be found in the authors’ previous work (Brown et al. 2021). This methodology generates one performance metric value for each asset in the analysis at every condition assessment point. Therefore, if a single asset was assessed three times over its service life, three performance metrics will be calculated and tracked in this analysis. This means that there are some cases where assets are tracked multiple times, but this provides strength to the analysis by increasing the sample data set.Query Weather Database and Calculating Cumulative TotalsThe AccuWeather database was first queried to match a local weather station with the latitude and longitude coordinates of the Air Force installation to calculate the cumulative climate exposure of each asset at each installation. The proprietary AccuWeather database contained weather data for 1,938 weather stations across the US Each weather station had the coordinates, and these could be matched with the coordinates for the 20 locations of interest. The average distance between the weather station and its corresponding Air Force installation was only 0.85 mi, and the maximum distance between any one weather station and Air Force installation was only 2.1 mi, so the climate data selected is indicative of the conditions experienced at the Air Force installations. Once the weather stations were linked with the Air Force installations, each location’s data could be mined for the climate variables of interest: HDD, CDD, Solar Irradiance, and H-55.This analysis matches each chiller and air handler unit with cumulative climate exposure between assessment dates. This methodology allows for a link to be made between each climate variable and the asset’s performance. First, each asset’s installation date is marked as the first day of interest, and a counter begins that sums the number of days between the installation date and the asset’s assessment date at which the condition data was recorded. This exact timeframe (number of days) is found in the climate database, and the cumulative amount of climate exposure for each variable described above is totaled for that same period. For example, if a chiller was installed on January 10, 2005 and was first assessed on January 10, 2010, the counter would return 1,826 days. The weather variable database is then queried to find January 10, 2005 and records the variable of interest for that day. The program then sums the number of accumulated climate units, e.g., HDDs, until the assessment date on January 10, 2010 (1,826 total days). This cumulative approach is meant to account for asset exposure between condition assessments. This cumulative value is paired with the performance of the asset calculated at that point in time. This methodology is followed for each asset’s assessment date at each installation for each climate variable of interest.Visualizations and Statistical AnalysisAfter calculating the cumulative climate exposure for each asset, scatterplots can be generated to inspect the relationship between asset performance and climate variables, as they enable easy visualization of trends. Scatterplots are generated for each asset (chillers and air handlers), each climate region (Arid, Temperate, and Cold), and each climate variable (HDD, CDD, Solar Irradiance, and H-55). The cumulative climate exposure is shown on each scatterplot on the horizontal axis, and asset performance is shown on the vertical axis. Each point on the scatterplot represents an individual asset. Select scatterplots are shown in the “Results” section. Results and all scatterplots are shown in the Appendix. In addition to scatterplots, a Pearson correlation coefficient (r) is calculated to measure the linear correlation between cumulative climate exposure and asset performance. For this analysis, an absolute correlation value less than 0.1 indicates no relationship (0.1>|r|), an absolute correlation value between 0.1 and 0.3 is considered a weak correlation (0.1≤|r|<0.3), a correlation coefficient between 0.3 and 0.5 is considered a moderate correlation (0.3≤|r|<0.5), and a value greater than or equal to 0.5 indicates a strong relationship (0.5≤|r|). These threshold values are general guidelines often cited in literature (Cohen 2013). Correlation coefficients are shown for each scatterplot as well as in Figs. 4 and 5 of the following section “Results.”In addition to correlation analysis, an analysis of variance (ANOVA) was performed to determine if a statistically significant difference in the mean performance metric of assets between the different factor levels is present. This test is performed for chillers and air handlers, and different factor levels, i.e., climate zone, asset manufacturer, and asset location (indoor versus outdoor unit), are tested within each asset group. ANOVA testing provides context to whether the different factor levels contribute to a difference in the average performance metric. The ANOVA results are explored in-depth in the next section “Results.”ResultsCorrelation VisualizationsAfter calculating cumulative climate exposure for the time period between assessments for assets, the data could be visualized in scatter plots. These scatterplots show the cumulative climate exposure on the horizontal axis and the zero (0) to one (1) performance metric values on the vertical axis. These figures provide a visual of any relationships that exist between the variables. On each scatter plot, each point represents a unique asset. Plots are color-coded by the manufacturer of the asset. Alongside each scatter plot, the correlation coefficient is shown for each manufacturer. As expected, most correlation values are negative, likely due to the climate variable’s inherent time component. As time passes, the total for each climate variable increases, which implies that more time passing is connected to assets aging. However, variation in the correlation value between manufacturers suggests that each climate variable and performance combination is different. Most relationships are weak to moderate, though some climate variables have a strong correlation to asset performance. All scatterplots are shown in the Appendix. Chiller units in the Cold climate zone (Fig. 2) are shown here to highlight strong correlations and assets with minimal dispersion between asset performance and cumulative climate exposure. These scatterplots show a negative linear trend in the data. Manufacturer A shows a strong correlation between all climate variables and asset performance. Manufacturer B has a moderate correlation between CDD and asset performance and a strong relationship between HDD, Solar Irradiance, and H-55. The tight dispersion shows the low variability that exists between cumulative climate exposure and asset performance for this climate zone.Air Handlers in the Temperate climate zone (Fig. 3) are shown to highlight results with more dispersion. The greater dispersion for these plots indicates that there is a high degree of variability within the data. These results show negative linear trends for the manufacturers across all climate variables. Manufacturer A shows a weak correlation for HDD and moderate correlation for CDD, Solar Irradiance, and H-55. Manufacturer B shows a weak correlation for HDD and a strong correlation for CDD, Solar Irradiance, and H-55. Manufacturer C shows a moderate relationship for HDD and shows weak relationships for CDD, Solar Irradiance, and H-55.Correlating Numerical ResultsThe selected scatterplots and correlation coefficients generalize the statistical relationships between cumulative climate exposure and asset performance for the three climate zones of study for chillers and air handlers. By further grouping assets by their location in relation to the facility they service—either indoor or outdoor units—further investigation can be performed to determine if the asset’s location plays a role in linking asset performance and cumulative climate exposure. Figs. 4 and 5 below provide an overview of this level of analysis. These figures show each asset’s correlation coefficient, in each climate zone, for each manufacturer, by location. These figures also contain the correlation coefficients shown previously in Figs. 2 and 3. The correlation coefficients in the figures are color-coded to correspond to correlation strength. A gray color indicates no relationship (0.1>|r|), a light orange indicates a weak correlation (0.1≤|r|<0.3), a light green color indicates a moderate correlation (0.3≤|r|<0.5), and a dark green color indicates a strong correlation coefficient (0.5≤|r|). The bold type shows the major grouping of the assets by manufacturer before grouping by indoor or outdoor units. Additionally, calculating the p-value for each correlation provides context to whether the correlation coefficient is statistically significant. Asterisks denote the statistical significance following each correlation value.Fig. 4 shows all the correlation coefficient values for chiller units and the statistical significance of each value. Overall, there are many moderate and strong correlation values between asset performance metric and cumulative climate exposure within the different climate regions. This figure also highlights that in some cases, grouping assets by their location in relation to the facility they serve (indoor or outdoor unit) strengthens the relationship. For example, the statistical significance of H-55 and asset performance of Manufacturer B assets increase when comparing indoor and outdoor units, as opposed to all units combined for the Cold climate zone. Across all climate variables and for both manufacturers, the Cold climate zone shows moderate to strong relationships between climate variables and asset performance. This result shows that asset performance is highly influenced by cumulative climate exposure within the Cold climate zone. In the Arid climate zone, both manufacturers show a moderate correlation between CDD and Solar Irradiance, showing that hot temperatures and sun exposure influence asset performance. For the Temperate climate zone, CDD and H-55 show moderate correlation levels to asset performance, indicating that hot temperatures and humid environments influence assets in the Temperate region.Fig. 5 shows the correlation coefficient values and statistical significance for the air handlers in the analysis. There are many moderate and strong correlation values across all climate zones, indicating that cumulative climate exposure influences asset performance. The Temperate region shows the strongest correlation values that might indicate that the Temperate region’s asset performance is highly influenced by cumulative climate exposure. Within the Temperate climate zone, CDD, Solar Irradiance, and H-55 appear to have the strongest correlation values across the three manufacturers meaning that asset performance in the Temperate zone is most affected by the cooling demand, exposure to the sun, and humidity. In the Arid climate zone, asset performance is most affected by CDD and Solar Irradiance, showing the highest correlation values, meaning that hot temperatures and exposure to the sun influence asset performance. Most correlation values are weak and moderate for the Cold climate zone, except for Manufacturer B, which shows strong correlation values for some outdoor units. These results show that, on average, the Cold climate zone’s asset performance is not highly affected by cumulative climate exposure, except for Manufacturer B.Across the analyses for both chillers and air handlers, some sample sizes are small when assets are grouped by location. Sourcing additional data could provide more strength to the correlation coefficients and make some relationships stronger and more statistically significant.ANOVA ResultsThe chiller ANOVA test results (Table 2) show that only the interaction element between climate zone and manufacturer produces different average performance metrics. This factor level where the p-value is lower than the critical p-value, 0.05 for this analysis, provides statistical evidence of a difference in means. This result suggests that Manufacturer A assets perform differently in the Arid climate zone from those in the Temperate climate zone and those in the Cold climate zone. The same is true of Manufacturer B, where assets perform differently in each climate zone. Overall, these results show that there are differences in asset performance across the different climate zones and between the two manufacturers.Table 2. ANOVA test results for chillersTable 2. ANOVA test results for chillersFactor sourceSum of squaresDegrees of freedomMean squaresF-statisticp-valueClimate zone0.010920.005460.210.8076Location of asset (indoor/outdoor)0.011810.011750.460.4979Manufacturer0.000710.000700.030.8681Interaction between climate zone & location of asset (indoor/outdoor)0.018120.009070.350.7014Interaction between climate zone & manufacturer0.470520.235279.210.0001Interaction between location of asset (indoor/outdoor) & manufacturer0.039710.039731.550.2127Error25.50489980.02556——Total26.23961,007———The same result is shown for air handlers (Table 3). This ANOVA test shows that the interaction element between climate zone and manufacturer impacts the asset performance metrics. The p-value for this factor is lower than the critical p-value of 0.05, which provides the statistical evidence. These air handler results show that each manufacturer performs differently in each climate zone, e.g., Manufacturer A assets perform differently in the Arid climate zone than those in the Temperate climate zone and differently from those in the Cold climate zone.Table 3. ANOVA test results for air handlersTable 3. ANOVA test results for air handlersFactor sourceSum of squaresDegrees of freedomMean squaresF-statisticp-valueClimate zone0.041520.020730.880.4149Location of asset (indoor/outdoor)0.051610.051612.190.1390Manufacturer0.033320.016660.710.4931Interaction between climate zone & location of asset (indoor/outdoor)0.040120.020040.850.4270Interaction between climate zone & manufacturer1.318140.3295213.990.0000Interaction between location of asset (indoor/outdoor) & manufacturer0.005420.002680.110.8924Error60.26552,5580.02356——Total62.04852,571———The ANOVA testing (Tables 2 and 3) highlights that the interaction element between climate zone and asset manufacturer is influential to asset performance, but other variables are not influential. Alone, the ANOVA results show that climate zone does not create performance differences among assets. Location of assets (indoor or outdoor units) does not create performance differences, and by itself, asset manufacturer does not create performance differences. Nevertheless, when investigating different manufacturers in different climate zones, performance differences are apparent. These results suggest which factor levels are influential in creating asset performance differences and which are not.DiscussionThe statistical analysis performed and detailed in the previous section “Results,” indicates that there is a moderate level of influence that environmental factors play in asset performance across both space and time. Many of the different groupings of assets showed moderate and strong correlation values. The ANOVA results show that climate zone and manufacturer of assets affect asset performance such that each manufacturer performs differently in each climate zone. These results are illustrated when cumulative climate exposure is plotted against asset performance, and correlation coefficient values show moderate associations between the variables. The combination of correlation testing and ANOVA testing show that asset performance is linked to climate exposure and that manufacturer selection is important because there are differences in how individual manufacturers respond to climate differences. The results can help facility managers determine which asset manufacturer provides the best performance for the climate zones in which their assets operate. By choosing the manufacturer that exhibits the best performance when faced with the most influential climate variables for their region, they can ensure they employ high performing assets that may ultimately lead to a lower total cost when factored into TCO models.Previous work concluded that there was insufficient evidence to support manufacturer selection decisions at an enterprise level for the Air Force. Using the technical performance metric for assets allowed for installation decisions to be made that are best for that specific location, but there was no clear decision at the enterprise level. Grouping Air Force installations by climate region shows that assets within the same climate zone react to climatic variables similarly. This climatic analysis provides further support that making manufacturer selection decisions at local installation-levels may make the most sense instead of enterprise-wide solutions.These investigatory results show that asset performance is not the same across all manufacturers of assets. Making manufacturer selection decisions based on a technical performance metric can be useful to a facility manager. The results may help guide operational decisions a facility manager needs to make throughout an asset’s life cycle. The influence of climate variables may impact these decisions, such as the effect on asset degradation. If a facility manager in a particular climate zone anticipates a specific degradation profile for their assets, based on the long-term averages of weather variables, and then the climate zone experiences extremes for these averages, a facility manager may be able to predict potential changes to their asset’s degradation profiles.Additionally, the climate zones and climate variables drive asset performance, and as such, degradation predictions could be partially informed with a climate-based assessment model. Adjustments to asset-anticipated service life could also be informed by performance expectations based on climate zones and future climate projections. As the average climate changes for some areas around the United States, and more extreme weather events occur more often and with greater intensity, the effect on asset performance could be predicted by relying on the relationships calculated here. Moreover, as climate change effects become more prevalent in some areas, understanding the link between climate and asset performance may strengthen.The methodology of choosing four climate variables to study not only provides the capability to study climatic influences but provides a measure of the operational use of the assets studied. The cumulative climate exposure variables provide an indication of the total amount of utilization over time and show that these variables could be used in lieu of sensor monitoring or meter readings if they are not available. The ancillary value of using cumulative climate variables provides a good approximation of asset usage. One limitation of this study is the limited scope of climate variables investigated. The decision to include HDD, CDD, Solar Irradiance, and H-55 as the variables of interest was based on the operational effects these variables have on chillers and air handlers; however, these four variables are not the only climatic factors that may affect chillers and air handlers. Future research could be focused on expanding the scope of variables to fully understand all climatic factors that may affect assets’ technical performance.In addition to investigating additional climate variables as an avenue of future research, additional asset types could be analyzed to investigate the role of climate on their performance. This study chose to analyze two asset types, chillers and air handlers; however, the described methodology could also be applied to other built infrastructure assets like boilers, transformers, generators, or condensers. The inclusion of additional asset types may reveal additional links to climate’s influence on asset performance, strengthening the results of this study.An additional recommendation for future research is to investigate the influence of time and asset age on performance, in addition to climatic influence. These investigatory results solely looked at the effect that cumulative climate exposure had on asset performance; however, it is important to decouple the effects of time and age in future analysis. Ideally, the amount of influence that asset age has on asset performance should be isolated, separate from the influence that cumulative climate exposure has on assets.This research also highlights the analytical capabilities that are available when organizations track and manage built infrastructure data. The USAF has more than ten years of condition assessment data available. Statistical analyses can be performed to show the relationship that exists between asset performance and climatic variables. Organizations that manage facilities and the accompanying assets on any level, whether it is a small organization that owns a few facilities or a large organization similar to the USAF that has a multitude of facilities spread out geographically, built infrastructure data can be leveraged to perform statistical analysis to help make data-driven decisions for their organization. Accurate data management policies can help organizations know and understand their assets to make the best decisions for their asset portfolios. This research also exposes the potential limitations that exist from incomplete data records. A large portion of the original data points had to be excluded from the analysis because there was missing data regarding the manufacturer of the asset. By implementing robust data management procedures, organizations can increase the amount of data available to them for analysis.ConclusionsThis research set out to examine the role that four climate variables, HDD, CDD, Solar Irradiance, and H-55, might play in asset performance when assessed via a technical performance metric. This analysis showed that all of these climate variables impacted chillers and air handler units in some way through the many different combinations of analyses that were targeted. In most cases, asset performance was negatively linked to the climate variables studied, which implies that climate variables influence asset performance such that they decrease asset performance. By comparing results by one of the three Köppen-Geiger climate zones (Arid, Temperate, Cold) that exist for the 20 Air Force installations of interest, the climate variables’ role on asset performance could be observed. Additionally, by looking at the asset’s location in relation to the facility it serves (indoor versus outdoor unit), an understanding could be made to see if asset placement plays a role in asset performance, which it does.Ultimately, this research builds on extensive research that already exists in the field for using TCO models to describe all costs of ownership for built infrastructure assets. By employing a technical performance metric that describes an asset’s performance based on condition, age, and variation in condition, an economic consideration can be factored into TCO models to account for this technical performance. A climatic analysis helps facility managers further understand their assets’ technical performance, specific to their climate zone. This analysis further links the impact of climate on built infrastructure assets and can provide another criterion for facility managers to use when making manufacturer selection decisions. This analysis also highlights the evaluation capabilities that are available when organizations employ rigorous data management programs to track and manage their infrastructure assets.Appendix. Additional Climate ScatterplotsAdditional Correlation Scatterplots are shown for the remaining climate zones. The Arid Climate Zone for Chillers (Fig. 6) and the Temperate Climate Zone for Chillers (Fig. 7) provide visualizations of the chiller data previously not shown in the text. The Arid Climate Zone for Air Handlers (Fig. 8) and the Cold Climate Zone for Air Handlers (Fig. 9) provide the remaining visualizations of Air Handler data previously not shown in the text.Data Availability StatementSome or all data, models, or code generated or used during the study are proprietary or confidential in nature and may only be provided with restrictions (e.g., anonymized data). The raw dataset consists of Sustainment Management System BUILDER data, which belongs to the US Air Force. Summary data could be provided as anonymized.References ASHRAE (American Society of Heating, Refrigeration and Air Conditioning Engineers). 2009. Climatic design information. Atlanta: ASHRAE. Bartels, L. B., L. Y. Liu, K. El-Rayes, N. El-Gohary, M. Golparvar, and M. N. Grussing. 2020. “Work optimization with association rule mining of accelerated deterioration in building components.” J. Perform. Constr. Facil. 34 (3): 04020033. https://doi.org/10.1061/(ASCE)CF.1943-5509.0001441. Beck, H. E., N. E. Zimmermann, T. R. McVicar, N. Vergopolan, A. Berg, and E. F. Wood. 2018. “Present and future Köppen-Geiger climate classification maps at 1-km resolution.” Sci. Data 5 (1): 180214. https://doi.org/10.1038/sdata.2018.214. Brown, S. L., S. J. Schuldt, M. N. Grussing, L. B. Bartels, M. A. Johnson, and J. D. Delorit. 2021. “Performance-based building system manufacturer selection decision framework for integration into total cost of ownership evaluations.” J. Perform. Constr. Facil. 35 (5): 04021049. https://doi.org/10.1061/(ASCE)CF.1943-5509.0001624. Cohen, J. 2013. Statistical power analysis for the behavioral sciences. Cambridge: Academic Press. Crawley, D. B. 1998. “Which weather data should you use for energy simulations of commercial buildings?” In Transactions-American society of heating refrigerating and air conditioning engineers. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers. Dao, B., S. Hasanzadeh, C. L. Walker, D. Steinkruger, B. Esmaeili, and M. R. Anderson. 2019. “Current practices of winter maintenance operations and perceptions of winter weather conditions.” J. Cold Reg. Eng. 33 (3): 04019008. https://doi.org/10.1061/(ASCE)CR.1943-5495.0000191. de Rubeis, T., S. Falasca, G. Curci, D. Paoletti, and D. Ambrosini. 2020. “Sensitivity of heating performance of an energy self-sufficient building to climate zone, climate change and HVAC system solutions.” Sustainable Cities Soc. 61 (Oct): 102300. https://doi.org/10.1016/j.scs.2020.102300. Dowds, J., and L. Aultman-Hall. 2015. “Barriers to implementation of climate adaptation frameworks by state departments of transportation.” Transp. Res. Rec. 2532 (1): 21–28. https://doi.org/10.3141/2532-03. Durán, O., M. Macchi, and I. Roda. 2016. “On the relationship of spare parts inventory policies with total cost of ownership of industrial assets.” IFAC Pap. OnLine 49 (28): 19–24. https://doi.org/10.1016/j.ifacol.2016.11.004. Forasté, J. A., R. Goo, J. Thrash, and L. Hair. 2015. “Measuring the cost-effectiveness of LID and conventional stormwater management plans using life cycle costs and performance metrics.” In Low impact development technology: Implementation and economics, 54–73. Reston, VA: ASCE. Grussing, M. N., L. Y. Liu, D. R. Uzarski, K. El-Rayes, and N. El-Gohary. 2016. “Discrete Markov approach for building component condition, reliability, and service-life prediction modeling.” J. Perform. Constr. Facil. 30 (5): 04016015. https://doi.org/10.1061/(ASCE)CF.1943-5509.0000865. Guest, G., J. Zhang, R. Atadero, and H. Shirkhani. 2020. “Incorporating the effects of climate change into bridge deterioration modeling: The case of slab-on-girder highway bridge deck designs across Canada.” J. Mater. Civ. Eng. 32 (7): 04020175. https://doi.org/10.1061/(ASCE)MT.1943-5533.0003245. Jazaeri, J., R. L. Gordon, and T. Alpcan. 2019. “Influence of building envelopes, climates, and occupancy patterns on residential HVAC demand.” J. Build. Eng. 22 (Mar): 33–47. https://doi.org/10.1016/j.jobe.2018.11.011. Peel, M. C., B. L. Finlayson, and T. A. McMahon. 2007. “Updated world map of the Köppen-Geiger climate classification.” Hydrol. Earth Syst. Sci. 11: 1633–1644. Pregnolato, M., A. Ford, V. Glenis, S. Wilkinson, and R. Dawson. 2017. “Impact of climate change on disruption to urban transport networks from pluvial flooding.” J. Infrastruct. Syst. 23 (4): 04017015. https://doi.org/10.1061/(ASCE)IS.1943-555X.0000372. Rehan, T., Y. Qi, and A. Werner. 2018. “Life-cycle cost analysis for traditional and permeable pavements.” In Proc., Construction Research Congress 2018. Reston, VA: ASCE. Roda, I., and M. Garetti. 2014. “The link between costs and performances for total cost of ownership evaluation of physical asset: State of the art review.” In Proc., Int. Conf. on Engineering, Technology and Innovation (ICE), 1–8. New York: IEEE. Roda, I., M. Macchi, and S. Albanese. 2020. “Building a total cost of ownership model to support manufacturing asset lifecycle management.” Prod. Plann. Control 31 (1): 19–37. https://doi.org/10.1080/09537287.2019.1625079. Roth, M. 2017. “Updating the ASHRAE climate design data for 2017.” ASHRAE Trans. 123 (Part 2): 80–89. Shi, F., X. Peng, Z. Liu, E. Li, and Y. Hu. 2020. “A data-driven approach for pipe deformation prediction based on soil properties and weather conditions.” Sustainable Cities Soc. 55 (Apr): 102012. https://doi.org/10.1016/j.scs.2019.102012. Tari, Y. H., S. S. Shamsabadi, R. Birken, and M. Wang. 2015. “Deterioration modeling for condition assessment of flexible pavements considering extreme weather events.” In Vol. 9437 of Structural health monitoring and inspection of advanced materials, aerospace, and civil infrastructure 2015, 943721. Bellingham, WA: International Society for Optics and Photonics.