Occupant thermal comfort and indoor environmental conditions in Mies van der Rohe’s S. R. Crown Hall

ABSTRACT There is little empirical data in the published literature on occupant thermal comfort in buildings from the Modern Movement in architecture. We present a field survey of occupant thermal comfort and indoor environmental conditions in Mies van der Rohe’s modern S. R. Crown Hall (1956) on the campus of Illinois Institute of Technology in Chicago, IL. Surveys were deployed to 557 student participants on four separate days, including two days in the cooling season and two days in the heating season, to assess their perceptions of thermal comfort throughout the space. Indoor air temperature, relative humidity and mean radiant temperature were measured concurrently in several locations throughout the space. Occupants reported high levels of dissatisfaction with comfort in the space (percent dissatisfied ranging 37%–54%), with somewhat counterintuitive results for this building type. Overcooling was apparent during warm weather and a combination of both over- and under-heating occurred during cold weather, contrary to what was expected in this building with a high thermal transmittance enclosure. There was also high spatial variability in comfort responses and measured indoor environmental conditions. Findings highlight the need to develop contextual approaches to meeting occupant comfort needs in this building while preserving architectural aesthetics/intent.


Introduction
The Modern Movement (Modern architecture) evolved in America during the first half of the 20th century, with a philosophical rejection of the tenets of traditional building in an era when energy resources were considered cheap and plentiful, expressing freedom in using new and innovative technologies and construction techniques and a new vision to create architecture reflective of its time (Cunningham, 1998).The materials and construction techniques utilized in these buildings were often basic steel, concrete and glass, which advanced the technical capabilities of their time but also presented challenges to achieving modern standards of energy efficiency (Timberlake, 2015) and thermal comfort (Requena-Ruiz, 2016).While many such buildings realized in the first half of the 20th century were originally designed to operate without air-conditioning systems and, thus, some have posited that energy or comfort in these buildings should not be evaluated relative to modern standards (Chang & Winter, 2015), the reality is that many such buildings have been retrofitted to incorporate heating and cooling systems and will continue to operate for many years to come.Therefore, various buildings from this era have recently been the object of interventions aimed at addressing preservation and challenges concerning adapting such Modern heritage buildings to contemporary needs (Tostões, 2022).
Despite our knowledge and expectations of thermal problems commonly encountered by such buildings in practice and attributed to large surface areas of single pane glazing, retrofitted heating, ventilating, and airconditioning (HVAC) system installations, and unplanned changes in building spaces over time away from initial programming, there is little empirical data in the published literature on indoor environmental conditions or thermal comfort in buildings from the Modern Movement in architecture, especially in the US.For example, Martínez-Molina et al. (2016) reviewed the literature on energy efficiency and thermal comfort in historic buildings, finding only a few studies of buildings constructed between the 1940s and 1980s, most of which were simulation-based and not necessarily focused on buildings from the Modern Movement (Fouseki & Cassar, 2014).Since that review, Galiano-Garrigós et al. (2022) investigated the expected thermal comfort in the modernist E.1027 house by Gray and Badovici (1929, France) using building simulation tools (Galiano-Garrigós et al., 2022), while others have used simulation tools to investigate daylighting and visual comfort (Iommi, 2019) and energy performance (González-Avilés et al., 2022) in Le Corbusier's Modern Movement residences.However, simulation-based studies of predicted comfort are limited because they rely on narrow assumptions and simplifications to attempt to capture a perception that is inherently more complex, especially in buildings that do not resemble simple climate chambers from which comfort models were derived (Nicol & Roaf, 2017).Lawrence et al. (2019) assessed energy use and occupant thermal comfort (via user surveys) in seven university buildings in the UK, including two post-war era concrete and glass buildings, Arts Tower (1965) and Western Bank Library (1959), that had been recently renovated (in 2011 and 2009, respectively), finding that both buildings were perceived to be cooler and less comfortable during the winter than other more recently constructed buildings, presumably due to their low thermal resistance enclosures (Lawrence et al., 2019).Data from other Modern Movement buildings are scarce.
Therefore, while there has been significant attention to monitoring indoor environmental conditions and assessing occupant thermal comfort in existing buildings, including in historic buildings that pre-date the Modern Movement (Martínez-Molina et al., 2016), research specifically addressing environmental conditions and occupant comfort in buildings from the Modern Movement is highly underrepresented in the literature.As such, indoor environmental quality (IEQ) and comfort evaluations are often of secondary importance in considering renovations and retrofitting strategies to such buildings in practice, as the primary motivating factor for interventions is often associated with the degradation of the envelope materials or reglazing of the metal frames, generally for structural or aesthetic reasons (Ayón et al., 2019).
A quintessential example of a highly glazed building from the Modern Movement is S. R. Crown Hall, home of the College of Architecture on the main campus of Illinois Institute of Technology (IIT) in Chicago, IL, US.Crown Hall was designed by Ludwig Mies van der Rohe and constructed in 1956.On 1 October 1997, it was declared a City of Chicago landmark.While buildings from the Modern Movement era are often considered too recently constructed for the general public to find consensus to be recognized as historic, heritage-worthy buildings to be preserved for future generations (Requena-Ruiz, 2016), Crown Hall was listed as a National Historic Landmark in the National Register of Historic Places in 2001, a rare recognition for a building that at the time had been standing for less than 50 years.It remains an educational building, consistent with its original design.The building mirrors the historic oneroom schoolhouse and reflects IIT's College of Architecture's curriculum, aligned with its unique models, values and norms that stem from the unique open studio culture.Within this open space, students sit at individual desks that are arranged in double rows that run east and west and that face north and south.
For years, students, staff and faculty have raised anecdotal issues surrounding the thermal comfort in Crown Hall.However, a literature search reveals that there are no published works empirically assessing occupant thermal comfort or indoor environmental conditions in Crown Hall.In fact, to the best of our knowledge, there are no published works empirically assessing occupant thermal comfort or indoor environmental conditions in any Mies buildings.Given this dearth of literature, the objective of this research is to empirically investigate occupant thermal comfort and indoor environmental conditions in S. R. Crown Hall during both heating and cooling seasons to capture a range of realistic operating conditions.The study involves indoor environmental measurements and thermal comfort surveys of daily users of the building over two distinct seasons with widely varying ambient weather conditions.Data are analysed specifically to (i) assess perceived comfort broadly across users in the building, (ii) assess the factors that influence perceived comfort across space and time and (iii) provide preliminary suggestions for addressing perceived discomfort.

Methodology About S. R. Crown Hall
The architecture of S.R. Crown Hall resembles a clear manifestation of the Modern Movement, with primary construction materials of steel, reinforced concrete, and glass (see Figure 1).The space alludes to Mies's Modernist ideals and his 'less is more' philosophy, described by the architect as 'the clearest structure we have done, the best to express our philosophy' (Docomomo, n.d.).S. R. Crown Hall is well-known for its innovative roof, suspended by four steel-plate girders carried by eight exterior steel columns, which allows for a universal space serving as one large room.The building is rectangular in shape, with dimensions of 36.5 m by 67 m and 5.6 m with external columns on a 1.5 m by 1.5 m grid.All four orientations feature a single-glazed enclosure.The landscaping surrounding the building was purportedly designed to disperse solar radiation during the summer season (when leaves are present) but to allow for passive solar heating during the winter season (when leaves have fallen).The predominantly Honey Locust tree canopy surrounding Crown Hall was installed shortly after its completion based on directions by then-faculty member Alfred Caldwell who espoused a regional approach to selecting tree species.The building is in a humid continental climate zone (Köppen classification) and ASHRAE Climate Zone 5A.This climate is characterized by large seasonal temperature differences, often humid summers, and cold winters.
The upper floor comprises a main 'center core' enclosed on three sides by free-standing 3 m high white-oak panelled walls designed to allow multiple uses and functions in the space, which often hosts guest events and visitors to the space.The main teaching spaces are studios at the east and west ends, with two free-standing glass cubicles for seminars, and staff and faculty offices located on the west and east sides introduced in 2013.Studio areas are characterized by desks that are arranged in lines.The location of investigation in this work is the studios on the main floor of the building, which occupy approximately half of the upper level of the building and is where most of the students work.The building also must now account for different comfort necessities that were not present in the initial design and construction.For example, at the time of its opening, the building hosted around 120 students; today the building regularly hosts over 550 students.
The building has also experienced significant modifications to meet the evolving needs of IIT's College of Architecture.In 2005, under careful oversight since it was already a National Historic Landmark, the building underwent a $15 million USD restoration/rehabilitation process (Sexton, 2017), which aimed to restore the building back to its original condition after nearly 50 years from early renovation work in 1976.The architectural firm of Krueck Sexton Partners collaborated on this project with Gunny Harboe of McClier.Additionally, Atelier Ten and Transsolar Klima Engineering conducted simulation studies estimating improvements in energy performance following façade, lighting and HVAC system renovations (Ten & Solar, 2003).The 2005 façade renovation work included restoring the original metal frame and complete glass replacement on the main building level.The upper clear glass was replaced from ¼ ′′ (6 mm) to ½ ′′ (13 mm) float glass.The lower translucent glass maintained the original ¼ ′′ (6 mm) clear-tempered sandblasted glass.In 2016, a new roof assembly was installed, including a metal roof deck, two ½ ′′ layers of roof board (the first layer was mechanically fastened to the roof deck and modified bitumen membrane and the second layer was adhered using low-rise polyurethane adhesive), and 60 mm thick reinforced PVC roof membrane adhered to the primed roof deck with contact adhesive.As shown in Figure 1, a solar array was introduced on the roof in 2018.Another prior simulation-based analysis on the building and its 2005 renovations predicted that the renovations would reduce energy use and improve thermal comfort (Zakrzewski, 2012); however, Crown Hall remains one of the most energy-intensive buildings on the IIT campus, with an energy use intensity (EUI) of 458.3 kWh/m 2 [145.5 kBtu/ft 2 ] recorded in 2019 (City of Chicago, 2023), and no empirical data on occupant comfort in this building exist to our knowledge.
The building is now served by three constant volume air-handling units (AHUs) located in the penthouse (and enabled with economizers), distributing air in the space through 66 diffusers and with the return air path surrounding the envelope perimeter, as shown in Figure 2. All three air-handling units are controlled by Delta Controls direct digital control (DDC) systems.Based on the building sequence of operations (SOO), the AHUs operate from 6 am until 9 pm and the building operator makes adjustments as needed.Given the large volume of the space, the building automation system utilizes an optimal start for both heating and cooling.The AHU starts up to two hours before the scheduled building opening time if the return air temperature is below the heating setpoint or above the cooling setpoint before the building is scheduled to be occupied.Two different types of heating systems are utilized: radiant floor heat and forced air from the ceiling.In winter, the floor heating system is only on when the outdoor air is too cold, and the building operator decides to manually turn on the pumps.This condition usually occurs when the outdoor air temperature is less than −6°C, and the indoor zone temperature does not reach beyond 19°C and does not meet the setpoint requirement.The heating setpoint only controls the operation of AHUs and not the floor heating.Consequently, the sequence of operations for winter accounts for night setbacks and the contribution of solar heat gain to warm up the building in the morning and early afternoon, meaning the space temperature is generally colder in mornings than in the afternoons to avoid overheating the space in the afternoons.The air-conditioning system was introduced in 1985-1986 for cooling of the upper level and it was upgraded in the 2005 renovation.The system essentially overrides the original hand-adjusted natural ventilation louvre system located at floor level on the upper level along the perimeter of the building.Overall, the system has trouble maintaining the space set point given the sequences of operations provided.

Thermal comfort surveys
A thermal comfort survey was prepared using ASHRAE Standard 55 as a basis (ASHRAE, 2017), and informed by prior field research investigations (Zagreus et al., 2004).The survey was deployed to participants during field investigations on four separate days spanning two seasons between November 2018 and September 2019, encompassing two fall (cooling season) surveys and two winter (heating season) surveys.Participants, who were all students in IIT's College of Architecture, completed the thermal comfort questionnaire during regular studio hours that took place from 14:00 to 18:00 (the survey was distributed at 16:30 and collected by 17:50 on each survey day).The survey time was chosen to capture commonly high usage of the space.Participants were given paper-based surveys that asked them to report their current thermal sensation vote, level of satisfaction with comfort and perceived level of productivity, each on a 7-point scale, which ranged from 'too cold' to 'too hot', 'very dissatisfied' to 'very satisfied' and 'interfere with productivity' to 'enhance productivity', respectively.Collection of these data allowed for calculating the Actual Mean Vote (AMV) and the Percentage of People Dissatisfied (PPD) among surveyed occupants (Fanger, 1972).Also, we recognize that there are other means of assessing productivity such as in-depth interviews as deeper subjective assessments or analysis of objective outputs as indirect assessments (e.g.worker output, test scores, etc.) (Al Horr et al., 2016), but were not utilized herein.The surveys also collected demographic data (i.e.age and gender), data on the participants' location on a floor plan of the space, data on each occupants' pattern of usage in the space, and included questions regarding sources of discomfort, subjective adaptive responses, and other general comments.A copy of the surveys is included in the online supplemental material; note that after the first three survey deployments, the first survey was modified slightly to that shown second in the online supplement.The study was approved by IIT's Institutional Review Board (IRB-2020-24).

Indoor environmental monitoring
Indoor environmental monitoring was conducted during each field investigation using Onset HOBO U12-013 two-channel temperature (accuracy: ±0.35°C from 0°C to 50°C) and relative humidity (RH; accuracy: ±2.5% from 10% to 90% RH) data loggers deployed at eight locations throughout the space (Figure 3).The vertical height of each data logger was at desk level (∼90 cm high).The data loggers were positioned to capture indoor spatial and temporal variability through readings logging at 1-min intervals over the course of approximately two days during each field investigation.Two data loggers were placed within the envelope perimeter zone (EPZ) at ∼3 m from the interior face of the exterior envelope and distributed among the two studio areas (i.e. one in the southwest corner and one in the southeast corner).The two data loggers closest to the enclosure also had external TMC20-HD temperature probes (accuracy: ±0.25°C from 0°C to 50°C) that were attached to two glazed surfaces of the building to capture interior enclosure surface temperatures (one in the southwest corner and one in the southeast corner).Six data loggers were distributed diagonally towards the interior of the building to capture gradients between the EPZ and the core (i.e.centre) zone (CZ).The data loggers located from the EPZ to CZ also had Onset TMC20-HD external temperature probes connected and placed inside a small (38 mm diameter) matte black plastic ping-pong ball to record a measure of globe temperature, which was used to estimate the mean radiant temperature using Equation (1).
where T MRT is the mean radiant temperature (°C), T g is the globe temperature (°C), T db is the dry bulb air temperature (°C), V a is the air velocity (not logged; assumed to be 0.01 m/s, but verified with periodic instantaneous readings), D is the globe diameter (0.038 m) and ε is the emissivity of the globe (assumed 0.9 for black surfaces).
Outdoor weather data during each sampling period were also collected from the online platform from Wunderground [12].

Results
Table 1 summarizes the four field investigations, including the measurement time frame, the predominant space conditioning mode during each survey (e.g.heating or cooling), concurrent outdoor environmental conditions collected during each occupant survey period (i.e. from 16:30 to 17:30), and the total number of survey responses and the gender breakdowns of survey respondents.A total of 557 participants completed the comfort surveys across the four measurement sessions, with the number of participants per survey session ranging from 118 to 163.There was a relatively consistent gender distribution across the four surveys, with the percentage of female respondents in each survey ranging from 41% to 47%.
Since the surveyed population were students in an architecture programme, there was some overlap in which some individuals completed more than one survey.Therefore, Table 1 also displays the turnover rate between the two cold weather surveys and between the two warm weather surveys.Overall, 71% of survey respondents on 6 March 2019 (survey #2) did not complete the 7 November 2018 survey (survey #1), meaning that 29% of the survey respondents were the same  individuals, although they were not necessarily seated in the same locations from one survey to the next.There was less turnover between the consecutive surveys in September 2019, with 41% of survey respondents on September 11, 2019 (survey #4) also completing surveys two days prior on 9 September 2019 (survey #2).
Outdoor temperatures during the two survey campaigns conducted in the heating season were both below freezing; average outdoor temperatures during the 16:30-17:30 time frame of each winter survey period were approximately 3°C on 7 November 2018, and −10°C on 6 March 2019.Average outdoor temperatures during the 16:30-17:30 time frame of each survey period conducted in the cooling season were approximately 25°C on September 9, 2019, and 30°C on 11 September 2019.To provide context for these ambient environmental conditions, Figure 4 shows the cumulative distribution of hourly outdoor temperatures for a typical meteorological year (TMY, 2007(TMY, -2021) ) in Chicago, IL, with peak hourly values from each of our four survey days in 2018 and 2019 overlaid on the plot (Climate.OneBuilding, 2023).The peak temperatures from our survey days are daily low temperatures for Surveys #1 and #2 (heating season) and daily high temperatures for Surveys #3 and #4 (cooling season).The results show that among the four survey days, 6 March 2019 (survey #2), is in the bottom 2% of hourly TMY data, representing one of the coldest days, very similar to a peak design day for heating.On the other hand, 11 September 2019 (survey #4) is in the top 99% of hourly TMY data, representing one of the warmest days, very similar to a peak design day for cooling.7 November 2018 (survey #1) and 9 September 2019 (survey #3) were less extreme, with a peak low temperature in the bottom 22% of hourly TMY data for survey #1 and a peak high temperature in the top 89% of hourly TMY data for survey #3, each representing a fairly typical cold and hot day, respectively, but not necessarily peak conditions.Thus, weather conditions during the survey periods are broadly representative of a wide range of climate conditions during heating and cooling seasons, while also capturing essentially peak design conditions for both seasons.To also understand occupants' thermal history and how this impacts their assessment of thermal comfort (Fadeyi, 2014), Table 2 compares additional weather parameters from each experimental survey day, denoted as actual meteorological year (AMY), to the same parameters from a TMY day with the same month and date as the test days.The list of variables includes temperature, relative humidity (RH), global horizontal radiation (GHR), direct normal radiation (DNR), diffuse horizontal radiation (DHR), wind direction, wind speed, total sky cover and the precipitation length.These data demonstrate that the weather conditions during our experimental survey days were reasonably in line with those from TMY.
Figure 5 provides a general understanding of the occupants' usage of the space obtained from the surveys.Approximately 40% of the surveyed population reported spending 20-30 h per week in the building, and most (52%) reported spending most of their time in the building between 12:00 and 18:00.
Figures 6-9 illustrate individual comfort survey responses from surveys #1 through #4, respectively, overlaid with their noted location on the floor plan of S. R. Crown Hall, which was accomplished by mapping where respondents marked their location during the time of the surveys.This allows for visualizing a distribution of thermal comfort survey responses across the space of the architecture studios in which surveys were deployed.Also provided in Figures 6-9 are useful metadata for interpreting the visualized data, including the survey date and time, concurrent outdoor weather conditions during the survey period, number of respondents, gender make-up of respondents, as well as the average indoor air temperature and estimated mean radiant temperature at each of the six spatially distributed indoor environmental measurement locations during the survey period.From these figures, one can observe relatively stark contrasts in indoor air and mean radiant temperatures between the locations near the exterior envelope versus locations closer to the core of the building during the cold weather (heating mode) surveys, while differences were not as drastic during the warmer weather (cooling mode) surveys.Moreover, these figures also show high variability in comfort responses within the space and apparent clusters of responses in some locations.These individual survey response data and environmental data from each monitoring location are analysed in further detail in subsequent sections.

Analysing thermal comfort survey responses
Results from the thermal comfort surveys are analysed further in Figures 10-12, presented as distributions of thermal sensation votes (TSV, Figure 10), level of satisfaction (LOS, Figure 11) and level of productivity (LOP,  Figure 12) across all participants, one for each survey session.Several striking results were observed.First, the mean thermal sensation was lower during warm outdoor conditions (i.e.surveys #3 and #4) when the central air-conditioning system was providing cooling as compared to during cold outdoor conditions during which the systems were in heating mode.The mean thermal sensation vote was −1.81 during survey #3 when the outdoor temperature was 25°C and −0.96 during survey #4 when the outdoor temperature was 30°C.The mean thermal sensation votes during the two cold weather surveys (surveys #1 and #2) were at approximately 0.5 points or more higher: −0.51 and −0.40, respectively.These counterintuitive observations of mean thermal sensation votes suggest that the space is likely overcooled by the central forced-air cooling system during periods of warmer weather.Interestingly, mean thermal sensation votes were not as cool during survey #4 when the outdoor temperature was 30°C (and the full day was closer to design day conditions) as compared to survey #3 when the outdoor temperature was 25°C.This suggests that the extent to which the building was overcooled was reduced at higher outdoor temperatures, albeit still overcooled even when operating at approximately ambient design conditions.
Second, these distributions of thermal sensation votes appear to be associated with a general level of dissatisfaction with thermal comfort in all four surveys.The fraction of people reporting thermal sensation votes within a typical range of acceptability (i.e. between −1 and +1) was only 53%, 71%, 33% and 66% for surveys #1 through #4, respectively, suggesting a high proportion of respondents feeling either warm or cool rather than neutral.Correspondingly, mean levels of satisfaction with thermal comfort during the two cold weather surveys (surveys #1 and #2) were −0.54 and −0.50, while mean levels of satisfaction with thermal comfort during the two warm weather surveys (surveys #3 and #4) were −1.25 and −0.48.The fraction of people reporting thermal comfort satisfaction votes within typical acceptability bounds (i.e. between −1 and +1) was only 53%, 63%, 46% and 58% for surveys #1 through #4, respectively, meaning that the percentage dissatisfied (i.e.out of those defined bounds) ranged from 37% to 54%.This means that in all four surveys spanning two different seasons, the level of satisfaction   with thermal comfort fell far below the typical acceptability criteria of 80%-90% of people satisfied (ASH-RAE, 2017).
Third, a substantial fraction of respondents reported that the air temperature in the space at least somewhat interfered with their productivity.The mean productivity votes were −0.87, −0.55, −1.12 and −0.6 across surveys #1 through #4.At least 25% of respondents rated either a −2 or −3 for their productivity vote in all four surveys (i.e.37%, 27%, 39% and 29% for surveys #1 through #4, respectively).These observations suggest that through a combination of overcooling during cooling seasons and perhaps a combination of overheating and under-heating during heating  seasons, depending on an occupant's location within the space, contributed to a high level of dissatisfaction with comfort and a feeling that the thermal conditions in the space at least moderately interfered with productivity, on average.

Gender differences in thermal comfort responses
Figures 13 and 14 show the population frequency distributions of thermal sensation votes (TSV), level of satisfaction (LOS) and level of productivity (LOP) responses, broken down by gender, for the two cold weather surveys (surveys #1 and #2 combined) and the two warm weather surveys (surveys #3 and #4 combined), respectively.Similarly, Table 3 shows mean values for each parameter and p-values resulting from Mann-Whitney Utests applied to male and female responses.
Female respondents reported lower levels of thermal sensation, lower satisfaction with comfort and greater negative impacts of comfort on productivity across all surveys and in both seasons, and most comparisons were statistically significant (p < 0.05).More specifically, in the cold weather surveys, the mean responses by male participants for thermal sensation, thermal satisfaction and impact on productivity were −0.27, −0.31 and −0.46, all of which were near 'neutral', while the mean responses by female respondents for each metric were −0.69, −0.93 and −0.97, respectively, all of which were at least 0.5 points lower than males (and all three comparisons were significant at p < 0.05).Similarly, the mean responses by male respondents for thermal sensation, thermal satisfaction and impact on productivity were −1.15, −0.56 and −0.52 during the warm weather surveys, while the mean responses by female respondents for each metric were −1.83, −1.37 and −1.41, all of which were again at least 0.5 lower than males (and all three comparisons again were significant at p < 0.05).These findings are in alignment with previous studies demonstrating that females tend to report higher sensitivity levels to cooler environments than males (Beshir & Ramsey, 1981;Kim et al., 2013;Langevin et al., 2015;Nakano et al., 2002;Parkinson et al., 2021).They also further demonstrate that while both genders reported cooler sensations during the warm weather (i.e.cooling season) than during the cold weather heating seasons, overcooling during warm weather had a greater negative impact on the perception of thermal comfort in female respondents than in male respondents.

Spatial differences in thermal comfort responses and environmental conditions
Given the differences in mean indoor air temperatures and estimated mean radiant temperatures, as well as the identification of clusters of individual comfort responses, that were observed between the indoor monitoring locations closer to the centre core of the building compared to those located closer to the perimeter near the building envelope (Figures 6-9), this section explores spatial differences in thermal comfort responses and environmental conditions in the building.For this analysis, we define two versions of an envelope perimeter zone (EPZ) that delineates between spaces that are near the exterior façade (and thus potentially influenced by environmental conditions near the façade) and spaces that are far from the exterior façade.In one scenario, we assumed an imaginary border between the envelope perimeter zone (EPZ) and the centre zone (CZ) was located at a depth of 4.5 m (15 ft) from the façade and in another scenario we assumed the EPZ border was located at a depth of 7.6 m (25 ft) from the façade (Leaman & Bordass, 2006) (shown in Figure 15).The shallower assumption represents a standard definition of an EPZ in modern construction, whereas the deeper assumption represents a definition of an EPZ in which the thermal impacts of the highly glazed modernist façade might be reasonably felt at a greater distance away from the interior surface of the façade.Anyone not located in a defined EPZ was thus located in the central zone (CZ) associated with that EPZ; thus, the CZ boundaries moved with the definition of EPZ.The marked locations of each survey respondent were used to locate them either in the EPZ or in the corresponding CZ.These two definitions of the depth of spatial gradient between façade adjacent EPZ and CZ locations allow for an exploration of the impacts of being in the perimeter zone in this building (with its low thermal resistance glass and steel façade) versus being located within the core of the building.
Mann-Whitney nonparametric U-tests were used to compare comfort survey responses between participants located in the EPZ and the CZ locations under the two assumptions for EPZ depth, as shown in Table 4.To evaluate whether any observed differences in comfort responses were attributable to differences in gender distributions among the two zones, Table 5 shows a comparison of the gender distribution of survey respondents who marked their location as in the EPZ or the CZ, assessed as the percentage of female respondents in each zone (this calculation was repeated for both EPZ depth scenarios).
During the two cold weather heating season surveys, the mean responses for thermal sensation, level of satisfaction and productivity interference were consistently lower for participants located in the EPZ than those located in the CZ, although only some comparisons were statistically significant (p < 0.05).For example, in survey #1, when the EPZ was defined as 4.5 m, the mean thermal sensation votes were −1.2 and 0.03 for  respondents located in the EPZ versus CZ, respectively, and were −0.8 and 0.29 for respondents located in the EPZ versus CZ, respectively, when the EPZ was defined as 7.6 m; both comparisons were statistically significant (p = 0.001).In survey #1 with a defined EPZ depth of 4.5 m, the gender distribution of those reporting both gender and location were similar among the EPZ and CZ (45% female in the EPZ and 48% female in the CZ), which suggests that the differences in thermal sensation between respondents in the EPZ and CZ for this survey (and at this EPZ depth) were not influenced by gender distribution but rather reflect true spatial differences.When the EPZ was defined as 7.6 m for survey #1, thermal sensation votes were still statistically significantly lower for respondents in the EPZ compared to the CZ despite the EPZ having a greater proportion of female respondents (39% female in EPZ and 56% in CZ), who, according to Figures 13  and 14, reported lower levels of thermal sensation.This further suggests that for this cold weather survey, lower thermal sensation votes in the EPZ at either assumed depth were likely largely attributable to the spatial proximity to the façade.
Conversely, although comfort responses in the EPZ were lower than the CZ for all comparisons for survey #2 (the second cold weather survey), none of the comparisons were statistically significant (although two comparisons, TSV and LOS, were borderline at p < 0.1).Regardless, both comparisons suggest that during the cold weather heating surveys, occupants located in the EPZ reported colder sensations, lower levels of satisfaction, and greater levels of productivity interference due to thermal comfort than those located in the CZ, suggesting that heat losses through the low thermal resistance envelope in this building negatively impacted comfort for those occupants who are located near the façade.Interestingly, the only significant spatial differences in comfort responses between occupants in the EPZ and CZ occurred during survey #1, when outdoor temperatures were milder than in survey #2 (3°C versus −10°C); this may be indicative of differences in heating system operation that yielded local differences in indoor environmental conditions or perhaps differences in adaptive responses among survey respondents.
During the first warm weather cooling season survey (survey #3), the mean responses for thermal sensation, level of satisfaction and productivity interference were consistently higher for participants located in the EPZ than in the CZ, with all comparisons statistically significant at p < 0.1 and both TSV and LOS comparisons significant at p < 0.05.This contrasts with the cold weather (heating season) surveys and suggests that there was a greater extent of overcooling in the CZ than in the EPZ.In contrast, those located in the EPZ may have benefited from the additional warmth experienced near the highly glazed facade.Moreover, when the EPZ was extended from 4.5 to 7.6 m, mean TSV and LOS votes decreased in the EPZ, suggesting that proximity to the CZ led to colder sensations, which further suggests overcooling was occurring in the core space.
There were smaller differences in comfort responses between those in the EPZ and CZ during survey #4, and no comparisons were statistically significant.Interestingly, the outdoor temperature was higher during survey #4 (30°C) than during survey #3 (25°C).This, combined with results from the two cold weather heating season surveys, in which significant differences in comfort responses were observed between those in the EPZ versus the CZ only for the milder 3°C day (but not for the −10°C day), suggests that the interactions between the heating and cooling systems that serve this building and heat transfer through the exterior façade of this building yield significant spatial differences in comfort for those located in the EPZ versus the CZ during relatively mild conditions, but not as much during more extreme (hot or cold) conditions.This may be indicative of equipment sizing issues in which heating and cooling systems may overheat or overcool different parts of the building, especially at part-load conditions.Notably, gender distributions between the EPZ and CZ during both warm weather surveys were similar (less than 6% absolute difference for all comparisons), which suggests that the observed differences reflect spatial differences and not gender differences.
To provide additional context to the observed spatial differences in thermal comfort responses, Figure 16 shows distributions of dry bulb temperatures (T db ) and mean radiant temperatures (T MRT ) measured in the EPZ and CZ during each of the four surveys.Box plots show distributions of 1-min interval data recorded between 16:30 and 18:00 during each survey.In this analysis, only the EPZ definition of 4.5 m is used; data from the sensors located at measurement stations 1C, 1E, 2C and 2W were used to characterize the CZ while data from the sensors located at measurement stations 1W and 2E were used to characterize the EPZ.
During the two heating mode surveys, the mean radiant temperature was consistently lower in the EPZ than in the CZ, with median values in the EPZ typically 3-4°C lower than in the CZ, which generally aligns with the lower thermal sensation votes observed in the EPZ during these surveys.Median dry bulb temperatures were within approximately 1°C of T MRT in both zones during the milder survey #1 (3°C outdoor temperature).Still, they deviated more drastically in both zones during the colder survey #2 (−10°C outdoor temperature).One noticeable difference between survey #1 and survey #2 is the higher zone dry bulb temperature in survey #2 which is due to the operation of the floor heating system in this cold day.However, T MRT values appear more representative of thermal sensations in this space than dry bulb measurements, better capturing the radiant asymmetry introduced in the space by the combination of modernist enclosure elements and non-original HVAC systems.
Contrary to the heating season, during the two cooling surveys, there were smaller differences in dry bulb and mean radiant temperatures measured between the EPZ and CZ, and smaller differences between dry bulb and T MRT within the EPZ or CZ.This suggests that there is a more uniform spatial temperature pattern of indoor environmental conditions in this space under cooling mode operation, likely attributable to the forced-air cooling system that also contributes to mixing air in the space.This is especially evident during survey #3 with milder ambient conditions (25°C outdoor temperature), which resulted in similar median temperatures between the EPZ and CZ and narrow ranges of temperatures within each zone.Median indoor dry bulb and mean radiant temperatures were also relatively low during this survey (i.e.approximately 21-22°C).Slightly lower temperatures in the CZ during this survey also support the findings of lower thermal sensation votes in the CZ than the EPZ during these surveys, even though the absolute magnitude of differences in temperatures was relatively small.Conversely, both dry bulb and mean radiant temperatures were higher during the warmer weather survey #4 (30°C outdoor temperature), and temperatures in the EPZ had higher variability than in the CZ.Worth noting is that mostly cloudy conditions persisted during the warmer surveys, which likely deemphasized the importance of envelope interactions during warm weather periods in this building relative to what might be expected on sunnier days.Thermal comfort in context: occupant responses and adaptive behaviours This section explores data collected from the comfort surveys regarding relationships between comfort responses, sources of discomfort, subjective adaptive responses and other general comments that participants made.First, Table 6 shows Spearman rank correlation coefficients between the three main quantitative comfort outcomes (i.e.TSV, LOS and LOP) for each of the four surveys.While none of the Spearman correlation coefficients were above 0.4, most comparisons among these parameters were significant at p < 0.05; and more comparisons were significant during cooling mode surveys than during heating mode surveys.These findings suggest that these measures of thermal comfort are generally correlated in this building and study population, but the overall strength of the relationship was not particularly robust.Thus, these findings also suggest that other factors may contribute to occupant satisfaction and the impact of the indoor environment on occupant productivity in this space than thermal sensation alone.The non-thermal drivers in occupant responses might be attributable to a variety of personal factors such as variations in clothing level, metabolic activity, age group and gender, cultural differences, expectations and physiology, or other workspace features.This is consistent with other prior work demonstrating that other non-temperature parameters can impact satisfaction and productivity such as noise and air quality, amount of space, visual privacy and other architectural features (Frontczak et al., 2012).
To further explore occupant responses and adaptive behaviours reported in the survey responses, Figures 17 and 18 summarize responses to contextual survey questions, including which factors that perceptions of thermal discomfort can be attributed to and which elements in the space occupants reported adjusting to control the comfort in their workspaces.Note that this portion of the survey was deployed for both heating season surveys (surveys #1 and #2) but only for the first cooling season survey (survey #3); results from all three are shown in Figure 17.The questions were then modified, and a slightly different set of questions were deployed during the second cooling season survey (survey #4, summarized separately in Figure 18).
Figure 17 summarizes the percentage of respondents from each of the first three surveys who, if they experienced thermal discomfort, reported which factors they thought contributed to their discomfort.Respondents could select as many factors as applicable.During the two heating season surveys (surveys #1 and #2), the most frequently reported factors were, paradoxically, 'my workspace is colder than other areas' (9% and  19%) and also 'vented air is too hot' (20% and 13%).These responses provide further evidence of the presence of hot and cold spots within the space during heating modes.Other frequently reported factors were 'cold floors and walls' (12% and 14%) and 'too much or too little air movement' (13% and 13%).
During the one cooling season survey (survey #3), over a third of respondents (38%) reported that the 'vented air is too cold' and over a quarter of respondents (27%) reported that 'my workspace is colder than other areas'; both responses provide further evidence of a high level of overcooling in this space during the cooling season.Figure 17 also demonstrates that approximately half of all respondents in both seasons reported adjusting their clothing to suit their comfort level, and consistently 18%-24% relocate to another place in the building  in search of comfort and 12%-22% reported leaving the building to work elsewhere to control comfort.
Figure 18 summarizes responses from just the last cooling survey (survey #4) in which participants were asked to rate their level of satisfaction with a broad range of comfort parameters, including building maintenance, visual comfort, and building cleanliness, sound privacy, temperature, noise level, building cleanliness, the comfort of furniture, ease of interaction and air quality (Brager et al., 2015), as well as a ranking of the parameters that respondents reported most impact their overall level of satisfaction with their workspace.In the left side panel of Figure 18, the shaded areas of the horizontal box plots show interquartile ranges, and the whiskers show minimum/maximum values; median values are also shown numerically in the figure.Temperature, comfort of furniture and furniture adjustability had the lowest satisfaction ratings among the broad range of comfort parameters, whereas the amount of space, colours and textures, visual comfort, amount of light and building cleanliness had the highest satisfaction ratings.Similarly, in the right-side panel of Figure 17, the most frequent parameter that respondents reported most impacted their level of satisfaction with their workspace was thermal comfort, followed by space furnishings.
Finally, additional comments were collected via the surveys, in which occupants indicated a combination of thermal and non-thermal reasons for discomfort and dissatisfaction in the space.Some of the noted comments include the following: -Too hot during winter and too cold during summer.
-I sit under the supply, which is warm but too much air movement, cold when I leave the workspace.-Its either too hot or too cold, never comfortable.-The heat is turned off at night, which makes the Crown freezing -Better furniture.
-My back always is sore because of the stools.I have started paying for chiropractor to try and make it better.-The wall is pouring in cold air from poor seals.
-Since I am close to the window, my area is cold and the heater over doesn't heat anything.

Discussion
As stated previously, this work describes what we understand to be the first empirical assessment of perceived occupant thermal comfort in Mies van der Rohe's iconic S. R. Crown Hall, and, to the best of our knowledge, the first such empirical assessment in any Mies building, and one of a small number of empirical assessments of comfort in any well-known buildings from the Modern Movement.Crown Hall and other buildings from its era are known, at least anecdotally and through practice, to encounter challenges in meeting modern thermal comfort or energy efficiency standards, especially as they have experienced changes in their occupancy, programming and systems to adapt to changing uses over time.However, this study provides empirical data and analyses of perceived comfort that can be useful for helping to better understand the extent of the problems in this building and inform potential solutions.It may be premature to propose specific retrofitting solutions, but some broad generalizations can be made, and there is some potential for winwin solutions that could improve comfort and reduce energy consumption at the same time.
Occupants reported high levels of dissatisfaction with thermal comfort across all four surveys and across both seasons, with high spatial variability.Particularly notable were markedly high levels of dissatisfaction with the indoor thermal environment during the cooling season due to excessively cool sensations.Such findings are contrary to expectations in Crown Hall with its high thermal transmittance building enclosure, which would typically be expected to yield excessively warm sensations in warm/hot weather, especially for those located within close proximity to the exterior walls and windows, as a glass and metal facade will influence thermal comfort by long-wave and shortwave radiation (Huizenga et al., 2006;La Ferla et al., 2020).On the contrary, during the summer surveys, occupants located closer to the exterior walls and windows expressed greater satisfaction with comfort, whereas those located in the core of the building reported colder sensations and greater discomfort.These findings demonstrate a general state of overcooling during the warm weather surveys.There are several possible reasons for overcooling, including but not limited to system oversizing, poor sequences of operation controls, and layout and type of the supply and return outlets.For instance, as Figure 2 shows, the system uses a plenum return distributed across the entire perimeter close to the window (a 10 cm gap between the drop ceiling and the ceiling) while the round supply ceiling diffusers are distributed evenly in the space.This layout allows a better air mixing of the cold supply air when it interacts with the warm fenestration surface in summer and provides a partially mixed and stratified air distribution from the floor to ceiling.This air diffusion in the perimeter zones enables the secondary air in the occupied zone to achieve a more comfortable air temperature.However, in the core zone, the supply and return layout leads to a fully stratified air in the room with the cold air being stagnant close to the floor.In addition, in this air diffusion pattern, the thermostat or any temperature control is less sensitive to load changes (Int-Hout, 2015).
Another solution to potentially improve the thermal comfort of the core zone occupants could be to increase the discharge air temperature (increasing the temperature setpoint for the core zone).For this building, the temperature difference between the discharge air and the room temperature is significant because of the high cooling load in the space in the afternoons in order for the cold supply air to travel further in the space (Arens et al., 2015).However, as Figure 2 illustrates, this may require re-arrangement of the AHU connections to the existing ductwork.Regardless, the presence of overcooling suggests that the cooling setpoint could be raised at least for the core zone air-handling unit during these periods to provide dual benefits of lower energy consumption and increased levels of comfort (de Dear et al., 2015;Hoyt et al., 2015;Kim et al., 2019;Nicol, 2004;Silva et al., 2013).It is worth noting that during these same two surveys, a small number of respondents reported sensations leaning towards 'too hot', which would likely increase at an extended set point and thus the comfort of that population may need to be addressed in a different way (e.g. via more personalized comfort approaches (Kim et al., 2018;Zhang et al., 2015)).Also consistent with prior literature (Lee & Ham, 2021), significant individual differences in comfort responses herein further demonstrate the potential utility of more personalized approaches to comfort.
Similarly, during the two heating season surveys, 18% and 25% of respondents reported sensations leaning towards 'too hot', which further suggests that energy savings may be possible by relaxing set points and meeting local comfort needs in other ways.However, interactions with the exterior enclosure were more predictable than during the cooling season surveys, as those located closer to the exterior reported colder sensations and lower levels of satisfaction than those in the core of the building.These results suggest that the combination of the highly glazed façade and the layout of heating and cooling systems in this building means that occupants located near the exterior enclosure are at a disadvantage for comfort during cold weather but at an advantage for comfort during warm weather, all else being equal.In addition, given the poor performance of ceiling diffusers during the heating season, utilization of a secondary system such as the floor heating that is currently installed in the building but not in operation at all heating times (i.e. it operates mostly during the coldest days) may improve the heating season thermal comfort of occupants.Results from the heating season environmental measurements also demonstrated that measurements of indoor mean radiant temperatures were more indicative of comfort, especially during the colder of the two survey days, than indoor air temperature.This is consistent with prior literature (Chaudhuri et al., 2016;Gan, 2001) and suggests that one potential solution to improve comfort in this space may be to deploy radiant temperature sensors throughout the space to better capture these effects and exert greater control over the heating systems.
Another potential improvement could be to the sequences of operation for this building.While the system is capable of measuring space temperature, it would be conservative to assume that the thermostat setpoint in the building is not the main controller of the room temperature, but rather that the space temperature is monitored by the return sensor, which should provide a fairly accurate reading on space temperature.The return air temperature difference variable, which is the difference between the return air temperature and the return air temperature setpoint, impacts the discharge air temperature and ultimately the thermal comfort satisfaction of the occupants.The concern is that in this building with a tall ceiling and subject to a high solar heat gain, it would have heavily stratified air in the cooling conditions, and the return air temperature would be measuring a temperature much higher than the space temperature the occupants would experience.Due to this, it would be recommended that either a thermostat-based control is enabled, or the space is monitored closely to determine the correlation, if any, between return air temperature and operative temperature that a thermostat would sense.This work is not without its limitations.For one, it must be acknowledged that occupant surveys of thermal comfort provide only snapshots in time, and results can change under different weather conditions, times of day and occupancy characteristics (Nicol & Roaf, 2005).Thus, while we attempt to use these data to draw broad conclusions about the space, we also acknowledge that our surveys are limited to the days, times and occupants who completed them.Two, we attempted to work closely with the building automation system to understand and collect additional variables such as temperature values in the AHUs, but due to the limitations in data recording and implementations in our existing system, we were not able to fully accomplish this task.Third, given the scope of this paper to focus primarily on the survey outcomes, we did not introduce any computational fluid dynamics or energy modelling, but it would be useful for future studies to understand how temperature distributions in the room can affect the thermal comfort and ultimately the energy use in this space.
Ultimately, as Mies's Modern S.R. Crown Hall and many other buildings from the Modern Movement have exceeded their design life of approximately 50 years and are now vulnerable to long-term degradation due to corrosion, moisture, and structural deterioration, innovative approaches to improving thermal and energy performance while preserving their experimental character are much needed (Carmichael, 2012).

Figure 1 .
Figure 1.(a) Descriptive images of S. R. Crown Hall at Illinois Institute of Technology; (b) the building studio area; (c) aerial view of the solar panels placed on the building roof; (d) description of the main floor space and plan and the allocation of the studio area; (e-g) Infrared images if the exterior south façade and interior of the building; (h and i) images of solar panel array located on the rooftop of the building.

Figure 2 .
Figure 2. (a) The location of AHUs, (b) the location of the ductwork, (c) the location of the ductwork above the drop ceiling and (d) the study area with the ceiling diffuser and the return plenum.

Figure 3 .
Figure 3. (a) Location of the study space and eight indoor environmental monitoring locations, (b) example interior glazed surface temperature sensor and (c) example indoor monitoring station setup.

Figure 4 .
Figure 4. Cumulative distribution of the outdoor air dry-bulb temperatures in Chicago, IL from TMY (2001-2021), with peak daily values for the actual survey days overlaid.Actual peak temperatures are minimum hourly peak temperatures on 7 November 2018 and 6 March 2019 (heating periods) and maximum hourly peak temperatures on 9 September 2019 and 11 September 2019 (cooling periods).

Figure 5 .
Figure 5. Participant responses to questions regarding time spent in the building and time of day typically spent in the building.

Figure 6 .
Figure 6.Summary diagram showing individual comfort survey responses overlaid with the location on the floor plan of S. R. Crown Hall where respondents marked their location during the time of the 7 November 2018 survey (survey #1), as well as the average indoor air temperature and estimated mean radiant temperature at each of the six spatially distributed indoor environmental measurement locations during this survey.

Figure 7 .
Figure 7. Summary diagram showing individual comfort survey responses overlaid with the location on the floor plan of S. R. Crown Hall where respondents marked their location during the time of the 6 March 2019 survey (survey #2), as well as the average indoor air temperature and estimated mean radiant temperature at each of the six spatially distributed indoor environmental measurement locations during this survey.

Figure 8 .
Figure 8. Summary diagram showing individual comfort survey responses overlaid with the location on the floor plan of S. R. Crown Hall where respondents marked their location during the time of the 9 September 2019 survey (survey #3), as well as the average indoor air temperature and estimated mean radiant temperature at each of the six spatially distributed indoor environmental measurement locations during this survey.

Figure 9 .
Figure 9. Summary diagram showing individual comfort survey responses overlaid with the location on the floor plan of S. R. Crown Hall where respondents marked their location during the time of the 11 September 2019 survey (survey #4), as well as the average indoor air temperature and estimated mean radiant temperature at each of the six spatially distributed indoor environmental measurement locations during this survey.

Figure 11 .
Figure 11.Distribution of level of satisfaction (LOS) responses from the four surveys in S. R. Crown Hall.LOS responses range from 'very dissatisfied' (−3, dark purple) to 'very satisfied' (+3, yellow).

Figure 13 .
Figure 13.Population frequencies for thermal sensation, level of satisfaction and level of productivity responses across the two cold weather surveys (surveys #1 and #2), by gender.Mean values by gender are shown in dashed lines.

Figure 14 .
Figure 14.Population frequencies for thermal sensation, level of satisfaction and level of productivity responses across the two warm weather surveys (surveys #3 and #4), by gender.Mean values by gender shown in dashed lines.

Figure 15 .
Figure 15.Description of two scenarios of a hypothetical border defining an envelope perimeter zone (EPZ) that delineates between spaces that are near the exterior façade and spaces that are far from the exterior façade: 4.5 and 7.6 m.

Figure 16 .
Figure 16.Box plot representing the spatial variation of the measured indoor environmental parameters during each of the four survey periods (time: 16:30-18:00).T db = dry bulb temperature; T MRT = mean radiant temperature; EPZ = envelope perimeter zone; CZ = centre zone.

Figure 17 .
Figure 17.Participant responses to survey questions associated with sources of discomfort and adaptive responses across the first three surveys (two heating seasons and one cooling season).

Figure 18 .
Figure 18.Contextual responses from survey #4 during the cooling season.Left: participant responses to rating their level of satisfaction with indoor comfort parameters.Right: participant responses to parameters that most impact their level of satisfaction in their workspace.

Table 1 .
Summary of the four field investigations in S. R. Crown Hall.

Table 2 .
Summary of weather data for the days of experiments using AMY and TMY data (average ± standard deviation).

Table 3 .
Gender breakdown in mean thermal sensation, level of satisfaction and level of productivity votes across the study surveys, along with p-values resulting from Mann-Whitney U-test of statistical significance.

Table 4 .
Statistical comparisons of the differences in thermal sensation, level of satisfaction and level of productivity votes for respondents in the envelope perimeter zone (EPZ) and in the centre zone (CZ) across the four surveys.

Table 5 .
Comparison of the gender distribution of respondents located in the envelope perimeter zone (EPZ) versus the central zone (CZ), assessed as the percentage of female (%F) respondents in each zone, calculated for the two EPZ depth scenarios.
BUILDING RESEARCH & INFORMATION

Table 6 .
Spearman Rank Correlation between thermal sensation votes, level of satisfaction and level of productivity.