Professional Documents
Culture Documents
net/publication/334697927
Design with Comfort: Expanding the psychrometric chart with radiation and
convection dimensions
CITATIONS READS
0 247
4 authors:
Some of the authors of this publication are also working on these related projects:
IoT@NUS - Smart Devices and Energy Savings for Buildings View project
All content following this page was uploaded by Eric Teitelbaum on 26 July 2019.
Abstract
We present a makeover of the psychrometric chart using a new color-shading
method that allows the chart to be considered comfortable based on the varia-
tion of non-air related comfort parameters such as mean radiant temperature,
air movement, and the transitional behavior of occupants. These represen-
tations allows us to think outside the thermal comfort box with the use of
innovative thermal design and comfort feedback for occupants. Based on
the Olgyay bioclimatic chart allowing architects to “Design with Climate”,
the new chart representations are applied over a wide range of conditions
to illustrate a physical basis for expanding comfort zones. An open-source
repository and web app is available for designers and researchers to reproduce
the charts and color-shading for their own projects.
Keywords: Psychrometric Chart, Data Visualization, Radiant Cooling,
Indoor Environmental Quality (IEQ), Thermal Comfort, Adaptive
Comfort, Visual Analytics
1 1. Introduction
2 The psychrometric chart is a widely-used visualization of the properties
3 of air for the purpose of building systems design and operations. Since the
4 nearly parallel publications of the first Mollier diagram in 1904 and the first
5 psychrometric chart by Willis Carrier in 1911, Anglo-American organizations
6 like ASHRAE have published the psychrometric chart extensively for use by
∗
Corresponding author email: eteitelb@princeton.edu
2
45 chart do not directly show tradeoffs between the mean radiant temperature,
46 air temperature, and air speed. Particularly, when invoking the definition
47 of operative temperature in equation 1, by inspection one can see that if
48 tr < ta < tskin as would likely be the case with a radiant cooling system,
49 the operative temperature increases when air speed increases. Such a rela-
50 tionship makes designing a radiant system with a particular setpoint range
51 difficult to extract from the operative temperature framework.
hr tr + hc ta
to = (1)
hc + hr
52 While in general, adding complexity to static psychrometric axes increases
53 in complexity and specificity quickly, the goal of this paper is to build on pre-
54 vious depictions of comfort zones, expanding the comfort zone to the entire
55 psychrometric chart allowing the future use of novel radiant and convection-
56 based systems. The online webtool component of this chart allows users
57 to dynamically change inputs, providing a platform to visualize more di-
58 mensions of the heat balance approach to thermal comfort, contextualizing
59 tradeoffs between air temperature, mean radiant temperature, and air speed
60 with greater detail than traditional adaptive comfort diagrams.
61 Building on the diagrams from Arens [4] and Kessling [5], this framework
62 is the first to truly expand the air based comfort zone and graphically provide
63 answers for how radiation can be used to compensate for any air tempera-
64 ture, or what is the lowest acceptable wind speed for a given metabolic rate.
65 Probing the corners of the psychrometric chart with a heat balance can allow
66 for creative system design, perhaps leveraging ‘free’ heat from evaporation
67 [12] or return condensate [13]. In short, this tool can allow for a simple
68 high-level analysis of the following conditions:
69 • High air temperature (> 30◦ C) compensated for with a low mean ra-
70 diant temperature and no air speed
71 • Low air temperature (< 15◦ C) with a high mean radiant temperature
73 • Required metabolic rate for the presence of wind, i.e. would certain air
74 speeds be uncomfortable in a cool gym
3
75 • Compared to tools such as the online CBE comfort tool [6], this tool
76 can produce a wider range of air speed and metabolic rate, although
77 they have yet to be experimentally verified.
92 2. Methodology
93 Crucial to the notion of designing with comfort and recoupling comfort
94 and architecture is the need to visualize the allowable dynamic shifts be-
95 tween comfort parameters when fixing comfort. Conveying this to designers
96 intuitively can enable integration between geometric form and comfort. Just
97 as the Olgyays’ research presented information on environmental dynamics
98 to enable designing with climate, there remains a gap still bridging comfort
99 and design, human and system, across all comfort variables. However, unlike
100 the 1960’s, we are no longer limited by two dimensional plots, and can ex-
101 plore the dynamism, on which climate, comfort, and controls can be linked
102 by design.
103 Color is a convenient parameter to help illustrate the dynamic nature
104 of comfort variables, particularly when used as a tool to demonstrate the
105 tradeoff between the change in one comfort variable and the required com-
106 pensation by another. Adding a degree of freedom to the thermal perfor-
107 mance and design of spaces imparts a similar degree of freedom to spatial
108 and programmatic design by opening the door to the separation of comfort
109 and ventilation air. For example, energy benefits have been demonstrated
4
110 when latent loads (dehumidification) can be separated from sensible loads
111 (cooling) [14, 15], as these two processes are linked with a conventional air
112 conditioning system yet often are very different in magnitude [16]. Similarly,
113 radiant systems [12] and personal comfort systems [17] can help separate
114 comfort conditioning from required ventilation air. However, recent research
115 has shown that in buildings with radiant systems, there often exists very little
116 separation between the air temperature and mean radiant temperature [18],
117 failing to take advantage of the separation of ventilation and comfort loads.
118 We hypothesize that visualizing the relationship between air temperature,
119 air velocity, humidity and radiant temperatures through color gradients is a
120 step towards using geometric design to engage directly with heat transfer for
121 designing with comfort.
122 Victor and Aladar Olgyay pioneered the modern approach to designing
123 with climate [2], recognizing that comfort is dictated by air temperature,
124 humidity, and motion, as well as radiation. These basic modes are always
125 either enhanced or diminished by local climate and associated thermody-
126 namics, and interact with humans through clothing level, metabolic rate,
127 skin temperature, and skin wettedness. The modality of these interactions
128 was fully characterized by P.O. Fanger [19], whose comfort model, still com-
129 monly used today, characterized how human variables on the one hand and
130 ambient variables on the other interact to produce comfort or discomfort.
131 However, air temperature and humidity are still the two independent vari-
132 ables that are manipulated, with the other parameters always being set as
133 variables which controlled for, an unnecessary restriction of the comfort zone
134 that does not inherently optimize comfort design performance.
135 Applying a reductionist approach to thermal comfort yields 7 key vari-
136 ables acting within 3 interconnected modes of heat and mass transfer (radia-
137 tive, convective with associated evaporative transfers, neglecting conduction
138 for this analysis). The 7 variables are air temperature, Tair , mean radiant
139 temperature, TM RT , relative humidity, %RH, skin temperature, Tskin , air
140 velocity, vair , skin wettedness, w, and metabolic rate, M R. Clothing level
141 is also an important factor to consider, however this is a variable parameter
142 that can easily be added later to calibrate a specific model, similar to one’s
143 sitting versus standing position.
144 For this abstracted model, a human comfort comfort condition is defined
145 as equality between heat removed and metabolic rate. Metabolic rates are
146 well defined and independent of size, and are therefore a robust metric to
147 serve as the basis of this analysis [20]. Specifically 1.2 met (69.8 W/m2 ) is
5
148 the metabolic rate of an individual performing light office work such as typ-
149 ing, and 2 met (116.3 W/m2 ) is one’s metabolic rate when walking briskly.
150 These values of metabolic rate were chosen to serve as the boundary of the
151 office space comfort zone, or in other words this range describes the operating
152 range for a building system to remove heat between 69.8 and 116.3 W/m2
153 from an individual.
154
155 Detailed calculations of the human heat balance presented used in this
156 paper are presented in the supplementary materials, and will be briefly re-
157 viewed here.
158 Heat transfer between a building and an occupant can occur through
159 radiative and convective exchanges, as well as convection linked explicitly to
160 evaporative mass transfer, and conductive exchanges. Conduction from the
161 person to a surface they are in contact with was neglected for this study,
162 since this type of conduction is highly environment-dependent and therefore
163 difficult to quantify, rather than the other three modes which are always
164 present and empirically calculable. The following equation demonstrates the
165 objective function for comfort computation.
Qevap + Qconv + Qrad = M R (2)
166 Each heat transfer (Q) is proportional to a temperature difference be-
167 tween either the air temperature, Tair , or the mean radiant temperature ,
168 TM RT , and the occupant’s skin or clothing temperature, Tskin , with the ap-
169 propriate heat transfer coefficient. The same is true for mass transfer through
170 evaporation, except using the partial pressure difference between water on
171 the skin and water in the air instead of a temperature difference. Other im-
172 portant parameters are skin wettedness, w, which is a dimensionless number
173 ranging from 0.06 to 0.8 that describes the fraction of skin covered in sweat
174 [20]; air speed, vair , which changes the rate of heat removed through convec-
175 tion and evaporation; emissivity of the skin or clothing, ε, which can vary
176 the amount of heat transfered radiatively. The clothing insulation level, clo,
177 was not set as a free parameter for this analysis, but will be added in future
178 versions of the software. Either the mean radiant temperature or air speed
179 is left as a free parameter, and then the required value for this parameter
180 for comfort is calculated and plotted on a psychrometric axis. This is done
181 automatically by the solver using the following order of operations.
182 1. The user is prompted to select for a “Mean Radiant Temperature” or
183 “Air Speed” output.
6
184 2. If the user selects “Mean Radiant Temperature”, air speed is the re-
185 quired input.
186 • If the air speed value is less than 0.2 m s−1 , then a free convection
187 solver is used to compute the heat balance. Otherwise, a forced
188 convection solver is used.
189 3. If the user selects the “Air Speed” output, the difference between the
190 air temperature and the mean radiant temperature, T air − T M RT , in
◦
191 C is the required input. For example, an input value of 5 means that
192 the mean radiant temperature is 5 ◦ C below the air temperature for
193 each air temperature on the calculated psychrometric chart. This value
194 may be 0, meaning the two are equivalent.
1
www.comfortch.art
7
Additional variables
Figure 1: Expanded and annotated psychrometric comfort zone - The background color
gradient represents the required mean radiant temperature to maintain comfort conditions,
at a given air temperature and humidity, holding metabolic rate constant at 1.2 met and
skin wettedness at 0.06 for free convection. All areas shaded with color can be considered
comfortable if the MRT is at the value that corresponds with the color legend. All graphics
in this publication can be interpreted in a similar way.
8
Metabolic Rate: 2 (met); Skin Wettedness: 0.06; Free Convection 14.8
100% 5.4
90%
25 80% 70%
13.4
60%
Humidity Ratio (g water/kg dry air)
20 6.0 12.0
50%
6.6
10.7
15 40%
9.3
8.4
30%
10
9.6 2
10. .8
7.8
10 4
7.9
11.2.0
20%
1
9.0
7.2
5
6.6
10%
123. .2 8
1 13.
6
0 5.2
5 10 15 20 25 30 35 40 MRT C
Air Temperature C
Figure 2: Expanded psychrometric comfort zone with elevated metabolic rate. The back-
ground color gradient represents the required mean radiant temperature at a given air
temperature and humidity, holding metabolic rate constant at 2 met and skin wettedness
at 0.06 for free convection.
224 2.1.1. Interpretation guide - the whole chart as the comfort zone
225 These charts eliminate a rigid comfort zone, and instead treat the entire
226 area covered by the gradient as a potentially comfortable region. So long as
227 the required temperature or air speed condition shown by the color gradi-
228 ent is met by the system, the energy balance is favorable for comfort. For
229 example, air temperature, relative humidity, and mean radiant temperature
230 measurements would be taken, using Tair and %RH to position the point on
9
Metabolic Rate: 1.2 (met); Skin Wettedness: 0.4; Free Convection 42.8
100% 21.0
90%
25 80% 70%
22.5 39.5
60%
Humidity Ratio (g water/kg dry air)
20 36.1
25.5 50%
32.7
15
24.0
28.5 40%
27.
0 29.3
31.5 30%
10 30.0
34.5 26.0
33.0 20%
37.5 36.0
5
39.0 22.6
10%
40.5
0 19.2
5 10 15 20 25 30 35 40 MRT C
Air Temperature C
Figure 3: Expanded psychrometric comfort zone with elevated skin wettedness rate. The
background color gradient represents the required mean radiant temperature at a given air
temperature and humidity, holding metabolic rate constant at 1.2 met and skin wettedness
at 0.40 for free convection.
10
231 the x, y axes, and the color of the point is chosen from the measurement of
232 TM RT at the location of an occupant. A mismatch between the color of the
233 plotted point and the color of the gradient at the location of the point would
234 indicate an uncomfortable scenario. Because there are a number of internal
235 thermoregualtion mechanisms the human body has available, thermal com-
236 fort would be achieved within a TM RT range of the point rather than at only
237 the precisely specified temperature.
238 If the air velocity is known, the same set of plots can be drawn for a
239 fixed air velocity, addressing the remainder of an occupant’s thermal comfort
240 needs with a modulated radiant temperature. Since this is a convection dom-
241 inated regime, the mean radiant temperature required to produce comfort
242 at even high air temperatures is not particularly large, and there is nearly
243 a one-to-one relationship with air temperature and required mean radiant
244 temperature.
245 vair = 0.4 m/s is a high air velocity, and also is an air velocity that
246 would be hard to maintain and likely wouldn’t be guaranteed with a passive
247 natural ventilation system. A lower air velocity is more realistic. Figure 5 is
248 a redrawn version of Figure 4 but with a lower vair = 0.2 m/s.
249 One of the major advantages of using the standard psychrometric chart
250 as a design aid is superimposing local climate data on the axes to calculate
251 demand, assess the feasibility of evaporative cooling, and generally design
252 for the local climate [2]. The same type of assessment can be incorporated
253 in these expanded psychrometric charts, by examining lines of maximum
254 potential for different types of cooling for supplying chilled water to a hy-
255 dronic radiant system. Reverting to the base case from Figure 1, which for
256 free convection shows the TM RT required for comfort at M R = 1.2 met and
257 w =0.06, lines are added in Figure 6. These lines represent three conditions.
258 First, when the comfort condition is provided with TM RT = Tair , as indi-
259 cated by Line A at approximately 22 ◦ C. What is particularly interesting
260 about this case is that this TM RT = Tair is a commonly held assumption,
261 however this data shows that it is most comfortable, i.e. requiring the least
262 additional thermoregulation by one’s body, for a very narrow subset of air
263 temperatures between 21 and 22 ◦ C. While this is a common setpoint for
264 office buildings, this line makes it clear why small fluctuations in radiant
265 temperature such as next to an exterior wall or cold window, fluctuations in
266 air velocity, i.e. under a duct, or fluctuations in even air temperature itself
267 can cause discomfort. This is a powerful insight gained with the inclusion of
268 the mean radiant temperature gradient that is missing in a standard chart
11
Metabolic Rate: 1.2 (met); Skin Wettedness: 0.06; vair: 0.4 m/s 36.3
100% 2
90%3.2
25 80% 70%
24. 24 34.1
8 0 .
60%
Humidity Ratio (g water/kg dry air)
20 26. 31.9
4 50%
28. 29.6
15 0
25. 40%
29. 6
6 27.4
31. 27.
2 2 30%
10
3 28.
3 2.8 8 25.1
34 3.6 30. 20%
.4 4
32.
5 0
22.9
10%
35
.2
0 20.6
5 10 15 20 25 30 35 40 MRT C
Air Temperature C
Figure 4: Expanded psychrometric comfort zone with forced convention. The background
color gradient represents the required mean radiant temperature at a given air temperature
and humidity, holding metabolic rate constant at 1.2 met and skin wettedness at 0.06 for
forced convection at vair = 0.4 m/s.
12
20.8
Metabolic Rate: 1.2 (met); Skin Wettedness: 0.06; vair: 0.2 m/s 27.2
100% 21.2
90%
70%
25 21.680%
26.2
22.0 60%
22.4
Humidity Ratio (g water/kg dry air)
20 25.1
50%
23.2
22.8 24.1
15 40%
24.0
23.6 23.0
24.8 30%
10
25.6 24.4
22.0
20%
25.2
5 26.4
20.9
10%
26.8
0 26.0 19.9
5 10 15 20 25 30 35 40 MRT C
Air Temperature C
Figure 5: Expanded psychrometric comfort zone with forced convention at a low velocity.
The background color gradient represents the required mean radiant temperature at a
given air temperature and humidity, holding metabolic rate constant at 1.2 met and skin
wettedness at 0.06 for forced convection at vair = 0.2 m/s.
13
Line C:
TMRT = Tdp
Line B:
TMRT = Twb
Line A:
TMRT = Tair
Figure 6: Redrawn version of Figure 1 (w =0.06, M R = 1.2 met, free convection), with
lines added to visualize potential cooling sources. Line A is Tair = TM RT , implying radiant
cooling is not required, or air source heat pumps would suffice. Line B represents TM RT =
Twb , implying the use of standard wet bulb evaporative cooling could effectively supply a
hydronic radiant cooling system for comfort in this range. Line C is TM RT = Tdp , which
shows the upper limit of M-Cycle evaporative cooling, but also the risk of condensation
with conventional radiant cooling panels. Reproduced with permission from [23].
14
280 points above this line would require the entire room to display a mean radi-
281 ant temperature below the dew point for comfort, or have a couple of panels
282 significantly below the dew point. There would likely be convective issues
283 associated with cooling the air rather than radiantly cooling the occupants,
284 or radiantly cooling opposite and adjacent surfaces below the dew point as
285 well [24, 21] However, new radiant technologies are currently under investiga-
286 tion to reduce the crossover between convection with a radiant surface while
287 simultaneously avoiding condensation in humid environments [25]. Likewise,
288 careful design of spaces with reflective surfaces opposite radiant panels can
289 mitigate condensation risk on non-activated surfaces.
290 Mean radiant temperature is defined as “the uniform temperature of an
291 imaginary enclosure in which the radiant heat transfer from the human body
292 is equal to the radiant heat transfer in the actual non-uniform enclosure” by
293 the ISO 7762 standard [26]. This means that mean radiant temperature itself
294 is used as an abstraction, removing view factor from radiant heat transfer cal-
295 culations in the built environment, since the mean radiant temperature refers
296 to an arbitrary space with a view factor of 1. Therefore, the lines in figure
297 6 are generated with a view factor of 1 and represent the best possible per-
298 formance scenario. In reality, colder temperatures (or warmer, for heating)
299 would be required as the view factor decreases from 1, and this calculation
300 can be implemented in future versions of the tool allowing designers to spec-
301 ify the view factor of a radiant system and the required supply temperature
302 calculated to produce the required mean radiant temperature. However, this
303 is still rooted in the mean radiant temperature abstraction, and does not
304 include any specifics of the human body or occupant orientation or spatial
305 location.
306 If radiant cooling below the dew point is deemed acceptable, line C also
307 indicates the theoretical limit of dew point evaporative cooling [27]. Anything
308 above this line would require further cooling to achieve radiant temperatures
309 below the dew point. Examples could be using a liquid desiccant to first
310 dehumidify air for evaporative cooling similar to the DEVap project [28],
311 to extract additional sensible cooling. Regeneration of the desiccant and
312 subsequent cooling would still need to be addressed.
313 2.2. Representation of air movement from free and forced convection
314 If assumptions are made about the mean radiant temperature in a space,
315 then the same methodology from the previous section can be applied for
15
Metabolic Rate: 1.2 (met); Skin Wettedness: 0.06; Tair = TMRT
100%
90%
25 80% 70% 1.8
60% 1.6
Humidity Ratio (g water/kg dry air)
20 1.4
50%
1.2
15 40% 1.1
30% 0.9
10
0.7
20%
5 0.5
10% 0.3
0 0.1
5 10 15 20 25 30 35 40 m/s
Air Temperature C
Figure 7: Expanded psychrometric comfort zone with ventilation. The background color
gradient represents the required air velocity and humidity, holding metabolic rate constant
at 1.2 met and skin wettedness at 0.06 for free convection. Additionally, it is assumed
that TM RT = Tair . Reproduced from [23].
316 calculating the air velocity required for achieving comfort if that is a tun-
317 able parameter. Such analysis could be useful for design of natural ventila-
318 tion systems without radiant systems. Operating under the assumption that
319 TM RT = Tair , one can again set metabolic rate and skin wettedness constant
320 and solve for the remaining parameter at each air temperature and relative
321 humidity, air velocity.
322 Personal ventilation systems are an exciting area of research [29] since
323 fan technology now draws very little energy. Focusing on the personal, local
324 environment is another strategy for saving energy as well, as making people
325 comfortable is ultimately the goal of conditioning, rather than making build-
326 ings comfortable as a proxy for human thermal comfort. Because convection
327 grows nonlinearly as a function of air velocity witnessed in the convective heat
328 transfer coefficient, convection can provide comfort conditions even in warm
16
329 environments. Knowledge of the required air velocity for thermal comfort
330 is vital to these models, and these visualizations show not only required air
331 velocities for each temperature and humidity, but also reflect the nonlinear
332 trend.
333 Figure 7 shows a gradient solving for required air velocity when the mean
334 radiant temperature is set equal to the air temperature. The metabolic rate
335 in this plot is 1.2 met with a skin wettedness of 0.06. Figure 7 shows rel-
336 atively low air speeds can provide comfort with outdoor air, however the
337 difficulty is ensuring the air is always supplied at the same temperature and
338 velocity. As alluded to in de Dear et al. [30], there is a neurophysiological
339 basis for feeling more uncomfortable when comfort conditions suddenly shift
340 to uncomfortable conditions, even if momentary. Especially with air veloc-
341 ity and temperature, small fluctuations can lead to a feeling of discomfort
342 because of the strong nonlinearity of forced convection.
343 Figure 8 shows the required wind velocity for an elevated metabolic rate
344 of 2 met but same skin wettedness of 0.06 and TM RT = Tair assumption as
345 in Figure 7. Again, many intuitively uncomfortable conditions could now be
346 considered comfortable with the correct air velocity.
347 In both figures, a lower bound of vair = 0.1 m/s and an upper bound of 2
348 m/s were chosen, as 0.1 m/s is typically the onset of forced convection, and
349 above 2 m/s, air velocities are noticeably uncomfortable.
17
Metabolic Rate: 2 (met); Skin Wettedness: 0.06; Tair = TMRT
100%
90%
25 80% 70% 1.8
60% 1.6
Humidity Ratio (g water/kg dry air)
20 1.4
50%
1.2
15 40% 1.1
30% 0.9
10
0.7
20%
5 0.5
10% 0.3
0 0.1
5 10 15 20 25 30 35 40 m/s
Air Temperature C
18
2.5
Metabolic Rate: 2.3 (met); Skin Wettedness: 0.6; Free Convection 38.0
100% 5.
90%0
25 80% 70%
7.5 32.7
60%
Humidity Ratio (g water/kg dry air)
20 27.4
12.5 50%
22.1
15
10.0
17.5 40%
15.
0 16.8
22.5 20.0 30%
10
25.0
27.5 11.4
20%
5
32.5 30.0 6.1
10%
35.0
0 0.8
5 10 15 20 25 30 35 40 MRT C
Air Temperature C
Figure 9: Expanded psychrometric comfort zone solving for the required TM RT at 2.3 met
and a skin wettedness of 0.6, that might be indicative the transition from a bike ride to
entering a building.
19
Metabolic Rate: 2.3 (met); Skin Wettedness: 0.6; Tair = TMRT
100%
90%
25 80% 70% 1.8
60% 1.6
Humidity Ratio (g water/kg dry air)
20 1.4
50%
1.2
15 40% 1.1
30% 0.9
10
0.7
20%
5 0.5
10% 0.3
0 0.1
5 10 15 20 25 30 35 40 m/s
Air Temperature C
Figure 10: Expanded psychrometric comfort zone solving for the required vair at 2.3 met
and a skin wettedness of 0.6, that might be indicative the transition from a bike ride to
entering a building.
366 a standard building scenario, yet provides insight into the trends for required
367 radiant temperatures.
368 Figure 10 shows how well convection is able to treat the same conditions
369 of 2.3 met and a skin wettedness of 0.6. Because the large partial pressure
370 difference between skin with such a high value for w is high, a significant
371 portion of the heat generated at 2.3 met can be shed to the air with even
372 modest air velocities. Therefore, forced convection is a better method for
373 cooling in such a transitional space rather than radiant cooling.
374 In developing the model for the expanded psychrometric framework for
375 transient scenarios, several areas were identified in which the framework can
376 demonstrate counter-intuitive phenomena. A prime example occurs for non-
377 steady-state cooling models, where an individual with w > 0.06, or in other
378 words an individual with a certain amount of sweat on their skin attempting
379 to shed heat to the environment will experience competing effects of radiant
20
Metabolic Rate: 1.2 (met); Skin Wettedness (complex): 0.4; Free Convection 25.9
100%
90%
25 80% 70%
24.0
60%
Humidity Ratio (g water/kg dry air)
20 22.0
19 50%
.0
20.1
15
17.0
20 40%
.0
18.2
30%
21
10
.0
22.
16.3
0
20%
23.
0
5 .0 .0
18 16 .0 14.4
15 10%
24
0 12.5
.0
5 10 15 20 25 30 35 40 MRT C
Air Temperature C
Figure 11: Expanded complex psychrometric comfort zone solving for the required TM RT
at 1.2 met and a skin wettedness of 0.4. Drier air requires a cooler TM RT due to the
competing effects of radiant and evaporative cooling.
380 and evaporative cooling. Because water is at best 20% transparent in the
381 domain that humans radiate, λpeak = 9.5µm. Therefore water on the skin
382 would be radiantly exchanging with a radiant cooling system at the wet bulb
383 temperature, rather than the skin temperature. This can be modeled as
384 effective surface area through which radiative heat transfer occurs at Tskin (1−
385 w) and skin = 0.95. While water is also highly emissive, water = 0.95,
386 the depressed temperature at the surface of the sweat beads approaching
387 Twb (w) will cause less radiative heat transfer to a radiantly cooled surface.
388 Theoretically, in such environments radiative cooling is less effective, and
389 there may exist interesting practical embodiments. This is modeled in figure
390 11, where an individual with a metabolic rate of 1.2 met and skin wettedness
391 of 0.4 is modeled for free convection heat and evaporative losses, solving for
392 the required mean radiant temperature for comfort.
393 Figure 11 demonstrates that as water on the surface of the skin gets
21
394 cooler, radiant cooling to chilled surfaces is less effective, since the depressed
395 surface temperature of the water towards the Twb reduces the temperature
396 difference driving radiant cooling. Such a competing effect would exist for
397 transient conditions, but helps inform where radiant systems could poten-
398 tially be least effective. Figure 11 includes a gradient over the range where
399 the approximation for Twb is best, above 3%RH [31]. This treatment of
400 exceptional conditions must be validated experimentally, but demonstrates
401 the robustness and internal consistency of this methodology, with tradeoffs
402 emerging from the fundamentals-driven physics engine.
403 3. Discussion
404 Many of the new regions defined as comfortable here are somewhat ex-
405 treme and counterintuitive, for example claiming that an environment with
406 an air temperature of 32 ◦ C and relative humidity of 70% could be comfort-
407 able with no air motion. Based on figure 1, comfort could be achieved with a
408 mean radiant temperature of 19.2 ◦ C. The inverse case of a high mean radi-
409 ant temperature with a low air temperature is intuitive, perhaps a campfire
410 at night or a radiant heater on an autumn night at an outdoor restaurant.
411 Yet no case of this type of mean radiant temperature to air temperature
412 separation exist in the ASHRAE thermal comfort database II [32]. However,
413 this does not mean that the conditions could not be designed with creative
414 thermal systems [12, 25]. Particularly with recent advances on materials for
415 radiant cooling [33, 34], there is a renewed need to understand how occupants
416 physically interact with comfort systems to realize energy savings with air
417 temperature setbacks [35].
418 Validating these regions can be conducted in climate chambers with stan-
419 dard instrumentation, to gain insight to whether these proposed domains are
420 perceived as comfortable by occupants. However, the regions proposed by
421 this paper are generally aligned with an adaptive comfort window. Take
422 the example where operative temperature is, for simplicity, is calculated us-
423 ing equation 3 from ISO 7726 [26]. Also assume that a natural ventilation
424 system is in place with no air heating or cooling system, such that the air
425 temperature in a building is equivalent to the air temperature outside.
√
tr + (ta × 10vair )
to = √ (3)
1 + 10vair
22
35.0 Adaptive Comfort, 1.0 - 1.3 met, vair = 0.3m s 1 40 Expanded Comfort, 1.0 - 1.3 met, vair = 0.3m s 1
80% Acceptability 80% Acceptability
32.5 90% Acceptability 90% Acceptability
30.0
27.5 30
25.0
22.5 25
20.0
20
17.5
15.0 15
10 15 20 25 30 35 10 15 20 25 30 35
Air Temperature C Air Temperature C
Figure 12: (a - left): The adaptive comfort zone with a derived expanded comfort zone
from a radiant system added. (b - right): The adaptive comfort zone transformed into a
mean radiant temperature y-axis, appended with the same expanded comfort range after
transformation.
426 Using the adaptive comfort 80 and 90% acceptability zones for 0.3 m s−1
427 air speed shown in figure 12a, for each operative temperature, to , and air
428 temperature, ta , equation 3 can be used to transform the y-axis of figure
429 12 from Operative Temperature to Mean Radiant Temperature, tr , shown in
430 figure 12. In parallel, the model presented in this paper was run for metabolic
431 rates of 1.0 to 1.3 met for a fixed skin wettedness of 0.06 and an air speed
432 of 0.3 m s−1 . The range of calculated mean radiant temperature required
433 for comfort between 10 and 90 %RH was used as an arbitrary boundary
434 condition. The resulting mean radiant temperature ranges were appended to
435 figure 12b, and then converted to operative temperature ranges and appended
436 to figure 12a.
437 There is good agreement between the two comfort zones, however the
438 standard adaptive comfort criteria are consistently wider than the expanded
439 comfort framework. This is likely due to the constant skin wettedness built
440 in to the expanded comfort model. Similarly, the heat removal required for
441 comfort when handled entirely by a radiant system necessarily narrows the
442 range of acceptible temperatures as the air temperature parameter becomes
443 fixed. Therefore, a primary contribution of this framework is a visualization
444 of the required range of radiant setpoints required to achieve the adaptive
445 comfort criteria with a radiant system using outdoor air with no sensible
446 gains.
447 Moreover, the high degree of overlap between the adaptive comfort frame-
23
448 work and this expanded comfort tool implies that while using a radiant
449 system alone to achieve comfort conditions at high air temperatures may
450 be novel, the resulting operative temperature that produces comfort is not
451 contrary to existing standards. Therein lies the true novelty of the tool,
452 separating comfort contributions to modes of heat transfer to provide design
453 flexibility.
454 4. Conclusion
455 This paper builds on the extensive development of comfort studies that
456 paralleled the commercialization of air conditioning technology. This devel-
457 opment led to a rigid comfort zone, defined by the limits of an air-based
458 approach to thermal comfort, which focuses on management of only temper-
459 ature and humidity . The major contribution of this paper is the expansion
460 of the thermal comfort zone, illustrated on a psychrometric chart, allow-
461 ing for dynamic manipulation and contributions from vair and TM RT . An
462 open-source code base was produced for climate designers to use establish
463 more flexible design of operating conditions controls and setpoints. Directly
464 considering TM RT can enable new ways to couple form and thermal perfor-
465 mance. This code is available in an open-source repository for understanding
466 dynamic relationships between all comfort variables.
2
https://github.com/buds-lab/psychrometric-chart-makeover
3
www.comfortch.art
24
479 acceptable windows, rather than a pure solution to allow for even larger
480 design windows. The mean radiant calculations presented herein still rely on
481 an abstracted mean radiant temperature. Future versions of the tool could
482 use a better human body representation or spatial description.
483 5. Acknowledgement(s)
484 This work would not be possible without Willis Carrier, the Olgyay broth-
485 ers, ASHRAE and the millions of engineers who have made the psychrometric
486 chart ubiquitous. The authors would like to acknowledge Axel Kilian for his
487 guidance during the model development.
488 6. References
489 [1] ASHRAE, ASHRAE Pocket Guide for Air Conditioning, Heating, Ven-
490 tilation, Refrigeration (I-P Edition), ASHRAE, 8th edition, 2013.
491 [2] V. Olgyay, Design with Climate, Princeton University Press, 1963.
492 [3] B. Givoni, Man, climate and architecture, Architectural science series,
493 Applied Science Publishers, London, 2nd ed edition, 1976.
500 [6] S. Schiavon, T. Hoyt, A. Piccioli, Web application for thermal com-
501 fort visualization and calculation according to ashrae standard 55, in:
502 Building Simulation, volume 7.
25
508 [9] S. A. Damiati, S. A. Zaki, H. B. Rijal, S. Wonorahardjo, Field study
509 on adaptive thermal comfort in office buildings in malaysia, indonesia,
510 singapore, and japan during hot and humid season, Building and Envi-
511 ronment 109 (2016) 208 – 223.
26
540 [18] P. R. J. W. S. S. Dawe, M., F. Bauman, Comparison of mean radiant
541 and air temperatures in mechanically-conditioned commercial buildings
542 from over 200,000 field and laboratory measurements, Submitted to
543 Energy and Buildings June (2019).
547 [20] E. Arens, H. Zhang, Thermal and moisture transport in fibrous materi-
548 als, Woodhead Publishing.
554 [22] D. Aviv, Forrest Meggers, Cooling oculus for desert climate – dy-
555 namic structure for evaporative downdraft and night sky cooling, 2017.
556 Manuscript in Review, CISBAT 2017.
560 [24] J. D. Feng, Design and control of hydronic radiant cooling systems,
561 Dissertation, University of California, Berkeley, 2014.
566 [26] I. ISO, 7726, ergonomics of the thermal environment, instruments for
567 measuring physical quantities, Geneva: International Standard Organi-
568 zation (1998).
27
572 [28] E. Kozubal, J. Woods, J. Burch, A. Boranian, T. Merrigan, Desiccant
573 enhanced evaporative air-conditioning (devap): evaluation of a new con-
574 cept in ultra efficient air conditioning, contract 303 (2011) 275–3000.
575 [29] H. Zhang, E. Arens, Y. Zhai, A review of the corrective power of personal
576 comfort systems in non-neutral ambient environments, Building and
577 Environment 91 (2015) 15–41.
578 [30] R. de Dear, Thermal comfort in natural ventilation—a neurophysiolog-
579 ical hypothesis, in: 2010 Windsor Conference: Adapting to Change:
580 New Thinking on Comfort, volume 6.
581 [31] R. Stull, Wet-bulb temperature from relative humidity and air tem-
582 perature, Journal of Applied Meteorology and Climatology 50 (2011)
583 2267–2269.
584 [32] V. F. Liina, T. Cheung, H. Zhang, R. de Dear, T. Parkinson, E. Arens,
585 C. Chun, S. Schiavon, M. Luo, G. Brager, P. Li, S. Kaam, M. A. Ade-
586 bamowo, M. M. Andamon, F. Babich, C. Bouden, H. Bukovianska,
587 C. Candido, B. Cao, S. Carlucci, D. K. Cheong, J.-H. Choi, M. Cook,
588 P. Cropper, M. Deuble, S. Heidari, M. Indraganti, Q. Jin, H. Kim,
589 J. Kim, K. Konis, M. K. Singh, A. Kwok, R. Lamberts, D. Loveday,
590 J. Langevin, S. Manu, C. Moosmann, F. Nicol, R. Ooka, N. A. Oseland,
591 L. Pagliano, D. Petr, R. Rawal, R. Romero, H. B. Rijal, C. Sekhar,
592 M. Schweiker, F. Tartarini, S. ichi Tanabe, K. W. Tham, D. Teli, J. Tof-
593 tum, L. Toledo, K. Tsuzuki, R. D. Vecchi, A. Wagner, Z. Wang, H. Wall-
594 baum, L. Webb, L. Yang, Y. Zhu, Y. Zhai, Y. Zhang, X. Zhou, Devel-
595 opment of the ashrae global thermal comfort database ii, Building and
596 Environment 142 (2018) 502 – 512.
597 [33] A. P. Raman, M. A. Anoma, L. Zhu, E. Rephaeli, S. Fan, Passive
598 radiative cooling below ambient air temperature under direct sunlight,
599 Nature 515 (2014) 540–544.
600 [34] O. Salihoglu, H. B. Uzlu, O. Yakar, S. Aas, O. Balci, N. Kakenov,
601 S. Balci, S. Olcum, S. Suzer, C. Kocabas, Graphene-based adaptive
602 thermal camouflage, Nano letters 18 (2018) 4541–4548.
603 [35] T. Hoyt, S. Schiavon, A. Piccioli, D. Moon, K. Steinfeld, Cbe thermal
604 comfort tool, Center for the Built Environment, University of California
605 Berkeley, http://cbe. berkeley. edu/comforttool (2013).
28
606 [36] R. J. de Dear, E. Arens, Z. Hui, M. Oguro, Convective and radiative heat
607 transfer coefficients for individual human body segments, International
608 Journal of Biometeorology 40 (1997) 141–156.
29
639 Applying a reductionist approach to thermal comfort yields seven key
640 variables acting within three interconnected modes of heat and mass transfer
641 (radiative, convective with associated evaporative transfers, neglecting con-
642 duction for this analysis). The 7 variables are air temperature, Tair , mean ra-
643 diant temperature, TM RT , relative humidity, %RH, skin temperature, Tskin ,
644 air velocity, vair , skin wettedness, w, and metabolic rate, M R. Clothing level
645 is also an important factor to consider, however this is a variable parameter
646 that can easily be added later to calibrate a specific model, similar to one’s
647 sitting versus standing position.
648 For this abstracted model, a human comfort comfort condition is defined
649 as equality between heat removed and metabolic rate. Metabolic rates are
650 well defined and independent of size, and are therefore a robust metric to
651 serve as the basis of this analysis [20]. Specifically 1.2 met (69.8 W/m2 ) is the
652 metabolic rate of an individual performing light office work such as typing,
653 and 2 met (116.3 W/m2 ) is one’s metabolic rate when walking briskly. These
654 values of metabolic rate were chosen to serve as the boundary of the office
655 space comfort zone, or in other words this range describes the operating range
656 for a building system to remove heat between 69.8 and 116.3 W/m2 from an
657 individual.
658 Heat transfer between a building and an occupant can occur through ra-
659 diative and convective exchanges, as well as convection linked explicitly to
660 evaporative mass transfer, and conductive exchanges. Conduction was ne-
661 glected for this study, since conduction tends to be environment-dependent
662 and therefore difficult to quantify, rather than the other three modes which
663 are always present and empirically calculable. The following equation demon-
664 strates the objective function for analysis.
30
672 fer.
Qevap = f (w, %RH, Tskin , vair ); Qconv = f (vair , Tair , Tskin ); Qrad = f (TM RT , Tskin )
(A.2)
673 Each mode is coupled by a dependence on skin temperature, and evaporative
674 and convective modes have a dependence on vair .
675 In models presented by Arens and deDear [20, 36], the three modes
676 are simply portrayed as potential-driven transfer mechanisms as depicted
677 in equation sets A.3 through A.5. Equation set A.3 shows Qconv divided
678 into a two regimes, forced and free convection. Free convection occurs for
679 vair ≤ 0.1 m/s and forced convection occurs for vair > 0.1 m/s. For this
680 study, free convection assumes an individual is sitting, and forced convection
681 assumes an individual is standing (walking) thereby changing the form of hc ,
682 the convective heat transfer coefficient [20].
Qconv = hc (Tskin − Tair ); hc,f ree = 0.78(Tskin − Tair )0.56 ; hc,f orced = 10.4vair
0.56
(A.3)
683 Equation set A.4 shows the dependence of evaporatively generated heat
684 transfer on convective heat transfer as shown in the he , evaporative heat
685 transfer, coefficient, in addition to partial pressure differences, P . In this
686 model, Pskin,sat is the partial pressure of water at the skin’s surface and
687 therefore at Tskin , and Pair is the partial pressure of water in the air at
688 Tair . For an individual that is not sweating but transporting moisture at a
689 background rate, it is assumed that w = 0.06. At the onset of sweating, this
690 number increases to a maximum practical threshold of 0.80 [20].
31
699 useful equation that allows reducing parameters among the modes of heat
700 transfer. While there is a nonuniform relaxation period during which the skin
701 temperature approaches the equilibrium value, a simple linear expression can
702 be defined to more accurately model Tskin as a function of Tair as follows in
703 equation A.6, which fits data with R2 = 0.99.
704 Using these fundamental relationships, a model was developed using Python
705 scripting that allowed for the fundamental parameters to be varied and the
706 resulting topology change plotted. Similar to the models created by Shukuya
707 [38] and Fanger [19], these topologies were created with the intent of defining
708 a comfort zone and performing a generalized system optimization. How-
709 ever unlike other work in the field, the intended goal of this analysis is to
710 provide the user with an intuitive picture of her surroundings, adding more
711 complexity than just air temperature and relative humidity.
32