A concept for non-uniform thermal irradiation emulation in an immersive virtual environment [original]

A concept for non-uniform thermal irradiation emulation in an immersive

virtual environment

Karsten Tawackolian

, Lukas Schmitt , Martin Kriegel

Technische Universit¨

at Berlin, Hermann-Rietschel-Institut, Energy, Comfort & Health in Buildings, Germany

ARTICLE INFO

Keywords:

Thermal comfort

Virtual reality

Indoor environmental quality

Non-uniform environment

ABSTRACT

This study addresses the challenge of emulating human thermal environments without the need to build physical

mock-ups. The aim is to complement the established technology of emulating visual environments with head-

mounted displays. For the user, the virtual environment (VE) emulates an arbitrary indoor environment

including visual and thermal aspects. In contrast to conventional climate chambers, we describe a concept to

emulate arbitrary non-uniform environments and room geometries without the need to construct or modify

separate physical mock-ups for each scenario being tested. By fully emulating air temperatures as well as the

thermal irradiation, a user can experience the same thermal sensation as in the actual environment. An air

conditioning system and nine supply air diffusers are used for the air temperature and humidity control. For

thermal radiation, 45 temperature-controlled surfaces are arranged in a hemicube around the user. An arbitrary

environment’s thermal irradiation is emulated by a projection of temperatures on the hemicube surface elements.

This study compares two methods for setting the surface temperatures in the VE to emulate thermal irradiation.

Finally, it presents the results of computational fluid dynamic simulations that benchmark the capability of the

VE to emulate a non-uniform indoor thermal environment. For the non-uniform thermal environment benchmark

scenario, the assumption of a uniform enclosure (using the surface average radiant temperature) resulted in a

surface root mean square error of 2.2 K (local deviation of up to 4.6 K) of radiant temperature on the person

dummy surface. The projection method was implemented for the VE, utilising temperature-controlled surfaces.

Depending on the projection method used to determine the surface temperatures, it was possible to decrease the

root mean square error of the radiant temperature in the emulated environment to a range of 0.4 K to 0.7 K.

1. Introduction

The quality of indoor environments affects human satisfaction and

health, productivity, and building energy use by influencing human

behaviour [1–2]. In an occupant survey of office workspaces, most

complaints were about acoustics/noise (54 % of respondents), followed

by thermal comfort (38 %) [3]. Nowadays, photorealistic visualisations

are used for decision-making in the early architectural design stages.

However, it is still challenging to demonstrate non-visual factors. This

bears the risk that non-visual aspects receive less attention. For example,

large glazed surfaces may be visually appealing but could cause thermal

discomfort due to radiation asymmetry or draughts [4]. Virtual envi-

ronments (VE) are an enabling technology to demonstrate and examine

concepts early and holistically [5–6]. Although the term virtual reality

(VR) became famous with the advent of three-dimensional visualisations

and head-mounted displays it encompasses the stimulation of multiple

senses [7–8]. Virtual environments that emulate thermal conditions, in

addition to visual aspects, could facilitate decision-making processes for

the built environment [9].

Fig. 1 illustrates the main factors and challenges of the present work

which is focused on the emulation of thermal conditions in an immersive

virtual environment. The task is to emulate a non-uniform thermal

environment where the user should experience the same thermal

sensation as in the actual environment. From a thermodynamic stand-

point, the human heat balance with the environment and the basic heat

transfer mechanisms need to be considered. A climate chamber concept

for a VE is examined that enables to control the radiative heat transfer as

well as the convective heat transfer. Thus, a laboratory for arbitrary non-

uniform indoor thermal environments is created. In combination with a

head-mounted display with headphones, it is then possible to emulate a

visual, acoustic, and thermal environment.

An air conditioning system is used to control the air temperature and

humidity of a virtual environment. However, human thermal sensation

* Corresponding author.

E-mail address: [email protected] (K. Tawackolian).

Contents lists available at ScienceDirect

Energy & Buildings

journal homepage: www.elsevier.com/locate/enb

https://doi.org/10.1016/j.enbuild.2024.114748

Received 14 March 2024; Received in revised form 26 July 2024; Accepted 30 August 2024

Energy & Buildings 322 (2024) 114748

Available online 1 September 2024

is also affected by the vertical air temperature distribution and air ve-

locities and turbulence caused by ventilation and natural convection (i.

e. draughts) [10,11].

Thermal radiation has also to be considered for thermal comfort

[12]. It is of paramount importance for radiative surface heating or

cooling systems [13] or for persons sitting close to large glazed surfaces

[4]. The temperatures and locations of surfaces in relation to the occu-

pant (view factors) are decisive for the radiative heat transfer [14]. In

the Nußelt (1928) [15] analogy for radiation heat transfer, an external

environment is projected onto a hemisphere, leading to the concept of

solid angles. Each visible surface patch of the environment is replaced by

a surrogate surface patch on the hemisphere with the same view factor.

Instead of a hemisphere, a hemicube can also be used [16]. The hemi-

cube projection was first introduced by Cohen and Greenberg (1985)

[17] for calculating diffuse illuminance (visual irradiation) on surfaces

but can also be applied for diffuse thermal irradiation.

In the proposed VR Lab (Fig. 2) temperature-controlled surrogate

surfaces arranged in a hemicube are used for projecting the thermal

irradiation of an enclosing environment. A benchmark case for this

approach of a non-uniform environment was defined and simulated with

computational fluid dynamic (CFD) simulations. Air temperatures and

velocities and radiation were simulated for the benchmark case and the

VR Lab. The selected benchmark is a surface-heated room with a cool

outer wall with windows. An occupant was either seated in non-uniform

conditions close to the cool wall or in nearly uniform conditions on the

opposite side of the room. The occupant’s distance to the side wall

(lateral) was 1 m and the distance to the back wall (dorsal) 0.5 m in both

cases.

2. Literature survey

2.1. Climate chambers

Climate chambers are widely used to provide reproducible condi-

tions for thermal comfort studies [18,19]. Schweiker et al. [20]

reviewed studies that included multiple types of stimuli, such as thermal

and visual. They found that studies with simultaneous stimuli were often

inconclusive, and the results were partly contradictory. Climate cham-

bers that are capable of emulating non-uniform thermal environments

are less common. Beyer et al. [21] noted that for thermal comfort, the

radiation asymmetry and the vertical air temperature gradient are

connected factors. In their subject study, a heated ceiling caused not

only a thermal radiation asymmetry but also a perceptible vertical air

temperature stratification. Gao et al. [22] re-assessed the currently used

comfort limits for asymmetries due to cool and warm walls and observed

smaller acceptance ranges for longer durations than previous studies. A

typical example of a climate chamber for investigating indoor heating

systems is described in the Appendix and used as a benchmark case in

this work. Conventional climate chambers require physical construction

or modification to create a specific visual or thermal environment. This

inspired the development of virtual environments that do not require

physical mock-ups for each scenario.

2.2. Virtual environments

With a head-mounted display, the visual aspects of an environment

can be emulated without needing to construct it, creating a virtual

environment (VE). Alamirah et al. [23] conducted a review of virtual

environments for comfort research and listed 13 environmental factors

Nomenclature

λAir thermal conductivity

Air density

Air dynamic viscosity

IThermal irradiation on a surface

FView factor for radation

TTemperature

TMRT Mean radiant temperature

UU-Value, heat transfer coefficient for walls and windows

y+Non-dimensional wall distance

Fig. 1. Thermal conditions.

Fig. 2. VR Lab.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

that were varied in past studies. Most studies were about visual aspects

(20) and the operative temperate was varied in 5 studies. None of the

reviewed studies varied the radiant temperature, humidity, air velocity,

and ventilation type. Table 1 summarises a literature survey on virtual

environments (VE) with a focus on indoor environments (In), air tem-

perature control (AT), radiation temperature control (RT), air velocity

control (AV) and thermal comfort (TC). Latini et al. [24–25], Saeidi et al.

[26], Yeom et al. [27] and Ozcelik and Becerik-Gerber [5] conducted

subject tests in office environments. They examined how wearing a

head-mounted display affects thermal sensation and the perception of

the virtual environment and largely found that the thermal experience of

participants was similar in the physical environment and virtual envi-

ronment. Vittori et al. [28] developed a laboratory for human comfort

and occupant-building interaction with a VR component. Compared to

the concept in the present study, they arrived at different design de-

cisions on the layout, the heating system, and the ventilation system.

The main difference is that their climate chamber was built to be easily

adaptable for different room layouts by modifying the physical structure

whereas the present study intends to create a climate chamber that can

emulate arbitrary non-uniform thermal environments without changing

the physical construction for each scenario. Previous studies successfully

used head-mounted displays to increase the flexibility in generating

visual environments but were still limited in the flexibility of generating

non-uniform thermal environments.

2.3. Devices for emulating thermal and airflow stimuli

Fans, infrared lights, and Peltier elements were used in the past to

generate stimuli for occupants in virtual environments [29]. Stimuli

were mainly used to increase immersion but not to establish precise non-

uniform indoor thermal conditions. Hülsmann et al. [30] equipped a

three-sided virtual environment (CAVE) with an array of ceiling-

mounted fans and infrared lights to emulate outdoor thermal and

wind stimuli. Moon and Kim [31] placed 20 controllable fans around a

user to emulate wind stimuli. Kulkarni et al. [32] developed a system to

generate a front-facing air movement to emulate outdoor wind condi-

tions. Verlinden et al. [33] employed an array of 8 fans to emulate wind

stimuli for a sailing simulation. Tolley et al. [34] built a controllable fan

array of 90 fans, also targeting outdoor wind simulations. The studies

encountered some issues with fan arrays to generate airflow stimuli. The

interaction of the fans and the thermal buoyancy of the user were

complicated to control. There was noticeable fan noise for fans installed

close to participants, hindering immersion. The air temperature of free-

standing fans corresponded to the ambient air temperature and the fans

created highly turbulent flow. For thermal comfort, air velocity, tem-

perature, and turbulence intensity are relevant, but only air velocities

can be controlled directly with fans. Indoor mechanical ventilation

systems (e.g. mixing ventilation or displacement ventilation) and natu-

ral or window ventilation have well-studied airflow structures that differ

from the local airflow structure generated by free-standing fans.

2.4. Research challenges and opportunities

Previously published implementations of virtual environments were

limited in controlling the environments thermal irradiation, including

surface temperatures and asymmetry. As noted by Zhao et al. [35], some

new energy-efficient concepts such as radiant heating and cooling create

non-uniform environments. The research and understanding of thermal

comfort in non-uniform environments and the combined effects of

various stimuli on human perception of environments are still in an early

stage. Consequently, there persists a need for further subject studies.

This is a motivating factor to create comprehensive virtual environments

(VE) that adequately include non-uniform thermal conditions.

The present concept has potential for the academic and industrial

sectors. In academic thermal comfort research, climate chambers are

typically employed for large-group subject studies to explore statistical

comfort relationships. Climate chambers are, so far, supplemented with

VR technology mainly for enhancing visual aspects. In the industrial

sector, there is interest in employing VR as a tool for demonstrating and

communicating designs for decision-making purposes. While this group

primarily focused on visual aspects in VR, they lack established methods

for incorporating thermal perception aspects. Our proposed concept

bridges these two groups, enabling the transfer of capabilities between

them. It allows the thermal comfort research group to explore novel

methods for representing thermal environments, while providing the VR

group with tools to integrate thermal perception into visualizations.

In the present use case example, a simulated occupant was seated

close to a cool external wall with a window while a local surface heating

element was placed on the back wall. This non-uniform environment

was emulated in the proposed new VR Lab by adopting the surface

temperatures of the climate chamber and controlling the supply dif-

fusers. The resulting quality of the emulation was benchmarked directly

by comparing the thermal irradiation and air temperatures that occu-

pants would experience in both environments.

Table 1

Literature survey on virtual thermal environments.

Study Year VE In AT RT AV TC Content

Latini et al.

[24,25]

2023 ✓ ✓ ✓ ✓ Office room:

Comfort and

productivity

Lyu et al.

[9]

2023 ✓ ✓ ✓ ✓ (Semi-)Outdoor

environment,

28 ◦C

Gao et al.

[22]

2023 ✓ ✓ ✓ ✓ Radiation

temperature

asymmetry due to

cool or warm walls

Vittori et al.

[28]

2022 ✓ ✓ ✓ ✓ (✓)✓Test room with a

VR component

Saeidi et al.

[26]

2021 ✓ ✓ ✓ ✓ Office room:18 ◦C,

24 ◦C, 29 ◦C

Beyer et al.

[21]

2019 ✓ ✓ ✓ ✓ Radiation

temperature

asymmetry due to

a heated or cooled

ceiling

Yeom et al.

[27]

2019 ✓ ✓ ✓ ✓ Office room:

Heating and

cooling from 20 ◦C

and 30 ◦C and

30 ◦C to 20 ◦C

Tolley et al.

[34]

2019 ✓ ✓ Fan array of 90

controllable fans

Ozcelik and

Becerik-

Gerber

[5]

2018 ✓ ✓ ✓ Office room, 18 ◦C

and 28 ◦C:

occupant

behaviour

Hülsmann

et al.

[30]

2014 ✓ ✓ ✓ Outdoor

temperatures and

wind emulation

with ceiling fans

and infrared lights

Verlinden

et al.

[33]

2013 ✓ ✓ Wind emulator for

a sailing

simulation with 8

fans

Kulkarni

et al.

[32]

2009 ✓ ✓ Custom wind

display device

Moon and

Kim [31]

2004 ✓ ✓ Wind sensation

emulation with 20

fans

VE: Virtual environment, In: Indoor application, AT: Air temperature control,

RT: Radiation temperature asymmetry control, AV: Air velocity/wind control,

TC: Thermal comfort.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

3. Method

3.1. VR Lab

The VR Lab (Fig. 3) is a single-user virtual environment laboratory,

designed to emulate and experiment with indoor environments. The

room is compact (3 m x 3 m x 3 m) to enable fast and precise control of

the thermal environment. It features 45 temperature-controlled surfaces

in a hemicube arrangement and 9 independent supply diffusers to

simulate typical ventilation scenarios. The temperature-controlled sur-

face elements include capillary tube mats for cooling and superimposed

heating fabric mats that enable a faster operative temperature control

(1.3 K/min) in comparison to radiators or floor heating (<0.1 K/min)

[36]. Four slot diffusers (0.45 m ×0.03 m) and a swirl diffuser (0.16 m)

at the ceiling allow to reproduce typical mixing ventilation scenarios.

Four displacement supply diffusers (0.925 m ×0.12 m) at the bottom

walls enable to reproduce displacement ventilation or down-draught

airflows. Floor and ceiling supply diffusers can be combined to sup-

port a vertical air temperature stratification if intended. For non-

uniform thermal conditions, the control of the surfaces and air supply

diffusers can be non-trivial, due to the interaction of the airflows and the

temperature-controlled surfaces. CFD simulations were therefore used

to initially determine and evaluate boundary conditions and to assess

the thermal conditions.

3.2. Benchmark case

Fig. 4 shows the benchmark case to be emulated in the VR Lab. The

environment is a reference climate chamber, described in the Appendix

together with the CFD simulations that were carried out for the bench-

mark reference. For an objective comparison, an occupant was repre-

sented by an elliptical person dummy placed at one of two possible

locations. The person dummy had a heat load of 85 W and the latent heat

transfer component was not simulated. The environmental thermal

irradiation was evaluated on the dummy surface. A vertical probe array

located 0.5 m in front of the dummy position with a height of 1.5 m was

chosen to compare air temperatures. The probe array was located near

the dummy but outside of its thermal plume. As a consistent simulation

approach was used to model the actual room and the VE, the thermal

emulation of the environment can be compared.

3.3. Radiant temperature

The mean radiant temperature TMRT of an environment is defined as

the temperature of an imaginary uniform blackbody enclosure with

which an interior surface would have the same radiation exchange as

with the actual environment [11]. Depending on the application, the

surface can be a small black sphere, a plane square, or the human body

surface which will lead to a single mean radiant temperature. It can also

be a surface patch (of a larger surface). This will result in a mean radiant

temperature distribution [37], which is the definition used in this work.

It results from the thermal irradiation I of an environment on the surface

patch:

(TMRT)4=I/

(1)

The radiation heat transfer of surfaces is related to the fourth power of

their temperatures by the Stefan-Boltzmann law (

is the Stefan-

Boltzmann constant). If we restrict ourselves to blackbody diffuse sur-

face to surface radiation, the mean radiant temperature definition that is

known for thermal comfort is obtained: [11]

(TMRT)4=∑

i=1

(Ti)4Fi(2)

The enclosing environment consists of n surface patches with a tem-

perature Ti. The view factors Fi are purely geometric and result from the

orientation, size and distance of the surface patches. The same mean

radiant temperature (and thus the same radiation heat transfer) is ob-

tained if surface patches are replaced with surrogate surfaces (having

different distance, orientation, and size) and the same view factors and

temperatures which is the basic concept for the geometric projection.

Furthermore, it is possible to obtain the same mean radiant temperature

with arbitrary other temperatures and view factors if the sum in Equa-

tion (2) remains the same. The simplest application of this idea is an

imaginary blackbody enclosure with a unity view factor and a constant

temperature. Achieving identical thermal irradiation for other enclo-

sures and non-uniform environments leads to a more generic optimisa-

tion problem. It is pertinent to note that while Equation (2) is prevalent

in the field of thermal comfort, Equation (1) is a more generic definition

of the mean radiant temperature that can be used to include direct ra-

diation sources.

3.4. Irradiation on persons

Simplified mannequins were used in the simulations to represent the

human heat exchange with the environment. Fig. 5 shows the simulated

mean radiant temperature on a thermal mannequin (geometry from

Sørensen and Voigt [38]) seated near the cool wall in the climate

chamber. Due to the cool wall, the mean radiant temperatures on the

right-hand side of the thermal mannequin were lower. There was also

self-irradiation, noticeable especially between the legs and arms. This

confounds the intended evaluation of the thermal environment’s irra-

diation as this is an effect of the mannequin geometry (and temperature)

and not of the thermal environment. It also varies for different arm and

Fig. 3. VR Lab. Left: concept. Right: Current state of construction.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

leg positions. Thus, the detailed thermal mannequin was unsuitable for

objectively comparing the thermal irradiation of two environments. A

compromise was devised because thermal environments should be

compared objectively in this work, but also the heat transfer of the

occupant should be simulated. To retain similar view factors and surface

heat transfer, the mannequin was replaced by an elliptical cylinder with

the same surface area and similar dimensions (outer dimensions 0.23 m

×0.46 m ×1.36 m, surface area 1.6m 2), see Fig. 6.

3.5. Projection of enclosing environments

Fig. 7 visualises the surface elements of the VR Lab. Directions (Left,

right, etc.) refer to the perspective of the occupant. The ceiling and side

walls were split into nine surfaces, respectively, according to the nine

temperature-controlled surfaces on each wall.

3.6. Geometric projection

As was pointed out by Nußelt (1928) [15], surface patches of an

environment can be projected geometrically to other surfaces with the

same view factor using the concept of solid angles. For the VR Lab, the

geometric projections originating from the centre of the seated occupant

are shown in Fig. 8 and Fig. 9. The centre of the seated occupant was

placed at a height of 0.6 m. This is a typical choice for thermal comfort

measurements with globe temperature sensors [39]. The projected sur-

faces corresponding to these solid angles in the reference environment

are visualised in Fig. 8 for the central element of each side of the VR Lab.

They were generated by constructing pyramids and determining their

intersection with the reference room surfaces. Temperatures for each of

the 54 surface patches were then evaluated as average surface temper-

atures on the corresponding surface patches in the reference room

(Fig. 9). The generalized mean with power four of the temperature was

used for averaging. The averaging neglected the variation of the view

factors on each surface patch. For position 1, the occupant was seated

Fig. 4. Benchmark case. The occupant position was either near the windows (position 1) or on the other side of the room (position 2).

Fig. 5. Mean radiant temperature for a thermal mannequin in the asymmetric

environment.

Fig. 6. To avoid self-irradiation, the thermal mannequin was replaced by an

elliptical dummy.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

close to the back wall. The surface temperatures in the VR Lab back wall

(green) therefore need to reproduce the higher temperatures on the

heated surface behind the occupant.

3.7. Optimisation approach

As an alternative more sophisticated approach, surface temperatures

were determined by optimisation. In the optimisation approach, the

mean radiant temperature distribution on the person dummy in the

reference configuration at position 1 (Fig. 14) was specified as a target

that should be emulated in the VR Lab. A parametric CFD simulation of

the VR Lab was set up. A surface-to-surface radiation calculation was

used, and the simulations took into account the local view factors of all

surface patches of the person dummy. For comparison, in the geometric

approach, only a reference point at a height of 0.6 m was used.

Furthermore, while the geometric projection approach assumes diffuse

radiation, the optimization approach only aims at reproducing the mean

radiation temperature distribution on the person dummy surface

(Equation (1). This could include direct radiation, although in the pre-

sent case, only diffuse surface radiation was present. A batch of CFD

simulations was carried out where the surface temperature set points on

the VR Lab’s surface elements were varied as free design parameters.

The goal for the optimiser was to minimize the average quadratic de-

viation of the mean radiant surface temperatures on the person dummy

from the reference surface temperatures. Therefore, the target mean

radiant surface temperatures should reproduce the reference and

thereby the irradiance of the reference environment. A separate CFD

simulation was carried out for each design iteration, i.e., each set of

boundary temperatures. The optimisation had 54 free parameters,

which were the specified temperatures on the surface elements. A total

of 10,000 design iterations were carried out, and the best iteration was

selected as the solution. As the VR Lab lacks a floor cooling, the opti-

misation approach was additionally carried out with a fixed floor tem-

perature as a constraint.

3.8. Supply air diffuser control

In the second step, the supply air temperatures and airflow rates

were specified to emulate the reference air temperatures and air veloc-

ities. For this purpose, numerical experiments were carried out with CFD

simulations. One possible supply diffuser configuration was selected for

position 1 and two configurations for position 2. Whereas for deter-

mining the surface temperatures, a systematic approach was used, the

supply airflows and temperatures were determined manually. An auto-

matic approach would, in principle, also be possible for determining the

supply airflow parameters. However, this would be computationally

costly because of the higher computational cost of fluid flow

simulations.

In the present reference room at positions 1 and 2, the air velocities

at a distance of 0.5 m in front of the occupant were below 0.06 m/s and

were considered insignificant for human thermal sensation. As a refer-

ence, the draught rate model in the standard ISO 7730 [40] has a lower

Fig. 7. Visualisation of the VR Lab surface elements.

Fig. 8. Geometric projection. The emulated reference room is behind the

VR Lab.

Fig. 9. Projected hemicube surface elements for position 1.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

air velocity limit of 0.05 m/s and the ASHRAE standard 55–2020 [39]

defines air velocities below 0.1 m/s as still air. The velocity field in the

vicinity of the person dummy was mainly determined by the thermal

plume generated by the heat of the person dummy and not by the

ventilation diffusers. To ensure low air velocities in the VR Lab, multiple

supply diffusers were used simultaneously for position 1.

4. Results

4.1. Surface temperatures and radiant temperatures

Fig. 10 shows the surface temperatures obtained from the simula-

tions of the reference case for Position 1 and Position 2. The surface

patches that were used to extract temperatures for the VR Lab in the

geometric projection approach are outlined by black boundaries. Fig. 11

shows the resulting mean radiant temperature on the person dummy

surface at a height of 0.6 m for the two positions. For position 1, the

radiant environment was asymmetric with a difference of about 2 K

between the left and right side of the person dummy and about 6 K

between the front and back. The temperature differences were well

within the allowed range by the ASHRAE standard 55–2020 of 10 K for

walls cooler than air and 23 K for walls warmer than air. The steep

gradients on the left and right sides result from the elliptic shape of the

dummy. For position 2, the mean radiant temperatures at 0.6 m varied

less than 0.5 K. The small difference between the front and back sides

was caused by the person dummy itself which was placed 0.5 m close to

the back wall. For position 1, the influence of the cold wall extended to

most of the dummy front surface, as seen when comparing the front

surface temperatures of position 1 with position 2.

Fig. 12 shows the projected surface temperatures in the VR Lab

determined for position 1 with the geometric approach on the left and

with the optimisation on the right. The highest temperatures occurred at

the warm back wall. At the ceiling, an influence of the ceiling lights was

visible and on the right and front side an influence of the windows. The

average projected surface temperature was 22.2 ◦C for the geometric

approach and 22 ◦C for the optimisation approach. The average tem-

peratures on each of the 6 surfaces calculated by both approaches

differed between −1.5 K (cold wall) and +1 K (floor). The optimisation

approach resulted in a more distinguished temperature distribution.

Some surface element temperatures exceeded the maximum tempera-

ture at the back wall of the reference room (up to 35 ◦C). Although this is

counter-intuitive and not possible with a geometric projection, it can

improve the emulation of the radiation heat exchange due to the limited

resolution of the hemicube surfaces.

The floor surface temperatures determined with the geometric

approach were on average 21.3 ◦C but varied up to ±1 K. For the

optimisation approach, even larger temperature variations resulted on

the floor. The results of the optimisation approach with a fixed floor

temperature of 21.3 ◦C are shown in Fig. 12 in the lower diagram. With a

fixed floor temperature, the other surfaces have to compensate for the

irradiation of the floor, leading to a different optimal solution.

Fig. 13 shows the results of the geometric and optimisation approach

for position 2. The distribution of the geometric approach was relatively

homogeneous, except for the influence of the ceiling lights. The homo-

geneous environment at position 2 does not pose a challenge in the

present context but can be used as a baseline in subject tests. The slightly

higher temperatures on the back wall were caused by the person dummy

which is also a heat source. The average boundary surface temperature

for position 2 was 21.5 ◦C and thus 0.6 K lower than for position 1. The

optimisation approach for position 2 resulted in a surprisingly complex

solution. Such solutions are possible due to the rather loose constraints

of Equation (2).

4.2. Irradiation

In Fig. 14, the resulting person dummy surface distributions of mean

radiant temperatures of the different projection approaches are

compared for position 1. While the asymmetry between the front and

back of the person dummy was reproduced with all approaches, the

optimisation approach resulted in a more pronounced asymmetric

Fig. 10. Surface temperatures of the benchmark room with the person dummy seated at position 1 or 2.

Fig. 11. Mean radiant temperature (BMRT) on the person dummy at a height of

0.6 m.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

distribution and a better agreement with the reference distribution on

the warm backside.

For the optimisation approach without and with fixed floor tem-

peratures, the root mean square error of the temperature on the dummy

surface from the reference surface temperatures was 0.4 K and 0.48 K,

compared to 0.7 K for the geometric approach. In comparison, a uniform

enclosure assumption using the surface average radiant temperature

would have caused a surface root means square error of 2.2 K and a local

error of up to 4.6 K. A further improvement was constrained by the

resolution of the hemicube surface elements (i.e., the number of

temperature-controllable surface elements) and the preset temperature

resolution of 0.1 K for the surfaces.

4.3. Air temperatures

As the surface temperatures were fixed by the emulated mean radiant

temperatures, the air temperatures were controlled with the supply

diffuser airflow rates and temperatures. For both positions, the surface

temperatures determined with the geometric approach were used. For

position 1, a suitable configuration of supply diffusers was selected and

Fig. 12. Projected surface temperatures in

◦C for position 1.

Fig. 13. Projected surface temperatures in

◦C for position 2.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

for position 2, two configurations (2a,2b) were selected using either

ceiling slot diffusers or the ceiling swirl diffuser. The same two ceiling

extract air openings were used in all cases. The parameters are sum-

marized in Table 2.

For position 1, due to the warm back wall (30 ◦C), additional cooling

had to be provided by a supply airflow temperature of 16 ◦C. Fig. 15

shows the temperatures on the back wall, the floor and in the median

plane. As the back wall was further away from the person dummy in the

VR Lab (1.5 m) than in the reference configuration (0.5 m), the projected

heated surface and thus its heat load was larger. Most of the supply

airflow at supply diffuser 1 raised at the warm wall and was needed for

cooling. The warm wall introduced a heat load of 720 W. Together with

the 85 W heat load of the person dummy, this resulted in a total specific

heat load of 90 W/m2. This was above the typical maximum possible

cooling load for wall-mounted supply diffusers of about 60 W/m2 [41],

where room airflows are stable. The room airflow was unsteady and the

simulation for position 1 had to be carried out as an unsteady simula-

tion. The issue could have been avoided by seating the person closer to

one of the walls in the laboratory, thereby reducing the projected size of

the heated surface. This was not done here because of the current focus

Fig. 14. Comparison of mean radiant temperatures for position 1 obtained with different approaches.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

on a general assessment of the VR Lab’s capability to emulate similar

challenging scenarios.

For position 2, the supply diffuser configuration was simpler as the

temperature distribution was more uniform.

When using the ceiling swirl diffuser, the supply airflow was found to

attach to the walls of the laboratory. As a result, the supply airflow was

heated by the walls before reaching the occupied zone and had to be

introduced at a temperature of 14 ◦C.

The air velocities 0.5 m in front of the person dummy were below

0.07 m/s and thus negligible, except for the swirl diffuser which caused

secondary air induction and higher local velocities of up to 0.15 m/s.

The selected airflow rate of 150 m

/h was near the design maximum

airflow rate of the installed swirl diffuser size (160 mm).

Fig. 16 shows the vertical temperature profiles in the VR Lab in

comparison to the benchmark case. The vertical line array used for the

temperature evaluation was placed 0.5 m in front of the occupant and

contained 50 evenly spaced points between the floor and a height of 1.5.

The line array position is indicated as grey line in Fig. 15. Reference

temperature profiles are plotted for comparison as dashed grey lines,

together with boundary markers for deviations ±0.2 K. Vertical tem-

perature profiles were well reproduced in the VR Lab and deviated less

than ±0.2 K from the references.

When using only the ceiling swirl diffuser, a uniform vertical air

temperature distribution was obtained. During the manual adjustment

for position 1, the general influence of the bottom and top diffusers was

noted. Increasing the flow rate for the top slot diffusers tended to lead to

a more homogeneous temperature distribution, as expected for mixing

ventilation and similar to what was observed with the swirl diffuser.

Increasing the flow rate at the bottom low velocity supply diffusers lead

to a more pronounced vertical temperature stratification in the lower

region.

A desired vertical temperature profile in the VR Lab can therefore be

obtained by selecting and adjusting the supply diffusers. The supply air

temperatures have to be controlled to obtain the desired room air tem-

peratures in dependence on the heat load caused by the temperature-

controlled surfaces and the occupant.

5. Discussion and limitations

The proposed concept allows to separate the control of the vertical

air temperature profile and radiation asymmetry. Thus, it can be used to

extend the scientific knowledge about thermal discomfort due to radi-

ation asymmetry and vertical air temperature gradients. In conventional

climate chambers, physical mock-ups are used to emulate indoor ther-

mal environments, even in the studies where the visual environment was

provided by a VR headset, limiting the possibilities for research. Mock-

ups can become unfeasible for large rooms, complex arrangements (e.g.

tiers in a theatre), or if many different scenarios are to be investigated. In

Table 2

Supply diffuser configurations.

Position

Position 2a (slot

diffusers)

Position 2b (swirl

diffuser)

Floor 1 45 m

/h − −

Floor 2 22 m

/h 40 m

/h −

Floor 4 22 m

/h − −

Ceiling 6 100 m

/h 100 m

/h −

Ceiling 7 100 m

/h − −

Swirl diffuser − − 150 m

Supply air

temperature

16 ◦C 17.5 ◦C 14 ◦C

Fig. 15. Simulated air temperature field in the VR Lab.

Fig. 16. Vertical air temperature profile in the VR Lab (solid lines) and the reference ±0.2 K (dashed lines) at a measurement rake positioned 0.5 m in front of the

person dummy.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

such cases, a full non-uniform thermal environment cannot be fully

emulated in conventional climate chambers because of constructional

limitations. The present concept overcomes this issue, by emulating

arbitrary thermal irradiation and air temperatures and velocities. The

benefit of the improved non-uniform thermal emulation can be assessed

objectively by evaluating the non-uniform thermal irradiation and the

non-uniform air temperature and velocity field of the actual environ-

ment under investigation. As indoor thermo-fluid-dynamic processes are

complex, fluid dynamic simulations are a useful tool for this purpose.

The control of the thermal irradiation required an approach to

specify the surface temperatures in the VR Lab. The geometric projection

approach could be carried out in real time. The alternative optimisation

approach is more powerful and can partly overcome the limited reso-

lution of the surface elements but also requires more computation. For

the CFD simulations, it was assumed that the surface temperature con-

trol achieves sufficiently uniform temperatures on the surface elements

which will be checked with thermal imaging measurements. The simu-

lations showed possible airflow fluctuations in the VR Lab for high in-

ternal heat loads which should be checked in experiments. The

simulations have also indicated strong interactions of the supply air-

flows with the temperature-controlled surfaces, especially when using

the swirl diffuser. This may lead to complications for the surface tem-

perature control which needs to be investigated in experiments. The

present configuration with 9 supply diffusers and 4 extract openings can

be used to emulate conventional mixing and displacement ventilation

scenarios. For other ventilation systems, such as personal ventilation,

additional air supplies can be added. The influence of the head mounted

display on the human thermal sensation was neglected here and needs to

be investigated in subject tests. The acoustic environment will be

implemented with headphones in the next step. The coordination of a

full interactive visual, thermal, and acoustic environment requires a

software solution that is still under development.

This study introduces technical aspects for implementing the thermal

environment in a novel virtual environment, setting the stage for future

investigations. Building on the previous research mentioned in Section

2.2, there remains a need for additional subject studies to fully realize

the potential of VR in this domain:

•Ecological validity and immersion: Is the thermal sensation experi-

enced by subjects in a non-uniform virtual thermal environment

consistent with that in a real non-uniform thermal environment?

•To what extent does the use of a VR headset influence participants’

thermal perception and comfort in virtual environments. How can

researchers account for or mitigate potential effects in their experi-

mental design?

•To what extent can VR technology be leveraged to refine and expand

thermal comfort models for complex built environments? How might

it contribute to the development of more context-specific thermal

comfort equations?

6. Outlook

In the present work, the influence of a cold wall in a surface heating

scenario was examined as a benchmark case for a non-uniform indoor

environment. More non-uniform scenarios can be considered in the

future such as a cooling scenario with hot internal surfaces, solar radi-

ation through glazed surfaces or sensible air movements due to high

supply diffuser velocities.

The next step will be validation subject tests. The CFD simulations

were useful in objectively examining the thermal environment in the VR

Lab and screening possible issues before carrying out subject tests. Based

on the present results some modifications of the present setup for a

subject study can be considered, including:

- A smaller heated surface in the reference room with the same heat

load to obtain higher surface temperatures and a more perceptible

temperature asymmetry (presently smaller than 6 K).

- A larger vertical air temperature stratification (presently smaller

than 1 K)

- A variable seating position of the occupant in the VR Lab to increase

flexibility.

The geometric projection method was fast but can still be improved

by utilizing the full space of possible solutions. The alternative optimi-

sation approach required a high computational effort. Training a neural

network for this purpose could be beneficial and enable real-time

optimal projection. The inputs of the neural network would be the

desired mean radiant temperatures at a fixed number of locations on the

person dummy surface, and the outputs would be the projected surface

temperatures in the VR Lab to obtain this distribution. The air temper-

atures and velocities in the VR Lab can be monitored directly for control

with a vertical sensor array.

7. Conclusion

This study proposed a new concept to fully emulate non-uniform

thermal environments without needing to construct physical mock-

ups. Computational fluid dynamics (CFD) simulations were used to

analyse and benchmark the thermal emulation. The results are sum-

marised as follows:

•For the selected non-uniform benchmark scenario, the assumption of

a uniform enclosure (using the surface average radiant temperature)

resulted in a surface root mean square error of 2.2 K (local deviation

of up to 4.6 K) of radiant temperature on the person dummy surface.

•To determine the surface temperature boundary conditions for the

emulated environment, a geometric method to project surface tem-

peratures can be used.

•For complex non-uniform environments, an alternative optimisation

approach can be used. It takes into account the full space of possible

solutions as well as experimental constraints such as the number and

resolution of available temperature-controlled surfaces.

•With the proposed concept of an enclosure with 45 temperature-

controlled surfaces in a hemicube arrangement, the surface root

mean square error of the radiant temperature of the emulated envi-

ronment was reduced to 0.4 K (local deviation of up to 0.8 K) for the

optimisation approach and 0.7 K (local deviation of up to 1.5 K) for

the geometric approach for the benchmark scenario.

Funding

This work is funded by the German Federal Ministry for Economic

Affairs and Climate Action in the framework of the research program

EnOB:GEnEff/BMWi 03EN1017A.

CRediT authorship contribution statement

Karsten Tawackolian: Writing – review & editing, Writing – orig-

inal draft, Visualization, Methodology, Investigation, Formal analysis,

Data curation. Lukas Schmitt: Writing – review & editing, Validation,

Supervision, Project administration, Formal analysis. Martin Kriegel:

Writing – review & editing, Supervision, Resources, Project adminis-

tration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence

the work reported in this paper.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

Data availability

Data will be made available on request.

Appendix

A.1 Benchmark climate chamber

To assess the capabilities of the VR Lab in emulating non-uniform indoor environments, a benchmark case was devised. A test case scenario was

selected for use in a later thermal comfort subject study. To highlight the unique capabilities of the VR Lab in reproducing non-uniform environments,

a benchmark scenario was included that features a defined surface radiation temperature asymmetry.

To enable reproducible test conditions, a second climate chamber was selected as the reference environment to be emulated. The selected Optemp

(operative temperature) climate chamber [42] was designed to assess the performance and thermal comfort of non-uniform room heating systems

such as radiators and fast electrical surface heaters. A 3D model of the room was constructed in Unity for use with a head-mounted display (Figure 17)

in later subject studies in the VR Lab. The climate chamber includes a mixing ventilation system with a ceiling swirl diffuser, electrical surface heating

and the possibility to simulate a cold outdoor wall with two windows. The external side of the simulated outdoor wall is enclosed in a separately

climatised chamber with a controlled external temperature. Electrically heated textile fabric elements on the walls can be used for heating and

generating non-uniform surface temperatures. They are selected here as an exemplary heating system that depends on thermal radiation. The full

climate chamber is located in a larger climatised hall.

Although the Optemp climate chamber was constructed with flexibility in mind, most parameters such as the window arrangement or room layout

and height cannot be modified in such a conventional climate chamber without cumbersome constructional modifications. This shows the potential

for the new concept of a VR Lab for arbitrary environments.

Fig. 17. Optemp climate chamber used for comparison tests. Left: physical chamber, right: VR model.

A.2 CFD simulations

Computational fluid dynamic (CFD) simulations were carried out for both the VR Lab and the benchmark case. The computational setup and

boundary conditions are summarised in Table A1. Steady-state Reynolds-averaged Navier-Stokes (RANS) simulations were carried out in Siemens

Simcenter STAR-CCM+17.02. Simulations were run for 10

iterations. The iterative convergence was checked with monitors for the temperature 0.5

m in front of the simulated person. In one case of a VR Lab simulation, unsteady simulations were carried out. They were found to be necessary because

sustained temperature and velocity fluctuations occurred at the monitor points during the simulation run. The unsteady simulations had a timestep of

0.05 s and were run for a physical time of 600 s. For time iteration, a second order implicit unsteady solver with 5 inner iterations was used. The

domain root mean square residuals for all solved quantities were smaller than 3 ×10

−6

at the end of each time step. The resulting mean convective

Courant number in the computational domain was 0.33, well below the recommended values of 1 for time discretisation error limiting and 50 for the

stability limit of the implicit solver. Time-averaging was carried out for 500 s (10.000 time-steps).

Table A1

Computational setup for the simulations.

Quantity Symbol Value

Air density

1.2 kg/m

Air dynamic viscosity

18 µ Pa s

Air thermal conductivity λ0.026 W/(m K)

Software Siemens Simcenter STAR-CCM+17.02 [43]

Fluid model Steady state, Incompressible, Reynolds-average Navier Stokes (RANS), Unsteady RANS (position1)

Buoyancy model Boussinesq Model

Turbulence model RKE Realizable k-

[44]

Radiation model Surface-to-surface grey radiation, Surface emissivity of 0.95 and reflectivity of 0.05 for all surfaces

Wall boundary treatment Smooth wall, Two-Layer, Buoyancy Driven [45]

(continued on next page)

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

Table A1 (continued)

Quantity Symbol Value

Wall boundary conditions Benchmark room: U-values or heat source

VR Lab: fixed temperatures (Fig. 12 and Fig. 13)

Person dummy: heat source, 85 W

Near wall cell distance y+<2

Mesh Trimmer mesher, 5 prism layers at walls

Optemp: max. cell size 7 cm, 4.3 ×10

cells

VR Lab: max. cell size 3 cm, 3.2 ×10

cells

Validation case: max. cell size 3 cm, 10 ×10

cells

A2.1 Validation of the simulation approach

Gilani et al. [46] validated CFD simulations for a scenario comparable to the present application (a room ventilated by displacement ventilation

and a single internal heat source) with measurements from Li et al. [47]. In the present work, validation simulations were repeated for the same test

case and boundary conditions and with the present model and meshing choice (Figure 18, simulation BC1). The boundary conditions for the floor and

ceiling temperatures were not measured but estimated by Gilani et al. at 24 ◦C and 24.2 ◦C. A second simulation with modified floor (23 ◦C) and ceiling

(25 ◦C) temperature was also carried out, leading to an improved agreement with the measurements (simulation BC2). In the scenarios of this study,

the vertical temperature stratification was less than 1 K, (compared to 4 K in the case from Li et al.), making it a less challenging condition for

modelling. An important aspect for the current application is the radiation model. Schmitt et al. [42] verified the suitability of a surface-to-surface

radiation model for the radiation heat transfer of the electrical surface heating mats in the OpTemp chamber (Figure 19). Given the even smaller

geometry of the VR Lab, the impact of radiation absorption by the room air can be disregarded, affirming the suitability of the surface-to-surface

radiation model for both scenarios.

Fig. 18. Validation for thermal stratification. Experiments from Li et al. [44]. BC1: boundary condition from Gilani et al. [42]. BC2: adapted floor and ceiling

temperature.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

Fig. 19. Validation for the heating fabric radiation heat transfer model [38].

A2.2 Benchmark room simulations

UCFD = (U−1−U−1

i)−1(3)

The room dimensions are 3.94 m ×5.8 m ×3.03 m. Two simulations were carried out where the dummy representing the occupant was either placed

near the cool outdoor wall (position 1) or on the opposite side of the room (position 2). To simplify the validation subject study in the VR Lab, furniture

was not included. The dummy’s lateral distance to the outdoor (or opposite) wall was 1 m and the dorsal distance to the back wall was 0.5 m. The

occupant was simplified as an elliptical cylinder dummy as explained in Section 3.3. At the outdoor wall, an outdoor temperature of −20 ◦C and U-

values for the walls (0.3 W/m

K) and windows (1 W/m

K) were specified, based on the structural composition of the walls. The cold outdoor

temperature was selected to obtain a heating load and a pronounced internal radiation temperature asymmetry. For all other walls, U-values were

specified (0.3 W/m

K) and an external temperature of 18 ◦C, based on the planned climatisation in the measurement hall that encloses the climate

chamber. In the CFD simulations, all internal thermal boundary layers were resolved. Therefore, adjusted (larger) U-values were used as boundary

conditions (Eq. (3). Internal U-values of U

=1/0.13 W/m

K were used for this correction, based on average indoor values.

The external boundary conditions are a source of uncertainty. Once the actual subject tests are carried out, existing wall temperature sensors and

thermal imaging will be used to measure actual room surface temperatures and obtain precise boundary conditions. The simulated heat loss through

the cold wall was about 215 W, through other unheated surfaces about 85 W and the ventilation heat loss was about 32 W. The operative temperature

in the centre of the room as determined by a 15 cm black globe temperature sensor was 21 ◦C. The black globe temperature sensor was modelled in the

simulation as a 15 cm adiabatic sphere placed in the room centre.

At the ceiling, there were two lights with a heat load of 72 W, each. Behind the person dummy position at the cool wall, a heated mat (1.5 m ×1 m,

average surface temperature of 29 ◦C) was placed. Out of the 100 W of the heated surface, 54 W were transmitted by radiation. In combination with the

cool outdoor wall, it was intended to obtain a distinct asymmetric radiation temperature. Supply air was introduced through a ceiling-mounted swirl

diffuser at a temperature of 18 ◦C and an airflow rate of 35 m

/h. The geometry of the swirl diffuser blades was resolved in the mesh to simulate the

swirling flow. The present type of swirl diffuser produced a downward flow that was not attached to the ceiling, which was also observed in ex-

periments. The simulated occupants were seated far enough from the swirl diffuser to avoid direct exposure to the supply diffuser airflow.

A2.3 VR Lab simulations

The supply and extract slot diffusers of the VR Lab were modelled as extruded channel geometry, thereby neglecting the grills. This level of

simplification was acceptable for a thermal comfort analysis as the ceiling diffusers mainly influenced the ceiling area, inducing a typical mixing

ventilation flow, and the bottom diffusers had low flow velocities.

When active, the swirl diffuser was meshed including the blade geometry to simulate the swirling supply airflow. The person dummy was placed

centrally, and had a heat load of 85 W, corresponding to the simulations of the reference room. Suitable supply airflow rates and temperatures were

selected by manual experimentation on preliminary coarse mesh simulations that were subsequently carried out on finer meshes. At the walls, fixed

temperatures were prescribed to emulate the corresponding reference environment.

A2.4 Mesh Study for the benchmark simulations

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

Fig. A19. Mesh study for the Optemp climate chamber simulation. Temperature profile 0.5 m in front of the person dummy, position 1.

Figure 19 shows the mesh study for the configuration with the dummy placed at position 1. Simulations were carried out with three meshes with 1

×10

, 4.3 ×10

and 10 ×10

cells. The maximum difference of the predicted temperatures at the reference position 0.5 m in front of the dummy

position 1 was less than 0.06 K between these stimulations. The mesh resolution with 4.3 ×10

cells was therefore selected as sufficiently resolved

mesh.

A2.4 Mesh Study for the VR Lab simulations

The mesh study was carried out for position 1 as it was most challenging because of the non-uniform environment. The mesh study was carried out

with three meshes with 1.6 ×10

, 3 ×10

and 10 ×10

cells. The maximum deviation of the predicted temperatures at the reference position 0.5 m in

front of the dummy position 1 was less than 0.13 K between these stimulations (Figure 20). Because of unsteady airflow conditions for this scenario

and the finite time averaging (500 s), an uncertainty contribution of the time-averaged temperatures of about 0.1 K was estimated. The mesh with 3 ×

cells was therefore selected as mesh resolution for the simulations.

Fig. 20. Mesh study for the VR Lab simulation for position 1. Temperature profile 0.5 m in front of the person dummy.

References

[1] Y. Song, F. Mao, Q. Liu, Human comfort in indoor environment: a review on

assessment criteria, data collection and data analysis methods, IEEE Access 7

(2019) 119774–119786, https://doi.org/10.1109/ACCESS.2019.2937320.

[2] Y. Zhu, S. Saeidi, T. Rizzuto, A. Roetzel, R. Kooima, Potential and challenges of

immersive virtual environments for occupant energy behavior modeling and

validation: a literature review, J. Build. Eng. 19 (2018) 302–319, https://doi.org/

10.1016/j.jobe.2018.05.017.

[3] T. Parkinson, S. Schiavon, J. Kim, G. Betti, Common sources of occupant

dissatisfaction with workspace environments in 600 office buildings, Build. Cities 4

(1) (2023) 17–35, https://doi.org/10.5334/bc.274.

[4] C. Huizenga, H. Zhang, P. Mattelaer, T. Yu, E.A. Arens, P. Lyons. Window

performance for human thermal comfort, 2006 https://escholarship.org/uc/item

/6rp85170.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748

[5] G. Ozcelik, B. Becerik-Gerber, Benchmarking thermoception in virtual

environments to physical environments for understanding human-building

interactions, Adv. Eng. Inf. 36 (2018) 254–263, https://doi.org/10.1016/j.

aei.2018.04.008.

[6] Y. Zhang, H. Liu, S.C. Kang, M. Al-Hussein, Virtual reality applications for the built

environment: research trends and opportunities, Autom. Constr. 118 (2020)

103311, https://doi.org/10.1016/j.autcon.2020.103311.

[7] A. Gallace, M.K. Ngo, J. Sulaitis, C. Spence, Multisensory presence in virtual

reality, multiple sensorial media advances and applications, IGI Global (2012)

1–38, https://doi.org/10.4018/978-1-60960-821-7.ch001.

[8] W.R. Sherman, A.B. Craig, Understanding Virtual Reality: Interface, Application,

and Design, Morgan Kaufmann, 2018.

[9] K. Lyu, A. Brambilla, A. Globa, R. de Dear, An immersive multisensory virtual

reality approach to the study of human-built environment interactions, Autom.

Constr. 150 (2023), https://doi.org/10.1016/j.autcon.2023.104836.

[10] P.O. Fanger. Thermal comfort. Analysis and applications in environmental

engineering, 1970.

[11] K. Parsons, Human Thermal Environments: The Effects of Hot, Moderate, and Cold

Environments on Human Health, Comfort, and Performance, CRC Press, Taylor &

Francis Group, 2014.

[12] X. Su, Y. Yuan, Z. Wang, W. Liu, L. Lan, Z. Lian, Human thermal comfort in non-

uniform thermal environments: a review, Energy Built Environ. (2023), https://

doi.org/10.1016/j.enbenv.2023.06.012.

[13] C. Zhang, M. Pomianowski, P.K. Heiselberg, T. Yu, A review of integrated radiant

heating/cooling with ventilation systems-Thermal comfort and indoor air quality,

Energy Build. 223 (2020) 110094, https://doi.org/10.1016/j.

enbuild.2020.110094.

[14] M.H. Vorre, R.L. Jensen, J. Le Dr´

eau, Radiation exchange between persons and

surfaces for building energy simulations, Energy Build. 101 (2015) 110–121,

https://doi.org/10.1016/j.enbuild.2015.05.005.

[15] W. Nußelt, Graphische Bestimmung des Winkelverh¨

altnisses bei der

W¨

armestrahlung, Zeitschrift Des Vereines Deutscher Ingenieure 72 (20) (1928)

673.

[16] P. Vueghs, P. Beckers, Presentation of the hemisphere method, WIT Trans. Eng. Sci.

53 (2006) 121–130, https://doi.org/10.2495/HT060121.

[17] M.F. Cohen, D.P. Greenberg, The hemi-cube: a radiosity solution for complex

environments, ACM Siggraph Comput. Graph. 19 (3) (1985) 31–40, https://doi.

org/10.1145/325165.325171.

[18] S. Van Craenendonck, L. Lauriks, C. Vuye, J. Kampen, A review of human thermal

comfort experiments in controlled and semi-controlled environments, Renew.

Sustain. Energy Rev. 82 (2018) 3365–3378, https://doi.org/10.1016/j.

rser.2017.10.053.

[19] A.L. Pisello, I. Pigliautile, M. Andargie, C. Berger, P.M. Bluyssen, S. Carlucci,

G. Chinazzo, Z. Deme Belafi, B. Dong, M. Favero, A. Ghahramani, G. Havenith,

A. Heydarian, D. Kastner, M. Kong, D. Licina, Y. Liu, A. Luna-Navarro, A. Mahdavi,

A. Nocente, M. Schweiker, M. Touchie, M. Vellei, F. Vittori, A. Wagner, A. Wang,

S. Wei, Test rooms to study human comfort in buildings: a review of controlled

experiments and facilities, Renew. Sustain. Energy Rev. 149 (2021) 111359,

https://doi.org/10.1016/j.rser.2021.111359.

[20] M. Schweiker, E. Ampatzi, M.S. Andargie, R.K. Andersen, E. Azar, V.M. Barthelmes,

C. Berger, L. Bourikas, S. Carlucci, G. Chinazzo, L.P. Edappilly, M. Favero,

S. Gauthier, A. Jamrozik, M. Kane, A. Mahdavi, C. Piselli, A.L. Pisello, A. Roetzel,

A. Rysanek, K. Sharma, S. Zhang, Review of multi-domain approaches to indoor

environmental perception and behaviour, Build. Environ. 176 (2020) 106804,

https://doi.org/10.1016/j.buildenv.2020.106804.

[21] M. Beyer, L. Schinke, G. Alessio, J. Seifert, M. De Carli, Investigations of (local)

thermal comfort as a function of radiation asymmetry and vertical air temperature

difference, E3S Web of Conferences 111 (2019) 02053, https://doi.org/10.1051/

e3sconf/201911102053.

[22] S. Gao, L. Yang, M. Shi, S. Zhao, X. Zhou, Y. Zhai, Human thermal comfort under

lateral radiant asymmetries, Energy Built Environ. 4 (4) (2023) 432–444, https://

doi.org/10.1016/j.enbenv.2022.03.001.

[23] H. Alamirah, M. Schweiker, E. Azar, Immersive virtual environments for occupant

comfort and adaptive behavior research - A comprehensive review of tools and

applications, Build. Environ. 207 (2022), https://doi.org/10.1016/j.

buildenv.2021.108396.

[24] A. Latini, E. Di Giuseppe, M. D’Orazio, Immersive virtual vs real office

environments: a validation study for productivity, comfort and behavioural

research, Build. Environ. 230 (2023) 109996, https://doi.org/10.1016/j.

buildenv.2023.109996.

[25] A. Latini, E. Di Giuseppe, M. D’Orazi, Development and application of an

experimental framework for the use of virtual reality to assess building users’

productivity, comfort, and adaptive behaviour, J. Build. Eng. 70 (2023) 106280,

https://doi.org/10.1016/j.jobe.2023.106280.

[26] S. Saeidi, G. Rentala, T. Rizzuto, T. Hong, N. Johannsen, Y. Zhu, Exploring thermal

state in mixed immersive virtual environments, J. Build. Eng. 44 (2021) 102918,

https://doi.org/10.1016/j.jobe.2021.102918.

[27] D. Yeom, J.H. Choi, Y. Zhu, Investigation of physiological differences between

immersive virtual environment and indoor environment in a building, Indoor Built

Environ. 28 (1) (2019) 46–62, https://doi.org/10.1016/j.buildenv.2019.03.013.

[28] F. Vittori, C. Chiatti, I. Pigliautile, A.L. Pisello, The NEXT. ROOM: Design principles

and systems trials of a novel test room aimed at deepening our knowledge on

human comfort, Build. Environ. 211 (2022) 108744, https://doi.org/10.1016/j.

buildenv.2021.108744.

[29] A.C. Da Silveira, E.C. Rodrigues, E.B. Saleme, A. Covaci, G. Ghinea, C.A. Santos,

Thermal and wind devices for multisensory Human-computer interaction: an

overview, Multim. Tools Appl. (2023) 1–28, https://doi.org/10.1007/s11042-023-

14672-y.

[30] F. Hülsmann, N. Mattar, J. Fr¨

ohlich, I. Wachsmuth, Simulating wind and warmth in

virtual reality: conception, realization and evaluation for a cave environment,

J. Virtual Reality Broadcasting 11 (10) (2014), https://doi.org/10.20385/1860-

2037/11.2014.10.

[31] T. Moon, G. Kim, Design and Evaluation of a Wind Display for Virtual Reality,

VRST ’04: Proceedings of the ACM symposium on Virtual reality software and

technology (2004) 122–128, doi: https://doi.org/10.1145/1077534.1077558.

[32] S. Kulkarni, C. Fisher, E. Pardyjak, M. Minor, J. Hollerbach, Wind display device

for locomotion interface in a virtual environment, in World Haptics 2009-Third

Joint EuroHaptics conference and Symposium on Haptic Interfaces for Virtual

Environment and Teleoperator Systems (2009) 184-189, doi: https://doi.org/

10.1109/WHC.2009.4810855.

[33] J.C. Verlinden, F.A. Mulder, J.S. Vergeest, A. de Jonge, D. Krutiy, Z. Nagy, B.J.

Logeman, P. Schouten, Enhancement of presence in a virtual sailing environment

through localized wind simulation, in 6th Asia-Pacific Congress on Sports

Technology (APCST) (60) (2013) 435-441, doi: https://10.1016/j.

proeng.2013.07.050.

[34] D. Tolley, T.N. Nguyen, A. Tang, N. Ranasinghe, K. Kawauchi, C.C. Yen,

Windywall: exploring creative wind simulations, in Proceedings of the Thirteenth

International Conference on Tangible, Embedded, and Embodied Interaction

(2019) 635-644, doi: https://doi.org/10.1145/3294109.3295624.

[35] Q. Zhao, Z. Lian, D. Lai, Thermal comfort models and their developments: a review,

Energy Built Environ. 2 (1) (2021) 21–33, https://doi.org/10.1016/j.

enbenv.2020.05.007.

[36] M. Schaub, L. Schmitt, M. Kriegel, Vergleich des Nutzungsgrads von

reaktionsschnellen, direktelektrischen W¨

armeübergabeformen mit konventionellen

Raumheizsystemen, HLH 4 (2020) 22–25, https://doi.org/10.37544/1436-5103-

2020-04-21.

[37] M. Hou, D. Aviv, A. Chatterjee, E. Teitelbaum, M. Rida, F. Meggers, D. Khovalyg,

Resolving indoor shortwave and longwave human body irradiance variations for

mean radiant temperature and local thermal comfort, Energy Build. (2023),

https://doi.org/10.1016/j.enbuild.2023.113581.

[38] D.N. Sørensen, L.K. Voigt, Modelling flow and heat transfer around a seated human

body, Build. Environ. 38 (2003) 753–762, https://doi.org/10.1016/S0360-1323

(03)00027-1.

[39] ASHRAE Standard 55 (2020), Thermal environmental conditions for human

occupancy, ASHRAE: Peachtree Corners, GA, USA, 2020.

[40] ISO7730-2005 Ergonomics of the Thermal Environment–Analytical Determination

and Interpretation of Thermal Comfort Using Calculation of the PMV and PPD

Indices and Local Thermal Comfort Criteria, International Organization for

Standardisation, Geneva, 2005.

[41] VDI (Association of German Engineers), VDI 3804:2009-03. Air-conditioning -

Office buildings (VDI ventilation code of practice), 2009.

[42] L. Schmitt, M. Schaub, S. Brandt, M. Kriegel, Occupancy-equivalent, operative

temperature control for an energy-saving operation of direct electric, low-capacity

heating systems, Build. Environ. 226 (2022) 109638, https://doi.org/10.1016/j.

buildenv.2022.109638.

[43] Siemens Digital Industries Software, Simcenter STAR-CCM+, version 2022.1,

Siemens 2022.

[44] T.H. Shih, W.W. Liou, A. Shabbir, Z. Yang, J. Zhu, A New k-epsilon Eddy Viscosity

Model for High Reynolds Number Turbulent Flows - Model Development and

Validation, NASA TM 106721, 1994.

[45] W. Xu, Q. Chen, F.T.M. Nieuwstadt, A new turbulence model for near-wall natural

convection, Int. J. Heat Mass Transfer 41 (1998) 3161–3176, https://doi.org/

10.1016/S0017-9310(98)00081-7.

[46] S. Gilani, H. Montazeri, B. Blocken, CFD simulation of stratified indoor

environment in displacement ventilation: Validation and sensitivity analysis, Build.

Environ. 95 (2016) 299–313, https://doi.org/10.1016/j.buildenv.2015.09.010.

[47] Y. Li, M. Sandberg, L. Fuchs, Effects of thermal radiation on airflow with

displacement ventilation: an experimental investigation, Energy Build. 19 (4)

(1993) 263–274, https://doi.org/10.1016/0378-7788(93)90011-I.

K. Tawackolian et al.

Energy & Buildings 322 (2024) 114748