Computation of the Mean Radiant Temperature


The Mean Radiant Temperature T_r is defined as the uniform temperature of an enclosure in which the radiant heat transfer is equal to radiant heat transfer in the real, non-uniform enclosure. Indeed, indoor and outdoor environments can be rather complex, integrating solar fluxes \varphi or different surface temperatures. The mean radiant temperature hence combines all radiant contributions in a unique indicator.

The mathematical expression is as follows:

\sigma \epsilon S T_r^4 = \alpha S \varphi + \sigma \epsilon \sum_{i=1}^{n} S_i F_i T_i^4

where F_i are the View Factors between surfaces at a temperature T_i and the studied body.

In the following sections, we will first discuss solar contribution to radiant temperature and then the contribution of infrared radiation.

Solar Contribution

A blunt method of calculating the short-wavelength component is to apply the projection factor method, which considers that the direct normal flux \varphi^\perp is applied on the human body with a projection factor taking into account the cylindrical nature of the human body:

\varphi = \alpha \varphi^\perp f_p(h)

where the projection factor depends on the solar height: f_p(h) = 0.233 \cos(h)+ 0.067.

This simple method can be refined by considering the solar contribution as divided into a diffuse part, \varphi_d , and a direct one, \varphi_b , treated separately due to their nature, as described in the following sections.

Direct flux

The direct flux arriving on a cylindrical individual can be computed analytically from the direct flux on a horizontal plane \varphi_b^h , knowing the solar altitude angle h and the angle of incidence i :

\varphi_b = \varphi_b^h \times \frac{\cos(i)}{\sin(h)}

The flux received on the cylinder is calculated analytically from the following geometrical configuration and the angle of incidence \theta with respect to the normal to the cylinder surface :

This translates into the following integral, the result of which is the projected surface orthogonal to sun rays, which by the nature of the cylinder is always azimuthal (i.e. in the axis of the sun's rays)

\phi_{b} = \int_{-\frac{\pi}{2}}^{\frac{\pi}{2}} \varphi_b \cos(\theta) r H d\theta = 2 \varphi_b r H = S_p \varphi_b

Elementary surface exposed to direct flux.

Note : in practice, using this method provides similar received solar flux values as the f_p method for direct flux (see section 4.1 of this paper)

The value of direct solar flux calculated by the f_p method is however slightly higher, as shown in the model comparison below.

Comparison of the "projection factor fp" method and of the integral computation for a complete year (8760 hours)

Diffuse flux

Perez’ diffuse solar flux model is used (more details), considering the circumsolar, horizon brightening and isotropic dome contributions, as per following figure:

Representation of Perez' model for diffuse flux with an individual in the center of the scene.

In the following, we hypothesize that the dome contribution ( L ) and horizon contribution ( K_2 ) are isotropic on the cylinder, and that the circumsolar, anisotropic, contribution ( K_1 ) behaves as if it were specular, lighting “directly” the scene (see sketch below):

Isotropic and non-isotropic contributions of the Perez diffuse flux model on a cylinder

The total diffuse flux received is hence computed as:

\phi_d = 2 r H \varphi_{d}^{K_1} + 2 \pi r H \times (\varphi_{d}^L +\varphi_{d}^{K_2} ) = S_p \varphi_{d}^{K_1} + S_c (\varphi_{d}^L +\varphi_{d}^{K_2} )

Numerical verification

The total solar flux received is then \phi = S_p\varphi_b + \phi_d in [W].

Comparing the analytical model of the cylinder presented above with those of a discrete, 12 faced cylinder (using the package pvlib), we obtain following results, showing a good agreement between both models.

Comparison of the numerical and analytical models

The discrepancies between models are given below and present an RMSE inferior to 0.76 [W/m²] for the three solar fluxes, which demonstrates the validity of the approach.

Differences between the numerical and analytical models for direct, diffuse and total solar flux.

Results for three summer days

An example of application of the method described hereinabove is plotted for three summer days on the figure below. The same orders of magnitude as in the experimental literature are found (Kantòr et al. 2018, Park & Tuller 2011) as well as the "daytime relapse" phenomenon identified by Kantòr et al. 2013. The comparison with the so-called "6-directions" method (also described here) yields similar results.

Received fluxes over three days

The analogy of the results with those obtained by the projection factor method, derived by Fanger for an individual, leads to the conclusion that the cylindrical integral method is suitable for modelling the flux received by humans.

The results obtained are of the same order of magnitude as those of the "6-directions" method, whose pitfalls were presented in Kantòr et al. 2018.

The advantage of the presented method is that it requires only the direct and diffuse horizontal fluxes, as well as the solar height, which are usually present in meteorological files or easily determined with freely available tools.

The method could be extended to an ellipse whose analytical formula is known in order to approximate other projection factor configurations proposed in the literature.

Infrared contribution

The calculation of the long wave contribution is made possible by determining the view factors between facing walls. For this purpose, the pyViewFactor code is used, as described in this paper from IBPSA France 2022 conference.


Taking a geometrical representation of an individual as an octagon, one wishes to calculate the contribution of different walls in a room. The radiation view factor is defined as the ratio of the fluxes received by a wall (or a set of walls) to the total flux "outgoing" from the emitting wall. Thus it can be calculated by summing the fluxes between the n cells of a discretized wall and the 9 faces of a the "individual" (8 faces of the octagon plus the cap):

F_p = \frac{ \sigma \epsilon \sum_{j=1}^{9} \sum_{i=1}^{n} S_i F_{ij} T_p^4 }{ \sigma \epsilon S T_p^4} = \frac{ \sum_{j=1}^{9} \sum_{i=1}^{n} S_i F_{ij}}{ S}

View Factors of the walls to an individual in the centre of a room

An application of this method is the calculation of the contribution of each wall to the radiation balance of an individual located in the centre of a room, as shown in the following figure.

Slice view of a room with and individual in its center.

The calculation is carried out for different room lengths: the results are given in the figure below. As the individual is positioned in the centre of the zone, it can be seen that the contribution of the vertical walls decreases significantly with the size of the room. After about ~6 metres, it can be seen that the contributions of the different walls are roughly equivalent: one third for the floor, one third for the ceiling and one third for the four walls of the room.

Contribution of the different walls according to the size of the room

View factors of walls depending on the position in a room

We are now interested in the influence of the various walls with regards to the individual's position on the floor. The same calculation is carried out for each cell of the floor mesh, by successively repositioning the individual-octagon at each cell center.

The following maps are obtained, giving the view factor of each wall towards the individual, depending on its position in the room:

View Factor of a wall depending on the individual position in the room
View Factor of a ceilling depending on the individual position in the room

Spatial Mean Radiant Temperature

This makes it possible to calculate the radiant temperature in the room as a function of the individual's position. To determine this, the equation is solved for T_r giving the radiant temperature of the room walls that would cause the same radiative exchange as the existing walls, with their respective areas and view factors:

\sigma \epsilon \sum_{i=1}^{n} S_i F_{i} T_r^4 = \sigma \epsilon \sum_{i=1}^{n} S_i F_i T_i^4


T_r = \sqrt[4] { \frac{ \sum_{i=1}^{n} S_i F_i T_i^4 }{\sum_{i=1}^{n} S_i F_{i} } }

Performing the calculation on a cubic room where the walls are at 20°C, except for one wall at 0°C, we obtain the following result, which shows a difference of ~6 K in radiant temperature difference.

MRT field in a cubic room with a window at 0°C and all other surfaces at 20°C

For a more complex configuration, e.g. a train maintenance facility, including windows, radiant panels and a train at the outside temperature, the following results are obtained:

Spatial distribution of the radiant temperature in a non-uniform environment (windows/radiant heating panel/walls/and a bus at outdoor temperature, indoors for servicing).
Studied geometrical configuration
MRT in the maintenance facility without radiating panels

Abaques de facteur de forme entre une personne et une paroi

Afin d’actualiser et de rendre plus lisibles les travaux de Fanger présentant le facteur de forme entre un individu et différentes paroi (voir l’exemple ci-après donnant le facteur de forme avec un plafond), nous allons utiliser le package PyViewFactor.

Abaque de facteur de forme entre une personne et un plafond (tiré de Fanger 1970)

La géométrie de l’individu est tirée de l’exemple du doorman de PyVista. Afin d’alléger les calculs, une version comprenant moins de cellules a été créée en utilisant le filtre decimate. À titre d’exemple, l’image ci-dessous donne le facteur de forme entre les facettes composant l’individu et le sol.

Dimensions utilisées pour le calcul de facteur de forme

L’abaque résultant est donné ci-dessous :

Abaque de facteur de forme d’un individu vers le sol en fonction de ses dimensions a,b

L’abaque de facteur de forme d’un individu vers un plafond de dimensions a,b=longueur,largeur et de hauteur c est donné ci-dessous.

Abaque de facteur de forme entre un individu et un plafond (a = longueur, b = largeur, c = hauteur)

Le facteur de forme entre un individu et une paroi tel que sur la figure qui suit est donné sur l’abaque ci-après (l’orientation de l’individu influe marginalement sur le résultat).

Paramètres de la géométrie utilisée
Abaque de factgeur de forme entre l’individu et la paroi verticale