Mitigating the Cold-Pane Effect: Empirical Evaluation of Active Glazing for Enhanced Thermal Comfort in Near Zero-Energy Buildings

1. Introduction

The global built environment is currently exposed to a dual pressure: rapidly increasing urban density and the urgent need to reduce greenhouse gas emissions. The building and construction sectors remain major contributors to global energy-related carbon dioxide emissions and final energy demand. [1,2,3,4]. Within this context, building operation is a dominant driver of energy use, with space heating accounting for a substantial share of total demand in European residential and commercial sectors [5,6,7]. This challenge is particularly acute in rapidly growing megacities characterized by high-density, highly glazed commercial high-rise typologies. Historically, the design paradigms of the late 20th and early 21st centuries favored expansive glass curtain walls to enhance daylighting and aesthetic appeal. However, these envelope configurations have introduced substantial thermal liabilities, resulting in acute microclimatic perturbations, high solar heat gains in summer, and severe steady-state transmission losses in winter, which fundamentally compromise perimeter thermal equilibrium. [8,9,10]

The dual seasonal thermal penalty of highly glazed envelopes is schematically illustrated in Figure 1.

The thermal performance of modern building envelopes is defined by a fundamental thermodynamic asymmetry: while contemporary opaque wall assemblies routinely achieve low thermal transmittances $U_{wall}\le$0.15 W/(m²·K) [11,12].The high-performance triple-glazed units typically remain constrained to significantly higher values $U_{glas}\approx 0.8-0.9$ W/(m²·K) [13,14]. This persistent thermal resistance gap establishes transparent components as the primary thermal vulnerability of the building envelope across all operational seasons. As visualised in Figure 2, the severe thermal transmittance differential between the opaque wall assembly and the triple-glazed unit induces acute microclimatic perturbations across the envelope interface.

During summer, expansive transparent facades transmit substantial solar radiation, significantly increasing indoor temperatures and cooling loads [15,16]. In hot climates, solar radiation ingress can create a marked indoor-outdoor temperature differential, raising the inner glass surface temperature ${\overline{T}}_{mr}$to as much as 43 °C and increasing the mean radiant temperature above the indoor air temperature [17]. Even highly insulated glazing can suffer from poor thermal performance in summer if solar control is inadequate, potentially causing indoor overheating [18]. The reliance on high-power air conditioning to counteract this solar heat gain drives building energy consumption far beyond that of opaque wall configurations [15]. Furthermore, the high solar reflectance of some advanced facades can cut cooling energy needs but may inadvertently increase outdoor heat stress in urban canyons [19].

In winter, the same glazing systems become a primary source of heat loss due to their characteristically high thermal transmittance compared to well-insulated opaque walls [20,21,22,23]. Under cold external conditions and standard indoor air setpoints, the internal surface temperature of passive triple glazing depresses. This low surface temperature triggers two distinct microclimatic failure modes:

-

radiant temperature asymmetry ${(∆T}_{pr}$). The cool glass surface creates directional long-wave radiative loss from an occupant’s body, depressing local ${\overline{T}}_{mr}$:

$${\overline{T}}_{mr}={\left(F_{p-w}{T}_{s, glass}^4+{\displaystyle \sum _{i=1}^n}{F}_{p-i}{T}_i^4\right)}^{0.25},$$

(1)

where $T_{s, glass}$ is the internal surface temperature of the active window pane, K; $T_i$represents the surface temperatures of surrounding opaque internal boundaries K; $F_{p-i}$ is the person-to-surface view factor$; F_{p-w}$ is the occupant-to-window geometric view factor; n is the total count of internal enclosure boundaries;

-

buoyancy-driven convective downdraughts. Room air in direct contact with the chilled glass undergoes rapid cooling and density increase. This chilled air mass detaches from the glass boundary, cascading downwards as a high velocity convective plume ($\nu \approx$ 0.15—0.30 m/s) [24]. Upon reaching the floor, it spreads laterally, penetrating the occupied perimeter zone (0.5–1.0 m) and elevating the draught rate at ankle height.

Consequently, passive glazing creates a thermodynamic conflict: maintaining a nominal air temperature (T_int= +20 $℃$) fails to prevent localized dissatisfaction (PPD > 10%) near the perimeter [25].

To overcome the physical limitations of static, passive barriers, “active” glazing technologies have emerged as a highly promising alternative. Electrically heated windows, which integrate a transparent conductive coating or an ultra-thin polymer-based heating foil within the glazing cavity, can actively modulate the glass surface temperature [26,27,28]. By regulating $T_{s, glass}$ to match or slightly exceed $T_{int}=20-22℃$, active glazing replaces the cold boundary with a controlled radiant surface, eliminating radiant temperature asymmetry and suppressing boundary-layer downdraughts [29,30]. Rather than calculating comfort metrics based on fixed window dimensions, thermal comfort criteria are specified as mandatory boundary conditions (e.g., ISO 7730 category B: $-0.5\le PMV\le PPD\le 10\%, ∆T_{pr}\le 10℃$). This allows direct mathematical optimization of the window-to-wall ratio (WWR) and active surface heat flux density (q_elec) [31,32]. Empirical studies show that active TGUs can direct up to 85% of generated thermal energy indoors, reducing overall heating demand by up to 65.6% under targeted control [33,34]. To render these systems fully sustainable and mitigate the risk of grid overloading during peak winter hours, contemporary research advocates for the integration of localised energy buffering. By coupling the active facade with decentralised renewable sources, such as building-integrated photovoltaics and advanced storage mediums, these systems offer a realistic pathway toward self-powering, high-performance architectural envelope [35,36].

To provide a comprehensive technical baseline of how passive, active, and dynamic glazing solutions influence building energy management and microclimatic stability, Table 1 systematically maps recent state-of-the-art literature.

As evidenced by the synthesized literature mapped in Table 1, dynamic technological strides have profoundly enhanced the energy autonomy and operational efficiency of active glazing systems. However, existing building physics research predominantly evaluates transparent active envelopes through macro-level building performance metrics, specifically bulk thermal transmittance (U-value), overall solar heat gain coefficients (SHGC), and integrated annual space-heating loads. While minimising transmission heat loss remains paramount for nearly zero-energy building regulatory compliance, the ultimate operational mandate of any architectural envelope is to sustain a stable, healthy, and thermally acceptable indoor microclimate for its occupants. Consequently, energy conservation strategies must not be decoupled from their direct biothermal impact on local human thermal comfort.

The complex conjugate heat transfer and biothermal interactions between the human body and the highly asymmetrical radiant field generated by a passive cold window, as opposed to an actively modulated transparent thermal barrier, remain insufficiently quantified in current literature. Traditional multi-pane glazing units, despite incorporating low-emissivity coatings and noble gas cavities, inherently induce acute localized discomfort during winter periods. This degradation stems from three primary physical mechanisms: depressed inner glass surface temperatures ($T_{s, glass}$), high plane radiant temperature asymmetry (${∆T}_{pr}$)., and buoyancy-driven convective downdraughts ($\nu$ > 0.20 m/s ).

In contrast, active electrically heated glazing transforms the transparent envelope from a structural thermal bridge into a controlled, low-temperature radiant heating surface. To address the existing knowledge gap, Fanger’s classic heat balance framework must be integrated directly into the thermodynamic evaluation of the envelope boundary layer. Fanger’s model provides the theoretical bridge between environmental heat fluxes and human thermal sensation, quantified via the predicted mean vote (PMV) and predicted percentage of dissatisfied (PPD) indices (ISO 7730; ASHRAE Standard 55).

Prior computational fluid dynamics and empirical investigations confirm that low surface temperatures on vertical transparent barriers generate intense convective cooling loops, initiating air detachment that severely degrades the microclimate in perimeter zones. Microclimatic analyses by Manz and Frank [66] alongside foundational turbulence models established by Fanger et al. [67], demonstrate that these boundary-layer downdraughts trigger unacceptable draught rates (DR > 20%) within immediate occupant spaces. Conversely, elevating $T_{s, glass}$ via conductive transparent oxide layers directly elevates the local mean radiant temperature (${\overline{T}}_{mr}$) and neutralises convective downdraught boundary layers, thereby expanding the usable floor area towards the envelope perimeter [68,69].

To bridge this critical knowledge gapbetween macro-level building energy efficiency and localized indoor environmental quality by formulating, empirically validating, and energetically optimizing an autonomous electro-thermal active glazing system designed to eliminate cold-pane effects and guarantee thermal comfort across severe sub-zero boundary conditions.

To achieve this overarching goal, the study addresses the following specific objectives:

-

to empirically quantify conjugate heat transfer mechanisms, surface temperature dynamics, and boundary-layer suppression across an active triple-glazed unit subjected to simulated exterior conditions down to −25.0 °C within a dual-zone climatic chamber;

-

to evaluate microclimatic impacts and human thermal comfort performance at critical workstation proximities (x = 0.5 m to 1.0 m) using high-resolution spatial mapping of Fanger’s indices (PMV, PPD, and DR) under both passive and active operating regimes;

-

to investigate transient control stability and energetic viability by assessing closed-loop PID response metrics, thermal inertia, and continuous electrical power density scaling across variable climate transitions;

-

to establish an architectural scalability model that determines the maximum permissible window-to-wall ratio WWR_max enabled by active thermal regulation, successfully reconciling high transparency demands with stringent indoor comfort mandates.

2. Materials and Methods

2.1. Dual-Zone Climatic Chamber, Envelope Component Configuration and Experimental Setup

To evaluate the conjugate heat transfer dynamics and localized microclimatic perturbations across the building envelope interface under severe winter operational regimes, empirical investigations were performed using a specialised dual-zone climatic chamber at the Laboratory of Energy-Efficient Technologies, Kielce University of Technology. The central dividing septum incorporates a highly insulated opaque wall envelope (${\mathrm{U}}_{\mathrm{w}\mathrm{a}\mathrm{l}\mathrm{l}}$=0.15 W/(m²⋅K) accommodating a representative modular double-cavity triple-glazed unit (TGU) with a clear vision aperture of 520 mm × 780 mm (aspect ratio H/W = 1.5) and a nominal thermal transmittance of ${\mathrm{U}}_{\mathrm{g}}$ =1.1 W/(m^2⋅K). The transparent assembly features a symmetrical 4/18/4/18/4 mm pane configuration (K-Glass ESG 4 / 18 Ar / Float 4 / 18 Ar / K-Glass ESG 4). The outer and inner boundary panes consist of 4mm toughened safety glass coated with a durable pyrolytic low-emissivity layer (K-Glass), whereas the central pane comprises an un-toughened 4 mm clear float glass. Perimeter structural integrity and gas containment are provided by 18 mm aluminium spacer bars backfilled with a molecular sieve desiccant, connected at the corners via primary butyl-sealed poly-vinyl chloride key connectors. Secondary weatherproofing, edge-seal protection, and mechanical adhesion are established using a single-component neutral-cure silicone sealant (Modesil NO 33, Lakma). The experimental facility comprises an exterior environmental compartment engineered to maintain sub-zero ambient temperatures down to $T_{out}=-25 ℃$ under controlled convective conditions, and an interior occupant compartment regulated at a constant indoor setpoint of T_in= 20 ℃ with a relative humidity of RH = 50% ± 2.5% (Figure 3).

Non-contact surface temperature distribution and thermal gradient uniformity across the entire transparent area were verified using a calibrated Testo 872s thermal imaging camera. Continuous heat flux density profiles (q_surf, W/m²) were recorded across the 3 x 3 grid using nine Hukseflux FHF05 sensors connected to the HIOKI LR8432 logger. (Figure 4). To systematically track boundary layer development, edge losses, and thermal heat injection from the heating element, the 18 sensors were mapped in Cartesian coordinate matrix across the 520 mm × 780 mm aperture (Figure 4b).

An active resistive heating element (ThermoTECH flexible heating foil, Conrad Electronic) with a nominal electrical power rating of $P_{nom}=60 W$ was centered horizontally along the lower boundary on Surface 5. In addition to gas and surface temperature monitoring, localized surface heat flux densities and effective thermal transmittance were continuously tracked using nine Hukseflux FHF05 thin-film flexible heat flux sensors possessing a nominal sensitivity of approximately 20 µV /(W/m²) (Figure 4b). Output signals from this flexible array were acquired and logged via a high-precision Hioki heat flow logger LR8432. In order to guarantee metrological fidelity and the exact reproducibility of the experimental setup, a comprehensive overview of the technical specifications, sensor accuracy, and functional roles of all utilized equipment is detailed in Table 2.

2.2. Microclimatic Comfort Assessment and Fanger Thermal Comfort Model

To comprehensively evaluate the indoor environmental quality benefits and potential localized discomfort risks induced by the active glazing assembly, multi-point microclimatic measurements were conducted within a specialized full-scale climatic chamber facility featuring an integrated external wall and window test bench. Measurements were performed across the full spectrum of standardized ISO 7730 vertical elevations: z = 0.1 m (ankle level), z = 0.6 m (lumbar height), z = 1.1 m (seated head height), and z = 1.7 m (standing head height)) at two critical horizontal offsets from the internal glass surface (x = 0.5 m representing the immediate proximity zone and x = 1.0 m denoting the standard human occupancy zone). All physical parameters were captured via a Testo 400 multifunction IAQ meter equipped with high-accuracy sensors (air temperature accuracy of $\pm 0.3℃$, air velocity accuracy of $\pm 0.003{\displaystyle \frac{m}{s}}+4\% m.v.$, and a class 1 globe thermometer). The thermal performance and resulting human thermal comfort indices of the active facade were systematically investigated across a wide-ranging spectrum of external boundary conditions (spanning from $T_{out}=-25℃ \mathrm{t}\mathrm{o} 5℃$). Complete parametric datasets for all evaluated external temperatures are compiled in the supplementary materials (Table S1–S6).

The thermal load on the human body (L) represents the differential between internal metabolic heat production and the total heat flux density dissipated to the surrounding domain, $H_{loss}$:

The aggregated internal and external heat loss density $H_{loss}$ is formulated as:

$$\begin{gathered} H_{loss} =3.96·{10}^{-8}{f}_{cl}\left[{\left(T_{cl}+273.5\right)}^4-{\left({\overline{T}}_{mr}+273.5\right)}^4\right]+f_{cl}{h}_{c }\left(T_{cl}{-T}_a\right)+3.05·{10}^{-3} \\ ·\left[5733-6.99\left(M-W\right)-\left.p_a\right]+0.42\left[\left(M-W\right)-58.15\right.\right]+1.7 \\ ·{10}^{-5}M\left(5867-p_a\right)+0.0014M(34-T_a) \end{gathered}$$

(2)

where $p_a$ is the partial water vapour pressure, Pa, $T_{cl}$ is the outer clothing surface temperature, ∘C, and $h_c$is the convective heat transfer coefficient, W/(m²⋅K). The clothing surface temperature $T_{cl}$ is determined iteratively via energy equilibrium:

$$T_{cl} =35.7-0.028\left(M-W\right)-I_{cl}·H_{loss}$$

(3)

$$PMV=(0.303e^{-0.036M}+0.028)·\left[\left(M-W\right)-H_{loss}(T_{a, }{\overline{T}}_{mr}, v,p_a,I_{cl}\right]$$

(4)

where $M$ is the metabolic rate, corresponding to sedentary office activity; $W$ is external mechanical work (W = 0 W/$m^2$), $p_a$is water vapour partial pressure, Pa; $T_{in }$is local air temperature, $℃$; ${\overline{T}}_{mr}$ is mean radiant temperature, $℃$; $f_{cl}$is the clothing surface area factor. Standard winter clothing insulation of $I_{cl}$= 1.0 clo (0.155 m² K/W) was assumed.

The predicted percentage of dissatisfied (PPD) index is directly linked to

$$PPD=100-95\mathrm{e}\mathrm{x}\mathrm{p}(-0.03353·PM{V}^4-0.2179·PM{V}^2)$$

(5)

Local convective boundary-layer acceleration and turbulence intensity (TU) were logged at ankle height (z = 0.1 m) using the Testo omnidirectional hot-wire anemometer. The local draught rate (DR, %) was evaluated as:

$$DR =\left(34-T_a\right) {\left(v-0.05\right)}^{0.62}\left(0.37·v·TU+3.14\right)$$

(6)

where TU is the empirical turbulence intensity, % logged at 10Hz., . TU=$={\displaystyle \frac{{\sigma }_v}{\overline{v}}}·100\%.$

Spatial compliance is categorized strictly against ISO 7730 thresholds:

Category A: $PPD\le 6\%; -0.2\le PMV\le +\mathrm{0,2};DR\le 10\%.$

Category B: $PPD\le 10\%; -0.5\le PMV\le +0.5;DR\le 15\%.$

Category C: $PPD\le 15\%; -0.7\le PMV\le +0.7;DR\le 20\%.$

2.3. Microclimate Characterisation and Boundary Assumptions

All microclimatic measurements were conducted under steady-state nocturnal winter design conditions, assuming zero short-wave solar irradiance (G_sol = 0 W/m²). The localized mean radiant temperature (${\overline{T}}_{mr})$ was derived directly from long-wave equilibrium energy balance captured by the Testo black globe thermometer probe following ISO 7726:

$${\overline{T}}_{mr}={\left[{(T_g+273.15)}^4+{\displaystyle \frac{1.1·{10}^8{v}^{0.6}}{e_g{D}^{0.4}}}(T_g-T_{in})\right]}^{1/4}-273.15$$

(7)

where $T_g$is the equilibrium globe temperature, $℃;$, $T_a$is the local air temperature $℃; v$ is the local air velocity m/s; $e_g$is the globe surface emissivity, $e_g$= 0.95; and D is the globe diameter, D= 0.15 m. The operational temperature was subsequently evaluated as$T_{op}=0.5(T_{in}+{\overline{T}}_{mr})$.

Plane radiant temperature asymmetry ${∆T}_{pr}$across the vertical glazing boundary was calculated in accordance with ISO 7730 based on directional surface emissivities and surface-to-point view factors ($F_{p-w}$):

$${∆T}_{pr} =\left|T_{pr, window}-T_{pr, interior}\right|$$

(8)

2.3. Analytical Derivation of the Maximum Permissible Window-to-Wall Ratio

To extend the empirical findings obtained from the modular active glazing specimen (820 mm x 560mm) to a full-scale building envelope configuration, an analytical optimisation model was formulated. Rather than evaluating the building envelope solely at a singular peak design load, the model evaluates the maximum permissible window-to-wall ratio (WWR_max) across a continuous outdoor temperature domain $-25.0{\le T}_{out}\le 5 ℃.$This thermal sweep encompasses the full operational spectrum, from extreme sub-zero peak winter conditions ($T_{out}$ = $-25.0$ $℃$) to typical mild winter and shoulder-season regimes ($T_{out}$ = 5.0$℃$).

The objective of the model is to derive the WWR_max that guarantees compliance with ISO 7730 Category A standards (maintaining PPD_perimeter ≤ 6% while simultaneously minimising active electrical energy consumption. Under a regulated indoor air setpoint of $T_{in}$ = 20.0 $℃$, Category A mandates a minimum spatial mean radiant temperature threshold of ${\overline{T}}_{mr, target}\ge 19.0 ℃$.

The inner surface temperature of the unheated glass pane (surface 6), $T_{s, glass passive}$is governed by the steady-state thermal resistance network between the indoor environment ($T_{in}$) and the variable outdoor cold sink ($T_{out}$):

$$T_{s, glass passive}=T_{in}-{\displaystyle \frac{U_g}{h_i}}(T_{in}-T_{out})$$

(9)

where $U_g$ =1.1 W/(m^2⋅K) is the nominal glazing thermal transmittance, and $h_i$is the standardized internal combined convective-radiative heat transfer coefficient in accordance with EN 673, $h_i$= 7.7 W/(m^2⋅K).

Upon energising the integrated ThermoTECH active heating layer at an electrical power flux density $q_{elec}$ with a thermal transformation efficiency of $\eta \approx$0.95, the modulated active inner surface temperature${ T}_{s, glass}\left(q_{elec},T_{out}\right)$ is expressed as:

$$T_{s, glass}\left(q_{elec},T_{out}\right)=T_{in}- {\displaystyle \frac{U_g}{h_i}}\left(T_{in}-T_{out}\right)+{\displaystyle \frac{\eta ·q_{elec}}{h_i}}$$

(10)

Expressing the spatial radiant field $T_{rad}$ as a function of the transparent envelope fraction (WWR = ${\displaystyle \frac{A_{glassing}}{A_{\mathrm{f}acade}}})$and the occupant-to-window view factor $F_{p-w}$ yields the dynamic envelope optimization constraint:

$$T_{rad}=\left(1-F_{p-w}·WWR\right)·T_{wall}+(F_{p-w}·WWR)·T_{s, glass}\left(q_{elec},T_{out}\right)\ge T_{rad, min}$$

(11)

where $T_{wall}$represents the internal surface temperature of the highly insulated opaque wall assembly, $U_{wall}\le 0.15 \mathrm{W}/{\mathrm{m}}^2·\mathrm{K}$.

Explicitly rearranging equation (11) yields the governing equation for $WW{R}_{max}$ under active dynamic regulation:

$$WW{R}_{max}\left(T_{out}, q_{elec}\right)={\displaystyle \frac{T_{rad,min}- T_{ wall}}{F_{p-w} (T_{s, glass}\left(q_{elec},T_{out}\right)-T_{wall} }}.$$

(12)

3. Results

3.1. Conjugate Heat Transfer and Surface Heat Flux Density Profiles

3.1.1. Spatial Experimental Heat Flux Profiles Under Extreme Sub-Zero Boundary Conditions

The spatiotemporal thermal behaviour of the room-facing glass surface (Surface 6) was mapped using a 3 x 3 array of thin-foil Hukseflux FHF05 heat flux sensors under controlled environmental conditions. The experimental protocol evaluated the glazing assembly across an external climate chamber temperature range of $-25.0{\le T}_{out}\le 5 ℃$. Comparative evaluations were conducted between the baseline unheated (passive) state and three active heating set points ($T_{heat}$=30$℃$, 40$℃$, and 45$℃)$.

Under the unheated extreme cold baseline condition (Figure 5a), the glazing assembly operates purely as a passive thermal barrier. In this state, heat flux density across all nine sensor locations remains strictly negative, oscillating between $-30.0$ $\mathrm{W}/{\mathrm{m}}^2$ and $-65.0\mathrm{W}/{\mathrm{m}}^2$, with a space-averaged mean of $\overline{q}$= $-18.4$ $\mathrm{W}/{\mathrm{m}}^2$across the central aperture. Heat transfer is governed by a series conductive-convective thermal resistance network driven by the macro-scale thermal differential between the ambient indoor environment ($T_{in}$ = 20 $℃$)) and the cold environmental chamber ($T_{out}=-25℃$). Spatial non-uniformity across the 3x3 grid is notable even in the passive state. Sensors along the bottom edge (C7, C8, C9) record consistently more negative values than those in the upper tier (C1, C2, C3) (Figure 4b). This spatial thermal heterogeneity stems directly from the interplay of two coupled heat transfer mechanisms. Firstly, conductive thermal bridging through the peripheral warm-edge spacer frame depresses localized surface temperatures along the glazing margin. Secondly, buoyancy-driven natural convection within the hermetically sealed argon cavity causes dense, cooled gas to cascade along the outer pane and pool near the lower spacer, thereby establishing a persistent, localized thermal sink at the base of Surface 5. The bottom tier of the glass assembly, tracked by Sensors 7, 8, and 9 (Figure 5a), suffers from the most severe thermal losses, with heat flux values ranging from $-45$ W/m² down to $-65$ W/m² (space-averaged mean $\overline{q}\approx -52 \mathrm{W}/{\mathrm{m}}^2)$. This significant acceleration of heat loss at the base of the glazing is driven by two synergistic physical phenomena. As argon gas within the cavity is cooled along the exterior pane (Surface 4), its density increases rapidly. The cooled gas cascades downward as a high-velocity boundary layer downdraft and pools at the bottom of the cavity. This cold gas reservoir creates a severe localized heat sink on the cavity side of Surface 5, dramatically steepening the conductive thermal gradient across the bottom section of the inner glass pane. Conductive short-circuiting through the lower warm-edge spacer assembly and frame interface further depresses the local surface temperature $(T_{S6,bottom}$). Consequently, cold room air cascades downward along Surface 6, creating a localized cold-draft zone that significantly degrades occupant indoor thermal comfort and increases condensation risk along the lower edge.

3.1.2. Parametric Heat Flux Response Across Active Thermal Setpoints

The experimental internal surface heat flux densities and corresponding mean inner pane temperatures are illustrated in Figure 6.

The empirical evaluation of the nine-node active glazing array (C1–C9) conducted within a controlled climatic chamber environment under varying thermal setpoints reveals profound spatial heterogeneities in heat transfer dynamics across the building envelope interface. As delineated in Figure 6, the internal surface heat flux density ($q$) exhibits a strong non-linear dependency on external ambient temperatures descending to $T_{out}=-25℃$, a condition primarily driven by coupled conductive and radiative interactions between the centre of glass and perimeter edge zones. Under low-temperature operation at a modest $T_{heat}=30℃$, all sensor nodes (C1–C9) remain within the negative heat flux regime, signifying a persistent net heat loss towards the internal environment, wherein the upper and central nodes experience severe thermal depression reaching values as low as $q$=$-69\mathrm{W}/{\mathrm{m}}^2$ at an external temperature of $T_{out}=-25℃$. Conversely, the perimeter edge nodes (C7–C9) demonstrate significantly attenuated heat loss profiles, a phenomenon attributable to frame-induced thermal bridging and localised insulation effects that modulate the convective exchange. Elevating the interlayer setpoint to $T_{heat}=40℃$ shifts the baseline thermal flux curves upward, thereby mitigating extreme negative heat fluxes at the upper perimeter; however, a pronounced convergence and overlapping of intermediate nodes (C2–C6) is observed under severe sub-zero boundary conditions of $T_{out}<15℃$, which indicates that uniform internal heating fails to adequately compensate for the multidimensional edge cooling effect in the absence of dynamic power modulation. This interplay reaches a critical juncture at the maximum experimental setpoint of $T_{heat}=45℃$, where a distinct thermodynamic phenomenon of gradient inversion occurs. Whilst the nodes (C1–C6) maintain a conventional negative heat loss profile even under extreme cold, the lower perimeter edge nodes (C8 and C9) transition into the positive quadrant, achieving peak inward heat flux densities of $q=43.1{\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2}}$ at $T_{out}=-25℃$.

3.1.3. Active Perimeter Heating Dynamics and Rayleigh–Bénard Convective Regimes Under Extreme Sub-Zero Boundary Conditions

Energising the ThermoTech polyester heating foil located along the lower perimeter of Surface 5 fundamentally restructures the conjugate heat transfer dynamics across the inner glass pane. Heat generated at the Surface 5 interface ${(q}_{gen})$ conducts transversely through the glass pane thickness ($d_g$) to Surface 6 according to Fourier’s law:

$$q_{glass}=-{\lambda }_{glass}{\displaystyle \frac{T_{s6}-T_{S5}}{d_g}},$$

(13)

As the active setpoint is systematically increased from $T_{heat}$=30$℃$ to $T_{heat}$=45$℃$ (Figures 5b–5d), the indoor surface temperature (${TS}_6$) elevates, shifting the entire heat flux matrix upward toward the zero-loss boundary ($q=0 \mathrm{W}/{\mathrm{m}}^2$) (Table 3).

At setpoint $T_{heat}$=30$℃$ and at $T_{out}$ = $-25.0℃$(Figure 6b) thermal losses along the bottom perimeter are significantly attenuated. The heat flux recorded by Sensor 8 (bottom center) rises from its passive baseline value of $-52 \mathrm{W}/{\mathrm{m}}^2$ to an average of $\overline{q}=-23.8 \mathrm{W}/{\mathrm{m}}^2$, cutting transmission losses along the lower edge by 54.2%.

At setpoint $T_{heat}=40℃$(Figure 6c) the system establishes a near-adiabatic thermal barrier along the lower central axis. Sensor 8 operates at a mean heat flux of $\overline{q}=-8.2 \mathrm{W}/{\mathrm{m}}^2$ (with transient minimum losses of $-0.95 \mathrm{W}/{\mathrm{m}}^2$). His localized thermal injection eutralizes the cold draft originating from the environmental chamber. In setpoint $T_{heat}=45℃$(Figure 6d) maximum active heating forces heat flux values across the lower glazing region to cross the zero-loss threshold $(q=0 \mathrm{W}/{\mathrm{m}}^2)$ into positive territory. Sensor 8 reaches an average flux of $\overline{q}=-$9.4 $\mathrm{W}/{\mathrm{m}}^2$ during steady state, but experiences transient positive peaks up to 14.9 $\mathrm{W}/{\mathrm{m}}^2$. Similarly, Sensor 9 reaches 14.6 $\mathrm{W}/{\mathrm{m}}^2$. At this setpoint, Surface 6 transitions from a net energy sink into an active low-temperature radiant panel, delivering direct net thermal gains into the indoor room.

A prominent feature of the transient flux response under high active setpoints ($T_{heat}=40℃$ and $T_{heat}=45℃$) is the appearance of high-frequency temporal signal oscillations across all sensor channels (Figure 6c and Figure 6d). This behavior is driven by fluid flow instabilities within the sealed argon-filled cavity II. When the lower portion of Surface 5 is heated to $T_{heat}=45℃$against an external pane temperature near $T_{out}=-25℃$, an intense horizontal temperature gradient $(∆T_{cavity}>60K$) is established across the cavity gap width (b). This differential elevates the local Rayleigh number above its critical value ${(Ra}_{crit}\approx {10}^3)$:

$$Ra={\displaystyle \frac{{g\beta (T}_{s_5}-T_{S4})b^3}{\nu ·\alpha }},$$

(1)

where $g$is the gravitational acceleration; $\beta$ is the volumetric thermal expansion coefficient of argon, $\nu$ is the kinematic viscosity, and $\alpha$ is thermal diffusivity.

Exceeding this critical buoyancy threshold incites Rayleigh–Bénard natural convection within the gas domain, effectively transitioning the argon gap from a quiescent state into a dynamically recirculating thermal regime. This fluid-dynamic transition proceeds via a dual-boundary-layer circulation mechanism. Firstly, a buoyancy-driven ascending plume develops along the heated lower segment of Surface 5. Argon gas immediately adjacent to this thermal boundary undergoes rapid volumetric expansion and consequent density reduction, generating a positive buoyancy vector that accelerates the fluid upward. Concurrently, a descending cold plume dominates the opposing boundary; upon impinging on the upper cavity limit and transferring thermal energy to the cold outer glazing (Surface 4), the gas experiences a sharp increase in density, cascading downward as a chilled down-draught.

Within the cavity core, these counter-flowing boundary-layer streams interact intensely, establishing a high-shear zone. This velocity differential induces flow instabilities characterized by periodic vortex shedding. As these turbulent eddies intermittently impinge upon Surface 5, they provoke localized, transient perturbations in the internal convective heat transfer coefficient ($h_{conv}$). Consequently, these thermal pulses propagate conductively through the glazing thickness and are recorded by the room-side sensors as the high-amplitude, high-frequency heat flux oscillations observed during active thermal regulation.

3.2. Thermographic Surface Mapping and Spatial Temperature Profiles on the Inner Glazing Surface Across Climatic Regimes

To evaluate the conjugate heat transfer dynamics and localised microclimatic perturbations across the triple-glazed unit, experimental investigations were conducted within a specialised dual-zone climatic chamber. Instead of presenting a continuous sequence of thermograms for every temperature increment, the analysis synthesises the system’s thermal response by focusing on four discrete, representative winter boundary conditions. This approach highlights the baseline performance under extreme frost while mapping the parametric trends across a broader operational spectrum. High-resolution thermographic images captured by the calibrated Testo 872s IR camera confirmed the spatial uniformity of the thermal boundary layer across the active sightline area 520 mm × 780 mm, while surface temperatures across the inner glazing were additionally measured using array-mounted Hukseflux sensors, with all experimental data continuously recorded via a high-precision Hioki data acquisition system.

Under the passive baseline regime at an external ambient temperature of $T_{out}=-25℃$ (Figure 7a), the room-side glazing surface (Surface 6) exhibited pronounced thermal bridging effects. Infrared thermography and synchronized hukseflux sensor—hioke logs revealed a highly heterogeneous temperature distribution governed by multidimensional conductive heat transfer through the structural aluminium spacer bars, coupled with internal convective down-draughts. Quantitatively, under this severe boundary condition, the minimum surface temperature at the lower-edge perimeter plummeted to $12.4 ℃$ (against a nominal indoor air temperature of $T_{in}$ = 20$℃)$. This sharp local depression generated an intense thermal sink, creating an elevated risk of surface condensation and significant occupant discomfort driven by localized radiant asymmetry.

; (b) Passive baseline ($T_{out}=0℃;P=0 W$; (c) Active heating setpoint $T_{heat}$=30 $℃$; (d) Active heating setpoint $T_{heat}$=40 $℃$ (isothermal lower boundary achieved); € Active heating setpoint $T_{heat}$=45 $℃$ (full surface transition to a radiant emitter).

Upon energising the lower-perimeter heating foil (Surface 5), the thermal profile of the triple-glazed unit was fundamentally restructured (Figure 7c–e). As the active set-point was sequentially increased from 30$℃$ to 40$℃$ and 45$℃$, both the thermographic and high-precision logging data captured a progressive upward propagation of the thermal front. At a heating set-point of $T_{heat}=30 ℃$ (Figure 7c), the minimum perimeter temperature rose moderately to $16.5 ℃$, effectively mitigating the severe thermal bridge identified during the passive baseline regime $X_{M8}=12.00$. Elevating the set-point to $T_{heat}=40 ℃$ (Figure 7d) successfully neutralised the local edge cooling effect, elevating the perimeter temperature to $18.2 ℃$, and establishing a near-isothermal temperature distribution across the lower quartile of the TGU aperture.

Under maximum thermal regulation ($T_{heat}=45℃$) (Figure 7e), the perimeter temperature reached $19.8 ℃,$causing the entire fenestration surface to transition into an active low-temperature radiant emitter that completely suppressed cold down-draughts. Extending the empirical evaluation across the broader climatic setpoints $-25.0{\le T}_{out}\le 5 ℃$, parametric thermographic and heat-flux sensor analysis confirms that the electrical power input required to maintain a strictly adiabatic boundary $(q =0 \mathrm{W}/{\mathrm{m}}^2$) on Surface 6 scales linearly with the external temperature depression. Specifically, maintaining neutrality at $T_{out}=10℃$ required a proportional reduction in heating power of approximately 38.5% relative to the $T_{out}=-25℃$ baseline. This highly predictable linear response verifies the operational stability of the system, enabling seamless integration with advanced building energy management systems for predictive, demand-responsive control.

To corroborate the thermographic findings and quantify the surface temperature distribution across the entire active aperture within the specialised climatic chamber, data gathered from the array of nine Hukseflux sensors and continuously logged via the Hioki high-precision data acquisition system were subjected to rigorous steady-state analysis. Figure 8 illustrates the extended steady-state thermal response of the room-side glazing surface, $T_{s, glass6},$spatially averaged across the sensor matrix as a function of external ambient temperature under varying active heating set-points (30 °C, 40 °C, and 45 °C). Following data validation and spatial processing, the internal glazing temperature exhibits a robust, highly linear correlation across all tested thermal load regimes. Decreasing the external ambient temperature from +5 °C down to -25 °C induces a proportional reduction in $T_{s, glass6}$, directly reflecting exacerbated conductive and convective heat losses through the building envelope.

The thermal performance curves systematically shift upward with incremental increases in the internal heater supply set-points. This demonstrates that under extreme conditions of $-25 °C$, elevating the set-point from $30 °\mathrm{C}$ to $45 °\mathrm{C}$ raises the mean internal glazing temperature from approximately $16.2 °\mathrm{C}$ to $17.0 °\mathrm{C}$. This consistent upward shift effectively mitigates downdraught phenomena and maintains inner surface temperatures well above critical dew-point thresholds. Furthermore, the near-parallel trajectories of the characteristic curves indicate stable, invariant heat transfer coefficients throughout the tested climatic range, confirming the reliability and repeatability of the dataset.

3.3. Microclimatic and Thermal Comfort Assessment Near the Active Glazing Assembly

Evaluating microclimatic conditions near active building envelopes is essential to ensure that integrated heating technologies successfully eliminate localized discomfort, such as cold downdraughts or thermal stratification, while maintaining adequate indoor environmental quality. Building upon the experimental framework and measurement procedures detailed in Section 2.2, the empirical data obtained from the full-scale climatic chamber tests are analyzed to quantify the spatial thermal performance of the active glazing assembly. It must be noted that while this controlled chamber environment enables precise parametric mapping under defined thermal boundaries, it inherently abstracts from stochastic real-world phenomena, such as variable solar irradiance, dynamic wind pressures, and transient internal gains. Consequently, the findings reflect idealized boundary interactions rather than full operational complexity. While complete numerical datasets for all tested boundary conditions are compiled in the supplementary materials (Table S1 and Table S2), the following subsections examine key microclimatic parameters, including local temperatures, air velocity, and thermal comfort indices (PMV, PPD, and DR), across representative operational regimes.

3.3.1. Perimeter Down-Draught Mitigation and Microclimatic Dynamics Under Extreme Sub-Zero Environmental Condition

Under extreme sub-zero environmental condition, the structural and thermal resilience of the active façade undergoes a rigorous stress evaluation. As visualised in the vertical profiles for the near-window boundary at x = 0.5 m (Figure 9), operating under a low heating setpoint ($T_{heat}=30℃$), results in local microclimatic conditions that are heavily penalised by radiant asymmetry and aggressive convective cooling along the internal glass pane. The low mean radiant temperature near the unconditioned perimeter causes the PMV at ankle level (z = 0.1 m) at x = 0.5 m to drop to $-0.78$, corresponding to an elevated PPD of 20.4% and a draught rate (DR) of 14.2%, thereby confirming the persistence of cold downdraught phenomena driven by natural convective cooling along the internal pane. Moving outward to x = 1.0 m (Figure 10), these adverse gradients attenuate moderately, yet local discomfort persists with an ankle-level PMV of $-0.62$ and PPD of 13.0%.

Elevating the transparent resistive heating layer setpoint to $40 ℃$ significantly mitigates these spatial heterogeneities, raising the indoor operative temperature profile and reducing local PPD below the 10% threshold recommended by ISO 7730 for standard conditioned spaces across both horizontal offsets. At the maximum operational setpoint $T_{heat}=45℃$the active glazing successfully transitions into an active radiant heating source. The radiant asymmetry is entirely reversed, raising interior surface and globe temperatures to generate a slight positive PMV (+0.39 at z = 1.1 m for x = 0.5 m, and +0.43 for x = 1.0 m) that effectively counteracts envelope-induced thermal dissatisfaction, albeit requiring sophisticated, dynamic modulation controls to prevent minor localized warmth sensations in immediate proximity to the façade.

3.3.2. Microclimatic Dynamics Under Intermediate Winter Boundary Conditions

Under moderate sub-zero environmental forcing ($T_{out}=-5℃$), the structural stress on the active façade diminishes significantly compared to extreme frost conditions, allowing for more energy-efficient operational regimes. As visualised in the vertical microclimatic profiles for the near-window boundary at x = 0.5 m (Figure 11), operating under a baseline heating setpoint ($T_{heat}=30℃$),) results in a mild suppression of local air temperatures, yielding an ankle-level (z = 0.1 m) local air temperature of 17.8 $℃$and a corresponding PMV of $-0.35.$Crucially, due to the reduced thermal head across the glazing assembly, the PPD at ankle level remains at 11.2%, hovering just above the 10% threshold stipulated by ISO 7730 for Category B environments, while the DR peaks moderately at 10.1%. Moving outward to the extended occupancy zone at x = 1.0 m (Figure 12), convective currents attenuate further, with baseline local discomfort subsiding such that the PPD fully complies with standard criteria across all elevations. The implementation of an intermediate active heating setpoint $T_{heat}=40℃$ completely eradicates localized cold discomfort, stabilising the indoor microclimate. At this setpoint, the ankle-level PPD drops well below 6%, and the thermal comfort indices achieve optimal uniformity across both horizontal offsets (x = 0.5 m and x = 1.0 m). Conversely, elevating the active glazing to the maximum operational setpoint ($T_{heat}=45℃$) under these intermediate external conditions leads to an over-correction. The thermal output generates strong buoyant plumes that elevate upper-level PMV values up to +0.72 at z = 1.7 m, accompanied by an elevated PPD of 8.2% driven by minor local warmth sensations.

3.3.3. Over-Heating Risks and Setpoint Modulation in Mild Regimes

Under mild transitional winter conditions ($T_{out}=0℃$), the thermal demand on the active building envelope decreases substantially, shifting the operational focus from severe cold mitigation to fine-tuned indoor environmental quality control and overheating prevention. As visualised in the vertical microclimatic profiles for the near-window boundary at x = 0.5 m (Figure 13), operating under a baseline heating setpoint $T_{heat}=30℃$ yields stable local air temperatures ranging from 18.3 $℃$at ankle level z = 0.1 m to 20.0$℃$ at head level z = 1.7 m.

Despite the mild exterior temperature, the PMV at the perimeter remains slightly negative, registering $-0.39$ at ankle level and improving to $-0.18$ at the upper boundary. Crucially, the PPD remains well below the 10% threshold mandated by ISO 7730, peaking at 8.5% near the floor and dropping to 7.4% at z = 1.7 m, while the DR remains acceptable with a maximum of 9.2%. Moving to the extended occupancy zone at x = 1.0 m (Figure 14), these profiles shift further towards thermal neutrality, with baseline PPD values stabilising near or below 6.2%.

Deploying an intermediate active heating setpoint $T_{heat}=40℃$ successfully eliminates any remaining sub-neutral bias, aligning indoor thermal conditions closely with optimal comfort standards. At this intermediate setting, the PPD across the vertical profile at x = 0.5 m flattens uniformly at approximately 5.1%, while global operative temperatures reach comfortable levels. Conversely, applying the maximum operational setpoint $T_{heat}=45℃$ under a mild external forcing of 0.0 $℃$leads to pronounced thermal over-servicing. The excessive radiant and convective output elevates upper-level PMV values up to +0.82 at z = 1.7 m (x = 0.5 m) and drives the PPD at x = 1.0 m as high as 19.0% due to local thermal dissatisfaction associated with overheating.

3.4. Transient Dynamic Response and Closed-Loop PID Control Stabilityand Power Demand Analysis

The transient thermal response of the active glazing assembly was evaluated by subjecting the system to standardized step-change perturbations in external boundary conditions, simulating sudden transition from mild $5.0 ℃$ to severe $-25.0 ℃$ winter design scenarios. The closed-loop control framework, managed by the Lumel RE72 PID controller, was monitored with a temporal resolution of $∆t=1 \mathrm{s}$using the HIOKI LR8432 data acquisition system. The primary feedback loop utilized surface-mounted Omega SA3-K thermocouples (Type K, Class 1) to provide real-time surface temperature $T_{si,feedback}$ to the PID algorithm. Simultaneously, the HIOKI logger captured the active power density $q_{heat}\left(t\right)$ by monitoring the RMS voltage drop across the ThermoTECH heating foil. This high-fidelity synchronous logging enabled the calculation of the controller’s transient performance indicators, providing a comprehensive metric of the system’s ability to maintain the target setpoint despite fluctuating external thermal loads. The derived performance metrics are synthesized in in Table 4.

An analysis of the dynamic performance indicators compiled in Table 4 reveals distinct behavioural trends governed by the macro-thermal inertia of the active triple-glazed units. As environmental severity increases, with external temperatures dropping from $5.0 ℃$ to $-25.0 ℃$ alongside proportional setpoint elevations, the required active heat flux density expands significantly. This operational shift results in a proportional elongation of both the rise time, which increases from 12.4 min to 22.6 min, and the settling time, scaling from 28.6 min to 51.8 min. Such responses highlight the robust thermal mass damping capacity inherent in advanced multifunctional glazing systems. Despite aggressive actuator modulation necessitated by severe sub-zero thermal gradients reaching down to$∆T=70K$, the refined PID tuning framework effectively restricts percentage overshoot to OS$\le$3.1%. This tight constraint prevents localized thermal expansion stresses that could otherwise compromise the structural integrity of pyrolytic low-emissivity coatings. Furthermore, the cumulative tracking penalty, quantified by the Integral of Time-weighted absolute error (ITAE) index over an evaluation horizon of $t_{eval}=60 min$. Furthermore, (ITAE) index scales from $4.12·{10}^3 K·m^2$ under mild operational conditions to $11.30·{10}^3 K·m^2$under severe winter environments, accurately capturing the extended transient adjustment phase required to stabilise massive thermal loads. Crucially, across all evaluated operational profiles, the steady-state error remains tightly bounded between ±0.03$℃$ and ±0.05 $℃$.

To maintain a raised inner surface temperature $T_{set}$ against aggressive external thermal forcing $T_{out},$ the active coating must supply a precise heat flux density. Under steady-state conditions, the required active heat flux compensates for the total outward thermal losses through the triple-glazed unit envelope, governed by the the overall thermal transmittance (U-value) of the assembly. The steady-state active heat flux density and total electrical power demand were quantified across the complete spectrum of environmental boundary conditions. Given the aperture area of A = 0.4056 m² and a maximum rated heating capacity of P_max= 60.0 W at a severe design gradient of $∆T$= 70.0 K, the thermal power demand scales linearly with boundary severity (Table 5).

As demonstrated in Table 5, the steady-state heat flux density increases from $q_{heat,ss}=52.83 \mathrm{W}/{\mathrm{m}}^2$, ${(P}_{el}=$ 21.43 W) under mild transition conditions ($T_{out}=5℃$) up to a peak design load of $q_{heat,ss}=147.93{\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2}},(P_{el} = 60.00W$) under severe winter conditions.

3.5. Parametric Validation of the Maximum Permissible Window-to-Wall Ratio

To evaluate the scalability and practical architectural impact of the proposed active electro-thermal glazing system, the empirical parameters validated on the laboratory test specimen were rigorously extrapolated to a standard architectural window module representative of modern facades $A_{wind st}=1.82 m^2,$ dimensions 1.23 m×1.48 m. Maintaining the validated area coverage ratio $C_r=0.142$, the active heating foil area was scaled proportionally to $A_{wind st}=0.258 m^2$. This methodological transition ensures that the local electrical power density $P_{foil}=350{\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2}}$translates into a consistent global effective active power flux density of $P_{akt, ef}=49.7{\displaystyle \frac{W}{m^2}}$= across the entire standard fenestration. Furthermore, the occupant-to-window angle factor was adjusted to reflect a typical perimeter workstation distance of 1.0 m $(F_{p-w}=0.23)$. The analytical matrix was executed across a continuous winter outdoor temperature domain from −25.0$℃$ to +5.0$℃$ with a ΔT=5.0K resolution. The boundary conditions incorporated an indoor air setpoint of $T_{in}==+20.0 ℃$, an internal convective heat transfer coefficient of $h_i$= 7.7 W/(m^2⋅K), and a highly insulated opaque wall thermal transmittance of $U_{wall}=0.12$W/(m^2⋅K). o reflect realistic engineering practices and address potential over-conservatism in architectural design, the optimization model was evaluated across three distinct indoor environmental quality thresholds based on ISO 7730 standards:

-

category A: strict comfort requiring a spatial mean radiant temperature $T_{rad,min}\ge 19℃$

-

category B: moderate comfort allowance, $T_{rad,min}\ge 18.5℃$

-

category C: relaxed threshold for rare, extreme weather events, $T_{rad,min}\ge 18℃$

The resulting inner surface thermal states and corresponding maximum permissible window-to-wall ratios WWR_max under both passive and active regimes are synthesized in Table 6.

As evidenced by the numerical data in Table 6, relying on conventional passive triple-glazed assemblies ${\mathrm{U}}_{\mathrm{g}}$ =1.1 W/(m^2⋅K). severely restricts architectural design freedom during severe winter conditions. At an external design temperature of $T_{out}=-25℃$, the inner glass surface temperature drops critically to +13.57$℃$, forcing architects to restrict the facade transparency ratio to a restrictive 24.6% to prevent localized thermal discomfort.In stark contrast, activating the closed-loop electro-thermal heating system elevates the inner glass pane temperature above the indoor ambient setpoint, maintaining ${ T}_{s, glass}\left(q_{elec},T_{out}\right)\ge +19.7 ℃$ even under extreme $T_{out}=-25℃$ boundary conditions. This effectively eliminates the cold radiant sink, allowing a fully unconstrained WWR_max = 100.0% across the entire temperature spectrum. However, in standard engineering practice, designing strictly for category A under highly infrequent meteorological extremes is rarely economically or practically justified. If design criteria are relaxed to ISO 7730 category B or C during these severe frost periods, the maximum permissible passive WWR expands significantly, reaching 60.7% and 98.6% respectively at $T_{out}=-25℃$. This highlights a fundamental architectural trade-off: vast transparent envelopes can indeed be achieved passively, provided that occupants accept minor, transient microclimatic degradation during extreme weather peaks. In stark contrast, activating the closed-loop electro-thermal heating system decouples facade architectural layout from these physiological compromises. By elevating the inner glass pane temperature above the indoor ambient setpoint (maintaining ${ T}_{s, glass}\left(q_{elec},T_{out}\right)\ge +19.7 ℃$ even at $T_{out}=-25℃$, the active system effectively eliminates the cold radiant sink. This allows a fully unconstrained WWR_max = 100.0%while rigorously sustaining premium Category A comfort across the entire climatic spectrum, albeit accompanied by operational electrical energy input.

4. Discussion

The empirical and analytical findings of this study provide a comprehensive quantitative evaluation of conjugate heat transfer and local biothermal interactions across active electrically heated triple-glazed units under severe winter operating conditions. While contemporary building physics literature extensively validates the macro-level energy efficiency of passive high-performance glazing systems through bulk thermal transmittance ($U_g$-values) and solar heat gain coefficients, the results of this investigation demonstrate that static thermal barriers remain constrained in their ability to maintain localized indoor environmental quality (IEQ) near transparent building envelopes.

4.1. Conjugate Heat Transfer and Surface Thermal Dynamics

Under the passive baseline regime $T_{out}=-25℃$, the room-facing glass surface ${(T}_{s6}$) exhibited severe thermal depressions down to 12.4$℃$, driving negative heat flux densities ranging from $-30.0 \mathrm{t}\mathrm{o} -65.0$ W/m², as mapped across the 3x3 Hukseflux FHF05 sensor array (Figure 4 b). In alignment with the foundational observations in [24,66,70], this pronounced surface cooling triggered acute radiant temperature asymmetry and buoyancy-driven convective downdraughts. Consequently, occupants situated within the perimeter zone (x = 0.5m) experienced elevated local discomfort, characterized by predicted mean vote values dropping to PMV=$-0.78$and predicted percentage dissatisfied indices exceeding PDD=20.4% (Figure 10 and Figure 11), thus violating ISO 7730 Category B and C comfort thresholds. Conversely, integrating an active resistive heating foil along the lower perimeter of Surface 5 fundamentally restructured the thermodynamic boundary layer (Figure 4a). As evidenced by the spatial heat flux mappings in Figure 5b–d and Figure6, modulating the internal surface temperature via closed-loop PID control successfully eliminated the cold-pane effect across all evaluated thermal load regimes. At an active setpoint of $T_{heat}=40℃$, the glazing assembly established a near-adiabatic thermal shield along the lower boundary ${(q}_{\mathrm{S}6}=-8.2 \mathrm{W}/{\mathrm{m}}^2$; Table 3), suppressing perimeter draught rates and reducing PPD below the mandatory 10% threshold. Operating at the maximum experimental setpoint ($T_{heat}=45℃$) induced a gradient inversion, transforming the fenestration surface into a low-temperature radiant emitter that delivered net inward thermal gains ${(q}_{\mathrm{S}6}$ up to +43.1$\mathrm{W}/{\mathrm{m}}^2$; Figure 6d). Furthermore, the steady-state thermal response curves illustrated in Figure 8 confirm a highly linear correlation between internal glazing temperatures and external ambient temperatures, verifying the predictive reliability of the thermal boundary model.

From a fluid-dynamic perspective, the emergence of high-frequency temporal heat flux oscillations under high active setpoints $T_{heat}=40℃$ and $T_{heat}=45℃$underscores the complex conjugate interactions within argon-filled cavities. When horizontal temperature differences across Cavity II exceeded critical Rayleigh number thresholds (Ra > ${10}^3$), the gas domain transitioned from a quiescent state into a dynamically recirculating Rayleigh–Bénard convection regime. Periodic vortex shedding and thermal plume interactions generated transient fluctuations in convective heat transfer coefficients, highlighting the necessity of advanced predictive control algorithms to stabilize active facade performance.

Authors should discuss the results and how they can be interpreted from the perspective of previous studies and of the working hypotheses. The findings and their implications should be discussed in the broadest context possible. Future research directions may also be highlighted.

4.2. Microclimatic Implications and Human Thermal Comfort Performance

A critical contribution of this study is the high-resolution spatial mapping of microclimatic parameters across standardized vertical elevations and horizontal offsets (Figure 9, Figure 10, Figure 11, Figure 12, Figure 13 and Figure 14). Under extreme sub-zero conditions $T_{out}=-25℃$, passive configurations create severe microclimatic stratification, where ankle-level discomfort z = 0.1 m dominates due to the convergence of cooled boundary-layer air plumes. Implementing intermediate active heating $T_{heat}=40℃$ successfully harmonized vertical temperature gradients, ensuring that PPD values remained strictly within ISO 7730 Category A and B limits (Figure 9c and Figure 10c, 11c, 12c). However, the microclimatic profiling under intermediate $T_{out}=-5℃$, and mild $T_{out}=0℃$, boundary conditions (Figure 11, Figure 12, Figure 13 and Figure 14) revealed that static high-temperature setpoints $T_{heat}=45℃$ can induce localized thermal over-servicing. Specifically, excessive radiant output under mild exterior forcing elevated upper-level PMV values up to +0.82 and increased PPD up to 19.0% due to local warmth dissatisfaction (Figure 14). This underscores the operational necessity of weather-compensated, adaptive PID control algorithms rather than fixed-temperature heating schedules.

4.3. Transient Control Stability and Energetic Viability

The closed-loop PID control framework demonstrated robust stability across all evaluated climatic transitions, as summarized in the dynamic performance indices compiled in Table 4. While rising external severity from from −25.0$℃$ to +5.0$℃$ naturally extended the thermal rise time from 12.4 to 22.6 min and settling time from 28.6 to 51.8 min due to the substantial thermal inertia of the triple-glazed assembly, the refined PID tuning restricted percentage overshoot to $OS\le 3.1\%$. This tight constraint is vital for preventing excessive thermal expansion stresses that could otherwise jeopardize the mechanical integrity of pyrolytic low-emissivity coatings. In terms of operational energy demand, steady-state active heat flux densities scaled proportionally with the external thermal gradient$∆T$, ranging from 52.83 $\mathrm{W}/{\mathrm{m}}^2$ ($P_{el}=21.43 W)$ at $T_{out}=5℃$ to a peak design load of 147.93 $\mathrm{W}/{\mathrm{m}}^2$ ($P_{el}=21.43 W$) at $T_{out}=-25℃$ (Table 5).

4.4. Architectural Scalability and Window-to-Wall Ratio Optimization

Extrapolating the empirical data to a standard architectural window module via the analytical optimization model yields profound implications for modern building (Table 6). Traditional passive glazing constraints force architects to restrict window-to-wall ratios WWR down to 24.6% at $T_{out}=-25℃$ to avoid severe thermal discomfort near the perimeter. By maintaining active inner surface temperatures above $19.7℃$through controlled electro-thermal regulation, the proposed system successfully eliminates cold radiant sinks, enabling an unconstrained maximum permissible window-to-wall ratio WWR_max=100% across the entire sub-zero temperature spectrum. This broadens the viable architectural design space, reconciling aesthetic and daylighting demands with stringent thermal comfort mandates. However, while achieving a fully unconstrained maximum window-to-wall ratio via active electro-thermal governance effectively eradicates cold radiant asymmetry and interior discomfort, the associated electrical power demand during extreme meteorological events $q_{heat,ss}\approx 148{\displaystyle \frac{\mathrm{W}}{{\mathrm{m}}^2}}$ warrants careful consideration regarding operational expenditures and grid stability. Unmitigated continuous operation under peak frost conditions would introduce a substantial burden on local electrical distribution networks. o reconcile high transparency with nearly zero-energy building (NZEB) mandates, the active glazing architecture must be integrated directly with building-integrated photovoltaics and localized electrical storage nodes. Because extreme sub-zero design temperatures typically coincide with high solar irradiation availability on vertical facades during clear winter days, BIPV generation can supply a significant fraction of the required low-voltage input. Furthermore, implementing advanced model predictive control frameworks ensures that thermal activation is neither continuous nor indiscriminate; instead, power is modulated dynamically—leveraging thermal inertia, targeted occupant zone occupancy, and predictive weather forecasting—to achieve effective peak shaving and maintain systemic energy neutrality.

4.5. Limitations and Future Research Directions

Despite the distinct advantages of utilizing a climatic chamber to isolate the thermodynamic performance of the active glazing system, certain methodological limitations must be addressed. While the controlled environment successfully eliminates external stochastic variables, it inherently restricts the direct replication of complex, multi-zone airflow interactions, dynamic occupant behaviours, and transient solar gains typical of fully occupied built environments. Additionally, idealized laboratory boundary conditions may deviate from the heterogeneous thermal fields encountered in real-world structures. Crucially, the empirical observations revealed that operating the active glazing at elevated surface set-points from $T_{heat}=40℃$ to $T_{heat}=45℃$ induces a transitional Rayleigh–Bénard convection regime within the argon-filled cavity, resulting in localized fluid-dynamic disturbances and transient heat flux oscillations. Furthermore, under milder winter conditions (e.g., $T_{out}=0℃$), unmodulated high-power inputs can cause thermal over-servicing, elevating the local PPD index up to 19% as a consequence of internal overheating. These findings underscore that active electro-thermal fenestration is highly sensitive to control methodologies; therefore, future implementation requires sophisticated, sensor-driven closed-loop algorithms (e.g., adaptive PID controllers) to dynamically modulate thermal output. Nevertheless, these controlled findings provide a robust and scientifically rigorous foundation for analytical optimisation models. Future research efforts will focus on long-term in-situ field monitoring within occupied nZEB testbed facilities to validate these active envelope strategies under genuine meteorological transients and adaptive user dynamics.

5. Conclusions

This study has comprehensively investigated the thermodynamic behaviour, conjugate heat transfer mechanisms, and localized microclimatic impacts of an autonomous, electrically heated active triple-glazed unit designed for nearly zero-energy buildings (NZEBs). By integrating Fanger’s thermal comfort framework with high-precision experimental data acquired from a dual-zone climatic chamber, the research bridges the critical gap between macro-level building energy efficiency and micro-level human thermal perception.

The principal findings of this investigation are summarized as follows:

-

under extreme sub-zero exterior boundary conditions, passive multi-pane glazing assemblies suffer from severe surface temperature depression ${ T}_{s, glass}\approx 12.4\mathrm{℃}$ along the lower edge perimeter, inducing acute radiant temperature asymmetry and buoyancy-driven convective downdraughts

-

energising the integrated low-voltage heating foil along the lower boundary of Surface 5 effectively restructured the boundary-layer dynamics. Active thermal regulation at set points from $T_{heat}=40℃$ to $T_{heat}=45℃$ successfully neutralized cold-pane effects, shifting the room-facing surface from a net thermal sink into a controlled, low-temperature radiant heating panel that delivered positive inward heat fluxes up to 43.1 $\mathrm{W}/{\mathrm{m}}^2$;

-

Spatial microclimatic mapping confirmed that active surface modulation successfully restored local thermal comfort within the critical occupant perimeter zone from x = 0.5 m to 1.0 m. The predicted percentage of dissatisfied (PPD) index was successfully maintained well below the ISO 7730 Category B threshold PPD < 10% across standardized vertical elevations;