In Situ Winter Performance and Annual Energy Assessment of an Ultra-Lightweight, Soil-Free Green Roof in Mediterranean Climate: Comparison with Traditional Roof Insulation

Abstract

Green roofs are effective passive strategies for enhancing building energy efficiency and indoor thermal comfort, particularly in response to climate change. This study presents an experimental and numerical assessment of an ultra-lightweight, soil-free green roof system for Mediterranean climates. In situ thermal monitoring was carried out on two identical test rooms in Rome (Italy), comparing the green roof to a traditional tiled roof under winter conditions. Results revealed a 45% reduction in thermal transmittance. These data were used to calibrate a dynamic TRNSYS 18 model and then applied to annual simulations of energy demand and indoor comfort across different roof configurations, including expanded polystyrene-insulated reference roofs. The model was calibrated in accordance with ASHRAE Guideline 14, achieving an MBE within ±10% and a CV(RMSE) within ±30% for hourly data, ensuring the simulation’s reliability. The green roof reduced cooling energy demand by up to 58.5% and heating demand by 11.6% relative to the uninsulated reference case. Compared to insulated roofs, it maintained similar winter performance while achieving summer operative temperature reductions up to 0.99 °C and PPD decreases up to 2.94%. By combining field measurements with calibrated simulations, this work provides evidence of the green roof’s effectiveness as a passive retrofit solution for Mediterranean buildings.

1. Introduction

The building sector plays a critical role in achieving global decarbonization goals, accounting for approximately 36% of final energy consumption and nearly 40% of CO₂ emissions worldwide [1]. In response to this environmental challenge, the European Union has established ambitious targets through the 2030 Climate and Energy Framework [2] and the Renovation Wave strategy [3], promoting deep energy renovation as a primary lever to enhance building performance and climate resilience [4,5].

Within this framework, nature-based solutions are increasingly acknowledged for their dual potential to improve thermal performance while simultaneously mitigating the urban heat island effect and enhancing environmental quality in urban areas [6,7,8,9]. Among these, green roofs have gained particular attention due to their ability to reduce building energy demand, regulate indoor temperatures, and deliver ecosystem co-benefits [10,11,12].

The role of green roofs is particularly relevant in Mediterranean urban environments [13,14], characterized by hot summers, increasing climatic stress, and dense built fabrics [15,16]. In these contexts, lightweight green roofs represent a viable solution for retrofitting existing buildings that lack the structural capacity to support conventional green roof systems [17,18,19]. While the existing Mediterranean literature on green roof performance is extensive [16], studies focusing specifically on soil-free or ultra-lightweight systems remain scarce. Such configurations, often based on synthetic growing media or modular pre-vegetated mats, are particularly relevant for retrofitting buildings with limited structural capacity. Investigations have examined the performance of soil-free and ultra-lightweight green roofs in different climates [20,21,22], yet experimental evidence under real Mediterranean operating conditions remains limited, underscoring the relevance of the present study. However, the thermal and energy performance of such systems is highly dependent on roof design, climatic conditions, and simulation assumptions [23,24,25].

Although several studies have assessed green roof performance through numerical simulations, most rely on non-calibrated models, which limits predictive reliability. As documented in a recent systematic review [16], there is a clear need for studies that simultaneously combine in situ experimental validation, high-resolution urban climate data, and calibrated annual energy simulations to assess the performance of lightweight green roofs in Mediterranean environments.

Moreover, a critical and often overlooked issue concerns the meteorological input data used in simulations. As thoroughly discussed in a previous study [26], the widespread reliance on standardized Typical Meteorological Year (TMY) datasets, typically derived from airport stations, often fails to capture the different thermal dynamics of densely built urban areas, where microclimatic variations can be substantial. This mismatch may lead to considerable inaccuracies in estimating the passive potential of green roofs, undermining the reliability of simulation-based approaches. Furthermore, TMY datasets may fail to capture recent climate trends or extreme weather events, which could potentially compromise simulation accuracy [27,28]. Periodic updates and site-specific calibration are therefore strongly recommended.

To overcome these methodological limitations, this study adopts an integrated approach that combines high-resolution experimental validation, calibrated dynamic simulations, and updated urban climate datasets representative of Mediterranean conditions.

The investigated green roof system, designed within the framework of the “GRINN-S” research project, is a soil-free, extensive solution optimized for buildings with limited structural load capacity. Its performance was experimentally assessed through in situ thermal monitoring under real operating conditions. Various techniques can be employed to monitor the surface temperature of green roof layers with heterogeneous compositions, including advances in non-contact methods such as spectral pyrometry [29]. In this study, the choice of contact-based sensors was made in accordance with the recommendations and best practices discussed in our recent systematic review on the thermal and energy performance assessment of Mediterranean green roofs [16].

The resulting dataset was used to calibrate a building energy model developed in TRNSYS 18 [30], which was subsequently validated according to ASHRAE Guideline 14 [31]. Once validated, the model was applied to a simplified residential prototype to simulate year-round thermal performance and energy demand under the climatic conditions of Rome (Italy), using urban weather data. The analysis included a comparative assessment with conventional roof insulation strategies aimed at quantifying differences in energy demand, thermal behavior, and indoor thermal comfort. By integrating empirical evidence, numerical accuracy, and site-specific climatic forcing, this study delivers robust and transferable insights to support the deployment of lightweight green roofs as a scalable and climate-responsive retrofit strategy for the Mediterranean building stock.

This work distinguishes itself from previous studies by combining high-resolution in situ winter monitoring, a rigorously calibrated TRNSYS model driven by site-specific urban climate data, and a comparative analysis with varying EPS (Expanded Polystyrene) insulation levels, providing new evidence on the seasonal performance of an innovative ultra-lightweight green roof in Mediterranean conditions.

The paper is organized as follows: Section 2 defines the aim and scope of the study. Section 3 describes the methodological approach, including both the in situ experimental setup and the dynamic simulation procedures. Section 4 serves a dual function: it presents the results of the thermal characterization and dynamic simulations, and critically discusses the key findings, with an emphasis on their practical applicability and contribution to the scientific literature. Finally, Section 5 summarizes the key contributions and outlines directions for future research.

2. Aim and Scope

This study advances the current state of knowledge on the thermal and energy performance of green roofs in Mediterranean climates by addressing several critical research gaps identified in the literature. In contrast to the majority of existing studies that rely on uncalibrated simulation models, this research integrates high-resolution in situ experimental monitoring with a rigorously calibrated TRNSYS 18 model, validated according to ASHRAE Guideline 14 standards. Moreover, it incorporates updated, site-specific urban weather data instead of conventional airport-derived TMY datasets, thereby capturing the distinct microclimatic conditions of dense Mediterranean urban areas and improving predictive reliability. It focuses on an innovative ultra-lightweight, soil-free green roof system specifically designed for buildings with limited structural capacity, which has been scarcely investigated in real operating conditions. The study provides a direct, systematic comparison between the green roof and reference roofs equipped with different levels of EPS insulation, assessing not only heating and cooling energy demand but also indoor thermal comfort indicators over an annual cycle. By integrating empirical validation, climate-responsive modeling, and a comprehensive comparative framework, this work aims to provide novel and transferable insights to support the scalable adoption of lightweight green roofs as a climate-adaptive retrofit strategy for the Mediterranean building stock.

The research framework is structured into three sequential and interlinked phases: (i) the empirical thermal characterization of a novel lightweight green roof system; (ii) the development and validation of a calibrated simulation model based on real boundary conditions; (iii) the comparative assessment of annual energy performance across multiple roof configurations under site-specific Mediterranean climates.

This approach was adopted to fulfill the following key research objectives:

• to quantify the potential reduction in energy demand resulting from the installation of an ultra-light green roof system on conventional buildings’ roofs;
• to assess the impact of the green roof on inner surface temperature profiles and indoor thermal conditions under real and simulated operating scenarios;
• to simulate how different roof insulation levels influence the thermal performance of a green roof-equipped building envelope.

This framework addresses a critical gap identified in the recent literature [16], specifically the lack of integrated approaches that combine empirical validation, site-specific urban climate data, and dynamic simulation.

3. Methodology

The green roof system analyzed in this study was developed within the “GRINN-S” project, a publicly funded initiative aimed at designing lightweight, nature-based solutions for energy retrofitting in Mediterranean climates. As illustrated in Figure 1, the prototype consists of a soil-free extensive green roof, featuring a synthetic substrate over a polyethylene root barrier, equipped with a subirrigation system and an automated control unit. Designed for buildings with limited structural capacity, it combines high thermal efficiency with minimal structural load impact. During the winter season, the vegetative layer enters a dormant phase, characterized by negligible metabolic activity and halted growth, thus eliminating the need for irrigation and allowing the system to function in fully passive mode. The full-scale installation enabled extended in situ thermal monitoring under real operating conditions, providing a robust experimental foundation for calibrating and validating a dynamic simulation model.

The methodological framework adopted in this study combines empirical monitoring and numerical simulation to evaluate the thermal and energy performance of an extensive ultra-lightweight green roof system. The approach is structured into two interrelated macro-phases described in the following sections.

3.1. Experimental Phase

The experimental campaign started during January 2025 near Rome (Italy), on two thermally conditioned test rooms located side by side (see Figure 2) and characterized by the same constructive technologies (solid bricks plastered on both sides, and a typical composite slab roof with hollow clay infill blocks and reinforced concrete ribs). The winter monitoring period was selected to meet ISO 9869-1 [32] requirements for U-value determination, ensuring a consistent indoor–outdoor temperature gradient above 10 °C, which minimizes the influence of variable solar gains and wind speed on measured heat flux. This stable thermal gradient allows for a more accurate estimation of intrinsic thermophysical properties. The two test rooms will be referred to below as P1 (the one characterized by a traditional clay-tile roof, serving as reference) and P2 (the one with the green roof).

Both test rooms were equipped with a controlled heating system based on infrared lamps regulated by thermostats set to maintain an indoor air temperature of 20 °C. Heat flux sensors were mounted on the inner side of the roofs, and thermocouples were installed to acquire surface and air temperature data at 10 min intervals. Figure 3 provides some pictures related to the installation phase.

Figure 4 illustrates the overall sensor installation layout. Heat flux sensors (Hukseflux, model HFP01 [33]) and contact surface temperature sensors (LSI Lastem, model EST124 [34]) were installed on the inner surface of the roofs. Within the test rooms, air temperature sensors (LSI Lastem, model EST033 [35]) were installed and shielded to prevent direct radiation from the heating system. Moreover, EST124 sensors were used to measure the temperature on the external surface of the conventional roof (P1), and the temperature beneath the green roof system, at the interface between the green layer and the structural roof (P2). Finally, an EST033 sensor (equipped with a radiation shield) was installed for measuring the external air temperatures.

The collected data was post-processed to calculate the thermal transmittance (U-value) of both roof configurations. The analysis followed the ISO 9869-1 standard [32], applying the progressive average method, which allows for the estimation of thermal transmittance under real operating conditions:

$$U={\displaystyle \frac{\sum q}{\sum (T_{int}-T_{ext})}}$$

(1)

where $q$ is the heat flux density, and $T_{int}$ and $T_{ext}$ are the internal and external air temperatures, respectively.

The three convergence criteria defined by the standard were applied to validate the measurements. In particular: (i) the measurement duration must be at least 72 h; (ii) at the end of the measurement period, the calculated thermal transmittance value must not differ by more than ±5% from the value obtained 24 h earlier; (iii) the analysis of data from the initial period equal to INT(2DT/3) days must yield a result that does not differ by more than ±5% from the value obtained by analyzing the data from the final period of the same duration (where DT is the total duration of the test in days, and INT denotes the integer part of the expression).

The assessment of measurement uncertainty was carried out in accordance with the principles of the ISO Guide to the Expression of Uncertainty in Measurement. The analysis considered the statistical variability of the measurements, with the standard error of the mean used to quantify the influence of random fluctuations. The individual uncertainty contributions from the heat flux sensor and from the internal and external air temperature sensors were combined by applying the law of propagation of uncertainties, using the root-sum-of-squares method. The expanded uncertainty, corresponding to a 95% confidence level, was obtained by applying a coverage factor k = 2, in line with international metrological practice.

3.2. Simulation Phase

The experimental dataset was used to calibrate a dynamic simulation model in TRNSYS 18. The model was constructed using TRNSYS Build to define the building envelope and internal conditions, and the dedicated Type 785 (see the TESS library for more details [36]) was used to represent the thermal dynamics of the green roof system. Experimentally measured indoor and outdoor air temperatures, recorded from 16–23 January 2025, were imposed as thermal boundary conditions.

Although the calibration dataset refers to winter conditions, this study represents a preliminary stage of an extensive monitoring campaign, for which only winter in situ measurements were available. As mentioned before, the winter period was selected as it ensures stable indoor–outdoor temperature gradients, in line with standard recommendations, allowing accurate estimation of the roof system’s thermophysical parameters. While certain properties may vary slightly with temperature and moisture content, the calibrated parameters provide a reliable basis for year-round simulations. The use of TRNSYS in this context remains robust, as the model explicitly incorporates dynamic processes relevant to summer conditions, such as solar radiation, evapotranspiration, and thermal inertia, ensuring that seasonal performance is realistically reproduced despite the winter-based calibration.

Model calibration was assessed using two statistical indexes recommended by ASHRAE Guideline 14: the Mean Bias Error (MBE) and the Coefficient of Variation of the Root Mean Square Error (CV(RMSE)). The equations of these indexes are reported below:

$$MBE={\displaystyle \frac{{\sum }_i\left(m_i-s_i\right)}{{\sum }_i{m}_i}}(\times 100)$$

(2)

$$CV(RMSE)={\displaystyle \frac{\sqrt{{\sum }_i\frac{{\left(m_i-s_i\right)}^2}{N}}}{\overline{m}}}(\times 100)$$

(3)

where $m_i$ and $s_i$ are the measured and simulated values, respectively. $N$ is the number of data points acquired, and $\overline{m}$ is the mean value of the measured data. According to ASHRAE Guideline 14, a simulation model is calibrated when the MBE falls within ±10% and the CV(RMSE) falls within ±30% for hourly data. From the calibrated model, effective thermophysical properties (thermal conductivity, specific heat capacity, and mass density) can be derived.

Finally, the validated green roof model was implemented in a simplified building simulation to evaluate annual energy performance and comfort conditions. The building model was intentionally kept simple, as it directly represents the test rooms used for the experimental campaign. This approach ensures consistency between the numerical and experimental setups and allows us to isolate the effect of the roof configuration on the thermal response. In particular, the reference building consisted of a single-zone volume (6-by-6 m square floor plan structure), characterized by a height of 3 m, with envelope stratigraphy based on the typological data from the TABULA project [37]. In particular, the vertical walls are characterized by solid bricks plastered on both sides. The structural part of the roof is composed of hollow clay blocks and concrete. Finally, the basement is constructed with reinforced concrete. The stratigraphy is reported in

Table 1. Construction layers, listed from the inner to the outer side.

Component	Material	Thickness [m]	Thermal Conductivity [W/mK]	Specific Heat Capacity [J/kgK]	Mass Density [kg/m3]
Walls	Plaster	0.020	0.870	1000	1600
	Solid brick	0.200	0.800	840	1600
	Plaster	0.020	0.870	1000	1600
Roof	Plaster	0.015	0.870	1000	1600
	Hollow clay block	0.120	0.400	840	800
	Concrete	0.040	1.400	1000	2200
Basement	Concrete	0.300	1.400	1000	2200

, together with the thermophysical properties of each layer. These properties were selected to be representative of typical construction materials used in Mediterranean buildings and fall within the standard ranges reported in [38,39].

To explore different retrofit scenarios, the roof stratigraphy was modified by adding a thermal insulation layer of expanded polystyrene (EPS) with three thicknesses (0.02 m, 0.04 m, and 0.06 m), followed by an additional 0.08 m concrete layer on top. The selected EPS insulation thicknesses were chosen to represent typical low-to-moderate retrofit scenarios applicable to existing buildings with limited structural load capacity, in line with the design constraints of the ultra-lightweight green roof system. These values also reflect common practice in the Mediterranean context for cost-effective interventions where structural reinforcement is not feasible. Thicker insulation layers were not considered, as they would fall outside the scope of lightweight retrofit strategies investigated in this study. In particular, the EPS is characterized by a thermal conductivity equal to 0.045 W/mK, a specific heat capacity of 1300 J/kgK, and a mass density of 28 kg/m³. In this new configuration, the green roof system was installed on top of the added concrete layer.

In all cases, the modeled building features a total glazed surface area of 12 m², equally distributed across the four façades. The windows are equipped with single-pane glass, 6 mm thick, characterized by a thermal transmittance of 5.69 W/m²K.

The HVAC system simulation was carried out by setting the indoor temperature to 20 °C for heating and 26 °C for cooling, while maintaining a constant indoor relative humidity of 50%. Internal heat gains from occupants were implemented using the predefined TRNSYS profile ASHRAE_115W-Person_AIII_24C, assigned as absolute gains. According to the model output, each occupant contributed approximately 252 kJ/h (70 W), distributed as 151.2 kJ/h (42 W) radiative and 100.8 kJ/h (28W) convective heat, reflecting a metabolic activity level consistent with ASHRAE Class AIII. In the dynamic simulation, an air exchange rate of 0.5 1/h was assumed, representing typical natural ventilation conditions for residential buildings without mechanical ventilation systems [40].

The wind speed ($v$) values from the climatic files were used to compute the external convective coefficients for surfaces, setting the following equation in TRNSYS Build [41]:

$$h_c=4v+4$$

(4)

In addition to assessing the thermal performance and energy impact of the green roof system, this study aimed to evaluate its influence on indoor thermal comfort conditions. To this end, the TRNSYS Build model was configured to compute several comfort-related variables, including Mean Radiant Temperature, Operative Temperature, Predicted Mean Vote (PMV), and Predicted Percentage of Dissatisfied (PPD). Throughout the annual simulation period, the clothing factor (clo) varied according to a simplified seasonal schedule, reflecting typical occupant adaptation to outdoor climate conditions. A value of 1.2 clo was assigned during the coldest months (January, February, and December), 0.9 clo during transitional periods (March, April and October, November), and 0.5 clo during the warmest months (from May to September). This time-dependent profile was implemented to improve the accuracy of PMV and PPD calculations. In addition, a metabolic rate of 1.2 met was assumed, corresponding to light sedentary activity, and a typical indoor air velocity of 0.1 m/s was applied in accordance with ISO 7730 [42] and ASHRAE 55 [43] recommendations.

Simulations were performed using urban weather data acquired from a meteorological station located in the city center of Rome (see Figure 5a). The area is characterized by mid-rise historical buildings (typically 4–6 stories), narrow streets, and limited open spaces. In such a dense urban fabric, wind speed is typically reduced by building-induced sheltering, while long-wave radiation exchange is affected by the high level of sky-view obstruction. As highlighted in [26], this approach enables a more accurate assessment of building thermal behavior under realistic urban heat stress conditions (Figure 5b–d illustrate the comparison between the weather conditions observed in the city center and at Fiumicino Airport (FCO)). By analyzing multiple roof stratigraphy, the study investigates how varying insulation levels affect the building’s energy demand in a Mediterranean climate.

Figure 6 illustrates the methodological flowchart, detailing the sequential steps and data exchange between the experimental phase and the simulation process.

4. Results and Discussion

4.1. Experimental Results

The analysis of the data acquired during the monitoring campaign allowed for a clear assessment of the time evolution of the physical quantities involved in the determination of the U-value, following the average method outlined in ISO 9869-1 and reported in Equation (1). The experimental results focused on the most representative winter monitoring period, the week from 16–23 January 2025.

Figure 7a displays the indoor air temperatures recorded in the two test rooms, along with the outdoor air temperature. The indoor conditions remained stable throughout the monitoring period, indicating effective environmental control, while the outdoor temperature exhibited the expected diurnal fluctuations typical of Mediterranean winter conditions. A consistent thermal gradient between indoor and outdoor environments was maintained, satisfying a key requirement for in situ U-value estimation. As reported in the literature [44], a minimum temperature difference of 10 °C is recommended to ensure a reliable heat flux across the envelope (this condition was met for the majority of the monitoring period in both P1 and P2).

Figure 7b shows the evolution of external surface temperatures at the roof level. In room P2, the sensor was positioned below the vegetated layer, at the interface between the green roof system and the structural slab, thereby capturing the buffered thermal response of the system. In contrast, the traditional roof in P1, being directly exposed, exhibited strong diurnal fluctuations, with sharp daytime peaks and rapid cooling at night, indicative of low thermal inertia. The green roof in P2 showed a markedly more stable thermal profile, with reduced temperature swings and higher average values, particularly during the coldest hours, confirming its thermal damping capacity.

Figure 7c reports the internal surface temperatures of the roof assemblies. The green roof configuration consistently maintained higher surface temperatures and exhibited reduced fluctuations compared to the traditional roof. This behavior confirmed the thermal buffering effect of the green system, driven by the combined contribution of vegetation, substrate, and added mass. The result is an improved stability of internal conditions and enhanced thermal comfort, particularly relevant during winter in Mediterranean climates.

Finally, Figure 7d illustrates the time evolution of U-value, computed using the progressive average method. After an initial transient phase, values stabilized within approximately 33 h, fulfilling the convergence criteria defined by ISO 9869-1, with a variation within the range ±5%. The green roof achieved a U-value approximately 45% lower than that of the traditional tiled roof, confirming its superior insulating performance and overall contribution to improving the thermal characteristics of the building envelope. Specifically, the U-value was 2.897 ± 0.017 W/m²K for the reference roof and 1.588 ± 0.010 W/m²K for the green roof, highlighting the substantial improvement in thermal transmittance achieved by the latter. Corresponding numerical values and validation criteria are summarized in

Table 2. Average thermal transmittance values and convergence criteria.

U-Value	P1	Criterion Satisfied	P2	Criterion Satisfied
U (last value)	2.897	☑	1.588	☑
U (24 h before)	2.867	☑ (1.0%)	1.579	☑ (0.6%)
U (first2/3; last2/3)	2.957; 3.027	☑ (−2.3%)	1.654; 1.719	☑ (−3.8%)

.

To provide deeper insight into the thermal behavior and performance benefits of the green roof system, the data acquired over the entire month of January was analyzed. The comparison between the two configurations clearly highlights the green roof’s effectiveness in mitigating surface temperature fluctuations and enhancing indoor thermal comfort throughout the monitoring period.

In terms of heat flux, the green roof significantly reduced thermal energy transfer through the roof structure. In January, the average heat flux measured in P2 was approximately 32% lower than in P1, confirming the insulating capacity of the green stratigraphy. Moreover, the indoor air temperatures in P2 were consistently higher, with an average difference of +2.3 °C, indicating improved thermal stability. A similar trend was observed for internal surface temperatures. In P2, the average surface temperature was 3.79 °C higher than in P1, further supporting the green roof’s role in moderating indoor thermal conditions.

Overall, the empirical data demonstrate the green roof’s capacity to attenuate heat fluxes and stabilize interior environments during winter, validating its effectiveness as a passive solution for improving both energy performance and occupant comfort in Mediterranean climates.

4.2. Simulation Results

The experimental data served as the basis for calibrating a dynamic simulation model developed in TRNSYS 18 (see Figure 8). The building geometry and internal conditions were defined using TRNSYS Build, while the thermal behavior of the green roof system was represented through the dedicated Type 785 component. The model was thermally driven by applying measured boundary conditions, specifically the indoor and outdoor air temperatures recorded during the monitoring campaign. In TRNSYS, Type 9 is a standard component used to import time-dependent data from external files and supply them as dynamic inputs to the simulation model. In this study, experimentally measured indoor and outdoor air temperatures, acquired from 16–23 January 2025, were employed as thermal boundary conditions. The comparison between the simulated and measured roof surface temperatures (extrados) for room P2 enabled the calibration of the green roof model. This calibration process allowed for the identification of equivalent thermo-physical properties (thermal conductivity, specific heat capacity, and mass density) characterizing the green roof system.

The experimental data allowed the following values to be obtained: a thermal conductivity of 0.265 W/mK, a specific heat capacity of 1380 J/kgK, and a density of 575 kg/m³. The accuracy of the calibration procedure was confirmed by comparing the simulated and measured surface temperatures at the extrados of P2 (see Figure 9), as well as by the resulting values of the calibration indexes. The model achieved an MBE of 2.8% and a CV(RMSE) of 8.7%, both complying with the acceptable limits established by ASHRAE Guideline 14.

In this ultra-lightweight, soil-free green roof, seasonal variations in thermal behavior are primarily influenced by vegetation activity rather than substrate moisture content. During summer, active evapotranspiration from the vegetation canopy can enhance cooling and slightly reduce the thermal conductivity of the underlying layers due to lower retained moisture. In winter, vegetation enters a dormancy phase, evapotranspiration is minimal, and increased moisture retention within the drainage and filter layers may slightly increase effective thermal conductivity. These effects are implicitly accounted for in the TRNSYS simulations through the modeling of plant shading, evapotranspiration, and moisture-related thermal changes, although future monitoring covering both summer and winter would allow direct validation for this specific system.

As mentioned in the methodology section, the validated green roof model was implemented in a simplified building simulation to evaluate annual energy performance and comfort conditions. The dynamic simulations provided insights into the comparative energy performance of the ultra-lightweight green roof and the reference roof across different insulation levels (0 cm, 2 cm, 4 cm, and 6 cm). The results, summarized in Figure 10, highlight the seasonal impact of the green roof on both heating and cooling energy demand.

In the absence of traditional insulation (Figure 10a), the green roof demonstrated a clear thermal advantage during the winter season, reducing heating energy demand by 11.6% compared to the reference roof. This reduction reflects the insulating effect of the multi-layer green roof system, which enhances thermal resistance and attenuates heat loss.

However, as conventional insulation is introduced (Figure 10b), the difference between the two solutions becomes negligible, indicating that adding an additional insulating layer within the roof stratigraphy offers limited improvements in terms of winter energy performance. However, it is worth noting that expanded polystyrene (EPS) is a petroleum-derived material, which raises concerns regarding its environmental sustainability and life cycle impact.

On the other hand, Figure 10c,d illustrate the outcomes in terms of cooling energy demands, documenting a contrasting trend where the green roof consistently outperforms the reference roof across all insulation levels. Without insulation (0 cm), the green roof achieves a 58.5% reduction in cooling demand, confirming its superior ability to mitigate solar heat gains through evapotranspiration and shading effects. Even with increasing insulation thickness, the green roof maintains a cooling advantage, though with decreasing percentage variations: −7.3% at 2 cm, −3.9% at 4 cm, and −2.7% at 6 cm. Similar percentage differences can be obtained by comparing the green roof without additional EPS to the reference roof equipped with insulating material.

To provide a more comprehensive and comparable assessment of the results,

Table 3. Normalized heating energy demand for the green roof and reference roof configurations under different roof insulation thicknesses.

Heating	No Insulation [kWh/m2]	Insulation 2 cm [kWh/m2]	Insulation 4 cm [kWh/m2]	Insulation 6 cm [kWh/m2]
Green Roof	189.68	180.68	179.59	179.16
Reference Roof	214.50	180.93	179.62	179.15

and

Table 4. Normalized cooling energy demand for the green roof and reference roof configurations under different roof insulation thicknesses.

Cooling	No Insulation [kWh/m2]	Insulation 2 cm [kWh/m2]	Insulation 4 cm [kWh/m2]	Insulation 6 cm [kWh/m2]
Green Roof	16.55	16.52	16.54	16.55
Reference Roof	39.90	17.82	17.22	17.00

present the heating and cooling energy demands, normalized per unit roof area, for the green roof and reference roof configurations as a function of roof insulation thickness.

The calibrated model confirms that the thermal performance of green roofs is not only comparable to that of conventional insulation during winter but also superior in summer, thereby enhancing annual energy efficiency and indoor thermal comfort. The findings underline a seasonal complementarity between green roofs and conventional insulation. For poorly insulated buildings, the green roof offers dual benefits, reducing both heating and cooling demands. In well-insulated scenarios, its primary contribution shifts towards summer performance, where conventional materials cannot replicate the thermal inertia, evapotranspiration, and surface temperature damping provided by green roofs.

Table 5. Monthly differences between the green roof (GR) and reference roof (REF) in mean radiant temperature (ΔMRT), operative temperature (ΔTOP), predicted mean vote (ΔPMV), and percentage of dissatisfied occupants (ΔPPD), under the uninsulated scenario. Values calculated as GR–REF.

Month	ΔMRT [°C]	ΔTOP [°C]	ΔPMV [-]	ΔPPD [%]
January	0.64	0.32	0.06	−1.37
February	0.68	0.34	0.06	−1.61
March	0.12	0.02	0.00	−0.44
April	−0.40	−0.44	−0.10	1.30
May	−0.73	−0.66	−0.15	−1.17
June	−0.85	−0.57	−0.16	−0.94
July	−0.99	−0.52	−0.15	−2.94
August	−0.92	−0.50	−0.14	−2.32
September	−0.78	−0.75	−0.22	4.47
October	−0.29	−0.30	−0.07	1.63
November	0.39	0.20	0.03	−0.75
December	0.70	0.35	0.06	−1.56

summarizes the monthly differences between the green roof (GR) and the reference roof (REF) in terms of indoor thermal performance indicators, namely: Mean Radiant Temperature (ΔMRT), Operative Temperature (ΔTOP), Predicted Mean Vote (ΔPMV), and Predicted Percentage of Dissatisfied (ΔPPD), under the uninsulated roof scenario.

The results highlight the seasonal responsiveness of the ultra-lightweight green roof system in the Mediterranean climate of Rome. During the colder months (January through March, and November to December), the green roof configuration consistently outperforms the reference roof, exhibiting higher mean radiant temperatures, ranging from +0.12 °C in March to +0.68 °C in February, and increased operative temperatures, from +0.02 °C to +0.34 °C in the same period. The highest mean radiant temperature of +0.70 °C was found in December. These temperature increases correspond to improved thermal comfort conditions, as evidenced by slightly higher PMV values (up to +0.06 in December) and reductions in PPD ranging from −1.61% (February) to −0.44% (March). These results suggest that the green roof enhances indoor thermal comfort during the heating season, even in the absence of conventional insulation. This performance is likely due to the thermal buffering effect of the green roof layers, which reduce heat losses by dampening temperature fluctuations and delaying heat exchange with the outdoor environment.

In the shoulder season (April), the green roof system shows a slight performance penalty. Both mean radiant and operative temperatures are lower than the reference case (ΔMRT = −0.40 °C, ΔTOP = −0.44 °C), resulting in decreased thermal comfort (ΔPMV = −0.10) and an increased share of dissatisfied occupants (ΔPPD = +1.30%). This behavior can be attributed to reduced solar gains combined with limited thermal insulation, which may lead to a cooling effect not always desirable in spring.

As temperatures rise (May), the green roof begins to regain its thermal moderation role. ΔMRT and ΔTOP remain negative (−0.73 °C and −0.66 °C, respectively), but this contributes to a perceptible improvement in summer comfort conditions, reflected in reduced PPD (−1.17%) and lower PMV (−0.15). This shift anticipates the favorable performance typically observed in green roofs during the cooling season. This trend continues and intensifies during the peak summer months (June through August), where the green roof consistently reduces indoor thermal loads. In July, the system achieves a maximum reduction in mean radiant temperature of −0.99 °C and a ΔPMV of −0.15, with a significant drop in ΔPPD (−2.94%), indicating a meaningful improvement in perceived comfort. Similarly, operative temperature is reduced by −0.50 °C in August, while PPD reductions range from −0.94% (June) to nearly −3% (July), confirming the capacity of the green roof to attenuate thermal stress during heat-intensive periods. Similarly to April, in September, a small comfort penalty was observed (ΔPPD = +4.47%), likely due to reduced solar gains during a period with still mild outdoor temperatures.

Overall, the ultra-lightweight green roof demonstrated a favorable impact on indoor thermal conditions across most of the year, particularly during the winter season, where it effectively reduced discomfort despite the absence of additional insulation. While performance during transitional months requires careful consideration, the observed cooling potential in May suggests promising behavior during the summer, which will be further investigated. These findings reinforce the role of lightweight, soil-free green roof systems as effective passive solutions for climate-adaptive retrofitting in Mediterranean regions. Their implementation can contribute to improved indoor comfort, reduced energy demand, and enhanced building resilience in a warming climate, especially when tailored to the local seasonal thermal dynamics.

Table 6. Monthly differences in indoor thermal comfort parameters between the green roof (GR) and reference roofs (REF) with three levels of EPS insulation (2 cm, 4 cm, 6 cm). Values refer to differences in mean radiant temperature (ΔMRT), operative temperature (ΔTOP), predicted mean vote (ΔPMV), and predicted percentage of dissatisfied occupants (ΔPPD), calculated as GR–REF.

Insulation	2 cm	2 cm	2 cm	2 cm	4 cm	4 cm	4 cm	4 cm	6 cm	6 cm	6 cm	6 cm
Month	ΔMRT [°C]	ΔTOP [°C]	ΔPMV [-]	ΔPPD [%]	ΔMRT [°C]	ΔTOP [°C]	ΔPMV [-]	ΔPPD [%]	ΔMRT [°C]	ΔTOP [°C]	ΔPMV [-]	ΔPPD [%]
January	0.03	0.01	0.00	−0.05	0.01	0.01	0.00	−0.02	0.01	0.00	0.00	−0.01
February	0.03	0.01	0.00	−0.06	0.01	0.01	0.00	−0.03	0.01	0.00	0.00	−0.02
March	−0.01	−0.01	0.00	0.01	−0.01	0.00	0.00	0.01	−0.01	0.00	0.00	0.01
April	−0.04	−0.03	−0.01	0.19	−0.02	−0.02	0.00	0.11	−0.02	−0.01	0.00	0.07
May	−0.09	−0.09	−0.02	0.15	−0.05	−0.05	−0.01	0.08	−0.03	−0.03	−0.01	0.06
June	−0.11	−0.09	−0.03	0.42	−0.06	−0.05	−0.01	0.23	−0.04	−0.03	−0.01	0.16
July	−0.11	−0.08	−0.02	−0.03	−0.06	−0.04	−0.01	−0.01	−0.04	−0.03	−0.01	0.00
August	−0.11	−0.08	−0.02	0.02	−0.06	−0.04	−0.01	0.02	−0.04	−0.03	−0.01	0.01
September	−0.08	−0.07	−0.02	0.81	−0.04	−0.04	−0.01	0.43	−0.03	−0.03	−0.01	0.29
October	−0.03	−0.02	0.00	0.12	−0.01	−0.01	0.00	0.07	−0.01	−0.01	0.00	0.04
November	0.01	0.01	0.00	−0.02	0.00	0.00	0.00	−0.01	0.00	0.00	0.00	0.00
December	0.03	0.01	0.00	−0.06	0.01	0.01	0.00	−0.03	0.01	0.00	0.00	−0.02

compares the indoor thermal performance of the green roof with three flat roof scenarios featuring different levels of EPS insulation (2 cm, 4 cm, and 6 cm).

Across all indicators (ΔTMR, ΔTOP, ΔPMV, and ΔPPD), the differences between the green roof and the insulated reference roofs remain minimal and nearly constant regardless of the EPS thickness. This confirms that the green roof’s contribution to indoor thermal comfort is not significantly influenced by the underlying insulation layer, and that increasing insulation levels does not enhance or compromise its performance. These findings suggest that the thermal behavior of the ultra-lightweight green roof is largely independent of additional roof insulation, reinforcing its applicability as a stand-alone passive solution, especially in Mediterranean area retrofitting scenarios.

In summary, the green roof performs comparably to a lightly insulated reference roof (EPS 2 cm) during winter and outperforms it during summer in terms of cooling effect and thermal comfort. Compared to well-insulated roofs (EPS 4–6 cm), the performance gap narrows significantly in both heating and cooling seasons. These findings reinforce the green roof’s potential as a passive design strategy, particularly suitable for retrofits where insulation is limited or where summer overheating is a major concern.

5. Conclusions

This study investigated the thermal behavior and energy performance of an ultra-lightweight, soil-free green roof system specifically designed for the Mediterranean climate, through an integrated experimental and numerical approach. The work addressed a critical gap in existing literature, which often lacks combined empirical validation and dynamic simulation calibrated with real boundary conditions. By merging high-resolution in situ monitoring data, urban climate datasets, and a validated TRNSYS simulation model, this research offers insights into the effectiveness of green roofs as passive systems for building energy retrofitting through comparative analysis with traditional insulation strategies applied to existing buildings.

The experimental campaign, conducted during the winter season, revealed that the green roof reduced thermal transmittance by approximately 45% compared to a traditional tiled, non-insulated roof. Surface and indoor air temperatures were consistently higher in the green roof test room, confirming the system’s thermal buffering capacity and its contribution to improved comfort conditions in cold weather. These empirical findings provided the basis for calibrating a dynamic thermal model in TRNSYS 18, which reproduced the measured data with high accuracy, achieving an MBE of 2.8% and a CV(RMSE) of 8.7%, in full compliance with ASHRAE Guideline 14 standards.

Once validated, the calibrated model was used to simulate annual building energy performance and indoor thermal comfort for different roof configurations (characterized by different insulation levels). The simulations confirmed the seasonal effectiveness of the green roof. In winter, under uninsulated conditions, it reduced heating demand by 11.6% compared to the reference case. More significantly, in summer, it led to a cooling demand reduction of 58.5%, thanks to its combined shading, evapotranspiration, and thermal inertia effects. When compared to insulated reference roofs (EPS layers of 2 cm, 4 cm, and 6 cm), the green roof maintained comparable performance in winter and continued to offer an advantage in summer, with cooling energy savings ranging from −7.3% to −2.7%, depending on insulation thickness.

Thermal comfort analysis further supported these findings. During the colder months, the green roof ensured higher mean radiant and operative temperatures, improving perceived comfort even in the absence of conventional insulation. In the warm season, it significantly reduced indoor thermal stress, with operative temperatures reduced by up to 0.99 °C and PPD values lowered by nearly 3% in peak summer conditions. Notably, the performance of the green roof was found to be largely unaffected by the presence or thickness of EPS insulation layers. The comparative analysis demonstrated that adding more insulation did not significantly alter the indoor comfort outcomes associated with the green roof, highlighting its robustness and potential to function effectively as a standalone passive solution.

The ultra-lightweight green roof system presented in this study proved to be a climate-responsive, structurally compatible, and energetically effective solution for improving both winter and summer performance in Mediterranean buildings. Its capacity to offer thermal benefits without relying on petroleum-derived insulation materials reinforces its potential role in sustainable retrofit strategies.

The absence of summer in situ data represents a limitation of this preliminary stage of the work. Future developments will include summer monitoring, which will allow both the integration of warm-season boundary conditions into the calibration process and a direct assessment of the simulation model’s accuracy when only winter data are available.

Furthermore, future research should incorporate a life cycle assessment of the green roof compared to conventional insulation strategies to quantify long-term environmental impacts. Moreover, exploring synergies between green roofs and other passive systems (such as natural ventilation or solar control elements) could unlock further potential in climate-adaptive building design. Lastly, long-term evaluations of system durability, vegetation dynamics, and maintenance requirements will be essential to ensure reliable and scalable deployment across the Mediterranean building stock.