From the literature reviewed above, it emerges that, despite the growing number of studies, comprehensive energy and economic analyses of solar-based prosumers in Mediterranean climates remain scarce [15], [16], [19], [27], [28]. In particular, existing works have largely overlooked the valorisation of surplus PV electricity through its conversion into both heating and cooling energy within a thermal prosumer framework. Moreover, the economic viability of such configurations under different pricing schemes – accounting for both electricity and thermal energy markets – has not been systematically investigated. Most existing research has predominantly focused on heating-dominated conditions, resulting in operational and economic assessments that differ significantly from those encountered in Southern European contexts [11], [20], [23]. In contrast, as shown in Calise et al. [41] Mediterranean regions present specific challenges and opportunities, including high cooling loads during the summer season and abundant solar irradiance throughout the year. In addition, another part of literature [32]—[40] has investigated the benefits of integrating HPs with PV systems, in some cases also considering their integration with district networks via seasonal storages. However, these studies typically frame the prosumer role only from an electrical perspective, focusing on electricity exchange with the grid, while neglecting the potential of prosumers to interact with thermal networks.

In the transition to a decarbonized society, the combination of building-integrated PV systems and reversible HPs connected to DHCN could represent a promising solution for enhancing power system flexibility [42] and reducing RES curtailment [6]. Then, the valorisation of surplus PV electricity through its conversion into both heat and cooling and exchange with DCHNs, particularly under realistic operating conditions and market frameworks, warrants further investigation. Addressing these gaps requires integrated modelling approaches capable of simultaneously capturing building energy demand, dynamic HP performance, and high-resolution PV generation. Such approaches are still rarely applied in a unified manner within Mediterranean contexts, limiting the ability to fully assess the techno-economic potential of thermal prosumers in these regions [24].

To address these gaps, this study presents an integrated energy and economic analysis of a renewable-based thermal prosumer connected to a DHCN in a Mediterranean context. An office building located in Palermo, Southern Italy, equipped with a PV-driven air-to-water HP, is analyzed as a case study. The proposed methodology combines detailed building energy demand modelling carried out in TRaNsient SYstem Simulation (TRNSYS) software [43], dynamic HP performance characterization using IMST-ART, and a high-resolution PV generation model developed in Python. Annual simulations are carried out to assess system performance and economic profitability coming from the valorisation of surplus PV electricity - after meeting the building’s air conditioning and electrical loads - through its conversion into heat or cold via a dedicated HP and subsequent supply to the local DHCN. Specifically, the main novelties of this work can be summarized as follows:

• The study adopts an integrated multi-domain modelling framework that combines detailed building energy simulation, dynamic HP modelling, and high-resolution PV generation. This approach enables a consistent representation of coupled electrical and thermal energy flows in prosumer configurations and, differently from most previous works, explicitly captures the interaction between on-site PV generation, reversible HP operation in both heating and cooling modes, and bidirectional energy exchange with DHCNs.

• Specific attention is devoted to Mediterranean climate conditions, where high cooling demand and strong solar availability significantly influence system performance. Furthermore, the work provides a comparative assessment of alternative strategies for the valorisation of surplus PV electricity, namely direct grid export and conversion into thermal energy for supply to the district network, thereby offering a structured framework to analyse prosumer operation in sector-coupled energy systems.

• Finally, the economic feasibility of the proposed configuration is investigated under different electricity and thermal energy pricing schemes through dedicated sensitivity analyses, providing insights into the conditions required for the effective deployment of PV-driven thermal prosumers.

The remainder of the paper is structured as follows. Section 2 describes the case study and the modelling methodology. Section 3 presents the investigated scenarios and the main assumptions. Section 4 discusses the energy and economic results, while Section 5 summarizes the main conclusions.

The case-study office building follows the reference specifications provided by ASHRAE Standard 90.1–2016 [44]. The methodology adopted for the modeling and simulation of the building has been previously described in a publication by some of the authors [45]; however, the main assumptions are briefly recalled here to ensure the self-consistency of the present work. First, the building is assumed to be situated in Palermo, in Southern Italy, and is composed of three floors above ground, with a floor-to-floor height of 3.96 m. Each level is divided into a central core zone and four perimeter zones. The glazed areas amount to 43 m² on the north and south facades, and 28.63 m2 on the east and west facades. The thermal transmittance of opaque and transparent envelope elements is summarized in Table 1; values were adjusted to reflect Italian building practice in accordance with UNI/TR 11552:2014 [46]. Occupancy density, as well as lighting and equipment power densities, are also collected in Table 1. Operating schedules for occupants, lighting, and equipment are defined daily, distinguishing between weekdays and weekends, and are derived from the profiles specified in ASHRAE 90.1-2016 [44]. The building was simulated in TRNSYS [15], with Outdoor Air temperature (ODT) for Palermo retrieved from the Meteonorm Database. The resulting cooling and heating demands are shown in Figure 1.

Figure 1.

Heating and cooling demand for the office building [45]

The description and modeling of the hydronic plant were already presented in previously published work by some of the authors [45]. For the sake of self-consistency, only the main features are briefly recalled here, while the reader is referred to the cited paper for a more detailed description and discussion.

The air-conditioning plant consists of two hydraulic loops, namely a primary and a secondary circuit, and includes a reversible air-to-water HP, a thermal energy storage tank, and a circulation pump. The secondary water loop is responsible for distributing heated and chilled water throughout the building and consists of circulation pumps, mixing and diverting valves, distribution piping, and fan coil units serving each thermal zone. The entire thermal system -ranging from the heat HP to the terminal fan coil units, as well as the building envelope and the thermal zones was implemented in TRNSYS.

The nominal cooling and heating capacities of the air-to-water HP were respectively equal to 321 kWc and 390 kWth, based on the standard rating conditions [47]. Table 2 summarizes the technical data available from a commercial catalogue.

The HP modeling was developed following an approach previously validated by the authors [48]. The simulations were carried out using IMST-Art software (version 4.0) [49], whose reliability in describing vapor-compression systems has been widely documented in the literature [50]-[53]. The HP performance in cooling modes was described through eq. (1) and eq. (2), which were used to evaluate the delivered cooling capacity and the corresponding electrical power consumption [48]:

Table 3 collects all the fitting coefficients for eq. (1) and eq. (2), together with Root Mean Square Error (RMSE) and Normalized Root Mean Square Error (NRMSE). Similar equations were developed for the heating capacity and absorbed power. For the sake of brevity, only fitting coefficients are shown in Table 3.

The electrical demand of the building is divided into two components: mandatory office loads and HP electricity consumption. Figure 2 shows the weekly electrical demand profile associated with mandatory office loads, i.e., all electrical uses excluding space conditioning provided by the HP [44]. The demand exhibits a clear weekday-weekend pattern, with pronounced peaks during working hours from Monday to Friday, corresponding to typical office activities such as lighting, IT equipment, and auxiliary systems. During weekdays, peak demand reaches approximately 120 kW, while nighttime consumption remains limited to base-load levels. On Saturday, both the peak magnitude and duration are reduced, reflecting partial occupancy and limited operational schedules. In contrast, Sunday demand is almost constant and significantly lower, indicating minimal building usage.

Figure 2.

Weekly electrical demand profile associated with mandatory office loads [44]

Figure 3a and Figure 3b illustrate the electrical demand of the air-to-water HP (red solid curve) for two typical days of the year: one in winter (January 16^th) and one in summer (August 4^th). The stepwise pattern of the HP electricity demand reflects the sequential activation of the four compressors, with on-off cycling shown in Figure 3c and Figure 3d. Looking at Figure 3c, during the heating season, compressors’ operation is driven by the temperature of the upper node of the thermal energy storage. A temperature setpoint of 50 °C with a ±2.5 °C deadband is adopted: as the storage temperature decreases, compressors are progressively activated, while they are sequentially switched off when the temperature rises above the upper threshold. Figure 3b and Figure 3.d show the corresponding results for a summer operating condition. In this case, the monitored variable is the temperature at the lower node of the storage, representative of the cooling operation, with a setpoint of 7 °C and a ±1 °C deadband. Similar to the winter case, the compressor activation sequence follows the deviation of the storage temperature from the setpoint, resulting in a modulated electrical demand of the HP. Overall, the figures highlight the strong coupling between storage temperature control, compressor sequencing, and electrical demand profiles, as well as the distinct operational behavior of the system under heating and cooling conditions.

The simulations showed that the total annual electricity consumption of the building is equal to 408,194.10 kWh/y. Of this amount, 353,439.99 kWh/y (86.59%) are related to mandatory electricity uses, such as lighting and equipment in the thermal zones, while the remaining 54,754.11 kWh/y (13.41%) are required for the operation of the HP.

Figure 3.

HP electricity demand and control behaviour for two representative days: winter operation (January 16th) and summer operation (August 4th).(a) Winter: HP electrical demand and mandatory office loads. (b) Summer: HP electrical demand and mandatory office loads. (c) Winter: compressor staging (four compressors) and thermal storage upper-node temperature with a 50 °C setpoint and ±2.5 °C deadband. (d) Summer: compressor staging and thermal storage lower-node temperature with a 7 °C setpoint and ±1 °C deadband

To assess the feasibility and optimal sizing of a rooftop PV system for the non-residential building under study, a dedicated Python-based simulation and optimization workflow was developed. The proposed methodology integrates geometric constraints of the available roof area, component specifications, meteorological inputs, and electrical load profiles to identify the PV configuration that maximizes on-site energy utilization and minimizes grid dependence. To better understand the procedure, the overall simulation and optimization workflow is schematically summarized in the block diagram reported in Figure 4, illustrating the interaction between geometric layout generation, PV performance modelling, load matching, and optimization loop.

Figure 4.

Description of the optimization routine for sizing of the PV plant

The flat rooftop was parameterized according to its effective geometric dimensions (49.92 m east-west, 33.29 m north-south), accounting for mandatory setbacks, maintenance corridors, and row spacing constraints. The PV layout was discretized by defining two decision variables: the number of module rows N_r (north-south direction), and the number of modules per row N_m (east-west direction). For each feasible pair (N_r, N_m), the total number of installed modules and the corresponding DC peak power were computed based on the selected commercial PV module (550 W_p, 2.10 m × 1.13 m).

Row spacing (pitch) was determined to prevent inter-row shading at solar noon during the winter solstice, adopting a conservative geometric criterion. For each feasible layout (N_r, N_m), the total number of installed modules, the corresponding DC peak power, and the ground coverage ratio were calculated. Configurations violating geometric or electrical constraints were discarded a priori. For each admissible configuration, inverter sizing was automatically performed to match the DC capacity while respecting practical DC/AC ratio limits, selecting commercially available inverter units, and aggregating their nominal AC power.

PV energy production was simulated using the open-source pvlib Python library. Consistent with the building simulation, meteorological inputs were derived from the PVGIS Typical Meteorological Year (TMY) dataset for Palermo [16], providing hourly values of global horizontal irradiance (GHI), direct normal irradiance (DNI), diffuse horizontal irradiance (DHI), ambient temperature, and wind speed. Plane-of-array irradiance was computed for each configuration, and module cell temperature was estimated using the Sandia Array Performance Model (SAPM), considering an open-rack glass-glass module. DC power production was calculated through the PVWatts performance model, accounting for irradiance and temperature effects, while AC power output was obtained by applying inverter efficiency and clipping effects [54]. The resulting PV AC power time series was generated at hourly resolution and aligned with the building electrical load profile.

For each candidate configuration, the hourly energy balance was evaluated as:

where P_load(t) represents the electrical load of the building and P_PV(t) is the PV system's AC power output at time t. The annual electricity imported (E_import) from the grid was computed as:

where T denotes the total number of hours in the year and ∆t = 1 h. While the annual electricity surplus that is exported to the grid (E_export) was computed as:

Based on the hourly energy balance, a set of energy-related key performance indicators (KPIs) was computed. The annual PV energy production (E_PV) and the annual building electrical demand (E_load) were obtained by integrating the corresponding hourly profile. The amount of self-consumed PV electricity (E_self) was defined as:

From these quantities, the self-consumption ratio (SCR) and the load coverage factor (LCF) were calculated as follows:

The PV system sizing problem was formulated as a discrete optimization over the set of geometrically feasible layouts, to minimize the annual electricity imported from the grid (i.e. min E_import)^.

This criterion was selected to directly enhance on-site utilization of PV generation and reduce dependence on external electricity supply, which is particularly relevant for non-residential buildings characterized by significant daytime electrical loads. The remaining KPIs were used to characterize the performance of the optimal configuration, but did not directly enter the objective function.

The energy KPIs associated with the optimal PV configuration were subsequently used as input for the economic assessment of the integrated PV-HP-DHCNs. Annual revenues were computed by distinguishing between self-consumed electricity, valued as avoided electricity purchase, surplus electricity exported to the grid, and surplus electricity converted into thermal energy through the air-to-water HP and sold to the district heating and cooling network. The annual cash flow in the y-th year was defined as follows:

where R_y represents total annual revenues, OPEX_y the operating expenditures, and REINV_y possible reinvestment costs.

The Net Present Value (NPV) over the system lifetime (N) was calculated as:

where r is the discount rate. The Internal Rate of Return (IRR) was obtained as the value of r satisfying:

Finally, discounted payback time was evaluated based on the cumulative nominal and discounted cash flows.

Figure 5 provides a schematic layout of the PV system installed on the roof of the office building, highlighting the usable area and access aisles. The PV plant supplies electricity to a dedicated air-to-water HP, which converts surplus PV power into heat or cold for on-site use and for export to the DHCN through a heat exchanger.

Figure 5.

Detailed energy flow exchanges between the PV plant and the HP used to provide heat & cold to a local DHCN

Two scenarios are analysed:

• Electricity Surplus to the grid: all surplus PV electricity is sold to the electrical grid at a given selling price.

• Thermal prosumer: surplus PV electricity is supplied to a “dedicated and newly installed” air-to-water HP, which produces hot water at 40 – 45 °C during the heating season (December – March) and chilled water at 6 °C during the cooling season (May– September), with the generated thermal energy exported to the DHCN. As shown in eqs. (12) to (13), the Coefficient of performance (COP) and the Energy Efficiency Ratio (EER) of the HP were parameterized using quadratic correlations derived from the performance of representative medium-size commercial units with nominal capacity between 250 – 300 kW, modelled by IMST-Art, accounting not only for the effect of ODT but also for the progressive reduction of thermal demand from the DHCN at milder ODTs, which influences the effective operating conditions and part-load efficiency of the machine.

The economic analysis of the proposed PV system was carried out over a lifetime horizon of 25 years, assuming a weighted average cost of capital (WACC) of 7% as the discount rate. The initial investment cost (CAPEX) was estimated at 1,400 EUR/kWp, inclusive of modules, inverters, installation, and balance-of-system, while annual operating expenditures (OPEX) were considered equal to 4% of CAPEX. A reinvestment equal to 10% of CAPEX was included in year 12 to account for inverter replacement or other revamping operations. Energy production was corrected by an annual performance degradation of 0.5%. The economic valorization of electricity was distinguished between self-consumed energy, valued at 0.15 EUR/kWh (avoided cost of electricity purchase), and surplus electricity exported to the grid, valued at 0.11 EUR/kWh. The latter reflects the average remuneration, which represents typical market schemes in the Italian regulatory context. For the case of the air-to-water HP, a CAPEX of about 250 EUR/kWc was assumed [55]. Furthermore, it was assumed a selling price of heat/cold of about 0.05 EUR/kWh.

All tariffs were assumed to grow annually at a nominal rate, compounding an inflation rate of 2% and a real escalation factor of 1.5%, resulting in an effective annual growth of approximately 3.5%. OPEX was assumed to escalate at the inflation rate only (2% per year). No fiscal incentives or tax effects were applied to maintain the analysis conservative and technology-focused. On this basis, yearly revenues, OPEX, and cash flows were computed, and the main financial indicators were derived, including Net Present Value (NPV), Internal Rate of Return (IRR), and Discounted Payback Time (DPBT).

Two sensitivity analyses were carried out to evaluate the robustness of the results with respect to key tariff parameters. The first analysis focused on the effect of the heat and cold selling price, varying it from the reference value of 0.05 EUR/kWh to 0.08 EUR/kWh and 0.11 EUR/kWh, to reflect different market and regulatory conditions. The second analysis investigated the impact of electricity price variability on both scenarios, by varying the self-consumption tariff – assumed equal to the avoided cost of grid electricity purchase -between 0.09 EUR/kWh (−40% relative to the baseline of 0.15 EUR/kWh) and 0.21 EUR/kWh (+40%), thus accounting for the significant volatility observed in electricity markets in recent years.

Figure 6 illustrates the electrical energy flows among the PV system, the HP, and the DHCN for two representative operating days: a winter day (January 16^th, Figure 6a) and a summer day (August 4^th, Figure 6b). During the winter day, PV electricity production is limited and mainly concentrated around midday, covering only a small fraction of the HP electrical demand (below 20%), which is therefore largely supplied by electricity imported from the grid during heating operation. Conversely, during the summer day, higher solar irradiance leads to a significant increase in PV generation, allowing more than 60 – 70% of the HP electrical demand to be directly supplied by on-site PV production during peak hours. As a result, a substantial PV surplus is observed, especially around midday, which can be exported to the grid or indirectly exploited through cooling production and interaction with the DHCN. Overall, the comparison highlights the strong seasonal variability of PV-driven prosumer operation and the role of HP and DHCN in enhancing PV self-consumption and system flexibility.

Figure 6.

Hourly electrical power flows among the PV system, HP, and grid/DHCN interface for two representative days: winter day (January 16th) (a) and summer day (August 4th) (b)

Energy results for one year of operation are presented in Figure 7a and Figure 7b. Figure 7a shows the hourly electricity surplus, highlighting a clear seasonal trend, with higher values during summer and peak values of approximately 185 kW. On an annual basis, the system achieved a PV generation of 363,109 kWh, covering 53.9% (LCF) of the building’s annual electricity demand (408,193 kWh). Of the PV generation, 220,239 kWh were self-consumed on-site (self-consumption ratio SCR = 60.7%), while 142,869 kWh were exported to the grid. Consequently, annual grid imports were reduced to 187,954 kWh.

Figure 7.

PV surplus and annual energy balance for the optimal PV layout. Hourly profile of PV surplus power over the year (a). Annual PV energy balance showing self-consumed electricity, exported electricity, and resulting grid imports (b)

Figure 8a shows the power flows in the thermal prosumer scenario, including HP heating and cooling output exported to the DHCN, and surplus PV power, which is exported to the grid when the HP connected to the DHCN is switched off. During the winter and mid-season periods, the system mainly operates in heating mode, with HP thermal power peaks reaching 450 kW. In these periods, PV surplus power is limited and occurs primarily during daytime hours in spring and autumn, with exported power typically below 150 kW. Conversely, during the summer, cooling operation dominates, with the HP reaching peak values of approximately 476 kW during high cooling demand conditions. In this period, the higher availability of solar generation leads to increased PV surplus, which can be either exported to the grid when the HP is inactive or indirectly valorized through cooling production and interaction with DHCN.

Figure 8.

Thermal prosumer operation: (a) hourly profiles of PV surplus electricity, and corresponding heat and cooling outputs delivered by the dedicated HP to the district heating and cooling network; residual electricity export to the grid when the dedicated HP is off; and (b) monthly and yearly energy results

Figure 8b shows monthly and yearly results. The system generates 47.03 MWh/y of heat and 168.52 MWh/y of cooling. This is mainly driven by the higher availability of PV electricity during the summer season compared to winter, resulting in a greater production of chilled water supplied to the DHCN, which is particularly beneficial in cooling-dominated climates. Meanwhile, the surplus electricity exported to the grid amounts to 72.19 MWh/y.

Figure 9 compares the economic performance of the “Electricity surplus to the grid” and “Thermal prosumer” scenarios in terms of Net Present Value, Discounted Payback Time, and Internal Rate of Return.

Results show that the electricity surplus to the grid scenario achieves the highest economic performance, with an NPV of approximately 350 kEUR, a DPBT of about 9 years, and an IRR of 16.21%. This outcome is mainly driven by lower system complexity (absence of the additional HP and connection to the DHCN) and the direct monetization of PV surplus electricity through grid export. In contrast, the thermal prosumer scenario yields a lower but still attractive profitability, with an NPV of around 240 kEUR, a DPBT of approximately 13 – 14 years, and an IRR of 11.90%. The reduction in economic indicators is associated with higher initial investment and operational complexity related to the HP operation and thermal interaction with the DHCN. However, this scenario enables the provision of flexible thermal services to the district network and power grid, which are not fully captured by purely economic metrics. Overall, the comparison highlights a trade-off between short-term economic profitability and system-level benefits, suggesting that the thermal prosumer configuration may become increasingly attractive under future market conditions with higher electricity prices, increased remuneration of flexibility services, or stronger incentives for sector coupling.

Figure 9.

Economic performance comparison between the electricity export and thermal prosumer configurations in terms of NPV, DPBT, and IRR

The sensitivity of the main economic KPIs (NPV, IRR, and DPBT) to two key tariff parameters is reported in Table 4 and Table 5, which examine the effect of the heat and cold selling price and the self-consumption electricity tariff, both for the electricity surplus to the grid and the thermal prosumers scenarios.

Regarding the heat and cold selling price, results confirm that this parameter exclusively affects the thermal prosumer configuration, as no dedicated HP is operated in the electricity surplus to the grid scenario, whose KPIs therefore remain fixed at the reference values (NPV = 346 kEUR, IRR = 16.2%, DPBT = 8.8 years) regardless of the assumed thermal tariff. The selling price of heat/cold was varied from 0.05EUR/kWh (reference case) up to 0.11 EUR/kWh. As shown in Table 4, for the thermal prosumer, increasing the selling price from 0.05 EUR/kWh (reference case) to 0.11 EUR/kWh raises the NPV from 239.78 kEUR to 438.63 kEUR and the IRR from 11.9% to 15.5%. At the same time, the DPBT decreases from 13.4 years to 9.4 years, confirming the strong sensitivity of this scenario to thermal energy remuneration.

On the other hand, regarding the self-consumption electricity tariff, both scenarios are significantly affected, as self-consumed PV electricity represents the dominant revenue component in both configurations, amounting to 220,239 kWh/y. As shown in Table 5, varying the tariff between 0.09 EUR/kWh (−40% relative to the baseline of 0.15 EUR/kWh) and 0.21 EUR/kWh (+40%) produces a wide range of outcomes. For the electricity surplus to the grid scenario, the NPV spans from 142.8 kEUR to 549.1 kEUR, with the IRR ranging between 11.1% and 20.9% and the DPBT between 6.5 and 14.6 years. The thermal prosumer scenario exhibits a similar directional response, with NPV ranging from 36.6 kEUR to 442.9 kEUR, IRR from 7.8% to 15.5%, and DPBT from 9.3 to 22.1 years. Notably, at the lowest tariff level (0.09 EUR/kWh), the thermal prosumer IRR of 7.8% falls marginally above the assumed WACC of 7%, indicating that economic viability is maintained but with limited margin. This result underscores that the thermal prosumer configuration is more exposed to unfavourable electricity market conditions, given the combined effect of higher investment cost and lower revenue diversification at reduced tariff levels. Overall, the self-consumption electricity tariff emerges as the most influential economic driver for both scenarios, reinforcing the importance of stable and favourable regulatory frameworks for distributed PV generation in the Mediterranean context.

Despite the comprehensive energy and economic assessment presented in this work, some limitations should be acknowledged.

• First, the analysis is based on a single non-residential case study located in a Mediterranean context. While representative of Southern European conditions, results may differ for other building typologies, occupancy profiles, or DHCN configurations. Future studies should extend the analysis to residential and mixed-use buildings and to districts with multiple interacting prosumers.

• Second, the economic evaluation relies on assumed electricity, heat, and cold prices, reflecting current market conditions. Although a sensitivity analysis was conducted on heat and cold selling prices, future electricity market dynamics, policy incentives, and regulatory mechanisms were not explicitly modelled. Further research should investigate dynamic pricing schemes, flexibility markets, and alternative business models for PV-HP-based prosumers in DHCNs.

• Third, the substation and DHCN interaction were modelled in a simplified manner, focusing on aggregated energy flows. Detailed thermo-hydraulic behavior, heat exchanger performance, pressure drops, and advanced control logics were not explicitly represented. Future work should integrate detailed substation and network models to assess hydraulic constraints, temperature management, and coordinated operation among multiple prosumers. Moreover, capital expenditures related to the substation should be included to enable a comprehensive techno-economic evaluation.

• Finally, the HP operation was governed by rule-based control strategies. Advanced control approaches, such as model predictive control or AI-based energy management, could further enhance PV self-consumption, reduce peak loads, and improve both economic and system-level performance.