**Low Enthalpy Geothermal Systems in Structural Controlled Areas: A Sustainability Analysis of Geothermal Resource for Heating Plant (The Mondragone Case in Southern Appennines, Italy)**

#### **Marina Iorio 1, Alberto Carotenuto 2, Alfonso Corniello 3, Simona Di Fraia 2,\*, Nicola Massarotti 2, Alessandro Mauro 2, Renato Somma 4,5 and Laura Vanoli <sup>2</sup>**


Received: 16 January 2020; Accepted: 5 March 2020; Published: 6 March 2020

**Abstract:** In this study, the sustainability of low-temperature geothermal field exploitation in a carbonate reservoir near Mondragone (CE), Southern Italy, is analyzed. The Mondragone geothermal field has been extensively studied through the research project VIGOR (Valutazione del potenzIale Geotermico delle RegiOni della convergenza). From seismic, geo-electric, hydro-chemical and groundwater data, obtained through the experimental campaigns carried out, physiochemical features of the aquifers and characteristics of the reservoir have been determined. Within this project, a well-doublet open-loop district heating plant has been designed to feed two public schools in Mondragone town. The sustainability of this geothermal application is analyzed in this study. A new exploration well (about 300 m deep) is considered to obtain further stratigraphic and structural information about the reservoir. Using the derived hydrogeological model of the area, a numerical analysis of geothermal exploitation was carried out to assess the thermal perturbation of the reservoir and the sustainability of its exploitation. The effect of extraction and reinjection of fluids on the reservoir was evaluated for 60 years of the plant activity. The results are fundamental to develop a sustainable geothermal heat plant and represent a real case study for the exploitation of similar carbonate reservoir geothermal resources.

**Keywords:** carbonate geothermal reservoirs; sustainable geothermal energy exploitation; southern Italy; numerical simulation

#### **1. Introduction**

The total (electric and thermal) potential energy of a geothermal reservoir depends on many variables such as thermal regime, geological structure, and hydrogeological properties. Recently, the use of groundwater, extracted via open-loop systems, is increasingly being considered as an efficient means for heating and cooling of buildings, especially in the heat pump systems.

This research concerns the geothermal carbonate reservoir, which makes the bedrock in the alluvial Mondragone plain, Southern Italy. The site has been extensively studied through the VIGOR project (coordinated by the Italian National Council Research and sponsored by the Italian Ministry of

Economic Development) dedicated to the evaluation of geothermal potential in four Italian Regions: Puglia, Calabria, Campania, and Sicily. In this plain, a well-doublet open-loop plant has been designed for the district heating of seven public schools of Mondragone town [1].

Low-temperature geothermal fields have been exploited over decades for industrial applications in Iceland, Hungary, China, Turkey, France, Germany, Russia, and other countries [2], and in recent times many authors propose the use of geothermal energy standalone or coupled with other renewable energy sources [3–5]. As an example, a classical low-temperature geothermal field of meteoric origin, in Kamchatka, Russia, has extracted thermal water since 1966 mainly in the mode of artesian flow to supply numerous swimming pools, district heating of two villages, greenhouse farming, and fish breeding [6]. The long-term exploitation (20 years) has been assessed finding an insignificant temperature decrease in the geothermal reservoir (0.5 ◦C) [2]. The geothermal power potential in a hot spring close to the Municipality of Isa, in Japan, has been analyzed through the gravity data estimating a power density of 30.4 kWe/km2 for a 20 year exploitation [7]. Another interesting investigation has been carried out in a volcanic research site in the North German Basin, where a well-doublet with one injection and one production well has been explored for future geothermal energy production [8]. Alternating the injection of low and high flow rates of water a rapid increase in the water level of an adjacent well has been observed as well as an increase of the overall productivity of the treated well [9].

With the exception of active volcanic areas, carbonate aquifers are the most important geothermal reservoirs in which low-to-medium temperatures are usually reached at significant depths from the ground level (>1000 m) (e.g., [10–16]).

In many countries (i.e., Southern Italy), fault-controlled carbonate extensional domains may present temperatures in the range of low or middle enthalpy (*T* < 150 ◦C), at relatively shallow depths (<0.5 km) [14,17–21]. In fact, fault-controlled geothermal sites are accompanied by hydrothermal activity, hot fluid circulation, and mineralization processes [22,23]. Fracturing and chemical dissolution of carbonates increase the permeability of rocks, enhancing the advection of hot fluids and generating heat and mass transport [24–28]. The setting of carbonate bedrock in Mondragone plain corresponds to this hydrogeological model.

In recent times, the installation of innovative pilot geothermal electric and heating power plants has been incentivized in Italy. In this framework, prior to drilling activities and plant design, knowing scientific-technical data related to the potential of geothermal reservoirs and the sustainability of the utilization represents a crucial task to assess the economic feasibility of the geothermal exploitation and the development of future project plants [29,30].

Other information can be obtained in such a way, such as the relevance of system boundary conditions during long-term utilization, the interference of fluids extraction and reinjection during production, and the effectiveness of geothermal production systems. Numerical simulation is recognized as a fundamental tool for the elaboration and assessment of using geothermal energy [31], which is otherwise considered to be highly risky [32–34].

In recent years, different multiphase simulation tools have been widely used to model the effects induced by the exploitation of geothermal resources [35–37]. Among them, the code TOUGH2®(Transport Of Unsaturated Groundwater and Heat) has been firstly applied to Wairakei (New Zealand) geothermal field [38–40] and subsequently to several other geothermal fields (i.e., [41,42]). Additionally, the Los Alamos National Laboratory Finite Element Heat and Mass Transfer (FEHM®) code has a good record of numerical modeling studies, mainly related to enhanced geothermal systems (EGS) [43–46], together with the U.S. Geological Survey code HYDROTHERM® [47–49]. The commercial software COMSOL®, in particular, is a widely used simulator that has been adapted (as for example with the link to MATLAB®) [50] to applications including geothermal studies and, in several cases, the evaluation of fault influence [23,51,52]. As the heat and mass transfer in the reservoir depends on the internal properties of the reservoir and of the surrounding rocks, there has been an increased effort to model geothermal porous reservoirs. However up to now, few sustainability exploitation analyses on fractured carbonate rocks have been done [53,54] and they mainly focus on

middle-high enthalpy and deep geothermal reservoirs, establishing several dynamic relations between the properties of the equivalent porous medium and fracture aperture. The exploitation of a low enthalpy geothermal system has been proposed by Cherubini et al. [55], who simulated a fractured geological system, composed of a single homogeneous layer with an inclined fault. They found that the width and permeability of the fault significantly affect fluid flow and thermal field, with a concentration of thermal perturbation within the fault plane.

#### **2. Research Object Description**

In the present paper, a sustainability analysis of the exploitation, by means of a well-doublet open-loop plant, of a low-temperature reservoir consisting of tectonically fractured and karstic limestone is carried out. This aquifer hosts an artesian groundwater body with a pressure larger than 1.70 bars, a wellhead temperature of 14 ◦C, and a natural water flow of 23.5 L/s. In the present work, two of the seven schools analyzed within the VIGOR Project are considered to be supplied. The characteristics of the system proposed to feed the two public schools are reported in Table 1.

**Table 1.** The proposed system for the feeding of two public schools.


A water flow of 6.00 out of 23.5 L/s was computed as sufficient to heat the considered schools (about 160 kW). The remainder is supposed to be by-passed and then mixed again to reinject the entire flow rate through a reinjection well [56]. A scheme describing the operation of the system is shown in Figure 1.

**Figure 1.** Scheme of the system operation.

The sustainability of such geothermal exploitation has been analyzed by a numerical simulation, comprising a one fault case and two faults case, using the commercial software COMSOL Multiphysics®to assess the reservoir thermal perturbation (for 60 years of system operation for heating needs) due to extraction and reinjection of fluids at a rate suitable to produce the required thermal power.

Due to low enthalpy values and high water pressure effects on the reservoir, mechanical deformations have been considered negligible. A further element of interest in the modeling is due to the position of the wells (of the extraction/injection) in the plant. In fact, for logistical constraints, the injection well is not downstream of the extraction well, but almost flanked by this with respect to the flow direction of the groundwater body. The numerical results obtained allow development of a solid conceptual model for the Mondragone geothermal reservoir exploitation and open the possibility to future applications to similar geothermal reservoirs which are widespread in the Mediterranean region.

#### **3. Hydrogeological Setting and Conceptual Model of the Geothermal Area**

The area of interest is located on the coastal plain of Mondragone town at the bottom of Mt. Petrino carbonate hill (Figure 2).

**Figure 2.** Localization and hydrogeological structural setting of the analyzed site. (Adapted from [1]).

This is at SW of Mt. Massico, which is a carbonate monocline trending SW bounded by high-angle normal faults along the margins of the Garigliano (NW), Volturno (SE), and Mondragone (SW) plains. These faults, several kilometers deep, have a throw estimated to be hundreds of meters [57–59]. Recent geophysical data have provided information about the bedrock of Mondragone plain. It is made by limestone and marly-arenaceous rocks and affected by several faults. The most important fault of this area crosses the whole plain from NW to SE and it is here coupled to other sub-parallel faults. The carbonate bedrock near Mt. Petrino has been found at a depth of –30 m below ground level

(b.g.l.), under pyroclastic-alluvial deposits. From this zone, moving towards SE, the bedrock shows instead a regular depth towards the sea [1] (Figure 2).

Boreholes' data in the plain have shown the presence of a powerful bank of Campanian Ignimbrite tuff almost continuous across the plain. This tuff has low permeability and therefore separates the pyroclastic-alluvial aquifer (very poor) above the tuff, from the underlain alluvial deposits, deeper and more permeable, generating confined conditions in this lower aquifer. This last aquifer receives groundwater from lateral and vertical flows from the carbonate hill of Mt. Petrino, which is not hydro-geologically connected with the Mt. Massico, and from the carbonate bedrock, both shallow and deeper groundwater flow from NE to SW, toward the sea [22,60] (Figure 2).

Groundwater below the tuff presents interesting chemical characteristics: temperatures and average electrical conductivities are above 20 ◦C and >20,000 μS/cm, respectively, while, in the neighboring areas, these parameters show lower values (i.e., conductivity around 1000 μS/cm). These anomalies can be explained by the rise of deep mineralized waters (rich in gas, above all CO2 of inorganic origin [61]) along the faults of the carbonate bedrock. These deep waters mix with groundwater below the tuff and increase their total dissolved solids and temperature, features that are distributed along the groundwater flow direction [22].

Additional data about the mineralized zone is derived from drilling, in the frame of VIGOR project, of a geothermal exploration well, 300 m deep, about 150 m from the base of Mt. Petrino (Figure 2). The first 167 meters of the well stratigraphy (Table 2) highlight that limestone starts under pyroclastic-alluvial deposits at about 30 m b.g.l., and that fracturing of crossed carbonate rocks reaches its maximum around 100–120 m b.g.l. (Table 3).

Table 2 reports the geothermal stratigraphy and the geophysical properties of the layers composing the analyzed domain. Density, thermal conductivity, and porosity were derived from literature, whereas the permeability was determined through "Lugeon tests" performed at different depths in the exploration well.

The experimental campaign has revealed the presence of artesian groundwater (under a maximum pressure of 1.70 bars) in the carbonate rocks, from which, in absence of confinement, about 23.5 L/s naturally flow [1]. After reaching the depth of 300 m, inside the well (and with a temporary casing) geophysical logs were performed (using a probe 2PFA-1000 / MATRIX converter) to measure temperature (Table 3) and verticality. During drilling, groundwater samples have been collected at different depths to analyze changes in chemical composition. The gases have been also sampled at two different sites, in correspondence of the geothermal exploration well and in its vicinity [60]. It was found that the chemical characteristics of groundwater along the depth are similar; the gases sampled are 98% CO2, whose average concentration is 1380 mg/L [1,60].

Combining all data, a geothermal model that controls the mineralization of groundwater at the base of Mt. Petrino can be proposed [1,10,22,26,60,61]. According to this model, the geothermal reservoir corresponds to the carbonate bedrock of the Mondragone plain. Near Mt. Petrino, deep hot gases (mainly CO2) rise along the faults of the bedrock, involving groundwater (sodium-bicarbonate type) typical of a reducing environment and of meteoric origin. From the carbonate bedrock, thermo-mineral groundwater spreads in the overlying pyroclastic-alluvial aquifer diluting gradually. Therefore, the geothermal process is closely linked to the faults, as shown by temperature increase, which occurs, as measured continuously along the well (Table 3), at the same levels of major fracturing degree. Therefore, it is very likely that these levels represent the preferential path along which the hot fluid rise occurs [60,61].




**Table 3.** Temperatures and lithology of the geothermal exploration well at different depths.

#### **4. Numerical Simulation of Geothermal Exploitation**

In order to investigate the sustainability of geothermal exploitation, a numerical model was developed to analyze the coupled multiphase thermo-hydraulic processes due to the extraction and reinjection of groundwater. The model was implemented within the Finite Element commercial software COMSOL Multiphysics®, which is a powerful tool to reproduce coupled or multiphysics phenomena.

#### *4.1. Governing Equations*

Geothermal energy exploitation involves the interaction of different physical phenomena occurring in the ground, such as fluid flow, heat transport, chemical transport, and mechanical deformation [32,34,62]. In this work, only heat and fluid flow in fully saturated porous media were analyzed, considering the conservation laws of mass, momentum, and energy in a porous medium, that represents the ground.

The conservation of mass for a fluid flowing through a porous medium is expressed as:

$$\frac{\partial}{\partial \mathbf{t}} (\rho \boldsymbol{\varepsilon}) + \nabla \cdot (\rho \mathbf{u}) = \, \mathbf{0}, \tag{1}$$

where ρ is the fluid density, ε is the porosity, **u** is the seepage (Darcy) velocity vector. Therefore, the terms ρε and ρ**u** represent the mass per unit volume within the porous matrix and the fluid mass flux, respectively. The flow velocity is low due to the low values of permeability and porosity of the domain. In this study, the porous medium flow was analyzed by using the well-known Darcy equation, which is the most common approach in geo-mechanic [63]. According to Darcy equation, the net flux across a face of porous surface, **u**, is linearly related to the pressure gradient, ∇p, as

$$\mathbf{u} = -\frac{\mathbf{K}}{\mu} \nabla \mathbf{p}\_{\prime} \tag{2}$$

where K is the permeability of the porous medium and μ the geothermal fluid dynamic viscosity. Since the fluid flow is coupled to heat transfer, the dependence of fluid properties on temperature is taken into account.

The Darcy's flow model is coupled with heat transport in a porous medium to study the temperature distribution under geothermal exploitation. In the heat transport equation, the local thermal equilibrium between the solid and fluid phases is considered. Therefore, the solid temperature, *Ts*, is equal to the fluid temperature, *Tf*. Moreover, heat transport between solid and fluid phases is considered to be negligible. The heat transport in the subsurface is described as [64,65]

$$\mathbf{T}\left(\rho\mathbf{c}\_{\rm p}\right)\_{\rm eq}\frac{\partial\mathbf{T}}{\partial\mathbf{t}} + \left(\rho\mathbf{c}\_{\rm p}\right)\mathbf{u}\cdot\nabla\mathbf{T} = \nabla\cdot\left(\mathbf{k}\_{\rm eq}\nabla\mathbf{T}\right),\tag{3}$$

where cp is the specific heat, k is the thermal conductivity. The term (ρcp)eq represents the volumetric heat capacity of the porous medium. The subscript eq indicates that these are equivalent properties since they are spatially averaged to account for the porous matrix according to

$$\left(\rho\mathbf{c}\_{\rm p}\right)\_{\rm eq} = \theta\_{\rm p}\rho\_{\rm p}\mathbf{c}\_{\rm p,p} + \left(1 - \theta\_{\rm p}\right)\rho\mathbf{c}\_{\rm p'} \tag{4}$$

$$\mathbf{k}\_{\rm eq} = \theta\_{\rm P} \mathbf{k}\_{\rm P} + \left(\mathbf{1} - \theta\_{\rm P}\right) \mathbf{k}\_{\rm \prime} \tag{5}$$

where θ<sup>p</sup> is the porous matrix volume fraction and subscript p indicates quantities that are related to the solid porous matrix.

#### *4.2. Boundary and Initial Conditions*

The boundary conditions employed in the present model refer to the ground and the well domains. For logistical reasons, the position of the injection well is not the traditional one, which is downstream of the extraction well [66]. The injection well is in fact placed, along the direction of groundwater flow, parallel to the extraction well and not far behind this (Figure 2). The top and bottom surfaces of the ground domain are considered impermeable to mass (*u* = 0), such as the lateral surfaces parallel to the groundwater flow. The average hydraulic gradient, (i), in the groundwater layer was estimated, according to Corniello et al. [22], with a NE–SW direction of the flow, determined from the iso-piezometric map of the area (Figure 2).

A prescribed velocity u was assigned on the filtering lateral surfaces of the wells. This value was calculated considering the water flow rate of extraction and injection, . *V*, measured during the experimental campaign. As initial condition for the groundwater flow equation, the hydraulic head, H, derived from experimental measurements, was imposed on the whole domain. The bottom ground surface was assumed to be adiabatic, whereas a convective heat flux is imposed on the top surface of the domain. The external temperature was assumed to be equal to 16.0 ◦C, which is the yearly average temperature of the area. As concerns the wells, a fixed temperature, T, equal to that of flow extraction and injection, is imposed on the lateral surfaces. The water temperature values measured along the exploration well [1,56] are illustrated in Figure 3. The soil was assumed to be in local thermal equilibrium with the groundwater, then the measured temperature profile (Table 3) was considered as the initial condition on the ground domain. All boundary and initial conditions are summarized in Figure 3 and Table 4.


**Table 4.** Boundary and initial conditions.

The experimental campaign revealed a sub-vertical fault zone, NW–SE oriented. The injection well is located at 135 m (Figure 2) from the production well, in the rearward direction. From the stratigraphical data of the exploration well, two different faults were found between 100–120 and 180–240 m b.g.l., according also to the results of the detailed geophysical survey. The deepest fault if projected towards the injection well may or may not meet this well at about 80 m depth. For this reason, two case studies were analyzed, one with the fault crossing the injection well and the other without. The fault zone presents the same characteristics of the layer identified by the number 10 in Table 2.

**Figure 3.** Sketch of boundary and domain initial conditions (for clarity the width of the wells is enlarged). P and I represent the production and injection wells, respectively.

#### **5. Results and Discussion**

The numerical model described in the previous section was used to study the sustainable use of the low-temperature geothermal field of Mondragone. In order to evaluate the sustainability of this use, the temperature field in the ground was modeled over a period of time in which the source was continuously used at the full capacity of the plant. As mentioned above, the analysis was carried out with and without the fault zone crossing the injection well, to assess its influence on heat transfer and fluid flow. The simulations that refer to a single fault are indicated as Case 1 from now on, whereas those with the domain that includes two fault zones are referred to as Case 2. Figure 4 shows a schematic representation of the domain considered for the analysis of Case 1 where a single fault, indicated as F1, is present. In Case 2 the domain is the same, with the addition of a fault zone crossing the injection well, as shown in Figure 5. For the sake of clarity, only the data concerning the fault crossing the injection well, indicated as F2, are reported in Figure 5.

The overall domain is defined as a rectangular parallelepiped of 163 m in depth with a basis equal to 250 × 200 m. Below this depth, the effect of the geothermal system is considered negligible due to the lower degree of fracturing in limestone. The fault zones have both a thickness of 20.0 m and a slope of 10◦ in relation to the vertical axis. The production and reinjection wells cross the fault cataclastic zone starting from 80.0 to 120 and 68.0 to 110 m b.g.l., respectively. Both the wells have a radius of 24.4 cm and a filtering section of 10.0 m in height from the top of the cataclastic zones.

The water properties, ρ, μ, *cp*, and *k* are considered temperature-dependent. For the ground domain, literature values of thermal conductivity, density, porosity, and permeability of the eight layers found during the well drilling were considered. The parameters used in the numerical simulation are reported in Table 4.

**Figure 4.** Computational domain in the case of a single fault (F1).

**Figure 5.** Computational domain in the case of double faults (F1 and F2).

A 3D unstructured mesh, refined near all the boundaries to capture the larger gradients of both flow velocity and temperature, was used. Indeed, the grid needs to be finer in the region of greatest change in the skin zone close to the well and can be coarser further out in the reservoir [67]. For the sake of clarity, the mesh used for the case of a single fault is reported in Figure 6.

**Figure 6.** Mesh used to analyze the case of a single fault.

The details of the meshes used for the two cases are shown in Table 5. A mesh sensitivity study was carried out to obtain grid-independent results from the calculations.


**Table 5.** Mesh details for Case 1 (single fault) and Case 2 (two faults).

In order to assess the impact of the geothermal exploitation on the ground domain, the groundwater velocity field and the ground temperature field were determined through the proposed model for a period of time of 60 years.

The temperature fields at different times in the *xz* plan cutting the domain in correspondence of the two wells' axes are reported in Figure 7 for the case of the single fault (Case 1) and in Figure 8 for the case of double (Case 2) fault zones. For the sake of clarity, the section cuts are also reported.

**Figure 7.** Temperature field of the sections *xz* crossing the wells, at different times, for Case 1 (single fault).

**Figure 8.** Temperature field of the sections *xz* crossing the wells, at different times, for Case 2 (double faults).

The effect of geothermal extraction is mainly contained in the area surrounding the production well. A moderate decrease in the ground temperature can be observed in the downstream region: the variation is similar in both Cases 1 and 2.

In the upstream section the temperature field remains almost unchanged in Case 1, whereas in Case 2 a temperature decrease can be observed in the region delimited by the two faults. The difference is due to the permeability and the thermal conductivity of the fault zone, which are larger than those of the surrounding material. However, in both cases after 10 years of operation the temperature field does not change any more (reaching steady state operation of the reservoir). The effect of injection is more pronounced than that of extraction in both cases. However, the thermal disturbance does not reach the superficial layers, since it is limited by the 5th layer which has lower values of permeability and thermal conductivity with respect to the other layers (Table 2). The injection in the fault slightly modifies the temperature distribution within the domain: the propagation of the thermal disturbance is lower in layers over the fault and higher towards the bottom of the domain. In both cases, the temperature field does not change any more after 20 years of operation, therefore the period of time needed to reach a steady condition is longer in the region surrounding the injection well than in the area of the production well. This is probably due to the higher difference in temperature between groundwater surrounding the injection well and injected water.

The temperature field, at different times, in the *yz* plan cutting the domain in correspondence of the axis of two wells is shown in Figures 9 and 10 for Cases 1 and 2, respectively. For the sake of clarity, the section cuts are also reported. In these plots, it can be noticed that the temperature decrease caused by the injection well is localized in the region of the domain behind the production well and it is less pronounced in Case 1. In Case 1 the temperature propagation is more pronounced towards the upper layers, whereas in Case 2 it is larger in the bottom of the domain.

The temperature distribution and the streamlines indicating the velocity field, at different depths and times, reported in Figure 11, clearly show that the direction of the thermal disturbance is significantly influenced by the groundwater flow. A stable temperature distribution is reached after 25 years at the lowest depths, whereas only 5 years are needed at highest depths. The thermal disturbance in the layer close to the top surface of the domain is concentrated in the region surrounding the injection well.

The fault located in correspondence of the injection well, F2, does not significantly affect the temperature distribution, except for the upstream region of the production well, where the temperature is slightly lower than in Case 1. Conversely, the fault influences the velocity field, due to the permeability, which is larger than in the adjacent layers. This variation in the velocity field is mainly responsible for the extension of the region affected by the thermal disturbance in Case 2.

At depths below 75.0 m from ground level the temperature and velocity fields are constant over time, therefore only their variation with depth was taken into account. Analyzing the plots in Figure 12, which are all related to the time of 60 years, it can be observed that the thermal disturbance is more extended in Case 2. In addition, at 155 m b.g.l. the thermal disturbance can be neglected in Case 1, whereas in Case 2 the region downstream of the injection well is still affected by the geothermal exploitation.

**Figure 9.** Temperature field of the sections *yz* crossing the wells, at different times, for Case 1 (single fault).

**Figure 10.** Temperature field of the sections *yz* crossing the wells, at different times, for Case 2 (double faults).

**Figure 11.** Temperature field at different depths and times, for Case 1 (single fault) and Case 2 (two faults).

**Figure 12.** Temperature field at different depths, for Case 1 (single fault) and Case 2 (two faults).

The Darcy velocity field and magnitude at different depths for Cases 1 and 2 are reported in Figures 13 and 14, respectively. The *xy* sections reported in these figures are related to different layers of the domain, in order to analyze the influence of their properties on the groundwater flow. In the first two layers of the domain, (refer to Figure 13a,b and Figure 14a,b), both the distribution and the magnitude of Darcy velocity do not change from Case 1 to Case 2, and result as constant in the entire *xy* plans considered. This is due to the low permeability of the third layer, which avoids the effect of the geothermal exploitation on the groundwater flow reaching the top of the domain. Indeed, the groundwater flow in the lower layers of the domain, which are characterized by higher values of permeability, varies in the regions where the production and injection wells are located (refer to Figure 13c,d and Figure 14c,d).

Considering the injection well, the velocity magnitude decreases upstream and increases downstream, whereas the opposite trend can be observed in the region surrounding the production well. In Case 2, a lower propagation of the fluid flow disturbance can be observed: this is due to higher permeability of the material in the correspondence of the fault, which facilitates the groundwater flow, which is more concentrated in the tectonic region. In addition, considering Case 2, the velocity magnitude varies differently in the zones of injection and production, probably due to the different vertical extension of the faults. The fault located in the injection area, F2, is more extended and its effect on the upper layers is larger.

In the sections that cut the wells (refer to Figure 13f,g and Figure 14f,g), the Darcy velocity field and magnitude are similar in Case 1 and Case 2. Obviously, the magnitude is significantly higher than at other depths due to the flow extraction and injection. A different trend can be observed in sections 10.0 m distant from the bottom of the wells in the vertical direction (refer to Figure 13e,h and Figure 14e,h), mainly in the areas surrounding the injection well. The fault affects the groundwater flow, whose magnitude increases in the area surrounding the injection well and decreases upstream of the production well. However, a vertical distance of 10.0 m is sufficient to observe a significant decrease in the Darcy velocity, which tends to the undisturbed value as the horizontal distance from the wells increases. At higher depths, the velocity magnitude continues to decrease (refer to Figure 13i,l and Figure 14i,l). The velocity magnitude is higher in the central region of the considered section, where the groundwater has the same direction of the undisturbed fluid flow in both cases, injection and extraction.

**Figure 13.** Darcy velocity field at different depths for Case 1 (single fault).

**Figure 14.** Darcy velocity field at different depths for Case 2 (two faults).

Finally, in Figure 15, the temperature over time is plotted at three different distances from the wells (10, 50, and 100 m) at a depth equal to half of the filtering sections of the wells, for the two cases analyzed.

**Figure 15.** Temperature over time at three different distances from the wells (10.0, 50.0, and 100 m), for two different depths (−95.0 and −75.0m).

Considering the production well (see Figure 15a,b) whose filtering section goes from −90 to −100 m, the geothermal exploitation causes a temperature decrease of around 2 ◦C during the considered period of time. The temperature becomes constant after 10 years in Case 1 and 20 years in Case 2. It is worth noticing that in Case 2, the temperature decrease at a distance of 100 m is slightly lower than in Case 1, suggesting a positive effect of the injection in the fault. A similar conclusion can be drawn analyzing the results related to the injection well (see Figure 15c,d). The variation in temperature with respect to the initial condition is lower than 1 ◦C, therefore significantly lower than that observed in the region of the production well. The period of time needed to reach a constant temperature is longer than for the production well: 20 years for the Case 1 and 25 years for the Case 2. The injection in the fault has a positive effect, reducing the temperature oscillations during the first years of exploitation in Case 2. The low decrease of the basin temperature is coherent with the results found in the literature related to the geothermal exploitation of low-temperature reservoirs [2].

So far, the numerical results show that the geothermal exploitation modifies the groundwater flow in correspondence with the filtering sections of the wells, whereas only a weak effect can be observed in the rest of the domain, in terms of speed and direction. In order to account for the possible presence

of a fault crossing the injection well, two case studies were analyzed. When the injection occurs in the fault the variation of the velocity field is lower, thanks to the permeability in correspondence with the fault. The injection in the fault positively affects the temperature distribution as well. The thermal disturbance does not reach the top layers of the domain due to their hydraulic and thermal properties in both cases. It is worth noticing that in the case of injection in the fault the disturbance tends to be more concentrated in the same fault region, decreasing its effect on the other layers of the domain. In addition, in this case, also the temperature oscillation over time observed at the depth of the wells is lower than in the case without the injection in the fault.

The proposed model allows estimation of the effect of geothermal exploitation not only on the reservoir but also on the operation of the technological system that uses the geothermal source for thermal energy production. The results of the simulations indicate a temperature decrease of around 2.00 ◦C in the extraction temperature during the considered period of time. Such a decrease will reduce the thermal energy available from geothermal exploitation. In the specific case analyzed in this work, where a heat pump was considered to supply heat to two schools, a decrease in the COP of the heat pump is expected.

#### **6. Conclusions**

The numerical results on the geothermal exploitation here reported are of global interest for the following reasons: (a) the nature of aquifer represented by carbonate rocks with strong anisotropy, due to the presence of several fault systems, with confined groundwater; (b) the environmental effects of an open-loop well-doublet plant are assessed considering the geometry of the wells (imposed by logistical constraints) which is designed in a different way from the one usually adopted, where the injection well is placed downstream of the extraction well. In fact, even if thermal feedback is present at the production well due to the position of injection well, the sustainability of the use of the geothermal resource over time (60 years here tested) is asserted by the numerical simulation. In fact, the main results show that the temperature decrease is only 2.00 ◦C in correspondence with the production well filtering section, and even less at greater depths.

It is likely that the sustainability is linked, considering the hydrogeological scheme, to the characteristics of groundwater flow (direction and pressure) and to the position of the filtering section of the extraction well, that is placed near the main cataclastic zone of a fault along which the permeability is higher than the surrounding rocks and the upwelling of deep hot fluids is easier.

Even assuming the presence of another fault at the injection well, the numerical simulation still indicates the full sustainability of the geothermal plant. It is worth noticing that in the case of injection in the fault the disturbance tends to be more concentrated in the same fault region, decreasing its effect on the other layers of the domain. Finally, in all the cases considered, the thermal disturbance does not reach the top of the domain due to the hydraulic and thermal properties of the rocks closest to the ground level.

The definition of the hydrogeological and stratigraphic scheme of the geothermal area together with the numerical modelling are therefore decisive either for the final choice of the well plant localization or for the evaluation over time of the environment effects of the use of the geothermal resource in faulted carbonate rocks, which are common in Southern Italy and peri-Mediterranean regions.

**Author Contributions:** Conceptualization, M.I., N.M., and L.V.; Data curation, M.I. and A.C. (Alfonso Corniello); Formal analysis, S.D.F.; Funding acquisition, M.I., A.C. (Alberto Carotenuto), N.M., and L.V.; Methodology, S.D.F. and N.M.; Project administration, M.I., A.C. (Alberto Carotenuto) and N.M.; Writing—original draft, S.D.F.; Writing—review and editing, M.I., A.C. (Alberto Carotenuto), A.C. (Alfonso Corniello), S.D.F., N.M., R.S., L.V., and A.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors gratefully acknowledge the financial support of VIGOR project (Valutazione del Potenziale Geotermico delle Regioni Convergenza) CUP: B72J10000060007, GEOGRID project CUP: B43D18000230007 and project M027061 funded by Italian Ministry of the Foreign Affairs.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **A Novel Ground-Source Heat Pump with R744 and R1234ze as Refrigerants**

**Giuseppe Emmi 1,\*, Sara Bordignon 1, Laura Carnieletto 1, Michele De Carli 1, Fabio Poletto 2, Andrea Tarabotti 2, Davide Poletto 3, Antonio Galgaro 4, Giulia Mezzasalma <sup>5</sup> and Adriana Bernardi <sup>6</sup>**


Received: 22 September 2020; Accepted: 26 October 2020; Published: 29 October 2020

**Abstract:** The energy-saving potential of heat pump technology is widely recognized in the building sector. In retrofit applications, especially in old and historic buildings, it may be difficult to replace the existing distribution and high-temperature emission systems. Often, historical buildings, especially the listed ones, cannot be thermally insulated; this leads to high temperatures of the heat carrier fluid for heating. In these cases, the main limits are related, on the one hand, to the reaching of the required temperatures, and on the other hand, to the obtaining of good performance even at high temperatures. To address these problems, a suitable solution can be a two-stage heat pump. In this work, a novel concept of a two-stage heat pump is proposed, based on a transcritical cycle that uses the natural fluid R744 (carbon dioxide) with an ejector system. The second refrigerant present in the heat pump and used for the high-temperature stage is the R1234ze, which is an HFO (hydrofluoro-olefin) fluid. This work aims to present the effective energy performance based on real data obtained in operating conditions in a monitoring campaign. The heat pump prototype used in this application is part of the H2020 Cheap-GSHP project, which was concluded in 2019.

**Keywords:** heat pump; R1234ze; CO2; geothermal; renewable energy; historic building; energy saving

#### **1. Introduction**

The European Union (EU) set ambitious targets to fight the issue of climate change, to guarantee the security of supply and increase competitiveness in the energy sector. The building sector, responsible for about 40% of total EU energy consumption, is a great contributor to greenhouse gas (GHG) emissions [1]. When considering the household sector, the share of the European final energy consumption is around 26%, with about 80% due to heating, cooling, and domestic hot water production [2]. An important step to meet the goals of reducing GHG emissions and increasing the renewable energy sources (RES) share can be obtained by adopting technologies that do not employ fossil fuels for the heating and cooling of buildings.

Heat pumps are electrical appliances that can be used for this purpose and allow an improvement in the efficiency of the energy system.

These machines can be employed when a replacement of the energy plant is needed for the retrofit of existing buildings [3,4]. For these applications, high-temperature heat pumps (HTHPs) can be used when heat is demanded at high-temperature levels (80–150 ◦C) and COP (coefficient of performance) values range between about 2.4 and 5.8 with a temperature lift of 95 to 40 ◦C, respectively [5]. Traditionally, HTHPs are built using a single-stage thermodynamic cycle, representing a challenge for obtaining favorable COPs with high-temperature heat production. In the literature, studies investigating the use of new refrigerants can be found [6,7], highlighting the need for the adoption of fluids with low GWP (global warming potential) that show acceptable system performance at high temperature.

Other studies analyzed different machine configurations, where the thermodynamic cycle is modified to increase the efficiency and reach higher temperature levels. Dai et al. [8] investigated five configurations of dual-pressure level condensations for exploiting waste heat, obtaining an improvement in COP of more than 9%, compared to traditional cycles. One possible layout is the cascade cycle, composed of two independent single-stage cycles, the high-temperature-stage cycle and the low-temperature-stage cycle [9]. The two cycles are connected by an intermediate heat exchanger that works as the evaporator for the high-temperature cycle and as the condenser for the low-temperature cycle.

Fine et al. [10] developed a numerical model of a solar-assisted cascade heat pump using photovoltaic thermal panels to provide the heat source and electricity to the system. Xu et al. [11] presented an experimental investigation on an air-source cascade heat pump operating in cold climate conditions, demonstrating a higher efficiency in comparison to other heat solutions at low ambient temperature and in a high water-supply temperature region. Yang et al. [12] proposed a single-fluid cascade air-source heat pump that can operate in different modes, and the analysis of the monitored data showed a linear increase in the heating capacity with the lower-stage compressor speed.

Le et al. [13] analyzed an air-source cascade heat pump coupled with thermal storage through laboratory and field results and dynamic simulation models, finding that the heat pump can be used to obtain a CO2 emission reduction (up to 57%) compared to the traditional boilers and obtain a seasonal COP of 2.12 with the direct heating of the retrofitted building. Mota-Babiloni et al. [14] presented an optimization of the intermediate temperature and the internal heat exchanger effectiveness in both stage cycles of a cascade HTHP using low-GWP refrigerants, finding the maximum COP (3.15) in the combination of pentane and butane.

A different layout can be adopted by including an ejector in the thermodynamic cycle, which allows the decrease in the compression work by reducing the throttling losses and the liquid overfeeding and lifting the compressor inlet pressure [15]. The ejector is a flow device with two intake ports and one discharge port. The primary high-pressure stream and the secondary low-pressure stream are mixed inside the ejector and discharged at some intermediate or back-pressure [16]. Brodal and Eiksund [17] investigated the heat pump performance with and without an ejector or a suction gas heat exchanger, as a modification to conventional transcritical CO2-based heat pump systems. They found that for pure CO2-based heat pumps, the systems with an ejector are more efficient than systems without, with an increase in the COP up to 19%. Liu and Lin [18] carried out a thermodynamic analysis of an air-source heat pump producing heat at two temperature levels and using a zeotropic mixture refrigerant (R1270/R600a) of three different configurations, while Besagni et al. [19] studied the influence of different refrigerants on ejector refrigeration systems.

A further way to enhance the efficiency of the heat pump could be to use the soil as the heat source, instead of the external air. These heat pumps are called ground-source heat pumps (GSHP) [20–23] and their advantages range from the high energy performance to environmental friendliness due to the exploitation of renewable energy sources and ease of integration with other energy systems. D'Agostino et al. [24] compared GSHPs to air-source heat pumps or a condensing boiler coupled to a chiller to analyze the energy savings of the investigated system that reaches up to 55% for one of the two locations. Christodoulides et al. [25] provided the cost analysis and comparisons between

an air-source heat pump and a ground-source heat pump showing that, for the case study in Cyprus, the air source HP is highly competitive, while Li et al. [26] theoretically demonstrated the general exergy convenience of GSHPs compared to air-source heat pumps.

In this work, the monitored data of a novel partial cascade GSHP are presented. Indeed, the heat pump project is currently involved in a patenting process (Patent application number: 102020000021097, presented on 7th September 2020). The heat pump configuration includes an ejector and two gas-coolers in series in the low-temperature cycle, where CO2 is used as the refrigerant in a transcritical cycle. The high-temperature-stage fluid is the refrigerant R1234ze. This last cycle is not properly the high-temperature one, as the analyzed layout of the heat pump is different from a common cascade cycle as described in the following section of the text.

#### **2. Materials and Methods**

The present paper provides the description and the energy analysis of a heat pump prototype, which uses low-GWP fluids as refrigerants. Moreover, the heat pump has been developed to supply the high-temperature heat carrier fluid to the terminal units used for heating a real building in Zagreb (Croatia). The first part of the paper summarizes the main information about the case study and the details of the heat pump from the mechanical and thermodynamic points of view. In the second part of the work, the results of the operating conditions of the heat pump are described and discussed. The energy analysis has been carried out using the measured data obtained from the monitoring system installed in the plant.

#### *2.1. The Building*

The ground-source heat pump has been installed in the building that hosts the Technical Museum Nikola Tesla in Zagreb. The entire complex is listed within the National Register of Cultural Property. The building area is located in the historic city center, which is protected by a special regime of conservation and heritage protection. The building complex is enlisted within the National Register of Cultural Property, by the decision of the Ministry of Culture of the Republic of Croatia in 2005, and it is located in the historic urban areas of the city of Zagreb. The national law on the Protection and Conservation of Cultural Goods forces any structural work planned for the museum to be greenlit by the Urban Office for the Protection of Cultural and Natural Monuments. A limited part of the museum was renovated, which was an exhibition room of 380 m<sup>2</sup> with a volume equal to 1463 m3, as shown in Figure 1.

**Figure 1.** Plan of the exhibition room (ZONE A) of the Technical Museum Nikola Tesla in Zagreb (Croatia). Source: [27].

The original plant system consisted of only six electrical heaters originally used for heating, and no devices for cooling were present. This had a serious impact on the environmental comfort of both visitors and personnel with a possible threat to the fire safety of the structure, which is made mainly of wood components both in its interior and exterior.

The refurbishment intervention was approved by the museum and municipal authorities and included the installation of a CO2 heat pump with a capacity of about 30 kW. The installed heat pump is a prototype device, specifically designed to provide water at high temperature to terminals.

The refurbishment also included the drilling of a borehole heat exchanger (BHE) field consisting of six 100 m-deep BHEs in the courtyard, plus the relative hydraulic connections. In order to assure heating and cooling in the exhibition room of the museum, 10 fan coils were installed [27].

Unlike the previous work, the present analysis investigates in detail the real behavior of the heat pump, using the data of the plant obtained from a monitoring campaign. The previous paper focused on a possible application of different management strategies for the air-conditioning of the plant and, in that context, a dynamic simulation tool was used for deriving the energy performance of the heat pump considering a mean seasonal coefficient of performance. The museum and the municipal authorities approved the retrofit of the plant after the involvement of the Museum in the Cheap-GSHP European Project [27].

#### *2.2. The Thermal Power Plant*

The reversible heat pump is a water-to-water machine that uses a geothermal field as the source and sink in heating and cooling operation, respectively. It consists of two cycles: The low-temperature cycle uses the natural fluid CO2 as the refrigerant, and the high-temperature cycle uses the hydrofluoro-olefin (HFO) refrigerant R1234ze.

The design of this machine came from the necessity to develop and build a heat pump generation system suitable for a retrofit application, where the heat carrier fluid is required at high-temperature levels, to be used in existing terminal units. As the existing terminal units were electric in the analyzed case study, the use of the heat pump presented several advantages. Indeed, the heat pump reduces the electric energy demand and, at the same time, can produce chilled water to cool the exhibition room in summer, while in the previous configuration, the building was not provided with a cooling system. The solution proposed in the present work is suitable for buildings characterized by a dominant heating thermal load profile. Usually, commercial heat pumps can be used for cooling, through the use of a four-way valve in the refrigerant circuit that reverses the thermodynamic cycle (the component that was used as the condenser is used as the evaporator). On the contrary, in this study, as the heat pump has a double cycle in cascade, the switch of the operating mode is obtained by changing the hydraulic circuit side, not the refrigerant circuit.

The switch of the operation mode, heating and cooling, is allowed using the two 4-way valves and is highlighted in the two schemes of Figure 2. The figure shows the connections of the thermal power plant located outside the building, inside a dedicated room used by the Museum as a showroom for didactic visits. The plant is divided into four main parts: The heat pump, the hydronic module, the borehole heat exchanger (BHE) field, and the distribution pipelines.

The refurbishment of the thermal power plant also included the drilling of a BHE field, consisting of six 100 m-deep ground heat exchangers positioned in the garden and the relative horizontal hydraulic connections. In order to assure heating and cooling provision to the exhibition room of the museum, 10 fan coils were installed. The fluid that circulates in the BHE field is a mixture of pure water and antifreeze. For ground-source heat pump applications, the antifreeze is normally used to prevent the freezing of the horizontal connections during the stop of the thermal power plant, or to prevent the damaging of the plant and of the heat pump when the borehole field is not properly designed and the ground is affected by thermal drift.

(**b**) Cooling Mode

**Figure 2.** Scheme of the thermal power plant of the system.

#### *2.3. The Heat Pump*

As can be seen from the scheme in Figure 2, the heat pump configuration differs from the more common double cascade thermodynamic cycles that can usually be found in the market. The heat carrier fluid exchanges heat with the two thermodynamic cycles by the use of two flat plate heat exchangers connected in series, while usually, the double cascade cycles use only one heat exchanger at the condenser side of the high-temperature cycle. For this reason, the configuration in the figure could be defined as a "partial" double cascade cycle. In the market, high-temperature transcritical heat pumps that use CO2 as the refrigerant are quite common, but among the main limits of these machines, the high values of temperature difference required at the condenser side limit the achievement of good energy performances. Consequently, the terminal units used for heating should work with low flow rates and high temperature drops to go below the critical temperature of the CO2. As widely known in the air-conditioning field, the only terminal units that can work below the critical temperature

are radiant systems. However, in these applications, the temperature drops do not exceed 3–4 ◦C, as suggested by the standards, in order to guarantee a uniform temperature of the radiant surface and good thermal comfort for the users. The layout of the heat pump prototype allows this last problem that limits the use of the transcritical cycle in the field of air conditioning of buildings to be overcome.

In this case study, the heat pump exploits the advantages of the medium/high-temperature cycle, using the CO2 refrigerant in the transcritical cycle and the heat exchanged at high temperature by the subcritical cycle, which uses the R1234ze refrigerant. The total heat flux exchanged with the heat carrier fluid is the sum of the heat flux exchanged at the gas cooler of the CO2 transcritical cycle and the heat flux exchanged at the condenser of the high-temperature cycle working with the HFO refrigerant. As is well known, the CO2 is not suitable for every application in refrigeration, air-conditioning, and heat pump systems, and it presents some limitations in order to guarantee good energy performance. The CO2 has a vapor pressure much higher than other traditional refrigerant fluids and its critical temperature is around 31 ◦C. This last characteristic leads to the impossibility of discharging heat to the external air through condensation in a subcritical cycle, when the air temperature, or in general, the heat carrier fluid, is above the critical temperature. The main consequence is that CO2 can be used efficiently for cooling applications when heat is discharged below the critical temperature, while at higher temperatures, when the cycle works at supercritical pressures, only gas cooling is possible. Therefore, in cooling applications, the use of the ground as the heat sink is a possible and efficient solution. On the other hand, when the heat pump operates in heating mode, the HVAC (Heating Ventilation Air Conditioning) applications could result more complex because, generally, the supply temperature for the terminal units is above 30 ◦C, except for radiant systems. However, CO2 is a natural fluid, its ODP (ozone depletion potential) is null, it is not flammable and not toxic (A1), and its GWP is equal to 1. For this reason, CO2 was also used in the past when environmental advantages and safety restrictions reasons overcame the energy drawbacks. The R1234ze refrigerant is an alternative fluid to the R134a and its environmental impact is lower if compared to the latter. The main properties of the two refrigerants are summarized in Table 1.


**Table 1.** Main properties of the refrigerants used in the heat pump [28].

The thermodynamic cycles of the heat pump are representedin Figure 3 in a temperature–entropy chart. The diagram shows, from a qualitative point of view, the operating points of the thermodynamic cycles during the heat pump operations. The letters and the numbers in the chart refer to the points reported in Figure 2, which shows the scheme of the heat pump's high and low-temperature cycles.

The properties of the main components of the heat pump are summarized in Table 2. The heat pump is a prototype because it is not available in the market yet, even though all the devices constituting the machine are employed in other applications. For example, the ejector used in the CO2 cycle is not commonly used in HVAC application, but rather in commercial refrigeration. One of the main objectives during the development of this prototype was to investigate the energy performance potential of this technology at the time of the project. In the last two years, the HVAC market started to introduce new devices and components properly developed for HVAC applications that use the CO2 as the refrigerant. These new products can certainly guarantee better results than those obtained with the technology available about 3–4 years ago and used in the prototype; therefore, the results of this study can be considered the starting point of this emerging technology. The investigated heat pump was thought to be used in plug-and-play applications, as the generation module and the hydraulic module have been included in a common box. The control of the heat pump works at two levels: The high-temperature stage is controlled by the regulation system that keeps the gas cooler outlet temperature at a setpoint value of 35 ◦C, with a dead band of 5 ◦C. If the temperature drops below 30 ◦C, the thermal efficiency of the high stage cycle decreases under acceptable values as the evaporating temperature is too low. However, the dead band of 5 ◦C prevents the continuous on/off of the HFO compressor. The compressor of the CO2 sets its working conditions by monitoring the return temperature of the heating and cooling terminal units. The volume flow rate of the BHE field circuit is modified according to the following principles: In heating mode, when the evaporator supply temperature decreases, the flow rate increases, and the control is done evaluating a set point evaporator pressure; in cooling mode, the flow rate is modified by keeping a constant temperature difference between the inlet and the outlet at the evaporator side. The volume flow rate at the user side of the heat pump is set to maintain a constant temperature difference between the supply and return temperatures of the circuit, to 5 ◦C and 10 ◦C in cooling mode and heating mode, respectively.

**Figure 3.** Thermodynamic cycles of the high-temperature heat pump.

#### *2.4. The Monitoring System*

The monitoring system consists of flow meters and temperature probes for the measurement of the heat fluxes exchanged between the heat carrier fluids and the refrigerants, at the user side (fan coil terminal units) and source side (ground heat exchangers field), respectively. Other parameters have been monitored in the thermodynamic cycle in order to evaluate the behavior of the cycle during the operation of the system. In particular, considering the CO2 cycle, the pressures of the refrigerant inside the gas cooler (GC) and inside the evaporator (EV) have been logged by the system. The electrical consumption of the heat pump, necessary for the evaluation of the COP and EER (Energy Efficiency Ratio), has been measured using two energy meters. The first electricity meter monitored the electrical energy demanded by the compressors and the auxiliary devices of the heat pump. These consisted of the auxiliary electric resistances present in the heat pump for the safety of the components during the stop of the plant in the heating period, in the unit controller and in other electric minor devices present inside the heat pump box. The second is located in the hydronic module of the thermal power plant, which has two pumps (user and source side). The electrical energy meter measures their energy consumption and the demand of the controller, which manages the operation of this system. As for the first energy meter, a small part of the total electrical energy consumption is due to other devices located in the hydraulic box, installed in the thermal power plant of the Museum.


**Table 2.** Properties of the heat pump and design conditions for the compressors.

The main properties of the monitoring units installed in the thermal power plant are summarized in Table 3. The experimental uncertainty of the measurement chain is within ±10% for the analyzed case study. The time step of the logging data has been set to 30 s.



#### **3. The Monitoring Data**

The thermal power plant of the air conditioning system has been monitored for about one year, from the beginning of March 2018 to the end of April 2019. The energy data and the values of pressures of the thermodynamic cycle have been analyzed to evaluate the thermal and energy behavior of the system in cooling and heating operation mode, with the aim of checking the trend of the pressures during the switch-on cycle of the system and the energy performance of the heat pump. The air conditioning system operation, during the monitoring period, was scheduled according to the opening days of the Museum, when possible. For this reason, when the building was closed to visitors, the heat pump switched on only a few times during the day to keep the water inside the tank at the set point temperature of about 67.5 ◦C in heating and 7 ◦C in cooling, as can be seen in the trends of the storage temperature. In some cases, with the purpose of gathering enough data for the energy analysis of the system, the system was switched on more often and independently from the opening hours of the Museum. Two examples of operating conditions of a day are shown in Figures 4 and 5 for the heating and cooling periods, respectively. In the chart, some details about the switching-on and -off of the heat pump are reported. Indeed, the trends of the temperature of the thermal storage and of the heat carrier fluids are highlighted.

**Figure 4.** Example of day—Heating mode.

**Figure 5.** Example of day—Cooling mode.

The main objective of this work was the investigation of the novel reversible heat pump producing high-temperature water for heating and chilled water for cooling a historic building during the winter and summer period, respectively. This goal was one of the main fields of investigation of the European Cheap-GSHP H2020 project.

#### **4. Results and Discussion**

In this section, the data of the monitoring campaign are discussed in detail, evaluating the behavior of the system from the thermal and electrical points of view. The operating conditions for some selected days and the energy fluxes involved in the system have been investigated. In particular, the heating mode and the cooling mode of the heat pump were considered. Some days have been chosen and analyzed considering the days with continuous operation and the days with only a few starts of the heat pump during the day.

In Figure 6, the trend of the system's temperatures and R744 pressures are represented for 18 November 2018. This is a representative day of continuous operation mode in heating. In particular, the chart shows the heat carrier fluid temperatures at the inlet and outlet of the high-temperature heat exchangers and the heat carrier fluid temperatures at the inlet and outlet of the evaporator coupled with the BHE field. The temperatures of the fluids have been measured in the points of the system, as represented in the scheme of the thermal power plant in Figure 2. The switch-on of the heat pump is controlled by the value of the temperature of the fluid in the thermal storage tank (Ttank). The setpoint

temperature of the storage tank is 67.5 ◦C with an upper dead band of 1.5 ◦C, so the heat pump is shut down when the temperature of the storage tank is higher than 69 ◦C. As can be seen in the chart during this day, the system works continuously, the heat pump starts for 29 times, and each on and off cycle lasts about 50 min. The pressure of the R744 transcritical cycle is shown in Figure 6b. The high pressure (HP—Point 3 of the scheme in Figure 2) of the cycle at the gas cooler varies between 95 and 80 bar while the low pressure (LP—Point 9 of the scheme in Figure 2) and suction pressure (SP—Point 6 of the scheme in Figure 2) at the compressor vary between 40 and 50 bar. A zoom of the on and off cycle of the heat pump is reported in Figure 7. In the charts, temperatures and pressures between 6:28 a.m. and 7:26 a.m. are reported. At the beginning of the on and off cycle, the high pressure rises to the maximum value, decreasing gradually until the stop of the heat pump when the setpoint temperature of the storage tank was reached. As can be seen in the chart in Figure 7a, before and during the first part of the cycle, the values of Tinwg\_gc <sup>+</sup> cd and Toutwg\_gc <sup>+</sup> cd were the same. This happens because the heat pump compressor was off until 6:37 a.m., but the circulator of the hydronic module started about two minutes before the compressor. Indeed, the compressor has a time delay of switch-on after the start of the circulator for safety and protection of the components of the heat pump. The details of the pressures of the R744 cycle are reported in Figure 7b. As can be seen in the chart, the SP and evaporator pressure (LP) are the same during the operation of the heat pump, as expected, while the SP is equal to the HP when the compressor is off because the two pressures are in equilibrium through the compressor. The LP is less than the other pressures because of the expansion valves present in the refrigerant loop.

**Figure 6.** System's temperatures (**a**) and pressures (**b**)—18 November 2018.

**Figure 7.** System's temperatures (**a**) and pressures (**b**)—18 November 2018—6:28/7:26.

In Figures 8–11, similar charts are presented for two representative days in December. As for 4 December 2018, the plant was working for 14.9 h, providing water to the storage tank at 67.8 ◦C on average. From the graph in Figure 8a, it can be noticed that the switching-on cycles were less frequent in the warmer hours of the day (the external air temperature is also shown in this figure). In this case, the high pressure of the cycle at the gas cooler also varied between 95 and 80 bar, while the low pressure and suction pressure at the compressor varied between 40 and 50 bar.

**Figure 8.** System's temperatures (**a**) and pressures (**b**)—4 December 2018.

**Figure 9.** System's temperatures (**a**) and pressures (**b**)—4 December 2018—8:00/9:00.

**Figure 10.** System's temperatures (**a**) and pressures (**b**)—29 December 2018.

**Figure 11.** System's temperatures (**a**) and pressures (**b**)—29 December 2018—9:00/10:00.

Similar considerations can be done for 29 December 2018, when the heat pump results to be operating nearly without interruptions. In this case, the electrical energy demanded by the compressors of the heat pump compressors during the whole day is 291 kWh, nearly 56% more than for the day of 4 December. In the same way, the electrical energy absorbed by the auxiliaries is nearly three times with 29 against 11 kWh for the day of 4 December. In addition, the thermal energy released at the user-side of the HP and withdrawn from the ground is about 54% higher for 29 December, with an amount of 631 and 350 kWh, respectively. As shown in Figure 11b, the high pressure of the cycle at the gas cooler varies between 80 and 85 bar, while the low pressure and suction pressure at the compressor are between 35 and 40 bar. In this last case, the values of LP are lower than the other two days because of the lower temperature of the heat source that affects the evaporating temperature and the relative pressure at the evaporator side. As can be seen in the charts, the inlet temperature of the heat carrier fluid at the evaporator moves from about 11 ◦C to about 8 ◦C. On the other hand, at the gas cooler, the steady-state pressure after the first phase of the switching-on of the heat pump is about 80 bar. This is due to the control system that regulates the outlet temperature at the gas cooler at 35 ◦C, as previously described in the Heat Pump section of the text.

Considering the months of November and December, the temperature of the water entering the condenser at the user side of the heat pump is about 60 ◦C and 63 ◦C, respectively. The water is then heated up to 67/71 ◦C, maintaining an average storage tank temperature of 67.5 ◦C. At the source side of the heat pump, the heat carrier fluid from the BHE field loop has a monthly average temperature of 13.5 ◦C in November and 8.4 ◦C in December. The fluid is then cooled inside the evaporator to 9 ◦C and 6 ◦C in November and December, respectively.

In Table 4, the values for the heat pump COP and for the system COP, for the investigated days in November and December, and for the selected on-off cycle of the heat pump are shown.


**Table 4.** *COPHP* and *COPsys* in heating period.

The *COPHP* of the heat pump is defined as the thermal energy delivered at the user-side of the heat pump (*Q1*), divided by the electrical energy absorbed by the compressors of the low (*Pcomp,L*) and high-temperature cycles (*Pcomp,H)*. On the other hand, the *COPsys* of the system is defined as the thermal energy delivered at the user-side of the heat pump (*Q1*), divided by the electrical energy absorbed by the compressors of the low and high-temperature cycles (*Pcomp,H* + *Pcomp,L*) and by the auxiliaries of the hydronic system (*Paux*).

$$\text{COP}\_{HP} = \frac{Q\_1}{P\_{comp,L} + P\_{comp,H}} \tag{1}$$

$$COP\_{sys} = \frac{Q\_1}{P\_{comp,L} + P\_{comp,H} + P\_{aux}} \tag{2}$$

During these days, the source temperatures are 12 ◦C, 9.5 ◦C, and 7.1 ◦C for November and 4 and 29 December.

In cooling mode, the cycle working with R1234ze is off and only the CO2 thermodynamic cycle is switched on. The hydronic module modifies its layout by changing the position of the two 4-way valves. In this configuration, the BHE field loop exchanges heat with the gas cooler heat exchanger while the user tank exchanges heat with the evaporator heat exchanger. Unlike the heating mode, the pressures of the CO2 cycle are different because the high pressure (HP) is set by the controller of the heat pump in order to obtain the optimal value, which is a function of the outlet temperature at the gas cooler heat exchanger. Therefore, in this case, the pressure depends on the temperature of the heat carrier fluid in the BHE field loop. Similarly, as shown for the heating mode, few days have been analyzed in detail to show the behavior of the heat pump from the temperature and pressure point of views.

In particular, the temperature profile and the operating pressures of the cycle have been analyzed for one day in June, July, and August. Moreover, the one on-off cycle of the heat pump can be seen in detail for each representative day.

The temperature of the water entering the evaporator and, therefore, at the user side of the heat pump is about 7.4 ◦C for the months of June and July, while it is an average of 7 ◦C in August. On the other hand, the temperature exiting the evaporator is set to 6 ◦C. Considering the source side of the heat pump, the temperature of the heat carrier fluid coming from the BHE field loop is around 17.7 ◦C in June on a monthly average, 18.8 ◦C in July, and 17.3 ◦C in August. Respectively, inside the gas cooler, its temperature is increased to 23.2 ◦C, 24.3 ◦C, and 22.4 ◦C.

The average monthly temperature of the water inside the storage tank is 6.6 ◦C in June, and 6.9 ◦C in July and August.

In Figures 12–17, the trends of the system's temperatures and R744 pressures are represented for 29 June, 19 July, and 22 August 2018. The charts show the heat carrier fluid temperatures at the inlet and outlet of the evaporator at the user-side of the HP and the heat carrier fluid temperatures at the inlet and outlet of the condenser coupled with the BHE field.

**Figure 12.** System's temperatures (**a**) and pressures (**b**)—29 June 2018.

**Figure 13.** System's temperatures (**a**) and pressures (**b**)—29 June 2018—12:55/13:55.

**Figure 14.** System's temperatures (**a**) and pressures (**b**)—19 July 2018.

**Figure 15.** System's temperatures (**a**) and pressures (**b**)—19 July 2018—11:40/12:40.

The representative day for June was 29, when the heat pump was on nearly 9 h. The high pressure of the cycle at the gas cooler varied between 45 and 70 bar while the low pressure and suction pressure were in the range of 25 and 50 bar, as for the other analyzed summer days.

Table 5 summarizes the EER values for the heat pump and for the heat pump coupled with the hydronic system, for the investigated days in June, July, and August, and for the selected on-off cycle of the heat pump.

**Figure 16.** System's temperatures (**a**) and pressures (**b**)—22 August 2018.

**Figure 17.** System's temperatures (**a**) and pressures (**b**) —22 June 2018—11:30/16:30.


**Table 5.** *EERHP* and *EERsys* in cooling period.

The EER of the heat pump is defined as the thermal energy withdrawn at the user-side of the heat pump (*Q*2), divided by the electrical energy absorbed by the compressor of the CO2 cycle (*Pcomp,L*). On the other hand, the EER of the system is defined as the thermal energy withdrawn at the user-side of the heat pump (*Q*2), divided by the electrical energy absorbed by the compressor of the low-temperature cycles (*Pcomp,L*) and by the auxiliaries of the hydronic system (*Paux*).

$$EER\_{HP} = \frac{Q\_2}{P\_{comp,L}}\tag{3}$$

$$EER\_{sys} = \frac{Q\_2}{P\_{comp,L} + P\_{aux}} \tag{4}$$

During these days, the source temperatures are 18 ◦C, 19.4◦ C, and 19.6 ◦C for June, July, and August, respectively, and it can be seen that higher source temperatures lead to lower EER values.

From the monitoring campaign, the EER and COP of the heat pump and of the entire system were evaluated. In particular, when considering the operation during the month of July, a heat pump EER of 3.55 and a system EER of 2.74 have been found. As for the heating operation, in December, the heat pump COP was 2.15, while the system COP was 2.01.

#### **5. Conclusions**

In the present work, the configuration and monitored data for a novel cascade GSHP developed within the H2020 Cheap-GSHP project, concluded in 2019, are presented. The heat pump, installed at the Technical Museum Nikola Tesla in Zagreb, uses CO2 and R1234ze as the working fluids in a transcritical cycle where the common expansion valve is replaced by the ejector technology and in the high-temperature cycle, respectively. During the heating season, the reversible heat pump provides high-temperature water used for the space heating of the historical building provided with fan coil terminal units, rejecting heat to the ground through the use of six borehole heat exchangers. Data for temperatures and pressures at the heat exchangers are shown for some representative days of the warm and cold seasons, and the logic controls regulating the operations of the heat pump are explained. Even though the heat pump is a prototype built with devices and components normally used in commercial refrigeration and not for HVAC systems, the results in terms of COP and EER values are not different if compared with standard high-temperature double-stage heat pumps present in the market. Moreover, the case study presented has several positive characteristics and innovations, such as the use of low-GWP refrigerant, CO2 (R744), and R1234ze.

**Author Contributions:** Conceptualization, G.E.; methodology, G.E., S.B., L.C. and M.D.C.; validation, G.E., S.B., L.C., F.P. and A.T.; formal analysis, G.E., S.B. and L.C.; investigation, G.E., S.B. and F.P.; resources, F.P.; data curation, G.E., S.B. and F.P.; writing—original draft preparation, G.E. and S.B.; writing—review and editing, M.D.C., F.P., A.G., D.P., G.M. and A.B.; visualization, G.E. and S.B.; supervision, G.E., S.B. and A.B.; project administration, A.B.; funding acquisition, A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** Cheap-GSHPs project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No. 657982.

**Acknowledgments:** The authors thank many people that contributed, participated and supported this work. For the work performed at the Technical Museum Nikola Tesla in Zagreb, the director Franuli´c and Eng. Branimir Prgomet, Davor Trupokovi´c, Assistant Minister, Ministry of Culture of Croatia, Vladimir Soldo, Faculty of Mechanical Engineering and Naval Architecture University of Zagreb, Eng. Ante Pokupˇci´c and Arch. Željko Kovaˇci´c, designer of the technical room, heat pump installer Eng. Franjo Bani´c, Trojanovi´c local REHAU staff, Veljko Mihali´c, Ines Franov Beokovi´c, Narcisa Vrdoljak and Jasna Š´cavniˇcar Ivkovi´c, Municipality and City office of Zagreb, Vesna Drasal from E-kolektor services, Miljenko Sedlar, Deputy Principal of Department for project implementation, Rut Carek, Secretary General, Croatian Commission for UNESCO.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
