An Integrated HBIM Simulation Approach for Energy Retroﬁt of Historical Buildings Implemented in a Case Study of a Medieval Fortress in Italy

: The Italian building stock consists of buildings mainly constructed until the mid-20th century using pre-industrial construction techniques. These buildings require energy refurbishment that takes into account the preservation of their architectural heritage. In this view, this work studies an innovative integrated modelling and simulation framework consisting of the implementation of Historical Building Information Modeling (HBIM) for the energy retroﬁt of historical buildings with renewable geothermal HVAC system. To this aim, the ﬁeld case study is part of a medieval complex in Central Italy (Perugia), as representative ancient rural o ﬀ shore architecture in the European countryside. The system involves of a ground source heat pump, a water tank for thermal-energy storage connected to a low-temperature radiant system, and an air-handling unit. The building heating energy performance, typically inﬂuenced by thermal inertia in historical buildings, when coupled to the novel HVAC system, is comparatively assessed against a traditional scenario implementing a natural-gas boiler, and made inter-operative within the HBIM ad hoc platform. Results show that the innovative renewable energy system provides relevant beneﬁts while preserving minor visual and architectural impact within the historical complex, and also in terms of both energy saving, CO 2 emissions o ﬀ set, and operation costs compared to the traditional existing system. The integrated HBIM approach may e ﬀ ectively drive the path toward regeneration and re-functioning of heritage in Europe.


Introduction
Nowadays, sustainable energy and cost efficiency solutions are widely accomplished in new constructions. On the contrary, in the case of historical building retrofit, deep multidisciplinary interaction is required for the preservation of cultural heritage value. This value usually represents a relevant constraint to face during the whole construction path. Accordingly, tailored low-impact solutions, typically to be tailored with the case-specific boundary conditions, must be thanks to its multidisciplinary and integrated approach [21]. In fact, it consists of the application of the principles of BIM to historical buildings renovation and retrofit [22]. Historical buildings, indeed, present some critical issues, e.g., the absence of detailed documentation on the project or the presence of multiple layers over the years. In this view, HBIM methodology allows to integrate, in a single three-dimensional model, building technical, design, and historical data, namely information about the degradation, the historical phases, and the performed interventions. On the one hand, many existing recent studies involve the use of the HBIM methodology to study the degradation and conservation of the cultural heritage [23][24][25], including integration with photogrammetry [26] and Geographic Information System (GIS) [27]. On the other hand, studies focus on the introduction into the BIM methodology of recurring elements and families typical of historical buildings [28,29]. With the aim to develop a tool to collect heritage information from interdisciplinary stakeholders, Palomar et al. [30] designed an online work platform that connects the intrinsic HBIM database with heritage documentary databases, which could be further improved in terms of modelling historic structures. Bolognesi et al. [31] analyzed different workflows to create a mathematical model for reproducing the geometry of a real historical vault, which can be parameterized in a BIM environment, as one of the main relevant elements characterizing historical buildings. Accordingly, Gigliarelli et al. [32] stressed the need of further standardization and interoperability between software in the HBIM workflow to provide solutions for the built heritage conservation.
In this panorama, this study, developed within the context of the H2020 GEOFIT project, has a twofold purpose. Firstly, an HBIM platform is proposed to design and model the historical case study building providing a comprehensive tool for the analysis of historical building. The approach, indeed, takes into account both historical priorities and retrofit actions for the building. In this view, different building elements, typical of historical buildings, are created from scratch in the model. Therefore, the models of such historical building elements are available to bridge this gap of BIM modelling for historical buildings. Secondly, in this context of analysis, the performance and the feasibility of innovative GSHP systems, integrating a water-to-water heat pump with ground heat exchangers, are preliminarily assessed when installed as energy retrofit strategy in the Italian historical building case study of the project. According to the goals of the GEOFIT project, novel horizontal slinky type ground source heat exchangers (GHEXs) coupled with an adsorption heat pump will be implemented in the Italian pilot. They will be placed lying flat in loops in a trench, then backfilled with in-situ sediment or with primarily sandy sediment. The loops will be overlapping at half of the loop diameter. These innovative heat exchangers allow geothermal systems to be implemented when drilling is too operatively difficult or expensive. However, in this preliminary study, the performance of a standard ground source heat pump system with vertical U-tube GHEXs, as a low temperature heat source, is analyzed. Therefore, the outcomes of this study are used to evaluate the theoretical feasibility of replacing a gas boiler with a geothermal heat pump, rather than the expected performance of the real pilot system. The real geothermal system will be specifically assessed following its comprehensive design and implementation in future studies.

Methods
To evaluate the achievable energy performance improvement associated to the implementation of a geothermal system, building dynamic energy simulation was carried out.
Firstly, the architectural model of the case study building was developed via HBIM, to have an integrated building model. Therefore, it was imported in the dynamic simulation platform, where the HVAC system was further modelled. The data exchange between the BIM and the dynamic simulation software was provided via gbXML schema version 6.0.1 [33] import capability. This interoperability allows saving the time needed to redraw the building geometry. The dynamic simulation model was validated, to allow a reliable comparison of the performance of the innovative HVAC system with the existing one. Firstly, the building envelope performance was Energies 2020, 13, 2601 4 of 21 validated based on the thermal zone air temperature. Secondly, the whole building-system model was validated based on the heating gas consumption for the entire building. Finally, the model was simulated when implementing the existing gas boiler or the geothermal system, with a water-to-water heat pump and ground heat exchangers, as HVAC system. Results were compared in terms of energy consumption, cost analysis, and CO 2 emissions.

Case Study
The case study building represents an ex stable building built in the second half of the XIX century ( Figure 1). It is part of the medieval complex of Sant'Apollinare fortress in Perugia, Italy Figure 1c. The building was recently refurbished through seismic and energy renovations. It was converted to an office building and it currently hosts the offices of a university research center dedicated to energy and sustainable development.

Case Study
The case study building represents an ex stable building built in the second half of the XIX century ( Figure 1). It is part of the medieval complex of Sant'Apollinare fortress in Perugia, Italy Figure 1c. The building was recently refurbished through seismic and energy renovations. It was converted to an office building and it currently hosts the offices of a university research center dedicated to energy and sustainable development.
The building consists of three floors: two above ground and a basement, for a total usable area equal to about 240 m 2 and a heated volume of about 700 m 3 . The basement is used as a laboratory, while the upper floors are intended for offices.
The structure is composed of the original external bearing walls, completely preserved, which are made of a mix of local stone and bricks. During the retrofit, windows were improved with lowemissivity glass. Moreover, external coating insulation using local recycled cork panels (0.1 m) was added to improve building envelope thermal performance, in addition to the good thermal inertia of the existing massive masonry. However, cork panels were not glued to the historical masonry, to preserve the historical aspect of the building and respect the principle of reversibility, according to which the retrofit intervention must be removed without damaging the original envelope preserved [34]. Finally, the original wooden structure of the roof was retrofitted and partially renovated before re-installation in situ. The building retrofit procedure was also recently awarded with the Gold Certification by GBC Historic Building™ (HB) protocol [35] as very first case in the world. This environmental certification allowed monitoring the impacts of the whole refurbishment and construction process, in terms of both sustainability of the construction process and preservation of the cultural value of the artefact during the restoration process [36]. The GBC HB protocol, indeed, addresses the issue of the The building consists of three floors: two above ground and a basement, for a total usable area equal to about 240 m 2 and a heated volume of about 700 m 3 . The basement is used as a laboratory, while the upper floors are intended for offices.
The structure is composed of the original external bearing walls, completely preserved, which are made of a mix of local stone and bricks. During the retrofit, windows were improved with low-emissivity glass. Moreover, external coating insulation using local recycled cork panels (0.1 m) was added to improve building envelope thermal performance, in addition to the good thermal inertia of the existing massive masonry. However, cork panels were not glued to the historical masonry, to preserve the historical aspect of the building and respect the principle of reversibility, according to which the retrofit intervention must be removed without damaging the original envelope preserved [34]. Finally, the original wooden structure of the roof was retrofitted and partially renovated before re-installation in situ. The building retrofit procedure was also recently awarded with the Gold Certification by GBC Historic Building™ (HB) protocol [35] as very first case in the world. This environmental certification allowed monitoring the impacts of the whole refurbishment and construction process, in terms of both sustainability of the construction process and preservation of the cultural value of the artefact during the restoration process [36]. The GBC HB protocol, indeed, addresses the issue of the enhancement of the energy efficiency and cultural value of the building heritage. The combination of these two aspects drives the choice of passive and active retrofit solutions to be implemented in the building.

HBIM Model
The architectural model of the case study building was developed in the BIM design authoring platform Revit (Autodesk) [37]. The HBIM model was developed to faithfully reproduce every historical component of the building, as depicted in Figure 2. Therefore, several building elements that are generally recurrent in historical buildings were created in Revit to model the case study building, since they are currently not implemented in Revit library. Ad hoc families were modelled starting from the existing templates, when possible, to bridge this gap. As further detailed in Appendix A for allowing readers' replicability of the proposed framework, the models for the following components were developed: roof clay bent tiles and tiles, that a modeled as a linear family to be arranged in lines on the roof; • barrel vault; • ancient wooden door; • wooden frame with beams and joists for roof and floor.
Energies 2020, 13, x FOR PEER REVIEW 5 of 21 enhancement of the energy efficiency and cultural value of the building heritage. The combination of these two aspects drives the choice of passive and active retrofit solutions to be implemented in the building.

HBIM Model
The architectural model of the case study building was developed in the BIM design authoring platform Revit (Autodesk) [37]. The HBIM model was developed to faithfully reproduce every historical component of the building, as depicted in Figure 2. Therefore, several building elements that are generally recurrent in historical buildings were created in Revit to model the case study building, since they are currently not implemented in Revit library. Ad hoc families were modelled starting from the existing templates, when possible, to bridge this gap. As further detailed in Appendix A for allowing readers' replicability of the proposed framework, the models for the following components were developed:

Building Thermal-Energy Model
Numerical analysis was performed within EnergyPlus v8.9 [38] simulation environment with DesignBuilder graphical Interface [39]. The building thermal-energy model was derived directly from the related HBIM model made in Revit, while the HVAC system was modelled directly in DesignBuilder. To facilitate the compatibility with the dynamic simulation software, the architectural model was simplified in Revit before export, as shown in Figure 3a. The final geometry of the building thermal-energy model after importing the gbXML file into DesignBuilder is depicted in Figure 3b.
The building model was divided into 17 thermal zones, according to the intended use. Three of them are free running, namely the technical room lodging the heating and the domestic hot water systems (basement), the technical room for the electrical and automation system (basement), and the stairwell. The basement is in direct contact with the ground on the front side.

Building Thermal-Energy Model
Numerical analysis was performed within EnergyPlus v8.9 [38] simulation environment with DesignBuilder graphical Interface [39]. The building thermal-energy model was derived directly from the related HBIM model made in Revit, while the HVAC system was modelled directly in DesignBuilder. To facilitate the compatibility with the dynamic simulation software, the architectural model was simplified in Revit before export, as shown in Figure 3a. The final geometry of the building thermal-energy model after importing the gbXML file into DesignBuilder is depicted in Figure 3b.  (Table 1). Windows, which were replaced during the energy retrofit, consist of low-emissivity double pane glass with Argon in the cavity (4 mm -15 mm -4 mm), characterized by U-value equal to 1.317 W/m 2 K. Moreover, the characteristics of the main opaque envelope components are summarized in Table 2.
(a) (b)  The schedules of occupancy and internal heat gains associated to electrical and lighting equipment, as well as HVAC system operation, were inputted according to the real typical use of the building considering office as main intended use. These schedules follow a typical profile of an office building with peak load by the end of the morning and during mid-afternoon. The building is occupied only on working weekdays and the occupancy schedule is shown in Figure 4. For occupancy density, a value of 0.025 people/m 2 was set, which means one person every 40 m 2 . This value represents the maximum possible occupation of the building. For office equipment, the following values were considered: • power density equal to 2 W/m 2 , representing the maximum achievable value; • emitted radiant fraction equal to 0.10; • schedule based on the occupancy schedule.
Air infiltration was set equal to 0.3 ac/h, according to the ASHRAE method [40]. Finally, heating system set-point temperature was set equal to 20 °C in each thermal zone, according to indoor thermal comfort standards [41] and building simulation analysis practice [42], except for the unheated zones. The building model was divided into 17 thermal zones, according to the intended use. Three of them are free running, namely the technical room lodging the heating and the domestic hot water systems (basement), the technical room for the electrical and automation system (basement), and the stairwell. The basement is in direct contact with the ground on the front side.
As for the envelope materials, transparent surfaces were modelled in terms of thickness, transmittance, g-factor, and reflectance to solar radiation, emissivity, thermal conductivity, and visible transmittance (Table 1). Windows, which were replaced during the energy retrofit, consist of low-emissivity double pane glass with Argon in the cavity (4 mm -15 mm -4 mm), characterized by U-value equal to 1.317 W/m 2 K. Moreover, the characteristics of the main opaque envelope components are summarized in Table 2.  The schedules of occupancy and internal heat gains associated to electrical and lighting equipment, as well as HVAC system operation, were inputted according to the real typical use of the building considering office as main intended use. These schedules follow a typical profile of an office building with peak load by the end of the morning and during mid-afternoon. The building is occupied only on working weekdays and the occupancy schedule is shown in Figure 4. For occupancy density, a value of 0.025 people/m 2 was set, which means one person every 40 m 2 . This value represents the maximum possible occupation of the building. For office equipment, the following values were considered: • power density equal to 2 W/m 2 , representing the maximum achievable value; • emitted radiant fraction equal to 0.10; • schedule based on the occupancy schedule.
Air infiltration was set equal to 0.3 ac/h, according to the ASHRAE method [40]. Finally, heating system set-point temperature was set equal to 20 • C in each thermal zone, according to indoor thermal comfort standards [41] and building simulation analysis practice [42], except for the unheated zones.

HVAC System Model
Firstly, the HVAC system was modeled in order to best represent the actual system installed in the case study building. The heating load of the building is satisfied by a gas boiler connected to radiant floor in all rooms, except for service rooms where wall radiators are installed. The water

HVAC System Model
Firstly, the HVAC system was modeled in order to best represent the actual system installed in the case study building. The heating load of the building is satisfied by a gas boiler connected to radiant floor in all rooms, except for service rooms where wall radiators are installed. The water distribution to the various terminal units is carried out thanks to a small water storage tank. The tank is maintained at the required set-point temperature by means of a plate heat exchanger between the gas boiler and the tank itself. Moreover, a 2000 L thermally insulated water tank is installed for domestic hot water. However, since this study focuses on the analysis of the sole space heating need, it was not accounted for in this study.
Furthermore, an air-handling unit (AHU) is used to supply fresh air to any zones. The system allows to partially recirculate air with inlets in office rooms, in the first and second floor, and outlet in the service rooms. To this aim, the AHU includes a cross-flow heat exchanger for the exhaust air and a water heating coil, connected to the gas boiler, that conditions the mixed air.
Therefore, the space heating system was modelled via an air loop and three water loops. For the proper modelling, each system was divided into two complementary parts: one for the supply side and the other for the demand side [43]. The three water loops are depicted in Figure 5 and organized as follows: • Hot Water Loop 1: the loop on the heat generation side of the system with the gas boiler on the supply side and the plate heat exchanger on the demand side. The set-point temperature of the heat transfer fluid in the system is set to 68 • C, which is the temperature required by the wall radiators and the AHU heating coil. The set-point temperature required by the radiant floor is considerably lower, i.e., 38 • C. The daily operation profile of the HVAC system is defined based on occupancy schedules and metabolic activity.
The air loop is similar to the hydraulic loops, as shown in Figure 6. The air terminal units, namely diffusers, are set up on the demand side and connected to the thermal zones. On the other hand, the supply side is composed by a half loop, with elements in series representing the technological equipment of the AHU. Moreover, as previously mentioned, the economizer on the exhaust path of outdoor air mixer allows free-cooling, especially during the intermediate seasons, when indoor and outdoor air temperature and relative humidity look suitable for contributing to the thermal load reduction.   The set-point temperature of the heat transfer fluid in the system is set to 68 °C, which is the temperature required by the wall radiators and the AHU heating coil. The set-point temperature required by the radiant floor is considerably lower, i.e., 38 °C. The daily operation profile of the HVAC system is defined based on occupancy schedules and metabolic activity.
The air loop is similar to the hydraulic loops, as shown in Figure 6. The air terminal units, namely diffusers, are set up on the demand side and connected to the thermal zones. On the other hand, the supply side is composed by a half loop, with elements in series representing the technological equipment of the AHU. Moreover, as previously mentioned, the economizer on the exhaust path of outdoor air mixer allows free-cooling, especially during the intermediate seasons, when indoor and outdoor air temperature and relative humidity look suitable for contributing to the thermal load reduction.
The management of thermal load in the zones where both the radiant floor and AHU units are installed, is arranged by assigning the priority to the radiant system. Typically, almost all the heating load is satisfied by the radiant floor, leaving to the AHU its main function of fresh air supply. The main components of the HVAC model [44] are detailed as follows: 1. Hot Water Boiler: the gas boiler, which can be characterized in terms of efficiency curves and type of fuel used, e.g., natural gas in this case. 2. Fluid-to-Fluid Heat Exchanger: type of heat exchanger that models a general heat exchange from fluid to fluid, whose heat transfer rate and efficiency are described in Equations (1) and (2), respectively: where m is the mass flow rate, cp is the specific heat capacity, TSupLoop,in and TDmdLoop,in are the fluid temperatures entering the heat exchanger on the supply side and on the demand side of The management of thermal load in the zones where both the radiant floor and AHU units are installed, is arranged by assigning the priority to the radiant system. Typically, almost all the heating load is satisfied by the radiant floor, leaving to the AHU its main function of fresh air supply.
The main components of the HVAC model [44] are detailed as follows: 1. Hot Water Boiler: the gas boiler, which can be characterized in terms of efficiency curves and type of fuel used, e.g., natural gas in this case.

2.
Fluid-to-Fluid Heat Exchanger: type of heat exchanger that models a general heat exchange from fluid to fluid, whose heat transfer rate and efficiency are described in Equations (1) and (2), respectively: where . m is the mass flow rate, c p is the specific heat capacity, T SupLoop,in and T DmdLoop,in are the fluid temperatures entering the heat exchanger on the supply side and on the demand side of the loop, respectively, NTU is the Number of Transfer Units [44], and R c is the flow thermal capacity ratio [44].

3.
Mixed Water Thermal Tank: water heater with only water storage function and, therefore, without auxiliary heater capacity. Due to the small volume, it is not affected by stratification phenomena and simulates a well-mixed water tank. The storage energy balance equation is as follows (Equation (3)): where and c p are water density and specific heat, respectively, V is the tank volume, T is the temperature of the water tank, and q net is the net heat transfer rate to the tank water.

4.
Low Temperature Radiant Variable Flow: represents the radiant floor when integrated as a tailored type of internal source within floor construction layers to calculate the relative conduction transfer function (CTF).

5.
Water Convective Baseboard: radiators installed in service rooms, which split the thermal power emitted into radiant and convective. 6.
Water Heating Coil: downstream heating coil in the AHU, equipped with a water controller to vary the flow of hot water depending on the thermal load. 7.
Air-to-Air Sensible Heat Exchanger: sensible heat recovery for ventilation system, characterized by an efficiency equal to 0.8 in design conditions.
In order to check the correct sizing of the HVAC system, the building was simulated during the month of February. Figure 7 shows the trend of indoor air temperature, mean radiant temperature, and operative temperature with respect to outdoor dry-bulb temperature for a representative thermal zone with hourly time step. The proximity among operative temperature, mean radiant temperature, and air temperature trends especially during the occupied hours confirms the proper operation of the heating system, also thanks to building thermal insulation and inertia. Furthermore, during occupied periods, indoor air temperature is slightly higher than mean radiant temperature, thanks to internal heat gains, while it falls below it during the free-running hours. month of February. Figure 7 shows the trend of indoor air temperature, mean radiant temperature, and operative temperature with respect to outdoor dry-bulb temperature for a representative thermal zone with hourly time step. The proximity among operative temperature, mean radiant temperature, and air temperature trends especially during the occupied hours confirms the proper operation of the heating system, also thanks to building thermal insulation and inertia. Furthermore, during occupied periods, indoor air temperature is slightly higher than mean radiant temperature, thanks to internal heat gains, while it falls below it during the free-running hours.

Geothermal System Model
To evaluate the potential benefits associated to the proposed innovative HVAC system, the gas boiler was then replaced with a ground source heat pump (GSHP). Therefore, the energy model was modified by adding a vapor compression water-to-water heat pump, with thermal power of 15 kW, and GHEXs, while maintaining the storage water tank connected to the radiant system and the airhandling unit. Accordingly, an additional water loop was added to model the two vertical U-tube GHEXs, as depicted in Figure 8. The U-tubes were characterized by a depth of 70 m and a diameter of 0.03 m.

Geothermal System Model
To evaluate the potential benefits associated to the proposed innovative HVAC system, the gas boiler was then replaced with a ground source heat pump (GSHP). Therefore, the energy model was modified by adding a vapor compression water-to-water heat pump, with thermal power of 15 kW, and GHEXs, while maintaining the storage water tank connected to the radiant system and the air-handling unit. Accordingly, an additional water loop was added to model the two vertical U-tube GHEXs, as depicted in Figure 8  Two types of mathematical models co-operate within EnergyPlus to simulate the operation of a GSHP: (i) the curve-fit model [45] and (ii) the parameter estimation-based model [46]. The first, classified as an equation fit component model by Hamilton and Miller [45], treats the system as black box using four dimensionless curves, which are useful to predict the heat pump performance in heating and cooling mode. In detail, the generalized least square method generates a set of coefficients, from the catalogue data used in the model, to simulate the behavior of the heat pump [47]. The second is a method between the black box and the deterministic model [46]. Specifically, a simplified model, based on thermodynamic heat and mass balance equations [48], is applied to each internal component of the heat pump. The required parameter values are defined by means of a multi-variable optimization algorithm using manufacturers catalogue data [49]. This latter method is suitable for alternative compressor heat pumps. However, when varying the type of compressor of the heat pump, the evaluation of the parameters has a high level of uncertainty. Therefore, in this study, the mathematical model selected for the heat pump was the curve-fit model. The model uses two non-dimensional equations or curves to predict the heat pump performance in heating mode [47]. The generalized least square method was used to generate a set of performance coefficients from the heat pump catalogue data at indicated reference conditions. Thereafter, the respective coefficients Two types of mathematical models co-operate within EnergyPlus to simulate the operation of a GSHP: (i) the curve-fit model [45] and (ii) the parameter estimation-based model [46]. The first, classified as an equation fit component model by Hamilton and Miller [45], treats the system as black box using four dimensionless curves, which are useful to predict the heat pump performance in heating and cooling mode. In detail, the generalized least square method generates a set of coefficients, from the catalogue data used in the model, to simulate the behavior of the heat pump [47]. The second is a method between the black box and the deterministic model [46]. Specifically, a simplified model, based on thermodynamic heat and mass balance equations [48], is applied to each internal component of the heat pump. The required parameter values are defined by means of a multi-variable optimization algorithm using manufacturers catalogue data [49]. This latter method is suitable for alternative compressor heat pumps. However, when varying the type of compressor of the heat pump, the evaluation of the parameters has a high level of uncertainty. Therefore, in this study, the mathematical model selected for the heat pump was the curve-fit model. The model uses two non-dimensional equations or curves to predict the heat pump performance in heating mode [47]. The generalized least square method was used to generate a set of performance coefficients from the heat pump catalogue data at indicated reference conditions. Thereafter, the respective coefficients and indicated reference conditions were used in the model to simulate the heat pump performance. The variables influencing the water-to-water heat pump performance are (i) entering load side water temperature, (ii) entering source side water temperature, (iii) source side water flow rate, and (iv) load side water flow rate. The governing equations for the heating mode are Equations (4) and (5): where C1 and D5 are the equation fit coefficients, T L,in is the entering load side water temperature, T S,in is the entering source side water temperature, T ref is the reference temperature, equal to 283.15 K (temperature unit of Kelvin is used to ensure a positive value of the ratio between the water inlet temperature and reference temperature even if the water inlet temperature drops below the freezing point), The inlet variables are divided by the reference conditions to allow the coefficients falling into a smaller range of values. Moreover, the value of the coefficient indirectly represents the sensitivity of the output to the specific inlet variable. The reference conditions used in the model must be the same as the reference conditions used to generate the performance coefficients. Assuming no losses, the source side heat transfer rate in heating mode (Q source,h ) can be calculated through Equation (6): The heat pump modeled in this study has a reference COP equal to 5 and the equation fit coefficients used in the model are listed in Table 3.
The model used to represent ground thermal behavior was developed Kusuda and Achenebach [50] as a function of time and depth. Therefore, it is correlated to the average annual soil surface temperature, the amplitude of the soil temperature variation along the year or the day with minimum surface temperature, and to ground thermal diffusivity. For the ground temperature, the deep monthly temperature available at the case study location (according to the weather file from the EnergyPlus database [51]) was considered. It is equal to 14 • C in every month. Accordingly, the ground temperature variation is null in this case. The vertical U-Tube heat exchangers model is based on the work of Eskilson [52]. The problem of heat transmission for boreholes is solved by means of a mixed numerical and analytical method that determines the response factors under constant initial and boundary conditions over a long time period. Yavutzturk and Spliter [53] provided the integration of this model for a short time step by considering also the thermal capacity of ground and grout, the thermal resistance of the pipe, and the fluid flow. In this work, the thermal response of the vertical U-tube GHEXs was determined as a predefined dataset of response factors, based on the above-mentioned properties and the boreholes field configuration. The configuration of the boreholes field has a significant influence on the response factors, which must be evaluated before every simulation [54].

Model Validation
The building thermal-energy model was validated by verifying the reliability of both the sole building envelope modelling and the whole building-system model. For the envelope performance validation, a specific weather file was developed by considering in-field monitored data in the month of August 2018. In fact, it was the only period in free-running conditions and without building occupancy. The climatic data for the construction of the weather file were collected thanks to a dedicated on-site weather station, which allows measuring outdoor dry-bulb air temperature and relative humidity, used to develop the weather file, and additional parameters. On the other hand, for the annual building-system model validation, a specific weather file was developed by considering in-field monitored data by the same weather station during the whole year 2018. Therefore, the annual dynamic simulations for the assessment of building thermal-energy performance were carried out when using the typical meteorological year (TMY) weather file from the EnergyPlus database [51].
For the envelope performance validation, the measured indoor air temperature was compared with the simulated one for selected thermal zones with available monitored data. In fact, the building is equipped with an integrated monitoring system able to measure air temperature ( • C), relative humidity (%), and velocity (m/s) in different office rooms within the building. The indoor monitoring setup is based on a XWeb Monitoring system that allows collecting the mentioned parameters with a frequency up to 10 min. The model accuracy was evaluated by means of the NMBE (Normalized Mean Bias Error) and CV(RMSE) (Variation of the Root Mean Square Error) indices, shown in Equations (7) and (8), respectively. Their analytical expression is represented as follows: where V states for value. These statistical indicators quantify the discrepancy between the simulated output data and the measured data [40]. According to the indications of the ASHRAE Guideline 14, a model can be considered calibrated when CV(RMSE) < 30% and NMBE < 10% [55], when sub-hourly values are considered. The annual building-system model validation was carried out through direct comparison with the data of gas consumption for heating available for the case study building, which cover one year following the commissioning of the system and the occupation of the building after the retrofit. For the purpose of this additional validation, the same indices were calculated when comparing the actual energy performance with the simulated one.

Building Envelope Performance Validation and Modelling
The building integrated modelling and multidimensional BIM integration represents the key general outcome of the paper, which procedure may be replicated for guiding and inspiring energy efficient, sustainable, and comfortable retrofits of historic buildings. Therefore, the whole modelling effort and replicable reported procedure included in the previous section represents itself a "lesson learnt" result of the research.
As regards the building envelope performance validation, the comparison between the simulated indoor air temperature and the values measured by the monitoring system shows an excellent accuracy of the model in the evaluated month, with NMBE equal to 2% and CV(RMSE) equal to 5%. The results of the validation for the monitored thermal zone mostly exposed to weather conditions in summer, i.e., an office room in the second floor above ground, are depicted in Figure 9 for the whole month of August 2018. Nevertheless, the building-system model behaves acceptably with a CV(RMSE) equal to 5% over a monthly period (with hourly time step simulations). Moreover, the NBME is equal to 4.4% and rather within the limits. Since at least three consecutive years of monitored consumption data are not yet available, it is not possible at this stage to definite this as an accurate validation. However, this annual validation allows verifying the consistency of the model with the same order of magnitude of the real building heating consumption.

Building-System Model Validation
As for the building-system model validation, the gas consumption data recorded in the first year of operation, shown in Figure 10, include the periods of commissioning and initial start-up of the HVAC system during December 2017 and the further commissioning during February-March 2018, when the outputs of the simulation model deviate significantly from the real data.   Nevertheless, the building-system model behaves acceptably with a CV(RMSE) equal to 5% over a monthly period (with hourly time step simulations). Moreover, the NBME is equal to 4.4% and rather within the limits. Since at least three consecutive years of monitored consumption data are not yet available, it is not possible at this stage to definite this as an accurate validation. However, this annual validation allows verifying the consistency of the model with the same order of magnitude of the real building heating consumption.

Building Energy Performance
The dynamic thermal-energy simulation shows that the predicted annual electricity demand for heating the case study building with the novel GSHP system is equal to 2920 kWh. The monthly demand distribution is reported in the Figure 11. Therefore, the building energy performance with the water-to-water heat pump and ground heat exchangers as heating system is compared with the actual scenario (gas boiler) in terms of source energy demand. Results, depicted in Figure 12, show an achievable energy saving for space heating up to more than 70% with the novel GSHP system, thanks to the exploitation of renewable ground thermal potential and the higher efficiency of this system. This finding stresses the relevant energy efficiency potential of the proposed innovative system, with total annual energy saving equal to about 12,121 kWh. actual scenario (gas boiler) in terms of source energy demand. Results, depicted in Figure 12, show an achievable energy saving for space heating up to more than 70% with the novel GSHP system, thanks to the exploitation of renewable ground thermal potential and the higher efficiency of this system. This finding stresses the relevant energy efficiency potential of the proposed innovative system, with total annual energy saving equal to about 12,121 kWh. Moreover, the two systems were compared in terms of economic and environmental benefits to provide a comprehensive operational assessment of the GSHP system. Specific whole economic assessment will be carried out once the technology is fully available into the market at the end of the project. Nevertheless, current operating-phase results confirm the significantly improved performance also in terms of costs and CO2 emissions savings. For the cost analysis, an average global cost equal to 0.22 €/kWh for electricity and an average global cost equal to 0.8 €/m 3 for natural gas were considered according to the energy market prices. Moreover, considering the current national energy mix, the consumption of electricity from the grid involves 0.44 kg/kWh of CO2 emissions, while the combustion of 1 m 3 of natural gas is responsible for 1.89 kg of CO2 emissions. The achievable economic and CO2 savings associated to the implementation of the GSHP system are depicted in Figure 13. In the case study context, the replacement of a gas boiler with a GSHP system allows annual CO2 savings associated to the need of space heating equal to 3032 kg CO2. Additionally, annual cost savings for heating are up to the 69%, corresponding to 1336 €/year.   Moreover, the two systems were compared in terms of economic and environmental benefits to provide a comprehensive operational assessment of the GSHP system. Specific whole economic assessment will be carried out once the technology is fully available into the market at the end of the project. Nevertheless, current operating-phase results confirm the significantly improved performance also in terms of costs and CO 2 emissions savings. For the cost analysis, an average global cost equal to 0.22 €/kWh for electricity and an average global cost equal to 0.8 €/m 3 for natural gas were considered according to the energy market prices. Moreover, considering the current national energy mix, the consumption of electricity from the grid involves 0.44 kg/kWh of CO 2 emissions, while the combustion of 1 m 3 of natural gas is responsible for 1.89 kg of CO 2 emissions. The achievable economic and CO 2 savings associated to the implementation of the GSHP system are depicted in Figure 13. In the case study context, the replacement of a gas boiler with a GSHP system allows annual CO 2 savings associated to the need of space heating equal to 3032 kg CO 2 . Additionally, annual cost savings for heating are up to the 69%, corresponding to 1336 €/year. Figure 12. Building heating energy saving in terms of monthly source energy demand with the ground source heat pump system compared to the gas boiler. Figure 13. Building annual savings for space heating operation costs and CO2 emission with the ground source heat pump system compared to the gas boiler.

Conclusions and Future Developments
Given the need to develop effective and low-impact solutions for retrofitting historical buildings, this work aimed at investigating a new holistic modelling and simulation framework working as a support tool for design and implementation of energy retrofits in historical buildings. An innovative geothermal HVAC technology and novel procedures of analysis specifically tailored for historical buildings were indeed integrated within a tailored HBIM approach. Therefore, the energy, environmental, and economic performance of a GSHP system was assessed as potentially ready for implementation in historical building retrofit, as heating system. This system was developed within the context of the ongoing H2020 GEOFIT project.
Building energy dynamic simulation was carried out to compare building performance with the ground source heat pump system with respect to the existing gas boiler. Results show considerable benefits associated to the implementation of the innovative system in terms of source energy savings (up to 73% heating energy need reduction), while maintaining the same operation and comfort Figure 13. Building annual savings for space heating operation costs and CO 2 emission with the ground source heat pump system compared to the gas boiler.

Conclusions and Future Developments
Given the need to develop effective and low-impact solutions for retrofitting historical buildings, this work aimed at investigating a new holistic modelling and simulation framework working as a support tool for design and implementation of energy retrofits in historical buildings. An innovative geothermal HVAC technology and novel procedures of analysis specifically tailored for historical buildings were indeed integrated within a tailored HBIM approach. Therefore, the energy, environmental, and economic performance of a GSHP system was assessed as potentially ready for implementation in historical building retrofit, as heating system. This system was developed within the context of the ongoing H2020 GEOFIT project.
Building energy dynamic simulation was carried out to compare building performance with the ground source heat pump system with respect to the existing gas boiler. Results show considerable benefits associated to the implementation of the innovative system in terms of source energy savings (up to 73% heating energy need reduction), while maintaining the same operation and comfort conditions for the occupants. Furthermore, CO 2 emissions savings reach about 69% of the total and operating costs are significantly cut accordingly. Therefore, the benefits achievable through the implementation of the proposed geothermal system in historical buildings are both energetic, economic, and environmental, thanks to the exploitation of renewable heat from the ground without compromising the architectural heritage of the complex.
A new HBIM platform is proposed to model the case study historical building. This modelling approach aims at considering historical building priorities together with energy/structural retrofit actions. The same level of careful detail is therefore reached through the proposed inter-operative modelling, design, and implementation. The proposed methodology demonstrated its effectiveness and will be implemented into the GEOFIT project with the purpose to act as demonstrator for historical building retrofits and foster similar interventions in Europe and outside EU. In fact, it combines indoor wellbeing, environmental sustainability, architectural preservation but also energy efficiency, building functionality, structural safety, and reliability, in order to preserve cultural heritage through an holistic approach. It indeed allows to take advantage of the same cultural heritage peculiarities for the development of new technologies with low architectural and environmental impact, such as the proposed geothermal and storage systems. This model will be further developed during the project thanks to the integration within a GEOBIM platform, i.e., an integrated environment where architectural and technical information are integrated with geographical information covering also the geothermal conditions of the building site. Additionally, economic assessments will be also carried out, once the technology implementation is finalized, in order to further integrate a whole evaluation protocol into the proposed platform, including maintenance and economic aspects, representing future developments of the work.
Another future development consists of the effective implementation of the real ground source heat pump system, involving horizontal slinky type GHEXs, in the case study historical building and the experimental assessment of the pilot performance.

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.

Appendix A
There are building elements recurrent in the majority of historical buildings and at the moment there is a lack of families in the Revit library [37] or downloadable from the web modeling them. This underlines that the software for BIM modeling are currently mainly used for modeling new buildings with a high degree of standardization. Therefore, in this study ad hoc families were created in Revit, starting from the templates provided by the program database, to represent these common components of historical buildings.
The first element is the column, which is modeled starting from the "Metric Column" template. The pedestal of the column is modeled using the command "Extrusion", while for the column itself the command "Revolution" is used. The top and bottom of both column and pedestal are constrained at a proper reference plane. The distances between the reference planes are then parameterized. Once the family is loaded into a project and placed with the command "Place Component" at the specific level, the height of the pillar is fixed with the command "Attach top/base".
The roof modeling is divided into two phases: the model of the slab and that one of the tiles. First, the stratigraphy of the roof is created from the "System family: base roof". Thereafter, the roof pitches are defined with the commands "Perimeter roof" and "Modify secondary elements". Roof clay bent tiles and tiles require the use of multiple families, nested together. Uppermost, four profile families are created with the template "Metric Profile". Then, the obtained profiles (two for the bent tile and two for the tile) are imported in two new families, starting from "Metric Generic Model" family template. Starting from two profiles, the "Solid Sweep Blend" command is used to achieve a single tile. The family containing the single tile is then loaded within a new family created from the "Metric generic model line based" template. Therefore, an array of tiles is created, and the number of tiles is parameterized in function of the length of the array itself. Finally, the grid for the positioning of the linear families is created through the family "Curtain Wall". The two linear families are then placed with the command "Place Component".
The barrel vault family is created starting from the template "Generic model line based". An empty volume is created with the command "Void-Extrusion". When the barrel vault family is positioned, this void cuts the host element. In order to create a family that can be used in multiple projects, the dimensions of the vault are parameterized. The extrusion representing the intrados of bricks of the vault is created using the same procedure. In the "Parameters and Categories of Families", the option "Cut with Voids when Loaded" is selected. Therefore, the barrel vault is placed with the command "Place Component" at the proper level.
The ancient wooden door is created through a slightly different procedure, namely using a "Local Model". The appropriate voids and solids are created with the command "Creates".
Finally, the wooden frame does not require the creation of ad hoc families. Firstly, the "Wood" family located in Revit "Italy" library inside the "Structural Frame" folder and the "Wood" subfolder is loaded. From this family, three types of needed beams are created. The "Ceiling Plants" views are fundamental for the correct positioning of the beams. Thereafter, in the "Structure" tab, the "Beam System" command is selected, and the lower surface of the floor is set as the reference plane. At this point, the outline of the area is drawn with the command "Outline" and the direction of the beams is specified with the command "Beam Direction".