Ground temperature monitoring for a coaxial geothermal heat exchangers ﬁ eld: practical aspects and main issues from the ﬁ rst year of measurements

Ground temperature at shallow depths (< 50 m) is not stable, nor in space, neither in time, and its behaviour is the result of the superimposition of e ﬀ ects of heat pulses of di ﬀ erent origin: solar, geothermal and anthropic. The correct assess-ment of ground temperature is a crucial point when designing a shallow geothermal energy system. In geothermal closed loop projects using short borehole heat exchangers, the ground temperature has more variability and a ﬀ ects the rate of heat extraction/injection. Monitoring of the ground temperature can therefore be useful in ground source heat pump projects to correctly understand the behaviour of a shallow geothermal reservoir subjected to heat extraction/injection. This paper illustrates the practical aspects and main issues occurred in the installation, testing and working phases of a monitoring system realised to record ground temperature in a geothermal application. The case study is a ﬁ eld of eight coaxial borehole heat exchangers, 30 m long, connected to a novel prototype of dual source (air and ground) heat pump.


Introduction
A closed loop geothermal circuit is designed to exchange heat with the ground within a specifi ed volume (Eskilson, 1987). Rocks, soils and groundwater are used to provide heat to the buildings, generally by ground source heat pumps (GSHP), or to receive and store their excess heat (Magraner et al., 2010). Shallow geothermal systems are designed to allow operational repeatability over the years, thereby avoiding thermal depletion of the soil (Focaccia et al., 2016). Due to the insulation from ambient weather, underground temperature is stable, thus enhancing the effi ciency and capability of the energy transfer in the heat pumps (Florides et al., 2013). The standard depth of a vertical closed loop borehole heat exchanger lies between 50 and 150 m, allowing a high portion of heat exchange surface to be in contact with aquifers, soils and rocks at a stable temperature (Aresti et al., 2018). Drilling work down to 50 m depth and beyond is the major cost of the entire system and it negatively impacts the market of shallow geothermal components: geo-exchangers and GSHP . Several countries and regional administrations put into action incentives and environmental laws to support the diffusion of this kind of environmentally friendly energy technology (Giambastiani et al., 2014). Mature markets exist in some countries, such as Sweden, Netherlands and Germany, but the GSHP systems are far from becoming an air conditioning standard technology. This is mainly due to the competition with district heating and cooling, natural gas boilers and air source heat pumps, which are environmentally friendly as well. The competition is becoming harder in recent years because of improvements in machinery effi ciency. Therefore, new solutions had to be thought of, implemented and tested on the geo-exchanger components, too.
Coaxial borehole heat exchangers (CBHE) are particularly suitable to be installed at low depths, thus reducing installation costs. The heat exchange surface is bigger with respect to single-U or double-U pipe BHE, but the reachable depth is limited. The H2020 GEO-TeCH Project aims to further decrease the installation costs of CBHE, by effi ciently adapting to GSHP specificities in regard to the hollow stem auger drilling technology, for the replacement of standard drilling techniques in alluvial plains . A second GEO-TeCH innovation, the dual source (air and ground) heat pump (DSHP), was supposed to be integrated with The Mining-Geology-Petroleum Engineering Bulletin and the authors ©, 2018, pp. 47-57, DOI: 10.17794/rgn.2018.5.5 CBHE, for a further decrease in installation costs. Using DSHP, it is theoretically possible to undersize the CBHE fi eld thanks to the combined use of air and ground sources. The active length of vertical CBHE, in contact with aquifers at stable ground temperature, can then be reduced, too. In consequence, the ambient seasonal variability can no longer be neglected. As an example, at least 30% of the active length of a 50 m deep CBHE, is located above the "neutral zone," the layer where ground temperature is not disturbed by ambient seasonality. This means that at least 30% of CBHE works with timevarying temperatures of the ground (Kurevija and Vulin, 2010).
Many knowledge gaps still exist for the correct design of DHSP linked to CBHE. Among them, it is worth noticing the following: • How does the ground behave in the presence of strong, unconventional, heat pulses (higher than in standard BHE design) but interspersed with frequent shutdowns (because of the activation of an air source)? • What will the effi ciency of the DSHP be on the short and long term? Will it be smaller, similar or higher than the standard GSHP? • To what extent will the impact of ambient seasonal variability on effi ciency of the system be? Will the negative impact of time varying temperature during DSHP ground working mode be bigger than the positive recovery to natural state during the DSHP air working mode? The CBHE behaviour can be accurately reproduced by simulating both the short-term and long-term response of the shallow geothermal reservoir. The short-term response is especially important in systems with high on/off operations, such as DSHP ones. For this purpose, the temperature variation of the surrounding ground must be well predicted, and it will depend both on the heat injected or extracted, and also on the ground thermal properties and the operating conditions of the BHE.
The thermal response of the ground and the amount of soil affected by the heat injection during a specifi c time period can be calculated and simulated in different ways. As an example, a possibility is to add a number of radial ground nodes and discretize the soil mass in small radial steps until the far-fi eld radius (the "penetration radius"), where the effect of the heat injection vanishes. When numerically modelling the heat transfer between a BHE and the ground, for each specifi c time period, the correspondent penetration radius should be suitable to allow reproducing the behaviour of the system accurately (Ruiz Calvo et al., 2015). For this purpose, experimental measurements (temperature and fl ow) of circulating fl uid inside the CBHE system and measurements of the ground temperature at different distances from its centre are necessary.
Monitoring ground temperature is an important issue in the shallow geothermal sector, both for research and professional purposes. Due to technology advancements and a decrease in costs, electronic measurement, registration and data transmission systems have been widely applied for many purposes, such as the validation of models (Tinti et al., 2017;Badenes et al., 2017), the control of system behaviour and effi ciencies (Montagud et al., 2011) and the management of the resource by local and regional environmental authorities (Hähnlein et al., 2013).
This paper shows the practical aspects and main issues arisen of a monitoring system of ground temperature installed on the fi rst prototype of DSHP connected to CBHE of the GEOTeCH Project (www.geotech-project.eu), whose results will be used for the validation of the coaxial borehole heat exchanger model.

Monitoring system used
To validate the models of DSHP fed by CBHE, a monitoring system of ground temperatures was prepared. The system is installed on-site to monitor the ground temperature and consists of an innovative tool, called Therm Array, based on Modular Underground Monitoring System (MUMS) technology developed by ASE S.r.l. This approach relies on a combination of sensors embedded in specifi cally moulded nodes, called Links, connected by an aramid fi bre and an electrical cable, thus forming an arbitrary long chain (Segalini et al., 2013;Segalini et al., 2014). Links can be customized, according to the situation, with different sensors able to record quantities such as displacements, water level variations and temperature. The whole monitoring apparatus is connected to an ASE801 Control Unit, which queries each different Link with an appropriate sampling frequency that could be changed accordingly to the monitoring needs (see Figure 1). Data collected is stored locally on a memory unit and sent to the mainframe server at the elaboration centre, where it is stored in a dynamic MySQL database with a daily multilevel backup system. Upon arrival on the central server, raw data is automatically elaborated and converted into physical units with a proprietary software routine specifi cally de- veloped for this purpose. The results are stored in a "parallel" database from which they can be accessed and analysed thanks to a dedicated web-based platform with private access, together with an FTP transmission to the end user.
In particular, the Therm Array presented in this paper equips a high-resolution thermometer (Therm Link) specifi cally designed for geothermal applications where it is necessary to monitor the soil temperature at different depths with high resolution. The sensor calibration is carried out thanks to a climate chamber where the instrumentation can be tested at different temperatures. The accuracy of the Therm Link is ±0.5°C on an operating range of -40 to +105°C; the resolution is 0.0078°C and the measure repeatability is ±0.015°C. It should be underlined that the resolution of the thermometer implemented in the monitoring system has particular relevance in this specifi c case study, since the instrumentation's main purpose is the measurement of temperature variations. In addition to the monitoring system previously presented, PT100 Class A thermoresistance probes were installed in order to monitor the temperature of water within the pipes fl owing inside the geothermal fi eld. These sensors operate according to the resistance measurement principle, relying on the fact that the material composing the probe features a well-known resistance-temperature relationship. In particular, as far as the PT100 sensors are concerned, this relationship is defi ned by an approximately linear trend, with a tolerance of 0.15°C at 0°C. (www.capitindustria.eu)

Tribano demo site
The fi rst prototype of new technology, together with appropriate monitoring, is located in the alluvial Po Plain, in the car park area adjacent to the HIREF S.p.A. factory (the producer and tester of DSHP), in Tribano (Province of Padua, Italy). In the Veneto Region, the use of pure water, as a circulating fl uid, is generally better accepted instead of additive water. When additives are necessary, only the use of propylene glycol is allowed, with a maximum concentration of 20% in volume. Distinctions in terms of legal authorization requests exist in Italy between vertical drilling shallower or deeper than 30 m. Moreover, when the drilling depth is more than 30 m, geologists should send information about the crossed soil layers to the National Institute for the Environmental Protection and Research (ISPRA -Istituto Nazionale per la Protezione e la Ricerca Ambientale) in Rome, according to the Italian Law 464/84 (Gazzetta Uffi ciale della Repubblica Italiana, 1984). In the Tribano demo site, propylene glycol is not used and the drilling depth is 30 m. Furthermore, the demo site is located far away from the two groundwater source protection areas of the Padua Province: High Padua (supplying drinking water to the city) and Euganean Hills (with natural thermal springs). The geology of the study area is comprised in the shallowest layers of unconsolidated, alluvial soils, with very fi ne grain size -silt and clay -of low permeability, locally interspersed by sandy layers of thickness: 10 m (8-18 m depth), 5 m (21-26 m depth), 3 m (44-47 m depth) and 16 m (58-74 m depth) (Pomarin, 2017). A phreatic aquifer can be found at around 2.0 m below the ground surface; a series of overlapped confi ned aquifers are crossed starting from a depth of 50 m. A hydrogeological study proved that drilling down to 30 m only crosses phreatic aquifers of poor quality, not suitable for drinking uses. Grouting was therefore not necessary for this geothermal system. However, its addition is always suggested to ensure a good thermal contact between the BHE and the ground. At the moment of the prototype installation, the GEOTeCH hollow stem auger drilling system was still not equipped with grouting devices. So, for this installation, grout was not injected and the procedure implied the ground collapsing around each CBHE, after removal of the drilling rods, which contained it. Due to low permeability of sediments along the borehole length, a signifi cant thermal impact due to advection phenomena for groundwater movement is not expected. Finally, the Tribano demo site is not in a potential area of archaeological fi ndings. The material excavated from the boreholes was mostly very fi ne sand with some clayey banding in the top of the borehole. All 8 CBHEs were connected to fl ow and return headers in the concrete collector pit. After the complete assembly of the borehole heat exchanger pipes, the whole system was fl ushed with water and pressure tested up to 6 bars. The demo site technical specifi cations are reported in Table 1.
Furthermore, 3 additional observation boreholes (OB) were realized to 15 m depth. They were equipped with pipes type PE100 SDR11 OD63, with the scope of hosting the Therm Links. The three OBs were installed in a straight line, two west (OB 1 and OB 2) and one east (OB 3) to BHE 8. This confi guration was chosen in order to have control points for the defi nition of penetration diameters, according to the coaxial borehole heat exchanger thermal modelling of Cazorla et al., 2018. BHE 8 was selected being at the eastern vertex of the borehole heat exchanger fi eld. In that way, OB 1 and OB 2, west to BHE 8, are located inside the area infl uenced by heat transfer between BHE and the ground. On the contrary, OB 3, east to BHE 8, is located at the external border of the infl uenced area. OB 1 is located exactly in the middle between BHE 6 and BHE 8, so major phenomena of superimposition of thermal effects are expected to be measured and registered in the long term. OB 2 is located as close as possible to BHE 8, with the aim to measure and register heat wave behaviour around the BHE for the defi nition of the penetration diameter. OB 3, located three meters away from the borehole heat exchanger fi eld, is used as reference point of the undisturbed ground thermal behaviour; no signifi cant temperature changes from natural sinusoidal behaviour are ex-The Mining-Geology-Petroleum Engineering Bulletin and the authors ©, 2018, pp. 47-57, DOI: 10.17794/rgn.2018.5.5 pected to be measured by Therm Arrays located in OB 3. Along each observation borehole, a Therm Array with 4 Therm Link was inserted. The four Therm Links were positioned at depths of 2, 5, 10 and 15 m each. Unfortunately, the pipe installed in OB 3 could not reach the same depth of OB 1 and OB 2, but it stopped half a meter above. In OB 3, Therm Links were correctly positioned at 2, 5 and 10 m, while the last Therm Link was laid down at pipe bottom. Finally, in order to correctly relate ground thermal behaviour with effective working of DHSP system, four measurement points on the circuit have been installed, by the use of PT100 sensors. PT100 have been installed: two on the head of BHE 8 (inlet and outlet) and two on the collector (inlet and outlet), the last two measuring the mixed water temperature from/to the 8 BHEs. The details and geometry of the monitoring system installed in Tribano is presented in Table 2.   Figure 2 shows the map of Tribano demo site, with evidence of location of 8 BHEs, 3 OBs, COL and DSHP. Figure 3 shows the details of installation.

Results
The monitoring system started operating from mid-November 2017, with an acquisition frequency of 5 minutes. This frequency was chosen to detect the intervals of switching on and off operations of the DSHP ground circuit. Therefore, from the 15 th of November 2017, it was possible to monitor the ground thermal behaviour. Figure 4 reports the ground temperature monitored from the 15 th of November 2017 to the 15 th of September 2018 (10 months).
Values of circuit temperature monitored by PT100 should be shown together with the working data of heat pump and are not the object of this paper.

Discussion
Ground temperature measured waves apparently have the same space-time behaviour in the three OB, and respect the typical sinusoidal model of Equation 1 (Baggs, 1983).
(1)  1, the thermophysical properties of ground layers (down to 15 m) and the geothermal heat fl ow in the area, the results of the ground temperature model do not match with the measurements (see Figure 5).
It was necessary then to analyse the curves separately. Three different data sets were tested: 1. Ground temperature at 2 m depth seems to be following a sinusoidal behaviour with an average of 17.5°C, amplitude 6.5 °C (which means, using the analytical model, an ambient amplitude of 10.5°C) and a time of a minimum of 40 days. By using such values, it leads to completely inaccurate results at 5, 10 and 15 m depth (see Figure 6). 2. Ground temperature at 5 m depth, on the contrary, seems to be following a sinusoidal behaviour with an average of 15°C, amplitude 4°C (which corresponds to an ambient amplitude of 6°C) and a time of minimum of 30 days. By using such values, it leads to quietly accurate results of the lower part of the sinusoidal wave at 2 m, while the wave at 10 and 15 m depth appears time shifted (see Figure 7).

UNMIG, 2017
Figure 5: Analytical model with standard climate data set applied to: OB 1 (upper left), OB 2 (upper right) and OB 3 (lower right). No correspondence can be found with the ground temperature measurements 3. Ground temperature at 10 and 15 m depth, fi nally, seems to be following a sinusoidal behaviour with the same average of ground temperature at 5 m depth, but a time of a minimum of 315 days (see Figure 8). The Root Mean Square Error (RMSE) between model results and measurements was calculated. The four different data sets of initial parameters were used to verify how well the analytical model fi ts with measured data: The   Standard Climate Data Set (see Figure 5), Data Set 1 (see Figure 6), Data Set 2 (see Figure 7) and Data Set 3 (see Figure 8).
RMSE results show that, using different initial parameters from the standard climate data set of Tribano, the fi rst part of the ground temperature measurements (roughly up to the month of May 2018) can be approximated by the analytical model, of sinusoidal behaviour, of Equation 1. The discrepancy between the climate data and ground temperature measurements should be explained by heat waves of anthropic origin (for example considering a heat loss to the ground, and subsequent storage, due to the works for the installation of the BHE fi eld and the parking lot). Further investigations will be approached by a multi-year measurement campaign, with more stable thermal conditions. On the contrary, the second half of the measured temperature (starting from May 2018) does not follow the model anymore. Temperature in all points is increasing, several degrees higher than the expected temperature by the analytical model, phenomenon particularly evident at 2 m depth. An anomalous heat pulse from the ground surface should be added, with subsequent damping with depth. The most probable explanation relies on the cover material used for the parking lot, black asphalt, subjected to sun radiation all day, especially in the summer season (see Figure 3). The past summer (June-September), there were 72 sunny days (75% of the total), with an average daily temperature of 27°C and peaks of 37°C. However, the daily impact is supposed to disappear at a depth of 1.5 -2.0 m, so the heat wave behaviour measured in the ground cannot be fully explained at the moment and further investigations will be undertaken. OB 2 is the nearest OB to the BHE 8, and so the most affected by heat pump operations heat pump operations. From May to September, the dual source heat pump worked regularly, in cooling mode, thus injecting heat to the ground. In OB 2, this effect is clearly visible, with temperature values at 5, 10 and 15 m depth higher of around 1°C than values measured in OB 1 and OB 3. Temperature measurements in the period 11 -27 July should not be taken into consideration, since the monitoring system faced unexpected issues, which were later solved. Measured values of OB 2 will be useful for the quantifi cation of penetration diameters of heat pulse from CBHE to the undisturbed ground. In the winter period, the work of DSHP was more discontinuous, being at the very initial and testing phase. Therefore, less evidence of geothermal heat pump operations could be found in the thermal footprint measured underground. Nonetheless, a decrease of 0.5°C from standard behaviour can be perceived at 5, 10 and 15 m depth from February 2018 to April 2018 in OB 2, which does not appear in OB 1 and OB 3. At 5 m depth, in OB 2, the temperature reaches a minimum of 13.5°C, while in OB 1 and OB 3 the minimum is no lower than 14°C. The setup and testing phase of the monitoring sys-tem faced some unexpected issues. In particular, a Digital Multiplexer was present at the beginning of the monitoring, which burned and was replaced several times. It was observed that the circulating water was electrostatically charged and introduced electrical shocks into the system. The problem was solved by moving the system away from the wells and discharging the current to the ground. After some months, the control units started to show a series of anomalies, resulting in several locks. The problems repeated over a long period of time when the cause was fi nally detected in a 380V three-phase transformer. This one was positioned next to the control unit and discharged the current on the metal bottom plate, which, in turn, was fi xed to the UMTS for data transmission. The router carried the mass through the power supply to the control unit, effectively blocking it. The problem was solved by disconnecting the router power supply and transmitting data via LAN network. This case is an example of the fact that any monitoring system, even advanced, can present problems linked to the real conditions of use that were previously unthinkable and diffi cult to identify on-site.

Conclusion
This paper presented the practical aspects related to the installation of a temperature monitoring system with the aim of understanding thermal behaviour of underground subjected to the work of a prototype of a dual source heat pump connected to a fi eld of shallow coaxial borehole heat exchangers. The setup of the monitoring system faced some problems linked to specifi c conditions of the test site, whose solution, presented in the paper, can help practitioners and BHE installers. The preliminary monitoring measurements of ground temperature evidenced a different behaviour from the one expected by applying the standard climatic model. The thermal impact on the ground of anthropic origin seemed to be affecting the ground thermal behaviour for the time period considered, resulting in an average temperature value higher and an amplitude value lower than the ones obtained by the analytic climate model. The most probable explanation for this phenomenon relies on the black asphalt of the parking lot, which substituted natural ground at the surface level. After the ignition of the dual source heat pump (in winter mode: preliminary testing, and in cooling mode: working at full capacity), the thermal effect to the ground could have been detected by the The Mining-Geology-Petroleum Engineering Bulletin and the authors ©, 2018, pp. 47-57, DOI: 10.17794/rgn.2018.5.5 monitoring borehole located 1 m away from one BHE. On the other hand, it seems that the heat wave did not have a signifi cant effect on the ground inside the BHE fi eld. Long term, multiyear, monitoring will be useful to appreciate the broad effect of the working of the system on the volume of the ground interested by the presence of BHEs. Further analysis on the records of temperature, fl ow rate of the circulating water in the pipe circuit and the heat pump parameters will be aggregated to the ground temperature data for a comprehensive assessment of the thermal behaviour of the shallow geothermal fi eld. Moreover, they will be used for the validation of the operative model of the innovative dual source heat pump and its coupling to the geothermal reservoir.