A novel Ground Source heat exchanger in an underground metro tunnel

Using Ground Coupled Heat Pump (GCHP) systems in urban areas can be particularly difficult due to space or legislative constraints, other than excessive costs due to drilling. To overcome these problems, the authors proposed to use Artificial Ground Freezing (AGF) probes, used for tunnel excavation, as Ground Heat Exchangers (GHE). It is a widespread practice to seal the probes in the tunnel after the completion. The conversion of existing AGF probes into GHE for GCHP allows us to avoid additional drilling costs, and other space or legislative constraints associated with the use of GCHP systems in urban areas. Such systems could be developed to use underground urban transportation tunnels for heating and cooling in smart cities. These systems were tested after the construction of two tunnels, as part of the GeoGRID project, in Piazza Municipio, Naples, Italy. The data obtained from the experimental setup have been analysed and used for validation of a finite element model developed by the authors to simulate heat transfer between the probes and the surrounding ground. A simplified pipe flow model was introduced, in combination with a 3D model of the ground, to reduce the complexity and computational effort to solve the discretized equations. The simulation results have been compared with experimental data, showing a good agreement, and observing differences in the range of 2 – 5 %. The model can therefore be used as a predictive tool for the development of this type of innovative heat exchanger.


Introduction
The goals set for decarbonization by the European Green Deal (EGD) aim to achieve carbon neutrality by 2050, and before that to reduce Greenhouse gas (GHG) emissions by 30% by 2030 compared to 1990 levels [1].30% of the GHG emissions are due to the use of fossil fuels for energy consumption across the world [2].Including renewable energy is necessary to reduce the dependence on fossil fuels, which is not in good proportion compared to the energy supplied by conventional fuels.Heating is crucial for the European energy sector as it is the major contributing sector in energy consumption.The energy source mix for heat production in the EU states is shown in figure 1 including renewables contribution [3].  Figure 1 shows that around 32% of the heat production is obtained by the renewables while the main contributor is the natural gas.If we specifically investigate the gross heat production mix of Italy which is shown in figure 2, we can see that the heat production in Italy highly depends on the supply of gas and then on solid fuels, which is an important cause of GHG emissions.It is worth mentioning that Italy is importing that gas from different countries and prices of the gas will likely increase under the current geopolitical situation between the Russia and the European countries.The energy transition is a gradual process that must include on one side more low carbon energy sources and on the other side technologies with higher efficiencies to reduce GHG emissions while reducing the energy consumption.Heat pumps being an energy efficient and economically viable technology in terms of energy consumption can be a good fit for the purpose.An advantage of heat pump is its versatile nature of compatibility with diverse renewable energy sources.Recently, ground source heat pump (GSHP) systems attracted many researchers because of their sustainable nature in terms of low energy consumption and minimal impact on environment in terms of GHG emissions [4], [5], moreover it can be used to exploit the low enthalpy sub-surface sources which can be used for heating/ cooling [6], [7].Among all GSHP types, one of the most convenient type of system is the Ground Coupled Heat Pump (GCHP) system which can further be categorized into a vertical and horizontal system based on the configuration of the loops laid into the ground [8].However, more than 50% of the cost associated with GCHP is due to ground loop installation, due to the drilling installation costs for the loops [9].These costs can be minimized by avoiding the expensive procedures of drilling.
Figure 3 shows the number of tunnels across Europe, many of those tunnels were built by using the Artificial Ground Freezing (AGF) technique [10].The advantageous aspect about underground tunnels constructed with AGF is the presence of probes used during the construction of tunnels [11], [12].The probes are then buried in the ground after the tunnel is completed.Conversion of these probes into ground heat exchangers (GHE) for coupling heat pumps and refrigeration system with the ground for heating/ cooling operations can reduce the drilling cost, but it is not easy.This novel idea was implemented under the GeoGRID project [13] developed in the region of Campania in southern Italy, with the support of EU funding.The idea was to convert available AGF probes into GHE with no or little modification to maximize the benefits associated with their use.During the project, experiments were conducted by converting some of AGF probes with GHEs [14].The numerical model validated in this work has been developed to evaluate the performance of the GHE with different conditions, to assess the potential for heat exchanges during the heating and cooling processes.The GHE has been numerically simulated to see the behaviour of the surrounding ground with the GHE [15].This paper presents a validation of the model with the experiments conducted at Piazza Municipio metro lines in the city of Naples, Italy.The developed model is based on the coupling of heat transfer in solid module with non-isothermal pipe flow module to reduce the computational time and model complexity during simulation.

Model & Equations
The numerical model developed in this work has been solved with commercial software, to simulate the effects of heat transfer between the fluid in the probes and the ground.The tunnel geometry has been obtained from the final construction and surroundings.After the tunnel geometry's construction, this was imported in COMSOL environment, where a 1D pipe flow model was combined with a 3D model of the ground and tunnel.The introduction of the pipe flow model reduces the overall complexity of the model and the computational time required for the simulations.Figure 4 shows the geometry details together with a picture of the activated tunnel and probes used in the experimental campaign.

Model Equations
To have a complete solution, two different physics are coupled in this model.Figure 4a presents the detailed model showing the heat transfer domain and the pipes.The thermo-fluid dynamics of the working fluid flowing inside the probes is modelled with the non-isothermal pipe flow (nipfl) onedimensional module, while the interaction of pipes with the surrounding ground is modelled as a 3D heat conduction module.The equation used for the solution of non-isothermal one-dimensional model is the energy conservation equation for an incompressible fluid flowing in a pipe shown as equation 1 [16].
Where,   (kg/m 3 ) is the fluid density, A (m 2 ) is the cross-sectional area of the pipe, Cpf (J/(kg.K)) is the specific heat of the fluid at constant pressure, u (m/s) is the velocity of the fluid in the pipe, T (K) is the temperature, kf (W/(m.K)) is thermal conductivity of the fluid, the second term on right hand side of the equation corresponds to the friction heat dissipation, and Qwall (W/m) takes into account the heat transfer from the pipe walls to the surrounding ground.A few assumptions are considered during the solution: i.
The flow through the pipes is incompressible, fully developed and one dimensional for the model.ii.
Frictional pressure drops are considered.iii.
All velocity components normal to the pipe axis are assumed to be zero.The energy conservation equation used to model the heat transfer in the ground is shown as equation 2.
Where   (kg/m 3 ) is the density of solid, Cps (J/(kg.K)) is the specific heat of solid at constant pressure, T (K) is the temperature, u (m/s) is the velocity field which is assumed to be zero near the probes because of the reduced permeability due to freezing operation during the excavation of the tunnel, k (W/(m.K)) is the thermal conductivity of the solid, and Qwall is the heat, which is exchanged with the pipes.In this model, pipe flow module was coupled with solid heat transfer module through temperature and the Qwall term in the above equations.

Initial and Boundary Conditions for the Model
The initial and boundary conditions considered in this work have been derived from the experimental setup, to have a realistic comparison of simulated and experimental data.The initial and boundary conditions employed in the simulations are shown in figure 5.The boundary conditions on both sides are assumed to be adiabatic because of the infinite domain, therefore: The undisturbed ground temperature, measured experimentally, is imposed to be constant and applied at the bottom of the domain.Tgrnd = 18.5 o C (4) The top surface of the domain is exposed to the ambient environment, therefore a convective boundary conditions, with assumption of natural convection, is assumed which is exchanging heat with the domain and can be written by the following equation.
= ℎ ( −   ) (5) In this equation, T represents the temperature of the domain while   is the ambient temperature, whereas 'h' represents convective heat transfer coefficient for natural convection and the value of 6 W/m 2 K is used during the calculation [17].The front and back of the tunnel are assumed to be thermally, insulated so there is no heat transfer in the direction normal to these surfaces.Initial conditions for the fluid flow are: Tin = T(t) (6) uin = u(t) (7) Tin and uin are the inlet temperature and velocity of the Heat Transfer Fluid flowing through the pipe respectively.The temperature and velocity are time dependent and, are obtained from the experimental data.1.28 Heat capacity Cp (kJ/(m 3 K)) 3150 Permeability kf (m/s) 10 -6 Ground is considered to be homogeneous and material properties are considered to be constant during the simulation.

Mesh Characteristics
The mesh sensitivity analysis has shown that the best balance between computational time and independence of the results from the grid is the one presented in Figure 6 and Table 2.
Table 2. Details of the mesh elements of the model  Figure 6 shows the overall positioning and shape of the Mesh elements in the model and shows the area where two physics models are interacting, the size of the elements is much finer as compared to the areas which are far from the probes.The model has been simulated for the month of July, with the above-mentioned boundary conditions, to observe the behaviour of the model.The experimental data were processed to be used as boundary conditions for the simulation model, and to obtain experimental data at the probes outlet to be used for validation of model.

Results & Discussions
In this section of the paper, results in terms of temperature field in the ground and transient thermal behaviour of the tunnel are discussed.In figure 7   Figure 8 shows the isotherms at the inlet and outlet of the tunnel.The average difference between each iso-surface is around 0.5 o C, and the temperature distribution can be seen through the computational domain.The surface adjacent to the pipes shows a temperature around 23 o C, while away from the tunnel, the temperature reduces to the average ground temperature of 18.5 o C. The main parameter employed to compare simulations with experiments is the outlet temperature of the working fluid from the GHE probes.Figure 9 shows the comparison between experimental data and numerical results in terms of outlet temperatures over time for different probes.The trend outlet temperature trend for all the probes is the same against the variation of the flow rate which shows that irrespective of the position of the probes the temperature may vary with variation in the flow rate of the HTF (Heat Transfer Fluid).It is due to the fact that the probes were fed in parallele.

Conclusions
The numerical model presented here has been validated with the experimental results obtained in correspondence of the probes installed in the tunnels at Piazza Municipio, Naples, Italy.A good agreement between the numerical and experimental results is observed.The model can be used to predict the actual behavior of the probes when used for actual building.The present ground heat exchanger system has economic benefits, related to the re-use of the freezing probes due to which the inflated cost of drilling can be minimized for such systems.Moreover, the model will be used to simulate the operation of GSHP coupled with the ground through the tunnel.

Figure 1 :
Figure 1: Energy mix for gross heat production in EU countries

Figure 2 :
Figure 2: Energy mix of Italy for gross heat production

Figure 3 :
Figure 3: Number of Underground tunnels in Europe

Figure 5 :
Figure 5: Boundary conditions on domainFurthermore, the material properties have been chosen considering the actual conditions, the properties of the tunnel material are shown in Table1[18].Table1.Ground properties adjacent to tunnel

Figure 6 :
Figure 6: Distribution of elements on a particular mesh (a), the fluid temperature in the probes is shown.The heat is transferred as the fluid flows through the pipe from the inlet towards the outlet.The inlet fluid temperature was 27 o C and at outlet the temperature dropped down to 26 o C. The temperature reduces along the pipe as heat is transferred to the ground via AGF probes converted into the ground heat exchangers (GHE).In figure7(b), the temperature distribution in the ground adjacent to the tunnel is shown throughout the computational domain.It is evident from the plot that the heat transfer occurs outwards from the tunnel.Temperatures near the tunnel is high and it reaches the ground temperature as the distance increases from the tunnel.

Figure 7 :
Figure 7: (a) Fluid temperatures in the probes, (b) Temperature distribution in the computational domain

Figure 8 :
Figure 8: Temperature profiles at the inlet and outlet sections of the tunnel

Figure 9 :
Figure 9: Comparison of experimental and numerical data for validation of model

Table 1 .
Ground properties adjacent to tunnel