 Research
 Open access
 Published:
Multiobjective optimization and longtime simulation of a multiborehole ground heat exchanger system
Geothermal Energy volume 12, Article number: 31 (2024)
Abstract
Multiobjective optimization and CFD simulation are conducted to optimize the design of a multiborehole ground heat exchanger (GHE) system and assess its longtime performance. The multiobjective optimization is performed to minimize the entropy generation number (EGN) and total cost rate by using various evolutionary algorithms, including NSGAII, GDE3, MOEA/D, PESAII, SPEAII, and SMPSO. NSGAII and GDE3 algorithms perform best in obtaining Pareto optimal solutions. Three prominent points on the NSGAII Pareto frontier, representing the results of singleobjective thermodynamic, singleobjective economic, and multiobjective optimizations, are simulated in three dimensions over three months. The trends of EGN variations extracted from the transient CFD simulation agree well with those from the steady analytical model. The EGN obtained from multiobjective optimization is 58.8% lower than the EGN obtained using singleobjective economic optimization and 1.9 times higher than that calculated from singleobjective thermodynamic optimization. Likewise, the total cost rate obtained from multiobjective optimization is 64.4% lower than the value obtained from singleobjective thermodynamic optimization and four times higher than that calculated using singleobjective economic optimization. The proposed optimization approach can be reliably applied to improve the design of multiborehole GHE systems.
Introduction
Global warming is a significant concern, and carbon dioxide emissions from fossil energy sources are its primary cause (Alkaff et al. 2016; Soltani et al. 2019a). Buildings are responsible for approximately 40% of the carbon dioxide emissions caused by fossilbased energy consumption (Hughes et al. 2011; Soltani et al. 2019b). A considerable amount of energy consumption and carbon dioxide emissions in buildings are related to HVAC systems (Razmi et al. 2018), which are predicted to increase due to world population growth (ÜrgeVorsatz et al. 2015).
Expanding the use of ground source heat pumps (GSHPs) for cooling and heating buildings could be an effective strategy to confront these issues. In recent decades, GSHPs have received increasing interest because of their high efficiency, low environmental footprint, long life expectancy, low maintenance requirement, and low running costs (Self et al. 2013; Zhu et al. 2014). Despite the popularity of these systems, GSHPs have some drawbacks, such as their high installation costs, infrastructure restrictions, and design limitations, that should be overcome to make these systems more convenient (Goetzler et al. 2012). In this regard, many studies have been conducted on improving the ground heat exchanger (GHE) design and thermal performance of the GSHP systems. These studies are recognized into analytical, experimental, numerical, or optimization categories (Javadi et al. 2019).
Numerical studies are the most common method of investigation for GHEs (Javadi et al. 2019) because of the high cost of experimental studies. Several numerical studies have been conducted on the longtime simulation of GHEs (Hein et al. 2016; Huang et al. 2021; Naicker and Rees 2020; Tang and Nowamooz 2018). Longtime numerical simulations are essential for the performance assessment, maintenance, and lifecycle stability of GSHP systems (Li et al. 2018). GSHP applications generally use multiple borehole heat exchangers (BHE) to enable a higher heat transfer rate (Marcotte et al. 2010; Qian and Wang 2014); therefore, models are expected to perform well for multiborehole systems. In addition, threedimensional models simulate such systems more precisely by considering the temperature variations along the depth, capturing interactions between boreholes, and accurately predicting the system performance (Pu et al. 2015; Rees and He 2013). Despite these facts, few studies have conducted longtime threedimensional numerical simulations for multiBHE systems because of their complexity and high computational cost. Choi et al. (2018) simulated the 10year operation of a GSHP consisting of 35 single Utube BHEs under a load imbalance condition. However, their model was simplified by considering a single BHE instead of the multiBHE system. Gultekin et al. (2019) investigated the influence of the arrangement and the number of boreholes on the thermal performance of a multiBHE field. In their longtime simulation, temperature changes along the borehole direction were ignored, and the threedimensional problem was simplified to a twodimensional one. The present work aims to bridge the gap by numerically simulating a threedimensional model of multiborehole GHE for a long time.
To date, many optimization strategies have been developed for the optimal design of GSHP systems to reduce initial investment costs and improve operating performance. An overview of the recent progress in this area was presented by Ma et al. (2020). Thermodynamic and economic objectives are frequently used for the optimal design of GSHP systems. Among the thermodynamic objectives, entropy generation minimization is widely used in modeling and optimizing GHEs to reduce thermodynamic imperfections from heat transfer and fluid flow irreversibilities (Bejan 1996; Cheng and Liang 2014). The annual, life cycle and total costs are often used as economic objectives (Ma et al. 2020). Several studies (Huang et al. 2014; Kord and Jazayeri 2012; Robert and Gosselin 2014; Soltani et al. 2021) have attempted to optimize GSHP systems using a single objective, economic or thermodynamic. The main problem with the singleobjective design optimization strategies is that system performance might need to be improved regarding the overlooked objective (Ma et al. 2020; Sayyaadi et al. 2009). On the other hand, in a multiobjective design optimization approach, thermodynamic and economic objectives are simultaneously optimized, which could improve the optimal design of GSHP systems. Evolutionary algorithms (EAs) in solving multiobjective optimization problems have been greatly expanded in recent years. These algorithms use a populationbased approach that evolves a set of solutions in each iteration to eventually provide Pareto optimal solutions welldistributed across the Pareto frontier (Emmerich and Deutz 2018). Sayyaadi et al. (2009) presented a multiobjective optimization method for a GSHP with a vertical GHE to minimize the system’s total cost and exergy destruction with eight decision variables using an evolutionary algorithm. Huang et al. (2015) conducted a multiobjective optimization to minimize entropy generation and system upfront cost using a genetic algorithm (GA). By comparing the proposed multiobjective design optimization with a singleobjective optimization based on entropy generation minimization for two case studies, it was concluded that multiobjective design could reduce total installation cost by 9.5% and increase energysaving by 6.2%. Among evolutionary algorithms, the nondominated sorting genetic algorithmII (NSGAII) has widely been used for multiobjective optimization studies (Song et al. 2019; Chang et al. 2023; Mostafazadeh et al. 2023) because of its high solving speed and proper convergence (Deb et al. 2002). Other evolutionary algorithms have rarely been used in energy issues. Employing various EAs in a multiobjective optimization approach based on total cost minimization and exergetic efficiency maximization for the cooling mode of a vertical GSHP system was presented by Keshavarzzadeh et al. (2020). The NSGAII algorithm provided more acceptable performance in their study than other evolutionary algorithms.
Considering the importance of multiobjective optimization in the appropriate design of GSHPs and the outstanding portion of the GHE in the total cost of the system, this study attempts to determine the optimum value of the multiborehole GHE design parameters concerning entropy generation and total cost rate as two objective functions. Entropy generation is nondimensionalized and quantified by entropy generation number (EGN) (Bejan 1996; Maheshkumar and Muraleedharan 2011). A semitwodimensional steadystate analytical model describes the GHE thermal behavior in the optimization process. In addition to NSGAII algorithm, five other evolutionary algorithms are used to define the optimum design parameters: speedconstrained multiobjective PSO (SMPSO), strength Pareto evolutionary AlgorithmII (SPEAII), Pareto envelopbased selection algorithmII (PESAII), multiobjective evolutionary algorithm based on decomposition (MOEA/D), and the third evolution step of generalized differential evolution (GDE3). The previous studies did not apply various evolutionary optimization algorithms to the GSHP system for simultaneous optimization of entropy generation and total cost rate. Five decision variables are considered to find the optimal design of a multiborehole vertical GHE system. Besides the multiobjective optimization of GSHP, threedimensional CFD simulation of GHEs assess the longtime performance of the optimized multiborehole vertical GHE system for 3 months. To the best of the authors’ knowledge, a combination of optimization and CFD simulation studies has yet to be conducted. The advantage of the threedimensional transient CFD simulation is comparing the results with the steadystate analytical solution and getting a better assessment of the optimized GHE system by obtaining features not considered in the optimization.
System description
GeoCool GSHP system located at the Universitat Politènica de València in Spain (Corberan et al. 2011; Montagud et al. 2011) is studied as illustrated in Fig. 1. GeoCool plant is a reversible watertowater GSHP that provides heating and cooling for a set of spaces with a total area of 250 \({\text{m}}^{2}\). The GHE of this system is made up of six vertical boreholes connected in parallel and arranged in a 2 × 3 rectangular grid over an area of 18 \({\text{m}}^{2}\). Each borehole contains a single highdensity polyethylene (HDPE) Utube, is filled with sand, and is finished with a bentonite layer on top. The present study only considers the cooling operation of this GSHP system. The specifications and design conditions of the system are provided in Table 1.
Three case studies are considered from the optimization of the system named A, B, and C; these cases represent the optimized system from thermodynamic, multiobjective, and economic points of view, respectively. The details of the three case studies will be provided in Sect. “Material properties and geometry”.
Analytical modeling for multiborehole GHE
Entropy generation modeling
Entropy generation leads to exergy destruction and loss of available work. Therefore, entropy generation minimization (EGM) is an effective optimization tool to minimize losses and maximize system performance (Bejan 1996; Cheng and Liang 2017). In a vertical GHE, entropy is generated by fluid friction along the tube and finite temperature difference between the fluid and the borehole, resulting in pressure drop and heat transfer losses, respectively (Bejan 1996). Therefore, minimizing entropy generation improves the system's hydraulic and thermal performances.
The following introduces the relations to obtain borehole wall temperature, outlet, and mean fluid temperatures. Next, the equations for pressure drop calculation, total entropy generation, and the economic model are presented.
Heat transfer analysis
The schematic of a single Utube vertical GHE, along with its main geometric parameters, including borehole radius (\({r}_{\text{b}}\)), inner and outer tube radius (\({r}_{\text{i}}, {r}_{\text{o}}\)), and borehole shank spacing (2D), is presented in Fig. 2. Two heat transfer processes occur in a BHE: heat transfer between the ground and the grout, and heat transfer between the grout and the fluid through the tube wall.
The analytical model representing the heat transfer process of the vertical GHE consists of the onedimensional infinite line source model (ILSM) and the twodimensional thermal resistance of the boreholes acquired from the line source approximation (Lamarche et al. 2010). The ILSM can correctly predict heat transfer in GHE if the operating time is higher than \(20{{\text{r}}_{\text{b}}}^{2}/{\alpha }_{\text{s}},\) which is 54 here (Yang et al. 2010). Also, the operating time is one year.
By considering a long timescale and constant heat flux in the vicinity of the borehole, Eq. (1) approximates the borehole wall temperature (\({T}_{\text{b}}\)) based on ILSM (Huang et al. 2014):
\({T}_{\text{s},0}\), \({k}_{\text{s}}\), \({\rho }_{\text{s}}\), \({c}_{\text{s}}\), and \({\alpha }_{\text{s}}\) are the undisturbed temperature, thermal conductivity, density, specific heat, and thermal diffusivity of soil (i.e., ground), respectively. \({r}_{\text{b}}\) is the borehole radius, t is the time, and \({\text{E}}_{1}(x)\) is the exponential integral function. \({q}{\prime}\) is the heat transfer rate per unit length:
where \(\dot{Q}\) is the design load of the GSHP system. N and \({L}_{b}\) are the number and depth of the boreholes.
The thermal resistance of the borehole determines the heat transfer rate between the circulating fluid and the grout material. The borehole thermal resistance (\({R}_{\text{b}}\)) in a single Utube BHE has three components (Lamarche et al. 2010):
Convection resistance (\({R}_{conv}\)) and conduction resistance (\({R}_{\text{cond}}\)) are defined as Lamarche et al. (2010):
and
in which, \({r}_{\text{i}}\) and \({r}_{\text{o}}\) are the inner and outer radii of the Utube. \({h}_{\text{f}}\) is the convection coefficient, and \({k}_{\text{t}}\) is the tube’s thermal conductivity. Grout’s thermal resistance (\({R}_{\text{g}}\)) is determined by the line source approximation (Lamarche et al. 2010):
where D is the halfshank spacing and \({k}_{\text{g}}\) is the grout’s thermal conductivity.
The coefficient of convection heat transfer (\({h}_{f}\)) is obtained through the Nusselt number definitions (Bejan 2013) for fully developed laminar and turbulent flows using the following relationships:
and
Equation (8) is valid for \(0.7<\text{Pr}<160\text{ and Re}>{10}^{4}\). Prandtl number and Reynolds number for the flow in the Utube are obtained from
and
in which, \({c}_{\text{f}}\), \({k}_{\text{f}}\), and \(\mu\) are the fluid’s specific heat, thermal conductivity, and dynamic viscosity, respectively. \({\dot{m}}_{\text{f}}\) is the mass flow rate of the fluid in a single BHE.
The thermal resistance between the borehole wall and the soil (\({R}_{\text{s}}\)), and the thermal resistance for multiple boreholes connected in parallel (\({R}_{\text{s},\text{N}}\)) are determined from Lamarche et al. (2010):
and
where B denotes the distance between the boreholes.
The arithmetic mean of the fluid’s temperature in a BHE (\({T}_{\text{f}}\)) is related to the undisturbed soil temperature, heat transfer rate, and thermal resistances through the following equations:
for a single BHE, and through:
for a multiBHE system.
\({T}_{\text{f}}\) is defined as:
where \({T}_{\text{f},1}\) and \({T}_{\text{f},2}\) represent the inlet and outlet temperatures, respectively, for the circulating fluid in BHE.
Pressure drop calculation
Pressure drop through a single Utube is approximated as Li and Lai (2013):
where \(f\) is the friction factor (Li and Lai 2013):
and
Entropy generation rate calculation
The entropy generation rate caused by temperature difference (\({\dot{S}}_{\text{gen},\Delta T}\)) is expressed as (Bejan 1996):
in which, \(\Delta T\) is the difference between \({T}_{\text{f},\text{m}}\) and \({T}_{\text{b}}\). The parameter \(\chi\) is the dimensionless temperature difference defined as \(\Delta T/{T}_{\text{f},\text{m}}\), and \({T}_{\text{f},\text{m}}\) denotes the mean temperature of the fluid (Bejan 1996):
The entropy generation rate caused by fluid friction (\({\dot{S}}_{\text{gen},\Delta P}\)) is expressed by Eq. (21) for an incompressible fluid under nonadiabatic conditions (Bejan 1996):
where \({\dot{m}}_{\text{f},\text{tot}}\) is the total mass flow rate through the multiborehole GHE system. The total entropy generation rate in the GHE (\({\dot{S}}_{\text{gen}}\)) is the sum of the hydrodynamic and thermal entropy generation rates components:
Economic modeling
For economic modeling of the GSHP system, the total revenue requirement (TRR) method presented by Sayyaadi et al. (2009) is employed. Based on the method, the levelized cost rate related to the capital investment and the operation and maintenance expenses of the kth system component (\({\dot{Z}}_{k}\)), and the levelized cost rate of the electricity provided for the overall system (\({\dot{Z}}_{\text{elec}}\)) are:
where \(\tau\) is defined as the annual operating hours in the cooling mode, and \({\text{PEC}}_{k}\) is the purchase equipment cost for the kth system component. \({\text{CC}}_{L}\), \({\text{OMC}}_{L}\), and \({\text{FC}}_{L}\) represent levelized annual carrying charges, levelized annual operation and maintenance costs, and levelized annual fuel costs, respectively; for more details see Sayyaadi et al. (2009). The purchase equipment cost for the GHE is determined by Eq. (25) (Sayyaadi et al. 2009), and the purchase equipment cost for the heat pump’s components is provided in Table S1 of the Supplementary Material:
\({L}_{\text{t}}\) is the Utube length. \({C}_{\text{t}}\) is the cost of the Utube per meter, set to be 1.6 €/m. \({C}_{\text{b}}\), the drilling and grouting cost per meter is 50 €/m.
Optimization
Figure 3 shows a block diagram of the optimization strategy, including three main steps used for the multiborehole GHE systems. After the energy and economic analysis of the system, the decisionmaking variables are selected among all the design parameters based on sensitivity analysis. The decisionmaking variables affect the objective functions more strongly than the other design parameters. The next step is formulating the optimization method. After developing the mathematical model of the GSHP system, the objective functions, decision variables, and optimization constraints are defined. The decision variables are then optimized through a variety of multiobjective evolutionary algorithms, which search for a set of Pareto optimal solutions.
Objective functions, decision variables and constraints
Two objective functions are considered: EGN and total cost. The optimization algorithms are used to minimize both objective functions simultaneously. EGN (\({N}_{\text{s}}\)) describes system irreversibilities is defined as:
The total cost of the GSHP system (\({\dot{C}}_{\text{tot}}\)) is expressed as:
Five influential design parameters, including borehole depth (\({L}_{\text{b}}\)), borehole radius (\({r}_{\text{b}}\)), number of boreholes (N), circulating fluid mass flow rate per Utube (\({\dot{m}}_{\text{f}}\)), and Utube outer radius (\({r}_{\text{o}}\)) have been selected as the decision variables in this study based on the results of a global sensitivity analysis conducted by Huang et al. (2014); these five decision variables are determined based on their sensitivity indices in terms of the EGN. It is worth mentioning that GHErelated costs are a significant part of the GSHP total cost. \({L}_{\text{b}}\) and N are the most influencing parameters on the cost of the GHE/GSHP system (Robert and Gosselin 2014).
A reasonable range should be used for the values of the decision variables to avoid unrealistic solutions. These constraints are determined based on the recommendations obtained from practical engineering projects (ASHRAE 2011; Banks 2012), shown in Table 2. The values of the lowsensitive parameters, which are considered constant in the optimization process, siterelated parameters, and grout thermal conductivity are presented in Table 3.
Two other constraints are used for the heat transfer rate per unit of length (\({q}{\prime}\)) and outlet temperature. The acceptable range for \({q}{\prime}\) is between 30 and 130 W/m proposed by Robert and Gosselin (Robert and Gosselin 2014). The recommended values for the minimum outlet temperature (\({T}_{\text{f},2,\text{min}}\)) and maximum outlet temperature (\({T}_{\text{f},2,\text{max}}\)) are determined by Eqs. (29) and (30) for the heating and cooling modes, respectively (ASHRAE 2011):
where \({T}_{\text{s},\text{min}}\) and \({T}_{\text{s},\text{max}}\) are the minimum and maximum soil temperatures over a year.
Multiobjective evolutionary algorithms
NSGAII is a multiobjective evolutionary algorithm based on genetic algorithm. In addition to the NSGAII algorithm, five other evolutionary algorithms, including SMPSO, SPEAII, PESAII, MOEA/D, and GDE3, are applied and compared.
SMPSO incorporates particle swarm optimization (PSO) and Pareto dominance to handle multiobjective optimization problems. This algorithm can store nondominated solutions found in a search in an external repository. It could enhance the exploratory ability of PSO by utilizing a timedependent mutation operator (Nebro et al. 2009). SPEAII is an improved version of SPEA. This algorithm behaves similarly to NSGAII (Zitzler et al. 2001). PESAII is a developed version of PESA. This algorithm uses a new selection method based on regions instead of individuals. The mechanism of this method is very similar to multiobjective particle swarm optimization by the difference that PESAII uses genetic operators instead of PSO operators. PESAII has an external archive for nondominated solutions; parents and mutants are selected from the created regions in this archive (Corne et al. 2001). The structure and mechanism of MOEA/D algorithm are different from other algorithms. It decomposes multiobjective optimization problems into some scalar optimization subproblem and optimizes them simultaneously. The number of subproblems is usually equal to the size of the population. The method assigns a weight vector to each subproblem. The evaluation is performed based on this vector. Finally, each subproblem theoretically provides a Pareto Front solution at the end of the search (Zhang and Li 2007). GDE3 is an extension of differential evolution that can support an arbitrary number of objectives and constraints. It provides a better distribution solution than the previous GDE version (Kukkonen and Lampinen 2005).
Model validation
The BHE was optimized using the entropy generation minimization method and genetic algorithm. A comparison between the results of Huang et al. (2014) and those from the present study is shown in Table 4. The aim of the optimization is to enhance the thermodynamic efficiency of the original system design (considered an unoptimized design) by minimizing EGN through modifying decision variables. The EGN, the optimization objective function, is compared for validation purposes. The comparison indicates that the method used herein provides a lower EGN, by 2.38%, compared to that obtained by Huang et al. (2014). The optimized values of the decision variables in this study and that of Huang et al. (2014) are slightly different, which stem from the different optimization models utilized in these studies.
Numerical simulations
From the optimization results of the studied GSHP system (to be presented in Sect. “Optimization results”), three prominent points labeled as points A, B, and C in Fig. 8 will be selected for CFD simulation as case studies. Transient 3D models of the selected case studies are simulated longterm. The model geometry is divided into four subdomains: the circulating fluid, the Utube walls, the grout, and the ground. The following assumptions are used for the simulations:

The ground is an infinite homogeneous medium for heat transfer.

The thermophysical properties of the ground, grout, Utube walls, and fluid are constant and independent of temperature.

The effects of underground water flows are neglected.

The circulating fluid is Newtonian and incompressible.

Heat losses due to radiation, convection, and evaporation from the ground surface are neglected.

The temperature of the ground surface is considered uniform and constant.
Transient fluid flow and heat transfer in the fluid and solid subdomains are simulated by solving the threedimensional and unsteady governing equations: conservation of mass, conservation of momentum, and conservation of energy.
conservation of energy for the fluid subdomain (Pletcher et al. 2012) is
where \(e\) is the internal energy per unit mass, \(Q\) is the heat produced per unit volume by external agencies and \(\varphi\) is the viscous dissipation term.
Conservation of energy for the solid subdomains (Pletcher et al. 2012) is
in which, \({\rho }_{i}\), \({c}_{i}\) and \({k}_{i}\) are the density, specific heat, and thermal conductivity of one of the solid subdomains, respectively. The solid subdomains include the ground, grout, and Utube walls.
The Reynolds numbers in the present study are higher than 10^{4}; therefore, the realizable k−ε model is selected as a turbulence model (Serageldin et al. 2018).
The finitevolume discretization method is used to approximate the governing equations. The numerical calculations are carried out using a doubleprecision and pressurebased solver. The SIMPLE algorithm is selected for pressure–velocity coupling, and a secondorder discretization scheme is used to discretize pressure, momentum, energy, k and ε equations. The convergence criteria are set to be \({10}^{8}\) for energy equation, and \({10}^{5}\) for the rest. The time step is considered 300 s with a total number of 26,574 steps.
Material properties and geometry
The simulated geometry consists of four subdomains: one fluid subdomain, which includes the water flowing inside the Utube, and three solid subdomains, including the Utube walls, grout, and the ground. A highdensity polyethylene (HDPE) Utube is placed inside the borehole. The space between the tube and the ground is filled with grout material, combining sand and bentonite. The thermal properties of the materials used in the fluid and solid subdomains are shown in Table 5. The geometric parameters and the size of the computational domains for cases A, B, and C are shown in Table 6. It should be noted that cases A, B, and C represent three prominent points of A, B, and C on the Pareto frontier presented in Fig. 8 and detailed in Table 10. Therefore, the values of the design parameters, i.e., \(N\), \({L}_{\text{b}}\), \({r}_{\text{b}}\), and \({r}_{\text{o}}\), for cases A, C, and B have been determined by optimization to minimize entropy generation number, total cost rate, or a combination of both targets, respectively. Other parameters not subject to optimization, such as \(D\), \(B\), and \(t\), are constant across three cases.
In all three cases, the boreholes are connected in parallel, with a distance of 3 m from each other. These arrangements closely align with the original design described in Sect. “System description”. Figure 4 illustrates the arrangements of boreholes in cases A, B, and C. As an example, the computational domain for Case B and the essential geometric parameters are shown in Fig. 5. The outermost boundaries of the soil subdomain are considered far enough from the boreholes to prevent the boundary conditions from affecting the results.
Mesh generation
The computational domain is divided into subdomains with two types of threedimensional meshes to reduce the number of cells, shown in Fig. 6 for Case B. Other cases have a similar mesh structure. An unstructured mesh is considered. Tetrahedral cells are applied for meshing the 180° bend of the Utube and the base of the grout and the ground. The sweep meshing method is used for the vertical parts of the tubes, plus the grout and the surrounding ground. A twodimensional horizontal crosssection of the domain was firstly discretized into the triangular cells, swept in the vertical direction to generate a threedimensional mesh, and divided into hundred slices to capture variations in the vertical direction. The cell sizes gradually decrease as moving from the outer boundaries toward the center. The smallest cell size of 2 mm is used for the 180° Utube bend, where strong temperature gradients are expected. An optimal mesh size is obtained by performing a mesh independency test for the three case studies under steadystate operating conditions. The outlet temperatures of Case B are presented in Table 7. The optimum is mesh#3 with \(1.65\times {10}^{6}\) cells because the temperature’s relative deviation for a finer mesh (mesh #4) is only about 0.1%. With a similar method, the optimum number of cells for Case A and Case C are determined to be \(4.42\times {10}^{6}\) and \(0.73\times {10}^{6}\), respectively.
Initial and boundary conditions
The initial temperature of the whole computational domain is equal to the undisturbed ground temperature, i.e., 19.6 \(^\circ{\rm C}\). In Table 8, a Dirichlet boundary condition with a constant temperature of 19.6 \(^\circ{\rm C}\) is assigned to the upper and bottom surfaces of the ground. In contrast, the adiabatic condition is used for the ground’s lateral surfaces. Constantmassflow and constantpressure boundary conditions are used for the inlets and outlets of the fluid subdomain, respectively. An additional timevarying Dirichlet boundary condition is also used for the inlets, which describes the incoming fluid’s temperature changes as a function of the heat transfer between the circulating fluid and the heat pump. The following equation expresses the timevarying inlettemperature boundary condition:
The heat rate (\({\dot{Q}}_{\text{b}})\) assigned to each borehole, for cooling operation:
\(\text{COP}\) is the average coefficient of the heat pump’s performance in the cooling mode, equal to 5 (Nam et al. 2008).
According to Eqs. (33) and (34), the inlet fluid temperature at each time step is a function of the cooling load and the outlet fluid temperature from the previous time step.
Model validation
An experimental dataset from the 5year operation of GeoCool GSHP (RuizCalvo and Montagud 2014), Table 9, has been applied to validate the longterm transient simulation. Despite a few quantitative differences, this experimental study (RuizCalvo and Montagud 2014) closely matches the physical processes, setup, and operating parameters used in the present numerical model, making it a suitable case for validation.
The system was monitored from February 2005, working 15 h daily and five days weekly. Data were collected every minute. The numerical model uses collected data for each borehole's inlet water temperature and mass flow rate as boundary conditions. The measured data for outlet water temperature is used for validating the model.
Figure 7 compares the average outlet flow temperature of six boreholes during April 2005 in the heating mood. The model predicts the outlet temperature with a maximum error of about 4.5%, showing the acceptable performance of the model. The numerical and experimental outlet flow temperature values for each borehole are compared in Fig. S1 of the Supplementary Material. The difference in outlet temperature between the numerical and experimental studies can be attributed to a 20% uncertainty in estimating the ground thermal conductivity, as reported in RuizCalvo and Montagud (2014). Additional contributing factors include experimental errors due to environmental conditions, as well as the accuracy of the data acquisition system and the temperature sensors.
Results and discussion
This study presents a multiobjective optimization of the GSHP system used in an institutional building in Spain and a 3D CFD simulation of its multiborehole GHE. The optimization results, presented in Sect. “Optimization results”, are obtained based on a steadystate analytical model using NSGAII and five other evolutionary algorithms. The simulation results of the same system, presented in Sect. “Numerical simulation results”, study the outlet fluid temperature distribution, the borehole wall temperature distribution, and the transient entropy generation have been studied for three cases over a long time. Finally, a comparison between the optimized and simulation results is presented.
Optimization results
NSGAII optimization
Figure 8 presents the Pareto frontier solution of the proposed GSHP system obtained by minimizing EGN (Eq. 27) and total cost rate (Eq. 28) simultaneously, using NSGAII multiobjective optimization algorithm. An increase in total cost accompanies a decrease in the EGN. As illustrated in Fig. 8, the minimum EGN occurs at point A (0.040), where the total cost rate is maximum (37.776 €/h). On the other hand, point C has the highest value of the EGN (0.192), while its total cost rate has the smallest value (3.329 €/h). Points A and C represent the optimal design points in singleobjective optimization, in which EGN and total cost rate are the objective functions, respectively.
In multiobjective optimization, selecting the desired optimal point among a set of optimal solutions on the Pareto frontier requires a decisionmaking procedure based on potential design limitations or financial considerations. One of the procedures for determining the desired optimum point is to use a hypothetical point indicated as the ideal point, Fig. 8 (Navidbakhsh et al. 2013). At this point, each objective has its optimal value and is independent of the other one; the ideal point cannot be located on the Pareto frontier because it is impossible to reach the optimum values of both objectives at the same time. Therefore, the closest point on the Pareto frontier to this ideal point is selected as the desired optimal point (Huang et al. 2015; Keshavarzzadeh et al. 2020; Ghanadi Arab et al. 2013). This point, presented as point B in Fig. 8, has an EGN of 0.079 and a total cost rate of 13.462 €/h. Point B is considered a tradeoff between EGN and total cost rate. The calculated results for the three prominent points (points A, B, and C) and the original design data are summarized in Table 10. EGN in points A and B are 77.4% and 55.4% less than the EGN in the original system; whereas for point C, EGN is 8.47% higher than that for the original design. From an economic point of view, the total cost rate for points A and B is 4.6 and 1.6 times more than the original case, respectively. However, the total cost rate in point C is 59% less than that of the original system. Therefore, the original system design is close to point C, which is the optimal design point based on the economic singleobjective optimization of the GSHP. However, the multiobjective design optimization shows that by increasing the total cost rate of the system by just 5 €/h, a significant reduction in the entropy generation number is achieved, which improves the thermal performance of the GHE.
Figure 9 presents scatter distributions of five decision parameters along the Pareto frontier, which illustrates the trends of each parameter and their impact on the GHE performance. It can also aid designers in the decisionmaking process. Figure 9a indicates that the borehole radius is nearly constant and located near its lower bound for all Pareto optimal points because an increase in the borehole radius increases the thermal resistance between the fluid and the ground. In Fig. 9b, the distribution of the number of boreholes is scattered uniformly between 3 and 10 because of the direct impact of the number of boreholes on both objective functions. Although increasing the number of boreholes of the GHE is desirable from a thermodynamic point of view, it leads to an increase in the total cost rate of the system similar to borehole depth, Fig. 9c. Although the spread of the optimal values of borehole depth is relatively uniform, they are mainly located near the upper and lower bounds. According to Fig. 9d, the optimum mass flow rate of the fluid in this study is around 0.4. In Fig. 9e, the distribution of the Utube outer radius is located near the upper bound for almost all optimum solution points. Larger tubes reduce the overall thermal resistance of the GHE and improve the heat transfer rate between the fluid and the ground.
Comparison of evolutionary algorithms
The Platypus library of Python was employed to compare the performance of different EAs in multiobjective optimization of the GHE. Python allows running several EAs simultaneously with the same random number generator seeds while saving CPU time with parallelization. For all algorithms, the direction of the objective functions is to minimize both EGN and total cost rates. The selection of the best optimization algorithm among various options is based on its ability to achieve a uniform distribution of optimal points and ensure diversity along the Pareto frontier (Keshavarzzadeh et al. 2020, 2019; Chen et al. 2010; Li et al. 2014). Figure 10a shows the NSGAII Pareto frontier discussed in the previous subsection, demonstrating excellent diversity and continuity in the distribution of optimal points. The Pareto frontier of GDE3, Fig. 10b, is similar to that of NSGAII, and there is not any considerable difference between them. The PESAII Pareto frontier, Fig. 10c, is almost identical to the previous two algorithms, although it is not as uniformly distributed as those generated by GDE3 and NSGAII. SPEAII algorithm yielded a discontinuous Pareto frontier, Fig. 10d, particularly for EGNs above 0.12 and below 0.04, indicating the unacceptable performance of this algorithm in these ranges. Similarly, the SMPSO algorithm results in a discontinuous Pareto frontier, as shown in Fig. 10e, with some data missing in two subranges. Finally, Fig. 10f clearly shows that MOEA/D algorithm provides an unstable behavior, failing to provide a welldefined Pareto frontier. Considering the notably diversity and uniform dispersion of optimal solutions along the Pareto frontier achieved by the NSGAII and GDE3 algorithms in contrast to the other algorithms, these two algorithms exhibit superior performance and reliability. These observations are consistent with earlier studies (Keshavarzzadeh et al. 2020, 2019) comparing evolutionary algorithms where the NSGAII algorithm provided acceptable results. The comparison emphasizes the capability of NSGAII as a dependable and efficient algorithm for addressing the complexities of multiobjective optimization in energy issues.
Numerical simulation results
Temperature distributions
Variations of the average fluid temperature at the outlet of the GHE during three months of cooling operation for the three case studies corresponding to points A, B, and C on the NSGAII Pareto frontier (Fig. 8) are shown in Fig. 11. The cooling load for all three case studies was 14 kW. Case C has the highest outlet fluid temperatures, followed by Case B. This difference between outlet temperatures is mainly due to the difference between the total borehole lengths (= borehole number × borehole depth). According to Table 6, the total borehole length for Case B is about onethird of the corresponding value for Case A and 3.9 times more than that for Case C. Shorter total borehole lengths lead to an increase in the heat flux per meter per borehole, which increases the outlet temperature of the circulating fluid in the GHE. Increasing the temperature of the fluid entering the heat pump system in the building increases the energy consumption of the heat pump and decreases its exergetic efficiency.
Figure 12 demonstrates the variations in the average temperature of the borehole wall. Similar to the results for outlet temperature, Case C has the highest borehole wall temperature, followed by Case B. Reducing the total borehole length from Case A to Case C increases the heat flux per meter of the borehole. The difference between the arithmetic average fluid temperature (\({T}_{\text{f}}\)) and the average borehole wall temperature is plotted in Fig. 13. After about three days, the temperature of the borehole wall and circulating fluid rises at the same rate, and the difference between them reaches an almost constant value of 1.6, 2.3 and 4.6 ℃ for cases A, B, and C, respectively. The highest temperature difference for Case C indicates the most significant thermal resistance between the fluid and the boreholes, leading to the lowest heat transfer rate. Using numerical results and Eq. (35), the values of the overall borehole resistance for the total length (\({R}_{\text{b}}^{*}\)) in cases A, B, and C are calculated as \(9.52\times {10}^{5}\), \(13.69\times {10}^{5}\) and \(27.38\times {10}^{5}\) \(\text{K}/\text{W}\), respectively. According to the simulation results of the temperature distributions in the GHEs, case study A demonstrates the best thermal performance, and case study C gives the weakest thermal behavior. These observations align entirely with the optimization results provided in Sect. “Optimization results”, Fig. 8; among all the optimal points across the Pareto frontier, point A emerges as the most optimal point from a thermodynamic point of view, and point C presents the weakest point from this aspect:
Transient entropy generation
Transient entropy generation is calculated using Eq. (22) as a function of time for the three case studies. From Sect. “Analytical modeling for multiborehole GHE”, the total entropy generation rate (\({\dot{S}}_{\text{gen}}\)) in a GHE is the sum of the thermal (\({\dot{S}}_{gen,\Delta T}\)) and hydrodynamic (\({\dot{S}}_{\text{gen},\Delta \text{P}}\)) entropy generation rates caused by heat transfer and fluid friction. Variations of all entropy generation rates during the 3month operation of the system for case studies A, B, and C are demonstrated in Fig. 14a–c, respectively. In all three studies, \({\dot{S}}_{\text{gen},\Delta \text{P}}\) remains almost constant and has a small contribution to the total entropy generation rate in the GHE. The average values of \({\dot{S}}_{\text{gen},\Delta \text{P}}\) for cases A, B, and C are 4.69, 1.48, and 0.34 \(\text{W}/\text{K}\), respectively. According to Eq. (16), the pressure drop along the Utube (\(\Delta P\)) is directly related to the borehole length. Therefore, reducing the borehole length from Case A to Case C decreases the pressure drop along the Utube. The maximum pressure drops for cases A, B, and C are 7.16, 4.94, and 2.76 kPa, respectively. This reduction in pressure drop, combined with reducing total mass flow rate (\(N\times {\dot{m}}_{\text{f}}\)) from Case A to Case C, reduces the entropy generation rate caused by friction based on Eq. (21). According to Fig. 14, the thermal entropy generation component has the major contribution to the total entropy generation rate for all cases during the entire system operation. The trends of the \({\dot{S}}_{\text{gen},\Delta T}\) variations are relatively similar for all three case studies. \({\dot{S}}_{\text{gen},\Delta T}\) increases for a short period after the system starts operating because of the increase of \(\Delta T\). With the continuing growth of \({T}_{\text{f},\text{m}}\) while \(\Delta T\) is constant, \({\dot{S}}_{\text{gen},\Delta T}\) slowly decreases with different slopes. The variations of \({T}_{\text{f},\text{m}}\) and \(\Delta T\) over simulation time for case studies A to C are shown in Figs. S2 and S3 of the Supplementary Material. Due to the highest \(\Delta T\) in Case C, the \({\dot{S}}_{\text{gen},\Delta T}\) during the entire system operation has the highest amount, followed by Case B. So, with the decrease in the total borehole length from cases A to C, despite the reduction of the \({\dot{S}}_{\text{gen},\Delta \text{P}}\) value, the value of \({\dot{S}}_{\text{gen},\Delta T}\) is increased. Since the temperature difference factor has a more substantial effect than the fluid friction factor on the total entropy generation rate, \({\dot{S}}_{\text{gen}}\) is also increased by decreasing the total borehole length.
Figure 15 compares the transient EGN variations for the three cases. After a slight increase of EGN for the early times of running, EGN gradually decreases and converges to a constant value. EGN is highest for Case C, with values ranging from about 0.11 to 0.14, and lowest for Case A, with values below about 0.05. EGN values for Case B lie between 0.04 and 0.08. A comparison between the EGN values at the end of the 3month simulation and the EGN values obtained from optimization for three cases is presented in Table 11. The trend of EGN variations for the three case studies is the same with both methods. The result acquired for case studies A and B agree to be within about 9.0% and 7.6%, respectively. There is a 29.7% difference between the two data sets for Case C. These quantitative differences are expected because the two methods are fundamentally different.
Conclusion
The design of a groundsource heat pump with multiborehole ground heat exchangers (GHE) for an educational building located in Spain is optimized using a variety of multiobjective evolutionary algorithms to minimize entropy generation number (EGN) and total cost rate simultaneously. In addition, the optimized multiborehole GHEs are numerically simulated using a threedimensional transient model for three months to study the variations of their thermal performance over time.

Based on the NSGAII Pareto frontier, the desired optimum values of EGN and total cost rate are 0.079 and 13.46 €/h, respectively (point B). Absolute EGN optimization yields the optimum solution at 0.040, where the total cost rate is at its highest value, 37.77 €/h (point A). Additionally, absolute economic optimization gives the optimum solution at 3.33 €/h, where the EGN is at its highest value, 0.192 (point C). The multiobjective solution achieves a remarkable 55.4% reduction in ENG while experiencing a 65.1% increase in the total cost rate compared to the original case

The selection among points A, B, and C as the final optimized solution depends on the priorities and specific requirements of decisionmakers and designers. If the primary concern is maximizing the thermodynamic performance of the GHE regardless of costs, then point A is the optimal choice. Conversely, point C is most suitable if the focus is just on minimizing costs. For scenarios where a balance between thermodynamic efficiency and costeffectiveness is desired, point B presents the ideal one.

NSGAII and GDE3 evolutionary algorithms perform superior to PESAII, SPEAII, SMPSO, and MOEA/D algorithms in generating the Paretooptimized solutions. MOEA/D algorithm is not an acceptable algorithm in this case.

The optimal values of the number and depth of boreholes are scattered uniformly between the specified lower and upper bounds; the optimum state is obtainable through various values between the defined bounds. The lower borehole radius and the higher tube radius contribute to attaining the optimized condition of the system.

The trend of entropy generation variations obtained from the 3D transient numerical simulation is similar to the semi2D steadystate analytical model, which affirms the efficiency of the optimization approach in determining the optimal configuration of BHE. Consequently, this approach provides designers with a reliable tool for choosing appropriate design parameters to achieve an optimum design.

Simulation results show that reducing the total borehole length lessens the entropy generation caused by fluid friction while increasing the entropy generation caused by temperature differences. The temperature difference has a more substantial effect than the fluid friction on the entropy generation in the BHE, so the total entropy generation increases by decreasing the total borehole length.
Availability of data and materials
The data generated or analyzed during the current study are available from the corresponding authors upon request.
Abbreviations
 \(B\) :

Borehole distance (\(\text{m}\))
 \(\dot{C}\) :

Cost rate (€/h)
 \(\text{C}\) :

Constant coefficient
 \(c\) :

Specific heat (\(\text{J}/\text{kg K}\))
 \({\text{C}}_{v}\) :

Volumetric heat capacity (\(\text{J}/{\text{m}}^{3}\text{K}\))
 \(\text{CC}\) :

Carrying charge
 \(\text{COP}\) :

Coefficient of performance
 \(D\) :

Halfshank spacing (\(\text{m}\))
 \(e\) :

Internal energy per unit mass (\(\text{J}/\text{kg}\))
 \(\text{f}\) :

Friction factor
 \(\text{FC}\) :

Fuel cost (€)
 \(h\) :

Convection coefficient (\(\text{W}/{\text{m}}^{2}\text{K}\))
 \(k\) :

Thermal conductivity (\(\text{W}/\text{mK}\))
 \(L\) :

Depth (\(\text{m}\))
 \(\dot{m}\) :

Mass flow rate (\(\text{kg}/\text{s}\))
 \(N\) :

Borehole number
 \({N}_{s}\) :

Dimensionless entropy number
 \(Nu\) :

Nusselt number
 \(\text{OMC}\) :

Operation and maintenance
 \(\Delta \text{P}\) :

Pressure drop (\(\text{Pa}\))
 \(\text{Pr}\) :

Prandtl number
 \(\text{PEC}\) :

Purchase equipment cost (€)
 \(\dot{Q}\) :

Design load (\(\text{W}\))
 \(Q\) :

Heat produced per unit volume (\(\text{J}/{\text{m}}^{3}\))
 \({q}{\prime}\) :

Heat transfer rate per unit length (\(\text{W}/\text{m}\))
 \(R\) :

Thermal resistance per unit length (\(\text{mK}/\text{W}\))
 \({R}^{*}\) :

Overall thermal resistance (\(\text{K}/\text{W}\))
 \(\text{Re}\) :

Reynolds number
 \(r\) :

Radius (\(\text{m}\))
 \({\dot{S}}_{\text{gen}}\) :

Entropy generation rate (\(\text{W}/\text{K}\))
 \(T\) :

Temperature (\(\text{K}\))
 \(\Delta T\) :

Temperature difference (\(\text{K}\))
 \(b\) :

Borehole/borehole wall
 \(\text{Cond}\) :

Conduction
 \(\text{Conv}\) :

Convection
 \(\text{elec}\) :

Electrical
 \(\text{f}\) :

Fluid
 \(\text{g}\) :

Grout
 \(\text{i}\) :

Inner
 \(\text{k}\) :

Component \(k\) th
 \(\text{L}\) :

Levelized
 \(\text{m}\) :

Average
 \(\text{max}\) :

Maximum
 \(\text{min}\) :

Minimum
 \(\text{o}\) :

Outer
 \(\text{s}\) :

Soil
 \(\text{t}\) :

Utube
 \(\text{tot}\) :

Total
 \(0\) :

Environmental condition
 \(1\) :

Inlet of Utube
 \(2\) :

Outlet of Utube
 \(\alpha\) :

Thermal diffusivity (\({\text{m}}^{2}/\text{s}\))
 \(\rho\) :

Density (\(\text{kg}/{\text{m}}^{3}\))
 \(\mu\) :

Dynamic viscosity (\(\text{kg}/\text{ms}\))
 \(\tau\) :

Time (s)
 \(\chi\) :

Dimensionless temperature
 \(\varphi\) :

Viscous dissipation term
References
Alkaff SA, Sim S, Efzan ME. A review of underground building towards thermal energy efficiency and sustainable development. Renew Sustain Energy Rev. 2016;60:692–713. https://doi.org/10.1016/j.rser.2015.12.085.
ASHRAE. ASHRAE handbookHVAC applications (SI). Atlanta: ASHRAE Inc.; 2011.
Banks D. An introduction to thermogeology: ground source heating and cooling. 2nd ed. Hoboken: John Wiley & Sons; 2012. https://doi.org/10.1002/9781118447512.
Bejan A. 37 minimization: the new thermodynamics of finitesize devices and finitetime processes. J Appl Phys. 1996;79:1191–218. https://doi.org/10.1063/1.362674.
Bejan A. Convection heat transfer. Hoboken: John Wiley & Sons; 2013. https://doi.org/10.1002/9781118671627.
Chang S, Feng G, Zhang L, Huang K, Li A. Multiobjective optimization of a photovoltaic thermal curtain wall assisted dualsource heat pump system. Appl Therm Eng. 2023;222:119845. https://doi.org/10.1016/j.applthermaleng.2022.119845.
Chen W, Yan J, Chen M, Li X. Diversity of Pareto front: a multiobjective genetic algorithm based on dominating information. J Control Theory Appl. 2010;8:222–8. https://doi.org/10.1007/s1176801072223.
Cheng X, Liang X. Discussion on the applicability of entropy generation minimization and entransy theory to the evaluation of thermodynamic performance for heat pump systems. Energy Convers Manag. 2014;80:238–42. https://doi.org/10.1016/j.enconman.2014.01.008.
Cheng XT, Liang XG. Role of entropy generation minimization in thermal optimization. Chin Phys B. 2017;26:120505. https://doi.org/10.1088/16741056/26/12/120505.
Choi W, Ooka R, Nam Y. Impact of longterm operation of groundsource heat pump on subsurface thermal state in urban areas. Sustain Cities Soc. 2018;38:429–39. https://doi.org/10.1016/j.scs.2017.12.036.
Corberan J, Finn D, Montagud C, Murphy F, Edwards K. A quasisteady state mathematical model of an integrated ground source heat pump for building space control. Energy Build. 2011;43:82–92. https://doi.org/10.1016/j.enbuild.2010.08.017.
Corne DW, Jerram NR, Knowles JD, Oates MJ. PESAII: regionbased selection in evolutionary multiobjective optimization. In: Proceedings of the 3rd annual conference on genetic and evolutionary computation. 2001. p. 283–90. https://doi.org/10.5555/2955239.2955289.
De Rosa M, RuizCalvo F, Corberán JM, Montagud C, Tagliafico LA. A novel TRNSYS type for shortterm borehole heat exchanger simulation: B2G model. Energy Convers Manag. 2015;100:347–57. https://doi.org/10.1016/j.enconman.2015.05.021.
Deb K, Pratap A, Agarwal S, Meyarivan T. A fast and elitist multiobjective genetic algorithm: NSGAII. IEEE Trans Evol Comput. 2002;6:182–97. https://doi.org/10.1109/4235.996017.
Emmerich MT, Deutz AH. A tutorial on multiobjective optimization: fundamentals and evolutionary methods. Nat Comput. 2018;17:585–609. https://doi.org/10.1007/s110470189685y.
Ghanadi Arab MM, Hajabdollahi M, Hajabdollahi H. MultiObjective optimization of a fin with twodimensional heat transfer using NSGAII and ANN. J Appl Mech Eng. 2013;2:1–7. https://doi.org/10.4172/21689873.1000117.
Goetzler W, Guernsey M, Kar R. Research and development roadmap: geothermal (groundsource) heat pumps. Burlington: Navigant Consulting Inc; 2012. https://doi.org/10.2172/1219848.
Gultekin A, Aydin M, Sisman A. Effects of arrangement geometry and number of boreholes on thermal interaction coefficient of multiborehole heat exchangers. Appl Energy. 2019;237:163–70. https://doi.org/10.1016/j.apenergy.2019.01.027.
Hein P, Kolditz O, Görke UJ, Bucher A, Shao H. A numerical study on the sustainability and efficiency of borehole heat exchanger coupled ground source heat pump systems. Appl Therm Eng. 2016;100:421–33. https://doi.org/10.1016/j.applthermaleng.2016.02.039.
Huang S, Ma Z, Cooper P. Optimal design of vertical ground heat exchangers by using entropy generation minimization method and genetic algorithms. Energy Convers Manag. 2014;87:128–37. https://doi.org/10.1016/j.enconman.2014.06.094.
Huang S, Ma Z, Wang F. A multiobjective design optimization strategy for vertical ground heat exchangers. Energy Build. 2015;87:233–42. https://doi.org/10.1016/j.enbuild.2014.11.024.
Huang Y, Zhang Y, Xie Y, Zhang Y, Gao X, Ma J. Longterm thermal performance analysis of deep coaxial borehole heat exchanger based on field test. J Clean Prod. 2021;278:123396. https://doi.org/10.1016/j.jclepro.2020.123396.
Hughes BR, Chaudhry HN, Ghani SA. A review of sustainable cooling technologies in buildings. Renew Sustain Energy Rev. 2011;15:3112–20. https://doi.org/10.1016/j.rser.2011.03.032.
Javadi H, Ajarostaghi SSM, Rosen MA, Pourfallah M. Performance of ground heat exchangers: a comprehensive review of recent advances. Energy. 2019;178:207–33. https://doi.org/10.1016/j.energy.2019.04.094.
Keshavarzzadeh AH, Ahmadi P, Safaei MR. Assessment and optimization of an integrated energy system with electrolysis and fuel cells for electricity, cooling and hydrogen production using various optimization techniques. Int J Hydrogen Energy. 2019;44:21379–96. https://doi.org/10.1016/j.ijhydene.2019.06.127.
Keshavarzzadeh AH, Zanjani AM, Gharali K, Dusseault MB. Multiobjective evolutionarybased optimization of a ground source heat exchanger geometry using various optimization techniques. Geothermics. 2020;86:101861. https://doi.org/10.1016/j.geothermics.2020.101861.
Kord AS, Jazayeri SA. Optimization and analysis of a vertical groundcoupled heat pump. Int J Renew Energy Res. 2012;2:33–7.
Kukkonen S, Lampinen J. GDE3: the third evolution step of generalized differential evolution. In: 2005 IEEE Congress on Evolutionary Computation. 2005. p. 443–50. https://doi.org/10.1109/CEC.2005.1554717.
Lamarche L, Kajl S, Beauchamp B. A review of methods to evaluate borehole thermal resistances in geothermal heatpump systems. Geothermics. 2010;39:187–200. https://doi.org/10.1016/j.geothermics.2010.03.003.
Li M, Lai AC. Thermodynamic optimization of ground heat exchangers with single Utube by entropy generation minimization method. Energy Convers Manag. 2013;65:133–9. https://doi.org/10.1016/j.enconman.2012.07.013.
Li M, Yang S, Liu X. Diversity comparison of Pareto front approximations in manyobjective optimization. IEEE Trans Cybern. 2014;44:2568–84. https://doi.org/10.1109/TCYB.2014.2310651.
Li W, Li X, Wang Y, Tu J. An integrated predictive model of the longterm performance of ground source heat pump (GSHP) systems. Energy Build. 2018;159:309–18. https://doi.org/10.1016/j.enbuild.2017.11.012.
Ma Z, Xia L, Gong X, Kokogiannakis G, Wang S, Zhou X. Recent advances and development in optimal design and control of ground source heat pump systems. Renew Sustain Energy Rev. 2020;131:110001. https://doi.org/10.1016/j.rser.2020.110001.
Maheshkumar P, Muraleedharan C. Minimization of entropy generation in flat heat pipe. Int J Heat Mass Transf. 2011;54:645–8. https://doi.org/10.1016/j.ijheatmasstransfer.2010.09.006.
Marcotte D, Pasquier P, Sheriff F, Bernier M. The importance of axial effects for borehole design of geothermal heatpump systems. Renew Energy. 2010;35:763–70. https://doi.org/10.1016/j.renene.2009.09.015.
Montagud C, Corberan JM, Montero Á, Urchueguia JF. Analysis of the energy performance of a ground source heat pump system after five years of operation. Energy Build. 2011;43:3618–26. https://doi.org/10.1016/j.enbuild.2011.09.036.
Mostafazadeh F, Eirdmousa SJ, Tavakolan M. Energy, economic and comfort optimization of building retrofits considering climate change: a simulationbased NSGAIII approach. Energy Build. 2023;280:112721. https://doi.org/10.1016/j.enbuild.2022.112721.
Naicker SS, Rees SJ. Longterm high frequency monitoring of a large borehole heat exchanger array. Renew Energy. 2020;145:1528–42. https://doi.org/10.1016/j.renene.2019.07.008.
Nam Y, Ooka R, Hwang S. Development of a numerical model to predict heat exchange rates for a groundsource heat pump system. Energy Build. 2008;40:2133–40. https://doi.org/10.1016/j.enbuild.2008.06.004.
Navidbakhsh M, Shirazi A, Sanaye S. Four E analysis and multiobjective optimization of an ice storage system incorporating PCM as the partial cold storage for airconditioning applications. Appl Therm Eng. 2013;58:30–41. https://doi.org/10.1016/j.applthermaleng.2013.04.002.
Nebro AJ, Durillo JJ, GarciaNieto, Coello CC, Luna F, Alba E. SMPSO: a new PSObased metaheuristic for multiobjective optimization. In 2009 IEEE Symposium on Computational Intelligence in Multicriteria Decisionmaking (MCDM). 2009. p. 66–73. https://doi.org/10.1109/MCDM.2009.4938830.
Pletcher RH, Tannehill JC, Anderson D. Computational fluid mechanics and heat transfer. 3rd ed. Boca Raton: CRC Press; 2012. https://doi.org/10.1201/b12884.
Pu L, Qi D, Li K, Tan H, Li Y. Simulation study on the thermal performance of vertical Utube heat exchangers for ground source heat pump system. Appl Therm Eng. 2015;79:202–13. https://doi.org/10.1016/j.applthermaleng.2014.12.068.
Qian H, Wang Y. Modeling the interactions between the performance of ground source heat pumps and soil temperature variations. Energy Sustain Dev. 2014;23:115–21. https://doi.org/10.1016/j.esd.2014.08.004.
Razmi A, Soltani M, Kashkooli FM, Farshi LG. Energy and exergy analysis of an environmentallyfriendly hybrid absorption/recompression refrigeration system. Energy Convers Manag. 2018;164:59–69. https://doi.org/10.1016/j.enconman.2018.02.084.
Rees SJ, He M. A threedimensional numerical model of borehole heat exchanger heat transfer and fluid flow. Geothermics. 2013;46:1–13. https://doi.org/10.1016/j.geothermics.2012.10.004.
Robert F, Gosselin L. New methodology to design ground coupled heat pump systems based on total cost minimization. Appl Therm Eng. 2014;62:481–91. https://doi.org/10.1016/j.applthermaleng.2013.08.003.
RuizCalvo F, Montagud C. Reference data sets for validating GSHP system models and analyzing performance parameters based on a fiveyear operation period. Geothermics. 2014;51:417–28. https://doi.org/10.1016/j.geothermics.2014.03.010.
Sayyaadi H, Amlashi EH, Amidpour M. Multiobjective optimization of a vertical ground source heat pump using evolutionary algorithm. Energy Convers Manag. 2009;50:2035–46. https://doi.org/10.1016/j.enconman.2009.04.006.
Self SJ, Reddy BV, Rosen MA. Geothermal heat pump systems: Status review and comparison with other heating options. Appl Energy. 2013;101:341–8. https://doi.org/10.1016/j.apenergy.2012.01.048.
Serageldin AA, Sakata Y, Katsura T, Nagano K. Thermohydraulic performance of the Utube borehole heat exchanger with a novel oval crosssection: numerical approach. Energy Convers Manag. 2018;177:406–15. https://doi.org/10.1016/j.enconman.2018.09.081.
Soltani M, Moradi Kashkooli F, DehghaniSanij A, Nokhosteen A, AhmadiJoughi A, Gharali K, et al. A comprehensive review of geothermal energy evolution and development. Int J Green Energy. 2019a;16:971–1009. https://doi.org/10.1080/15435075.2019.1650047.
Soltani M, Kashkooli FM, DehghaniSanij A, Kazemi A, Bordbar N, Farshchi M, et al. A comprehensive study of geothermal heating and cooling systems. Sustain Cities Soc. 2019b;44:793–818. https://doi.org/10.1016/j.scs.2018.09.036.
Soltani M, Farzanehkhameneh P, Kashkooli FM, AlHaq A, Nathwani J. Optimization and energy assessment of geothermal heat exchangers for different circulating fluids. Energy Convers Manag. 2021;228:113733. https://doi.org/10.1016/j.enconman.2020.113733.
Song Y, Zou M, Chen X, Deng J, Du T. Parameter optimization of passive heat supply tower of ground source heat pump based on NSGAII. Sol Energy. 2019;190:453–64. https://doi.org/10.1016/j.solener.2019.08.043.
Tang F, Nowamooz H. Longterm performance of a shallow borehole heat exchanger installed in a geothermal field of Alsace region. Renew Energy. 2018;128:210–22. https://doi.org/10.1016/j.renene.2018.05.073.
ÜrgeVorsatz D, Cabeza LF, Serrano S, Barreneche C, Petrichenko K. Heating and cooling energy trends and drivers in buildings. Renew Sustain Energy Rev. 2015;41:85–98. https://doi.org/10.1016/j.rser.2014.08.039.
Yang H, Cui P, Fang Z. Verticalborehole groundcoupled heat pumps: a review of models and systems. Appl Energy. 2010;87:16–27. https://doi.org/10.1016/j.apenergy.2009.04.038.
Zhang Q, Li H. MOEA/D: a multiobjective evolutionary algorithm based on decomposition. IEEE Trans Evol Comput. 2007;11:712–31. https://doi.org/10.1109/TEVC.2007.892759.
Zhu N, Hu P, Xu L, Jiang Z, Lei F. Recent research and applications of ground source heat pump integrated with thermal energy storage systems: a review. Appl Therm Eng. 2014;71:142–51. https://doi.org/10.1016/j.applthermaleng.2014.06.040.
Zitzler E, Laumanns M, Thiele L. SPEA2: improving the strength Pareto evolutionary algorithm. TIKreport, vol. 103. 2001. https://doi.org/10.3929/ethza004284029.
Acknowledgements
Not applicable.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or notforprofit sectors.
Author information
Authors and Affiliations
Contributions
SS: simulation, validation, data analysis, preparation of figures, and drafting; KG: conception, interpretation and discussion of results, editing and revising the manuscript; MS: conception, interpretation and discussion of results, editing and revising the manuscript; JN: editing and revising the manuscript; MD: conception, editing and revising the manuscript. All the authors read and approved the final manuscript.
Corresponding authors
Ethics declarations
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Sarshar, S., Gharali, K., Saffaripour, M. et al. Multiobjective optimization and longtime simulation of a multiborehole ground heat exchanger system. Geotherm Energy 12, 31 (2024). https://doi.org/10.1186/s40517024003109
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s40517024003109