Experimental Investigation on CIRCE-HERO for the EU DEMO PbLi/Water Heat Exchanger Development

The present paper describes the experimental campaign executed at the ENEA Brasimone Research Centre aiming at supporting the development of a PbLi/water heat exchanger suitable for the lithium–lead loops of the dual coolant lithium lead and the water cooled lithium lead breeding blankets of the EU DEMO fusion reactor. The experiments were performed in a test section named HERO, installed inside the main vessel of the lead–bismuth eutectic-cooled pool-type facility CIRCE. The test section hosts a steam generator bayonet tube mock-up in relevant scale, which was selected as a promising configuration for DEMO purposes. For the thermal-hydraulic characterization of the component, five tests were executed at different water pressures (6, 8, 12 MPa, two tests at 10 MPa), and liquid metal flow rates (40, 33, 27, 20, 10 kg/s). The experimental outcomes proved the technological feasibility of this novel steam generator and its suitability for the DEMO PbLi loops. The activity was completed with a post-test analysis using two versions of the system code RELAP5. Because the experiments were executed with lead–bismuth eutectic, a scaling analysis is proposed to find the equivalence with PbLi. RELAP5 code was applied to recalculate the experimental data using PbLi as working fluid.


Introduction
Within the EUROfusion R&D program, financed by the EU Horzon 2020 programme, for the realization of the demonstration fusion power reactor (DEMO) [1], a study was performed to support the development of a lithium-lead (PbLi)/water heat exchanger (HEX). Such a component is in charge of removing the nuclear heat deposited in the PbLi of the dual coolant lithium lead (DCLL) breeding blanket (BB) [2], producing steam suitable to properly feed the turbine unit, ensuring an efficient conversion of the thermal power to electricity. Moreover, the same design is suitable for removing the power deposited in the PbLi in case of unbalance in the water coolant lithium lead (WCLL) BB [3]. One of the configurations selected for DEMO purposes was the steam generator with bayonet tubes (SGBT) [4]. The SGBT concept stands out for its safety features, i.e., the reduced occurrence of water-Pb/Pb-alloy interaction [5][6][7], thanks to a double physical separation between the coolants and an easier monitoring of eventual leakages by pressurizing the separation region with inert gas.
The double wall concept has been studied since the 1960s for sodium reactor applications [8], with the aim of increasing the safety margin of the nuclear power plants by reducing the occurrence of liquid metal/water interaction. In fact, the use of Steam Generators (SG) with a single wall implies the possibility of a liquid metal-water reaction in case of tube rupture, negatively impacting the operation and safety of such plants. Besides the enhanced safety, the use of double wall tubes experienced in experimental breeder reactor (EBR) I and EBR II highlighted that despite the higher initial cost respective to the single wall tubes, the cost in terms of plant life and maintenance are lower [8], making the double wall solution convenient not only for safety concerns but also from an economical point of view.
The double wall concept, improved with the introduction of leakage monitoring abilities and the bayonet tube geometry, has been proposed for Generation IV heavy liquid metal fast reactors [9][10][11]. In particular, the solution turned out to be interesting for the pool-type lead fast reactors, where the steam generators are directly installed inside the reactor tank, acting as interface between the nuclear island and the secondary system. The adoption of a double wall allows the use of an intermediate loop and an intermediate heat exchanger to be avoided.
Similar considerations about the safety concerns can be applied to the fusion applications involving the use of liquid metals as coolant. For these reasons, the features of the SGBT make it suitable not only as SG for the DCLL BB PbLi loop, but also exploitable as HEX for the WCLL BB PbLi loop.
Since the SGBT solution to be connected to the PbLi system is prototypical, a set of experimental tests were planned and executed at ENEA Brasimone Research Centre (R.C.) to demonstrate the technological feasibility and performances of the component [12][13][14]. The test section (TS) involved in the experimental campaign was HERO (Heavy Liquid mEtal pRessurized water cOoled tubes) [15], which hosts a relevant scale mock-up of an SGBT consisting of a bundle of seven double-wall bayonet tubes with a dedicated leakage monitoring system. This test section, designed at ENEA Brasimone R.C. and implemented in the main vessel of the lead-bismuth eutectic (LBE)-cooled pool-type facility CIRCE (CIRColazione Eutettico) [16], deals with the investigation of the thermalhydraulic features of prototypical systems and components, providing at the same time an experimental database suitable for thermal-hydraulic validation of numerical tools (i.e., system codes and Computational Fluid Dynamic codes). The HERO SGBT unit was connected to a dedicated secondary loop that provides the feedwater at high temperature and pressure (up to 18 MPa). Both primary and secondary systems were instrumented for monitoring and control of the facility operation, as well as to acquire a complete set of experimental data.
The thermal-hydraulic characterization of the HERO SGBT was executed through a set of five experimental tests [12] with different water pressures (6 MPa,8 MPa, two tests at 10 MPa, 12 MPa). In each of the five tests, an experimental sensitivity analysis was carried out, changing the mass flow rate of the LBE in four tests (40,33,27,20, 10 kg/s) and the flow rate of SGBT feedwater in the repeated test at 10 MPa (LBE mass flow rate constant at 27 kg/s). The experimental outcomes were used to demonstrate the technological feasibility of the prototypical SGBT, proving its suitability for the PbLi loops of DCLL BB and WCLL BB.
On the basis of the secondary circuit layout and the HERO SGBT geometry, a numerical post-test analysis was performed using two versions of the RELAP5 code: the RELAP5-3D © Ver. 4.3.4 [17] and a modified version of the RELAP5/Mod3.3 [18,19], which includes heavy liquid metals as working fluids. The numerical outcomes were compared with the experimental results in order to assess the code capability in predicting thermal-hydraulic performances of the component and to study the heat transfer in liquid metal, supporting the code validation process for fusion applications.
Finally, a further analysis was carried out, applying scaling methods to assess a correspondence between the LBE and the PbLi [12]. The scaling was performed in two different ways, i.e., preserving the convective heat transfer (method #1) and preserving both the thermal power and the temperature difference (method #2).
The present paper aims at describing the HERO SGBT main features, along with the primary and secondary systems of the CIRCE facility. Then, the experimental tests are presented along with the experimental outcomes achieved, focusing on the thermal-hydraulic performances of the HERO SGBT when operated in different working conditions. The computational analysis and the scaling methods for LBE and PbLi are presented to provide a complete thermal-hydraulic analysis of the HERO SGBT.

CIRCE-HERO Layout
The pool-type LBE-cooled facility CIRCE is an experimental platform located at ENEA Brasimone R.C. [20,21] and used to perform large-scale integral tests [22] and component thermal-hydraulic characterization [13,23]. The facility consists of a cylindrical main vessel, heating and cooling systems for LBE, storage and transfer tanks, and auxiliary systems for LBE circulation and gas recirculation. The main vessel has a height of about 8500 mm, an inner diameter of 1170 m, and a wall thickness of 15 mm. The vessel stores an LBE amount of about 70 tons (up to 500 mm from the top flange), covered with inert gas (argon) maintained at~0.2 barg.
The HERO TS is installed within the CIRCE main vessel from the top through a coupling flange. The CIRCE main vessel along with the HERO TS main components are shown in Figure 1. The LBE flows upwards, passing through the heating source called the fuel pin simulator (FPS, red in Figure 1) consisting of 37 electrically heated pins, then it reaches the fitting volume (green) and it flows through the riser (yellow) up to the separator (gold), where the LBE free level reaches about the middle height of the walls. From the separator, the LBE enters the shell side of the SGBT (blue) and it flows downwards, coming back into the pool. The LBE circulation is performed thanks to the difference in height of the thermal barycenters of FPS and SGBT (natural circulation) or by means of an argon injection device, consisting of a nozzle connected to a dedicated argon line and placed at the riser inlet section, which allows it to enhance the LBE circulation (gas-enhanced circulation). The mixture of LBE/argon flows upward through the riser. Then, once that the mixture reaches the separator, the LBE is naturally separated from the argon, which migrates in the cover gas of the main vessel while the pure LBE flows in the shell side of the SG. Thanks to the separator, it is possible to avoid the argon inlet inside the steam generator.
The HERO SGBT is an experimental mock-up that uses heavy liquid metal (i.e., LBE in CIRCE) as the primary working fluid (shell side) and pressurized water as the secondary coolant (tube side). It is designed to study the thermal-hydraulic performances of an SGBT in nuclear applications involving heavy liquid metals as the primary coolant. The component has undergone several experimental campaigns in a broad range of operating conditions, supporting in such a way the development of heavy liquid metal technologies for fission [22,23] and fusion [12,13] applications, as well as the validation process of numerical tools [24][25][26].
The technical draw of the HERO SGBT unit is depicted in Figure 1. It is composed of: • The helium chamber (ex-vessel), for pressurizing the AISI316L powder gap with inert gas; • The steam chamber (ex-vessel), collecting the steam arising from the bayonet tubes (BTs); • The tube bundle, composed of 7 BTs with an active length of 6000 mm, arranged with a triangular pitch in a hexagonal shroud; and • The top flange on the top of the main vessel, which sustains the steam and helium chambers, the BTs, and the hexagonal shell.
Each of the seven BTs consists of four coaxial tubes, as depicted in Figure 2: the slave tube, through which the feedwater flows downward; the first and second tube, forming the annular region where the steam is produced flowing upward; and finally, the third tube, which forms with the second tube a gap pressurized with helium at~0.8 MPa to detect any leakages and is filled by AISI316L powder. The adoption of the stainless-steel powder allows the heat exchange to be increased and, thus, the overall performance of the component, thanks to its good thermal conductivity. The slave and first tube form a first annular gap filled by air acting as an insulator to avoid steam condensation. The hydraulic connection of the SGBT with the LBE circulating in the primary system is obtained by means of six openings drilled into the SG hexagonal shroud 300 mm above the separator bottom, which allow the primary coolant inlet on the SG shell side. A dedicated once-through secondary loop is installed to feed the SGBT with demineralized water at high temperature and pressure (up to 18 MPa). The water loop, depicted in Figure 3, is composed of a feeding line composed of a demineralizer, a volumetric pump, a helical pre-heater, and a manifold for the water distribution among the seven BTs. Downstream of the SGBT, the steam chamber is connected to a discharge line that allows the exit of the steam produced by the SG to the environment. Furthermore, upstream of the manifold, a bypass line is installed and used for the operational transients of the system (i.e., startup and shutdown phases). A helium line is also installed for pressurizing the AISI316L powder gap of the BTs at~0.8 MPa.
Both the primary and secondary systems are instrumented for monitoring and operating the facility, as well as to acquire experimental data. On the primary system, 119 TCs are distributed along the pool to capture mixing and stratification phenomena [16]. Each component of the TS is equipped with thermocouples located at the inlet and outlet sections (e.g., fitting volume, riser, separator) for their thermal-hydraulic characterization. The FPS and the SGBT shell side is more deeply instrumented by bulk TCs and wall TCs, positioned at the inlet/outlet sections and at different elevations in different sub-channels for better monitoring of the temperatures along their active lengths. Details on the FPS TC positions can be found in [27], whereas the TCs installed along the LBE side of the SGBT are reported in Figure 3 (red and azure identification codes for bulk and wall temperatures, respectively) in order to facilitate the comprehension of the experimental outcomes. A Venturi flow meter (VFM) is installed upstream of the FPS to measure the LBE mass flow rate. Bubble tubes injecting argon [28] are installed for absolute and differential pressure measurement within the main vessel and along the TS.  • Five TCs located in the central tube, named T0 and set at different levels (+500 mm, +1500 mm, +3000 mm, +4200 mm, +6000 mm, assuming the bottom part of the bayonet tube is 0 mm), aiming to characterize water vaporization; • Seven TCs at the exit of each bayonet tube annular riser; • Three TCs at the steam chamber outlet to detect eventual condensation and radial stratification; and • Four differential pressure transmitters to measure the pressure across 4 BTs in order to characterize pressure drops in single-and two-phases flow conditions.
More details on the CIRCE-HERO instrumentation can be found in [15,16,23].

EUROfusion Experimental Campaign
Within the EUROfusion Work Package Balance of Plant task, an experimental campaign was executed as part of the research activities undertaken at ENEA Brasimone R.C. to investigate the PbLi technology [29][30][31]. A test matrix of five tests was set up for the CIRCE-HERO facility in order to obtain experimental data in operating conditions relevant for the PbLi/water HEX [12] of the EU DEMO fusion reactor. In the following, the five tests are presented and the main outcomes of the experimental campaign are described. For each test, the boundary conditions to be included in the test matrix were defined by means of a preliminary numerical simulation using the system thermal-hydraulic (STH) code RELAP5-3D © Ver. 4.3.4 [4,32]. Table 1 summarizes the conditions assumed for the five tests. Each test is marked by the initials EF-T, meaning EUROfusion-Test, followed by the test number.
The tests focused attention on the HERO SGBT and its thermal-hydraulic performances. In particular, the component was tested on the basis of the secondary side operating parameters, i.e., pressure, feed water inlet temperature, and flow rate. For this purpose, the secondary system was managed assuming four different values of pressure: 8 MPa in EF-T1, 10 MPa in EF-T2 and EF-T5, 12 MPa in EF-T3, and 6 MPa in EF-T4, kept constant at the HERO outlet section by properly regulating the V3 valve (see Figure 3). These pressure values were selected in order to perform a full characterization of the component in different operating conditions. At the inlet of the BTs, the feedwater temperature was set by assuring few degrees of sub-cooling respect to the saturation temperature at the corresponding pressure, assuming in such a way the values of 280 • C in EF-T1; 300 • C in EF-T2, EF-T3 and EF-T5; and 250 • C in EF-T4. The water pre-heating was realized by managing the power supplied by the helical heater (see Figure 3). In accordance with the feedwater temperature, the LBE SG inlet temperature was also changed in order to achieve a relevant temperature difference between the primary and secondary coolant, but avoiding at the same time an excessive ∆T along the double wall thickness of the BTs. For these reasons, an LBE temperature of 450 • C was set in EF-T1; 480 • C in EF-T2, EF-T3 and EF-T5; and 430 • C in EF-T4. The LBE temperature at the SG inlet section was kept constant by managing the power supplied by the FPS. Furthermore, an experimental sensitivity analysis was performed by changing the mass flow rate of the LBE (tests EF-T1, EF-T2, EF-T3, and EF-T4) and the flow rate of the feedwater in the steam generator (EF-T5). In particular, the LBE mass flow rate reached an initial value of~38 kg/s (gas-enhanced circulation regime), from which it was reduced in a total of 5 steps, regulating the properly injected argon flow rate, up to a final value of~6/~8 kg/s (natural circulation regime). The LBE mass flow rate operating range in the forced circulation regime was selected to reproduce the liquid metal velocities expected in the DCLL BB PbLi/water heat exchangers, which are in the range of 0.5-0.1 m/s. In fact, considering the geometry of the HERO tube bundle and the shell side flow area, the LBE average velocities achieved with the LBE mass flow rate values reported in Table 1 were in the range of 0.15-0.5 m/s. In natural circulation, the LBE velocity field is expected to be lower than 0.1 m/s, thus represening the PbLi flow conditions in the WCLL BB PbLi loop (velocities lower than 0.1 m/s). The EF-T5 focuses the attention on the secondary loop, in particular on the feedwater flow rate, which was decreased from an initial value of~0.275 kg/s to~0.175 kg/s in a total of 4 steps. In such a way, it is possible to test the performances of the steam generator when different feedwater flow rates are supplied, evaluating the degree of superheated steam that the component is capable of producing. The experimental conditions reached during the tests are presented in Table 2, which reports the average of the measured values. Details about the LBE mass flow rate average values measured by the VFM and LBE SGBT average inlet temperature are reported in Figure 4, where the x-axes represent the steady states (SS) for each test. The steady state is defined as the time lapse in which the desired test conditions are reached and the main operating parameters (i.e., LBE and water mass flow rates and temperatures, water pressure) remain constant. In EF-T1 and EF-T2, the LBE mass flow rate was in the range of~38-39 kg/s, whereas it changed from 38.5 kg/s to 4.5 kg/s in EF-T3 and from 30.5 kg/s to 5.5 kg/s in EF-T4. In EF-T5 it was constant at the corresponding designed value of 27 kg/s. Concerning the LBE SGBT inlet temperature, it reached the designed value in EF-T3 and EF-T5, whereas in the other tests it was~2-4 • C higher than the designed temperatures. A lower value was reached in SS5 of EF-T3, where the LBE average inlet temperature was~473 • C (see Figure 4b). It can also be observed for the feedwater temperature that in EF-T2 and EF-T3 it was~4-5 • C lower than the designed values (see Tables 1 and 2), whereas for the other cases it reached the values expected in Table 1. Finally, the water pressure reached the same values of the designed ones for all the tests, and the water flow rate reached values slightly higher in EF-T1 and EF-T2 (difference of 0.02 kg/s) and close to the designed ones in the other tests.

Experimental Tests Results
The experimental outcomes are hereafter presented for all tests and for all steady states, focusing attention on the LBE side of the SG in order to characterize the heat transfer in the liquid metal side and on the water outlet conditions.

Primary Side Outcomes
The LBE temperatures are reported in Figure Figure 2) in order to represent the thermal field along the LBE side of the SG active length. In tests EF-T1 to EF-T4, the LBE outlet temperatures (TC-0X-L00) reached the highest average values during SS1. Then they decreased in the next SSs, coherently with the LBE mass flow rate reduction, achieve the lowest values in SS5. The outlet temperatures achieved in the tests were different because the LBE SGBT inlet temperature was different. In particular, the outlet temperature As a consequence, it can be noticed that from SS1 to SS5 there was an increase in the temperature difference between the inlet and the outlet sections of the active length. The ∆T inlet-outlet increased from 86.3 • C (SS1) to 148.7 • C (SS5) in EF-T1, from 84.2 • C to 162.5 • C in EF-T2, from SS1 90 • C to 162.5 • C in EF-T3, and from 98 • C to 162.5 • C in EF-T4. In EF-T5, the LBE average outlet temperature was about 374 • C during SS1,~364 • C during SS2,~380 during SS3, and~382 • C during SS4 (see Figure 9a). The temperature difference between the inlet and outlet of the SG active length was about 100 • C for all the steady states of EF-T5. It can be noticed also that considering the instrumented levels, the LBE temperature difference was smaller among the lower levels (i.e., among 0.0 mm, +1500 mm, +3000 mm, and +4200 mm), whereas it was larger in the higher levels, i.e., between +4200 mm and the SG inlet (6000 mm). These differences along the SG active length were mainly due to the feedwater, which entered with some degrees of sub-cooling, and this delayed the beginning of the vaporization, which occurred at higher levels. At such levels, thanks to the high efficiency of the vaporization process, the fraction of thermal power removed from the LBE was higher and the LBE temperature differences were larger as a consequence.       Taking into account the information on the LBE mass flow rate reported in Table 2 Figure 9b. From EF-T1 to EF-T4, the fraction of thermal power removed was the highest in SS1, whereas the lowest fraction was achieved in SS5. In particular, the power removed decreased from~465 kW (SS1) tõ 200 kW (SS5) in EF-T1, from~520 kW to~200 kW in EF-T2, from~485 kW to~100 kW in EF-T3, and from 450 kW to 130 kW in EF-T4. In EF-T5 (Figure 9b), the highest values of thermal power removed were reached during SS1 with~420 kW and SS2 with~443 kW, whereas SS3 and SS4 were characterized by a lower power fraction of~380 KW and 370 kW, respectively.

Secondary Side Outcomes
Concerning the secondary loop, Figure 10a reports the average temperature at the steam chamber outlet section for all the tests. It can be observed that in all the tests and relative steady states, except for EF-T5, the outlet temperatures reached values equal to or slightly above the saturation temperature corresponding to the operating pressure of each test. This means that in the reproduced operative conditions, the SGBT was capable of producing saturated steam. In EF-T5, the outlet temperature underwent a relevant increase from SS1 to SS4 as consequence of the progressive reduction of the feedwater flow rate injected in the BTs. SS1 was characterized by an outlet temperature of 318 • C, 7 • C degrees above the saturation temperature at a pressure of 10 MPa. In the following steady states, the temperature increased to~337 • C in SS2 and~390 • C in SS3, up to a final value of~422 • C in SS4, which was well above the saturation limit of 311 • C. These results prove that the SGBT is capable of reaching high degrees of superheating, producing high temperature steam suitable for the inlet in a turbine. In a two-phase flow system, it is possible to evaluate the thermodynamic quality based on a balance between the specific enthalpy of liquid and steam, respectively, as shown in Equation (1): where: • x t is the thermodynamic quality; • h is the average specific enthalpy of the mixture liquid/steam; • h v is the enthalpy of the steam in saturation conditions at the pressure of the system; and • h l is the enthalpy of the liquid water in saturation conditions at the pressure of the system.
Since the feedwater inlet conditions (pressure and temperature) are measured during the experiment, h v and h l are well defined, whereas h can be estimated by applying the thermal balance equation, assuming the power removed from LBE by the secondary coolant and the feedwater flow rate (Table 2) as a reference for the calculations. Figure 10b reports the outcomes of this calculation for all the tests. For EF-T1, in SS1, SS2, and SS3 the highest steam production was achieved (x t~0 .8/~0.9), whereas SS5 was characterized by a small fraction of steam generated (x t~0 .4) due to the low thermal power removed from the primary system. The highest values were also achieved in SS1 and SS2 for EF-T2, where x t was slightly above 1, indicating the complete vaporization of the water, with a slight superheating. SS3 was characterized by a complete vaporization (x t~1 ), whereas in SS4 and SS5 the thermodynamic quality was~0.85 and~0.4, respectively. The same values were reached in EF-T3, except for SS5, where the thermodynamic quality was very low (about 0.1). In EF-T4, the values of x t remained below 1 during all the tests. In SS1 the highest steam production was achieved (x t~0 .85), then it decreased to~0.75 in SS2,~0.65 in SS3, and~0.5 in SS4, reaching a final value of~0.2 in SS5. Finally, in EF-T5, the thermodynamic quality x t was above 1 for all the steady states, proving that the superheated steam was produced by the SGBT for the entire duration of the test. In particular, the initial value in SS1 was about 1.1; then it increased coherently with the feedwater flow rate reduction, reaching the values of~1.4 in SS2 and SS3 and~1.6 in SS4.

RELAP5 Post-Test Analysis
A numerical model of the secondary system including the SGBT was set up and used to perform a numerical analysis with two versions of the STH code RELAP5: the standard version of RELAP5-3D © Ver. 4.3.4 [17] and a version of RELAP5/Mod3.3 [34], which was modified by implementing the fluid properties of three liquid metals (Pb, LBE, PbLi) and three heat transfer correlations for heavy liquid metals [18,35,36]: Seban-Shimazaki (used for non-bundle geometry) and Ushakov and Mikityuk (used for bundle geometries). The 1-D model includes the SGBT, with the seven double wall BTs modeled separately, the feedwater line, the bypass line, and the discharge line [13]. Each BT consists of a single pipe component simulating the slave tube, and an annulus component, which represents the annular region between first tube and second tube. Furthermore, an equivalent single pipe simulates the LBE shell side of steam generator. The water and LBE inlet thermodynamic conditions and the environmental conditions for the steam discharge are defined using time-dependent volumes, whereas water and LBE mass flow rate are set by means of timedependent junctions. The spatial division of the SG and secondary system components was realized by taking into account the correct positions of the instrumentation placed along the loop, as well as the wall and bulk TCs installed in the active length of the SGBT. The full nodalization can be found in [16].
Dedicated heat structures simulate the thermal coupling between the annular riser of each BT and the LBE equivalent channel. Standard RELAP5-3D uses the Westinghouse correlation for liquid metal convective heat transfer in the HERO tube bundle zone [37,38], whereas the Ushakov correlation is used in RELAP5/Mod3.3. The BT downcomer was instead assumed to be thermally insulated with respect to the annulus. The feeding line, bypass line and discharge line were also assumed to be thermally insulated. The water warm-up inside the helical pre-heater was simulated by introducing an additional heat structure, which reproduced a heating source uniformly distributed along the helical pipe length, as well as along the pipe thickness.
For the validation purpose, the analysis presented in this section is focused on the heat transfer within the liquid metal side of the SGBT, comparing the computational outcomes with the experimental data acquired on the primary side of the TS.
For the simulation of each test, the experimental boundary conditions reported in Table 2 were assumed in the input deck. In particular, LBE and water mass flow rates were assumed by considering the measurements of the VFM and the TFMs, respectively. The temperatures at the HERO SG inlet section were assumed by considering the average of the temperatures measured by the TC-SG-0X for the LBE and TC-TX-I for the water. Finally, the pressure at the outlet section of the secondary loop was assumed constant for the duration of each test and set equal to the measure of PC-L3-1.
First, the RELAP5-3D results were compared with the average values of the experimental temperatures obtained along the LBE side of the SG and the wall temperatures for all tests and steady states.
The results of the LBE temperature for all the tests are reported in Figure 11a, Figure 12a, Figure 13a, Figure 14a, Figure 15a. The results show that, in general, the code predicted the LBE temperature with good accuracy, as minor differences were observed between the numerical values and the average experimental acquisition. A discrepancy was found in SS4 of EF-T3 and SS4 of EF-T5, where RELAP5-3D code overestimated the experimental temperatures of~10/~15 • C. It was also observed in EF-T4 that the temperatures were overestimated by the code, about~10/~20 • C. Such discrepancies were probably due to an unbalanced distribution of the LBE flow in the sub-channels, which could not be reproduced by the numerical model since it simulates the LBE channels with a single 1-D equivalent pipe.
The comparison between the measured wall temperatures and RELAP5-3D results are presented in Figure 11b, Figure 12b, Figure 13b, Figure 14b, Figure 15b. It can be observed that in all the instrumented sections the wall temperatures were~10/~20 • C lower than the LBE sub-channel temperatures. This difference was clearly visible in SS1, SS2, SS3, and SS4, whereas in SS5 the difference was less pronounced because of the low LBE mass flow rate, and thus a low velocity, which led to a more homogeneous radial thermal field. The differences in temperatures between the LBE and wall tubes were also predicted by the RELAP5-3D results. Nevertheless, the comparison between the experimental and numerical values demonstrates that there was a difference between the results because the code underestimated the temperature. The RELAP5-3D code underestimated the temperatures by~15/~20 • C in EF-T1, EF-T2, and EF-T4. In EF-T3, the RELAP5-3D temperatures were generally lower than the experimental values and minor differences were observed, with the exception of SS3 and SS4 at section +1500 mm, where the RELAP5-3D value was greater than the experimental value of~10 • C. This last discrepancy was probably due to a not-uniform LBE flow distribution among the sub-channels. As already mentioned, the numerical model simulates the LBE channel with a 1-D equivalent pipe. This means that the sub-channels effects are not reproduced and the resultant thermohydraulic behavior is a consequence of the averaged values. Finally, in EF-T5, the wall temperatures calculated by RELAP5-3D were in agreement with the measured ones, except in SS1, where the code underestimated the wall temperature by~20 • C.  The discrepancies between the experimental and numerical wall temperatures could be addressed to the simplified model, which uses an LBE equivalent pipe, and could not reproduce the sub-channel effects. Other reasons could be the uncertainty related to the acquisition of the experimental measurements, eventual unbalanced distribution of the LBE flow rate in the sub-channels and feedwater among the seven bayonet tubes [25], as well as the uncertainty related to the heat transfer correlations used by RELAP5-3D.   The thermal conductivity of the HERO double-wall gap could also be a source of uncertainty. In fact, the thermal conductivity of the AISI316L powder + He used to fill the gap is a function of the temperature and is influenced by several factors, i.e., powder compaction, grain size and growth, and thermal cycling [39]. Dedicated experimental campaigns performed at ENEA Brasimone R.C. allowed the experimental correlations reported in [40] to be derived. Such correlations were taken as reference for the simulations since they were used for the HERO SGBT design.
Finally, during the operation of the system, the LBE coolant may have formed compounds that attached to the outer surface of the tubes composing the HERO bundle. This could have caused an additional thermal resistance [41], which could have enlarged the discrepancies between the experimental and numerical outcomes. Nevertheless, this hypothesis can be verified only with a direct inspection during the next refurbishment of the facility.
The same simulations were performed using RELAP5/Mod3.3 in order to compare the performances of the codes. The results of the simulations were compared with the RELAP5-3D outcomes in Table 3 for all the steady states of EF-T3 in terms of the LBE temperatures and wall temperatures along the SG shell side. Table 3. EF-T3: RELAP5-3D/RELAP5/Mod3.3 (in bold) simulation results comparison. The results demonstrate that there was good agreement between the two codes. Small discrepancies of~0.5 • C/~1.5 • C between codes can be observed due to the different heat transfer correlations used (i.e., Westinghouse correlation for RELAP5-3D and Ushakov correlation for RELAP5/Mod3.3) and the fact that the RELAP5/Mod3.3 code considers the reference properties of the LBE reported in [33], which were updated respective to those available in RELAP5-3D [17]. The comparison presented in Table 3 is representative of the ones carried out for all the other tests.

Analysis of the Experimental Data in Case of PbLi Working Fluid
The CIRCE experimental facility, and thus the steam generator test section, uses LBE as primary fluid, whereas PbLi is the fluid selected for applications in DEMO. It should be noted that both fluids rely on the same formulations of the Nusselt number. Therefore, the following sections report an analysis of the correspondence between LBE and PbLi, preserving the convective heat transfer (method #1) and both the thermal power and temperature difference (method #2).

Method #1
Method #1 aims to maintain the same conditions of convective heat transfer across the active length of the SG, preserving the Nusselt number. This condition can be satisfied by changing the experimental mass flow rate, taking into account the thermo-physical differences between LBE and PbLi. The reference parameter to apply this method is the Peclet number. First, the Peclet number was evaluated for each test from the available experimental data. The LBE density, thermal conductivity, and heat capacity were calculated on the basis of the LBE average temperature and in accordance with the correlations reported in [33]. Then the Peclet number was calculated. The same parameters (density, thermal conductivity, and heat capacity) were also calculated, considering the PbLi as the working fluid. Assuming the same values as the Peclet, it was possible to recalculate the equivalent theoretical PbLi mass flow rate as reported in Table 4. A new RELAP5-3D run was set using the model introduced in Section 5 and running the calculation using the PbLi as the working fluid. Unlike the liquid metal mass flow rate, the initial and boundary conditions of the simulations corresponded to those summarized in Table 2.
The simulation outcomes are presented in Table 4, in which the experimental and RELAP5-3D Nusselt numbers are compared for EF-T1 and EF-T2. The Nu was calculated with the heat transfer correlation used for liquid metals in a rod bundle (Westinghouse correlation): Nu = 4.0 + 0.33(P/D) 3.8 (Pe/100) 0.86 + 0.16(P/D) 5 , where • Nu is the Nusselt number; • P/D is the pitch-to-diameter ratio (in HERO geometry, P/D = 1.42) of the rods; and • Pe is the Peclet number.
Comparing the experimental and computed values, it emerged that the errors were in range of 26.1-33.6% for EF-T1 and 24.1-38.2% for EF-T2. These discrepancies could be partially due to the uncertainty related to the use of Equation (2) for the calculation of Nu and to the differences between the thermo-physical proprieties of RELAP5-3D (i.e., c p = 180.5 J/kgK at 400 • C) and the most updated values in [42] (i.e., c p = 187.8 J/kgK at 400 • C). Neglecting the contribution of the fluid properties, the errors decreased in the range of 18-29% for EF-T1 and 15.7-28.7% for EF-T2. Similar differences were found by applying the scaling method #1 to the other tests.

Method #2
This method preserves the power exchanged and the difference of temperature across the SG. The experimental mass flow rate is changed considering the thermo-physical differences between LBE and PbLi. The reference parameter considered for the correspondence is the heat capacity. The PbLi mass flow rate (see Table 5) was re-calculated and used as a boundary condition to perform simulations with RELAP5-3D, using PbLi as the working fluid. In this way it was possible to compare the experimental data with the simulations involving PbLi. The simulation results are summarized in Table 5, where they are compared with the LBE temperatures at the SGBT outlet section in EF-T1 and EF-T2. Absolute errors were within the range of 3.8-7.8 • C for EF-T1 and in the range of 2.4-7.2 • C for EF-T2. Taking into account the thermo-physical propriety differences between the ones from RELAP5-3D and the most updated values in [42], the expected discrepancies in the calculation of the outlet temperatures were in the range of~4-6 • C. This implies that, neglecting this contribution, the errors became consistent with the uncertainty of the experimental measurements. Similar differences were found by applying scaling method #2 to the other tests.

Conclusions
Within the R&D program for the DEMO development, a research activity was addressed to develop a PbLi/water HEX capable of efficiently removing the thermal power from the DCLL BB and for the PbLi loop of the WCLL BB. An experimental and numerical investigation was performed involving the HERO SGBT at ENEA Brasimone R.C. with the aim of investigating the thermal-hydraulic characteristics of the component, improving the knowledge and the experience in terms of design and operations, and providing an experimental database to support the validation process of numerical tools. The paper describes five characterization tests in which the HERO SGBT was involved in the pooltype LBE-cooled facility CIRCE. The results of the research activity led to the following conclusions: • During each test, the steady state conditions and the designed boundary conditions needed to test the SG were achieved. In the LBE side (primary side), the FPS managed to supply the thermal power necessary to balance the power removed by the SGBT and the fraction of power lost from the CIRCE pool toward the environment, keeping the LBE temperature at the SG inlet section as close as possible to the target values (maximum discrepancy~4 • C). In EF-T5 the steam outlet temperature was well above the saturation limit of 311 • C (i.e.,~337 • C in SS2 and~390 • C in SS3, up to the final value of~422 • C in SS4). These results demonstrate that the SGBT is capable of reaching high degrees of superheating, producing high temperature steam suitable for the inlet in a turbine.

•
The thermal power removed by HERO for each SS and for all the tests was assessed by applying the thermal balance equation: The highest fraction was achieved in EF-T2-SS1 (~520 kW), whereas the lowest fraction was achieved in EF-T3-SS5 (~100 kW).

•
A numerical post-test analysis using the STH codes RELAP5-3D © Ver. 4.3.4 and RELAP5/Mod3.3 was performed by exploiting the outcomes of the experimental tests. Both versions of the code were demonstrated to have the capability to simulate this type of component, in particular as regards the liquid metal side. Minor differences (~0.5-1.5 • C) were observed between the two codes. The most relevant discrepancies were observed from the comparison of the SGBT wall temperatures calculated by the codes and the experimental ones. The reasons for such discrepancies could be found in the numerical model setup, which could not reproduce the LBE sub-channel effects and possible unbalance in the LBE flow. The analysis of the results also highlighted a possible source of uncertainty in the powder + He gap and its thermal conductivity, which was influenced by different factors (i.e., powder compaction, grain size and growth, thermal cycling) during the HERO SGBT operation. The eventual formation of LBE compounds around the HERO tubes and the consequent increase of thermal resistance could also enhance such discrepancies. Verification of this last hypothesis will be made when the HERO SGBT is dismantled. • Two scaling analysis approaches were proposed to find an equivalence between LBE and PbLi. RELAP5-3D code was applied to recalculate the data of the experiments using PbLi as the working fluid. Considering method #1, which preserves the convective heat transfer, and comparing experimental and calculated values, it emerged that the errors in terms of Nu were in the range of 26.1-33.6% for EF-T1 and 24.1-38.2% for EF-T2. These discrepancies could be partially due to the uncertainty related to the correlation used by RELAP5-3D for the calculation of Nu and to the differences of the thermo-physical proprieties between the ones from RELAP5-3D and the most updated values. Neglecting the contribution of the fluid properties, the errors decreased in the range of 18-29% for EF-T1 and 15.7-28.7% for EF-T2. Considering method #2 and thus preserving the thermal power and the difference of temperature, the code predicted differences in the range of 4-6 • C on the outlet temperatures, thus neglecting the contribution of the PbLi properties. The errors of the code results were in range of −1.8 • C and 0.6 • C. The same differences were found by applying the scaling methods to the other tests. Of the two methods, method #2 seemed to be more appropriate for the purposes of the present activity, since it preserved the thermal power exchanged, and thus the main figure of merit, which represented the efficiency of the component.
In conclusion, the activity carried out highlighted that the component has good capability to remove the thermal power from liquid metal, producing at the same time high quality steam suitable for feeding a steam turbine unit. The good efficiency, combined with the enhanced safety features given by the double wall tubes, make the SGBT concept suitable for DEMO purposes.