Community Research and Development Information Service - CORDIS

FP7

CSP2 Report Summary

Project reference: 282932
Funded under: FP7-ENERGY

Final Report Summary - CSP2 (Concentrated Solar Power in Particles)

Executive Summary:
The CSP2 project proposes to use dense gas-particle suspensions -DPS- (approximately 30-40% of solid) in tubes as alternative heat transfer fluid (HTF) in order to extend working temperature and to decrease environmental impact of standard HTF used in concentrating solar power (CSP) plants. The assembly of these tubes will constitute the solar absorber (receiver) placed at the top of a central receiver CSP system. The radiation absorbed by these tubes submitted to the concentrated solar radiation is transmitted by conduction, convection and radiation towards the suspension, which warms up by contact with the hot walls. This suspension circulates easily between the inlet and the outlet of the solar receiver and transports the absorbed energy towards the energy storage. Therefore this HTF enables to dissociate the production of electrical or thermal energy from the solar energy collection (as currently done in molten salts CSP plants, and cannot be done with other HTF such as gas or steam). The proposed HTF is totally safe; it permits higher possible operating temperature (700°C) and does not risk solidification at low temperature. The project includes fundamental studies on particle hydrodynamics and heat transfer in small diameter tubes, construction and testing of a laboratory 150 kW solar pilot and process scaling up to different sizes.
In the frame of the project, the influence of the various operating hydrodynamic parameters was studied. The behavior of particles in the vertical tube was pointed by tracking technique (PEPT as Positron Emission Particle Tracking). There exists a significant downward particle flow near the tube wall. Three models were developed in parallel: 1/ A simplified engineering model for the preliminary design of 57 MWth and 290 MWth solar receivers. 2/ A detailed two-phase Euler-Euler model for dense gas- particle systems to investigate the heat transfer in slowly rising bubbling fluidized beds used as HTM for CSP plants. 3/ A detailed heat and mass transfer model to predict the flow and heat transfer behavior of the DPS inside sun rays-impinged tubes.

The Consortium implemented successfully on-sun the pilot solar loop in the CNRS 1MW solar furnace. One hundred hours of on-sun experiments were run totally, with experiments lasting up to 8 hours continuously and solid flow rate in the range 0.660-1.760 T/h. Particle outlet temperature up to 700°C (500°C in average) was obtained, and calculated cavity thermal yield was up to 90%.

Two reference cases were considered for scale-up studies: 57 MWth and 290 MWth. The simplified model was used for integration and annual performance studies of various thermodynamic cycles. It shows the potential for plant efficiency increase of Brayton s-CO2 and combined cycles. The scale up and economic assessment was developed for a 10 MWe CSP plant, operating with Rankine cycle, based on typical meteorological year in Ouarzazate 2014. The plant construction total cost and O&M cost were estimated. The environmental impact assessment showed such a power plant suffers very limited powder-related environmental effects, no explosion risk or fire hazards, and very low equipment erosion because of the low gas velocities. Together with both expected improved thermal efficiency and reduced parasitics, the CSP2 concept can be considered as environmentally friendly.

Project Context and Objectives:
1.2.1 Context
The CSP2 project proposes to use dense gas-particle suspensions -DPS- (approximately 30-40% of solid) in upward flow in tubes as alternative heat transfer fluid (HTF) in order to extend working temperature and to decrease environmental impact of standard HTF used in concentrating solar power (CSP) plants. The assembly of these tubes will constitute the solar absorber (receiver) placed at the top of a central receiver CSP system. The radiation absorbed by these tubes submitted to the concentrated solar radiation is transmitted by conduction, convection and radiation towards the suspension, which warms up by contact with the hot walls. This suspension circulates easily between the inlet and the outlet of the solar receiver and thus ensures the transport of the absorbed energy towards the devices of energy storage and transformation. Moreover, these particles may be used as a storage medium after being heated and stored in a storage vessel, thus enabling to dissociate the production of electrical or thermal energy from the solar energy collection, as it is currently done in central receiver CSP plants working with molten salts, and as cannot be done by other fluids such as gas or steam. But the big differences with molten salts are the totally safe medium involved in the proposed HTF, the higher possible operating temperature and the absence of solidification at low temperature.
This concept is not completely new, it was proposed in France in the 80’s , , on more recently in the USA and in Germany for example. DPS are an innovative alternative to the heat transfer fluids classically used in solar power plants, which suffer from several limitations: limiting range of operating temperatures, toxicity and dangerousness, low transfer capacity (for gas). Such a system, which transposes the methodology developed by Partner 1b (CNRS-LGC), was never proposed before for solar power plants; Partners 1a and 1b patented it jointly in 2010 .
1.2.2 Objectives

The scientific and technical objectives of CSP2 project range from the concept validation, at lab-scale, of the dense gas-particle suspension as new heat transfer fluid, to the construction and operation of a complete solar pilot loop including new solar receiver, and the concept extrapolation to central receiver solar power plant.
They can be detailed as follows:
1/ Hydrodynamic study of the dense gas-particle suspension (DSP) concept, for application to tube-bundle solar receiver.
2/ Design, construction, and operation under real conditions of a solar pilot scale receiver based on the DSP concept as heat transfer fluid with high performances, and included in a complete hot loop.
3/ Development of a local model to describe the heat transfer and the hydrodynamics, and of a global model to describe the complete solar receiver.
4/ Identification of key issues concerning solid handling in such a system.
5/ Development of a system approach for scaling up the concept in association with different thermodynamic cycles.
6/ Design of a central receiver solar power plant based on the DSP concept as advanced heat transfer fluid (HTF), estimation of the resulting electricity cost and its comparison with that from existing systems.

These objectives are summarized in Table 1.

Table 1: Scientific and technical objectives of the CSP2 project
Item Scientific Objectives Technical Objectives
Flow of particles To understand the equilibrium of forces governing the particle flow stability To operate a tube bundle with variable flow rate of solid particle suspension
Heat transfer To measure the radial movement of particle governing the wall-suspension heat transfer.
To model the wall-suspension heat transfer process To obtain (and measure) high values of wall-suspension heat transfer coefficient (about 500 W/m2.K) in order to reduce the temperature difference
Solar receiver To model the suspension movement and dynamic energy exchange in dense circulating solid suspension, To build and to operate a complete solar receiver loop and to measure its thermal performances
Solar plant To derive scaling laws from hydrodynamic and thermal modeling after validation To propose technical choice at large scale (a few tens MW) and a comparative cost assessment

In addition, quantitative objectives of the CSP2 project are listed in Table 2.
Table 2: Quantitative initial objectives of the project
Item Expected achievements
Solid mass flow rate inside the receiver tubes 1-2 T/h
Flow stability Design point ± 20%
Solar receiver power At least 100 kWth
Solar receiver efficiency 70%
Particle suspension operating temperature 500°C-750°C
Scaling power 10-50 MWe

Project Results:

1.3.1 Objectives
The CSP2 project proposes to use dense gas-particle suspensions -DPS- (approximately 30-40% of solid) in tubes as alternative heat transfer fluid (HTF) in order to extend working temperature and to decrease environmental impact of standard HTF used in concentrating solar power (CSP) plants. The radiation absorbed by the tubes submitted to the concentrated solar radiation is transmitted by conduction, convection and radiation towards the suspension, which warms up by contact with the hot walls. This suspension circulates easily between the inlet and the outlet of the solar receiver and thus ensures the transport of the absorbed energy towards the devices of energy storage and transformation. These particles may be used as a storage medium, thus enabling to dissociate the production of electrical or thermal energy from the solar energy collection. The key points of the project were the hydrodynamic study inside the tubes (WP2) after constructing lab-scale setups (WP1), the pilot construction and implementation (WP3), the thermal study of the solar loop, especially of the solar receiver, and the modeling with different approaches (WP4), and the scale-up and environmental impact studies (WP5).
1.3.2 Main scientific and technical results

The two tasks composing WP1 were completed as planned: 1/ The cold and hot mocks-up were designed, constructed and implemented. 2/ The hot solar pilot loop was designed accordingly to the required specifications for DPS circulation in a 16-tube bundle set in a solar receiver that is adapted to the flux map at the focus and to the facility geometry.

The hydrodynamic study (WP2) of DPS in upward flow in tubes was developed implementing both the cold and the hot mocks-up. Stable gas-solid flow conditions have been found, and the control of solid flow rate was achieved. The influence of the various operating parameters (aeration flow rate, fluidization flow rate, solid flow rate, aeration tap positions) was studied and particle-to-wall heat transfer coefficient was studied. The behavior of particles in the vertical tube was studied by tracking technique (PEPT as Positron Emission Particle Tracking), difficult to implement near the tube wall. Activated SiC particles were used as tracer, and were successfully located in 3D. The radial movement of the particles was characterized, the time at the wall and axial and radial velocity profiles were determined.

The pilot solar loop was constructed with some delay due to extended time of on-sun pre-study of the 1-tube receiver rig, and solid handling problem solving. It was delivered in PROMES premises and commissioned in the end of June 2014 (WP3). In the frame of WP4, after several months of ambient operation permitting small problem solving and experimental parameters determination, the pilot rig was successfully operated on-sun for about 100 hours, with experiments lasting up to 8 hours continuously and solid flow rate in the range 0.660-1.760 T/h. Particle outlet temperature up to 700°C (500°C in average) was obtained, and calculated cavity thermal efficiency was up to 90%. The modeling of the solar receiver encompasses the development of models at two levels of fidelity. A detailed model of the heat and mass transfer within the receiver was developed by PROMES and ETH Zurich. PROMES modeled the cavity receiver (to predict the flow and heat transfer behavior of the DPS inside a receiver tube) whereas ETH modeled the solid particle flow inside the tubes (by a large-scale multiphase flow model including heat transfer). The overlap between the models and the coordination between partners permitted the necessary exchange of inputs between the models. IMDEA has been working on the simplified engineering model, using inputs from the detailed model. This global model was used for the preliminary design of a cavity receiver of various powers (from 50 MWth to 290 MWth).
WP5 considered 10 MWe and 50 MWe as the reference cases for scale-up studies and increase in performance expected with advanced thermodynamic cycles. This latter study done by IMDEA showed the potential of Brayton s-CO2 and combined cycles. Concerning scaling up on the basis of standard Rankine cycle, the model was used for integration and annual performance studies. The receiver thermal performance is in the range 65-75%. The scale up and economic assessment was developed for a 10 MWe CSP plant, operating with Rankine cycle, based on typical meteorological year in Ouarzazate 2014 (TOM). The plant construction total cost was estimated as 105.32 M€ (indirect cost 40 %), and O&M cost estimation 8 M€ per year. This leads to LEC (Levelized Cost of Electricity) = 34.6 c€/kWh, that is about 2 fold that of molten salt. Uncertainties in the plant make the LEC an illustrative value only. The environmental impact assessment showed such a power plant suffers very limited powder-related environmental effects, no explosion risk or fire hazards, and very low equipment erosion because of the low gas velocities. Together with both expected improved thermal efficiency and reduced parasitics, the CSP2 concept can be considered as environmentally friendly.

1.3.3 Experimental study of particle flow properties (WP2)

1.3.3.1 Hydrodynamics
The cold lab-scale mock-up built in the LGC premises includes 2 vertical tubes so that conditions can be studied to guarantee even particle suspension distribution in the tubes. Experiments confirmed the process ability to ensure the upward flow of dense particle suspension in a bundle of vertical tubes in parallel.

The gas velocity in the dispenser fluid bed (DiFB) has no influence on the hydrodynamic behavior for velocities larger than the minimum bubbling velocity (Umb). The aeration flow rate increases the suspension void fraction in the tubes, thus decreasing the driving pressure required for by the suspension flow. Helium tracing study showed that the gas flow rate coming from the DiFB and flowing through the tubes increases with the solid flow rate. The gas pressure drop due to wall-to-particle friction increases with the solid flow rate. The steady solid flow in the tubes, evenly distributed among them, requires:
The gas velocity in the DiFB must be more than the particle minimum bubbling velocity.
The aeration velocity in tubes must be more than 5 times the particle minimum bubbling velocity.
Aeration must be the same in all tubes.

1.3.3.2 Radial movement of particles
The hydrodynamic study developed in this Task is concerned with the description of the particles’ behavior inside the vertical tubes to understand the particle motion, the particle mean residence time at the wall and therefore the heat transfer mechanism from the solar heated surfaces to the DPS. The particle motion was studied by SURRREY who used Positron Emission Particle Tracking (PEPT), the only non-intrusive method that is capable of providing detailed and relevant information about the particle circulation within the tubes.

Time at the wall
The key aim of this work is to determine the time spent by particles close to the wall region due to its obvious implications in the wall-to-suspension heat transfer in the tube. For determining this time, we define a region close to the wall, of chosen thickness but typically 2mm in the work to date, where we can locate the particle and follow its radial coordinate with time in the 30 mm i.d. tube. Therefore to constrain the particle to this shell, the current location rxz(i) needs to satisfy rxz(i) ≥ (RR-Rr) being RR (15 mm) the outer radius and Rr (2 mm) the shell radius. The problem in determining time at the wall is that each location has an intrinsic special uncertainty of ~ 0.5mm so that it is not possible to be certain that the particle is within the wall region using one location alone. The approach taken was to determine the distribution of times-at-the-wall on the basis of 1 location, 2 consecutive locations, 3 consecutive locations and so on. This method allows the time in the wall region to be determined with a precision of 10 ms.

Clearly the certainty with which one can determine whether the tracer is within the wall layer and how long it spends there increases with the number of locations chosen as the criterion. As the air velocity is increased the time close to the wall notably decreases. The mean residence times vary with the aeration gas velocity and position in the tube (Figure 1).

Figure 1: Comparison of PDF of the time close to the wall determined with experiments operating at different aeration velocity

Numerical simulations
The experimental technique for accessing to the particle motion in the solar receiver according to Positron Emission Particle Tracking (PEPT) was compared to 3D numerical simulation via an Eulerian n-fluid approach with NEPTUNE_CFD code.

Both numerical predictions and PEPT measurements describe an upward flow at the center of the transport tube with a backmixing (downward) flow near the wall that strongly influences the solar to particles heat transfer mechanism.
Comparisons of the radial profiles of the time averaged vertical and radial velocity (Figures 2 and 3) present a satisfactory accordance and demonstrate the capability of NEPTUNE_CFD software to predict this singular upflow bubbling fluidized bed (UBFB).

Figure 2: Comparing simulation and experimental results: Vertical solid velocity vs. radius.
Figure 3: Comparing simulation and experimental results: Radial solid velocity vs. radius.

1.3.4 Construction and commissioning of a 150 kW pilot solar receiver (WP3)

COMESSA delivered the 150 kWth pilot solar receiver on July 8-9, 2014. The installation and assembly at the focus of the CNRS 1MW solar furnace took place at Week 29 and PROMES and COMESSA performed jointly the first commissioning of the pilot plant (Figures 4 and 5). Ambient temperature and on sun tests, with and without solid pre-heating, were conducted during the period July 20 to 31st, 2014. In a few operating days, positive results were obtained: on-sun experiments in closed circuit lasted between 1 hour and 3 hours, and did not permit to reach thermal equilibrium. The system stands very well small DNI reduction; small clouds have very little impact on tube outlet particle temperature. Particle mass flow rate was about 1 T/h, solid temperature increase in tubes 200-250°C and outlet temperature up to 580°C. Then the facility had to be dismounted and displaced from the solar furnace focal room, in order to be modified. Several improvements to optimize the plant operation and efficiency were identified and solutions were found and applied. Then the pilot was operated at ambient temperature after several set-up modifications to master the system hydrodynamics.

Figure 4: Pilot rig commissioning
Figure 5: 3D view of the pilot solar rig

1.3.5 Experimentation and modeling of a solar receiver with dense suspension of particles (WP4)

1.3.5.1 Experimental determination of the pilot solar receiver performances
CNRS-PROMES ran a series of experiments at ambient conditions with the 16-tube pilot rig operated in a technical hall (that-is-to say not in the focal room). Operation lasted several tens of hours and was developed in the mean time between commissioning and on-sun campaign (September 2014 to February 2015). It permitted to master the steady operation of dense particle suspension (DPS) in upward flow in the 16 parallel vertical tubes, and the starting and stopping phases. The various technical problems that were encountered during the commissioning were solved thanks to these experiments at ambient, several measurement gauges were changed or adapted, several measurement devices were calibrated, the insulation of vessels and tubes was completed, and the data acquisition software was improved. The setup was installed again in the focal room in March 2015, and on-sun series of experiments were run for totally more than 60 hours during the weeks 13-21 (March 22 to May 22) 2015, with on-sun operation of the 150 kW pilot solar loop lasting steadily up to more than 8 hours in several cases.
Experimental procedure
Three different settings of the heliostat field of the 1 MW solar furnace have been selected to vary the inlet power at the cavity inlet. Table 1 shows the different settings and their corresponding normalized solar flux densities and inlet powers at the cavity entrance and Table 2 summarizes the fluidization flow rates used at ambient for the dispenser, collector and cooling fluidized beds, respectively.
Table 1: Solar flux density and power at the cavity entrance for the 3 heliostat field settings
Setting Normalizeda mean solar flux density at the cavity inlet, [kW/m²] Normalized inlet power [kWth]
A 1048 78.6
B 1429 107.2
C 1772 132.9
a “Normalized” means referred to a 1000 W/m2 DNI.
Table 2: Fluidization flow rates at ambient
Fluidized bed Gas flow rate, [Nm3/h] Gas velocity (2umb) [m/s] u / umf
Dispenser (DiFB) 32 0.013 2.7
Collector (CoFB) 18 0.013 2.6
Cooling (CoolFB) 19 0.013 2.6
Flux density measurement at the cavity entrance
The solar field includes 63 heliostats. Three selections of heliostats among those constituting the solar furnace field were chosen, providing an inlet power at the cavity entrance ranging from 63 to 142 kWth, which corresponds to a solar flux density in the range 1-1.8 MW/m² at the receiver entrance. The tubes were irradiated through a 15x50 cm slot. Due to the specificities of this solar furnace, the uneven flux distribution at the cavity entrance was predicted. Each value of the normalized flux density at the cavity entrance showed in Table 3 was calculated as the mean value of the 33 power values given by the calorimeter and divided by the calorimeter surface. The homogeneity variation for a given heliostat field setting is given by the flux variation, and Figure 3 displays the solar flux distributions for the 3 solar field settings. The variation of measured values due to wind or partial shadows (due to cloud cover) was less than 8%.

Table 3: Normalized inlet power and flux density at the cavity entrance for each heliostat field setting
Setting Number of heliostats Normalized mean solar flux density at the cavity entrance, [kW/m²] Maximal flux variation during measurement [%] Normalized inlet power at the cavity entrance, [kWth]
A 19 1048 13 78.6
B 27 1429 2 107.2
C 32 1772 1 132.9

Figure 3: Flux density distribution at the cavity entrance: ① setting A; ②setting B; ③setting C
Operating parameters
The main system operating parameters are the solar power at the cavity entrance, the particle flow rate and the particle temperature in the DiFB. Aeration flow rate was fixed evenly for all tubes at 0.09sm3/m²/s during all experiments. Three selections of heliostats among those constituting the solar furnace field were chosen, providing an inlet power at the cavity entrance ranging from 63 to 142 kWth, which corresponds to a solar flux density in the range 1-1.8 MW/m² at the receiver entrance. The solid mass flow rate varied from 600 to 1800 kg/h, and the corresponding mass flux from 17 to 44 kg/m²/s. The maximal value of the mass flux measured was 47 kg/m²/s. This value was limited by the rotary maximum flow rate and not by the process itself. For this kind of technology the solid flux can be as high as 700 kg/m²/s (in tubes with i.d. 28 mm). The solid temperature in the DiFB varies from 43 to 184°C. The solid temperature was regulated through a water-cooling coil submerged in the CoolFB. Table 4 displays the ranges of operating parameters and the ranges of experimental results.

Table 4: Operating parameter ranges and corresponding experimental results
Parameter ranges Experimental result ranges
Inlet power F p G p T DiFB ΔTDiFB-ColFB Ttubes surface T p,i T p,o T ColFB ΔP/m
kWth kg/h [kg/m²/s] [°C] [°C] [°C] [°C] [°C] [°C] [Pa/m]
63-142 662-1759 17-44 43-184 137-335 366-625 69-251 217-495 188-433 11340-11780
Steady state experiments selection
The criteria to select the steady state experiments were as follows: on the one hand the stability of the solid mass in the DiFB, which reflects of the solid flow rate stability, and on the other hand the temperature stability, evaluated through the particle temperature increase between the dispenser and the collector fluid beds (δΔTColFB-DiFB). δΔTColFB-DiFB is the difference between the maximal temperature increase and the minimal temperature increase during steady state. The highest value of δΔTColFB-DiFB during all steady state conditions was less than 1,6 °C/min, and for 76 % of the experiments it was less than 0.5 °C/min. Twenty steady state experiments, with stable irradiation conditions, were selected for calculating the thermal efficiency (ƞth).
Particle suspension temperature during steady state periods
We achieved previously DPS temperature higher than 750 °C implementing a 1-tube-solar receiver, with particles electrically preheated up to 503 °C in the lack of closed loop circulation. In the present study, 693°C was reached at the outlet of at least one of the tubes, and the mean temperature of the 16 tubes reached 495 °C, with solar heating only. As explained before, the flux distribution in the cavity was uneven because of the solar furnace specificities (the solar flux on the various tubes may vary by a factor of at least 3), and as a consequence the temperature distribution was also uneven. Actually, similarly to the 1-tube solar receiver, the particle outlet temperature cannot overpass 750 °C at the outlet of the hottest tubes since the stainless steel must remain below 900 °C for safety reasons. Consequently, this highest acceptable temperature constraint on the hottest tubes results in low particle temperature at the outlet of the tubes submitted to low solar irradiation. This is why the mean solid temperature at the receiver outlet is limited. At the tube outlets, during steady state periods, the particle temperature difference between the hottest tube and the coldest tube (ΔThottest-coldest tube) was higher than 130 °C and it went up to 390 °C. The mean particle increase ΔThottest-coldest tube for the steady state experiments is given in Table 5. The temperature difference between the hottest tube and the coldest tube does not depend on the particle flow rate and on the inlet power at the cavity entrance. Figure 6 plots an example of particle mean temperatures at all tube outlets during a 55-minute steady experimental run, and the solid temperature distribution at the tube outlets at the hottest moment of this experiment. The mean particle outlet temperature over the 16 tubes was 442 °C.
Table 5: DPS temperature difference between the hottest tube and the coldest tube, [°C]
ΔT hottest-coldest tube Mean ΔT hottest-coldest tube Max ΔT hottest-coldest tube Min
140-350 390 130

Actually, the particle flow rate was measured globally and there was no individual measurement in the tubes. It was assumed that existing sonic nozzles distribute uniformly the aeration air in the 16 tubes, thus the solid flow rate. All particles issued from the irradiated cavity are collected in the ColFB, where their temperature is homogenized. The DPS mean temperature at the tube outlets was calculated assuming that the solid flow rate is uniformly distributed in the 16 tubes, but in reality the total solid mass flow rate is the only available accurate measurement. This is why the temperature increase between the DiFB and the ColFB (ΔTColFB-DiFB) was considered as the reference temperature increase. Figure 7 plots the temperature increase between the dispenser fluid bed and the collector fluid bed for the three considered heliostat field settings. For the mass flux calculation, the flow rate was supposed to be uniformly distributed between the tubes. The mean temperature increase between the DiFB and the ColFB (ΔTColFB-DiFB) ranged between 137 and 335 °C, with a mean value of 218. The tube length between these fluidized beds is 2.27 m, but only 1m of the total length is irradiated. The maximal value of the mean temperature increase, 335 °C, corresponds to an experimental run with 21 kg/m²/s as solid flow rate and 105 kW/m² as solar flux density at the cavity entrance (φcavity entrance), whereas the minimum value, 137, corresponds to an experiment with solid flow rate 37 kg/m²/s, and inlet power 63 kWth. Obviously, for the three considered inlet solar powers, it can be seen that the temperature increase ΔTColFB-DiFB decreases with the solid mass flux. Moreover, for a given solid mass flux, the higher the solar power at the cavity inlet, the higher the temperature increase.

Figure 6: Solid temperature at the hottest point & mean solid temperature distribution at the tube outlets during 55-minute steady experiment (aeration = 0.09 sm3/m²/s, mass flux = 35 kg/m²/s solar power at the entrance = 142 kWth)
Figure 7: Temperature increase between the DiFB and the ColFB versus solid mass flux, for the 3 solar power ranges

Thermal efficiency
In concentrating power solar plants (CSP), the heliostat field represents one of the main capital costs. By reducing the size of the solar field for a given capacity, the capital cost will be reduced. The thermal efficiency (ƞth) plays an important role in the improvement of the plant performance. Determination of experimental thermal efficiency was one of the main objectives of this series of experiments. The determination of the thermal efficiency is detailed bellow, and includes the calculation of the various parameters affecting ƞth.
Experiment selection
The criteria to select the experiments for thermal efficiency calculation were the same as for the steady state selection, and with stable DNI conditions. Figure 8 plots an example of steady state, with stable temperatures and flow rates, and appropriate irradiation conditions for thermal efficiency calculation.

Figure 8: Mean solid outlet temperature, temperature increase ColFB to DiFB, & DNI versus time
Thermal efficiency calculation
Experiments were carried out under various ranges of the operating parameters, i.e. the solid flow rate and the solar flux density at the cavity entrance (φcavity entrance). The thermal efficiency is defined as:
(1)
Where ϕDPS represents the power transmitted to the DPS (power extracted by the particles) and ϕin represents the solar power at the cavity entrance.
Three different settings of the heliostat field have been chosen to have different inlet powers at the cavity inlet. The receiver tubes are irradiated through a 0.15 m x 0.50 m slot (cavity entrance) set at the focus plane. The measurement data required for ƞth calculation that are temperature, pressure (necessary for the particle mass flow rate) and direct normal irradiation (DNI), were measured continuously during the experiments, and recorded every second. The inlet power for any experiment is calculated as:
ϕ_in=φ_(cavity entrance)∙S (2)
with ϕin solar power at the cavity inlet, φ the solar flux density at the cavity entrance, S the slot area and DNI the direct normal irradiation.
The flux density at the cavity entrance represents the flux density for a given DNI value. To obtain φ value, the normalized flux density (φN cavity entrance) is multiplied by the DNI value and divided by 1kW/m². The extracted power is defined as:
ϕ_(DPS 16 Tubes)=F_p ∫_i^(T_o )▒〖C_(p,p) dT≈F_p∙C_(p,p)∙〗(T_(p,ColFB)-T_(p,DiFB)) (3)
with Fp the particle mass flow rate, Cp,p the particle heat capacity, Tp,ColFB the particle temperature in the collector fluidized bed, and Tp,DiFB the temperature in the dispenser fluidized bed.
The air heat capacity is neglected in front of that of the solid. The latter is calculated with a polynomial expression established from the values given by the NIST database :
c_(p,p) (T_p )=a〖T_p〗^3+b〖T_p〗^2+cT_p+d (4)
with Tp in K and a = 2.25 10-7 J/kg.K4 , b = -9.88 10-4 J/kg.K3, c = 1.62 J/kg.K2, d = 320 J/kg K.
The temperatures used in the enthalpy balance were chosen as detailed hereafter. As pointed out before, there is a solid recirculation phenomenon in the vertical tubes: most particles flow upward in the tube center region, whereas particles flow downward near the tube wall. Therefore, in order to take into account the transferred heat to the particles going down into the DiFB, the temperature inlet considered for the enthalpy balance is the mean temperature of the four thermocouples placed in the DiFB. This is indeed more logical than considering the mean temperature of the six measured tube inlets (tubes 1, 4, 8, 9, 13 and 16). Regarding to the outlet particle temperature, all 16 tubes are equipped with thermocouples but there is only one global measure of the gas flow rate, since there are no individual mass flow meters. For better accuracy reasons, the solid outlet temperature is taken as the mean of the two thermocouples placed in the ColFB, because of its homogenized temperature. This induces a thermal efficiency underestimation, since the particle temperature at the tube outlet is between 14 °C and 89 °C higher than in the ColFB, with average 44 °C.
The thermal efficiency was calculated in two different ways. First, by applying Equations (1) to (4) to time dependent values of the data concerned (DNI, Tp,ColFB, Tp,DiFB, Fp,c p,p), in a way that we obtain instantaneous values of the thermal efficiency, and finally the value given is the main value of the instantaneous values obtained during the steady state. Second, by calculating ƞth from mean values of the time-dependent values during steady state. The difference was less than 0.05 %, with a mean value of 0.01 %. This fact shows the steady state definition consistency.
In the plots given hereafter in this report, information on the measurement uncertainty is given through error bars for each experimental point. The error was estimated as follows:
From Eq. (1), the relative error on the thermal efficiency ηth is
(5)
And from Eq. (3), the relative error on the heat flux ϕ_( DPS 16 Tubes) is
(6)
The uncertainty on Fp was estimated from an experimental correlation Fp versus ∆PDiFB-Tank. It is taken as the maximal difference between the measure and the calculated value by the correlation; then the relative error on Fp was always less than 5 %. The relative error on Cp,p (issued from the NIST database) is 5 %. The uncertainty on sheathed K thermocouple indications (Tp) is ± 2.5 °C up to 333 °C and ± 0.75 % over 333 °C. Finally, based on the accuracy of the 25 mm calorimeter used for flux density measurements, the relative error on ϕin is estimated as 2%.
Figure 9 plots the thermal efficiency ƞth as a function of the solid flux for the three solar power ranges at the cavity inlet. ƞth ranges from 50 to 90 %, in all cases it increases with the solid flux.

Figure 9: Cavity thermal efficiency versus solid mass flux for the 3 solar power ranges at the cavity entrance
Figure 10 plots the thermal efficiency versus the solid temperature increase ΔTColFB-DiFB, classified by solid flux and inlet powers. ΔTColFB-DiFB has a positive effect on the thermal efficiency for all series but one (that with solid flux 20-22 kg/m²/s and inlet power 78-80 kWth).

Figure 10: Cavity thermal efficiency versus temperature increase ΔTColFB-DiFB
Transient state analysis
Particle temperature
During the transient periods, the solid temperature reached 755 °C at the outlet of one of the hottest tubes and the mean temperature reached 590 °C at the tubes’ outlet (Figure 11) with solid mass flux 30 kg/s/m², and an average solar power at the cavity entrance 83 kWth. Figure 11 shows the solid temperature distribution at the tube outlet and the mean temperature of the 16 tubes. The solar flux unevenness limited the mean temperature reached by the particle flow, similarly to the steady state periods. This is why the solid temperature overpasses 700 °C only at the outlet of tubes 5 and 12, and it overpasses 600 °C in only 5 out of 16 tubes. The difference in DPS temperature at the tube outlets was as high as 246 °C.

Figure 11: Instantaneous solid temperatures at the tube outlets and their average during a transient experiment (mass flux rate = 30 kg/m²/s, aeration = 0.09 sm3/m²/s)

System response to DNI changes
The system response to DNI (Direct Normal Irradiation) variations was studied when irradiation conditions changed, either quickly (cloud) or slowly (sunset approach). The system stands very well irradiation fluctuations; small clouds have very little impact on the mean particle temperature at the receiver (tubes) outlet. Figure 12 plots the transient DNI and solid and wall temperatures for an experiment with variable irradiation conditions. The purple line represents the average of all values recorded by the thermocouples soldered on the tubes. Between 12:48 and 12:53, the DNI drops under 150 W/m², and the tubes’ surface temperature falls from 397 °C to 232°C (ΔT= 165 °C), while the DPS temperature at the tubes outlet drops down from 257 °C to 200 °C (ΔT= 57 °C), and the solid temperature in the ColFB varies from 245°C to 226 °C (ΔT= 19 °C). There are two explanations to these phenomena: one the one hand, the receiver has a good thermal inertia, and on the other hand, the system is self-regulated in this domain: at 12:48, when the DNI drops down, the solid mass flow rate decreases also without any operator’s action from 1.25 T/h to 0.96 T/h. When the sky clears again, at about 13:00, the solid flow rate rises by itself from 0.96 T/h to 1.2 T/h.

Figure 12: Transient DNI and temperatures (solid mass-flow rate = 1.25; 0.96; 1.2 T/h; mass flux = 31; 24; 30 kg/m²s; mean normalized solar power = 77 kWth)
System response to changes on the mass flow rate
Table 6 shows the three steady state periods reached when the operator changed the mass flow rate on purpose, for the mean temperature of tube wall surfaces, the mean solid temperature at the 16 tube outlets and the temperature increase between DiFB and ColFB, and the recorded DNI. During the steady state period ‘I’, a 36 kg/m².s solid mass flux leads to ΔTDiFB-ColFB = 216 °C. When the operator decreases the mass flux from 36 to 31 kg/m².s (14% reduction) by changing the rotary valve frequency, ΔTDiFB-ColFB increases from 216 to 228 °C that is 6 % increase. When the mass flux is reduced to 21 kg/m².s, the temperature increase grows to 269 °C, which means 23 % increase of ΔTDiFB-ColFB, for 48 % solid flux decrease, in comparison to the initial steady state ‘I’.
Table 6: Operating parameters and results of the experiment shown in Figure 16
Steady state Time of starting/ending G p ϕin ΔT DiFB-ColFB T p,o
[hh:mm] [kg/m².s] [kWth] [°C] [°C]
I 15:00/16:00 36 108 216 359
II 16:10/17:00 31 104 228 369
III 17:10/18:15 21 98 269 403

1.3.5.2 Modeling and heat transfer studies

Heat transfer studies
The extracted heat flux for both series of experiments (with the single-tube receiver and the multi-tube receiver) was calculated to control their coherency. In the single-tube DPS solar rig studies carried out first, the heat flux transmitted to the particles ( QSingle-Tube ) is calculated as:

Q_(Single-Tube)=〖 Φ〗_DPS/S_INT =(F_p 〖∙c〗_(p,p)∙(T_(p,o,center)-T_(p,iDiFB) ))/S_INT (7)

with Fp the particle mass flow rate, ϕ DPS the heat transferred to the particle/s, cp,p the particle specific heat capacity calculated at the mean particle temperature, Tp,o,center the particle temperature at the tube center at the cavity outlet, Tp,i DiFB the particle temperature at the tube inlet inside the DiFB and S INT the tube internal surface area. In the 16-tube receiver case, because of the uneven solar flux distribution in the cavity, one or two tubes only receive enough flux to work under the same thermal conditions as the single-tube receiver. Therefore, the hottest tube of each experiment was chosen for the enthalpy balance with the multi-tube receiver, to compare the heat extraction capacity of both solar rigs. The heat flux was calculated like for the one-tube rig, taking the DiFB temperature as the inlet temperature, and the DPS temperature at the tube outlet as the outlet temperature:

Q_(Hottest tube)=〖 Φ〗_DPS/S_INT =(F_p 〖∙c〗_(p,p)∙(T_(p,o,Hottest tube)-T_(p,iDiFB) ))/S_INT (8)

The inlet power has no influence on the heat flux when the hottest tube temperature is used for the enthalpy balance (Figure 13). But as mentioned before, the inlet power has a positive effect on ΔT ColFB-DiFB when the outlet temperature (ColFB) is that of all the particles. Finally, as shown in Figure 14, both data series are strongly coherent. For the single-tube receiver, the maximal extracted flux for a 45kg/m² solid mass flux was 160 kW/m². For the multi-tube solar receiver, the maximal extracted flux was 143 kW/m², with solid mass flux 44 kg/m².

Figure 13: Heat flux extracted versus particle mass flux for the single-tube receiver and the hottest tube of the multi-tube receiver
Figure 14: Extracted heat flux versus solid mass flux for the single and multi-tube setups

Modeling
1/ ETHZ developed a detailed two-phase Euler-Euler model for dense gas- particle systems, to investigate the heat transfer in slowly rising bubbling fluidized beds used as HTM for CSP plants. The model was built on the open-source code OpenFOAM and includes conduction, convection, and radiation heat transfer. The detailed two-phase model calculates, based on the determined volume- averaged radiation properties of the SiC particle suspensions, the effective radiation properties at each time step depending on the solid volume fraction in each computational cell. Therefore, the model captures the penetration of radiation into the suspension through bubbles at the wall and the absorption or scattering of radiation close to the hot wall due to the high extinction coefficient of the dense suspension. By including radiation heat transfer, the developed model can be applied to heat transfer in gas-particle systems with a wider range of wall and suspension temperatures than existing models. Furthermore, the model can provide quantitative data on the relevant hydrodynamic and heat-transfer mechanisms at a level of detail that is beyond existing experimental techniques.
2/ PROMES-CNRS modeled the receiver cavity under the name “Surface-to-Surface” -S2S- with ANSYS FLUENT version 15.0 using the radiosity method, and using the solar flux density maps of the previously developed MCRT model as boundary conditions. The surface view factors were computed by the software. For the flow, the laminar model was used. The study was conducted in steady state. The model considered that air was transparent to radiation and that all surfaces were diffuse grey bodies. The insulating material and Pyromark® paint emissivities in the infrared region differed from their emissivities in the solar spectrum. The insulating material emissivity was taken as an average of those of silica and calcium carbonate that are its main components and it was set to 0.2. The Pyromark® 's emissivity was 0.88. The used geometry was the same as that used in Solfast-4D model. The mesh was created using ANSYS Design Modeler. It comprised 2 334 000 tetrahedral cells with 1 cm edges. The tubes inside was not modeled and, therefore, not meshed. The tubes and insulating material thickness were not simulated and appropriate boundary conditions were set instead, as explained in the next sub-subsection. Polynomial functions of the temperature were computed to determine the air properties. The difficult part of this modeling was to set boundary conditions able to reproduce the physical phenomena occurring inside the receiver cavity. The cavity aperture was set as free surface (pressure inlet in FUENT) at atmospheric pressure (101,325 Pa) and 10 °C. This boundary was not only an inlet but also an outlet since the air could enter and exit freely, following the convection created by its heating inside the cavity. The fact that an inlet condition was imposed did not prevent the air exiting. The modeled system was limited to the cavity volume, thus no wind velocity could be imposed. But given the small size of the opening (15 cm x 50 cm), we believe that wind effect was very limited. This entrance was set as a blackbody at 10 °C for the radiation exchange. A no-slip condition was imposed to all the walls inside the cavity (the term “wall” applies to the insulating material as well as to the tubes).

The solar flux density impacting the walls that was previously determined with Solfast-4D was interpolated for each face of the mesh thanks to User Defined Functions (UDFs). Each surface had its specific UDF. The solar flux density could not be computed as a wall heat flux because it would have prevented adding the other heat exchanges through the walls that are the conduction heat loss through the insulating material and the heat transfer to the DPS circulating inside the tubes. Thus, the solar flux density was transformed into an equivalent heat generation rate inside the walls. A virtual wall thickness of 1 m was imposed and therefore the value of the heat generation rate in W/m3 was equal to the solar flux density in W/m2. Note that any virtual wall thickness could have been chosen as long as the product thickness x heat generation rate remained equal to the solar flux density. The absurdity of having a tube thickness larger than the diameter was not a problem since this thickness was virtual. The heat loss by conduction through the insulating material was modeled by an equivalent convection heat transfer with the environment at 10 °C (thermal resistance equal to the sum of the conduction resistance and convection resistance). The heat transfer with the DPS circulating inside the tubes was also modeled by a convection heat transfer, but since the DPS temperature increased along the tubes' height, a temperature profile was imposed thanks to a UDF instead of a constant temperature.

Model validation
1/ ETHZ performed an extensive verification and validation procedure in which the hydrodynamics and heat transfer of the detailed two-phase Euler-Euler model were examined. A single-phase flow verification was performed by using the analytical solution of the fully developed velocity profile in a two-dimensional pipe flow. The analytical and numerical results were in excellent agreement. An extensive grid-refinement and time-average study was done for a bubbling fluidized bed and the converged results were compared to experimental and numerical results from literature. The transient convective heat transfer was verified by using the analytical solution of forced convection in a one-dimensional packed bed. The radiation heat-transfer model was verified by comparing the incoming irradiation in a square enclosure predicted by our model with the analytical solution of the P1-approximation. For both heat-transfer verification studies, the analytical and numerical results were in excellent agreement. Steady state conduction combined with radiation was validated with experimental results from literature using an annular packed bed formed by concentric tubes and cordierite spheres. The model showed good agreement with the experimental results over a large range of temperatures.
After different verification and validation studies, the model was then compared to on-sun experimental results of the single-tube plant of PROMES using a dense gas-particle suspension as HTM. The comparison included the wall-to-bed heat-transfer coefficient (Figure 15), the pressure drop, and the solid fraction in the riser for different operating conditions.

Figure 15: Comparison of the experimental and numerical values of the wall- to-bed heat-transfer coefficient as a function of the aeration riser velocity. The numerical results are for different air-leakage velocities.
2/ CNRS-PROMES worked on validating the developed model with the experimental results obtained with the pilot rig operating on-sun.

Model validation method
Two comparison elements between the model and the experiments were used: one for the insulating materials, to validate the convection heat transfer coefficient applied as boundary condition, and the other one for the tubes, to validate the convection heat transfer coefficient as well as the temperature profile. During the experiments, thermocouples were inserted in the insulating material in the east and west cavity walls. On each side, one was close to the cavity interior wall; the other was located 7.5 cm from the cavity. These thermocouples allowed estimating the conduction heat loss density passing through those walls. The convection heat transfer coefficient set at the boundary was validated when the same heat loss density was observed in the model. For the tubes, the comparison between the model and the experiments could be done at two levels: the heat flux transmitted to the DPS and the temperatures inside the cavity. But actually, the value of the heat flux calculated from the experimental measurements was unsure. The power absorbed by the particles was calculated by an enthalpy balance. But a question remained concerning the method to calculate the outlet power (power transferred to particles): which temperature should be chosen as reference temperature for the enthalpy balance? The ColFB temperature or the average of the DPS temperatures measured in the tubes at the cavity outlet? The problem of using the ColFB temperature is that it included the heat losses that occurred between the cavity outlet and the Collector fluidized bed, as well as the ColFB losses, thus underestimating the heat flux received by the DPS. In the other case, the problem is that the solid flow repartition between the 16 tubes was not known. Therefore using the average of the DPS temperatures measured in the tubes would give too much weight to tubes with low solid flow rate and too little weight to those with high solid flow rate if significant differences existed. It was unknown if the second calculation option led to an overestimation or an underestimation of the power absorbed by the DPS.

Due to this uncertainty, the comparison between the model and the experiments was done with the temperatures measured inside the cavity, and more specifically, those measured at the back of the tubes. Indeed, these measurements were not affected by direct solar irradiation, contrarily to those at the tubes' front.

Case study
Characteristics and boundary conditions
A specific experimental run was selected among the experimental runs performed that led to steady conditions, and it was modeled. Modeling was performed with a normalized average solar flux density of 1 MW/m2 at the cavity inlet, corresponding to an actual 83.9 kW solar power entering the cavity. The corresponding maps of solar flux density on the cavity walls were computed with the solar furnace model. The total solid flow rate was 974 kg/h. The heat flux transferred to the DPS estimated with the ColFB temperature was 56.7 kW and that estimated with the average tube outlet temperature was 72.7 kW. The temperatures in the Dispenser Fluidized Bed (DiFB) and Collector Fluidized Bed (ColFB) were 132 °C and 359 °C, respectively. The average DPS temperature in the tubes at the cavity inlet and outlet were 197 °C and 421 °C, respectively. The average tube back temperatures at the inlet and outlet were 488 °C and 512 °C, respectively. The heat loss density through the insulating material at the thermocouples' locations on the east and west sides were 555 W/m2 and 589 W/m2, respectively.
Model validation
The model was validated when the simulated temperatures satisfactorily matched the temperatures measured at the tubes rear, in the middle, at the bottom and at the top of the cavity. The results can be seen in Figures 16, 17 and 18 for the middle, bottom and top, respectively. The largest errors produced by the model at this tuning stage were an overestimation of 63 °C in the middle, and underestimations of 79 °C at the bottom and 127°C at the top. The average simulated temperature was underestimated by 15 °C in the middle and overestimated by 33 °C at the bottom and 31 °C at the top. The model did not match the results perfectly, a finer tuning would be required to improve the fit, but it was close enough to draw conclusions from these results.

Figure 16: Comparison of measured and simulated temperatures measured at the tubes’ rear in the middle of the cavity

Figure 17: Comparison of measured and simulated temperatures measured at the tubes’ rear at the top of the cavity
Figure 18: Comparison of measured and simulated temperatures measured at the tubes’ rear at the bottom of the cavity

Results
The results presented here focus specifically on the distribution of the incoming solar power between reflection losses, convection and radiation losses through the cavity aperture, heat loss through the insulating material and heat transferred to the DPS inside the tube. Solfast-4D model showed that 1.3 kW among the 83.9 kW solar heat flux entering the cavity exited the system after multiple reflections without being absorbed. The total heat loss through the cavity opening was 13.9 kW, with 10.9 kW due to convection and 3.0 kW due to radiation. The heat loss through the insulating material was 2.3 kW. The tubes absorbed the remaining 66.6 kW, which were transferred to the DPS. These results show that the receiver cavity was well designed since only 1.6 % of the incoming solar energy was lost without being absorbed. The receiver is also well insulated (2.8 % heat loss through the insulating material). The cavity shape limits the radiation heat loss through the opening to 3.5 %. The main heat loss comes from the air circulation generated by its heating when it comes into contact with the cavity walls. This convection heat loss through the cavity opening is as high as 13 %. Finally, 79 % of the incoming solar energy is transferred to the DPS. The solar power entering the cavity is distributed as shown in Figure 19. The heat flux transferred to the DPS estimated by enthalpy balance with the Collector FB temperature was underestimated by 15 % (56.7 kW in front of 66.6 kW given by the model). It was 72.7 kW when estimated similarly with the mean DPS temperature at the tubes’ outlets (9 % overestimation, thus closer estimation).

Figure 19: Distribution of the solar power entering the cavity.

1.3.6 Scale-up, economic & environmental impact assessment (WP5)

1.3.6.1 Plant design with DSP technology
IMDEA Energy group focused on the development of a methodology to be used for the design and modeling of a concentrated tower solar power plant using DPS (dense particle suspension) as the heat transfer fluid. The design methodology was focused on obtaining a first approach of the performance of all the main components of the solar plant in three different ways: design point, annual performance and post-processing. It consisted in a main MATLAB routine managing these three features in a forward direction according to the plant specifications given, including the heliostat field. Once MATLAB calculated successfully the overall design of the plant and its components at nominal conditions, the annual performance of the plant was run. The main routine prepared the required data from the design process to be sent to TRNSYS plant model. When the annual simulation was finished, MATLAB was used to collect the simulation outputs generated in text files to start post-processing tasks. Design methodology was described in detail in deliverable D 5.1.
ETHZ used the verified and validated model for a parameter study to investigate the influence of the riser-wall temperature and the riser diameter on the wall-to-bed heat-transfer coefficient, the solid temperature, the solid mass flow rate, and the solid fraction in the heated riser section. The parameter study showed that for a constant solid mass flow rate an increase of the riser wall temperature leads to a decrease of the heat-transfer coefficient. This results from a strong decrease of the logarithmic mean temperature difference between the inlet and outlet of the heated riser section. Furthermore, the increased suspension temperature causes a gas-phase expansion and a resulting decrease of the solid fraction in the heated riser section. While keeping the gas velocity at the riser inlet and the riser- wall temperature constant, an increase of the riser diameter from 36 mm to 72 mm reduces the solid temperature difference over the heated riser section by about 30 %. This results from an increase of the total heat capacity rate by about 62 %, while the total power transmitted to the suspension increases only by about 27%. In addition, increasing the riser diameter above the maximum bubble size prevents slug flow, leads to a more uniform bubble size distribution, and reduces pressure oscillations, which is favorable for the power plant operation. Although the wall-to-bed heat-transfer coefficient gives important information, it is not necessarily an indicator for good performance of a CSP plant. Moreover, due to its dependence on several parameters, the heat-transfer coefficient should always be considered in relation to other properties like the solid temperature or the solid mass flow rate.
For Geldart-Baeyens Group A particles, the dynamic equilibrium between bubble coalescence and splitting leads to an increase of the bubble size towards a maximum size as they move in the axial direction. In the present case, the maximum bubble size is similar to the smallest considered riser diameter of driser = 36 mm. When the radial extension of a bubble approaches the riser diameter, it will lead to slug flow where the complete riser cross-section is covered by the gas phase. With a riser diameter of driser > 36 mm, a bubble with a diameter of about 36 mm rises freely and particles can pass the bubble on both sides. The instantaneous solid fraction field was compared for the different riser diameters in a riser section of 1.0 m height. For the smallest riser diameter, some bubbles cover almost the entire cross section that leads to slug flow. Increasing the riser diameter avoids, especially for driser = 72 mm, these slugs. Reducing slug flow by increasing the riser diameter decreases the bed expansion and increases therefore the average solid fraction in the riser. In addition, the reduced slug-flow tendency leads to a more uniform bubble size distribution and reduces pressure oscillations, which is a favorable effect for steady operation of such a CSP plant.
1.3.6.2 Flowsheeting and system analysis
The aim of this task is to analyze the DSP the flowsheeting and system analysis of a central receiver solar power plant based on DSP technologies. The work is based on the cases studied before and the analysis has been carried out collecting all the inputs developed during the CSP2 Project and collected in the document “Critical design issues for a dense particles suspension solar power plant scaling up” by PROMES-CNRS and EPPT. This document collects all the parameters obtained during the trials in the receiver prototype in Odeillo and the prototype scaling up made by PROMES-CNRS. These obtained parameters are experimental and the scale up of the receiver has been made thanks to the experimental experiences of PROMES-CNRS.
With all these inputs TORRESOL has analyzed a Global System, including the systems and subsystems theoretically necessary to develop an industrial plant with DSP technology. The starting input for this design and analysis has been the receiver and the scale up in charge of PROMES-CNRS. In the analysis of an industrial plant TORRESOL has assumed as valid the parameters given in the “Critical design issues for a dense particles suspension solar power plant scaling up” report (1), and the scale up of the receiver realized in the same document.
As the same time IMDEA summarized the main features of proposed power block configurations for a scale-up plant design methodology (proposed in Deliverable D5.1). The design methodology was applied for a solar power plant based in Ouarzazate (Morocco) using a particle suspension receiver design model described in deliverable D4.3 considering both the cavity and external configuration, for a given thermal power of 57 MWth and 290 MWth that correspond to typically a 10 MWel and a 50 MWel power plant. The receiver was coupled to a thermal energy storage system design with a solar multiple of 2 and 6 hours of storage capacity, according to specifications in D5.1. In order to extract the thermal energy from the DPS (Dense Particle Suspension), a solid particle heat exchanger was designed with 95% efficiency as reported in D5.3. For the power block, several configurations were investigated from the standard but optimized Rankine and Brayton cycles till more advanced concepts such as Combined Cycles or Brayton cycles using new working fluids such as Helium or supercritical Carbon Dioxide. All of these cycles were seen as having the potential for significant efficiency and/or cost improvements over contemporary CSP plants, and preliminary analysis of these cycles was included as part of deliverable D5.1 in order to establish the design process. As many of these cycles were non-standard, new components were developed within the TRNSYS simulation tool; models of the air and helium Brayton cycles were encoded as TRNSYS components.
Power cycles were optimized for design conditions using MATLAB tool, obtaining the plant layout, mass flows, temperatures and pressures distributions in all the components. High efficiency strategies such as multistage intercooled compression, multistage reheated expansion and regeneration were considered for plant layout. This data was used as input for the transient modeling performance developed using TRNSYS.
Depending on the power block defined in the previous stage, different plant models were built in TRNSYS Simulation Studio by combining, editing and creating black-boxes functions, called Types. In order to operate the solar power plant during transient conditions, a control strategy was introduced for the heliostats field, the receiver, the storage tanks and turbine. Proposed control assumed a defocusing strategy when the receiver reaches its limit power at absorber surface. On the receiver side, DPS mass flow was modified in order to keep receiver wall temperature below the safety limit, being required a 5% of input thermal power to start up the particle circulation. According to the solar multiple chosen, half of the available thermal energy was sent to the storage system while the other half was diverted towards the power block. During power block start-up and in the early morning hours the storage tanks supposed to be at minimum levels. So, in that working period power plant operated as if there were not a storage system to raise the tank levels. The main negative consequence of that approach was that fluctuations at this morning hours forced the turbines to work in transient mode. After 11am, storage tank took an active role in the operation tactic because fluctuations of incident solar power were covered by the energy storage to work the maximum number of hours at design conditions (full-load). Once the sun was set, power plan continued operating until the hot tank of heat transfer fluid was at a 5% of its capacity. Turbine performance was controlled monitoring the level tanks, input thermal power at the receiver and at the turbines inlet or steam generator performance. Cold start-up turbines began if a minimum thermal power required of 25% was reached. This period lasted 30 minutes and 50% of thermal power was needed for warm up, in other words it was not be transformed to electricity. Warm stand-by situations were contemplated in the model but with no extra energy required to maintain this operating condition.
The design methodology described above, the TRNSYS plant diagram and the aforementioned control strategy were used to analyze several power blocks coupled to medium (57 MWth) and large thermal power (290 MWth) configurations. Overall cycle efficiencies ranging from 31% to 48% at design conditions were found, what translated into sun-to-electric efficiencies meant from 17% to 26.4%. The detailed lists for both efficiencies definitions are listed as standard Brayton (31%-17%), Brayton Helium (34%-19%), Standard Rankine (41%-23%), Combined Cycle (42%-24%), Brayton High Temperature (48%-22%) and Supercritical Carbon Dioxide Brayton cycle (48%-26.4%). The nominal design and transient results for the abovementioned power cycles are detailed in Deliverable D5.2 and Table 7 summarizes them.
Table 7: Main parameters for 57MWth (about 10 MWel) power plant modeling, performances at nominal conditions
Nominal Rate Operation Units Std. Rankine Std. Brayton Brayton
DPS 750C Brayton
DPS 1,000C Brayton
He. Combined
Cycle Brayton sCO2
Heliostats efficiency [%] 72.1 72.1 72.1 67.8 72.1 72.1 72.1
Receiver efficiency [%] 82.3 80.7 77.3 72.2 81.0 83.1 79.7
Thermal power to storage [MW] 23.4 23.0 - 20.6 23.1 23.7 22.75
Thermal power to power block [MW] 23.4 23.0 - 20.6 23.1 23.7 22.75
HTX efficiency [%] 95.0 95.0 95.0 95.0 95.0 95.0 95.0
Net Electrical power [MW] 9.07 6.84 7.66 9.38 7.45 40.47 10.43
Net Power cycle efficiency [%] 40.8 31.26 37.1 47.9 33.9 42.6 48.25
Total efficiency [%] 22.95 17.28 19.38 22.30 18.81 24.22 26.40

1.3.6.3 Economic Assessment
TORRESOL was in charge of the economic assessment, which was addressed based on the inputs from previous tasks. In the Deliverable D5.2 is explained that the study and analysis realized by TORRESOL includes mechanical assumptions in the behaviour and handle of the heat transfer fluid due to the state-of-the-art is not matured enough especially in dealing with this kind of HTF in this kind of applications including the requirements and performance needed to be useful in a thermo solar plant (such us heat losses in the equipment, pumping performance for the HTF, corrosion and abrasion at high temperature,...). For that TORRESOL has made an analysis from theoretical point of view without going in depth into the design, capabilities and resistances of equipment where the HTF is going to flow. The cost analysis was made for a 57 MWth receiver plant with the DSP technology, similarly to previous analyses. It was based on a NREL’s reference document and a SANDIA Report .
The general flow chart of the process in a DSP Technology concept Plant is given in Figure 20.

Figure 20: Schematic view of a CSP plant. 1-DPS receiver; 2-Hot storage tank; 3-Cyclonic separator; 4-Redler conveyor; 5-Heat exchanger; 6-Cyclonic separator; 7-Redler conveyor; 8-Cold storage tank; 9-Redler conveyor; 10-Buffer tank; 11-Air blower; 12-Heat recuperator

Cost estimation
Receiver
The cost of the receiver will be proportional to the size and working temperatures, which will imply material selection. A 57 MWth receiver can be estimated as 175 €/kWth, including the receiver panels and the associated auxiliary equipment and receiver erection.
Solar field
For a 57 MWth Receiver, the total mirror surface can be estimated in about 140.000 to 150.000 m2, depending mainly on the final receiver configuration and performance or mirror reflectivity. The solar field cost, including the mirrors, structures, actuators, controls... will depend on the heliostat configuration and number. Cost per square meter of mirror can be estimated as 150 €/m2.
Thermal storage system (TES)
For estimating the thermal storage cost, the following data were considered: Hot tank temperature storage: 650 ºC. Cold tank temperatures storage: 350 ºC. Thermal Energy Storage: 6 hours. Considered sand thermal data were: specific heat: 1 kJ/kg K at 350 ºC and 1.14 kJ/kg K at 650 ºC; density: 3.210 kg/m3. As the particle will not be completely packed, there will be some space between the grains, so the effective storage density will be considered 20% lower than the solid density: 2,568 kg/m3. Therefore the 10 MWe plant will require 1,900 tons of sand, assuming that it will be about 5% of not usable sand (tank corners, circulating salt,...) that should be stored, and 10% volume margin, the size of the thermal storage tanks will require at least 850 m3 for a 10 MWe plant. The cost of the thermal storage will depend not only on the size but also on the construction characteristics, like type of soil, tank shape or materials, and can be around 23 €/kWh of stored thermal energy for the 10 MWe plant. In this cost estimation, the cost of the sand is not included.
Power block and BOP
For estimating the power plant cost, it is assumed that the SGS for particle receiver plant will have similar size and characteristics than a molten salt SGS system of same power, as working conditions of the water/steam side will be the same. Molten salt specific systems cost can be compensated by higher particles systems and temperatures, that can require special materials. The estimated cost for the power plant and BOP can be 1,750 €/kWe for a 10 MWe plant.
Other costs
In this section all the necessary works to build a plant are collected, such as general civil works, engineering works, construction, start-up,... The property costs are not included in this analysis. These costs are all the issues necessary to realize the plant investment, such as land, renting, leasing or purchase; all kind of taxes; necessary permits and licences (for project, construction and operation phase); financial costs; insurances...
O&M costs
The O&M costs collected in this section are an approach to the DSP Technology. As commented before, the state-of-the-art of this technology in global terms (a complete industrial plant) is incipient and there is no reference in terms of feasibility and O&M capabilities about the behavior of systems and subsystems that compose a plant with this technology. The results in the O&M cost are completely estimated and are subjected to the good performance and good operation of this technology. The estimated cost for the O&M can be 1,350 €/kWe per annum for a 10 MWe plant.
Global estimation
The global estimation can be summarized as shown in Table 8.

Table 8: Cost estimation of a DPS power plant

Levelized cost of energy
The Levelized Cost of Energy (LEC) is simulated using a free tool that can be found in the NREL (National Renewable Energy Laboratory USA) website .
The NREL’s tool is a calculator that allows introducing several parameters of an energy plant and the the LEC is calculated for these parameters. The parameters are:
Financial issues
Period to be analyzed. In this case 25 years of operation.
Discount Rate: in US projects, it is the discount that federal agencies offer for this kind of renewable project. 2,5% is put following the tool indications.
Renewable Energy System Cost and Performance
Capital cost in $/kW.
Capacity factor of the plant
Fixed O&M costs in $/kWh per annum.
Variable O&M costs in $/kWh per annum.
Heat Rate. For renewable energy systems that do not require fuel, this variable is 0.
Fuel cost in $/MMBtu. For such energy system that does not require fuel, this variable is 0.
Today's Utility Electricity Cost:
Electricity price in cts$/kWh. 9.84 cts$/kWh has been selected as an average in USA according to http://www.eia.gov/electricity/state/.
Cost Escalation Rate. Projected nominal/real cost escalation rate of utility purchased electricity over the length of the analyzed period. If entered a nominal Discount Rate, a nominal Escalation Rate is entered. If nominal, the cost escalation rate inclusive of general inflation is entered.

Finally, the calculation results are:
Levelized Cost of Utility Electricity in cts$/kWh
Simple Levelized Cost of Renewable Energy in cts$/kWh, which will be the result for the DSP Concept Plan.
Currency exchange used for this calculation was 1,00 USD = 0,880643 EUR.. The calculated LEC is 34.6 cts/kWh that is about 2 fold that of molten salt, but uncertainties in the plant make the LEC an illustrative value only.

1.3.6.4 Energy and environmental impact Assessment
EPPT was in charge of evaluating the energy and environmental impact of Dense Particle Suspension (DPS or UBFB) solar power plant. A number of considerations need to be accounted for when assessing the environmental impact.
The sensible heat stored is higher with powders than with thermal oil and molten salts due to “unlimited” upper and lower temperature of the mineral. It could even be increased implementing PCM (phase change material), with tubular encapsulation. EPPT keeps working on this possibility.
Studies were developed in CSP2 project with SiC, but other materials such as cristobalite – crystalline SiO2- can be thought of. Indeed, agglomeration is a problem for SiC but it is not for cristobalite.

As mineral powders, there exists neither explosion risk nor fire hazards. In addition, there is no toxicological effect provided dust is contained and abated by high temperature filtration. SiC as well as cristobalite contain no fibers. Nonetheless, mask using is necessary, especially for cristobalite. Finally, in DPS system the rate of breakage (thus attrition) is very low because of (or thanks to) low excess velocities and orifice velocity.
Globally SHE (Safety, Health and Environment) is good for such powders at low velocities. Moreover, the overall assessment of powder UBFB (especially cristobalite) is much lower than thermal oil and molten salts (pollution free, less costly...) and comparable to CO2 and steam, but permitting thermal storage contrarily to the latter.
For recycle loop, pneumatic transport must be excluded, screw conveyors hardly stand extrapolation since they cannot permit lifting in a 120 m high tower, mechanical transport (bucket elevator) is best since it generates very low attrition and wear.

Potential Impact:
Potential impact of the CSP2 project may be defined in two main directions:
1. Innovative components and configurations for CSP plants.
2. Solar thermochemistry: solar heating and solar-assisted chemical treatment of particles.
1.4.1 Innovative components and configurations for CSP plants
CSP2 project has validated at TRL 4 dense suspension of fluidized particles as new heat transfer fluid that can be used for direct thermal energy storage at temperature higher than 560°C, the upper limit operation temperature of molten salt. The working temperature range was extended to 750°C. This temperature is well adapted to be associated to s-CO2 cycles with efficiency larger than 50% and hybrid combined cycles with efficiency of 50-55% (see Figure 21). In this latter case the solar fraction is about 60% and it can be increased with further improvement of particle temperature. We think that 850°C particle temperature is attainable with two main improvements:
- Implementing ultra-high temperature alloys.
- Adding fins inside the heat transfer tubes in order to increase the heat exchange coefficient from 1000 W/m2.K to about 2000 W/m2.K.
Moreover new thermodynamic cycles can be proposed in order to fit perfectly with solar heat temperature level. As a consequence these improvements will results in conversion efficiency increase and cost reduction.

Figure 21: The CSP2 concept associated with a combined cycle.

In this domain, the next step is the technology validation at TRL 5. It should include:
(1) Demonstration of a multi-megawatt, high temperature particle loop, including the high temperature solar receiver, the heat storage and the heat discharge system, in a relevant environment;
(2) Tests of the whole power plant;
(3) Assessment of the various highly efficient thermodynamic cycles that can be coupled with a particle thermal loop, and evaluation of the gain in efficiency with respect to current technologies;
(4) Scale-up plan and feasibility study of a ~100 MWel commercial plant.

Consequently the main impacts will be:

➢ Reducing the technological risks for the next development stages (demonstration);
➢ Increasing significantly the technology performance;
➢ Reducing life-cycle environmental impact;
➢ Nurturing the development of the industrial capacity to produce components and systems and opening new opportunities;
➢ Contributing to strengthening the European industrial technology base, thereby possibly creating growth and jobs in Europe;
➢ Reducing renewable energy technologies installation time and cost and/or operational costs, hence easing the deployment of renewable energy sources within the energy mix;
➢ Increasing the reliability and lifetime while decreasing operation and maintenance costs, hence creating new business opportunities;
➢ Contributing to solving the global climate and energy challenges;
➢ Improving EU energy security;
➢ Making variable renewable electricity generation more predictable and grid friendly, thereby allowing larger amounts of variable output renewable sources in the grid;
➢ Bringing cohesion, coherence and strategy in the development of new renewable energy technologies.
1.4.2 Solar thermochemistry: solar heating and solar-assisted chemical treatment of particles
The constraints of particle thermal treatment processes are completely different from the usual needs of power production using concentrating solar technology. Energy intensive industries, such as e.g the cement industry (based on CaCO3 calcination), need the major part of their energy input as thermal heat and they are (behind the power industry) the biggest energy consumers and CO2 emitters. The cement, lime and clay sector represents more than 10% of antropogeneous CO2 emission.
Apart calcination, the CSP2 technology may be applied to the thermal transformation of minerals in general, and of such thermo-chemical transformations occurring at temperatures between 700 °C and 1000°C.
Several mineral transformations fall within these objectives, i.e.:
CaCO3 to CaO at 800 to 950 °C (as applied in lime and cement kilns, e.g. Lhoist, Carmeuse, Cemex and others); CaMg(CO3)2 - dolomite - to CaO.MgO at 650 to 750 °C (as applied by e.g. Dolomeuse-Lhoist, Trier Kalk- und Dolomit Werke), pretreatment of low quality phosphate rock to remove hydrocarbon contamination at ~700 to 750 °C (as widely applied in the USA and Morocco); removal of organic pollutants from clays at 600 to 750 °C [as applied by e.g. Hepworth Ceramics (UK) and Wienerberger (B)]; activation of pozzolanic clays at 700 to 750 °C (as applied to make pozzolan more suitable for producing cement and lime-based mortars); in roasting pyrite (ZnS) to ZnO at 650 to 900 °C (as applied by Umicore and others); in the direct reduction of hematitic cinder to hematite at 530 to 600 °C (as initially applied by Montecatini); in the synthesis of AlF3 from Al(OH)3 at temperatures around 650 to 750 °C; and in different other applications.
Due to the wide range of applications, a specific market size cannot be evaluated, however the cement and lime market are clearly the biggest targeted markets.
The cement production by region is illustrated in Figure 22. The world cement production was 3000 Mt in 2009; it reached about 3900 Mt in 2013 and it will likely reach 4800 Mt in 2016. Total tonnage produced in EU 27 in 2013 amounted to over 132 million tons .

Figure 22: Evolution of world cement production by regions.

The cement production is growing very fast in Asia and Africa. This latter continent is as well after Europe a clear target for the application of the solar-induced conversion technology.
In EU 27 in 2006, the lime production was estimated at 28 million tons, roughly 12% of the 227 million tons produced worldwide.
The issues related to solar treatment of reactive particles and particularly calcination will be addressed by the new European project SOLPART (High Temperature Solar-Heated Reactors for Industrial Production of Reactive Particulates), project N 654663 (2016-2019). For this application 950°C is the targeted temperature.
The general concept is illustrated in Figure 23 as proposed in .

Figure 23: Tower concept for the integration of a solar calciner in a cement plant.

List of Websites:
http://www.csp2-project.eu

Contact

FLAMANT, Gilles (Head of Laboratory)
Tel.: +33 4 68 30 77 58
Fax: +33 4 68 30 29 40
E-mail

Subjects

Energy Saving
Record Number: 187985 / Last updated on: 2016-08-10