4. MODELLING AND RESULTS
4.4 Validation of simulation results
4.4.4 CLFR result validation
Gharbi et al. (2011) compared the annual optical efficiency of the concentrators of parabolic troughs and CLFRs using DNR data for the Hassi Rmel region of Algeria. They found the annual collector efficiency of the parabolic trough to be 55.8% while the CLFR returned a value of 45.8%. They conclude that the optical and thermal efficiency of the CLFR is lower because of the greater influence of the incidence angle and cosine factor on this type of collector when compared to that of the parabolic collector.
Giostri et al. (2011) compared the annual performance of commercial Fresnel collectors with parabolic trough collectors using Therminol VP-1 as the HTF. In addition they investigated a DSG case for the Fresnel collectors. Thermoflex software was used with the following design parameters to facilitate the comparison:
i. DNI of 900 W/m2,
ii. Incidence angle equal to zero, iii. Solar multiple equal to one, iv. No thermal storage,
v. Net power output of 50 MW, vi. Evaporative cooling.
The model was set up with an optical efficiency of 67% for the Fresnel collector based on manufacturer data from Novatec Biosol (2011). An optical efficiency of 75% was used for the parabolic trough collector based on Eurotrough ET100 information from Geyer et al.
(2002). Using these parameters they found that at design conditions the optical, thermal and piping efficiencies of the parabolic trough collectors exceed those of the linear Fresnel collectors. The parasitic losses of the CLFR plant were lower than those of the parabolic trough plant. The overall efficiency of the parabolic trough plant was higher than that of the CLFR plant, resulting in the parabolic trough plant generating 3% more gross electrical power than the CLFR plant. The net power output per square meter of the CLFR plant was found to be higher than that of the parabolic trough plant, which agrees with the results obtained in this study.
Giostri et al. (2011) also performed an annual simulation for the location of Las Vegas, Nevada, USA in Thermoflex, with the input and output parameters shown in Table 4-19.
Table 4-19: Comparison of Giostri et al. (2011) simulations in Thermoflex to those of this simulation in SAM
Simulation Inputs
Giostri et al. (2011) parabolic trough
simulation
This parabolic trough simulation
Giostri et al. (2011) CLFR simulation
This CLFR simulation Simulation Software Thermoflex System Advisor Model Thermoflex System Advisor Model Site [Town, Province,
Country]
Las Vegas, Nevada, USA
Upington, Northern Cape, South Africa
Las Vegas, Nevada, USA
Upington, Northern Cape, South Africa Site coordinates [Lat.,
Long.]
36°10’30”N,
115°08’11” W 28°4’S, 21°3’E 36°10’30”N,
115°08’11” W 28°4’S, 21°3’E Average Temperature
[°C] 19.8 21.5 19.8 21.5
Annual averaged DNR
[kWh/m²/annum] 2592 2829.8† 2592 2829.8†
Nameplate/Total
capacity [MWe] 50 100 50 100
Power cycle cooling
method Evaporative Air-cooled Evaporative Air-cooled
Solar multiple 1 2 1 1.79
Hours of thermal
storage 0 6 0 0
Aperture area [m²] 235,899 860,010 268,596 862,848 Total land area [km²] 0.684 3.141 0.595 1.381 Simulation Outputs
Giostri et al. (2011) parabolic trough
simulation
This parabolic trough simulation
Giostri et al. (2011) CLFR simulation
This CLFR simulation Annual electric output
[MWh] 97,818 401,839 74,454 393,984
† CRSES DNR values used in the final model in section 4.3.
Based on their results, Giostri et al. (2011) note that while the linear Fresnel collector is capable of collecting more solar energy than an equivalent parabolic trough collector, it fails to reflect that energy efficiently to the receiver. Therefore, particularly at high insolation incidence angles the CLFR plant is unable to match the parabolic trough plant’s optical efficiency. This, in combination with the higher transverse angle losses of CLFR plants in winter, results in the lower annual electric output. These effects are modelled by SAM (Figure 4-16), however the fossil fuel backup of the modelled CLFR plant guarantees nominal power block output throughout most of the day, removing some of the advantage of the parabolic trough’s higher optical efficiency.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
1 252 503 754 1005 1256 1507 1758 2009 2260 2511 2762 3013 3264 3515 3766 4017 4268 4519 4770 5021 5272 5523 5774 6025 6276 6527 6778 7029 7280 7531 7782 8033 8284 8535
Optical Efficiency
Hour
Parabolic trough collector average optical efficiency, Hourly
Central receiver solar field optical efficiency, Hourly
CLFR collector optical efficiency, Hourly
Figure 4-16: Average hourly optical efficiency of parabolic trough, central receiver and CLFR Morin et al. (2012) performed a comparison of parabolic trough and CLFR power plants using ECOSTAR (developed by the DLR) and ColSim (Fraunhofer ISE in-house software).
Meteorological input data for Daggett, California, USA were used for their annual performance simulations. Their approach was to optimize the aperture area of both technologies in order to match the LCOE. An average was taken from the ECOSTAR and ColSim results of Morin et al. (2012) and compared to this study in Table 4-20.
Table 4-20: Comparison of Morin et al. (2012) simulations in ECOSTAR and ColSim to this study
Simulation Inputs
Morin et al. (2012) parabolic trough
simulation
This parabolic trough simulation
Morin et al. (2012) CLFR simulation
This CLFR simulation Simulation Software ECOSTAR and
ColSim System Advisor Model ECOSTAR and
ColSim System Advisor Model Site [Town, Province,
Country]
Daggett, California, USA
Upington, Northern Cape, South Africa
Daggett, California, USA
Upington, Northern Cape, South Africa Site coordinates [Lat.,
Long.] 34°51’N, 116°49’W 28°4’S, 21°3’E 34°51’N, 116°49’W 28°4’S, 21°3’E Average Temperature
[°C] 19.8 21.5 19.8 21.5
Annual averaged DNR
[kWh/m²/annum] 2791 2829.8† 2791 2829.8†
Nameplate/Total
capacity [MWe] 50 100 50 100
Power cycle cooling
method Evaporative Air-cooled Evaporative Air-cooled Hours of thermal
storage 0 6 0 0
Aperture area [m²] 269,000 860,010 371,000 862,848 Total land area [km²] 0.941 3.141 0.742 1.381 Simulation Outputs
Morin et al. (2012) parabolic trough
simulation
This parabolic trough simulation
Morin et al. (2012) CLFR simulation
This CLFR simulation Annual electric output
[MWh] 110,500 401,839 93,500 393,984
† CRSES DNR values used in the final model in section 4.3.
Morin et al. (2012) noted that during periods of high insolation around solar noon the CLFR was forced to discard considerable excess energy due to its larger aperture area per receiver.
They also observed that the optical losses inherent to the design of the CLFR reduce the amount of thermal energy delivered to the power block during early morning and late afternoon periods (high insolation incidence angles). This resulted in reduced operating time for the plant, which in turn led to the lower annual electric output of the CLFR plant in comparison to the parabolic trough plant.
In this study an identical fossil fill fraction was used across the parabolic trough, central receiver and CLFR models. This was required to match the power dispatch schedules.
However, since the CLFR model does not have a TES system, it does not need to use a portion of the fossil fill fraction to charge TES system tanks. Therefore it used the complete fossil fill fraction to ensure nameplate electrical output is maintained throughout the day.
Thus the backup fossil fuel boiler compensated for some of the optical losses which resulted in reduced operating time and electric output in the models of Morin et al. (2012). Setting the fossil fill fractions to zero across the board results in the CLFR and parabolic trough net annual electric output agreeing more closely with the result of Morin et al. (2012).
In sections 4.4.1 and 4.4.2 comparisons were carried out between the parabolic trough and central receiver CSP plants simulated in this study and those simulated by Wagner (2012) in SAM. Wagner (2012) describes the SAM model of the CLFR plant in more detail, noting the non-normal orientation of the primary reflectors towards incoming beam radiation as the primary optical loss of the CLFR. He also mentions the significant seasonal variation of optical performance of the CLFR. As mentioned in section 4.4.2 his modelling approach is different to the one adopted here. However, insight can be gained from the comparative results of his work despite this difference in approach. The results of his parabolic trough and CLFR models are compared to the results of this study in Table 4-21.
Table 4-21: Comparison of Wagner (2012) simulation metrics to those used in this study
Simulation Inputs
Wagner (2012) parabolic trough
simulation
This parabolic trough simulation
Wagner (2012) CLFR simulation
This CLFR simulation Site Phoenix, Arizona,
USA
Upington, Northern Cape, South Africa
Phoenix, Arizona, USA
Upington, Northern Cape, South Africa Nameplate electrical
output (gross) [MWe] 150 115 158 107 Fossil fuel backup No Yes No Yes
Solar multiple 1.46 2 1.7 1.79 Hours of thermal
storage 1 6 0 0
Field aperture area
[m²] 850,302 860,010 1,208,000 862,848 Heat transfer fluid Solar salt Solar salt Water/steam Water/steam
Power cycle cooling
method Air-cooled Air-cooled Air-cooled Air-cooled Power cycle efficiency 0.378 0.3774 0.38 0.38
Boiler operating
pressure [bar] 100 100 110 110
Simulation Outputs
Wagner (2012) parabolic trough
simulation
This parabolic trough simulation
Wagner (2012) CLFR simulation
This CLFR simulation Annual water usage
[m3] 72,237 82,467 31,560 77,452 Capacity factor [%] 24.8 45.9 23.0 44.7 Annual electric output
[MWh] 299,700 401,839 299,700 393,984
The results of Wagner’s (2012) SAM simulation reflect the optical limitations of CLFR technology in comparison to parabolic trough technology. This is expressed by the much larger solar field aperture area required to obtain the same annual electric output. This study also reflects the higher aperture area required from a CLFR plant to obtain a similar nameplate electrical output to a parabolic trough.
Wagner’s (2012) results also reflect the seasonal variation of CLFR optical performance via the lower capacity factor in comparison with the parabolic trough. Due to this study using fossil fuel backup, this seasonal variation was compensated for. With fossil fuel backup the parabolic trough plant capacity factor is 2.7% higher than the CLFR plant. Removing the fossil fuel backup, the CLFR plant capacity factor becomes 32.6% lower than the parabolic trough plant. This is similar to Wagner’s (2012) result.
The CLFR plant model deploys its fossil fuel backup in such a manner that it compensates for some of its optical inefficiencies. Despite this, its resultant net annual electric output is still lower than the central receiver and parabolic trough plant models. Removing the fossil fuel backup results in the CLFR plant model producing 32.6% less net annual electric output than the parabolic trough plant model. These results agree on average with the work of Giostri et al. (2011) and Morin et al. (2012) as shown in Table 4-22.
Table 4-22: Estimation of annual electric output deviation
Simulation Output Giostri et al. (2011) parabolic trough simulation
Giostri et al. (2011) CLFR simulation
Percentage difference in parabolic trough and
CLFR result Annual electric output
[MWh] 97,818 74,454 31.4%
Simulation Output Morin et al. (2012) parabolic trough simulation
Morin et al. (2012) CLFR simulation Annual electric output
[MWh] 110,500 93,500 18.2%
Average difference between parabolic trough and CLFR annual electric output 24.8%
Simulation Output
This parabolic trough simulation with fossil fuel
backup
This CLFR simulation with fossil fuel backup
Percentage difference in parabolic trough and
CLFR result Annual electric output
[MWh] 401,839 393,984 2%
Simulation Output
This parabolic trough simulation without fossil
fuel backup
This CLFR simulation without fossil fuel backup
Percentage difference in parabolic trough and
CLFR result Annual electric output
[MWh] 356,654 242,570 47%
Average difference between parabolic trough and CLFR annual electric output 24.5%