Professional Documents
Culture Documents
Real-Time Optimization of A Solar-Natural Gas Hybrid Power Plant To Enhance Solar Power Utilization
Real-Time Optimization of A Solar-Natural Gas Hybrid Power Plant To Enhance Solar Power Utilization
Authorized licensed use limited to: Israel Electric Corporation. Downloaded on January 03,2023 at 15:27:44 UTC from IEEE Xplore. Restrictions apply.
producing electricity through the high pressure (HP) turbine. through regulating hot air flow from the secondary firebox.
The outlet steam passes through the reheater maintains the When the solar irradiance is low, the boiler ramps up gas
steam temperature (at 824 K). The steam is divided between firing to maintain the desired temperatures and fulfill the
the closed heat exchanger and IP turbine. The outlet steam is demand of electricity. On the other hand, when there is high
further divided between the low-pressure (LP) turbine and DNI, the boiler backs down. Hence, the secondary firebox
low temperature boiler feedwater (BFW) preheater based ramps up gas firing to get heated air in order to maintain the
upon controlling the BFW temperature. The outlet steam desired temperatures
from the LP turbine is condensed to water and sent to the
high temperature BFW preheater. Then the steam passes
through the closed heat exchanger and solar preheater. The
steam outlet of the solar preheater enters the economizer to
achieve the desired temperature.
The steam from the economizer outlet recirculates back to The hybrid model ran for a year of simulation time. The
the boiler. Fundamental energy and mass conservation input variables that were perturbed were the temperature set
equations are used to simulate and design different point of the solar collector, HTF output temperature (Tsp),
components of the plant. solar irradiance (DNI), fraction of HTF entering the steam
generator (f), and the ambient temperature (Tamb). Here DNI
and Tamb are disturbances. Tsp and f are decision variables.
The output collected is the total solar power (Sp) from the
plant. The perturbed data was collected every four hours to
2) Control of the system ensure that steady state was reached before data collection.
Hence, a model fit was obtained from the data where solar
Due to the inherently transient nature of solar energy and power is the function of the input variables as stated in
other disturbances in the system like solar irradiance, ambient equation 1.
temperature, wind speed, etc., it is essential to have good
control in the system to maintain desired temperatures and Sp = f (Tsp , DNI, Tamb, f) (1)
power output. The solar trough uses a feed forward plus
The following equation fits with the data obtained from the
feedback control to maintain the HTF temperature output at
model:
the desired set point. The feedback controller responds by
maintaining the mass flow rate in response to the transmitter Sp = 349 − 0.02 Tsp + 0.00124 DNI − 2.4 Tamb −
of solar irradiance. PID control is used through IMC tuning 0.03f + 4.16 Tsp2+ 1.7 DNI 2+ 0.004 Tamb 2 −7.5
to offset disturbances. The steam drum has a master f2+ 0.0001 DNIf (2)
controller that regulates the steam flow rate based on the
power output signal from the slave controller. The
superheater and reheater maintain the steam temperature
3003
Authorized licensed use limited to: Israel Electric Corporation. Downloaded on January 03,2023 at 15:27:44 UTC from IEEE Xplore. Restrictions apply.
. percent meaning the total flow is directed to the steam
generator.
The multiple least squares method was used for fitting the
data. Figure 3 shows the parity plot predicted power with the Figure 4: Optimized result for DNI=400 W/m2 and ambient
temperature of 300 K
actual power produced for the validation of the dataset. The
R2 obtained is 0.95, suggesting it is a good fit for the
quadratic model. Although the model does have some
correlated residuals, the model is reasonably accurate, given
that it is restricted to be quadratic. In order to optimize the
model, solar power is maximized in terms of inputs using
quadprog in Matlab as the quadratic program (QP) solver.
Tsp sets the outlet temperature of the HTF from the
parabolic trough. At higher solar irradiance, the PID
controller of the parabolic trough increases the mass flow
rate to control the temperature of HTF at the desired set
point. f is the split fraction of HTF entering the steam
generator. In the non-optimized model, Tsp is set to a
constant value of 665K and f is kept at 100%, meaning the
full HTF flow is directed towards the steam generator. With
that functional dependence, the general form of the
optimization is:
Max Sp
Figure 5: Optimized result for DNI=100 W/m2 and ambient
590 < Tsp <665 (temperature set point in K) temperature
3004
Authorized licensed use limited to: Israel Electric Corporation. Downloaded on January 03,2023 at 15:27:44 UTC from IEEE Xplore. Restrictions apply.
Figure 8: DNI profile for January 1st and June 1st for Las Vegas
In general, the total mass flow rate of the HTF follows the
DNI pattern, meaning higher DNI leads to a total higher mass
flow rate and vice versa. On January 1st, the mass flow rate of
HTF in the steam generator is minimal or zero at lower DNIs.
The optimizer directs the majority of the flow to the solar
preheater. As the DNI increases, more than 500 W/m2 the
mass flow rate peaks to a maximum in the steam generator
because both the DNI and the fraction entering steam
generator f are high. Subsequently, the flow drops back to the
minimum when the DNI starts falling towards the end of the
day. The solar irradiance reduces as the sun goes down; the
mass flow rate of HTF also peaks up at the solar preheater
Figure 7: Power production optimized versus non-optimized at and subsequently drops to the minimum level. This follows
different DNIs the surface plots as mentioned earlier in Figures 4, 5 and 6.
3005
Authorized licensed use limited to: Israel Electric Corporation. Downloaded on January 03,2023 at 15:27:44 UTC from IEEE Xplore. Restrictions apply.
C. Results for a year
On June 1st a small amount of HTF mass flow in the solar In Figure 13, which shows the solar fraction for the
preheater at the beginning of DNI rise is observed, and the optimized and non-optimized case, there is a general trend of
rest of the HTF flow is diverted toward the boiler in Figure the solar fraction increasing during the summer and
10. From Figure 8, it is observed that the DNI is greater than decreasing towards the winter. This is because summer
zero and less than 500 W/m2 for a short period of time both months have higher solar irradiance days compared to winter.
in the beginning and end of the day, enabling the optimizer The optimized case always has a higher solar fraction
to make an impact for a short period. This explains the small regardless of the month. However, the improvement of solar
mass flow rate of HTF into the solar preheater. fraction solely depends on the DNI patterns throughout the
month.
As opposed to Figure 13, we observe from Figure 14 that
the overall efficiency of the plant has a general trend of
B. Solar Power Production
decreasing during the summer and increasing during the
winter. Because natural gas is more energy efficient fuel
compared to solar, and in winter, more natural gas is used, as
there is lower solar irradiance. However, the overall
efficiency of the optimized plant is slightly higher than the
non-optimized case. The effect of overall efficiency increase
is less in optimization because the plant is a 100 MW
capacity and solar fraction results indicate that natural gas is
the major contributor to energy production. If the demand for
electricity is low (less than 100 MW) less power would be
needed and the solar fraction will increase. In that case,
optimization will have a higher impact on the overall
efficiency.
Figure 11: Solar Power production in January 1st optimized versus Solar to electric efficiency (STE) is the amount of energy
non-optimized model produced from the solar part of the plant divided by the total
energy obtained from the sun. No general trend was observed
On January 1st as shown in Figure 8 significant portion of for STE over the months in Figure 15. In summer, the
the day the DNI is below 500 W/m2. Hence, for a large optimized solar to electric efficiency is higher than the non-
portion of the day optimization improves the solar power optimized case. In winter, they are slightly higher. This again
production compared to the non-optimized plant as relates to the fact of having a higher solar share during the
observed in Figure 11. The solar power production summer enables the optimizer to obtain a higher STE.
improves by 25.56 % for the optimized case.
Figure 12: Solar Power production in June 1st optimized versus non-
optimized model
3006
Authorized licensed use limited to: Israel Electric Corporation. Downloaded on January 03,2023 at 15:27:44 UTC from IEEE Xplore. Restrictions apply.
easy to implement on existing natural gas plants without the
need of extensive capital investment by sharing infrastructure
and integrating energy.
The quality of the optimization depends on the data fitting
of the model. An R2 value of 0.95 implies that this model is a
good data fit. Hence, the model optimizes the solar power
output well. By performing an optimized operation for a day,
the hybrid plant demonstrated significant (25%)
performance improvement when there was low solar
irradiance at winter. At summer or at higher DNI, the
optimized plant performance was better in comparison to the
non-optimized plant but to a lesser extent (1%). This agrees
with the study performed by Usaola J et al [11]. The
Figure 14: Monthly efficiency graph for optimized versus non- dynamic real-time optimization inputs optimum set points
optimized case
for the decision variables and ensure to maximize the solar
output by energy integration.
The yearly results of the solar fraction, STE and overall
efficiency demonstrates that the optimized model increases
the performance through improving solar fraction (10.7%)
and solar to electric efficiency (10%). The overall efficiency
improvement is less (1.5%) due to optimization. However,
the efficiency improvement depends on the capacity of the
plant. Hence, the optimized hybrid model can ensure the
reliability of the power supply during cloudy or night, when
there is not enough solar power available and the optimized
results improve the solar power production leading to
greener energy.
REFERENCES
[1] K. Powell and J. Hedengren, “Dynamic optimization of a solar
thermal energy storage system over a 24 hour period using
weather forecasts,” Control Conf. (ACC …, 2013.
[2] G. Bakos and C. Tsechelidou, “Solar aided power generation of a
300 MW lignite fired power plant combined with line-focus
Figure 15: Monthly STE graph for optimized versus non-optimized parabolic trough collectors field,” Renew. Energy, 2013.
case [3] K. Powell, K. Rashid, K. Ellingwood, and J. Tuttle, “Hybrid
concentrated solar thermal power systems: A review,” Sustain.
Energy …, 2017.
[4] D. Mills, “Advances in solar thermal electricity technology,” Sol.
Table 1 Yearly summary of improvement Energy, vol. 76, no. 1–3, pp. 19–31, Jan. 2004.
[5] Y. Zhao, H. Hong, and H. Jin, “Evaluation criteria for enhanced
solar–coal hybrid power plant performance,” Appl. Therm. Eng.,
Optimized Non- Improvement vol. 73, no. 1, pp. 577–587, 2014.
[6] A. Lee, O. Zinaman, and J. Logan, “Opportunities for Synergy
optimized Between Natural Gas and Renewable Energy in the Electric
Power and Transportation Sectors,” 2012.
Solar 7.2% 6.5% 10.7% [7] N. Fulton, M.; Melquist, “Natural Gas and Renewables: The Coal
Fraction to Gas and renewables Switch is on!,” Nat. Gas Renewables,
2011.
Solar to 11% 10% 10% [8] J. Peterseim, A. Tadros, U. Hellwig, and S. White, “Increasing the
electric efficiency of parabolic trough plants using thermal oil through
efficiency external superheating with biomass,” Energy Convers., 2014.
[9] M. E. Demir and I. Dincer, “Development of an integrated hybrid
Overall 32% 31.5% 1.5% solar thermal power system with thermoelectric generator for
efficiency desalination and power production,” Desalination, vol. 404, pp.
59–71, Feb. 2017.
[10] S. Wilcox, “National Solar Radiation Database 1991--2005
Update: User’s Manual,” 1991.
IV. CONCLUSION [11] J. Usaola, Operation of concentrating solar power plants with
storage in spot electricity markets, vol. 6, no. 1. Institution of
Solar power adds some interesting dynamics to the hybrid Engineering and Technology, 2012.
plant due to its intermittent nature. Natural gas enables the
plant to supply power with reliability, supplementing solar
energy when there is not enough supply. CSP technology is
3007
Authorized licensed use limited to: Israel Electric Corporation. Downloaded on January 03,2023 at 15:27:44 UTC from IEEE Xplore. Restrictions apply.