Anda di halaman 1dari 11

Proceedings of the ASME 2015 9th International Conference on Energy Sustainability

ES2015
June 28-July 2, 2015, San Diego, California

ES2015-49310

DESIGN OF A SECONDARY CONCENTRATOR FOR A


SMALL PARTICLE HEAT EXCHANGE RECEIVER

Olivier Berchtold Fletcher Miller


Alstom Renewable Combustion and Solar Energy Laboratory,
1
5102 Birr (AG), Switzerland Department of Mechanical Engineering
San Diego State University,
San Diego, CA, USA

ABSTRACT peak power hitting the window is found to increase very


The design of a secondary concentrator for the Small slightly only. This means the maximum window design
Particle Heat Exchange Receiver (SPHER) using a Monte specifications do not need to be adjusted dramatically to be able
Carlo Ray Tracing (MCRT) method is discussed in this paper. to accommodate the average power increase. First indications
Applying basic MCRT rules, a modular solver logic for using the MCRT output in preliminary receiver simulations
secondary concentrators is established. The logic is coded into suggest increased receiver efficiency and a boost in receiver
FORTRAN subroutines to be compatible with MIRVAL, a ray outlet gas temperature. The combined effect of these
trace code for heliostat fields created by Sandia National improvements is expected to raise overall power generation
Laboratories. Based on a 3D Compound Parabolic Concentrator efficiency by improving the gas- / steam turbine combined
(3D-CPC) example the power of the simulation based on the cycle efficiency.
Sandia heliostat field calculations is demonstrated.
The results of the simulations are used to calculate the INTRODUCTION
solar flux distributions in the ideal 3D CPC inlet and outlet Electricity alone accounted for 18.1% of the total
planes as well as the incident ray distribution hitting the worldwide energy consumption in 2012 (1). The consumption
secondary concentrator. Code modifications to account for has quadrupled within the past 40 years (Figure 1) with no stop
surface irregularities and spectral reflectivity are implemented to this trend expected in the near future. On the contrary
in the appropriate FORTRAN subroutine. Using the automated electricity is seen as a key component to help render the
simulation routines first the optimal receiver tilt angle and transportation sector more sustainable over the coming decades.
secondly the secondary concentrator acceptance angle are
determined. These parameters combined with the fixed
secondary concentrator outlet radius – which is determined by
the SPHER window diameter - fully constrain the 3D CPC
geometry. The flux maps generated using MATLAB post
processing on the derived concentrator results clearly indicate
the strengths and weaknesses of the specific concentrator and
heliostat field combination. 1
The influence of the secondary concentrator on the
window incident flux distribution and window transmission,
absorption and reflection properties is evaluated. Early findings
using the code suggest higher yearly average power entering
the receiver when compared to a non-secondary case. The Figure 1: Total final electricity consumption by sector from
reason for this effect is found in increased heliostat efficiency 1971 to 2012 [Mtoe] (International Energy Agency IEA,
towards the edges of the heliostat field. At the same time the 2014)
Governments have to balance sustainability requirements
(energetic, financial and social) to reduce energy consumption
1
Research performed at SDSU C&SEL while enrolled as a student at while not compromising the financial and social integrity of
ETH Zürich

1 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


their strategies for the future. Renewable energy production lies
at the heart of these endeavors.
The US Department of Energy Sunshot initiative is an
effort to make solar energy cost competitive with more
traditional non-renewable energy. The project discussed in this
paper is based on the concentrating solar power technology and
aims to use a novel receiver design to achieve higher solar to
electrical conversion efficiency and thus a lower levelized cost
of electricity.

The SPHER (Small Particle Heat Exchange Receiver (2))


uses a carbon black aerosol to volumetrically absorb
concentrated sunlight atop a tower in a concentrated solar
power plant. The absorbed energy causes particles to oxidize
and finally creates a clear, hot gas stream which exits the
Figure 3: NSTTF heliostat field at Sandia Laboratories
pressurized receiver cavity (Figure 2). The gas stream is then
(http://energy.sandia.gov/?page_id=11924, 2014/10/15)
used in a Brayton cycle to generate electricity. In case the solar
power receiver outlet temperature is not sufficient for efficient
RECEIVER HEAT BALANCE
cycle operation (due to time of day, cloud transients, etc.) a
combustor is added to the cycle to further increase the receiver The absorption efficiency for a windowed cavity receiver
outlet gas temperature (i.e., turbine inlet temperature) and neglecting conduction and convection losses can be deduced
improve the availability of the power plant. from a heat balance across the receiver aperture (3). Following
The current 5MW thermal input receiver is designed to be the line of argumentation by Steinfeld and Schubnell the
placed and tested on top of Sandia Laboratories’ National Solar absorbed power is given by the incident solar power on the
Thermal Testing Facility (NSTTF) tower (Figure 3) during a receiver cavity multiplied with the absorption coefficient of the
later project stage. cavity. This absorption coefficient 𝛼𝑒𝑓𝑓 is higher than the
absorption coefficient 𝛼 of the cavity walls themselves because
incident radiation can be submitted to multiple reflections
inside the cavity.
The power leaving the cavity through the aperture on the other
hand is defined by radiation losses only in this simplified setup.
It is given by the cross-sectional area of the aperture multiplied
by the cavity temperature 𝑇 (considered homogenous) to the
fourth power multiplied with the emittance of the receiver 𝜀𝑒𝑓𝑓
and the Stefan-Boltzmann constant 𝜎 = 5.67 ∗ 10−8 W/m2-K.
Combining these definitions into an efficiency reading by
dividing through the total incident cavity power, we get the
receiver absorption efficiency:

Figure 2: Schematic of pressurized volumetric receiver with 𝛼𝑒𝑓𝑓 ∗ 𝑃𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 − 𝜀𝑒𝑓𝑓 ∗ 𝜋 ∗ 𝑟 2 ∗ 𝜎 ∗ 𝑇 4


secondary concentrator used by the Small Particle Heat 𝜂𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 =
𝑃𝑖𝑛
Exchange Receiver (SPHER) project
In case the entire power incident on the cavity makes it into the
This paper presents San Diego State University’s (SDSU’s) aperture of the cavity, 𝑃𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 is equal to 𝑃𝑖𝑛 .
Combustion and Solar Energy Laboratory’s Monte Carlo ray If no useful power is extracted from the receiver and the
tracing analysis for the secondary concentrator of the SPHER. radiated power is equal to the incident solar power 𝑃𝑖𝑛 , the
Previous projects (13), (14), (15), (15), (16) have successfully highest possible temperature, the stagnation temperature 𝑇𝑠𝑡𝑎𝑔
used secondary concentrators to increase the receiver incident is reached.
solar flux concentration ratio. Unique to the SPHER is the use
of a consumable absorber, which makes it ideally suited to a 4
secondary concentrator since – in contrast to receivers based on 𝛼𝑒𝑓𝑓 ∗ 𝐼 ∗ 𝐶
𝑇𝑠𝑡𝑎𝑔 = ( )
tubes, recirculated particles, or porous materials - the small 𝜀𝑒𝑓𝑓 ∗ 𝜎
particle receiver effectively has no upper limit on the flux that it
can withstand. In this case 𝐼 is the direct normal incident radiation (DNI) and
𝐶 denotes the actual mean flux concentration ratio across the
aperture. Heliostat efficiency and heliostat land use factor are

2 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


included in the combined concentration ratio. This ratio is 𝛿𝜂 5 4
different from the purely geometric concentration ratio and also = 0 => (𝑇𝑜𝑝𝑡𝑖𝑚𝑎𝑙 ) − 0.75 ∗ 𝑇𝑒𝑛𝑣 ∗ (𝑇𝑜𝑝𝑡𝑖𝑚𝑎𝑙 )
𝛿𝑇
contains absorption within the air along the optical path of the 𝑇𝑒𝑛𝑣 ∗ 𝐼 ∗ 𝐶
− =0
sun rays. 4𝜎
For the following calculations we assume our cavity to be
ideally black, thus 𝛼𝑒𝑓𝑓 = 𝜀𝑒𝑓𝑓 = 1. This assumption allows The optimal temperature for each concentration ratio is found at
for a conservative estimate of radiation heat losses seeing as a the maximum efficiency point in Figure 4. Linking the maxima
black body is an ideal, diffuse emitter leading to the highest for each concentration ratio leads to a curve of maximum
possible emission at all frequencies. efficiencies for varying solar concentration ratios. It is evident
In our application the concentrated solar radiation is used to that the left part of each efficiency curve is limited by the
provide process heat to drive a gas turbine coupled to an Carnot efficiency while the drop experienced towards higher
electric generator. The conversion of the heat to work is thus temperatures is caused by re-radiation losses.
limited by the Carnot efficiency given by
1
C = 300
𝑇𝑒𝑛𝑣 C = 500
𝜂𝑐𝑎𝑟𝑛𝑜𝑡 = (1 − )
𝑇 0.8 C = 1000
C = 1500

Combined efficiency
where 𝑇 is the hot inlet temperature and 𝑇𝑒𝑛𝑣 is the C = 2000
environment or low temperature of the conversion cycle. 0.6 Maxima
The optimal solar to work efficiency is thus given by the
product of the abovementioned efficiencies: 0.4

𝜂𝑠𝑜𝑙𝑎𝑟−𝑤𝑜𝑟𝑘 = 𝜂𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 ∗ 𝜂𝑐𝑎𝑟𝑛𝑜𝑡


𝛼𝑒𝑓𝑓 ∗ 𝑃𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 − 𝜀𝑒𝑓𝑓 ∗ 𝜋 ∗ 𝑟 2 ∗ 𝜎 ∗ 𝑇 4 0.2
=
𝑃𝑖𝑛
𝑇𝑒𝑛𝑣 0
∗ (1 − ) 0 500 1000 1500 2000
𝑇 Temperature [K]

Replacing 𝛼𝑒𝑓𝑓 and 𝜀𝑒𝑓𝑓 with 1 this becomes: Figure 4: Combined absorption and Carnot efficiency

𝜂𝑠𝑜𝑙𝑎𝑟−𝑤𝑜𝑟𝑘 = 𝜂𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 ∗ 𝜂𝑐𝑎𝑟𝑛𝑜𝑡


MONTE CARLO RAY TRACING
𝑃𝑎𝑝𝑒𝑟𝑡𝑢𝑟𝑒 − 𝜋 ∗ 𝑟 2 ∗ 𝜎 ∗ 𝑇 4
= The calculations in this paper make extensive use of Monte
𝑃𝑖𝑛
𝑇𝑒𝑛𝑣 Carlo Ray Tracing (MCRT). MCRT is a numerical algorithm
∗ (1 − ) that works by repeating a given problem calculation for a large
𝑇 number of possible parameters and thus calculates a
probabilistic solution to the problem (5). Simply speaking the
The most important point to be taken from this formula is that
algorithm uses repeated random sampling over and over again
the total solar to power conversion efficiency is a balance
to calculate a relatively accurate solution to the problem.
between maximizing the incident radiation through the aperture
In our case we use the algorithm to trace random rays from the
and minimizing the losses caused by re-radiation off the walls
sun to the heliostat field and on into the receiver. By applying
through this same aperture.
statistics it is then possible to determine a qualitative and
In case useful power is extracted from the receiver cavity the
quantitative distribution of the rays’ behaviour. This means we
steady-state temperature of the receiver for a given power input
can calculate the fraction of solar power hitting the heliostats,
drops. This means that the re-radiation losses decrease
the secondary concentrator and the receiver and make reliable
significantly as they depend on the receiver temperature to the
performance predictions.
4th power. This in turn means the optimal cavity opening radius
The software implemented for this research uses MIRVAL (6)
increases.
heliostat field calculations as input. The advantage of using this
To determine the optimal operating temperature (still assuming
code lies in the efficiency provided by the FORTRAN
blackbody cavity) the derivative of the efficiency can be
programming language combined with powerful 21 st century
calculated which leads to the following polynomial to be solved
computers while the main downside is the lack of a graphical
(4):
user interface and post processing software. The design of the
secondary concentrator solver evolved from a simple single
case solver to a MIRVAL integrated solution (Figure 5)
including Mecit’s window solver (7).

3 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


In consequence all the secondary concentrator results were
post-processed with newly designed MATLAB routines which
allow both statistical and graphical evaluation of the results. In
a next step coupling the radiation distribution exiting the
window with an existing ANSYS FLUENT receiver model is
envisioned.

2.5
Single CPC
2

1.5

Multi CPC

ZCPC
1

0.5

0
1
Integration into MIRVAL
-0.5
0.5

-1 0
-0.5
Coupling with window 0 -0.5
0.5
1 -1
YCPC

Coupling with FLUENT XCPC


Figure 6: 3D CPC (Compound Parabolic Concentrator)

Figure 5: Simulation design flow chart Summing up the 3D CPC with an acceptance angle of 𝜃𝑚𝑎𝑥 its
most important characteristics are given by:
SECONDARY CONCENTRATOR MODEL 𝑟𝑜𝑢𝑡
A heliostat field is limited in the maximum achievable peak and 𝑓=
1 + sin(𝜃𝑚𝑎𝑥 )
average concentration ratio due to the incident angle spread
from the sun (8), heliostat tracking and surface deviations (9) where 𝑓 stands for the focal length of the parabola, 𝑟𝑜𝑢𝑡 is the
and further minor effects like absorption of solar energy in the 2D CPC outlet radius and 𝜃𝑚𝑎𝑥 denotes the CPC acceptance
air. half angle. The other characteristic parameters for the CPC can
In order to increase the maximum solar concentration ratio in be deduced by applying basic geometry, leading to the CPC
the receiver aperture a secondary concentrator is used. The length 𝐿 and the CPC inlet radius 𝑟𝑖𝑛 :
concentrator sits on top of the receiver tower and further
concentrates the solar energy incident from the heliostat field 𝑟𝑜𝑢𝑡 (1 + sin(𝜃𝑚𝑎𝑥 ))cos(𝜃max )
before it enters the receiver. 𝐿=
We use a non-imaging 3D CPC (Compound Parabolic sin2 (𝜃max )
Concentrator) (10) which consists of a rotated 2D CPC The 2D 𝑟𝑜𝑢𝑡
CPC itself offers the maximum theoretically possible geometric 𝑟𝑖𝑛 =
sin(𝜃𝑚𝑎𝑥 )
concentration ratio and can easily be described mathematically
(11). In essence it consists of a parabola that is tilted by the
Alternatively, the length is given by:
desired CPC acceptance angle around its global minimum in
the plane of reference and then mirrored along its original (non-
tilted) axis of symmetry. 𝐿 = (𝑟𝑖𝑛 + 𝑟𝑜𝑢𝑡 )cot(𝜃𝑚𝑎𝑥 )
In 3 dimensions the optimal solution is not evident. The 3D
The relationship between the z coordinate and the radius of
CPC was independently discovered by multiple parties in the
each point on the 3D CPC surface is given by the following
1970s (Baranov and Melkinov 1966, Hinterberger and Winston
1966, Ploke 1966 (12). Previous projects have successfully equation:
made use of 3D CPCs (Figure 6) which are created by rotating
a 2D CPC around the axis of symmetry or approximations of (𝑟𝑐𝑜𝑠(𝜃𝑚𝑎𝑥 ) + 𝑧𝑠𝑖𝑛(𝜃𝑚𝑎𝑥 ))2 + 2𝑟𝑜𝑢𝑡 (1 + sin(𝜃𝑚𝑎𝑥 )2 𝑟
this shape (DLR (13), Weizmann Institute of Science (14), PSI − 2𝑟𝑜𝑢𝑡 cos(𝜃𝑚𝑎𝑥 ) (2 + sin(𝜃𝑚𝑎𝑥 ))𝑧
(15), European Commission (SOLGATE project (16))). − 𝑟𝑜𝑢𝑡 (1 + sin(𝜃𝑚𝑎𝑥 ))(3 + sin(𝜃𝑚𝑎𝑥 )) = 0

4 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


𝑓𝑧 = 2𝑟𝑠𝑖𝑛(𝜃𝑚𝑎𝑥 ) cos(𝜃𝑚𝑎𝑥 ) + 2𝑧𝑠𝑖𝑛2 (𝜃𝑚𝑎𝑥 )
− 2𝑟𝑜𝑢𝑡 cos(𝜃𝑚𝑎𝑥 )(2 + sin(𝜃𝑚𝑎𝑥 ))

where 𝑓𝑦 is the same as 𝑓𝑥 due to symmetry. The surface normal


is then given by:

𝑓𝑥
2
√𝑓𝑥2 + 𝑓𝑦2 + 𝑓𝑧2
𝑛𝑥 𝑓𝑦
(𝑛𝑦 ) = 2
𝑛𝑧 √𝑓𝑥2 + 𝑓𝑦2 + 𝑓𝑧2
𝑓𝑧
2
( √𝑓𝑥2 + 𝑓𝑦2 + 𝑓𝑧2 )

The reflected ray direction is given by mirroring the incident


Figure 7: Important 3D CPC parameters ray on the surface. Knowing the surface normal a simple way to
determine the reflected ray’s direction vector is given by
Solving for the intersection point of the 3D CPC and an subtracting twice the surface normal component of the incident
incoming ray can be done analytically. It is however ray direction vector from the incident direction vector:
recommended to use a numerical method like the bisection
method instead seeing as the analytical solution yields a 4th 𝑣𝑥,𝑟𝑒𝑓𝑙𝑒𝑐𝑡𝑒𝑑 𝑣𝑥,𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡
order equation with algebraically unfavourable coefficients (see 𝑣
( 𝑦,𝑟𝑒𝑓𝑙𝑒𝑐𝑡𝑒𝑑 ) = (𝑣𝑦,𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 )
(17) for the derivation). Solving the exact 4th order equation can 𝑣𝑧,𝑟𝑒𝑓𝑙𝑒𝑐𝑡𝑒𝑑 𝑣𝑧,𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡
yield a bigger error than an appropriate numerical method. The 𝑛𝑥 𝑣𝑥,𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 𝑛𝑥
bisection method for example is very simple to implement and 𝑛 𝑣
− 2 ( 𝑦 ) (( 𝑦,𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 ) ∗ (𝑛𝑦 ))
still converges towards the exact solution linearly. 𝑛𝑧 𝑣𝑧,𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 𝑛𝑧
In our case the implementation of the bisection method to solve
for the radius of the intersection point was realized. The method The new reflected vector direction is combined with the new
works by choosing a starting point for the ray in the CPC inlet point of origin (intersection point) to continue the tracing of the
plane and calculating its trajectory length all the way to the ray either to the exit aperture or the next intersection point.
CPC outlet plane. The length determined is used as the Rays can enter the CPC almost at the maximum radius and
maximum initialization value 𝑡𝑚𝑎𝑥 for the bisection method almost parallel to the surface of the CPC. These rays are
while the origin in the inlet plane is considered to have length 0 subjected to an almost infinite number of reflections. To
and 𝑡𝑚𝑖𝑛 = 0. prevent the code from slowing down drastically in this case an
In each ensuing iteration step the ray segment under exception is added to the code which discards of rays with more
investigation is split in halves. An inequality then checks than a specified number of reflections. For this work rays with
whether the corresponding radius for this new centre position is more than 10 reflections were deleted.
within or outside the CPC by inserting the calculated z value Figure 8 shows a simplified flow chart of the secondary
into the equation and solving the resulting quadratic equation concentrator solver routine.
for the radius 𝑟. The boundaries are then readjusted according
to the result, rejecting half the ray segment. Once a maximum
tolerance threshold is met (1E-8m) between 𝑡𝑚𝑖𝑛 and 𝑡𝑚𝑎𝑥 the
routine declares the midpoint of the last two calculated
boundaries to be the intersection point.
In the next step the surface normal in the intersection point is
determined. This is achieved by calculating the gradient of the
parameterized equations of the 3D CPC.
The individual derivatives of the parameterized equations are:

2 sin(𝜃𝑚𝑎𝑥 ) cos(𝜃𝑚𝑎𝑥 ) 𝑧𝑥
𝑓𝑥 = 2 cos 2 (𝜃𝑚𝑎𝑥 )𝑥 +
𝑟
2𝑟𝑜𝑢𝑡 (1 + sin2 (𝜃𝑚𝑎𝑥 ))𝑥
+
𝑟

5 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


ACQUIRE
MIRVAL MIRVAL.FLX
RAY 12
OUTPUT
DATA

10 800

COORDINATE
TRANSFORMATION
8
600

Month
6
CHECK
Bad coordinates,
400
INLET ERROR!
direction or incident angle
CONDITIONS 4

200
2
CHECK
STRAIGHT TO EXIT?
0 0
0 5 10 15 20 25
Solar time [h]

CALCULATE
Figure 9: Averaged direct normal insolation data for
INTERSECTION MIRVAL use [W/m2]

An acceptance angle variation (Figure 10) showed the potential


CALCULATE
of each secondary concentrator configuration. The rays are split
RAY
REFLECTION
into categories corresponding to their behavior upon
YES, ray exits secondary without reflection concentrator interaction. Rays can be accepted meaning they
NO, ray hits concentrator again successfully exit the concentrator at the exit plane, however
CHECK
they can also be directly declared unsuccessful due to either
Discard ray,
RAY
REJECTED?
rejected missing the aperture, being outside the acceptance angle. In
case of rays that are within the acceptance parameters (angle,
NO, ray is still travelling towards exit entry position) it is still possible for rays to be rejected (exit
through the entry plane) or absorbed in the reflection process.
CHECK
STRAIGHT TO EXIT?
In the latter case a probabilistic approach is used to discard the
preconfigured fraction (absorptivity) of the rays. For each
Ray successfully exits secondary concentrator after reflection reflection process (in case of detected intersection between a
CALL
ray’s trajectory and the concentrator surface) this procedure is
WINDOW
SUBROUTINE
applied before calculating any reflected ray properties. The
green line indicates the decrease of CPC inlet radius with a
growing acceptance angle. The same trend can be observed for
Figure 8: Secondary concentrator solver flow chart the length of the concentrator which decreases significantly.

RESULTS Acceptance angle comparison


100 5
The calculations make use of local sunshape information which
relates the sunrays’ angles upon leaving the sun and their
80
respective power. To each ray a wavelength is attributed upon
creation fitting the wavelength distribution of solar power
Percent of rays

hitting the surface of the earth. Direct normal insolation data 60


Rin [m]

collected at Southwest Solar Research Park in Phoenix Arizona


for the years 2011 and 2012 is used for all calculations. The
40
resulting monthly averaged input data for MIRVAL is shown in
Figure 9. Accepted
20 Missing aperture
Outside acceptance angle
Rejected after reflection
Absorbed after reflection
0 0
0 10 20 30 40 50 60
Acceptance angle [deg]
Figure 10: Secondary concentrator ray fraction behaviour,
March 21, 1200 hours, rout = 0.85m

6 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


The square shaped heliostats were set to have a reflectivity of
Adding the CPC outlet radius as a second variable a detailed 96 percent, a slope error of 0.12 percent and a semi-diagonal of
efficiency map for outlet radius / acceptance angle 4.311m to approximate the heliostats used at NSTTF.
combinations is created (Figure 11).

Acceptance angle comparison


1 90

0.9 80

0.8
RCPCOUT [m]

70

0.7
60

0.6
50

0.5 Figure 12: Slope error calculation, original (yellow) and


40 deflected ray (red). (Wikimedia Commons, modified)

30 40 50 60
Acceptance angle [deg] 22
Yearly average accepted power by heliostat [kW]
Figure 11: Percentage of rays accepted, March 21, 1200 200
hours, rout = 0.85m 20

The influence of using a secondary concentrator can be 150


Width of field (E/W) [m]

18
evaluated by looking at the single heliostat efficiency (Figure
13). The CPC used throughout this paper was defined to use an
outlet radius of 0.85m and an acceptance angle of 46 degrees. 100 16
These two measures fully constrain the ideal 3D CPC with a
resulting concentrator length of 1.96m and an inlet radius of
14
1.18m. This secondary concentrator configuration promises 50

good results while retaining acceptable mechanical dimensions.


In order to account for geometrical imperfections to the ideal Tower 12
shape a surface slope error is introduced (17). Whenever a ray 0

is reflected off the inside surface of the concentrator a spatially 10


distributed error term (normal distribution) is determined in -100 -50 0 50 100
Depth of field (N/S) [m]
dependence of θ and φ (Figure 12) leading to a slightly deviated
reflected ray. Using separate random numbers in the θ and φ 22
Yearly average accepted power by heliostat [kW]
directions allows calculating a 3-dimensional distribution rather 200
efficiently. For each original ray (yellow) a new ray direction 20
(red) is determined. This allows the modelling of imperfections
in secondary concentrator specularity, slope, shape, alignment 150
Width of field (E/W) [m]

2 18
and tracking with the combined error amplitude 𝜎𝑒𝑟𝑟
determined by the following formula (18).
100 16
2 2 2 2 2 2
𝜎𝑒𝑟𝑟 = 𝜎𝑠𝑝𝑒𝑐 + 4𝜎𝑠𝑙𝑜𝑝𝑒 + 4𝜎𝑠ℎ𝑎𝑝𝑒 + 4𝜎𝑎𝑙𝑖𝑔𝑛 + 4𝜎𝑡𝑟𝑎𝑐𝑘
14
50
Using these models the yearly averaged power for a distinct
configuration (heliostat field, window and secondary
12
concentrator dimensions) is calculated by randomly distributing Tower
0
the traced rays for the Monte Carlo calculations not only
throughout the day but at the same time also throughout the 10
-100 -50 0 50 100
year. This feature is implemented directly within the core Depth of field (N/S) [m]
MIRVAL code. Figure 13: All year heliostat average power without CPC
(above) and with CPC (below) (sun-rise to sun-set average)

7 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


The results for a specific concentrator design are very different
The simulations conducted allow the visualization of power in depending on the time of day and year due to the variable solar
the inlet plane (Figure 14), the instantaneous and yearly position in the sky.
averaged power incident and absorbed (Figure 15) by the CPC
reflective surface and the CPC outlet plane ( Figure 16). All the 4
x 10
values are obtained for an aiming point that’s centered on the 1.5
aperture plane. In the case of the application of a secondary
concentrator the aiming plane remains in the same position on 1

North [m]
4
top of the tower (CPC aperture plane) while the receiver is
moved backward by the length of the concentrator. In the real 0.5
3

Up [m]
world this might not always be possible depending on the
length of the secondary concentrator. 0
2
Solar flux in aperture plane [W/m2] 6 -0.5
x 10
1.5
9
-1 1
8 Total Power absorbed = 128.3927 kW
1 -1.5 0
-1 0 1
7
West [m]
0.5 6
Total Power absorbed = 128.3927 kW
5
Up [m]

0
4
4
x 10
-0.5 1.5 3 2

1 2 1.5
North [m] North [m]

4
-1 4
Total Power hitting aperture = 5.9235 MW x 10 1
0.5 1
2
1.5 3
Up [m]

Total Power hitting window = 5.3195 MW 0.5


-1.5 0 0 1.5
-1.5 -1 1 -0.5 0 0.5 1 1.5
4
West [m] 2 0
-0.5 1
0.5
Figure 14: Power distribution hitting the aperture plane,
3
1 2
Up [m]

-1 hours
March 21, 1200 0.5
0 1
2 0
A detailed -1.5investigation
-0.5 -1 into0 the window
1 optical0 and 0
mechanical design and efficiency was conducted at SDSU in 2 1
West [m] 1 -1
separate work-1(19), (7). This work resulted in a spherically 0
curved window design. 1 -2
-1.5 Up [m] -1
0
-1 0 1 0 West [m]
The power distribution in the aperture plane offers a simple 1
way to analyze the potential West of[m]a secondary concentrator Figure 15: CPC absorbed power,
-1 March 21, 1200 hours
0
solution. Figure 14 clearly shows the limited use of the 3D CPC -2
Up [m]surface absorption
The CPC reflective -1
plots can be used to
at noon on March 21. This is down to the almost right angle
between the sun’s position and the aperture plane. The target analyze and optimize the secondary concentrator West [m]
geometry.
plane on top of the tower measures 6m by 6m in order to Figure 15 clearly shows improvement potential in the 3D CPC
capture all the rays that could potentially enter a secondary geometry. One possibility is to save material by truncating the
concentrator. The white circle in Figure 14 corresponds to the CPC and by further eliminating inefficient reflective surfaces.
actual window dimensions. All the rays within this circle This measure can also significantly reduce manufacturing cost
directly hit the window in case the secondary concentrator is of the concentrator due to less reflective surface on the one
left aside. All the power outside this window radius could hand but also saved weight and thus savings in concentrator
potentially be captured by an infinitely large secondary mounting hardware.
concentrator.

8 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


6 6 6 6
Total Power =Total Power
3.5122 MW= 3.5122 xMW
10 x 10 x 10 x Power
10 Power evolution
evolution
4 4
1 1 4 4

0.5 0.5 3.5 3.5


Up [m]
3 3
Up [m]

0 0
2 2
3 3
-0.5 -0.5
1 1
-1 2.5 2.5
-1

Total power [W]


Total power [W]
0 0
-1 -1
0 10 1
West [m] West [m]
2 2
6
Total Power =Total Power
2.8755 MW= 2.8755 xMW
10
6 x 10

1 1 4 1.5
4 1.5

0.5
0.5 3
3 1
Up [m]

1
Up [m]

0
0 2
2
-0.5 0.5
-0.5 0.5 CPC
1 CPC
1 no CPC
-1 no CPC
-1
0 0
-1 0 0 1 0 0 0.2 0.4 0.6 0.8
-1 0 1 0 0.2 0.4 0.6 0.8
West [m] Radius [m]
West [m] Radius [m]
Figure 17: Radial power distribution comparison in the
Figure 16: CPC outlet power distribution, June 21, 1600 cavity entry plane for June 21, 1600 hours
hours, with CPC (top) and without CPC (bottom)
Total incident power = 5319.546 kW
Looking at Figure 16 it is clear the CPC has a big influence on
the flux distribution entering the receiver cavity. The figure 0.5
itself shows the cavity inlet / secondary concentrator outlet
z [m]

0.2

y [m]
plane before the rays hit the window. In addition to the shift of 0 0
the position of the rays in this plane also their directions change
0.5
significantly. The visualization of this change has however been 0.5 -0.5
0
omitted in this paper due to the difficulty of interpretation of 0
-0.5 -0.5
the resulting vector plots. The analysis is possible however and y [m] -0.5 0
x [m]
is being conducted at SDSU. x [m]
Figure 17 shows a different representation of the incident
Total data contained
power = 5319.546 kW 6
within Figure 16 with the resulting radial power distribution x 10
8
across the cavity entry plane being mapped.
The effect of the CPC on the window ray distribution is seen in 0.5 6
z [m]

Figure 19 and Figure 19.


[W/m 2]

0.2
y [m]

0 0 4

0.5 2
0.5 -0.5
0
0
-0.5 -0.5 0
y [m] -0.5 0 0.5
x [m]
x [m]
Figure 18: Window incident power, March 21, 1200 hours,
without secondary concentrator

9 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Hour Total Power Absorbed Peak Power Absorbed
Looking at the window incident power figures it is clear that
the power distribution on the window is changed significantly No CPC CPC % gain No CPC CPC % gain
2 2
by the secondary concentrator. The central peak power spot is [kW] [kW] [kW/m ] [kW/m ]
split into two off-center focal spots. This is due to the rays 08:00 7.993 8.125 1.65 9.677 10.660 10.16
hitting the opposite side of the concentrator and then being
reflected onto the window. 12:00 17.271 15.731 -8.92 29.067 19.383 -33.32
The most notable results regarding the use of a 3D CPC on the 16:00 7.62 7.750 1.71 8.570 10.523 22.79
5MW receiver in terms of absolute power are summed up by Table 2: Comparison of power absorbed by window with
Table 1 and Table 2. and without CPC, March 21

Hour Total Power Transmitted Peak Power Transmitted Of further interest is the decrease in absorbed power within the
No window during peak power situations. The effect is less
CPC CPC % gain No CPC CPC % gain pronounced during off peak periods with the increase in
2 2 transmitted power by around 16% resulting in an increase of
[kW] [kW] [W/m ] [W/m ]
absorbed power of about 2%.
08:00 2356.9 2742.2 16.35 2622.5 4008.5 52.85
12:00 5161.3 5254.4 1.80 7921.1 7044.0 -11.07 CONCLUSION
16:00 2249.5 2614.1 16.21 2466.9 3857.4 The final MCRT secondary concentrator solver and the
56.37
Table 1: Comparison of power transmitted by window with MATLAB post processing power have been employed to
and without CPC, March 21 acquire various SPHER project specific important results. The
influence of this optimized secondary concentrator setup on the
receiver incident power has been evaluated in detail (20).
Total incident power = 5453.049 kW Using single heliostat6projections on the inlet aperture plane of
x 10
the secondary concentrator the receiver tilt angle and the
8
optimal acceptance angle for the secondary concentrator were
0.5 established. Detailed
6 parameter studies were conducted and the
z [m]

[W/m 2]
0.2 results displayed in efficiency maps.
y [m]

0 0 In the same step 4the influence of the secondary concentrator


0.5 setup on the pressurized receiver window has been examined.
-0.5 2
0 0.5 In this process studies for varying time frames during the solar
0 year have been looked into. The final results point towards a
-0.5 -0.5 0
y [m] -0.5 0clear0.5
power gain using the secondary concentrator optimized
x [m]
x [m]
for this experimental power plant. Most importantly the power
wer = 5453.049 kW x 10
6 gains are realized at off peak plant load. This means the
8 maximum power rating for most plant components does not
need to be adjusted. On the other hand higher efficiency plant
0.5 6 operation should be possible due to more steady input power.
[W/m 2]

A further beneficial effect of the secondary concentrator has


y [m]

0 4
been discovered while investigating its effect on the window.
2 While increasing the window incident and transmitted power
0.5 -0.5
during off peak energy production the average power absorbed
0
-0.5 0 by the window was actually found to drop.
-0.5 0 0.5 Summing up, the use of a secondary concentrator for the
x [m]
x [m] SPHER project makes sense from a pure efficiency point of
Figure 19: Window incident power, March 21, 1200 hours, view neglecting any financial or structural considerations.
using CPC
ACKNOWLEDGMENTS
The power transmitted through the receiver window is only
The authors acknowledge the support by the Department of
increased by about 2% during peak load situations. During off
Energy with its SunShot program under the Award # DE-
peak operation, however, significant power gains are
EE0005800. We also are thankful to Cliff Ho at Sandia
demonstrated by the early morning and late afternoon cases on
Laboratory for providing a copy of the MIRVAL code which we
March 21.
modified for our use. Ahmet Murat Mecit (19) (formerly of the
SDSU C&SEL and now at Solar Turbines) provided useful
discussion and assistance about the window ray trace code.

10 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


REFERENCES 14. Segal, A. and Epstein, M. Truncation of the Secondary
Concentrator (CPC) as Means to Cost Effective Beam-Down
1. International Energy Agency IEA. 2014 Key World Energy System. Journal of Solar Energy Engineering. 2010, 132(3).
Statistics. 2014. 15. Cooper, T., et al. Performance of compound parabolic
2. Hunt, A. J. A New Solar Thermal Receiver Utilizing Small concentrators with polygonal apertures. Solar Energy. 2013,
Particles.. Atlanta, GA : Proc. Int. Solar Energy Society Conf., Vol. 95, pp. 308-318.
1979. pp. 1362-1366. 16. Directorate-General for research SOLGATE, solar hybrid
3. Steinfeld, A. and Schubnell, M. Optimum Aperture Size gas turbine electric power system. 2005. EUR 21615, Contract
and Operating Temperature of a Solar Cavity-Receiver. Solar ENK5-CT-2000-00333.
Energy. 1993, Vol. 50, pp. 19-25. 17. Gui-Long, D. Numerical investigation of the solar
4. Fletcher, E. A. and Moen, R. L. Science 197. 1977, p. 1050. concentrating characteristics of 3D CPC and CPC-DC. s.l. :
5. Wikipedia. [Online] [Cited: 03 26, 2015.] Solar Energy 85, 2011. 2833‐2842.
http://en.wikipedia.org/wiki/Monte_Carlo_method. 18. Cooper, T. and Steinfeld, A. Derivation of the Angular
6. Leary, P. L. and Hankins, J. D. A user's guide for MIRVAL Dispersion Error Distribution of Mirror Surfaces for Monte
-- A computer code for comparing designs of heliostat-receiver Carlo Ray-Tracing Applications. ASME Journal of Solar
optics for central receiver solar power plants. Livermore : Energy Engineering. 2011, pp. 044501-1 - 044501-4.
Sandia Laboratories, 1979. 19. Mecit, A. M. Optical Analysis and Modeling of a Window
7. Mecit, A. M. and Miller, F. Optical Analysis of a Window for Small Particle Receiver using the Monte Carlo Ray Trace
for Solar Receivers Using the Monte Carlo Ray Trace Method. Method. San Diego State University : s.n., 2013.
Minneapolis, MN : ASME, 2013. ES-FuelCell2013-18186. 20. Berchtold, O. The Analysis of a Secondary Concentrator
8. Buie, D., Dey, C. J. and Bosi, S. The effective size of the for the Small Particle Heat Exchange Receiver using the Monte
solar cone for solar concentrating systems. Solar Energy 74. Carlo Ray Trace Method. San Diego : s.n., 2014.
2003.
9. Cooper, T. and Steinfeld, A. Derivation of the Angular
Dispersion Error Distribution of Mirror Surfaces for Monte
Carlo Ray-Tracing Applications. 2011, ASME J. Sol. Energy
Eng., Vol. 133, pp. 044501-1 - 044501-4.
10. Welford, W. T. and Winston, R. High Collection
Nonimaging Optics. San Diego, California 92101 : Academic
Press, Inc., 1989.
11. Winston, R., Minano, J. C. and Benitez, P. Nonimaging
optics. Saint Louis, Mo, USA : Elsevier Science & Technology,
2004.
12. De Winter, F. Solar Collectors, Energy Storage and
Materials. s.l. : MIT Press, 1990.
13. Pritzkow, W. The Hexagonal Compound Parabolic
Concentrator Design and Construction. Stuttgart : DLR, 1993.

11 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/31/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Anda mungkin juga menyukai