Download - CLFR Mills
-
8/3/2019 CLFR Mills
1/21
Solar Energy Vol. 68, No. 3, pp. 263283, 20002000 Elsevier Science Ltd
Pergamon P I I : S 0 0 3 8 09 2 X ( 99 )0 0 0 6 8 7 All rights reserved. Printed in Great Britain0038-092X/00 /$ - see front matter
www.elsevier.com/locate/solener
COMPACT LINEAR FRESNEL REFLECTOR SOLAR THERMALPOWERPLANTS
,DAVID R. MILLS* and GRAHAM L. MORRISON***School of Physics, University of Sydney, Sydney 2006, Australia
**School of Mechanical and Manufacturing Engineering, University of New South Wales, New South
Wales 2052, Australia
Received 25 November 1998; revised version accepted 30 August 1999
Communicated by LORIN VANT-HULL
AbstractThis paper evaluates Compact Linear Fresnel Reflector (CLFR) concepts suitable for large scalesolar thermal electricity generation plants. In the CLFR, it is assumed that there will be many parallel linearreceivers elevated on tower structures that are close enough for individual mirror rows to have the option ofdirecting reflected solar radiation to two alternative linear receivers on separate towers. This additional variablein reflector orientation provides the means for much more densely packed arrays. Patterns of alternating mirrorinclination can be set up such that shading and blocking are almost eliminated while ground coverage ismaximised. Preferred designs would also use secondary optics which will reduce tower height requirements.The avoidance of large mirror row spacings and receiver heights is an important cost issue in determining thecost of ground preparation, array substructure cost, tower structure cost, steam line thermal losses, and steamline cost. The improved ability to use the Fresnel approach delivers the traditional benefits of such a system,namely small reflector size, low structural cost, fixed receiver position without moving joints, and non-cylindrical receiver geometry. The modelled array also uses low emittance all-glass evacuated Dewar tubes asthe receiver elements. Alternative versions of the basic CLFR concept that are evaluated include absorberorientation, absorber structure, the use of secondary reflectors adjacent to the absorbers, reflector fieldconfigurations, mirror packing densities, and receiver heights. A necessary requirement in this activity was thedevelopment of specific raytrace and thermal models to simulate the new concepts. 2000 Elsevier ScienceLtd. All rights reserved.
1. INTRODUCTION This paper describes a new design approach
that came from a realisation that trough technolo-The majority of direct solar electricity worldwide
gy was near its design limits and that fundamentalis generated in nine large solar thermal electric
changes to the absorbing surface and collectorplants in California built by Luz International
configuration were needed for large scale im-Limited (LIL), with design and major solar com-
plementation of solar thermal power. New highponents produced by Luz Industries Israel (LII).
temperature selective surfaces with very lowThese solar fields have achieved better than 97%
emissivity for evacuated tube solar absorbers haveavailability over more than 10 years of operation
been developed (Mills, 1991; Zhang and Mills,and have successfully demonstrated the technolo-
1992). Mills and Keepin (1993) suggested that
gy, but in a financial environment requiring these surfaces could be used in new low con-substantial government subsidy. Advanced ver-
centration designs to reduce system costs, andsions of the Luz LS3 collector technology may
increase performance as lowering the absorbingsoon reduce unsubsidised electricity generation
surface emissivity allows greater flexibility incost to about ECU0.085 per kWh (EC, 1998).
geometric concentration as a design variable.However, these costs are still too high for many
Mills and Keepin used polar axis trough collec-markets. Even in markets where there is renew-
tors as the example of a low concentrationable energy obligation legislation, significant cost
collector, but there are several new design con-reductions in solar thermal power systems are still
figurations that could use advanced evacuatedneeded to compete against other renewable energy
tubes. In this paper, we advocate the use of angrid-electricity systems using waste biomass fuel
advanced form of linear Fresnel reflector as aor wind energy.
more cost-effective alternative to parabolic troughor parabolic dish systems. The peak solar energy
radiant flux concentration of such systems canAuthor to whom correspondence should be addressed. Fax:161-2-9351-3577; e-mail: [email protected] range between 50 and 150% of that used by LS2
263
-
8/3/2019 CLFR Mills
2/21
264 D.R. Mills and G.L. Morrison
and LS3 systems (23:1 and 26:1, respectively). and secondary concentrators, both of which allevi-
Higher concentrations are being investigated ate the requirement for a small primary image size
using advanced secondary reflector systems and and very high optical concentration.
will be reported in future publications. A more recent effort to produce a tracking
Linear Fresnel Reflector was made by the Israeli
Paz company in the early 1990s (Feuermann and2. LINEAR FRESNEL REFLECTOR Gordon, 1991; Feuermann, 1993). Although in-
TECHNOLOGYtended for 1508C operation, this technology is the
Geometrically, the ideal reflectors to use with closest in the literature to that proposed here. This
single receivers of solar energy are continuous array exhibited, among others, aberration difficul-
reflectors, usually parabolic for linear axis sys- ties caused by the movement of reflectors around
tems, or paraboloidal for two axis systems. Large an axis parallel to but displaced from the reflector
continuous reflectors or lenses can be simulated optical axis. The analysis approach taken by the
by small elements distributed over a plane thus Paz system developers was to use an optical ray
avoiding the problems associated with very large trace program for a system with finite reflector
reflectors. Baum et al. (1957) discussed large sizes. This is also the method used in the course
two-axis solar tracking systems of this type, but of the current study.
the first to apply this principle in a reasonably One fundamental difficulty with the Linearlarge linear system for solar collection was Fran- Fresnel Reflector (LFR) technology is the avoid-
cia (1968), who developed both linear and two- ance of shading of incoming solar radiation and
axis tracking Fresnel reflector systems. This work blocking of reflected solar radiation by adjacent
showed that elevated temperatures could be reflectors. Shading and blocking can be reduced
reached using such systems. Following this, Riaz by using higher absorber towers, which increases
(1976) developed theory associated with two-axis cost, or by increasing absorber size, which allows
systems, which was soon accompanied by addi- increased spacing between reflectors remote from
tional work by Vant-Hull and Hildebrandt (1976), the absorber. The latter leads to increased ground
Abdel-Monem et al. (1976), Lipps and Vant-Hull usage relative to collector area and also increases
(1977), Lipps and Vant-Hull (1978), Igel and both thermal losses and shading by the absorber.
Hughes (1979) and Dudley and Workhoven(1978, 1979). The work by Riaz can be adapted to
linear systems, and he discusses shadowing ef- 3. COMPACT LINEAR FRESNEL REFLECTOR
fects in a general way. Wei (1980, 1981) discusses (CLFR)
simplified calculations for two-axis systems.3.1. Basic conceptMuch of this work was associated with early
modelling of the US Central Receiver programme Compact Linear Fresnel Reflector (CLFR) tech-
which culminated in Solar One, a 10 MW(e) nology is, in effect, a second type of solution for
two-axis tracking solar power plant constructed in the Fresnel reflector field problem which has been
the early 1980s. However, Di Canio et al. (1979) overlooked until now. The classical linear Fresnel
of the FMC Corporation produced a detailed system has only one linear absorber on a single
project design study for a linear plant of between linear tower, and therefore there is no choice
10 MW(e) and 100 MW(e), with a mirror field on about the direction of orientation of a givenone side of a 1.68-km linear cavity absorber reflector. However, if one assumes that the size of
mounted on 61 m towers. Vant-Hull (1991) sug- the field will be large, as it must be in technology
gests that the increased image size and lowered supplying electricity in the multi-MW class, it is
concentration for ray incident angles not perpen- reasonable to assume that there will be many
dicular to the linear axis would permit no advan- linear absorbers in the system. If they are close
tages over the Central Receiver plants. However, enough, then individual reflectors will have the
the FMC report itself acknowledges these optical option of directing reflected solar radiation to at
shortcomings but says these are compensated by least two absorbers. This additional variable in
lower costs of manufacture and maintenance; the reflector orientation provides the means for much
authors were aware of Central Receiver develop- more densely packed arrays, because patterns of
ment going on in parallel at the time and proposed alternating reflector inclination can be set up suchuse of the same generating system. Since this that closely packed reflectors can be positioned
report was made, substantial advances have been without shading and blocking. The interleaving of
made in the areas of spectrally selective absorbers mirrors between two linear absorber lines is
-
8/3/2019 CLFR Mills
3/21
Compact Linear Fresnel Reflector solar thermal powerplants 265
Fig. 1. Schematic diagram showing interleaving of mirrors without shading between mirrors.
shown in Fig. 1. This arrangement minimises tube. The absorber of the glass evacuated tube
beam blocking between adjacent reflectors and is connected to the central steel pressure tube
allows higher reflector densities and lower ab- by a heat transfer fin. The evacuated tubes
sorber tower heights to be used. Land or roof area exhibit very low radiative losses and arecost is in many cases not a serious issue, but inexpensive, the current cost of a 1.2-m long,
available area can be restricted in industrial or 45-mm diameter evacuated tube is |US$15.
urban situations. Avoidance of large reflector Low array maintenance costs due to ease of
spacing and high towers is an important cost issue access for cleaning, and the capability to
when one considers the cost of ground prepara- remove the single ended evacuated tubes with-
tion, array substructure, tower structure, steam out breaking the heat transfer fluid circuit.
line thermal losses, and steam line cost for This paper investigates alternative versions of
installation next to an existing fossil fuel generat- the CLFR concept to determine which are worthy
ing plant where the objective is the retrofit of a of further development. Areas of study include
low pollution steam source. field orientation, absorber orientation, absorber
The CLFR power plant concept proposed in structure, usage of auxiliary reflectors adjacent tothis paper is intended to reduce costs in all the absorbers, reflector packing density, and tower
elements of the solar array. The following features height.
enhance the cost effectiveness of this system3.2. Horizontal tracking axis arrayscompared with trough technology.
Flat or elastically curved glass reflectors CLFR arrangements can include analogues of
mounted close to the ground are used to horizontal EastWest axis, NorthSouth axis and
minimise structural costs. Costly sagged glass polar axis parabolic troughs. In the latter, the
reflectors are avoided. plane of the CLFR array is inclined toward the
The heat transfer loop is separated from the equator at the latitude angle, and would require an
reflector field and fixed in space thus avoiding inclined support structure or favourable ground
the high cost of flexible high pressure lines or inclination. A scaled layout of a CLFR system
high pressure rotating joints required in trough with a 50-m tower spacing, 10-m high absorberand dish systems. and 48 mirrors, each 1 m wide, is shown in Fig. 2.
The heat transfer fluid is water, and passive The high-density arrangement of reflectors shown
direct boiling heat transfer is proposed to in Fig. 2 is such that the reflectors are separate but
minimise parasitic pumping losses and the would be close to touching if all were tilted to the
need for flow controllers. Steam supply may horizontal. Lower densities of reflectors may be
either be directly into the power plant steam more cost-effective in some cases. The scale of
drum or via a heat exchanger. Steam can also the moving elements is relatively small, even
be supplied in a similar manner for power though the unit scale of the overall system is very
plant preheating cycles. The steam delivery large compared to other linear concentrator con-
conditions considered in this study are 3508C figurations; the pictured array is equivalent to a
and 16 MPa wet steam. parabolic trough of focal length of around 10 m. An absorber composed of a pressure tube In this project most of the investigation was
containing the high pressure water, mounted done on a horizontal EastWest axis array using a
inside an advanced all-glass evacuated Dewar vertical absorber system consisting of a vertical
-
8/3/2019 CLFR Mills
4/21
266 D.R. Mills and G.L. Morrison
Fig. 2. Layout to scale of a CLFR array with 48 mirror rows, 50-m absorber spacing and absorbers 10 m above the primary
reflector field. Scale marks are in metres. Mirrors near each tower are trained on it alone because close packing can be achieved
without blocking, mirrors in the middle of the two absorber rows have alternating directions.
wall of all-glass evacuated tubes illuminated from inclination required. The stationary reflector redi-
both sides, or a horizontal absorber tube array rects rays that have not yet hit the finite length
illuminated from one side. primary mirror but would be intercepted by an
infinite length primary mirror and also collects
3.3. Inclined NorthSouth and polar axis rays that have been reflected from the primaryarrays mirror but have struck the reflecting wall before
A CLFR array can also be inclined toward the striking the finite length absorber. Inclination2
equator to increase winter and annual collection. markedly improves collected energy per m of
A NorthSouth axis array inclined at the latitude reflector for locations outside tropical latitudes.
angle (a polar axis tracking array) will yield close Inclining the array necessitates spacing between
to the optimal annual performance. Performance the inclined reflector arrays to avoid winter shad-2
simulations for different tracking axis orientation ing, and decreases output per m of ground area
and inclination are given in this study. An in- occupied compared to a horizontal NorthSouth
clined array would be similar to that shown in array. Inclined NorthSouth arrays have a flatter
Fig. 3, in which a short NorthSouth array is seasonal output profile compared to horizontal
tilted toward the equator. A stationary reflector at arrays, however, they require a more expensivethe back end of each array is used to reduce the substructure than horizontal arrays. The lower
Fig. 3. Inclined CLFR field with an inverted receiver. The stationary vertical reflector wall improves winter collection.
-
8/3/2019 CLFR Mills
5/21
Compact Linear Fresnel Reflector solar thermal powerplants 267
input in the early morning and late afternoon due evacuated tubes, since rays reflected from the
to the raised artificial horizon for an inclined array glass cover tubes or absorber elements could find
is accounted for in the analysis that follows. their way to other tubes. Primary, secondary and
tertiary reflections were tracked in the absorber. In
this study reflector slope errors were not consid-
4. SOLAR COLLECTOR OPTICAL ered.
MODELLING
4.2. CLFR receiver optical modelling4.1. Raytrace modelling
Two primary receiver types are proposed forA raytrace model was used to generate optical this technology. The first uses all-glass evacuated
collection maps in terms of transverse and longi- absorber tubes in a vertical rack illuminated fromtudinal incidence angles. The concentration maps both sides. The second uses tubes in a single-and beam radiation data were used as inputs to sided horizontal receiver facing downward.thermal modelling routines. A two-dimensional
model was used to generate concentration data for 4.2.1. Optical modelling of vertical evacuated
a series of slices through the array. The two- tube receiver rack. The first configuration consid-
dimensional solutions were then assembled to ered was a receiver with a vertical absorber
generate three-dimensional maps of the optics of protected by evacuated absorber tubes that werethe systems. Absorber angular response and inter- illuminated from both sides as shown in Fig. 4.
actions with the curved surfaces of the glass All-glass evacuated tubes having an outer cover
Dewar tubes was accurately modelled. The three tube diameter of 45 mm and a cover tube thick-
dimensional map of optical concentration and ness of 1.5 mm were evaluated. The absorber
absorption as a function of two orthogonal inci- surface on the outside of the inner glass tube is
dence angles was used in a radiation and thermal 1200 mm long, with a diameter of 37 mm. The
model developed in TRNSYS (Klein et al., 1996). University of Sydney has licensed the basic
The ray tracing model incorporated a branch- selective coating technology to a manufacturer in
ing ray concept for modelling reflections in an China (Turbosun, 1998), and large volume low
array of evacuated absorber tubes. A complex cost tubes having such dimensions are available.
incremental raytrace model was used to establish Due to the use of single ended absorber tubes thethe optimal orientation of each mirror row for a array can only operate as a boiler and thus cannot
given solar incidence angle. A model that in- generate super heated steam.
cluded tracking of a ganged field of mirror rows A potential difficulty with this arrangement is
was also used. The branching ray model was that evacuated spaces between the inner and outer
necessary to gauge the optical absorption ef- glass tubes allow radiation to pass through the
ficiency of arrays of closely spaced all glass absorber rack. Such losses could be significant.
Fig. 4. Vertical Dewar-type absorber tube banks illuminated from both sides. Fluid flow occurs entirely in fixed tubing.
Feedwater is introduced from the header pipe into the branch pipes enclosed by the evacuated glass tube. Boiling occurs in the
branch pipe and we saturated steam leaves via the header.
-
8/3/2019 CLFR Mills
6/21
268 D.R. Mills and G.L. Morrison
absorber surface because much of the available
absorber surface faces other absorber tubes.
The absorptance of such an arrangement varies
with the size of the gap between the tubes. A
raytrace program was developed which follows
primary ray paths through the tube assembly,
together with ray reflections from the glass tubesurfaces, as many reflections are collected on the
second bounce. It was found that the absorptance
of the tube array with 25-mm spacing could be
approximated quite well by a flat plate receiver
covered by a flat glass sheet (Fig. 6). The best
configurations were about 2.5% better than a flat
plate although the flat plate was the best perform-Fig. 5. Double row tube arrangements of branch tubes en-
er at normal incidence. A tube spacing of 25 mmveloped by all-glass evacuated tube absorbers. (a) Close
was chosen for analysis (equivalent to a singlepacked zig-zag absorber array, (b) absorber array with 2.5-cmline spacing of 0.64 diameters), as the hemispheri-gap between tubes.
cal absorptance for this tube spacing is almostidentical to a flat plate absorber (Fig. 7). This
However, because the tubes are inexpensive it choice also allowed use of a flat plate receiver in
was possible to consider a staggered double row subsequent modelling as a convenient approxi-
tube configuration (Fig. 5) with high tube density mation to the evacuated tube receiver. More
and very high optical interception. In a single row detailed analysis may be needed to determine tube
of touching evacuated tubes the gap between the spacing sensitivity, but a spacing of 0.64 diame-
inner and outer tubes amounts to 18% of the face ters is expected to be close to the optimum. In a
area of the absorber. For vertical tube racks a double row vertical absorber system the tube
solution to the gap loss problem is to use a pitch, or interval between tube centres in each
zig-zag double row of evacuated tubes (Fig. 5b). row, is 45 mm1 25 mm5 70 mm. The number of
Such a receiver traps light passing through gaps tubes per lineal metre of receiver using a 25-mmbetween the inner and outer tubes of the vacuum spacing and two rows is 28.6 / m.
tube fitted over the branch tubes. The absorber A second approach to the vertical evacuated
surface consists of both sides of the tube rack, as tube gap loss problem is to use secondary reflec-
in Fig. 4. The absorptance as a function of tube tors behind the tubes. This requires two separated
spacing shown in Fig. 6 is dictated primarily by rows of tubes with individual non-imaging reflec-
the plane aperture (face area) of the receiver, not tors for each tube between the two rows. This will
the circumferential area of the evacuated tube allow an increased tube spacing and a decreased
tube related cost. However, the optical efficiency
will always be less than that of the quasi-vacuum
flat plate receiver due to both increased ray
spillage and absorption in the reflector. The
benefit of increased tube spacing with secondaryreflectors depends upon component costs. Costs
are not presented in this paper, but our estimate is
that the increased optical losses and reflector
absorption would require an increase in plant size
which will be in excess of the evacuated tube and
pressure pipe related cost savings. Similarly, the
40% increase in tube numbers needed for a
smaller tube spacing of 5 mm would cost more
than the 3% performance increase gained. There-
fore, for modelling of vertical receivers, a 25-mm
zig-zag spacing was assumed.Fig. 6. Angular absorptance of zig-zag tube racks withdifferent tube spacing compared to a flat absorber covered by a
4.2.2. Optical modelling of horizontalglass sheet. The acceptance of the zig-zag rack averaged overall angles is almost identical to a flat plate absorber. evacuated tube receiver rack. In horizontal receiv-
-
8/3/2019 CLFR Mills
7/21
Compact Linear Fresnel Reflector solar thermal powerplants 269
Fig. 7. Hemispherical absorptance of double row absorber relative to a flat plate absorber with a glass cover.
2ers, the absorber face area is 1.2 m per lineal A secondary reflector can be used underneath
metre because it is single-sided (facing down- the horizontal absorber to enhance optical collec-
ward). One option is to use the same zig-zag tube tion and increase concentration. A schematic
arrangement as the vertical absorber. This would diagram of one such reflector design is shown in
achieve the same high absorptance but would be a Fig. 8 without an upper backing reflector. This
very expensive arrangement because only one uses a horizontal array of tubes with a backing
side of the rack (the underside) is receiving solar reflector above the tubes to collect rays passingradiation. Alternatively a backing reflector behind between the absorbers. The bifurcated secondary
a single row of tubes could be used. A system reflector system is designed such that most rays
with a tube pitch of 49 mm (to avoid tubes from the primary reflector strike the absorber
touching) requires only 20.4 tubes per lineal directly and do not incur an absorption loss on
metre of absorber and has negligible penalty in this secondary reflector. Only rays on the
optical efficiency. Approximately 1 / 4 of the rays periphery of the reflected beam use the secondary
find their way through the gaps in the worst case reflector.
of normal incidence, but these can be reflected4.3. Field raytraceback with high efficiency, and rays from lower
angles mostly strike the tubes directly. This Optimisation of the mirror field requires con-
absorber configuration would be less costly than sideration of the two possible positions of each
the zig-zag arrangement, but optical performance mirror row for each solar radiation input angle,and heat losses would be similar. corresponding to two absorber targets. The model-
Use of a non-imaging CPC backing reflector ling process involved beginning raytracing with
would also reduce the number of tubes and tube an arbitrary starting configuration for the linear
related costs. However, as this is a more open elements in the mirror field, then flipping the first
pitch tube arrangement with less direct tube mirror in the field and raytracing the whole field
absorption and increased reflector absorption the again. Having chosen the best position for the first
overall gain in cost-effectiveness will be minimal reflector, the second reflector was flipped and the
because the entire array must be enlarged in order best position chosen for it. This process was
to achieve the same output. In this paper we have repeated for the entire field, and for all incidence
evaluated the performance of the horizontal ab- angles. To account for the finite size of the solar
sorber systems using an inverted vacuum insu- source a secondary raytrace between 60.758 waslated flat plate receiver model which would be carried out, which increased the number of rays
similar in performance to the high density, tube by a factor of five. The resulting raytrace compu-
configuration. tations were very large. For a 48-mirror row
-
8/3/2019 CLFR Mills
8/21
270 D.R. Mills and G.L. Morrison
Fig. 8. Secondary reflector for a horizontal tube rack. Rays from the outer edge of the primary Fresnel reflector use the secondary
reflector more because of beam spread.
array, the model included 48 mirrors, two posi- configuration was 84% of the total beam radiationtions, 1000 rays, 19 angles and five beam spread arriving between the linear towers. The 2D inter-
divisions amounting to 9.12 million rays for each ception factor increases with incidence angle
simulation of the field optical characteristics. because gaps between the reflectors are covered
The optical specifications used were:
normal incidence absorptance of the evacuated
tube absorber surface5 0.94;
refractive index of transparent coating on the
mirror5 1.47;
reflectance of mirror surfaces5 0.95;
mirror segment width5 1 m.
In operating systems, reflectivity will be lostdue to dirt deposited on the mirrors between
cleaning operations. The raytrace in this study
was performed on the basis of clean mirrors.
Optimum performance in a CLFR is obtained
by directing each tracking mirror strip to the best
receiver for the time of day. This implies that
each mirror row must have independent tracking.
However, the simplest and cheapest mechanical
arrangement is to have many mirror rows me-
chanically attached to each other and run from a
single motor. Thus it is important to know the
performance penalty associated with mirror rowganging, and this is modelled for some of the
configurations. Having mirrors of identical curva-
ture could lower the array cost, and this was also
investigated.
Fig. 9a and b shows optical concentration maps
for two field configurations. A wide spacing (Fig.
9b) allows each mirror to gather more energy
without blocking or shading, and the output is
flatter throughout the day. A dense mirror con-
figuration (Fig. 9a) approaches the 2D cosine
collection characteristic of a horizontal flat plate Fig. 9. Ray trace map of CLFR solar radiation flux con-collector. The peak interception of beam radiation
centration, (a) absorber 15 m above a 48-mirror array, (b)for solar radiation arriving from a direction per- absorber 10 m above a 24-mirror array with mirror segmentspendicular to the CLFR array plane in the dense spaced by one mirror width.
-
8/3/2019 CLFR Mills
9/21
Compact Linear Fresnel Reflector solar thermal powerplants 271
by adjacent reflectors for low angles of incidence. designed to accept an incidence angle modifier
Peak flux concentration of the configurations map with up to 50 incidence angles in both the
considered without including optical losses (com- longitudinal and transverse planes. The optical
parable to peak geometrical concentration for a map data was generated using the ray tracing
trough) is 35:1 for the 48-mirror array. The routine described in Section 4.1.
geometrical concentration of the Luz LS2 collec- The radiation processor in TRNSYS was used
tor was 23:1 (aperture to absorber tube circumfer- to compute beam radiation from hourly globalence). Higher concentrations can be achieved with radiation records. For analysis in locations where
the Fresnel system, but the concentration is only hourly solar radiation was available the
limited by the heat flux capacity of the Dewar Reindl et al. (1990) radiation model was used to
tubes. compute beam radiation in terms of the clearness
index and the solar elevation. Radiation on the
tracking surface was calculated using the Hay and5. THERMAL MODELLING OF CFLR
Davies (1980) model which accounts for both
Solar radiation and thermal simulation models circumsolar and non-isotropic diffuse radiation
of the collectors were developed in the TRNSYS using an anisotropy index to quantify the portion
modelling environment (Klein et al., 1996). For of diffuse radiation considered as isotropic. Radia-
this project a series of extensions were developed tion data for various sites modelled is shown inwithin TRNSYS to simulate the linear Fresnel Fig. 10. For Dubbo, 1-min time step beam
concentrating collector. Due to the modular nature radiation measurements were available. The mea-
of TRNSYS these new routines were integrated sured data were converted to normal incidence
with the existing data handling and solar radiation beam radiation for the horizontal or inclined
analysis routines that are built into TRNSYS. planes of the Fresnel mirror field being consid-
The primary routines that were used to simulate ered. The annual solar radiation at each of the
concentrating solar collector performance were as design sites is shown in Table 1.
follows.
Radiation processor, nonisotropic radiation 5.1. Heat loss from an array of evacuated tubesdistribution model (TRNSYS TYPE16).
The absorber of the proposed linear Fresnel
Extended optical map-based solar collector collector consists of a rack of evacuated tubemodel for Fresnel concentrator (Morrison,absorbers mounted in two rows so that the
1997).absorber is equivalent to an evacuated flat plate
Nonlinear heat loss solar collector model forabsorber. The heat loss from the evacuated tube
evacuated tube absorber (Morrison, 1997).rack was determined from the measured charac-
The collector thermal mass was modelled interistics of single evacuated tubes (Harding et al.,
TRNSYS using an instantaneous collector ef-1985). The heat loss modes are as follows.
ficiency model coupled to a zero heat-loss Conduction through the insulated header.
storage-tank (TYPE4 tank). This procedure used Conduction through the glass envelope at the
the proven TRNSYS tank routine and solver toopen end of the tube.
include the effect of thermal capacitance, rather Conduction through the metal retainer near the
than developing a complex collector model with
closed end of the absorber tube.built in capacitance. This model follows the start Radiation from the absorber.
up and shut down transients at the beginning andThe heat loss (Q ) from a single tube can be
lend of each day and transient temperature effectsexpressed as
during cloudy periods. The transient effects, due
to thermal capacitance within the absorber, were Q 5 k (T 2T )1 k (T 2T )l 1 m a 2 s a
found to reduce the annual output of the collector4 4
1 k (T 2T ) W/ tube. (1)3 s aarray by 3 to 6% depending on the array con-
centration.For a single tube of the Sydney UniversityTo model the CLFR performance the TRNSYS
design (1.4 m long), Harding et al. (1985) havecollector routine was modified to include a spe-2
shown that k 50.26 W/(K m ), k 50.039 W/(Kcification of optical concentration through a biaxi- 1 229 4
al incidence angle modifier map (see Fig. 9). This tube) and k 58.5310 W/(K tube) where k is3 1was implemented via an extension of the optical the header heat loss factor, k is the conduction2mode 4 option in the extended TRNSYS TYPE1 heat loss factor for conduction heat loss from the
solar collector model. The new routine was absorbing surface to the top of the tube and
-
8/3/2019 CLFR Mills
10/21
272 D.R. Mills and G.L. Morrison
Fig. 10. Beam radiation at Australian design sites.
9through the retainer clip, k is the radiation heat The header heat loss coefficient k for a close3 1loss factor, is the absorber emissivity 5 a1 packed tube rack is
bT where a and b are coefficients determineds 2pkfrom measurement of surface properties (Fig. 11),]]
]9k 51 d2T is the absorber surface temperature, and T iss m]ln S Ddthe mean fluid temperature in the evacuated tube. 1
The tube array proposed for the absorber of the5 0.343 W/ (K m run of header) (2)
CLFR presents a continuous outer face to the
surroundings, thus the radiation heat loss is where k is the header insulation conductivityequivalent to that of a flat plate evacuated collec- (0.06 W/m K), d , d are the inner and outer1 2tor. The radiation heat loss per unit face area of diameters of the insulation (100-mm steam tubethe close packed array will be the same as the with 100 mm insulation).radiation heat loss per unit circumferential area of The heat loss due to conduction around the topa single evacuated tube as measured by Harding of the tube and through the retainer clip is givenet al. (1985). The header heat loss per tube will bybe reduced due to a closer tube spacing than
considered by Harding. Q 5 k (T 2T ) W/ tube. (3)C 2 s a
Table 1. Annual solar irradiation data for Australian design sites2
Location Latitude Climate Annual irradiation MJ /(m day)
Global Diffuse Total at Beamhorizontal latitude angle
Longreach 238 S Dry desert 21.8 7.6 23.1 22.2Dubbo (1994) 328 S Dry inland 19.4 6.5 22.6 24.6
Sydney 348 S Temperate coastal 16.9 7.0 18.6 16.3Wagga 358 S Temperate 17.7 7.1 19.3 17.2
inland
-
8/3/2019 CLFR Mills
11/21
Compact Linear Fresnel Reflector solar thermal powerplants 273
Fig. 11. Emissivity of selective surfaces.
The conduction heat loss per unit face area of double sided zig-zag evacuated tube rack (28.6
the rack is tubes / m) is
2 Q5 0.143 T 2T 1 0.93 T 2T 1 6.44s d s dQ 5Nk (T 2T ) /L W/ m (4) m a s aC 2 s a28 4 4 2
310
T2
T W/ m . (6)s ds awhere N is the number of tubes per metre lengthof rack (28.6/m for the vertical tube configuration
The selective surfaces investigated in this pro-and 20.6 for horizontal tube configuration in this
ject are two formulations of a stainless steel/study), L is the length of tubes51.2 m active
aluminium nitride cermet with a copper reflectorlength. The radiation from a single tube is given
layer (SS/Cu) and two formulations of a stainless4 4by k T 2T W/tube.s d3 s a steel/aluminium nitride cermet with a2
The surface area of a single tube is 0.132 mmolybdenum reflector (SS/Mo). The temperature
hence the radiation heat loss per unit surface areadependence of the emittance of the four selective
of a single tube or per unit face area of a tubesurfaces being considered for this application is
rack isshown in Fig. 11. The trade off between high
4 4 absorptance and low emittance is considered inQr5 k T 2T /0.132s d3 s a the analysis.
28 4 4 25 6.443 10 T 2T W/ m (5)s ds a The heat loss per unit face area from the
evacuated tube rack absorber for the CLFR sys-
The overall heat loss per unit face area of a tem is shown in Table 2 for a range of absorber
Table 2. Heat loss per unit face area of evacuated tube rack using stainless steelcopper selective surface (a50.93)2
Absorber surface Heat loss W/ mtemperature Header Tube conduction Tube radiation Total
8C
100 10 64 32 105200 22 143 146 312
300 34 223 423 681400 47 303 988 1337500 59 383 2020 2462
-
8/3/2019 CLFR Mills
12/21
274 D.R. Mills and G.L. Morrison
Table 3. Heat loss per unit mirror area for alternative CLFRsystems
Absorber surface Heat loss per unit area of mirror2
temperature W/ m8C CLFR 24 CLFR 36 CLFR 48
100 10.5 7.0 5.3200 31.2 20.8 15.6300 68.1 45.4 34.0400 133 89.2 66.9500 246 164.1 123
temperatures. The heat loss from the absorber perFig. 13. Cross section through Dewar-type evacuated tube andunit mirror area for three CLFR reflector packingpressure pipe, dimensions in mm.
densities is shown in Table 3. CLFR24 refers to a
CLFR system using 24 mirror rows in the 50-m
space between two towers, CLFR36, 36 mirrors in Schmid et al. (1990) have shown that a large
the same space, etc. temperature difference will occur between the
absorber surface and the fluid in the pressure tube
unless a fin system is used in the clearance space6. ABSORBER CONFIGURATIONSbetween the evacuated tube and the pressure tube.
The absorbers in the CLFR systems are single-6.1. Heat transfer in absorberended evacuated absorber tubes mounted horizon-
tally (Figs. 12 and 13) or vertically (Fig. 4). The The CLFR systems considered in this study
steel pressure tube inside the evacuated tube can have concentrations (mirror area / face area of the2
be either a single-ended tube or a flow through absorber surface) up to 20:1. For 900 W/m beam
U-tube system. A single ended system will be intensity the incident radiation flux on the ab-2
better at the radiation flux levels typical of this sorber will be 18 kW/m . The tube density
absorber system because a U-tube of sufficient adopted for the CLFR system is 28.6 tubes / m
diameter to provide adequate feedwater flow with each tube having 1.2 m exposed length hence
could not be easily fitted inside the inner glass the maximum heat transfer will be 755 W/tube.tube. The absorber surface of the Dewar-type This maximum heat transfer will occur only when
evacuated tube is on the vacuum side of the inner the sun is directly overhead of the mirror array.
glass tube. The absorbed energy must be con- To minimise heat loss the temperature drop
ducted through the 1.5-mm wall of the inner between the absorber surface and the inner pres-
borosilicate glass tube, then across the clearance sure tube must be minimised.
gap between the glass tube and the pressure tube Thermal resistance between the absorber sur-
and then through the pressure tube wall, Fig. 13. face and the water / steam working fluid is due to
The pressure tube is supplied with feed water conduction through the absorber glass wall;
from the header and returns steam through the heat transfer across the gap between the ab-
same opening. To obtain separation between the sorber tube and the pressure tube (via an
liquid and gas streams a slight inclination from internal fin);
the vertical or horizontal is required. The flow conduction through the pressure tube wall;configuration has been successfully demonstrated convection into the boiling water in the pres-
in a number of prototype systems (Mills, 1991). sure tubes.
Fig. 12. Transverse cross section through absorber rack, the evacuated absorber tube can be vertical or near horizontal and can be
easily slipped off the boiler tubes.
-
8/3/2019 CLFR Mills
13/21
Compact Linear Fresnel Reflector solar thermal powerplants 275
The temperature drop across the absorber wall temperature drop across the pressure tube wall is
DT is given by DT 51.3 K.1 3The convective heat transfer coefficient inside
dQ o the pressure tube during pool boiling will be very]] ]DT 5 ln (7)S D1 2pk L d high and the temperature drop will be small.g i
Fouling of the inner surface of pressure tube
where Q is the heat transfer 755 W/tube maxi- should not significantly add to the overall thermalmum; d is the outer diameter of inner tube ofo resistance between the absorber surface and thevacuum envelope537 mm; d is the inner diam-i steam.eter of inner tube of vacuum envelope534 mm; L The fin system will have additional thermalis the length of vacuum tube51.2 m; k is theg resistance due to the contact resistance betweenborosilicate glass conductivity51.1 W/m K. The the glass tube and the fin, and between themaximum temperature drop across the glass ab- pressure tube and the fin. The outer contactsorber tube is DT 57.7 K.1 between the glass and the fin has not been a
Heat transfer between the glass absorber tube problem in low temperature tubes using a similarand the pressure tube wall is via conduction type of fin. The contact resistance with thethrough the fin system and convection and radia- pressure tube could be minimised by using a fin
tion through the gap. The fin system could consist system consisting of a web sandwich so that thereof a circumferential plate and radial elements as is full circumferential contact over both the glassshown in Fig. 13. The fin system is a combination absorber and the pressure tube. This fin systemof two split cylindrical sections 0.5 mm thick with would be inserted into the evacuated tube andradial fins 0.2 mm thick and 4.5 mm long between twisted as it is slipped over the pressure tube. Thethe two cylinders. A finite element analysis of contact resistance can also be reduced by increas-conduction in the outer ring and through the radial ing the number of fins beyond the 16 fins atfin has indicated that it is equivalent to a simple 11.258 pitch considered in this analysis. At the
20.2-mm-thick fin that is 7.5 mm long. If the radial maximum beam intensity of 900 W/m when thefins were formed from 0.2-mm copper on an beam is normal to the array the temperature drop11.258 pitch the temperature drop (DT ) across the2 across the absorber tube is 13.2 K1the contactfin system (assuming all heat transfer is through
temperature drop.the fins as radiation and convection through the In the following analysis an overall temperaturegap will be minimal) is given by drop of 20 K has been assumed for beam radia-
2tion input of 900 W/m . The development of the
Qabsorber tube will require assessment of a number]]DT 5 l (8)2 NLt kf f of configurations to determine the most effective
finning and contact arrangement.where Q is the heat transfer per tube5755 W; tfis the fin thickness50.2 mm; k is the finfconductivity5350 W/m K (copper at 3508C); N 7. COLLECTOR PERFORMANCE ESTIMATESis the number of fins around the circumference5
In this section, array performance simulation is16; l is the equivalent length of fins between theused to select optimum CLFR configurations. The
glass wall and the pressure tube5
7.5 mm; L is basic selection calculations were carried out forthe width of fins5length of vacuum tube51.2 m.Sydney, Australia, but it was found that per-For a copper fin system the temperature drop isformance relativities are maintained at all sites.
DT 54.2 K.2This is because all of the sites in this study haveThe temperature drop across the pressure tubelow circumsolar radiation and therefore plants inwall is given bythese sites will operate similarly to a plant in
Sydney. Because of the low circumsolar radiationdQ o]] ]DT 5 ln (9)S D3 in Australia, it is sensible to optimise performance2pk L dw i
on the basis of a direct beam capture half angle of
0.758. Diffuse and circumsolar radiation outsidewhere Q is the heat transfer per tube5755 W; dothis range is ignored in the simulation, but there isis the outer diameter of pressure tube525.4 mm;
little energy in this angular range in Australia. Ind is the inner diameter of pressure tube519 mm;ipractice there will be beam spread due to mirrork is the conductivity of pressure tube520 W/mwirregularities and mirror tilt error and slightK (chrome steel at 3508C); L is the length ofdifferences in diffuse collection between differentpressure tube in the evacuated tube51.2 m. The
-
8/3/2019 CLFR Mills
14/21
276 D.R. Mills and G.L. Morrison
absorber configurations but these factors have between the improved optical collection and
been ignored in this initial investigation. In addi- increased thermal losses for a larger receiver
tion, beam spread for rays with a significant surface. The receiver size that delivered maximum
directional component parallel to the array linear energy collection was evaluated for a 50-m wide
axis has not been included, but is not serious for a array using 36 mirror rows. Annual delivered
concentrator of this receiver aperture with mini- energy for different absorber tube lengths is
mal aiming error. However, in some climates, shown in Fig. 14 for both horizontal and verticalnotably tropical regions and parts of the Northern absorbers. The operating temperature in each case
Hemisphere, beam radiation is more forward was 3208C. In neither case does the energy
scattered than in Australia and the amount of collection vary significantly with absorber size.
circumsolar radiation is greater. For such loca- The size of the horizontal absorber was found to
tions the acceptance angle of the collector must be optimal at 1.2 m tube length, and the vertical
therefore be larger to collect both the direct and absorber at 1.0 m tube length. The available
near circumsolar components. It is possible to evacuated tubes are 1.2 m long but an absorber
alter the design to enlarge the absorber for this 1.0 m long can be constructed by angling the
purpose. This is not done in this paper, as it leads tubes in the rack.
to a relatively greater capital cost and increased
7.1.2. Secondary reflector. Secondary reflec-thermal loss. tors can be used near the receiver to capture
reflected solar radiation that would otherwise have7.1. CLFR configuration assumptionsmissed the receiver. The ends of the secondary
A 50-m wide array is assumed with mirror rowreflector reposition the edge of the receiver aper-
densities of 24, 36 and 48 rows per absorber line.ture to better face the primary reflector array. One
Each mirror was taken as 1 m wide with 0.95 mconfiguration of a horizontal absorber (1.2 m
reflecting width and 25 mm edge structure. Eachwide, 12.5 m absorber height, 36 mirror rows)
configuration was evaluated for absorber heightswas chosen to test optimal secondary reflector
of 10 m, 12.5 m, and 15 m. An equal spacing oflength. After the ray trace, an annual performance
mirror rows is assumed and a space 1 m wide iscalculation was performed for each secondary
left clear on either side of the array centre linereflector length all using the same receiver con-
under the absorber for access and maintenance. figuration. The optimal secondary reflector isBecause the effective aperture does not change
similar to that shown in Fig. 8, and is quite short.as a simple cosine function as in a parabolic
The gap between the secondary reflector andtrough (because gaps between reflectors are
absorber allows rays to strike the absorber direct-blocked by adjacent mirrors at high angles of
ly, decreasing secondary reflector absorption loss-incidence), a nominal peak effective aperture was
es. A similar calculation was carried out for aselected. The peak effective apertures of the
vertical absorber with a short circular (orarrays were calculated on the basis of the maxi-
parabolic) section secondary reflector as shown inmum energy that could be collected by an ab-
Fig. 15. The optimal secondary reflector lengthssorber which picks up all beam ray spillage, and
depend on absorber height. However, for sub-is 87.5% of the curved mirror surface area: for 48
m of reflector, this is 42 m of nominal peakaperture in a space of 50 m. This presentation is
analogous to a parabolic trough, where the peak
aperture is used as a parameter for collector
attributes such as cost and energy collection.
There are several options for mirror construc-
tion and mounting. For the purposes of this study
the mirrors are assumed to be a self-supporting
laminated structure using silvered microsheet
glass as the reflector. The auxiliary reflectors2
adjacent to the absorber are assumed to be 2 m
per lineal metre of the absorber.
7.1.1. Optimal absorber length. One important
issue is how large to make the absorber relative to Fig. 14. Useful delivered thermal energy as a function ofthe array field dimensions. There exists a trade off absorber length for a 50-m-wide array.
-
8/3/2019 CLFR Mills
15/21
Compact Linear Fresnel Reflector solar thermal powerplants 277
Fig. 15. Vertical evacuated tube receiver using secondary reflector above the absorber rack.
top secondary reflector effectively angles thesequent analysis the secondary reflector wasreceiver aperture toward the array, reducing rayscaled with absorber size.spillage at the image boundary. Mirror rows close
7.2. Performance optimisation for Sydney to the absorber line deliver a smaller image and
would be aimed at the receiver itself, avoidingHaving sized the secondary reflectors and thesecondary reflector losses. Mirror rows under-absorber length, it is now possible to determineneath the absorber would use the secondarythe optimum absorber and mirror field configura-reflector.tions. NorthSouth axis and polar axis versions
Table 4 shows the annual thermal energywere used as the basic configurations. The per-
delivery. As tower height increases, performanceformance of EastWest systems was also investi-also increases due to reduced mirror row shading.gated as a variation on the basic design. TheHaving fewer field mirror rows allows moreprimary cases calculated are described in Sectionscollection by each row, but reduces energy col-7.2.17.2.4 using Sydney solar radiation data. Thelected by the fixed size absorber array. This is theselective surface of the tubes was taken as stain-original configuration proposed, but it is out-less/copper with normal incidence absorptance of
0.93. performed both by horizontal primary reflector
fields having horizontally oriented absorbers as7.2.1. Vertical evacuated tube absorber with shown in Fig. 8, and polar inclined fields, Fig. 3.
overhead secondary reflector and horizontal
NorthSouth primary reflector field. This configu- 7.2.2. Vertical evacuated tube absorber with
ration was found to perform better than a vertical overhead secondary reflector and polar axis
absorber without a secondary reflector, Fig. 4. The tracking NorthSouth primary reflector field. Thissecondary reflector is circular (or parabolic, they is a version of the vertical absorber array placed
are almost indistinguishable) in curvature as upon a NorthSouth structure inclined at the
shown in Fig. 15 and is sized according to results latitude angle to improve winter performance
of a large number of annual simulation runs (Figs. 3 and 16). If an end reflector is used at the
determining optimal solar radiation capture. The upper end of each segment of the array the system
Table 4. Performance of vertical evacuated tube absorber with overhead secondary reflector and horizontal NorthSouth primaryreflector field, output per unit area of mirror
Number of mirror rows Thermal energy delivery2
in the primary array MJ /(m day)
10 m tower 12.5 m tower 15 m tower
24 5.54 5.61 5.5736 5.31 5.32 5.3248 4.64 4.80 4.91
-
8/3/2019 CLFR Mills
16/21
278 D.R. Mills and G.L. Morrison
Fig. 16. Polar axis CLFR array.
would closely approximate the performance of a This configuration delivers the highest collection
segment of an infinite array. Table 5 shows that per unit aperture area (Table 7) of all the CLFR
the annual thermal energy delivery of this system configurations considered and is equivalent to
is substantially better than for the horizontal about 70% of a two-axis tracking paraboloidal
reflector configuration (Section 7.2.1). dish.
7.2.3. Horizontal evacuated tube absorber with7.3. Design variationsunderneath secondary reflector and horizontal
NorthSouth primary reflector field. This arrange- There are a number of design variations thatment yields substantially improved performance have the capability of improving performance or(Table 6) compared to the horizontal mirror field lowering cost. In the following, each variation iswith a vertical absorber, since fewer rays are evaluated for Sydney.missed due to a larger total receiver aperture. The
tubes would be tilted slightly off horizontal to 7.3.1. Use of more sophisticated secondary
ensure natural circulation of feedwater into the reflector design to decrease absorber size. The
tubes. configuration shown in Fig. 17 uses a short
horizontal absorber and an upper secondary semi-7.2.4. Horizontal evacuated tube absorber with circular reflector to provide a horizontal aperture
underneath secondary reflector and polar axis for a bifurcated lower secondary reflector. In this
tracking NorthSouth primary reflector field. This case, the absorber rack width is reduced to half
is a polar version of the NS horizontal evacuated the size of that for the previously describedtube absorber array placed upon a structure in- horizontal case. Losses are similar, because loss
clined at the latitude angle facing the equator. can take place from both sides of the receiver, but
Table 5. Vertical evacuated tube absorber with overhead secondary reflector and polar axis primary reflector field
Number of mirror rows Thermal energy delivery2
in primary array MJ /(m day)
10 m tower 12.5 m tower 15 m tower
24 7.07 7.16 7.1136 6.82 6.83 6.8848 5.99 6.22 6.40
Table 6. Performance of horizontal evacuated tube absorber with secondary reflector and horizontal NorthSouth primaryreflector field
Number of mirror rows Thermal energy delivery2
in primary array MJ /(m day)
10 m tower 12.5 m tower 15 m tower
24 6.56 7.06 7.3536 6.15 6.65 6.9648 5.22 5.86 6.23
Table 7. Performance of horizontal evacuated tube absorber with secondary reflector and polar axis tracking primary reflectorfield
Number of mirror rows Thermal energy delivery2
in primary array MJ /(m day)10 m tower 12.5 m tower 15 m tower
24 8.26 8.82 9.1636 7.76 8.36 8.7148 6.59 7.41 7.88
-
8/3/2019 CLFR Mills
17/21
Compact Linear Fresnel Reflector solar thermal powerplants 279
from the absorber and it is these mirror elements
that produce most of the beam spread at the
absorber. The envelope of rays diverge from the
mirror because of the finite angular size of the
solar disk. If the mirror is moved closer to the
absorber, fewer rays are spilled even if the focal
length of the mirror is incorrect. This means that acurvature correct for the outer mirrors should also
work for mirrors closer to the absorber and
explains why the results for beam collection for
the constant curvature are almost the same as for
variable curvature.
The question arises as to whether flat mirrorsFig. 17. Schematic of an upper and lower secondary reflectorarrangement for a horizontal tube rack. can be used for the field to further simplify
production and cleaning. Because the mirror has a
the absorber cost is halved. However, optical physical width of 0.95 m, the half image of the
sun at the absorber will fall outside 0.95 by 0.738,efficiency is reduced because of increased optical
or by about 0.32 m for a mirror at the centre oflosses in the dual secondary reflectors, which nowthe field. This means the total image extent is 0.95intercepts a greater fraction of incoming rays. Them1(230.32 m)51.59 m, about 0.39 m greatermain disadvantages of this approach are the verythan the absorber aperture size. The situation islarge unit size of the collector field 100-mworse for mirrors far from the receiver. Therefore,wide fields would be required for a 1.2-m longin spite of the small curvature, it is necessary totube and the high solar radiation flux on eachcurve the mirrors to capture all of the beamtube. There are several methods for using reflec-radiation, or else increase the absorber size by thetors in this way, however, at this stage, noorder of 50% if flat mirrors are to be used.definitive avenue for improving cost-effectiveness
The performance of horizontal and polar mirrorhas been identified.fields using optimised fixed curvature and flat
7.3.2. Use of a larger receiver to increasemirrors are compared to those of variable curva-acceptance angle. A simulation was performed on ture in Table 8. The optimised constant curvature
the standard array with the absorber doubled in performs only 0.50.6% lower than the variablesize. The net result was a 3% decrease in de- curvature, but the flat mirror is 13% lower. The
livered energy due to increased thermal losses net result is that constant curvature elastically
formed mirrors will be used because of highfrom the larger absorber. However, increasing theperformance and simplicity in production. Largeabsorber size also increases cost. Hence it isconcentrating systems typically require mirrorimportant to find the smallest receiver assemblyelements with different curvature in differentsize suitable for the radiation conditions applyingsections of the mirror field. In the proposedat the location of interest. Adoption of increasedconstant curvature system, all reflector elementsabsorber size may allow this technology to beare identical, and moulding or sagging of glass isused in climates with a high circumsolar fraction,not required. This represents a very low cost andas it increases the acceptance angle at a minorpractical option. Due to the very slight curvaturecost in shading and reflector absorption. Note thatrequired (|30 m radius) the mirrors will be asin place of varying the receiver size, one mayeasy to clean as flat mirrors.simply vary the field size for a given absorber
width for different locations.7.3.4. Ganging of mirror rows. The mirror
rows in the primary array can be constrained since7.3.3. Use of a standard mirror curvature. The
curvatures required in the field mirrors are small
but important in their effect on performance. AnTable 8. Average daily delivered energy for different mirror
optimisation calculation was carried out using the shapes15-m tower and 48 mirror row configuration Mirror shape Delivered energy
2
modified for use with constant curvature mirrors MJ/(m day)in all sections of the array. Table shows results for Horizontal array Polar arraya constant focal length of 30 m for a 50-m wide Variable curvature 6.23 7.88
Constant curvature 6.20 7.84array. A 30-m focal length was selected since theFlat 5.40 6.85outer field mirrors are approximately this distance
-
8/3/2019 CLFR Mills
18/21
280 D.R. Mills and G.L. Morrison
Table 9. Annual energy delivery of alternative tracking sys- Table 10. Annual delivered energy for different evacuatedtems (vertical absorber) tube selective surfaces, 24 mirror CLFR system in Sydney
Tracking mode Delivered energy Selective surface Annual energy delivery2 2
MJ /(m day) MJ /(m day)
Horizontal array Polar array SS / Cu a50.93 7.35SS/Cuha50.95 7.30
Row tracked 6.20 7.84SS/Mo a50.93 7.04
Optimal ganged 6.18 7.83SS/Mo a50.95 7.07
they all move together through the same tracking treatments is shown in Table 10 for Sydney
angle, even though the absolute angle of each conditions. Annual performance is shown for the
mirror row is different from the others. The four new selective surfaces. The new high tem-
advantage of this is that the cost of the tracking perature selective surfaces developed by Sydney
system may be reduced. The disadvantage is that University all have significantly better perform-
performance is also reduced due to shading ance than the cermet surface used in the Luz LS3
between some of the fixed mirror lines, because SEGS plant (earlier SEGS systems used chrome
the optimal allocation of mirror rows to the black selective surfaces). The stainless steel and
alternative towers changes throughout the year. copper surface with a slightly lower absorptance
Thus, there is a cost / performance trade off that delivered more energy than the surface with thehas to be assessed. higher absorptance. This shows that a small
A ganged mirror field configuration was opti- decrease in absorptance can be traded off for a
mised for annual collection and compared against significant decrease in emissivity. For the stainless
the standard configuration with mirrors switching steel molybdenum surface the loss of performance
between absorbers to minimise shading. The due to a lower absorptance was not compensated
annual energy delivery of the ganged field is only by a matching heat loss reduction due to lower
0.2% less than the independently row-tracked case emissivity, but the performance of the two formu-
(Table 9). The mirror arrangement of the ganged lations is very close. For a higher concentration
configuration used was approximately the optimal system with 36 or 48 mirrors per tower, higher
arrangement of the unganged configuration at absorptance will be preferred.
equinox (in unganged fields some mirrors change 7.4. Performance for different climaticabsorbers to optimise performance on a seasonalconditionsbasis). The mirror arrangement was fine-tuned by
trial and error with each change being evaluated The performance of the horizontal absorber
by an annual performance calculation. Although a with a 15-m tower and 24- or 48-mirror array in
ganged field might lead to lower capital cost, a both horizontal field and polar field configurations
non-ganged configuration has practical advan- was evaluated for a range of climatic conditions
tages, since focusing can be finely tuned, and all in Australia (Table 11). The sites ranged from
mirrors can be aligned vertically in hailstorms, or cloudy coastal conditions (Sydney), dry and clear
horizontally in high winds. Independently tracked inland sites ( Dubbo) and latitudes ranging from
mirror lines can also be aligned or inverted for the dry tropics to 358 S. Surprisingly good
cleaning. During absorber maintenance, arbitrary performance was obtained for Dubbo, however it
sections of the mirror array can be realigned to must be noted that the weather data for Dubbo areother absorbers, maintaining output, and individ- based on one particular year (1994) of measured
ual rows can be aligned vertically to provide 1-min data. The performance for other locations is
walk-through paths. A single control system could based on 1-h time step typical meteorological year
control many hundred drive motors of this slow- weather data, which gives a good estimate of the
moving tracking system. The issue of ganging of long term performance. Thus, Table 11 shows the
mirrors must therefore rest as a minor issue that performance during 1994 in Dubbo and the long
needs to be resolved during detailed equipment term average performance for the other locations.
and operation design. In the remainder of this The variation of annual output in Dubbo as a
study it is assumed that the unganged system is function of array slope for a NorthSouth align-
used. ment is shown in Fig. 18. The effect of the
shortening of the solar day on an inclined receiver7.3.5. Effect of selective surface properties. was accounted for in the system analysis. A polar
The performance of the 24-mirror CLFR system axis array has 22% higher output than a horizontal
with different evacuated tube selective surface array. It can be seen that the performance is not
-
8/3/2019 CLFR Mills
19/21
Compact Linear Fresnel Reflector solar thermal powerplants 281
Table 11. Annual average delivered thermal energy for different locations
Location Latitude Annual average delivered energy2
MJ/(m day)
Horizontal Polar Horizontal Polar24 mirrors 24 mirrors 48 mirrors 48 mirrors
Longreach 238 S 11.3 12.6 9.49 10.7Dubbo 328 S 11.4 14.1 9.49 11.9Sydney 348 S 7.35 9.16 6.23 7.88Wagga 358 S 7.95 9.49 6.70 8.12
therefore conceivable that both array types may
find geographical niches.
8. CONCLUSIONS
This paper has evaluated Compact Linear Fres-
nel Reflector (CLFR) concepts suitable for large
scale solar thermal electricity generation plants,and recommends the concepts most suitable to
pursue. In the CLFR, it is assumed that there will
be many parallel linear receivers that are closeFig. 18. Annual energy delivery as a function of array slope to enough for individual mirror rows to have thethe North, for Dubbo, latitude 328 S.
option of directing reflected solar radiation to two
linear receivers on separate towers. This addition-
al degree of freedom in mirror orientation can
allow closely packed mirror rows to be positionedsensitive to inclination within 5 degrees of theso that shading and blocking are almost elimi-latitude.nated.
7.5. Seasonal variation of CLFR performance
The avoidance of large mirror spacings andtower heights is an important issue in determiningPolar and horizontal arrays can have markedlythe cost of ground preparation, array substructuredifferent seasonal performance variation, with thecost, tower structure cost, steam line thermalpolar type having much better winter performancelosses, and steam line cost. The improved abilityat high latitudes. The improved winter perform-to use the Fresnel approach still delivers theance of polar arrays at high latitudes (Figs. 19 andnormal benefits of such a system, namely small20) may ensure its adoption, but for dry semi-reflector size, low structural cost, fixed receivertropical locations such as Longreach (Fig. 21), theposition without moving joints, and non-cylindri-winter performance of the horizontal array iscal receiver geometry. The modelled array usesreasonable as the sun is higher in winter. It is
Fig. 19. Seasonal performance of 48 mirror arrays in Sydney, latitude 348 S.
-
8/3/2019 CLFR Mills
20/21
282 D.R. Mills and G.L. Morrison
Fig. 20. Seasonal performance of 48 mirror arrays in Dubbo, latitude 328 S.
advanced all glass evacuated tubular absorbers for vertical and horizontal configurations was
with low emittance selective coatings. determined.
The CLFR concepts evaluated in this study (b) For optimum receiver performance a small
included absorber orientation, absorber structure, secondary reflector is required to reorient the
the use of secondary reflectors adjacent to the receiver aperture to a more favourable angle to
absorbers, mirror field configurations, mirror receive rays from the outer edge of the mirror
packing densities, and tower heights. A necessary field, and also to provide a slight degree of
requirement in this activity was the development concentration. The size of the secondary reflec-
of specific raytrace and thermal models to simu- tor is limited when the increased reflection and
late the new concepts. The primary results of the shadowing losses outweigh additional collec-
evaluations together with discussion of implica- tion.
tions are as follows. (c) A NorthSouth polar field option was also
(a) A new absorber tube configuration has been evaluated. This configuration has a high ef-developed using a purpose designed multi- ficiency and good unit aperture collection but
branched raytrace model. It was important to lower ground usage efficiency than the
have high absorber efficiency because rays lost horizontal mirror field collectors, because
between tubes necessitate enlargement of the spaces must be left between rows to avoid
entire array to compensate. The best orienta- shading. Quite large structures will be required
tion of the planar absorber (a rack of evacuated (at least 20 m along the slope) and mainte-
Dewar-type absorber tubes) was found to be nance will be more difficult than with the
horizontal rather than vertical as originally horizontal configuration. However, seasonal
proposed. The optimum size of the absorber performance is more uniform than with the
Fig. 21. Seasonal performance of 48 mirror arrays in Longreach, latitude 238 S.
-
8/3/2019 CLFR Mills
21/21
Compact Linear Fresnel Reflector solar thermal powerplants 283
Feuermann D. (1993). Analysis and Evaluation of the Solarhorizontal North South array for locationsThermal System at the Ben-Gurion Sede Boqer Test Center
outside of the tropics. for Solar Electricity Generating Technologies, Israel Minis-try of Energy and Infrastructure, Jerusalem, Final Report,(d) Optimised ganged mirror row versions ofContract No. 88169101; July.the CLFR were found to perform almost as
Feuermann D. and Gordon J. M. (1991) Analysis of a two-well as the row tracked version that can flip stage linear Fresnel reflector solar concentrator. J. Solar
Energy Eng. ASME 113, 272279.between targets.Harding G. L., Yin Z. and McKay D. W. (1985) Heat(e) It was found that a standard mirror curva- extraction efficiency of a concentric glass tubular collector.
ture could be used on all primary array reflec- Solar Energy 35, 7179.Hay J. E. and Davies J. A. (1980) Calculation of the solartors with negligible effect on performance, this
radiation incident on an inclined surface. Canadian Solargreatly simplifies manufacture.
Radiation Workshop, 5972.Igel E. A. and Hughes R. L. (1979) Optical analysis of solar
facility heliostats. Solar Energy 22, 283295.AcknowledgementsThis project was carried out with finan- Klein S. A. et al. (1996). TRNSYS. A Transient Systemcial support of the NSW Department of Energy through the Simulation, University of Wisconsin, Madison.State Energy Research and Development Fund, and the Lipps F. W. and Vant-Hull L. O. ( 1978) A cellwise method forfinancial assistance of his Royal Highness Prince Nawaf Bin the optimisation of large central receiver systems. SolarAbdul Aziz of the Kingdom of Saudi Arabia through the Energy 20, 505516.Science foundation of the University of Sydney. Mr. Wesley Morrison G. L. (1997). TRNSYS extensions (TRNAUS), Uni-Stein of Pacific Power was involved with the initiation of the versity of New South Wales, STEL 1/97.
project and provided technical input. Mr. Tony Monger Mills D. R. (1991) High temperature solar evacuated tube for(Sydney University) supplied data on circumsolar characteris- applications above 3008C. In Energy Research and De-tics for Australian climates. velopment Corporation Final Grant Report[1368, Depart-
ment of Primary Industries and Energy, Canberra, ACT2601, Australia, December.
Mills D. R. and Keepin W. (1993) Baseload solar power.REFERENCESEnergy Policy 21, 841893.
Abdel-Monem M. S. et al. (1976) A new method for collector Reindl D. T., Beckman W. A. and Duffie J. A. (1990) Diffusefield optimisation. Heliostech. Dev. 1, 372. fraction correlation. Solar Energy 45, 17.
Baum V. A., Aparasi R. R. and Garf B. A. (1957) High power Riaz M. R. (1976) A theory of concentrators of solar energysolar installations. Solar Energy 1, 6 13. on a central receiver for electric power generation. J. Eng.
Di Canio D. G., Tretyl W. J., Jur F. A. and Watson C. D. Power ASME 98, 375385.(1979). Line- focus Solar Thermal Central Receiver Re- Schmid R., Collins R. E. and Pailthorpe B. A. (1990) Heatsearch Study, FMC Corporation, Santa Clara, CA, Final transport in Dewar-type evacuated tubular collectors. SolarReport 1977-79 DOE/ET/20426-1. Energy 45, 291300.
Dudley V. E. and Workhoven R. M. (1978). Summary Report:
Turbosun, Beijing Turbosun Energy Technology DevelopmentConcentrating Solar Collector Test Results. Collector Mod- Co. Ltd, 3rd floor, No. 2B Shaoaoju, Taiyanggong, Cha-ule Test Facility, Sandia National Laboratories, Alberquer- oyang District, Beijing 100029.que, NM, Report No. SAND78-0816; May. Vant-Hull L. L. and Hildebrandt A. F. (1976) Solar thermal
Dudley V. E. and Workhoven R. M. (1979). Summary Report: power system based on optical transmission. Solar EnergyConcentrating Solar Collector Test Results. Collector Mod- 18, 3139.ule Test Facility (CMTF). January December1978, Sandia Vant-Hull L. L. (1991). In Passage in Solar Power Plants,National Laboratories, Alberquerque, NM, Report No. Winter C. J., Sizmann R. L. and Vant-Hull L. L. (Eds.), p.SAND78-0977; August. 114, Springer-Verlag, Berlin.
EC (1998) Solar thermal power: sustainable solutions for Wei L. Y. (1980) General formula for the incidence factor of asunbelt markets. In Document of the European Communities solar heliostat receiver system. Appl. Opt. 19, 31963199.
DG XVII, p. 17, The Franklin Company Associates, 192 Wei L. Y. (1981) A simplified method for calculation ofFranklin Rd., Birmingham B30 2HE, UK, ISBN 0-9524150- concentration characteristics of a solar tower system. Solar1-1; 8.5 cents of an ECU per kWh(e) is estimated for the Energy 26, 559562.planned THESEUS project in Crete. Zhang Q. -C. and Mills D. R. (1992) High solar performance
Francia G. (1968) Pilot plants of solar steam generation selective surface using bi-sublayer cermet film structures.systems. Solar Energy 12, 51. Solar Energy Mater. Solar Cells 27, 273290.