Applied Energy 228 (2018) 1531–1539

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Experimental evaluation of a prototype hybrid CPV/T system utilizing a T

nanoparticle fluid absorber at elevated temperatures
⁎
Todd Otanicara, , John Dalea, Matthew Orosza, Nick Brekkea, Drew DeJarnettea,
Ebrima Tunkarab, Kenneth Robertsb, Parameswar Harikumarc
a
Department of Mechanical Engineering, The University of Tulsa, Tulsa, OK 74104, USA
b
Department of Chemistry, The University of Tulsa, Tulsa, OK 74104, USA
c
Department of Physics, The University of Tulsa, Tulsa, OK 74104, USA

H I GH L IG H T S G R A P H I C A L A B S T R A C T

• Experimental demonstration of hybrid

CPV/T collector operating over 100 C
using nanoparticle filter with flowing
fluid.
• Nanoparticle filter utilizes UV/Visible
absorption with gold particles and in-
frared absorption with ITO.
• Thermal and electrical efficiency of
61% and 4% respectively at 110 C.

A R T I C LE I N FO A B S T R A C T

Keywords: Novel approaches for solar energy conversion continue to garner interest as a potential thermal and electrical
Nanoparticles energy source. Additionally, the need for systems capable of producing thermal energy at temperatures up to
Concentrating solar power 300 °C is growing as a means to provide process heat to industry and distributed generation for small com-
Photovoltaics munities. An approach that has seen recent increased interest is the hybrid concentrating photovoltaic/thermal
collector that can co-produce electricity and heat energy above 100 °C. One technique for this is to use nano-
particles in the heat transfer fluid to spectrally filter off wavelengths poorly utilized by the photovoltaic com-
ponent. Here, we have demonstrated the first on-sun operation of a nanoparticle based hybrid CPV/T solar
collector at temperatures exceeding 100 °C using a combination of gold and indium tin oxide nanoparticles in
Duratherm S flowing in the receiver, with an aperture area a full order of magnitude larger than other tests. At
14× concentration the system achieved a photovoltaic efficiency of 4% while achieving a peak thermal effi-
ciency of 61% with an outlet temperature of the fluid of 110 °C.

1. Introduction the limited ability to use the full solar spectrum, PV continues to de-
crease in cost and increase in worldwide installation. Unfortunately, the
Conventional single junction photovoltaic (PV) cells only convert a cost of batteries makes energy storage cost prohibitive particularly at
fraction of the full solar spectrum to electricity leading to a large por- industrial and utility scales. Conventional thermal absorbers, such as
tion of the incoming sunlight being reflected or thermalized. Despite those used in flat plate solar collectors or parabolic troughs (CSP), are

⁎
Corresponding author.
E-mail address: todd-otanicar@utulsa.edu (T. Otanicar).

https://doi.org/10.1016/j.apenergy.2018.07.055
Received 10 May 2018; Received in revised form 12 July 2018; Accepted 13 July 2018
0306-2619/ © 2018 Published by Elsevier Ltd.
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

highly efficient at capturing solar energy through the full solar spec- Xu et al. proposed and modeled the performance of a hybrid PVT
trum and the thermal energy is readily stored at much lower costs as system based around a triple junction AlGaLnP/LnAlGaAs/GaAs PV
thermal energy. CSP type systems in particular though are significantly cell, encapsulated between sapphire plates, deployed at the focal point
more costly than their PV counterpart, neglecting any significant pri- of a dual axis tracking dish concentrator [12]. The triple junction cell
cing premium on dispatchability [1]. Because of the dispatchability of efficiently converts as much of the visible and UV spectrum as possible,
thermal energy and the low cost of PV systems there has been sig- while transmitting approximately 75% of the out-of-band (mostly IR)
nificant recent effort at hybridizing the two systems. One of the primary light to a thermal absorber. The concentrated flux intensity is non-
challenges is the competing thermal nature of the two systems; PV is uniformly distributed across the PV cells, increase sing from ∼50 to
most efficient at low temperatures and CSP when used for electrical 580×, from the outer edge to the center of the PV cells. Passive cooling
power production increases in efficiency with temperature. Hybrid with aluminum fins and active cooling with embedded water channels
systems that are going to operate at higher temperatures, as required by were considered and predicted average cell temperatures of 87 °C and
process heat needs or for electricity production, will need to have novel 67 °C respectively. The cell module efficiency is reported as a function
optical and thermal designs such that the PV and thermal system are of in-band photon energy and was predicted to be 48.2% and 49.1% for
uncoupled or weakly coupled, through some form of spectral splitting. the passive and actively cooled conditions respectively. Thermal effi-
An overview of spectral splitting technologies can be found in the re- ciency of the proposed hybrid PVT system was not considered or re-
cent review by Mojiri et al. [2], and details of hybrid CPV/T systems are ported.
in extensive detail by Ju et al. [3]. Recently, Widyolar et al. compiled a Another approach that uses selective absorption, achieved with a
comprehensive analysis of single and two stage CPV/T designs using thin film absorbing filter was demonstrated by Stanley et al. [13]. In
silicon and gallium arsenide cell technologies and highlights some of this approach a semiconductor doped glass is embedded within a
the key techoneconomic challenges facing CPV/T systems particularly transparent glass tube that contains the working heat transfer fluid
as they relate to electrical power production [4]. Below we focus on (propylene glycol) absorbing below 700 nm and above 1100 nm. The
reviewing experimental CPV/T work in the literature. system employed a parabolic trough concentrator that provided 42×
Early experimental work focused on CPV/T receivers where thermal geometric concentration ratio. PV cells were Sunpower Maxxeon cells
energy is waste heat removed from the PV cell. One such application connected in series. During the experimental tests the observed PV ef-
was by Coventry et al. who used a parabolic trough to concentrate ficiency was 3.6–4.0%, while the observed thermal efficiency was
(37× geometric, and 30× PV) light onto a Si PV cell receiver [5] where ∼35% at working fluid temperatures of 120 °C, approaching the max-
waste heat was recovered via a water-glycol flow loop embedded in the imum working temperature due to onset of flow boiling. The authors
back of the receiver. The system achieved a net efficiency of 69%, with note that the addition on vacuum insulation around the primary ab-
the PV and thermal components operating at 11% and 58% efficiency sorber tube could lead to enhanced efficiency.
respectively with the thermal fluid reaching temperatures less than The volumetric approach outlined above by Stanley et al. [13,14] is
80 °C [6]. While this work demonstrated a CPV/T system, the working a novel approach for volumetric selectively using a glass insert, while a
thermal fluid limits uses for low temperature heating applications and number of others have proposed volumetric selective absorbers using
no spectral splitting was employed. nanoparticles [7,15–18]. The vast majority of the nanoparticle based
CPV/T systems that achieve much higher working fluid tempera- CPV/T arrangements have been numerical studies [17,19], focused on
tures are well studied numerically [7–9], but limited applications at the design of the optical filters [7,15,20], including nanoparticle
higher temperatures have been experimentally tested. Additionally, the characterization, thermal packaging, and system design [9,21–23].
number of experimental CPV/T systems that employ spectral splitting Using a volumetric approach has some advantages over dielectric sur-
are fairly limited. Below, we review experimental tests of systems em- faces that reflect light onto flat surfaces for absorption. First, there
ploying spectral splitting and filtration. exists a heat transfer advantage by directly absorbing light in the fluid
Kandilli proposed and tested a hybrid system using a parabolic dish versus absorbing on a surface followed by subsequent convection.
that focused solar energy onto a “hot mirror” which transmitted se- Second, the use of ultra-small particles such as nanoparticles effectively
lected wavelengths to a silicon PV cell while reflecting the remaining to eliminates any angle of incidence optical losses in the thermal absorber.
a conventional CSP tube mounted to the side of the dish [10]. In this For dielectric surfaces not only does the bandedge change but also
application maximum fluid outlet temperatures reported were 55 °C. optical losses can be exaggerated at non-normal angles of incidence.
The system, comprised of three dishes, resulted in a thermal power of Below we outline the limited number of on-sun experimental work for
141 W and electrical power output from the cells of 18.4 W. Solar nanoparticle CPV/T types of systems.
thermal and PV cell efficiency were not directly reported. Spectral filtration using a nanoparticle fluid filter has been experi-
Widyolar et al. investigated a hybrid PVT system that incorporates a mentally demonstrated by Crisostomo et al. [24]. The hybrid CPV/T
set of PV secondary mirrors at the focal point of a parabolic trough system reported employs a linear Fresnel lens to focus (8.3× & 16.6×,
concentrator that reflect a portion of the solar spectrum back onto a thermal and PV area-based concentration ratios respectively) light onto
traditional thermal absorber allowing the system to co-produce elec- a Si PV module. The spectral filter/thermal receiver is a glass tube si-
tricity and thermal fluid at ∼365 °C under ∼60× geometric con- tuated directly in front of the PV module. The PV module is not actively
centration ratio [11]. The secondary PV mirror is made from GaAs PV cooled in the arrangement tested so all thermal gains are the result of
cells attached to an aluminum extrusion that is actively cooled, instead the nanofluid absorber. Varying concentrations of silver nanodiscs in
of use a dichroic material the mirror effect relies on GaAs natural band water were pumped through the thermal receiver. Notably, the particles
gap cut off of 870 nm to reflect approximately 92% of the low energy used here are primarily absorbing in the visible spectrum. A maximum
light back to the thermal receiver, this entire system is installed in an temperature of 70 °C was obtained during the on-sun experimental
evacuated glass cylinder [11]. Peak thermal efficiency is reported as testing. Thermal and PV efficiencies are not reported individually, but a
roughly 37%, with the fluid inlet and exit temperatures around 310 °C peak overall efficiency of 33.2% was obtained, a significant improve-
and 350 °C respectively. A proper vacuum could not be achieved in the ment over pure water. The authors also experimentally demonstrated
receiver due to complications with incorporating the PV receiver, so the thermal decoupling between the nanoparticle based fluid absorber
reported thermal performance was adjusted, assuming a properly and the PV cell over the range of temperatures tested (35–55 °C mean
evacuated receiver. The maximum PV performance was 8% although fluid temperature) [24].
the reported PV efficiency of 8% does not correspond to the thermal An et al. experimentally investigated a fluid based spectral filter for
results as it was taken from a partly cloudy day under substantially CPV/T systems based on polypyrrole [25]. The experimental system
reduced flux. consisted of a linear Fresnel lens concentrating (10.2×) light onto a

1532
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

Fig. 1. Schematic diagram of a hybrid CPV/T solar collector that utilizes nanoparticle spectral filtering.

cylindrical quartz tube which was mounted in front of a Si PV receiver knowledge this is the first experimental test of a hybrid CPV/T system
with integrated water-based cooling channels. The quartz cylinder was using nanoparticle spectral filtration tested at operational temperatures
filled with varying concentrations of polypyrrole nanoparticles (which exceeding 100 °C with a flowing fluid, achieved through utilizing heat
primarily absorb across the visible spectrum) and water. It should be transfer fluids capable of operation up to 300 °C. Further, this work is
noted that in this experimental study the nanoparticle fluid was stag- novel as it is the first demonstration utilizing a mixture of nanoparticles
nant in the tube and a water flow loop was used to maintain the PV capable of primary absorption peaks in the visible and the infrared.
receiver temperature below 40 °C. The prototype was tested by placing Experiments are conducted on-sun to measure both thermal and PV
it on sun for ten minutes at a time and recording the PV electrical ef- efficiency, with comparisons to detailed performance models previously
ficiency and observed fluid temperature rise. The on-sun experiments developed. Additionally, the collector demonstrated here operated at an
demonstrated increased temperature rise, and decreased PV efficiency, aperture area that is a full order of magnitude larger than other na-
as the concentration of polypyrrole is increased. The authors reported noparticle based works.
an overall efficiency (based upon temperature rise, not flowing fluid) at
a peak of 25.1%, with a peak PV efficiency of 10.5%. Unfortunately, the 2. Nanoparticle based spectral filter fluid CPV/T system design
peak fluid temperature is not reported, only the maximum temperature
rise of 38 °C [25]. Extensive work was previously done to synthesize and characterize
An et al. followed up their initial study with another study using the nanoparticles and heat transfer fluids for this application
same experimental setup but with Cu9S5 nanoparticle suspended in [15,20,27,28]. Typically, the fluids by themselves are poor absorbers
oleylamine fluid with back contact monocrystalline solar cells [26]. particularly at low wavelengths (UV–visible bandgaps) and in select
Notably, this work is the first work to demonstrate the concept ex- regions in the infrared which has hampered any usage [28]. Recent
perimentally at temperatures exceeding 100 °C, but the fluid is not developments in nanotechnology have led to nanoparticles that can
flowing through the thermal receiver. The system design is similar to selectively filter solar energy [16]. This has led to the development of
prior work with a reported concentration ratio of 8.2. The authors re- nanoparticle based CPV/T collectors that can directly absorb solar en-
ported a peak total efficiency of 34.2% with the PV efficiency at 11.8% ergy into the working fluid [6–11]. Here we focus on a CPV/T system
and the thermal efficiency at 22.6%. This experiment was novel in that utilizing a nanoparticle based fluid filter, PV, and aluminum extrusion
the temperature of the fluid exceeded 100 °C with a peak temperature integrated on a parabolic concentrating trough as illustrated in Fig. 1.
recorded of 124.9 °C. The experiment was conducted in a similar The filter is placed in the path of the concentrated light to directly
fashion to prior work with the fluid not actually flowing in the receiver, absorb solar energy not efficiently utilized by the PV cell. A fluid is
so the thermal efficiency is only an estimate. Notably, they discuss the pumped through the extrusion upon which the PV cell is mounted
low thermal performance and a subsequent study points toward the where it actively cools the PV cell by absorbing waste heat. Integration
need for additional absorption in UV and visible regime [23]. of PV and thermal absorption enhances the overall system efficiencies
To our knowledge, the works of Crisostomo et al. [24] and An et al. by utilizing the full solar spectrum and optimally capturing each photon
[25,26] represent the only on-sun experimental tests of hybrid CPV/T as electricity or heat. The spectrally selective nanoparticles are de-
systems using fluid based spectral absorption. The aforementioned signed to absorb the irradiance below the PV band gap while remaining
works did successfully decouple the PV and thermal components in transparent to energy above the band gap.
hybrid CPV/T systems and provided increases in the state-of-the-art in
these types of CPV/T systems. Notably, the prior work on spectrally
2.1. Nanoparticle based thermal receiver
selective fluids for hybrid CPV/T was limited to either low temperature
operation or to a case where the temperatures are above 100 °C but the
The primary thermal receiver is comprised of the flowing nano-
fluid is stagnant.
particle doped working fluid. The heat transfer fluid selected for the
Here, we experimentally demonstrate the performance of a hybrid
base fluid of the filter is Duratherm S. Duratherm S is a polydimethyl
CPV/T system that utilizes a parabolic trough to concentrate light onto
siloxane based oil and was chosen based on its optical properties re-
a nanoparticle fluid filter placed in front of PV receiver. To our
lative to the cell bandgap, long term thermal stability of the optical

1533
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

properties, and chemical similarity to the nanoparticle surfactant for 3. Experimental setup
stabilization purposes [28]. Duratherm S has a maximum bulk tem-
perature in an open system (exposed to air) of 204 °C, which can be 3.1. On-sun testing
extended to 315 °C under inert gas, and up to these temperatures no
boiling occurs. The solar fluid filter proposed here is a combination of The proposed thermal receiver was constructed and integrated into
gold nanospheres and indium tin oxide (ITO) nanocrystals (nominally a 14× concentrator architecture and PV arrangement based on the
hexagonal crystals). Silica coated nanospheres were provided by na- Cogenra T-14 system located at The University of Tulsa North Campus
noComposix, nominally 50 nm gold diameter with a 25 nm silica shell. (36.17°N, −95.96°W). The T-14 employs a flexible mirror with a series
In-house synthesis of indium tin oxide nanoparticles (ITO) was per- of flat glass plates on a reflective aluminum substrate to form a cost-
formed to have greater control over the optical properties in the in- efficient parabolic trough collector system. The prototype trough was
frared (IR) spectral region. Synthetic procedures for ITO were per- deployed with the line focus aligned along an East-West axis to limit
formed by following the works of Garcia et al. [29], Gilstrap et al. [30], walk-off. The PV receiver and thermal receiver are both on separate
and Tunkara et al. [31]. One mmol of indium acetylacetonate was flow loops to monitor temperature, pressure, flow rate across each
mixed with 0.2 mmol of tin acetylacetonaate and 3 mmol of myristic system. The working fluid for the PV receiver is DowFrost™ Propylene
acid. Twenty milliliters of octadecene (solvent) was added to the mix- Glycol. The working fluid in the thermal receiver is the nanoparticle
ture and the whole mixture was purged with argon gas. Purging was doped Duratherm S, it is delivered to the glass extrusion by three
done five consecutive times each followed by degassing at zero atmo- flexible high-temperature stainless steel braided hoses. T-type thermo-
spheric pressure. After completion of the purging and degassing, the couples are installed where the working fluid diverges and converges,
reaction mixture was finally maintained at a vacuum set point of 15 atm into and out of the flexible hosing, as shown in the schematic, Fig. 3, as
and the temperature of the reaction increased to 110 °C and was stirred well as along the inlet and the outlet of the PV receiver. Mass flow rate
for two hours. The temperature was increased to 295 °C and 10 ml of a and fluid temperature can be controlled on both the extrusion and filter
previously degassed and heated solution of oleyamine in octadecene flow loops. Mass flow rate is measured using a FTI Flow Technologies
was added to the mixture. The temperature of the reaction was reduced FT4-6 turbine flowmeter. Error associated with the equipment in the
to 240 °C and maintained for an additional one hour. Particles were setup is outlined in Table 1.
precipitated out of the solution by adding absolute ethanol. The as- Direct normal irradiance was calculated as the difference of the
synthesized ITO nanoparticles (∼15 nm diameter) were not soluble in global and diffuse irradiance measurements recorded by two Hukseflux
Duratherm S. Therefore, in order to solubilize the ITO particles in SR11 pyranometers (uncertainty of ± 1.8%), where the diffuse irra-
Duratherm S, one will need to replace the OLAM with a Duratherm S diance is found using a shadowband on one of the pyranometers. The
compatible ligand, in this case (6–7% aminopropylmethylsiloxane) di- pyranometers are mounted on the back surface of the PV receiver such
methylsiloxane copolymer, 80–120 cSt, a PDMS-type ligand, through that they are normal to the solar irradiance. I-V data for the PV receiver
refluxing. Prior work on modeling the optical properties revealed that were taken using a SolMetric PVA-1000S handheld I-V curve tracer.
volume fractions for the gold spheres and ITO were found to be Fig. 4 shows the concentrating collector and the nanoparticle based
1.65 × 10−7 and 1.39 × 10−5, respectively [15]. While the modeling thermal absorber installed in the experimental setup. For on-sun testing
provides a guideline for finding the volume fraction, variations in the collector is rotated so that the receiver is fully focused on the re-
synthesis to synthesis requires adjusting the fluid as it is prepared to ceiver. All pumps are turned on to begin flow through the receivers.
achieve the desired optical properties [31]. Belt heaters are turned on for the nanoparticle fluid flow loop to set the
With the fluid designed, the thermal receiver can be designed to inlet temperature for tests involving the flowing nanoparticle fluid.
integrate with the primary optics and PV receiver. The thermal receiver Data were taken when steady state temperature operation is observed
is 1.2 m in length with a 15.6 cm aperture. It is constructed of three in the nanoparticle filter fluid (the temperature is not changing by more
glass extrusions with rectangular cross section to provide the working than 1 °C for 15 min). In cases without the nanoparticle filter fluid (PV
area needed to cover the 15.6 cm aperture of the PV receiver. A ray receiver only testing) the data were taken once the system is brought
trace simulation was conducted using SolTrace to understand the light on-sun.
impingement on the PV receiver. This simulation only accounts for
Fresnel reflection and refraction associated with the glass extrusions 3.2. Fluid optical properties
and base fluid but doesn’t consider the absorption associated with the
nanoparticles. Assuming an incident intensity of 1000 W/m2, the Spectra of the nanoparticles were acquired over the wavelength
SolTrace results, Fig. 2, indicate an average irradiance of ∼9150 W/m2 range from 300 to 4000 nm using both UV–Vis and Fourier Transform
on the receiver, with areas of denser flux concentration peaking at IR (FTIR) spectroscopy. Spectra were measured in the UV–Vis from 300
∼14,750 W/m2, corresponding to the expected concentration of the to 1400 nm using a Shimadzu UV–Vis 2600, and FTIR spectra were
primary optics of the Cogenra T-14 system. Overlaid in Fig. 2 is the measured from 855 to 4000 nm with a Perkin Elmer Spectrum Two. All
position of the thermal receiver glass extrusions samples were measured in a quartz cuvette with a path length of 10 mm
prior to testing on-sun to quantify the absorption levels relative to the
design.
2.2. PV receiver
4. Results and discussion
The PV receiver utilizes shingled strings of Choshu c-Si cells custom
designed for the Cogenra T-14 system. The cells are encapsulated to an 4.1. Optical properties of fluid
aluminum extrusion with cooling channels for heat removal. Each ex-
trusion represents a module rated at 99.1 Voc, 11.5 Isc, under AM1.5D Using the spectrometers outlined in the previous section the trans-
spectrum per Cogenra. Because the length of thermal receiver doesn’t mission/absorption spectrum of the prepared spectral filters can be
match the length of the original Cogenra modules the module was measured. Fig. 5 shows the spectral intensity distribution of the
spliced at a bypass diode at the length of thermal receiver and the cells AM1.5D standard, as well as absorbtance multiplied by the spectral
not illuminated were removed from the electrical circuit. intensity for both the designed fluid and prepared fluid used in the
experimental test. Also included is the spectrum of pure Duratherm S,
which can be seen to have only minimal absorption across the entire
solar spectrum. While the model provides a desired volume fraction for

1534
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

Fig. 2. Ray trace simulation results of flux profile on PV receiver with thermal receiver tube position overlaid.

Fig. 3. Schematic of experimental setup, (a) lo-

cation of thermocouples on filter components,
shaded section represents insulation. (b) 1 air
heat exchanger, 2 extrusion fluid reservoir, 3
belt heater, 4 centrifugal pump, 5 rotary mixer in
nanofluid reservoir, 6 belt heater, 7 recirculating
chiller, 8 gear pump, 9 shell and tube heat ex-
changer, 10 collector trough.

Table 1 each particle, both synthesis routines results in suspensions of con-

Details of Experimental Equipment and Associated Error. centrated particles in Duratherm S. The as synthesized solutions are
Measured parameter Instrument Error diluted and mixed until a mixture with the desired optical properties is
achieved. The experimental filter fluid closely matches the modeled
Irradiance Huskeflux SR-11 Pyranometers 1.8% fluid, transmitting 59% (versus a desired 76%) of photons in the
Temperature T-type thermocouples 1 °C 600–1150 nm ideal transmission band and absorbing 85% (versus a
Flowrate FTI Flow Technolgies FT4-6 Turbine 0.05% of
Flowmeter reading
desired 80%) of the remaining photons. The less than ideal performance
Optical Properties Shimadzu UV–VIS 2600 N/A in the transmission window of the experimental fluid is attributed to
Perkin Elemer FTIR Spectrum Two gold particle agglomeration, resulting in increased absorption. The gold
I-V Curves SolMetric PVA-1000S 0.5% nanoparticles, provided by nanoComposix, have been designed for long
term high temperature stability which results in some agglomeration,
visually apparent by some larger particles settling out in the bottom of

1535
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

typical behavior, as average cell temperatures increases, efficiency de-

creases. Additionally, as the concentration increases from 7× to 14×,
the PV receiver efficiency decreases substantially. The additional flux
exaggerates shading effects from contacts on the cell face, while also
increasing operating temperature and interconnect losses. When pure
Duratherm is introduced, a reduction in the PV operating temperature
and efficiency is observed. This decrease in temperature is attributed to
Fresnel reflective losses from the introduction of the filter glass tubes,
and limited absorption of the IR spectrum due to the Duratherm S. An
additional decrease in PV operating temperature and efficiency is ob-
served after the nanoparticle fluid is introduced to the system. The
decrease in efficiency is associated with the increased absorption of the
visible spectrum associated with the introduction of nanoparticles. In
the ideal case the filter should operate closer to that of pure Duratherm
S but the Gaussian distribution of the absorption peak for ITO and gold
results in absorption in the desired PV transmission window that ad-
versely affects the PV efficiency. Agglomeration also leads to broad-
ening of plasmon peaks, which further reduces filter performance re-
lative to the ideal.
Fig. 4. Picture of collector prototype and nanoparticle fluid in filter, top- the
The hybrid PVT model developed earlier [22] predicts a PV effi-
concentrator and associated flow loops, and bottom- the glass extrusions with ciency of approximately 13% for the same system design working with
nanoparticle fluid in front of the PV cells. the modeled fluid spectrum in Fig. 5. As shown in Fig. 6, the PV receiver
nominally can achieve a 13% efficiency without the filter in place at
temperatures below 60 °C. The difference in efficiency between the
modeled and experimental results is due to a number of factors. First,
and the most significant, is the difference in the modeled spectral
properties of the thermal receiver versus the actual tested fluid. As
shown in Fig. 5, the transmittance in the PV window is reduced by 17%
over the modeled fluid resulting in less photon energy to the PV re-
ceiver. Second, the model doesn’t account for any non-uniformity in
solar flux on the PV receiver. Although the SolTrace model points to a
fairly uniform flux on the solar cells any irregularity in operation from
that could result in significant changes in the PV performance. Third,
and last, the model doesn’t account for module level impacts to PV
receiver efficiency resulting in lower than predicted PV performance.
Fig. 7 shows the on-sun IV curves for the PV receiver with and
without the nanoparticle thermal receiver. As can be seen the in-
stallation of the nanoparticle thermal receiver results in a significant
decrease in the short circuit current due to the increased reflection with
Fig. 5. Spectral intensity of AM1.5 D spectrum and the intensity absorbed by the glass tubes and absorption from the nanoparticle fluid resulting in
the Modeled and As Created experimental nanoparticle fluid based. less flux to the receiver. Additionally, the thermal receiver results in an
increase in the open circuit voltage, resulting from the reduced tem-
the container over long times of stagnation. perature of the PV receiver, again due to the significant absorption of
photons prior to the PV receiver.
4.2. PV receiver
4.3. Thermal receiver
On-sun data were collected for the PV receiver using a handheld IV
tracer where the PV receiver is operated: by itself (no thermal receiver- All data collected for the thermal performance of the receiver were
PV only), with the thermal receiver only using Duratherm S, and with taken after the prototype system had reached steady-state operation.
the nanoparticle fluid in the thermal receiver. PV receiver efficiency is Steady state operation was assumed to occur after the fluid inlet and
defined in Eq. (1). outlet temperatures changed less than 1 °C over a 15 min period. The
prototype system thermal receiver on-sun thermal output is calculated
Pmax
ηPV = using Eq. (2) [34]:
DNI ·C·A (1)
̇ p (THTF , out −THTF , in )
Qth = mc (2)
where Pmax is the maximum power from the I-V curve measurement,
DNI is the measured direct normal irradiance, C is the concentration where ṁ is the mass flow rate, cp is the specific heat of the Duratherm S
ratio and A is the area of the cells. A ± 3.6% manufacturer instrument in the thermal receiver, THTF,out and THTF,in are the averaged, measured
error is accounted for in the irradiance measurements, and 0.5% asso- temperatures at the inlet and exit of the thermal receiver. The solar
ciated with the handheld IV tracer [32,33]. The concentration ratio thermal efficiency can then be determined from Eq. (3) [34]:
used in all the calculated efficiencies (thermal and electric) is based on
Qth
the area, i.e. the ratio of the concentrator aperture to the receiver ηth =
DNI ·C·A (3)
aperture.
Fig. 6 details the PV receiver efficiency with and without the optical here, A area is defined as total occupied space of the filter (0.185 m . A2

filter under operating temperatures. The data were collected over a error of ± 1 °C was accounted for in thermocouple data [35]. Thermal
number of experimental test days, more details on the test conditions gain isn’t reported for the PV receiver because the temperature rise
can be found in the supplemental information. The PV receiver exhibits recorded along the length of the PV receiver (for the tested flow rate) is

1536
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

Fig. 6. On-Sun PV Receiver Efficiency (open squares-7× concentration, no filter; open circles-14×, no filter; filled squares-14× concentration, pure Duratherm S
thermal receiver; filled triangles-14× concentration, Duratherm S + nanoparticles thermal receiver).

Fig. 7. Representative IV curve of PV receiver with and without the nano-

particle thermal receiver.
Fig. 8. Comparison of numerical and experimental thermal efficiency for solar
thermal receiver (Experimental: triangles - 0.013 kg/s, squares – 0.05 kg/s,
within the experimental uncertainty of the thermocouples. circles – 0.06 kg/s; Numerical: solid line – 0.013 kg/s, dashed line – 0.05 kg/s,
Fig. 8 compares the experimental and modeled thermal efficiencies dash-dot line – 0.06 kg/s).
as a function of temperature and irradiance. The model results utilize
the prior work developed by our group in Brekke et al. [22] by as- tube where it is possible to get air to flow through the gap resulting in
suming the average ambient temperature from the experimental tests, heat loss. Along the edges of the receiver, the model also assumes
the corresponding mass flow rate experimental data sets, a direct perfect insulation, but this was not achieved in practice. To limit heat
normal irradiance corresponding to the measured values, and the loss along the edges wind blockers, aluminum metal plates not in
modeled fluid. The model results and experimental data at the lowest contact with the glass but providing a wind-shield, were installed along
flow rate, 0.013 kg/s, share a similar slope but have drastically different the long axis of the thermal receiver but these are not insulated and also
thermal efficiencies. This is primarily due to the model under predicting had slight gaps that would allow airflow. In the model the internal
the convective heat losses that dominate at the low flow rates here. The cavity between the thermal receiver and the PV cell improves the solar
model assumes that the thermal receiver is perfectly insulated on all thermal efficiency by limiting heat transfer in the same way a con-
sides except the side facing the incident flux. In reality, the filter is ventional solar thermal collector would, but this requires the back of
comprised of three separate tubes, with minimal gaps between each the edges to be well insulated and for the air gaps to be eliminated.

1537
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

operating at temperatures over 100 °C where the fluid is flowing in the

thermal receiver. A number of works don’t directly report the PV effi-
ciency or thermal efficiency directly, or report it on a basis of exergetic
efficiency, merit function, or on an area/flux basis different from the
full sun. All efficiencies in the table are reported on a total irradiance to
the system aperture basis. For the cases where that data are not directly
reported it is possible to calculate the efficiency. Raush and Chambers
reported 20.05 W of electrical power from 717 W/m2 of irradiance for
an electrical efficiency of 3.19% [36]. The work by Kandilli reported
141 W of thermal power and 18 W of electrical power under 734 W/m2
of irradiance resulting in thermal and electrical efficiencies of 22.9%
and 2.98% respectively [10]. Crisostomo et al. reported results for
thermal performance in terms of useful heat gain per unit length
(without specified irradiance data) so it is not possible to back out
thermal efficiency [24]. It is possible to calculate PV efficiency for the
Fig. 9. Thermal Efficiency as a function of average thermal receiver tempera- Crisostomo et al., which had a peak PV efficiency of 5.9% with the
ture for a mass flow rate of 0.05 kg/s. lowest volume fraction of nanoparticles [24]. While the system here
reports high thermal efficiency it is at the expense of PV efficiency, due
When the thermal receiver flow rate is increased to 0.05 kg/s the to significant absorption in the PV window. Increasing the PV efficiency
differences between the model and experimental results is reduced with through better fluid suspension would result in significantly less
the model still over predicting performance due to the previously noted thermal efficiency presenting one of the important design tradeoffs in
reasons regarding additional heat loss mechanisms. At low flow rates the hybrid CPV/T system. The thermal efficiencies achieved here are
these impacts are exacerbated, but as flow rate is increased the model the result of significant absorption both in the desired IR and UV
and experiment come into line. The modeled and experimental data are windows but also in the undesired PV window. This is easily seen in
within the experimental error at a flowrate of 0.06 kg/s. This flowrate comparison to the work by Stanley et al. which used a filter with very
also corresponds to the optimal flowrate in the model for minimizing sharp cutoffs that resulted in much less thermal absorption [13]. While
thermal losses and maximize absorption thermal gain. Operating the the cell efficiencies are lower the cells used here were optimized for
experiment at flowrates further from this local minimum will cause the 14× concentration versus the cells used there which were off the shelf
difference between the model and the experimental data to be ex- 1-sun Sunpower Maxeon cells.
acerbated increasing thermal losses.
Fig. 9 demonstrates the thermal efficiency of the receiver at dif- 5. Conclusion
ferent operating temperatures. As can be seen there is a significant
decrease at higher temperatures resulting from the limited thermal Here we have demonstrated the first on-sun operation of a nano-
insulation in this design. The results do indicate the potential for op- particle based hybrid CPV/T solar collector at temperatures exceeding
erating nanoparticle based systems at temperatures exceeding 100 °C. 100 °C with a flowing fluid at an aperture an order of magnitude larger
Table 2 presents a comparison of the of known hybrid CPV/T sys- than prior work. Further, the collector demonstrated here operated at
tems that employ spectral filtration or splitting for which there exists an aperture area that is a full order of magnitude larger than other
on-sun experimental testing, the experimental data for Coventry are nanoparticle based works. While the system was operational with both
also included as a comparison to non-splitting approaches. As can be measurable thermal and electrical output it was not without opera-
seen the vast majority of spectral splitting approaches have been op- tional problems. Adhesion of gold nanoparticles on the glass walls of
erated at temperatures well under 100 °C. Those that have operated the thermal receiver was observed due to the affinity of the silica shells
above 100 °C have used splitting approaches such as dichroic mirrors to the glass. While the shells are necessary for stability in the fluid at
[36], GaAs cells [11], semiconductor doped glass [13], and Cu9S5 na- high temperatures (> 100 °C) the buildup on the walls is not desirable
noparticle suspension with a stagnant fluid [26]. To our knowledge this and further work should focus on maintaining high temperature stabi-
is the first system using nanoparticles in the heat transfer fluid lity while preventing this buildup. The prior work on numerical mod-
eling provided a fairly accurate prediction of the system performance at

Table 2
Performance comparison of nanoparticle fluid, spectrum splitting, hybrid CPV/T systems.
Ref. Spectrum splitting Concentrator CR Aperture area Working fluid Max PV Max thermal Max thermal
method [m2] efficiency efficiency temperature [°C]

[5] None Parabolic Trough 37 1.8 Water + Anti-freeze 11% 58% < 80
[11] GaAs secondary mirror Parabolic Trough 60 5 Therminol VP-1 8.50% 47% 375
[36] Dichroic Mirror Parabolic Trough 796 (PV) 0.876 R245FA NR NR 121
[10] Hot Mirror Dish 15 (T) 0.28 × 3 Water NR NR 55
109.5
(PV)
[24] Silver nanodisc + fluid Linear Fresnel 8.3 (T) 0.25 Water NR NR < 70
16.6
(PV)
[13] Doped glass Parabolic Trough 42 2.4 Propylene Glycol 4% 35% < 130
[25] Polypyrrole nanofluid Linear Fresnel 10.2 0.27 Water 10.70% 36%* < 60
(no flow)
[26] Cu9S5 particles + fluid Linear Fresnel 8.2 0.27 Oleylamine 11.8% 22.6%* < 130
(no flow)
This study Gold & ITO particles + Parabolic Trough 14 5 Duratherm S 5% 61% < 115
fluid

1538
T. Otanicar et al. Applied Energy 228 (2018) 1531–1539

higher flow rates, but further refinement is needed for the heat losses in system. Energy Convers Manage 2013;67:186–96.
the model and experiment to properly match. Additionally, the authors [11] Widyolar A, Abdelhamid Bennett, Jiang Mahmoud, Winston Lun, Yablonovitch
Roland, Scranton Eli, et al. Design, simulation and experimental characterization of
would recommend a flow geometry for the thermal receiver to be a a novel parabolic trough hybrid solar photovoltaic/thermal (PV/T) collector.
cylinder for strength and durability reasons as well as to allow better Renew Energy 2016;101:1379–89.
thermal insulation. This was not possible in the T14 concentrator design [12] Xu Q, Ji Y, Riggs B, Ollanik A, Farrar-Foley N, Ermer JH, et al. A transmissive,
spectrum-splitting concentrating photovoltaic module for hybrid photovoltaic-solar
due to the wide flux profile. Many challenges still exist in the devel- thermal energy conversion. Sol Energy 2016;137:585–93.
opment of a filter fluid acting as a thermal absorber and spectral filter [13] Stanley C, Mojiri A, Rahat M, Blakers A, Rosengarten G. Performance testing of a
but results are promising for further development in applications with spectral beam splitting hybrid PVT solar receiver for linear concentrators. Appl
Energy 2016;168:303–13.
temperatures below 300 °C. The primary challenge remains, making a [14] Mojiri A, Stanley C, Rodriguez-Sanchez D, Everett V, Blakers A, Rosengarten G. A
nanoparticle suspension with long-term stability. Above 300 °C these spectral-splitting PV–thermal volumetric solar receiver. Appl Energy
systems are unlikely to be deployed without significant breakthroughs 2016;169:63–71.
[15] DeJarnette D, Tunkara E, Brekke N, Otanicar T, Roberts K, Gao B, et al.
in stability as well as novel fluids that can operate without significant
Nanoparticle enhanced spectral filtration of insolation from trough concentrators.
pressurization which places challenges on transparent (glass) tubes. Sol Energy Mater Sol Cells 2016;149:145–53.
Further innovations in thermal packaging are also required to limit heat [16] Otanicar TP, DeJarnette D, Hewakuruppu Y, Taylor RA. Filtering light with nano-
loss for volumetric solar collectors. Future studies should focus on in- particles: a review of optically selective particles and applications. Adv Opt Photon
2016;8:541.
creasing the temperature further, improving the thermal packaging of [17] Brekke N, Otanicar T, DeJarnette D, Harikumar P. A parametric investigation of a
the design to minimize thermal losses, and an improved design that is concentrating PV/T system with spectral filtering utilizing a 2-D heat transfer
simpler to manufacturer and is based around cylindrical glass tubes model. J Sol Energy Eng 2016.
[18] Hjerrild NE, Mesgari S, Crisostomo F, Scott JA, Amal R, Taylor RA. Hybrid PV/T
which support higher pressure and will be more robust. Model efforts enhancement using selectively absorbing Ag–SiO2/carbon nanofluids. Sol Energy
should focus on more detailed prediction of the heat loss mechanisms Mater Sol Cells 2016;147:281–7.
that more accurately represent the as-built system. Another area of [19] Otanicar TP, Taylor RA, Telang C. Photovoltaic/thermal system performance uti-
lizing thin film and nanoparticle dispersion based optical filters. J Renew Sustain
improvement is in the solar cell. While the 5% PV efficiency here is low, Energy 2013;5:33124.
primarily caused by increased absorption in the visible from a non-ideal [20] DeJarnette D, Otanicar T, Brekke N, Hari P, Roberts K. Selective spectral filtration
nanoparticle fluid, use of back contact solar cells with higher perfor- with nanoparticles for concentrating solar collectors. J Photon Energy
2015;5:57008.
mance could also increase cell performance. [21] Hassani S, Saidur R, Mekhilef S, Taylor RA. Environmental and exergy benefit of
nanofluid-based hybrid PV/T systems. Energy Convers Manage 2016;123:431–44.
Acknowledgments [22] Brekke N, Dale J, DeJarnette D, Hari P, Orosz M, Roberts K, et al. Detailed per-
formance model of a hybrid photovoltaic/thermal system utilizing selective spectral
nanofluid absorption. Renew Energy 2018.
This work is supported in part by ARPA-E FOCUS program grant [23] An W, Li J, Ni J, Taylor RA, Zhu T. Analysis of a temperature dependent optical
number DE-AR0000463. Results of this work do not necessarily reflect window for nanofluid-based spectral splitting in PV/T power generation applica-
the views of the Department of Energy. tions. Energy Convers Manage 2017;151:23–31.
[24] Crisostomo F, Hjerrild N, Mesgari S, Li Q, Taylor RA. A hybrid PV/T collector using
spectrally selective absorbing nanofluids. Appl Energy 2017;193:1–14.
Appendix A. Supplementary material [25] An W, Zhang J, Zhu T, Gao N. Investigation on a spectral splitting photovoltaic/
thermal hybrid system based on polypyrrole nanofluid: preliminary test. Renew
Energy 2016;86:633–42.
Supplementary data associated with this article can be found, in the [26] An W, Wu J, Zhu T, Zhu Q. Experimental investigation of a concentrating PV/T
online version, at https://doi.org/10.1016/j.apenergy.2018.07.055. collector with Cu9S5nanofluid spectral splitting filter. Appl Energy
2016;184:197–206.
[27] Otanicar T, DeJarnette D, Brekke N, Tunkara E, Roberts K, Harikumar P. Full
References spectrum collection of concentrated solar energy using PV coupled with selective
filtration utilizing nanoparticles. MRS Adv 2016;1:2935–40.
[1] Branz HM, Regan W, Gerst KJ, Borak JB, Santori EA. Hybrid solar converters for [28] Looser R, Vivar M, Everett V. Spectral characterisation and long-term performance
maximum exergy and inexpensive dispatchable electricity. Energy Environ Sci analysis of various commercial Heat Transfer Fluids (HTF) as Direct-Absorption
2015. Filters for CPV-T beam-splitting applications. Appl Energy 2014;113:1496–511.
[2] Mojiri A, Taylor R, Thomsen E, Rosengarten G. Spectral beam splitting for efficient [29] Garcia G, Buonsanti R, Runnerstrom EL, Mendelsberg RJ, Llordes A, Anders A, et al.
conversion of solar energy—a review. Renew Sustain Energy Rev 2013;28:654–63. Dynamically modulating the surface plasmon resonance of doped semiconductor
[3] Ju X, Xu C, Han X, Du X, Wei G, Yang Y. A review of the concentrated photovoltaic/ nanocrystals. Nano Lett 2011.
thermal (CPVT) hybrid solar systems based on the spectral beam splitting tech- [30] Gilstrap RA, Capozzi CJ, Carson CG, Gerhardt RA, Summers CJ. Synthesis of a
nology. Appl Energy 2017;187:534–63. nonagglomerated indium tin oxide nanoparticle dispersion. Adv Mater
[4] Widyolar B, Jiang L, Winston R. Spectral beam splitting in hybrid PV/T parabolic 2008;20:4163–6.
trough systems for power generation. Appl Energy 2018;209:236–50. [31] Tunkara E, DeJarnette D, Muni M, Otanicar T, Roberts KP. Optical properties of
[5] Coventry JS. Performance of a concentrating photovoltaic/thermal solar collector. colloidal indium tin oxide suspended in a thermal fluid. J Phys Chem C
Sol Energy 2005;78:211–22. 2018;122:5639–46.
[6] Coventry JS. A solar concentrating photovoltaic/thermal collector; 2004. [32] Hukseflux. Hukseflux Thermal Sensors User Manual DR-Series Pyrheliometers; n.d.
[7] Taylor RA, Otanicar TP, Rosengarten G. Nanofluid-based optical filter optimization p. 1–47.
for PV/T systems. Nat Light Sci Appl 2012;1. [33] Solmetric. Solmetric PV Analyzer-1000S Specification Sheet. Solmetric; 2016.
[8] Brekke N, Otanicar T, DeJarnette D, Hari P. A Parametric investigation of a con- [34] Duffie JA, Beckman WA. Solar engineering of thermal processes. Wiley; 2006.
centrating photovoltaic/thermal system with spectral filtering utilizing a two-di- [35] O. Engineering. T-Type thermocouple reference tables; n.d.
mensional heat transfer model. J Sol Energy Eng 2016;138:21007. [36] Raush J, Chambers T. Initial field testing of concentrating solar photovoltaic (CSPV)
[9] Hassani S, Taylor RA, Mekhilef S, Saidur R. A cascade nanofluid-based PV/T system thermal hybrid solar energy generator utilizing large aperture parabolic trough and
with optimized optical and thermal properties. Energy 2016;112:963–75. spectrum selective mirrors. Int J Sustain Green Energy 2014;3:123–31.
[10] Kandilli C. Performance analysis of a novel concentrating photovoltaic combined

1539