Nuclear Engineering and Design 238 (2008) 3080–3085

Contents lists available at ScienceDirect

Nuclear Engineering and Design

journal homepage: www.elsevier.com/locate/nucengdes

Flow visualization in a pebble bed reactor experiment using PIV and refractive
index matching techniques
Yassin A. Hassan ∗ , E.E. Dominguez-Ontiveros
Department of Nuclear Engineering, Texas A&M University, College Station, TX, USA

a r t i c l e i n f o a b s t r a c t

Article history: In the advanced gas-cooled pebble bed reactors for nuclear power generation, the fuel is spherical coated
Received 28 March 2007 particles. The energy transfer phenomenon requires detailed understanding of the flow and temperature
Received in revised form 11 January 2008 fields around the spherical fuel pebbles. Detailed information of the complex flow structure within the bed
Accepted 11 January 2008
is needed. Generally, for computing the flow through a packed bed reactor or column, the porous media
approach is usually used with lumped parameters for hydrodynamic calculations and heat transfer. While
this approach can be reasonable for calculating integral flow quantities, it may not provide all the detailed
information of the heat transfer and complex flow structure within the bed. The present experimental
study presents the full velocity field using particle image velocimetry technique (PIV) in a conjunction with
matched refractive index fluid with the pebbles to achieve optical access. Velocity field measurements
are presented delineating the complex flow structure.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction
The Generation IV International Forum findings relative to
the future nuclear systems (sustainability, security and reliabil-
ity, economy, non-proliferation and physical protection) have given
new impetus to graphite-moderated high-temperature gas-cooled
reactors (HTGRs). The high modular HTGR concept exhibits inher-
ent safety features due to the low power density and the large
amount of graphite present in the core which gives a large thermal
inertia in the event of accidents as loss of coolant. These passive
concepts were first introduced in German HTR-Module (pebble
fuel) design (Lohnert, 1990; Lohnert and Reutler, 1983). The fuel
design of fissile kernels coated with carbon and silicon carbide lay-
ers mixed with graphite is suitable for reaching very high burnup
and ensures a full confinement of volatile fission products during
normal and abnormal situations. Other characteristics of HTGR are
the capability of providing high temperature heat and suitability
for various power conversion cycles.
In pebble bed reactor cores, the gas flows around randomly
distributed spheres. The understanding of such complex unsteady
flows is important. This requires a variety of analysis techniques and
simulation tools. These range from simple one-dimensional mod-
els (Rousseau and Greyvenstein, 2002) that do not capture all the
significant physical phenomena to large-scale three-dimensional
∗ Corresponding author. Tel.: +1 979 945 7090.
E-mail address: y-hassan@tamu.edu (Y.A. Hassan).
computational fluid dynamics (CFD) codes (Van Staden et al., 2002).
A huge number of grids is needed to resolve the flow structure
around the spheres that require huge CPU time and memory.
Unfortunately, there are only few studies (Calis et al., 2001) which
represent a similar flow around randomly distributed spheres as in
pebble bed modular reactors (PBMRs) under high Reynolds number
flow conditions. Furthermore, most of the turbulence models that
were used for these simulations are eddy viscosity models which
does not resolve the flow field appropriately where curved flows
exist. There is a scarce of the experimental velocity data for valida-
tions of these models. The aim of the present experimental study
is to provide some data for code validation and to investigate the
complex of the flow structure.
A typical packed bed is a column where a fluid is flowing through
packed grain-like material. One important and particular applica-
tion involves energy transfer as in the advanced gas-cooled pebble
bed reactors for nuclear power generation. In this case, a “fuel
compact” is in the form of spherical coated particles. A partic-
ular approach to obtain flow velocity fields of interstitial space
in a packed bed configuration is the use of the particle tracking
velocimetry (PTV) technique (Hassan et al., 1992). One of the diffi-
culties using this method when used in porous media is the need for
optical access of the medium. In order to overcome this complexity,
the solid particles used as packing material and the fluid flowing
through the column can be chosen in such a way that their refrac-
tive indices are matched. In such way, the optical access required for
PTV technique can be achieved. Other optical measuring techniques
including qualitative flow visualization, point measurements such

0029-5493/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2008.01.027
 Y.A. Hassan, E.E. Dominguez-Ontiveros / Nuclear Engineering and Design 238 (2008) 3080–3085 3081

Fig. 1. PMMA beads immersed in (a) air; (b) water; (c) p-cymene; (d) RI matched sodium iodide solution.

as laser Doppler velocimetry (LDV). However, this technique pro-
vides one point measurement and not a full field measurement such
as particle image velocimetry (PIV).
2. Refractive index matching
list of solid materials commonly used in experimental facili-
ties where optical transparency is needed. These solids can be
matched with a compatible liquid in Table 2. Additional information
such as dynamic viscosity, density, reactivity, hazard level among
other parameters of importance for each liquid are presented in
Table 2.

Several other studies have used matching of refractive index

(MRI) approach to achieve optical accessibility. There are many 3. Experimental setup
factors that influence the decision of which materials to use
such as toxicity, flammability, reactivity, viscosity, density and A vertical column (3 cm × 3 cm × 35 cm) was constructed out of
cost. Hendriks and Aviram (1982) used an aqueous zinc iodide PMMA and randomly packed with 4.7-mm PMMA beads. The col-
solution to match the refractive index (RI) of polymethyl methacry- umn was connected to a closed loop with a reservoir and pump. The
late (PMMA), RI = 1.4873, Dybbs and Edwards (1984) along with flow was controlled by means of a flow valve in the loop between
Northrup et al. (1991) used mixtures of silicone oils, while, Chen the pump and the column.
and Fan (1992) used an aqueous sodium iodide solution. Narrow et A flow meter (McMillan 100 series) was connected to the loop
al. (2000) developed an equation to predict the RI of the solution as to measure the flow rate. The flow loop and reservoir were filled
function of temperature, concentration and wavelength. Haam et al. with p-cymene and pumped through the loop. The PTV system
(2000) used p-cymene to match PMMA particles for laser-Doppler used a laser as an illumination source. The laser sheet impinged
anemometry. Recently, Daviero et al. (2001) used a sodium chloride perpendicular to the column’s axial plane. A high-resolution/high-
solution for refractive index matching. speed camera (4700 fps) was placed perpendicular to the laser
After reviewing the literature several promising liquids that sheet plane. The camera had an equivalent viewing area of about
meet the criteria of acceptable toxicity, flammability, reactivity, 0.6 cm × 0.5 cm. This configuration allowed us to measure velocity
viscosity, density and price are selected. p-Cymene (1.49), ethyl fields in the X–Y plane with reasonable time resolution. The flow
(1.36) and benzyl (1.54) alcohol mixture, and an aqueous sodium was kept constant during the data acquisition process and temper-
iodide solution (1.49 at ∼60% concentration) that matched with ature was monitored using a type-k thermocouple installed inside
PMMA (1.4873). The isopropyl alcohol (1.38) and methylnaphtha- the loop’s reservoir tank. The flow was allowed to flow for at least
lene (1.615) mixture and diethylphthalate (1.51) were matched with 30 min before the image acquisition started. Various flow rates were
soda-lime glass (1.504). Additionally, a sample of fluoropolymer investigated, the calculated Re range between 50 and 550 based on
(FEP) was tested in water to determine the matching refractive
index quality. PMMA beads, soda-lime glass beads and (FEP) tubes
were immersed in their respective RI matching fluids. Using an
Abbe refractometer and visual inspection it was determined that Table 1
Matching refractive index solids
the best match was accomplished using a 60% volume aque-
ous sodium iodide solution as shown in Fig. 1. However, small Material Refractive Index Reference
particles were found in the solution that could cause problems Boroscilicate (Pyrex) 1.47–1.49 Budwig (1994)
with particle imaging techniques. For this reason p-cymene was Optical glasses 1.45–1.96 Budwig (1994)
selected. p-Cymene does react with the PMMA and should care- Acrylic (PMMA) 1.49 Budwig (1994)
fully be monitored if it is used. The alcohol solutions also proved Polycarbonate 1.58 Budwig (1994)
FEP 1.33 Present
to be prohibitively corrosive with the PMMA and soda-lime glass. Teflon AF 1.29–1.31 Present
The matching of the diethylphthalate to the soda-lime glass was K8 glass 1.51 Pokusaev et al. (2004)
not close enough to be effective in reducing the light refrac- Styrene–divinyl-benzene 1.5903 Cui and Adrian (1997)
tion. The matching between the water and FEP was reasonable; Silica gel 1.452 Cui and Adrian (1997)
Fused quartz 1.4584
however, the transparency of FEP was poor. Table 1 shows a
 3082
Table 2
Matching refractive index of fluids and their properties

Fluid Refractive Index CAS # Hazard Fire Reactivity Density Boiling Flash point Dynamic Cost Reference
(g/cm3 ) point/(melting (◦ C) viscosity (cP)
point) (◦ C)

Y.A. Hassan, E.E. Dominguez-Ontiveros / Nuclear Engineering and Design 238 (2008) 3080–3085
Water 1.33 1 100 1
Glycerin 1.47 56-81-5 1 0 1 1.262 182 160 934 $$ Budwig (1994)
Zinc iodide solution 1.33–1.62 10139-47-6 3 0 1 4.74 (445) 1–10. $$$$ Hendriks and Aviram
(1982)
Sodium iodide solution 1.33–1.5 (60%) 7681-82-5 2 0 1 3.67 (651) $$ Narrow et al. (2000)
Potassium thiocyanate sol. 133–1.49 333-20-0 1 0 0 1.89 (173) 1–2.4 $ Budwig (1994)
Ammonium thiocyanate sol. 1.33–1.5 1762-95-4 1 0 1 1.3 (152) 1–2.1 $$ Budwig (1994)
Sodium thiocyanate sol. 1.33–1.48 540-72-7 2 0 1 1.74 (287) 1–7.5 $ Budwig (1994)
Kerosene 1.45 8008-20-6 2 2 0 0.8 175–325 82 $ Budwig (1994)
Silicone oil mixture 1.47 63148-58-3 0 0 0 1.102 316 90–110 $$$$ Dybbs and Edwards (1984)
and Northrup et al. (1991)
Mineral oil 1.48 8020-83-5 0 0 0 0.862 $ Dybbs and Edwards (1984)
Turpentine 1.47 8006-64-2 3 0.87 150 35 1.49 Dybbs and Edwards (1984)
Solvent naphtha 1.5 0.67 Dybbs and Edwards (1984)
Soybean oil 1.47 8001-22-7 0.93 69 $$$ Dybbs and Edwards (1984)
Olive oil 1.47 0.92 84 Dybbs and Edwards (1984)
Castor oil 1.48 8001-79-4 0.96 313 986 $$$ Dybbs and Edwards (1984)
Tung oil 1.52 8001-20-5 0.93 $ Dybbs and Edwards (1984)
Cassia oil 1.6 Budwig (1994)
Dibutylphthalate 1.49 84-74-2 2 1 0 1.043 340 171 $ Budwig (1994)
Tetraline 1.541 119-64-2 2 2 1 0.969 204 75 $$$ Budwig (1994)
Benzyl alcohol 1.54 100-51-6 2 1 1 1.044 90 96 5 $$$ Cui and Adrian (1997)
Ethyl alcohol 1.362 64-17-5 2 3 1 0.79 78 14 $$$ Cui and Adrian (1997)
Methylnaphthalene 1.615 90-12-0 3 2 0 1.001 240 82 $$$$ Cui and Adrian (1997) and
Wedd (2003)
Chloronaphthalene 1.632 90-13-1 2 1 0 1.194 111 121 $$$$ Cui and Adrian (1997)
200 fluid (Dow Corning) 1.404 1 3 0 0.816 152 30 1 Stohr et al. (2003)
550 fluid (Dow Corning) 1.4935 0 1 0 1.065 35 101 125 Northrup et al. (1991) and
Stohr et al. (2003)
556 fluid (Dow Corning) 1.46 3 1 0 0.98 65 101 20 Stohr et al. (2003)
710 fluid (Dow Corning) 1.533 0 1 0 1.11 35 101 500 Stohr et al. (2003)
Cargille RI matching fluids 1.3–2.3 $$$$$ Cargille-Sacher (2005)
d-Limonene 1.473 5989-27-5 3 2 0 0.843 176 48 0.8 $$$$ Nguyen et al. (2004)
Methylsalicylate 1.526 119-36-8 2 1 1 1.184 219 96 4.09 $$$$ Nguyen et al. (2004)
Diethylphthalate 1.504 84-66-2 0 0 0 1.118 294 156 12 $$ Nguyen et al. (2004)
p-Cymene 1.491 99-87-6 2 2 0 0.857 177 47 0.876 $$ Haam et al. (2000)
Isopropyl alcohol 1.378 67-63-0 2 3 1 81 12 81 $$ Wedd (2003)

All presented levels in the table are ranked in a 0–5 scale being 0 the lowest level.
 Y.A. Hassan, E.E. Dominguez-Ontiveros / Nuclear Engineering and Design 238 (2008) 3080–3085 3083

Fig. 2. The experimental facility.

particle diameter using the following equation:

Dp Vs 
Rep = (1)
(1 − s)
where Dp is the particle diameter, Vs is the superficial velocity, ␳ is
the density of the fluid (853 kg/m3 ), ␧ is the porosity (0.31), and ␮
is the dynamic viscosity of the fluid (1.195E−3 Pa-s). Several flow
rates and different pebble sizes are being investigated. Fig. 2 shows
the setup of the experimental facility.
Fig. 4. PTV example image of particles scattered light at the center of the packed
4. Experimental results bed.

The use of p-cymene allowed PTV measurements to be per-

in this investigation showed the velocity distribution at the pore
formed at the center of the packed bed square channel. The results
scale level. Various pore geometries were observed at the column’s
midplane. These geometries elucidate the complexity of the flow
paths along the pore. Two types of the gap geometry between the
pebbles are shown in Fig. 3. Vortices were identified inside the pores
between beads. The preliminary results indicate that the amount
and size of the vortices increase with the pore size. Vorticity maps
were calculated from the obtained velocity fields at the pore level.
Higher vorticity zones were identified near the boundary of the
packing material.
The inhomogeneities in the pebble arrangements play an impor-
tant role in determination the flow structure between the pebbles
and consequently the heat transfer. In these measurements the liq-
uid was seeded with tiny tracers of 6 ␮m and uniformly mixed with
the fluid before entering the test section. After acquiring the images,
a tracking routine was utilized to obtain high accuracy velocity
vectors (Estrada-Perez, 2004). The merits the algorithms are (1)

Fig. 3. Examples of the gap shape types between the spheres as pore 1 and pore 2 Fig. 5. Example of measured velocity field at the center of the packed bed (color
are identified. scale indicates velocity magnitude) for Rep = 2.
3084 Y.A. Hassan, E.E. Dominguez-Ontiveros / Nuclear Engineering and Design 238 (2008) 3080–3085
Fig. 6. Vortex identification example at pore 1, Rep = 500.
Particles image identification is based on particle mask correlation
and gray scale thresholds. (2) Particle centroid location to sub-
pixel accuracy is achieved using one of three different techniques:
three-point Gaussian interpolation, 2D Gaussian regression, and
center of mass technique. (3) Path tracking of tracer particles is
achieved through a correlation operation using the whole range
of image gray-scale values. (4) Improved image analysis by the
use of multi-shaped regions with different image analysis param-
eters on each region. (5) Flow statistics are calculated using the
Fig. 7. Average velocity fields for Rep = 50 and 500.
randomly detected velocity vectors eliminating errors due to inter-
polation.
An example of the acquired images through the center plane
of the packed bed is shown in Fig. 4.The white spots present the
scattered laser light from the tracers. An example of the obtained
velocity field from the tracking routine is presented in Fig. 5.
The use of state-of-the-art equipment in conjunction with par-
ticle tracking velocimetry techniques allows the study of vortex
formation at pore level. Fig. 6 shows an example of an identified
vortex in the pore geometry 1. The measured velocity field is repre-
sented by vectors in the picture where streamlines are calculated
and overlapped with a solid red line color.
The measured velocity fields show the absence of vortex in the
pore for Rep < 80. However, the vortex is evident for Rep = 500. Fig. 7
shows the average velocity field for two of the cases considered
in this study. In the first case, the vector field shows a preferential
direction going upward of the image. In addition, there are flow
paths going from left to right of the picture passing through the
center of the pore. The second case shows the upward preferen-
tial velocity but the flow path at the center of the pore evolves
in a vortex with a diameter equivalent to one-tenth of the parti-
cle diameter. The velocity magnitude at the center of the pore has
considerable differences in magnitude with respect to the velocity
magnitude outside the center. The zone considered as the pore cen-
ter is marked with a black dotted line in Fig. 7.The former velocity
difference is up to one order of magnitude for both cases. How-
ever, the fully developed recirculation zone is only present for
Rep > 300.
 Y.A. Hassan, E.E. Dominguez-Ontiveros / Nuclear Engineering and Design 238 (2008) 3080–3085
5. Conclusions
An experimental investigation was performed to study the com-
plex structure through the voids of packed bed. Most of the previous
studies were restricted to understand the global parameters such
as pressure drop, drag force exerted on particles and overall poros-
ity of the bed. Detailed qualitative and quantitative information of
velocity fields were obtained using particle tracking velocimetry
and matched refractive index fluid. The velocity distribution in a
randomly packed bed was obtained. The obtained data would be
useful for enhancing our understanding of flow through packed
bed and to be utilized in the computational fluid dynamic code
validation.
References
Budwig, R., 1994. Refractive index matching methods for liquid flow investigations.
Experiments in Fluids (17), 350–355.
Calis, H.P.A., Nijenhuis, J., Paikert, B.C., Dautzenberg, F.M., Van Den Bleek, C.M., 2001.
CFD modeling and experimental validation of pressure drop and flow profile in
a novel structured catalytic reactor packing. Chemical Engineering Science 56,
1713–1720.
Cargille-Sacher Laboratories Inc., 2005. http://www.cargille.com.
Chen, R.C., Fan, L.-S., 1992. Particle image velocimetry for characterizing the flow
structure in three-dimensional gas–liquid–solid fluidized beds. Chemical Engi-
neering Science 47 (13/14), 3615–3622.
Cui, M.M., Adrian, R.J., 1997. Refractive index matching and marking methods
for highly concentrated solid–liquid flows. Experiments in Fluids (22), 261–
264.
Daviero, G.J., Roberts, P.J.W., Maile, K., 2001. Refractive index matching in large-scale
stratified experiments. Experiments in Fluids 31, 119–126.
Dybbs, A., Edwards, R.V., 1984. An index matched flow system for measurement of
flow in complex geometries. In: Proceedings of the International Symposium
on the Application of Anemometry to Fluid Mechanics, Lisbon, Portugal, pp.
171–184.
3085
Estrada-Perez, C., 2004. Improved particle tracking algorithm (PTV) using mask cross
correlation technique. M.S. Thesis. Texas A&M University.
Haam, S.J., Brodkey, R.S., Fort, I., Klaboch, L., Placnik, M., Vanecek, V., 2000. Laser
Doppler anemometry measurements in an index of refraction matched column
in the presence of dispersed beads part I. International Journal of Multiphase
Flow 26, 1401–1418.
Hassan, Y.A., Blanchat, T.K., Seeley Jr., C.H., Canaan, R.E., 1992. Simultaneous velocity
measurements of both components of a two-phase flow using particle image
velocimetry. International Journal of Multiphase Flow 18, 371–395.
Hendriks, F., Aviram, A., 1982. Use of zinc iodide solutions in flow research. Review
of Scientific Instruments 53 (1).
Lohnert, G., 1990. Technical design features and essential safety-related properties
of the HTR-module. Nuclear Engineering and Design 121 (2), 259–275.
Lohnert, G., Reutler, H., 1983. The modular HTR-a new design of high temperature
pebble bed reactor. Nuclear Energy-Journal of the British Nuclear Energy Society
22 (June (3)), 197.
Narrow, T.L., Yoda, M., Abdel-Khalik, S.I., 2000. A simple model for the refractive
index of sodium iodide aqueous solutions. Experiments in Fluids 28, 282–283.
Nguyen, T.T., Biadillah, Y., Mongrain, R., Brunette, J., Tardif, J.C., Bertrand, O.F., 2004. A
method for match the refractive index and kinematic viscosity of a blood analog
for flow visualization in hydraulic cardiovascular models. Journal of Biomedical
Engineering (126), 529–535.
Northrup, M.A., Kulp, T.J., Angel, S.M., 1991. Fluorescent particle image velocimetry:
application to flow measurement in refractive index-matched porous media.
Applied Optics 30 (21).
Pokusaev, B.G., Karlov, S.P., Shreiber, I., 2004. Immersion tomography of a gas–liquid
medium in a granular bed. Theoretical Foundations of Chemical Engineering 39,
1–5.
Rousseau, P.G., Greyvenstein, G.P., 2002. One-dimensional reactor model for the inte-
grated simulation of the PBMR power plant. In: Meyer, J.P. (Ed.), Proceedings of
1st International Conference on Heat Transfer Fluid Mechanics and Thermody-
namics. Kruger Park, South Africa.
Stohr, M., Roth, K., Jahne, B., 2003. Measurement of 3D pore-scale flow in index-
matched porous media. Experiments in Fluids (35), 159–166.
Van Staden, M.P., Janse Van Rensburg, C., Viljoen, C.F., 2002. CFD simulation of helium
gas cooled pebble bed reactor. In: Meyer, J.P. (Ed.), Proceedings of 1st Inter-
national Conference on Heat Transfer Fluid Mechanics and Thermodynamics.
Kruger Park, South Africa.
Wedd M.W., 2003.“Determination of PSD using laser diffraction”. http://www.erpt.
org/032Q/Wedd-00.html.