Journal of Food Engineering 190 (2016) 80e93

Contents lists available at ScienceDirect

Journal of Food Engineering

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

Review

A review of the use and design of produce simulators for horticultural

forced-air cooling studies
Gabe P. Redding a, Angela Yang b, Young Min Shim b, Jamal Olatunji b, Andrew East b, *
a
Biological Systems Modeling Limited, Palmerston North 4442, New Zealand
b
Centre for Postharvest and Refrigeration Research, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand

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

Article history: Forced air cooling studies are hindered by the practical difficulties and cost associated with pallet scale
Received 10 December 2015 investigations. Produce simulators are a potential solution for overcoming some of these difficulties.
Received in revised form Despite their common use in precooling research, there are a lack of design guidelines for such simu-
28 April 2016
lators in the literature which in some case has led to their inappropriate use. This review examines
Accepted 19 June 2016
Available online 21 June 2016
previous use of produce simulators in produce cooling studies in the context of transport phenomena
theory. This allows constraints to be proposed which facilitate the design of produce simulators that can
replicate transport phenomena in horticultural produce. These include the requirements of geometric
Keywords:
Produce
and thermal property matching as well as the practical constraints imposed on produce simulators. It is
Simulator suggested that future computational modelling work is well suited to the establishment of produce
Cooling simulator design guidelines which incorporate complex effects such as the impact of variation in produce
Design geometry on flow. The material engineering challenges posed by produce simulators are also discussed.
Heat transfer © 2016 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2. Simulation of heat transfer in produce during forced air pre-cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3. Geometric matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4. Evaporative cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5. Addition of respiratory heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6. Thermal property matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.1. Single-domain produce simulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.2. Multi-domain produce simulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.3. Materials for thermal property matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
7. Simplifying cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

1. Introduction post harvest due to continued physiological, biochemical, and

microbiological processes (Verboven et al., 2006; Dehghannya
The quality of horticultural produce progressively deteriorates et al., 2010). Immediate chilling of produce after harvest partially
arrests these processes and is used to maintain quality and prolong
storage (Castro et al., 2004). Such chilling is known as precooling
and its optimisation is vitally important to the food industry
* Corresponding author. (Brosnan and Sun, 2001), with up to 20% of all perishable foods
E-mail address: A.R.East@massey.ac.nz (A. East).

http://dx.doi.org/10.1016/j.jfoodeng.2016.06.014
0260-8774/© 2016 Elsevier Ltd. All rights reserved.
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
Nomenclature
a thermal diffusivity (m2/s)
Bi Biot number, hL/k
c heat capacity (J/kg K)
E activation energy of carbon dioxide respiration (J/mol)
V evaporative heat loss as percentage of heat load
Fo Fourier number, at/L2 or at/ro2
h convective heat transfer coefficient (W/m2K)
DHvap latent heat of vaporisation (J/kg)
k thermal conductivity (W/mK)
kt transpiration mass transfer coefficient (kg/kg s Pa or
kg/m2 s Pa)
kair air film mass transfer coefficient (kg/m2 s Pa)
kskin skin mass transfer coefficient (kg/m2 s Pa)
L characteristic length, half the shortest dimension was
used for produce (m)
M percentage moisture loss from produce (%)
r density (kg/m3)
P static pressure (Pa)
Pair partial pressure of water in air (Pa)
Psurf partial pressure of water at produce surface (Pa)
f volume fraction of filler material
q heat of respiration (W/kg)
Q heat of respiration as a percentage of heat load
R the gas constant 8.314 (J/mol K)
being lost due to a lack of appropriate refrigeration (Defraeye et al.,
2015a).
In order to slow spoilage processes it is desirable to cool produce
as rapidly as possible, provided this does not cause chilling injury to
the produce (Valente et al., 1996) and is done in a homogenous
fashion. Non-uniform cooling results in under or over-cooling of
produce in different locations, which leads to variation in product
quality and associated product loss (Ferrua and Singh, 2009d;
Dehghannya et al., 2010). Hence, minimisation of product loss can
generally be achieved by cooling as rapidly and uniformly as
possible. For forced air cooled systems, the rate and uniformity of
cooling is ultimately determined by the flowrate, distribution and
temperature of the air throughout the entire packaging structure
(O’Sullivan et al., 2014).
One of the key parameters which effects airflow within horti-
cultural packaging is the size and location of vents. Enhancement of
cooling via design of vents is constrained by the requirement that
the packaging be of sufficient strength to protect produce from
damage during handling, transport and storage (Delele et al., 2008;
Defraeye et al., 2015a). Though requirements of structural integrity
have historically been prioritised over those which promote rapid
and uniform cooling (Castro et al., 2004; Ferrua and Singh, 2009a),
ventilated package design has still been the subject of a great deal
of research as summarised in the review of Pathare et al. (2012).
This review demonstrated that there is considerable variation in
the design recommendations for ventilated packages made across
various studies.
The lack of consensus in design recommendations is partly the
result of the challenges associated with studies in this field which
involve geometrically complex packaging systems and biologically
variable produce. The experimental determination of airflow
around individual products is difficult to achieve without disturb-
ing the packaging arrangement itself (Dehghannya et al., 2010;
O’Sullivan et al., 2014) and these difficulties have promoted the
81
r radial distance from center (m)
ro radius (m)
r* dimensionless distance from centre (r/ro)
T temperature (K)
DT change in temperature associated with a period of
precooling (s)
t time (s)
Dt time period of precooling (s)
q temperature (
C)
u flow velocity vector (m/s)
n kinematic viscosity (m2/s)
Y fraction of unaccomplished change in temperature,
(TT∞)/(TiT∞)
z respiratory production rate of carbon dioxide (mL/
kg hr)
zo respiratory pre-exponential for the production rate of
carbon dioxide (mL/kg hr)
Subscripts
a air
c composite material
f filler material
i initial
m matrix material
∞ Ambient
p produce
use and development of non-invasive flow measurement tech-
niques in packaging studies (Ferrua and Singh, 2008; O’Sullivan
et al., 2014).
Practical considerations of expense and time result from the fact
that experiments are required to be on a full pallet scale and are
subject to the seasonal availability of produce. This has limited the
scope of many studies (Ferrua and Singh, 2008). Experimental
replication of cooling studies is hindered by the biological nature of
produce (Castro et al., 2004). Degradation of produce, for example,
can change the heat transfer properties of produce during experi-
mentation. Surface evaporation is a significant form of heat loss for
some produce (Chuntranuluck et al., 1998a) and moisture loss from
products may mean that this effect differs over time. To avoid such
issues, replicate runs can be performed with fresh batches of pro-
duce, though this is costly, time consuming and subject to any
biological variation in parameters such as shape and ripeness
which will in turn lead to variability in replicate experiments.
These problems of experimental cost and time are magnified
when considering designing packaging to ensure maintenance of
temperature in transport systems such as refrigerated container
ship holds. In these cases full containers may be required to
conduct a useful experiment (e.g. Tanner and Amos, 2003; Defraeye
et al., 2015b), resulting in considerable cost for a single replicate.
Hence, cost becomes a limiting constraint when studying these
systems (Tanner et al., 2002a; Ambaw et al., 2013).
The difficulties associated with experimentation have led many
studies to focus on the alternative approach of developing mathe-
matical models capable of predicting the flow fields and heat
transfer within packaged commodities during forced air cooling
(Ferrua and Singh, 2008). These mathematical approaches have
been the subject of multiple reviews (Verboven et al., 2006;
Dehghannya et al., 2010; Delele et al., 2010; Ambaw et al., 2013;
Zhao et al., 2016). Earlier studies such as those of Talbot (1988)
made the simplifying assumption that the packed structure was a
82 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
porous medium, however this assumption has been shown to be
invalid in many common packages where the container to produce
diameter ratio is less than 10 (Pathare et al., 2012). Continued ad-
vancements in numerical modelling techniques and computational
power have greatly reduced the need for such simplifying as-
sumptions and have allowed examination of transport phenomena
which can be highly localised in packaged and palletised horticul-
tural produce (Alvarez and Trystram, 1995; Verboven et al., 2006;
Dehghannya et al., 2010; Delele et al., 2010; Ambaw et al., 2013;
Defraeye et al., 2015a).
For the purposes of reducing the complications associated with
experimentation and also for mathematical model validation, many
studies have also utilised produce simulators (Minh et al., 1969;
Alvarez and Trystram, 1995; Chuntranuluck et al., 1998b; Alvarez
and Flick, 1999a; Alvarez and Flick, 1999b; Vissotto et al., 2000;
Alvarez et al., 2003; Castro et al., 2004; Vigneault and Castro,
2005; Allais et al., 2006; Delele et al., 2008; Ferrua and Singh,
2008). Produce simulators are generally employed with the
objective to sufficiently replicate the flow field and/or temperature
profiles that would be produced by real produce, often given
specified simplifying assumptions (e.g. negligible heat added by
produce respiration). A produce simulator which can achieve these
objectives must also satisfy a number of practical requirements.
These include being economical and convenient for use in studies
conducted on a pallet scale across a range of produce. Such a pro-
duce simulator would facilitate more rapid research progress by
overcoming the limitations associated with large scale studies
involving real horticultural produce.
Produce simulators are seen to be the most appropriate way to
conduct experiments for determining the effect of different pa-
rameters on horticultural packaging design (Vigneault and Castro,
2005) as well as for validating mathematical models which are
being increasingly used in this area of research (Pathare et al., 2012;
Ambaw et al., 2013). Highly effective produce simulators would
likely play an important role in turning optimal horticultural
packaging design into a reality.
Given the exponential increase in research interest in horticul-
tural packaging design over the past 20 years (Fig. 1), the potential
need for produce simulators has never been greater. However,
despite the use of produce simulators in horticultural studies and
their potential utility to the industry, there has been very little
research specifically targeted at the effective design of produce
simulators. This has increased the likelihood of inappropriate use
and haphazard experimental design involving produce simulators.
The objective of this review is to highlight the challenges that need
to be addressed in produce simulator design by reviewing their
Fig. 1. Web of Science™ citations for the topic keywords horticultural, produce,
packaging and design.
current use in the context of design theory as it applies to the forced
air precooling of horticultural produce.
2. Simulation of heat transfer in produce during forced air
pre-cooling
Of the variety of different methods available for post-harvest
cooling operations, forced air cooling is the most common
(Ambaw et al., 2013). Fig. 2 shows a graphical representation of a
tunnel cooler, the most common type of forced air cooling device
(O’Sullivan et al., 2014). Produce is palletised and stacked in
ventilated packages and this physical structure leads to the uneven
distribution of refrigeration air and subsequent cooling heteroge-
neity due to localised heat transfer (Alvarez and Trystram, 1995;
Dehghannya et al., 2010; O’ Sullivan et al., 2014). Achieving a suf-
ficiently homogenous distribution of refrigeration air is the subject
of a large volume of research, and can be facilitated in a variety of
ways including adjusting package stacking arrangements, baffling,
periodic flow reversal, airflow bypass, manipulation of packaging
ventilation design and the positioning of produce within packages
(Aswaney, 2007; Ferrua and Singh, 2009a; Dehghannya et al., 2010;
Pathare et al., 2012; O’ Sullivan et al., 2014).
In forced air cooling, heat transfer within produce is multifac-
eted (Fig. 3). Heat is removed from produce by conduction to the
product surface followed by convective heat transfer from the
produce surface over which cold air is forced by a fan. Additional
heat may be lost from the surface of produce via moisture loss and
the associated removal of latent heat of evaporation
(Chuntranuluck et al., 1998a). Finally, continued post-harvest
respiration also adds heat to produce. Net radiation effects on
packaged products are small relative to experimental error and, if
required, can be sufficiently described as pseudo convection
(Alvarez and Flick, 1999b; Tanner et al., 2002b; Le Page et al., 2009).
Ferrua and Singh (2009b, c) showed that transport phenomena
during forced air precooling applications can be modelled by
decoupling the momentum transport from the transport of energy
and mass, and that the flow field within and between packages can
be sufficiently described as the steady, laminar and incompressible
flow of air for use in subsequent heat and mass transfer models
(Ferrua and Singh, 2009a). Under these conditions, flow within the
system can be described by the following forms of the continuity
and Navier-Stokes equations respectively,
V$u ¼ 0 (1)
VP
u$Vu ¼  þ n a V2 u (2)
ra
The thermophysical properties of both air and produce can be
considered to be constant over the duration of precooling due to
the relatively modest temperature range and small losses in
moisture (Tanner et al., 2002b; Dehghannya et al., 2010; Ferrua and
Singh, 2009a). It can also be assumed that energy transfer in air due
to the transfer of enthalpy associated with the diffusion of water
vapour is negligible (Ferrua and Singh, 2009a). Under these as-
sumptions heat transfer in the air domain is described by,


vTa
ðrcÞa þ u$VTa ¼ ka V2 Ta (3)
vt
and in the produce domain by,
vTp
ðrcÞp ¼ kp V2 Tp þ Q (4)
vt
The boundary condition at the air produce interface can be
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
Fig. 2. A tunnel cooler, the most common type of forced air cooling device.
Fig. 3. A schematic of heat transfer mechanisms in produce.
described by,
   
kp VT ¼ h Tp  T∞ þ kt Psurf  Pair DHvap
where the heat transfer coefficient can be described as a function of
fluid velocity or by some other appropriate correlation (Nusselt-
Reynolds-Prandtl correlations are typically used). If the produce
surface water vapour pressure (Psurf) is not sufficiently constant, a
suitable mass transfer model which appropriately describes water
transport through the produce domain as a function of time will
need to be incorporated. Alternatively, it may be sufficient to
incorporate the effect of evaporation on heat transfer by removing
the 2nd term on right hand side of Equation (5) and incorporating
evaporative effects into the heat transfer coefficient (Alvarez and
Flick, 1999b).
83
It is implicit from Equations (1) through (5) that, for a produce
simulator to be capable of reproducing the absolute temperature
profiles of real produce, the flow, heat and mass transfer phe-
nomena must be sufficiently replicated. Specifically, the geometry,
evaporative cooling, produce respiration, and thermal properties of
each structure within the produce must be sufficiently matched.
Table 1 reports some of these properties and other relevant pa-
rameters for a range of horticultural produce.
The following sections of this review detail the use of produce
simulators in a sample of previous literature studies in relation to
the replication of these phenomena. Table 2 summarises this
comparison for the studies for which the thermophysical proper-
ties of the produce simulators employed were reported. It should
be noted that although the use of produce simulators is common in
postharvest research, these simulators are rarely the focus of such
research. Instead, they are typically employed as experimental
devices assisting in the meeting of a specific objective. For these
reasons it is not feasible to exhaustively review the previous use of
produce simulators. Rather, what is intended to be a representative
sample is discussed here.
3. Geometric matching
Geometric matching refers to the degree to which a produce
simulator matches the shape and size of produce. Since momentum
transport can be addressed as decoupled (Ferrua and Singh, 2009b,
c), geometric matching alone will allow the replication of flow
(5) fields, without the need to replicate thermal properties. Decoupling
further permits modelling procedures to be simplified and more
computationally feasibly by following the step by step approach of
(i) determination of air flow (ii) determination of transfer co-
efficients from flow fields and (iii) determination of heat and mass
transfer (Le Page et al., 2009). This type of modelling approach has
also been utilised in studies examining the air cooling of meat
products (Hu and Sun, 2000, 2001; Mirade et al., 2002; Trujillo and
Pham. 2006; Pham et al., 2009).
Attempts to utilise produce simulators which deliberately
replicate the shape of specific produce are sparse. Valente et al.
(1996) used avocado shaped produce simulators in their study of
heat transfer in avocados. All of the studies in Table 2 used idealised
 84
Table 1
Thermophysical parameters for various produce.

k (W/m K) a (x 106 m2/s) r (kg/m3) c (J/kgK) rc L (103 m) Bi q (103 W/kg) z0d (mL/kg hr) Ed (kJ/mol) Source(s)
(103 J/m3 K)

Apples 0.45a 0.14 840 3810 3200 19e44 0.46e4.11c 20e61 (at 10
C) 4.96  1013 65.7 Sweat (1974), ASHRAE (2006), Tabatabaeefar and
Rajabipour (2005)
Blueberries 0.54 0.14e0.15a 990 3720e4040 3682e4000 2e8 0.04e0.62c 101e183 (at 15
C) 9.53  1017 92.9 Marai et al. (2012), MacGregor (2005), ASHRAE (2006),
Oliveira et al. (2012)
Cherries 0.51e0.53a 0.13 1050 3730e3850 3917e4043 15 0.31e1.24c 81e148 (at 15
C) 4.96  1015 75.3 Milosevic and Milosevic (2012), ASHRAE (2006)
Kiwifruit 0.40e0.45a 0.11 940e1040 3893 3659e4049 21e26 0.51e2.73c 39 (at 10
C) 2.70  1011 58.7 Chen et al. (1998), ASHRAE (2006), Razavi and
BahramParvar (2007), Oliveira et al. (2012)
Melon 0.57 0.15a 930 4030 3747 64b 1.24e4.72c 22e46 (at 10
C) 1.42  1010 50.5 Eifert et al. (2006), Sweat (1974), ASHRAE (2006),
Oliveira et al. (2012)
Peaches 0.45a 0.12 930 3910 3636 23e44b 0.56e4.11c 47 (at 10
C) 1.12  1017 87.6 Sweat (1974), ASHRAE (2006), Alibeyglu et al. (2013)
Potatoes 0.50e0.51a 0.13 1040e1070 3670 3817e3927 28b 0.60e2.35c 20e62 (at 10
C) 4.54  108 41.6 Maurer and Eaton (1971), ASHRAE (2006)
Watere 0.14a

G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93

0.582 1000 4198 4198 n/a n/a n/a n/a n/a Incropera and Dewitt (2006)
Range 0.40e57 0.11e0.15 840e1070 3670e4040 3200e4049 2e64 0.04e4.72 20e183 n/a n/a n/a
a
Calculated from Equation (2) (k ¼ arc).
b
Calculated from reported surface area by assuming spherical. Values of thermal diffusivity and thermal conductivity are intended to represent the bulk portion of the produce.
c
Bi values calculated across a convective heat transfer coefficient range of 11e42 W/m2K which is the range reported by ASHRAE (2006) for a variety of produce being air cooled. The characteristic length used is half of the
shortest dimension.
d
Values taken from Exama et al. (1993) except for kiwifruit which was calculated from the work of Heyes et al. (2010).
e
The thermophysical properties of water at ~7
C are provided on Table 1 for comparison.

Table 2
Previous use of produce simulators, in the literature.

k (W/m K) a (106 m2/s) rc (103 J/m3 K) L (103 m) Bi Size Respiration Evaporation Application Source
matched simulation simulation

Acrylic spheres 0.21a 0.12 1731 38 2.00e7.60b n/a No No Determination of heat transfer Minh et al. (1969)
correlations for forced air cooling design
3% carraghenan gel spheres 0.86a 0.21 4153 38 0.49e1.86b n/a No Yes Model validation and development Alvarez and Trystram (1995)
of control design strategy for
produce cooling
Tylose filled aluminium cylinders 0.50 0.13a 3890 36e71 0.79e5.96b n/a No Yes Validation of chilling time prediction Chuntranuluck et al. (1998b)
model
Phenolic vinyl spheres 0.68 0.37a 1857 26 0.42e1.60b n/a No No Determination of cooling rate of Vigneault and Castro (2005)
produce for indirect airflow assessment
Celluloid spheres filled 0.52 0.12a 4153 19 0.40e1.53b n/a No No Validation of a mathematical Allais et al. (2006)
with 3% carraghenan gel model of mist chilling of foods
a
Calculated from Equation (2) (k ¼ arc).
b
Bi calculated across the same range of h described in Table 1. The characteristic length used is half of the shortest dimension.
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
spherical or cylindrical geometries, and were intended to be
applicable to produce which sufficiently approximates these
shapes, and/or used for the purposes of model validation.
Chuntranuluck et al., 1998b found little difference in heat transfer
observed in their produce simulators as a result of variation in
shape among the three idealised geometries investigated (spheres,
cylinders and slabs). The experiments conducted by Chuntranuluck
et al., 1998b involved placing produce simulators in a wind tunnel
which allowed homogenous flow to be delivered, and suggests in
such cases it may be sufficient for produce simulators to approxi-
mate the shape of real produce, however this is unlikely to be the
case in complex packaging structures.
The lack of available data means that the impact of variation in
produce geometry on the flow fields and subsequent heat transfer
in produce may be a valuable area of future investigation. The in-
tricacies of geometric variation in real produce dictate that such
data may be most readily tractable in silico using modern compu-
tational modelling techniques. Indeed, the determination of design
guidelines for produce simulators which consider the aspects of
geometry, evaporation, respiration and thermal matching is well
suited to the domain of computational modelling. It should be
noted that the accurate replication and mathematical description of
produce geometry is not necessarily a simple task and remains an
active area of research (e.g. Olatunji et al., 2015; Rogge et al., 2015).
Although produce simulators which replicate produce shape are
rare, most reported use of produce simulators attempts to
approximate produce size. This is evident in a comparison of
Tables 1 and 2 which shows the characteristic lengths of produce
simulators to be comparable with those of real produce. Although
the matching of characteristic length will likely allow approxima-
tion of internal heat transfer, the failure to match geometry will
result in different flow conditions which will result in some degree
of failure to replicate external convective heat transfer. Further-
more, for studies involving complex packaging structures, objects
of similar characteristic length but different shape may not be able
to be packed in the same arrangement or bulk volumes, resulting in
variation in both flow and thermal mass within a specified volume.
The simplicity afforded by replicating flow using thermally un-
matched produce simulators has seen them used in this capacity to
validate CFD models of flow through ventilated horticultural
packages (Ferrua and Singh, 2008; Delele et al., 2008) and for the
determination of heat transfer correlations and heat transfer co-
efficient determination. Minh et al. (1969) evaluated Nusselt-
Reynolds-Prandtl heat transfer correlations of bulk potatoes for
use in the design of precooling systems. As a comparative standard,
correlations were also evaluated for acrylic spheres which were
dispersed throughout the bed of potatoes. Significantly different
transfer coefficients and heat transfer correlations were found be-
tween spheres and potatoes which were positioned in similar lo-
cations in the airflow field. Although the thermophysical properties
of the acrylic spheres (Table 2) were not well matched to produce,
differences in heat transfer correlations are not expected to be due
to this, but to geometric differences (shape, dimension, roughness)
between potatoes and spheres. This is because for geometrically
similar objects such correlations are only dependant on the ther-
mophysical and flow properties of the fluid (Alvarez et al., 2003).
This was demonstrated by Valente et al. (1996) who measured
similar heat transfer coefficients for both real avocados and avo-
cado shaped aluminium objects with substantially different ther-
mal properties to produce.
Similar experiments were carried out by Alvarez and Flick
(1999b) to establish heat transfer correlations for approximately
spherical produce stacked in bins. PVC balls were used as produce
simulators along with aluminium balls and cylinders equipped
with thermocouples. PVC balls were also employed by Alvarez et al.
85
(2003) to examine flow heterogeneity and its relationship to heat
transfer coefficients. Although neither of these studies used mate-
rials that are thermally matched to produce, the results would be
expected to apply to geometrically similar produce and are an
appropriate use of produce simulators which are not thermally
matched.
Since flow fields can be difficult to measure in stacked produce,
Vigneault and Castro (2005) devised a methodology in which the
average approach velocity around a produce simulator could be
calculated from the rate at which the simulator cooled. The produce
simulators employed were solid phenolic vinyl spheres which had
similar size and thermal conductivity to that of produce but were
relatively poorly matched for thermal diffusivity and volumetric
heat capacity. The authors suggested that the size and thermo-
physical properties of a given produce would need to be matched
before it could be used to deduce airflow within a group of produce
or other produce simulators. However, theory suggests that a
geometrically matched produce simulator could have any known
thermal properties and be used to deduce airflow provided internal
temperature and air temperature local to the produce simulator
were known/measured. In theory, this provides a method by which
flow field data can be collected without impacting on the flow field
itself.
4. Evaporative cooling
As shown in the produce surface heat transfer boundary con-
dition (Equation (5)), the heat lost from produce by evaporation is
proportional to the driving force between the vapour pressure of
water in the air stream and that at the produce surface. These will
be determined by the temperature and relative humidity of the air
stream and the water activity of the produce surface respectively. In
addition, the rate of water loss at the surface of produce is related to
the resistance provided by the skin to transpiration and hence is
dependent on the type and state of the product (Maguire et al.,
2001).
Of the studies summarised in Table 2 only Alvarez and Trystram
(1995) and Chuntranuluck et al., 1998b attempted to simulate the
effects of evaporative cooling in their produce simulators. The
produce simulators employed by Alvarez and Trystram (1995) were
composed of 3% carrageenan gum, the exposed surface of which
allowed evaporation from a fully wetted surface to be simulated.
Chuntranuluck et al., 1998b simulated evaporative effects by
wrapping produce simulators in a layer of cloth which was soaked
in a wetting agent. A fully wetted surface was maintained during
the course of experimental trials by pumping a gentle flow of
wetting agent over the surface of simulators. Different surface
water activities were achieved by using various salt solutions as
wetting agents. Difficulties associated with crystallisation of these
salt solutions were reported by the authors. The complexity of this
system for the replication of evaporative effects would be chal-
lenging to implement on a large scale. A single immersion of a cloth
wrapped produce simulator in wetting agent at the beginning of a
trial may be a viable alternative to the replication evaporative ef-
fects during the course of produce precooling.
Le Page et al. (2009) used wetted plaster casts in their work
developing CFD models of stacked food products. These casts
maintained a constant surface water activity of 1.0 and were used to
provide a model of water transport through food and subsequent
surface evaporation. Although the thermal properties of the casts
were not detailed, similar systems may be useful as produce
simulators.
The studies of Alvarez and Trystram (1995), Chuntranuluck et al.,
1998b and Le Page et al. (2009) all simulated evaporative effects of a
fully wetted surface. In reality, all produce will vary somewhere
86 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
between the two extremes of a continuously fully wetted surface
and one where there is no surface evaporation and hence such
simulation is likely to result in greater evaporative losses than that
which would be observed for real produce. For some produce,
evaporation may initially be more significant at the beginning of
the pre cooling process but become less significant as the surface
moisture loss outstrips mass transport of water to the produce
surface. This effect has been demonstrated by Ansari et al. (1984)
for the cooling of apples and potatoes. In addition, as the product
temperature decreases, the partial pressure difference between the
product and airstream also reduces rapidly, meaning that the
importance of evaporative losses becomes less as cooling pro-
gresses. Evaporative heat losses are also less likely to be important
in products that have natural water loss resistance (Chuntranuluck
et al., 1998a). For example, Chuntranuluck et al. (1998c) showed
that evaporative heat losses resulted in peeled carrots cooling
significantly faster than unpeeled carrots whose modest skin
resistance was sufficient for them to behave essentially as a non-
evaporative surface.
A humidified air stream also reduces the driving force for
evaporation and hence reduces evaporative heat losses. Horticul-
tural cooling and storage processes are typically carried out at
relative humidity levels of >85% (Tanner et al., 2002c). Although
these humid environments minimise moisture losses from pro-
duce, the high latent heat of vaporisation of water dictates that very
little moisture loss is required to make a significant contribution to
precooling heat losses. Ferrua and Singh (2009c) reported that
moisture losses of approximately 0.5% from strawberries contrib-
uted 15e26% of the total heat lost during a 2 h precooling period.
For small fractional losses in moisture and assuming the latent
heat of vaporisation of water is constant over the temperature drop
associated with precooling, the percentage of heat removed from
produce can be expressed as a percentage of the heat load as,
M DHvap
V¼ (6)
cp DT
Over the duration of precooling typically used for common
agricultural products the loss in mass is less than 0.1% (Tanner et al.,
2002a). At a level of 0.1% moisture loss (M) Equation (6) can be used
to estimate the contribution of evaporative heat losses as approx-
imately 2e7%. The lower and upper end of this range were calcu-
lated at DHvap ¼ 2400  103 J/kg, cp ¼ 4040 J/kg K, DT ¼ 30
C, and
DHvap ¼ 2500  103 J/kg, cp ¼ 3670 J/kg K, DT ¼ 10
C respectively,
and provide a low and high estimate of variability in these pa-
rameters. The range in heat capacity comes from the observed
range of produce reported in Table 1. The range of 2e7% shows that
it is unlikely for typical precooling that evaporation represents a
significant portion of the heat removed, however, the significance
of evaporative heat losses increases rapidly as moisture loss in-
creases, with a percentage evaporative contribution ranging from
10 to 34% at 0.5% moisture loss. This range encompasses the 26%
contribution reported by Ferrua and Singh (2009c) at a similar
moisture loss level.
The percentage moisture loss from the product can be difficult
to determine experimentally as it can be confounded by water
absorption by corrugated fibreboard packaging and surface
condensation once cold. Alternatively, it can be estimated from the
rate of moisture loss taken over the duration of precooling as,
 
M ¼ 100kt Psurf  Pair Dt (7)
Estimation of the percentage evaporative contribution via
Equations (6) and (7) will produce a conservatively high estimate of
the upper bound of this value since moisture loss is assumed
constant over the period of precooling.
ASHRAE (2006) reports values of the transpiration mass transfer
coefficient (kt) for a variety of produce per unit mass. For example,
apples are reported as having kt values ranging from 35 to
58  1012 kg/kg s Pa. It should be noted that kt is also usefully
expressed per unit area for some purposes (see Equation (5) for
example). Using an average apple mass of 0.165 kg (Tabatabaeefar
and Rajabipour, 2005) which would have a surface area of
approximately 0.0169 m2 (Clayton et al., 1995), the range listed for
apples above can be expressed per unit area as, 34 to 57  1011 kg/
m2 s Pa. This range is included by the range reported by Maguire
et al. (2001) of 1e261  1011 kg/m2 s Pa based on a literature
survey of values reported for 24 different varieties of apple. The
width of this range highlights that even within the same type of
produce there is large variability and that values of kt are specific to
the circumstances under which they were measured and should be
used with caution between systems (Maguire et al., 2001). There-
fore, such values should only be considered appropriate for per-
forming preliminary design calculations. For this reason, the
transpiration mass transfer coefficient is more typically estimated
by assuming it to consist of an air film mass transfer coefficient and
a produce skin mass transfer coefficient,
1
kt ¼ 1
(8)
kair
þ k1
skin
Modelling kt in this way allows the effects of air flow and skin
resistance to be separately accounted for when estimating kt, and
various methods of estimation including calculation from the heat
transfer coefficient using the classical Lewis relation (Lewis, 1922)
are discussed elsewhere for produce systems by Becker et al.
(1996), Valente et al. (1996), Chuntranuluck et al. (1998a),
ASHRAE (2006), Ferrua and Singh (2009a), Le Page et al. (2009),
and Dehghannya et al. (2010).
It should also be noted that there is potential for moisture loss to
alter the thermal properties of produce during precooling. How-
ever, given the typically small moisture losses associated with
precooling, these properties are usually assumed constant and
unaffected by moisture loss (Tanner et al., 2002b; Ferrua and Singh,
2009c). Furthermore, moisture losses of a magnitude that would
significantly affect thermal properties would likely result in prod-
uct loss through, for instance, shrivel development (Maguire et al.,
2001; Aswaney, 2007) or firmness degradation (Paniagua et al.,
2013).
Ultimately, whether or not evaporation plays a significant role
will be determined by the particular precooling system. However, it
is likely that for many precooling processes the effects of evapo-
rative cooling can be omitted from produce simulators without a
major effect on the observed heat transfer. For many applications
the effect of evaporation on temperature profiles and/or the time
required to achieve a certain degree of cooling is of most impor-
tance and may be small in some cases even where the contribution
relative to the overall heat load seems significant. Once again,
design guidelines in this regard will be very useful and can likely be
established using computational modelling.
5. Addition of respiratory heat
None of the studies reported in Table 2 attempted to incorporate
the addition of heat of respiration into the produce simulators
employed. Indeed, the assumption that heat added via produce
respiration is a negligible portion of the heat load is commonly
made (Tanner et al., 2002a; Ferrua and Singh, 2009a; Dehghannya
et al., 2010). Tanner et al. (2002a) suggest that in a typical pre-
cooling process the heat added by respiration is ~0.5% of the heat
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
load.
Assuming the respiratory heat generation is constant, the
approximate heat added by respiration over the period of pre-
cooling can also be expressed as a percentage of the heat load,
100qDt
Q¼
cp DT
Table 1 gives values of the heat of respiration for various pro-
duce at temperatures approximately mid-range to the typical pre-
cooling process. Using these values, a minimum and maximum
range for Q can be calculated as varying from 0.03 to 22%. Where
the lower and upper values were calculated at q ¼ 20  103 W/kg,
Dt ¼ 0.5 h, cp ¼ 4040 J/kg K, DT ¼ 30
C, and q ¼ 183  103 W/kg,
Dt ¼ 12 h, cp ¼ 3670 J/kg K, DT ¼ 10
C respectively. The range in
heat capacity again comes from the observed range of produce
reported in Table 1.
The calculated range is very wide and whilst this range includes
the value of 0.5% suggested by Tanner et al. (2002a) as being typical
of produce cooling, it also extends well beyond this. The width of
this range is due almost entirely to variation in the duration of
precooling (the upper value of this range becomes 0.90% if 0.5 h is
used instead of 12 h), showing that the assumption of a negligible
contribution of respiratory heat made by most studies is much
more likely to be valid over shorter durations of precooling, but
likely to be untrue for lengthy precooling processes. This will be
more likely in areas of maldistribution of flow which results in slow
cooling. In these cases, the contribution of evaporative cooling
would also be more significant.
However, as with evaporative losses, because the interest in
produce simulators is primarily to replicate temperature profiles
during precooling, the effect of respiration on temperature needs to
be quantified and design guidelines established. Using an analytical
treatment, Awbery (1927) showed that an apple placed in a
refrigerator had only 0.02
C excess temperature at its centre as a
result of respiration after 2 h of cooling. Similar research utilising
computational models should be able to produce appropriate
design guidelines demonstrating the impact of product respiration
on temperature at times scales of interest to produce precooling.
Because evaporation and respiration represent heat loss and
addition respectively, it is possible that their net effect may result in
an overall negligible effect on heat transfer in some applications.
The constant values of q used in calculating the range above will
also tend to provide a conservative overestimate of the contribution
of respiration, since respiratory heat generation reduces with
temperature. This effect can be incorporated into dynamic models
using Arrhenius type temperature dependence which studies have
shown describes the relationship between produce respiration and
temperature well (Bhande et al., 2008; Benítez et al., 2012;
Waghmare et al., 2013, Azevedo et al., 2015), where the produc-
tion of carbon dioxide can be described in the form,
 in ice water for gel formation and returned to ambient conditions
E
z ¼ z0 exp
RT
Carbon dioxide production rates can be converted to rates of
respiratory heat generation based on the amount of energy
released during respiration. Using a value of 470 kJ/mol of respi-
ratory carbon dioxide produced (Gross et al., 2004), the ideal gas
law, and Equation (10), the rate of respiratory heat generation can
be calculated at a given temperature as,
87

fz0 exp E
RT
q¼ (11)
T
where f is a conversion factor equal to 1.59 (J hr K/s mL). Values for
the pre-exponential factor and activation energies can be readily
calculated from measured carbon dioxide production rates at
various temperatures. Gross et al. (2004) provides a source of such
(9)
measured data for fruit and vegetables. Using data from a variety of
literature sources, Exama et al. (1993) calculated the pre-
exponential factor and activation energies for different produce,
some of which are displayed in Table 1.
Other approaches to calculating the rate of respiratory heat
generation include use of the empirical relationship reported by
Gaffney et al. (1985),
q ¼ a(T255.35)b (12)
This relationship has a high correlation coefficient and so is
widely used to calculate the heat of respiration of food products
(Tanner et al., 2002b). The product specific constants, a and b, can
be determined experimentally or estimated from existing data. For
example, Tanner et al. (2002b) obtained these values via regression
of data obtained from Hardenburg et al. (1986).
6. Thermal property matching
6.1. Single-domain produce simulators
Single domain produce simulators are constructed of one ma-
terial, or if constructed of multiple materials, only one part has
significant resistance to heat transfer and thermal mass. These
simulators can be employed to replicate heat transfer in produce
whose heat transfer properties are sufficiently homogenous
throughout.
In order to replicate internal heat transfer, a thermally matched
produce simulator will sufficiently match the thermal conductivity,
thermal diffusivity and volumetric heat capacity of the produce
being simulated. These thermal properties are related by,
k ¼ a(rc) (13)
and hence matching any two of these properties will also match the
third.
The range reported in Table 1 shows that the bulk thermal
properties of produce vary little across different types of produce.
The mid-range values of k ¼ 0.49 W/mK, a ¼ 0.13  106 m2/s and
(rc) ¼ 3625 kJ/m3 K are considered here to be representative of
typical produce.
A number of studies have attempted to use single domain pro-
duce simulators which sufficiently replicate the thermal properties
of produce. Alvarez and Trystram (1995) used spheres composed of
3% carrageenan gum solution to simulate approximately spherical
produce cooling in bins. Gel spheres were prepared by injecting the
solution into aluminium moulds which were subsequently cooled
(10) before experimental runs. The heat transfer resistance and thermal
mass of the aluminium mould can be considered negligible and
hence these produce simulators are considered single-domain. The
thermal diffusivity of the gel was about 1.5 times greater than that
of typical produce, whilst the volumetric heat capacity was very
similar to that of produce (Table 2). Since these produce simulators
were spherical and do not incorporate respiration or evaporation,
the cooling profile can be analytically represented by (Incropera
88 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
and Dewitt, 2006),
X
∞   1  
sin dn r *
2
Y¼ Cn exp dn Fo *
(14)
n¼1
d n r
where,
4½sinðdn Þ  dn cosðdn Þ
Cn ¼ (15)
2dn  sinð2dn Þ
And the values of dn are the positive roots of the equation,
1dncot(dn) ¼ Bi of which the 1st root is sufficient to describe heat
transfer at longer time intervals (Fo > 0.2).
Using Equation (14) it can be calculated that, at a convective
heat transfer coefficient value of 27 W/m2K (mid-range of the
values reported by ASHRAE, 2006 for cooling of a wide variety of
produce), the spheres used by Alvarez and Trystram (1995) would
take 90 min to cool to 87.5% (7/8th) of the overall temperature
change at the centre when cooled from 25 to 0
C. If the spheres
were thermally matched to typical produce (k ¼ 0.49 W/m K,
(rc) ¼ 3625 KJ/m3K) this time would be 104 min. Hence the pro-
duce simulator predicts cooling to be approximately 14 min too fast
as a result of thermal mismatching (this discrepancy will vary at
different convective heat transfer coefficients). Whether or not this
is significant will depend on the intended application, which in the
case Alvarez and Trystram (1995) was to demonstrate a design
strategy for controlling forced air cooling of produce. Because the
produce simulators were not thermally matched, absolute tem-
peratures and cooling rates should at least be extrapolated to other
systems with this in mind.
Table 2 shows the values reported by Alvarez and Trystram
(1995) and Allais et al. (2006) for the thermal conductivity and
diffusivity of 3% carraghenan gel differ, indicating that these values
may need to be more accurately quantified. The value reported by
Allais et al. (2006) of 0.52 W/m K is much closer to that of typical
produce than the value of 0.86 W/m K reported by Alvarez and
Trystram (1995). If this value is instead used in the calculation of
the time to cool to the centre to 87.5% of the overall temperature
change when cooled from 25 to 0
C, this time increases to 109 min
and the cooling is now 5 min too slow as a result of thermal mis-
matching. Reducing the thermal conductivity even further so that it
matches typical produce at 0.49 W/m K actually makes the match
worse by increasing the cooling time to 112 min. This is because the
reduction in thermal conductivity acts to slow heat transfer, which,
in this case is the same effect as increasing the thermal mass of the
produce simulator which is larger than that of typical produce. In
the case of the simulators used by Alvarez and Trystram (1995), a
produce simulator with thermal conductivity of 0.58 W/mK leads
to the time required to achieve 87.5% temperature change
becoming equal to that of typical produce. If the time required to
achieve a specified degree of temperature change at a specific
location is a suitable measure of cooling (as opposed to matching
the entire temperature profile), then this raises the possibility of a
produce simulator with thermal conductivity and volumetric heat
capacity combinations which are either both higher or both lower
than produce. However, such a simulator would only achieve this
match at a single flow condition. In the example above, the produce
simulator with thermal conductivity of 0.58 W/m K is matched at
h ¼ 27 W/m K and becomes 3 min too slow at h ¼ 20 W/m K and
4 min too slow at h ¼ 15 W/m K. The sensitivity of the time to
achieve a certain fraction of temperature change to changes in
thermal properties is further discussed in Section 7 below.
It is also important to note that although the study of Alvarez
and Trystram (1995) employed gel spheres, the majority of
spheres in the study were air filled PVC balls of the same size as the
gel spheres. These were chosen to give similar aerodynamic
behaviour. Although they would achieve this, their different ther-
mal properties will affect heat transfer within the system such that
the temperature profiles would be expected to be different to what
would have been observed had all spheres been composed of gel.
The magnitude of any such discrepancies is a worthy subject of
further investigation and if not too large may permit large scale
experiments with a few thermally matched produce simulators
amongst a bulk of aerodynamic placeholder simulators.
Chuntranuluck et al., 1998b devised an extremely comprehen-
sive produce simulator for use in validating their chilling time
prediction model. Intended to be a general food analogue, simu-
lators were constructed of various size aluminium cylinders filled
with Tylose. Aluminium was chosen to confine Tylose to the desired
shape without introducing significant heat transfer resistance or
thermal capacity. Table 2 indicates that the Tylose simulator of
Chuntranuluck et al., 1998b is very closely thermally matched to
typical produce. This suggests that Tylose based simulators with
housings which match the shape and size of specified produce
would represent a very good produce simulator from a technical
perspective, provided the resistance to transfer of the housing was
negligible or well matched to the produce. The practicality of such a
produce simulator would be impacted by its stability. For example,
if frequent cleaning and refilling were required, regular use on a
large scale may not be feasible. In addition, by their nature any
water based produce simulator is a candidate for shelf stability is-
sues associated with microbial growth.
6.2. Multi-domain produce simulators
Accurate representation of some produce may require it to be
split into multiple subdomains, such as skin, flesh and core. This
will be required for any region in which the thermophysical prop-
erties differ significantly enough from adjacent regions. If such
subdomain differentiation is required, then a separate set of
transfer equations can be written for each subdomain (Equation (4)
with new subdomain properties plus internal boundary conditions)
which in turn dictates that a produce simulator will need to
replicate the geometry and thermophysical properties of each
subdomain if the internal temperature profiles of produce are to be
replicated.
It appears that multi domain produce simulators in the litera-
ture have not necessarily been deliberate attempts to mimic
structural components of produce. Rather they have been restricted
to plastic housings employed as a convenient container for another
water based material. Whether deliberate or not, such produce
simulators may be a convenient way of replicating heat transfer in
produce where the properties of the skin differ significantly from
the bulk produce.
Allais et al. (2006) used simulators composed of 3% carraghenan
gel housed in celluloid spheres to validate their mathematical
model of stacked spheres during mist chilling (a variation of forced
air cooling where the air stream contains suspended water drop-
lets). Although the carraghenan gel provides a reasonable thermal
match with produce, unlike aluminium, the resistance and thermal
mass of the housing is unlikely to be negligible in this case. The
effectiveness of the produce simulator will depend on whether the
housing sufficiently matches the bulk produce, or the skin of the
produce if this is sufficiently different from the bulk. Sufficient
replication of the thermal properties of the skin of produce may
have played a role in the work of Vissotto et al. (2000) who
employed PVC balls filled with 3% carrageenan gum as produce
simulators and found that the temperature profiles produced dur-
ing precooling compared well to those of oranges. This is despite
some degree of thermal mismatching in carrageenan gum based
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
systems as discussed above. In this case the PVC housings may have
improved the match by simulating the skin of oranges which have a
low thermal conductivity of 0.18 W/m K (ASHRAE, 2006), which is
very similar to that of PVC (Table 3). Vissotto et al. (2000) also
attempted to produce set gel based simulators from carrageenan
gum without PVC housings, but experienced issues associated with
movement after stacking, evaporation and fungal growth.
Castro et al. (2004) used a similar experimental setup to Alvarez
and Trystram (1995) with the objective of examining which pa-
rameters effect the cooling rate and uniformity of stacked horti-
cultural produce. 28 PVC balls filled with 3% agar-agar solution
were distributed throughout a stacking of 224 hollow PVC balls.
Whilst the thermal properties of 3% agar are similar to those of
water (Denys et al., 2000) and hence produce (see Table 1), the
authors concluded that the presence of air filled balls would affect
heat transfer and dictate that their findings should be used care-
fully. Practical difficulties were also encountered, with the authors
attributing significant experimental errors to difficulties in uni-
formly filling PVC balls with agar solution and recommended that
the accuracy and reproducibility of such studies could be improved
through the design of more stable produce simulators.
Due to the potential increased complexity of creating multi
domain produce simulators, it should be verified that the ther-
mophysical properties are sufficiently variable to justify this. In
many cases, despite stark differences in the visual or structural
properties of regions of produce, the thermal properties do not
differ drastically. An example comes from a comparison of the pulp
and stone of avocado. These structures differ in composition and
also have obvious visual and physical differences. However, data
reported by Valente et al. (1996) shows the thermal conductivity of
the pulp and stone to be 0.43 and 0.45 W/m K respectively, with
volumetric heat capacitates of 3380 kJ/m3 K and 3240 kJ/m3 K
respectively. Hence, despite stark differences these two structures
appear to be closely thermally matched. The difference in assuming
a central stone within such an object had similar properties to the
pulp may be negligible for the purposes of most applications.
Another example comes from the thermal diffusivity of kiwifruit
flesh and core which Chen et al. (1998) found to be similar at
1.14  107 and 0.92  107 m2/s respectively. The water contents
of these structures are also similar at approximately 85% and 75%
respectively (MacCrae et al., 1989), and hence their volumetric heat
capacities would also be expected to be similar. This would suggest
an overall thermal similarity, and whether or not these similarities
are sufficient to justify a single domain produce simulator would be
worth considering before pursuing a more complex alternative.
Indeed, reasonable thermal matching would be expected wherever
different structures have similar water contents which will dictate
to a large degree the thermophysical properties. The thermophys-
ical properties of water at approximately 7
C are provided in
Table 1 for comparison, and show the expected similarity between
water and produce.
Whether single or multi-domain, the sensitivity of produce
simulators to thermal mismatching in pre-cooling is an important
topic for future research which, once again will be well suited to
computational modelling. These aspects are discussed further in
Section 7 when various simplifying assumptions are permitted.
Table 3
Thermophysical properties of common thermoplastics.
Type of thermoplastic k (W/m K) r (kg/m3)
High density polyethylene (HDPE) 0.543 (at 36
C) 950
Low density polyethylene (LDPE) 0.31 (at 25
C) 918
Polypropylene (PP) 0.11 (at 25
C) 908
Polyvinyl Chloride (PVC) 0.19 (at 25
C) 1400
89
6.3. Materials for thermal property matching
The studies discussed here which achieved reasonable thermal
matches between bulk produce properties and simulator used
water based materials (Alvarez and Trystram, 1995; Chuntranuluck
et al., 1998b; Castro et al., 2004; Allais et al., 2006). This is because
full thermal matching is constrained by the requirement of
matching volumetric heat capacity. The high water content of
produce dictates that its volumetric heat capacity is similar to that
of water. As very few materials have volumetric heat capacities as
high or higher than water, produce simulators have naturally been
water based.
The difficulties associated with the preparation and stability of
water based produce simulators, make produce simulators con-
structed of alternative materials a desirable proposition. However,
it is unlikely that a pre-existing material will be initially well
matched to the desired produce. Composite materials offer a po-
tential way to construct a desired simulator material and overcome
this issue.
A simple two component composite material is constructed by
combining a base matrix material with a filler material. The volu-
metric heat capacity of a two component composite can be
described by the law of mixtures,
ðrcÞc ¼ fðrcÞf þ ð1  fÞðrcÞm (16)
The law of mixtures is generally insufficient to predict the
thermal conductivity of composite materials. Many empirical and
experimental models describing the effective thermal conductivity
of composite systems have been reported in the literature. The
applicability and effectiveness of such models have been sum-
marised by Progelhof et al. (1976) and Kumlutas et al. (2003). The
field continues to develop and utilise modern modelling techniques
such as neural networks (Bhoopal et al., 2012).
An example of such relationships is the classical model of
Maxwell (Maxwell, 1873),
2  3
kf þ 2km þ 2f kf  km
kc ¼ km 4   5 (17)
kf þ 2km  f kf  km
Composite design can be achieved by solution of these equa-
tions. At any given filler fraction, for a produce simulator of speci-
fied desired thermal properties and a base matrix material of
known thermal properties, Equation (16) can be used to solve for
the required volumetric heat capacity of a filler material and
Equation (17) for the required thermal conductivity of the filler (in
the case of Maxwell’s equation this requires an iterative solution).
In practice, a thermal conductivity model that is appropriate to
the particular design scenario being investigated should be sought
and its limits of applicability should be understood. For example,
Maxwell’s model is able to predict the effective thermal conduc-
tivity of composites well at lower filler concentrations (Ebadi-
Dehaghani and Nazempour, 2012; Kumlutas et al., 2003) and as-
sumes that the filler particles are randomly distributed and non-
interacting homogenous spheres in a homogeneous medium
c (J/kg K) rc (103 J/m3 K) Source
1800e2700 1710e2565 Kumlutas et al. (2003)
1900e2300 1744e2111 Luyt et al. (2006)
1920 1743 Ebadi-Dehaghani and Nazempour (2012)
840e1170 1176e1638 Ebadi-Dehaghani and Nazempour (2012)
90 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
(Kumlutas et al., 2003).
Thermoplastics are an attractive proposition for use as a com-
posite matrix material due to their ability to be moulded into shape
and recast into different shapes if required. Furthermore, the
modification of thermoplastic properties through the creation of
composites is common. This includes research into the modifica-
tion of thermal conductivity, especially by combining with metals
(Mamunya et al., 2002; Kuriger and Alam, 2002; Kumlutas et al.,
2003; Weidenfeller et al., 2004; Boudenne et al., 2005; Luyt et al.,
2006). Little or no research has investigated modification of the
volumetric heat capacity via the creation of thermoplastic
composites.
The thermal conductivity of thermoplastics is similar to that of
produce, while the volumetric heat capacity is typically less than
half (Table 3). According to the design equations above, a produce
simulator built on a thermoplastic matrix would require a filler
material with thermal mass higher than water but with comparable
thermal conductivity. The rarity of this type of material may limit
the feasibility of thermoplastic based produce simulators. Ther-
moplastic produce simulators may be better suited to Biot matched
simulators, as discussed in Section 7.
Other potential matrix materials for the creation of composite
produce simulators are those with high volumetric heat capacity.
These include stainless steel and ceramic materials. Some ceramic
materials have volumetric heat capacities comparable to and
greater than water/produce (Warren, 1992) and thermal conduc-
tivity is typically much higher than that of water based materials
(Carter and Norton, 2013). The same is true of stainless steel, hence
both ceramics and stainless steel could in theory be used to create a
composite produce simulator by adding a low thermal mass, low
thermal conductivity material. The greater the thermal conduc-
tivity of the initial matrix above that of produce, the greater the
fraction of low thermal conductivity filler required to reduce the
thermal conductivity to the levels of produce. The more filler
required, the closer its volumetric heat capacity will have to be to
that of produce. Hence, in these cases a low thermal conductivity
material with thermal mass comparable to water/produce may be
required, and once again the rarity of this type of material may be
problematic.
However, even if ideal filler materials do not exist, such com-
posites may be worth investigating, especially if the sensitivity of
produce simulator temperature replication permits a certain de-
gree of property mismatching. Complex composites composed of
more than two materials may also provide potential solutions.
It is finally interesting to note the potential of using produce
itself as a produce simulator material. Potatoes, for example have
similar thermal properties to other produce (Table 1) and might be
useful to simulate different produce which is less available and/or
more expensive.
7. Simplifying cases
Effective and convenient produce simulators which replicate the
effects of evaporation, respiration and include multiple domains of
varying thermal properties will likely be difficult to achieve from a
practical perspective. The expense of such simulators may also limit
their application, especially for research on a pallet scale.
The complexity of produce simulator design reduces consider-
ably if evaporation and respiration can be assumed negligible, and
the produce thermal properties can be considered sufficiently ho-
mogenous to justify a single-domain produce simulator. Although
further investigation is required, the information discussed in this
review suggests that these assumptions may be valid for some
produce. This section considers the implications for simulator
design when these assumptions can be considered valid.
Fig. 4 shows the sensitivity of the duration required to achieve
87.5% centre cooling of typical produce to variation in some pa-
rameters when cooled from 25 to 0
C. Calculations were carried
out using Equation (14) with parameters values held constant at
h ¼ 27 W/m2 K, k ¼ 0.49 W/m K, (rc) ¼ 3625 kJ/m3 K, and Bi ¼ 1.82,
except when they were varied as the dependant variable.
Although the absolute magnitude of these predictions are
technically applicable to spherical geometry, the observed re-
lationships offer useful information for produce simulator design.
Fig. 4a shows the cooling time is highly sensitive to variation in the
convective heat transfer coefficient, increasingly so as flow condi-
tions stagnate at low values. This serves to highlight the importance
of replicating flow through geometric matching in produce simu-
lator design, but also the importance of the challenge posed in
ventilated packaging design trying to achieve homogenous flow to
all parts of the package in order to achieve cooling uniformity.
Fig. 4b shows cooling time rapidly becomes highly sensitive to
variation in thermal conductivity at values less than that of typical
produce (~0.5 W/m K), but becomes relatively insensitive as ther-
mal conductivity increases beyond this level. The opposing effects
of thermal conductivity and volumetric heat capacity are clearly
evident through comparison of Fig. 4b and c, with increasing
thermal conductivity reducing cooling time and increasing volu-
metric heat capacity increasing it. These figures do suggest that
there is some scope for thermal mismatching with small deviations
in thermal conductivity and volumetric heat capacity in the vicinity
of produce not leading to large deviations in cooling times. The
linear sensitivity to volumetric heat capacity is described by,
FoðrcÞro2
t¼ (18)
k
where t is the time required to achieve a specified degree of
completion (Y) and Fo is the dimensionless time at the same degree
of completion (Fo is constant when (rc) is varied). Hence for any
system where the time required to achieve a certain degree of
completion is known (either by prediction or experimentally), the
time required to achieve the same degree of completion as a result
of variation in volumetric heat capacity can be calculated as,
ðrcÞ2
t2 ¼ t1 (19)
ðrcÞ1
This relationship can be utilised to quickly determine the effects
of mismatching of the volumetric heat capacity of a produce
simulator on the cooling time.
Under the assumptions of negligible evaporation, respiration
and a single homogenous domain, further simplification of produce
simulator requirements may be permitted at extremes of the ratio
of internal to external heat transfer resistances in horticultural
produce. This ratio is described by the Biot number,
Bi ¼ hL/k (20)
and can be used to divide the heat transfer processes involved into
three categories (Vigneault and Castro, 2005), Bi ≪ 1, Bi [1 and
Bi z 1.
When Bi ≪ 1, convection limits heat transfer. In this case,
assuming geometry and airflow are mimicked by the produce
simulator, a temperature profile which matches that of produce can
be achieved using any thermal conductivity, provided the volu-
metric heat capacity is matched. Hence, only a partial thermal
match is required.
When Bi [ 1 conduction limits heat transfer. In this case a
temperature profile which matches that of produce can be achieved
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93 91

Fig. 4. Sensitivity of the duration required to achieve a 87.5% cooling of typical produce to variation in some parameters. Calculations made utilising Equation (14). Parameters held
constant at h ¼ 27 W/m2K, k ¼ 0.49 W/mK, (rc) ¼ 3625 KJ/m3K, and Bi ¼ 1.82 when not the dependant variable and cooling is from 25 to 0
C. Black circles indicate the location of
typical produce.

by only matching thermal diffusivity, provided the Bi numbers are Y ¼ f(dimensionless position, Fo, Bi) (21)
equal. However, if flow and geometry are also to be matched, then
Bi equality can only be achieved through matched thermal con- where Y is the fraction of unaccomplished change in temperature,
ductivity. Hence, in order to match the thermal diffusivity, the
volumetric heat capacity would also need to be matched. The net T  T∞
Y¼ (22)
result is that the requirements of matching flow and geometry force Ti  T∞
this scenario back to one where full property matching is required,
which offers no simplification for the purposes of produce simu-
lator design.
Using Equation (20), Fig. 5 plots the Bi number for each produce
item in Table 1 as a function of convective heat transfer coefficient.
Where a range of L and/or k have been given in Table 1, mid-range
values have been used in the calculations of Fig. 5. The Bi lines of all
produce in Fig. 5 cross the line Bi ¼ 1 (dotted line in Fig. 5) during
the convective heat transfer coefficient range of 11e42 W/m K re-
ported by ASHRAE (2006) for a variety of produce being cooled by
forced air chilling, which demonstrates that the case of Bi z 1 is
likely to apply to typical produce systems. Table 1 also gives a Bi
range for each individual produce item calculated across the heat
transfer coefficient range of 11e42. For blueberries this range is
from 0.04 to 0.62 (low range value is calculated at low h, L and high
k and vice versa for the high range value, thus creating a conser-
vatively wide range). This range also falls into the category of Bi z 1,
except for blueberries at the extreme low end of the range.
Therefore, the simplifying case of Bi ≪ 1 will not be valid to apply in
produce simulator systems. Rather, the dimensionless temperature Fig. 5. Biot numbers of various produce as a function of convective heat transfer co-
will be a function of dimensionless position, dimensionless time efficient. Values for L and k are taken from Table 1. Where a range is given in Table 1 the
and Biot number. That is, mid range value has been used. 1The lines for potatoes and kiwifruit are visually
indistinguishable. The dotted line is that of Bi ¼ 1.
92 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
and dimensionless time is defined by the Fourier number as,
∝t
Fo ¼
L2
In this case, if a system composed of produce simulators is of the
same scale and geometrically similar, then for a given flow field it
will produce the same dimensionless temperature profile as the
system it is simulating provided the thermal conductivities are
matched. This is because the Biot numbers will be matched. This
allows the possibility of utilising produce simulators which only
match the thermal conductivity of the produce being simulated to
carry out experiments, with post processing reconstruction of the
absolute temperature profile. Table 3 suggests high density poly-
ethylene (HDPE) would be well suited to this purpose given its
thermal conductivity is similar to produce and it can be moulded
into any shape desired. Fig. 4d also shows the sensitivity of the
dimensionless time required to achieve 87.5% temperature change
to variation in the Biot number. At Biot numbers greater than ~2
this dimensionless time is relatively insensitive to variations in the
Biot number, indicating that there may be some scope for Biot
mismatching when operating at Biot numbers around and beyond
this level.
Furthermore, for these systems, heat transfer theory can be used
to show that, given geometric similitude, two systems of different
characteristic length can also produce the same dimensionless
temperature profiles. For example, this will be achieved in a given
system if characteristic length is scaled by a factor of s, fluid velocity
is scaled by a factor of 1/s and the thermal conductivity is matched.
This allows for the possibility of carrying out experiments on a
smaller scale and only partially thermally matching produce, which
may offer further practical advantages.
8. Conclusions
Produce simulators have the potential to accelerate research
progress, especially in the areas of postharvest cooling and tem-
perature control operations. However, a survey of the literature has
shown there is a lack of appropriate design guidelines for produce
simulators, which in some cases has resulted in inappropriate use.
Consideration of transport theory as it applies to forced air
cooling allows design constraints to be established for produce
simulators. Geometric matching allows the replication of flow
fields. Combined with thermal property matching this allows the
replication of heat transfer in systems in which evaporative and
respiratory heat losses are negligible. Future work examining the
extent to which geometric and thermal parameters can be
matched/mismatched and still achieve suitable replication of the
desired transport phenomena is recommended, as is further
research examining the scenarios under which the simplifying as-
sumptions of negligible evaporation, respiration and produce ho-
mogeneity are considered appropriate. These issues can likely be
addressed with the assistance of modern computational modelling
which allows aspects like the effect of variation in the geometry of
produce to be established in silico, where such information might
be otherwise difficult to obtain experimentally.
Practical constraints including the ability to be produced on a
large scale at low cost, convenient experimental setup, long life and
reusability are also important to produce simulator systems. If
multiple types of different produce are to be simulated the ability to
be reshaped and manipulate thermophysical properties may also
be desired. Meeting these requirements represents a materials
engineering challenge to which composite materials are one po-
tential solution.
Acknowledgements
This work is an output of the Ministry of Business, Employment
(23)
and Innovation funded Fibreboard Packaging Design research
project (MAUX 1302).
References
Alibeyglu, S.Z., Ghaffari, H., Alipasandi, A., 2013. Estimating surface area of three
peach varieties using allometric relationships obtain from project area. Int. J.
Agron. Plant Prod. 4 (8), 1978e1984.
Allais, I., Alvarez, G., Flick, D., 2006. Modelling cooling kinetics of a stack of spheres
during mist chilling. J. Food Eng. 72 (2), 197e209.
Alvarez, G., Flick, D., 1999a. Analysis of heterogeneous cooling of agricultural
products inside bins Part I: aerodynamic study. J. Food Eng. 39 (3), 227e237.
Alvarez, G., Flick, D., 1999b. Analysis of heterogeneous cooling of agricultural
products inside bins Part II: thermal study. J. Food Eng. 39 (3), 239e245.
Alvarez, G., Trystram, G., 1995. Design of a new strategy for the control of the
refrigeration process: fruit and vegetables conditioned in a pallet. Food Control.
6 (6), 347e355.
Alvarez, G., Bournet, P., Flick, D., 2003. Two-dimensional simulation of turbulent
flow and transfer through stacked spheres. Int. J. Heat Mass Transf. 46,
2459e2469.
Ambaw, A., Delele, M.A., Defraeye, T., Ho, Q.T., Opara, L.U., Nicolaï, B.M., Verboven, P.,
2013. The use of CFD to characterize and design post-harvest storage facilities:
past, present and future. Comput. Electron. Agric. 93, 184e194.
Ansari, F.A., Charan, V., Varma, H.K., 1984. Heat and mass transfer analysis in air-
cooling of spherical food products. Int. J. Refrig. 7 (3), 194e197.
ASHRAE, 2006. Thermal properties of foods. In: ASHRAE Handbook e Refrigeration
(SI). American Society of Heating, Refrigerating and Air Conditioning Engineers.
Aswaney, M., 2007. Forced-air precooling of fruits and vegetables. Air Cond. Refrig. J.
57e62. January-March.
Awbery, J.H., 1927. The flow of heat in a body generating heat. Lond. Edinb. Dublin
Philosophical Mag. J. Sci. 4 (22), 629e638.
Azevedo, S., Cunha, L.M., Fonseca, S.C., 2015. Modelling the influence of time and
temperature on the respiration rate of fresh oyster mushrooms. Food Sci.
Technol. Int. 21 (8), 593e603.
Becker, B.R., Misra, A., Fricke, B.A., 1996. Bulk refrigeration of fruits and vegetables,
Part I: theoretical considerations of heat and mass transfer. Int. J. HVAC R Res. 2
(2), 122e134.
Benítez, S., Chiumenti, M., Sepulcre, F., Achaerandio, I., Pujola , M., 2012. Modeling
the effect of storage temperature on the respiration rate and texture of fresh cut
pineapple. J. Food Eng. 113 (4), 527e533.
Bhande, S.D., Ravindra, M.R., Goswami, T.K., 2008. Respiration rate of banana fruit
under aerobic conditions at different storage temperatures. J. Food Eng. 87 (1),
116e123.
Bhoopal, R.S., Sharma, P.K., Kumar, S., Pandey, A., Beniwai, R.S., Singh, R., 2012.
Prediction of effective thermal conductivity of polymer composites using an
artificial neural network approach. Special Top. Rev. Porous Media 3 (2),
115e123.
Boudenne, A., Ibos, L., Fois, M., Majeste, J.C., Gehin, E., 2005. Electrical and thermal
behaviour of polypropylene filled with copper particles. Compos. Part A 36 (1),
1545e1554.
Brosnan, T., Sun, D., 2001. Precooling techniques and applications for horticultural
products e a review. Int. J. Refrig. 24 (2), 154e170.
Carter, C.B., Norton, M.G., 2013. Conducting charge or not. In: Ceramic Composites:
Science and Engineering. Springer, New York.
Castro, L.R., Vigneault, C., Cortex, L.A.B., 2004. Container opening design for horti-
cultural produce cooling efficiency. J. Food, Agric. Environ. 2 (1), 135e140.
Chen, X.D., McLellan, D.N., Rahman, M.S., 1998. Thermal diffusivity of kiwifruit skin,
flesh and core measured by a modified fitch method. Int. J. Food Prop. 1 (2),
113e119.
Chuntranuluck, S., Wells, C.M., Cleland, A.C., 1998a. Prediction of chilling times of
foods in situations where evaporative cooling is significant e Part 1. Method
development. J. Food Eng. 37, 111e125.
Chuntranuluck, S., Wells, C.M., Cleland, A.C., 1998b. Prediction of chilling times of
foods in situations where evaporative cooling is significant e Part 2. Experi-
mental testing. J. Food Eng. 37, 127e141.
Chuntranuluck, S., Wells, C.M., Cleland, A.C., 1998c. Prediction of chilling times of
foods in situations where evaporative cooling is significant e Part 3. Applica-
tions. J. Food Eng. 37, 143e157.
Clayton, M., Amos, N.D., Banks, N.H., Morton, R.H., 1995. Estimation of apple fruit
surface area. N. Z. J. Crop Hortic. Sci. 23 (3), 345e349.
Defraeye, T., Cronje, P., Berry, T., Opara, U., East, A., Hertog, M., et al., 2015a. Towards
integrated performance evaluation of future packaging for fresh produce in the
cold chain. Trends Food Sci. Technol. 44, 201e225.
Defraeye, T., Verboven, P., Opara, U.L., Nicolai, B., Cronje, P., 2015b. Feasibility of
ambient loading of citrus fruit into refrigerated containers for cooling during
marine transport. Biosyst. Eng. 134, 20e30.
Dehghannya, J., Ngadi, M., Vigneault, C., 2010. Mathematical modelling procedures
for airflow, heat and mass transfer during forced cooling of produce: a review.
Food Eng. Rev. 2, 227e243.
 G.P. Redding et al. / Journal of Food Engineering 190 (2016) 80e93
Delele, M.A., Tijskens, E., Atalay, Y.T., Ho, Q.T., Ramon, H., Nicolai, B.M., et al., 2008.
Combined discrete element and CFD modelling of airflow through random
stacking of horticultural products in vented boxes. J. Food Eng. 89, 33e41.
Delele, M.A., Verboven, P., Ho, Q.T., Nicolaï, B.M., 2010. Advances in mathematical
modelling of postharvest refrigeration processes. Stewart Postharvest Rev. 6 (2),
1e8.
Denys, S., Ludikhuyze, L.R., Van Loey, A.M., Hendrickx, M.E., 2000. Modeling
conductive heat transfer and process uniformity during batch high-pressure
processing of foods. Biotechnol. Prog. 16 (1), 92e101.
Ebadi-Dehaghani, H., Nazempour, M., 2012. Thermal conductivity of nanoparticle
filled polymers. In: Hashim, A. (Ed.), Smart Nanoparticles Technology. Intech.
Eifert, J.D., Sanglay, G.C., Lee, D., Sumner, S.S., Pierson, M.D., 2006. Prediction of raw
produce surface area from weight measurement. J. Food Eng. 74 (4), 552e556.
Exama, A., Arul, J., Lencki, R.W., Lee, L.Z., Toupin, C., 1993. Suitability of plastic films
for modified atmosphere packaging of fruits and vegetables. J. Food Sci. 58 (6),
1365e1370.
Ferrua, M.J., Singh, R.P., 2008. A noninstrusive flow measurement technique to
validate the simulated laminar fluid flow in a packed container with vented
walls. Int. J. Refrig. 31, 242e255.
Ferrua, M.J., Singh, R.P., 2009a. Modeling the forced-air cooling process of fresh
strawberry packages, Part I: numerical model. Int. J. Refrig. 32, 335e348.
Ferrua, M.J., Singh, R.P., 2009b. Modeling the forced-air cooling process of fresh
strawberry packages, Part II: experimental validation of the flow model. Int. J.
Refrig. 32, 349e358.
Ferrua, M.J., Singh, R.P., 2009c. Modeling the forced-air cooling process of fresh
strawberry packages, Part III: experimental validation of the energy model. Int.
J. Refrig. 32, 359e368.
Ferrua, M.J., Singh, R.P., 2009d. Design guidelines for the forced-air cooling process
of strawberries. Int. J. Refrig. 32, 1932e1943.
Gaffney, J.J., Baird, C.D., Chau, K.V., 1985. Influence of airflow rate, respiration,
evaporative cooling, and other factors affecting weight loss calculations for
fruits and vegetables. ASHRAE Trans. 91 (1), 690e707.
Gross, K.C., Wang, C.Y., Saltveit, M., 2004. The Commercial Storage of Fruits, Vege-
tables, and Florist and Nursery Stocks. United States Department of Agriculture.
Handbook 66, Washington DC.
Hardenburg, R.E., Watada, A.E., Wang, C.Y., 1986. The Commercial Storage of Fruits,
Vegetables, and Florist and Nursery Stocks. United States Department of Agri-
culture. Handbook 66, Washington DC.
Heyes, J.A., Tanner, D.J., East, A.R., 2010. Kiwifruit respiration rates, storage tem-
peratures and harvest maturity. Proc. Int. Symp. Postharvest Pacifica 2009
(880), 167e173.
Hu, Z., Sun, D.W., 2000. CFD simulation of heat and moisture transfer for predicting
cooling rate and weight loss of cooked ham during air-blast chilling process.
J. Food Eng. 46, 189e198.
Hu, Z., Sun, D.W., 2001. Effect of fluctuation in inlet airflow temperature on CFD
simulation of air blast chilling process. J. Food Eng. 48, 311e316.
Incropera, F.P., Dewitt, D.P., 2006. Transient conduction. In: Fundamentals of Heat
and Mass Transfer, fourth ed. John Wiley & Sons, New York.
Kumlutas, D., Tavman, I.H., Coban, M.T., 2003. Thermal conductivity of particle filled
poluethylene composite materials. Compos. Sci. Technol. 63 (1), 113e117.
Kuriger, R.J., Alam, M.K., 2002. Thermal conductivity of thermoplastic composites
with submicrometer carbon fibers. Experimental heat transfer. A J. Therm.
Energy Generation, Transp. Storage Convers. 15 (1), 19e30.
Le Page, J., Chevarin, C., Kondjoyan, A., Daudin, J., Mirade, P., 2009. Development of
an approximate empirical-CFD model estimating coupled heat and water
transfers of stacked food products placed in airflow. J. Food Eng. 92, 208e216.
Lewis, W.K., 1922. The evaporation of a liquid into a gas. Trans. Am. Soc. Mech. Eng.
14 (7), 445e446.
Luyt, A.S., Molefi, J.A., Krump, H., 2006. Thermal, mechanical and electrical prop-
erties of copper powder filled low-density and linear low-density polyethylene
composites. Polym. Degrad. Stab. 91 (7), 1629e1636.
MacCrae, E.A., Bowen, J.H., Stec, M.G.H., 1989. Maturation of kiwifruit (Actinidia
deliciosa cv Hayward) from two orchards: differences in composition of the
tissue zones. J. Sci. Food Agric. 47 (4), 401e416.
MacGregor, W., 2005. Effects of air velocity, air temperature, and berry diameter on
wild blueberry drying. Dry. Technol. 23, 387e396.
Maguire, K.M., Banks, N.H., Opara, L.U., 2001. Factors affecting weight loss of apples.
Hortic. Rev. 25, 197e234.
Mamunya, Y.P., Davydenko, V.V., Pissis, P., Lebedev, E.V., 2002. Electrical and ther-
mal conductivity of polymers filled with metal powders. Eur. Polym. J. 38,
1887e1897.
Marai, S., Ferrar, E., Civelli, R., 2012. Postharvest cold chain optimisation of little
fruits. In: Proceedings of the 2012 COMSOL Conference, Milan.
Maurer, A.R., Eaton, G.W., 1971. Calculation of potato tuber surface area. Am. Potato
93
J. 48, 82e87.
Maxwell, J.C., 1873. A Treatise on Electricity and Magnetism. Clarendon Press,
Oxford.
Milosevic, T., Milosevic, N., 2012. Fruit quality attributes of sour cherry cultivars.
ISRN Agron. http://dx.doi.org/10.5402/2012/593981.
Minh, T.V., Perry, J.S., Bennett, A.H., 1969. Forced-air precooling of white potatoes in
bulk. ASHRAE Trans. 75 (2), 143e152.
Mirade, P.S., Kondjoyan, A., Daudin, J.D., 2002. Three-dimensional CFD calculations
for designing large food chillers. Comput. Electron. Agric. 34, 67e88.
Olatunji, J., Love, R.J., Shim, Y.M., Ferrua, M.J., East, A.R., 2015. Numerical determi-
nation of kiwifruit shape, volume and surface area using ellipsoid approxima-
tion and outside diameter function. In: Proceedings of the International
Congress of Engineering and Food, Quebec City, Canada, June 14th- 18th.
Oliveira, J.M., Lessio, B.C., Morgante, C.M., Santos, M.M., Augosto, P.E.D., 2012. Spe-
cific heats of tropical fruits: caja, cashew apple, cocoa, kiwi, pitanga, soursop
fruit and yellow melon. Int. Food Res. J. 19 (3), 811e814.
O’Sullivan, J., Ferrua, M., Love, R., Verboven, P., Nicolai, B., East, A., 2014. Airflow
measurement techniques for the improvement of forced-air cooling, refriger-
ation and drying operations. J. Food Eng. 143, 90e101.
Paniagua, A.C., East, A.R., Hindmarsh, J.P., Heyes, J.A., 2013. Moisture loss is the
major cause of firmness change during postharvest storage of blueberry.
Postharvest Biol. Technol. 79, 13e19.
Pathare, P.B., Opara, U.L., Vigneault, C., Delele, M.A., Al-Said, F.A., 2012. Design of
packaging vents for cooling fresh horticultural produce. Food Bioprocess
Technol. 5, 2031e2045.
Pham, Q.T., Trujillo, F.J., McPhail, N., 2009. Finite element model for beef chilling
using CFD-generated heat transfer coefficients. Int. J. Refrig. 32, 102e113.
Progelhof, R.C., Thrine, J.L., Ruetsch, R.R., 1976. Methods for predicting the thermal
conductivity of composite systems: a review. Polym. Eng. Sci. 16 (9), 615e625.
Razavi, S.M.A., BahramParvar, M., 2007. Some physical and mechanical properties of
kiwifruit. Int. J. Food Eng. 3 (6) http://dx.doi.org/10.2202/1556-3758.1276.
Rogge, S., Defraeye, T., Herremans, E., Verboven, P., Nicolai, B.M., 2015. A 3D contour
based geometrical model generator for complex-shaped horticultural products.
J. Food Eng. 157, 24e32.
Sweat, V.E., 1974. Experimental values of thermal conductivity of selected fruits and
vegetables. J. Food Sci. 39, 1080e1083.
Tabatabaeefar, A., Rajabipour, A., 2005. Modeling the mass of apples by geometrical
attributes. Sci. Hortic. 105, 373e382.
Talbot, M.T., 1988. An approach to better design of pressure-cooled produce con-
tainers. Proc. Fla. State Hortic. Soc. 101, 165e175.
Tanner, D.J., Cleland, A.C., Opara, L.U., Robertson, T.R., 2002a. A generalised math-
ematical modelling methodology for design of horticultural food packages
exposed to refrigerated conditions: part 1, formulation. Int. J. Refrig. 25, 33e42.
Tanner, D.J., Cleland, A.C., Opara, L.U., Robertson, T.R., 2002b. A generalised math-
ematical modelling methodology for design of horticultural food packages
exposed to refrigerated conditions: part 2, heat transfer modelling and testing.
Int. J. Refrig. 25, 43e53.
Tanner, D.J., Cleland, A.C., Opara, L.U., Robertson, T.R., 2002c. A generalised math-
ematical modelling methodology for design of horticultural food packages
exposed to refrigerated conditions: part 3, mass transfer modelling and testing.
Int. J. Refrig. 25, 54e65.
Tanner, D.J., Amos, N.D., 2003. Temperature variability during shipment of fresh
produce. Acta Hortic. 599, 193e203.
Trujillo, F.J., Pham, Q.T., 2006. A computational fluid dynamic model of the heat and
moisture transfer during beef chilling. Int. J. Refrig. 29, 998e1009.
Valente, M., Chamberel, A., Cordonnier, J., Pumborios, M., 1996. Finite element
modelling of heat transfer in avocados. Int. Agrophys. 10, 123e129.
Verboven, P., Flick, D., Nicolai, B.M., Alvarez, G., 2006. Modelling transport phe-
nomena in refrigerated food bulks, packages and stacks: basics and advances.
Int. J. Refrig. 29, 985e997.
Vigneault, C., Castro, L.R., 2005. Produce-simulator property evaluation for indirect
airflow distribution measurement through horticultural crop package. J. Food,
Agric. Environ. 3 (2), 67e72.
Vissotto, F.Z., Kieckbusch, T.G., Neves Filho, L.C., 2000. Pre-cooling of model prod-
ucts with forced air. Braz. J. Food Technol. 3, 1e10.
Waghmare, R.B., Mahajan, P.V., Annapure, U.S., 2013. Modelling the effect of time
and temperature on respiration rate of selected fresh-cut produce. Postharvest
Biol. Technol. 80, 25e30.
Warren, R., 1992. Ceramic-Matrix Composites. Blackie, Glasgow and London.
Weidenfeller, B., Hofer, M., Schilling, F.R., 2004. Thermal conductivity, thermal
diffusivity and specific heat capacity of particle filled polypropylene. Compos.
Part A Appl. Sci. Manuf. 35 (1), 423e429.
Zhao, C.J., Han, J.W., Yang, X.T., Qian, J.P., Fan, B.L., 2016. A review of computational
fluid dynamics for forced-air cooling process. Appl. Energy 168, 314e331.