Journal of Food Engineering 242 (2019) 21–30

Contents lists available at ScienceDirect

Journal of Food Engineering

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

Evaluation of the cold chain management options to preserve the shelf life of T
frozen shrimps: A case study in the home delivery services in Taiwan
Nodali Ndraha, Wen-Chieh Sung, Hsin-I Hsiao∗
Department of Food Science, National Taiwan Ocean University, Keelung, Taiwan

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

Keywords: Increasing demand for chilled and frozen temperature food products led to the growth of home delivery cold
Temperature fluctuation chain services. This service is becoming popular because of its convenience and affordability. However, home
Cold chain delivery cold chains could face difficulties, especially in temperature control during the delivery of packages,
Shelf life such as too frequent door opening and closing, which can raise considerable food quality and food safety issues.
In this study, we evaluated the temperature situation during transportation by four major home delivery service
providers in Taiwan. Automatic temperature data loggers were used to record the temperature profile during
transportation. The obtained data were then simulated with various scenarios to estimate the remaining shelf life
and food loss probability by using the Monte Carlo algorithm with 10,000 iterations. The results showed that
abusive temperature in real conditions could reduce the remaining shelf life of frozen shrimp by more than 70%.
Lowering the maximum temperature to 15 °C or even to 7 °C had almost no impact on preserving the shelf life or
reducing the quality of frozen shrimp. Among the developed scenarios, better preservation of remaining shelf life
could be obtained by narrowing the fluctuation to −18 ± 3 °C. This scenario is recommended as an integral
part of temperature management control in the home delivery cold chain, particularly in frozen food. These
findings may help the food managers in the food cold chain industry to preserve the quality of low-temperature
food product. Additionally, this study may be applied to manage the temperature in home delivery cold chain in
other country with a similar condition.

1. Introduction Moreover, the shifting of consumer preferences and the rise of e-

commerce as a channel for purchasing a wide variety of food products
According to a report from the FAO, roughly one-third of global pose new challenges, especially in delivering highly perishable food.
food production, valued at more than $750 billion annually, was lost or The cold chain industry, therefore, is required to provide an effective
wasted annually (Gustavsson et al., 2011). These losses mainly result logistic system, which should not only deliver food products safely, but
from the lack of proper facilities (i.e. inadequate storage and cooling also be able to preserve the quality of food, and reach the destination on
facilities), improper post-harvest handling procedures, poor processing, time in a cost-effective way (Kuo and Chen, 2010). In most cases, re-
poor attitudes of consumers, lack of information on market demand, tailer and/or convenience store contracts with the logistic provider to
and insufficient training for personnel working in the cold chain facilitate the delivery of their products to the consumer. The logistics
(Gustavsson et al., 2011). provider facilitates the shipment of small packages to general con-
Furthermore, increasing demand for chilled and frozen temperature sumers and corporations in the domestic market which could be tracked
food products led to the growth of cold chain services globally. In 2017, throughout the journey, under so-called ‘home delivery services’ (Chou
the market value of global cold chain was estimated to be USD 189.92 and Lu, 2009). This service is becoming popular because of its con-
billion, and it was projected to reach USD 293.27 billion by 2023 venience and affordability. However, home delivery cold chains could
(Marketandmarkets, 2018). Miller (2016) mentioned that the devel- face difficulties, especially in temperature control during the delivery of
opment pace of this industry was driven by the expansion of the food packages, such as too frequent door opening and closing, which can
cold chain industry, support from the relevant governmental agencies, raise considerable food quality and safety issues. Consequently, the
and the innovation of the infrastructure facilities utilized in the cold food products may be exposed to undesired temperature conditions. As
chain. pointed out by many researchers, exposing food products to

∗
Corresponding author.
E-mail address: hi.hsiao@ntou.edu.tw (H.-I. Hsiao).

https://doi.org/10.1016/j.jfoodeng.2018.08.010
Received 5 March 2018; Received in revised form 6 August 2018; Accepted 9 August 2018
Available online 11 August 2018
0260-8774/ © 2018 Elsevier Ltd. All rights reserved.
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

temperature abuse may considerably increase the amount of quality

(Margeirsson et al., 2012)

(Tingman et al., 2010)

loss by the time the product reaches its destination (Ashby, 2006;
Bogataj et al., 2005) and could lead to food waste (Tromp et al., 2016).

(Mai et al., 2011)

(Mai et al., 2012)

Huge efforts have been made to improve the temperature manage-
ment control in the food cold chain during the last two decades; how-

Reference
ever, it still remains difficult to realize. Mercier et al. (2017) point out
that the temperature abuse primarily occurs due to lack of precooling
practice, improper temperature management during transportation,

- Faster quality deterioration and microbial growth occurred to samples positioned at container

- No diversity on temperature of samples in the refrigerated container during 8 h transportation

- Shelf-life of product along with temperature fluctuated within 2.0 °C was two months shorter
temperature setting in the storage during display at retail centers and in

- Different temperature were observed during transport, depending on box sample positions
- Abuse temperature shortened the shelf life of chilled and super-chilled samples to 8.7–8.8

- Longer distribution at sea transport and abuse temperature at air transport shortened the
household refrigerators, and improper commercial handling practices.

days and 9.8–10.8 days, which was expected to be longer than 10 days and 15 days,
To improve the temperature management in the food cold chain, the
authors suggest further improvement related to precooling uniformity,
food inventory management based on time-temperature measure, and

- Shelf life reduction of 1.5–3 days was observed due to temperature abuse.
food cold chains in developing countries. Ndraha et al. (2018) stated
that many studies related to food cold chain focus more on chilled food
compared to frozen food products. The authors suggest that the appli-
cation of the developed system in real-time temperature monitoring
along the food cold chain should be further investigated. Moreover,
continuous temperature monitoring along the cold chain was suggested
to improve the temperature management (Abad et al., 2009). As re-
ported by Göransson et al. (2017), food quality could be better pre-
served and the shelf-life of a particular food product could be predicted
accurately if the time-temperature is fully monitored. Enabling real-

remaining shelf life of haddock fillets

time temperature monitoring would help the managers in the food cold
chain to make correct decisions, so that necessary action can be taken in
a timely fashion.
Several legal requirements and guidelines have been developed to
govern the temperature management in the food cold chain. The Codex
suggests using adequate equipment for storing refrigerated or frozen

than within 0.5 °C

food (CAC, 2003). In the European Union, the temperature of fishery

respectively.
products during transportation should not be warmer than −18 °C with
possible short upward fluctuations of not more than 3 °C (EC, 2004).
Key finding

corners

Similarly, frozen food in the United States should have an internal

product temperature of −18 °C or colder during transportation (US
FDA, 2017). In Australia, frozen food should not be warmer than
−18 °C during transportation and storage between −22 °C and −30 °C
- Total viable psychotropic

is recommended (AFGC, 2017). In Taiwan, food business should ensure

the frozen food products during transportation and storage have the
- Sensory evaluation

- Sensory evaluation

- Sensory evaluation

- Sensory evaluation

temperature of −18 °C or colder (TFDA, 2017).

Evaluated parameters

In particular, the cold chain industry is growing rapidly in Taiwan.

This industry was reported to have a market value of more than NTD
200 million (∼USD 6.8 million) in revenue and distributed NTD 1.7
counts

- TVBN

- TVBN

billion (∼USD 57.9 million) worth of a various types of food products

from 2011 to 2012 (MOEA, 2013). Given these facts, the evaluation of
temperature performance in food cold chains in Taiwan is therefore
necessary. In this study, we present a case study of frozen shrimp
transported by the home delivery cold chain. Shrimp was chosen be-
Example of shelf life assessment for seafood at transport chain.

Chilled: 0.5 °C Super-

Frozen: −18 ± 2 °C

cause of increasing demand for this type of product in Taiwan. We

evaluated the temperature management options to preserve the re-
chilled: −1 °C
Chilled: 0 °C

Chilled: 0 °C

maining shelf-life of frozen shrimp allowed during transportation from

a quality perspective by employing the kinetic model has been devel-
oped by Tsironi et al. (2009). Previous studies have demonstrated that
TRa

abuse temperature during transport may shorten the shelf life and
promote microbial growth (see Table 1). The kinetic model has been
Type of product

Haddock fillets

widely used to evaluate the quality changes and predict the remaining
Tilapia fish
Cod fillets

Cod loins

shelf life of various seafoods, including blue shark slices and arrow
squid (Giannoglou et al., 2014), chilled tuna fish (Tsironi et al., 2008),
Temperature requirement.

and packed gilthead seabream fillets (Tsironi et al., 2011). Besides, to

the extent of our knowledge, no available research has been done to
Sea and road transport
Supply chain involved

evaluate the temperature performance regarding the home delivery

Air and sea transport

Transport and frozen

cold chains in Taiwan.

Fig. 1 shows the general operational overflow of home delivery
storage

services in Taiwan. The home delivery service in Taiwan offers delivery

Transport

of various packages, including low-temperature products (i.e. chilled or

Table 1

frozen). It has unique service characteristics compared to the other cold

a

chain services, such as cold storage warehouse or distribution center. It

22
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

Consumer/
customer 1 Drop-off 1
Regional Regional
Transportation 1 Transportation Drop-off 1
center 1 center 2
Consignee 1
Consumer/
Drop-off 2 Transportation
customer 2
Transportation 2 Drop-off 2
Consumer/ Consignee 2
Drop-off n
customer 3
Drop-off n
Consumer/
Transportation n Consignee n
customer n

Sensitive to temperature abuse Sensitive to temperature abuse

Fig. 1. Operational flow of home delivery services in Taiwan.

facilitates door-to-door pickup and delivery service, and provides million), respectively. Although all these companies have their own
pickup call service, or the customer can take their package to the ser- storage stations, their methods differ in collecting the packages. While
vice office/drop-off area (Chou and Lu, 2009). As shown in this figure, Company 4 collects the packages from the consumer/customer location,
abusive temperature can occur primarily during package collection Companies 1, 2 and 3 collect the packages from the drop-off point (i.e.
from consumer/customer and package delivery to the consignee. The convenience stores 7–11, OK-Mart, or Family-Mart). Additionally, each
more locations visited to collect the packages from consumer/customer of these companies has its own strategy for delivering the packages
places or drop-off points, the more probable that temperature abuse involving, for example, the number of transit centers, the number of
could occur; the same condition also could occur during delivery of the drop-offs, and the number of vehicles; these may differ among them.
packages to the consignee. Additionally, managing temperature in the Two boxes were delivered by each home delivery service in the
home delivery cold chain is critical due to the population density and same shipment. Each box containing 600 g of shrimp sample. The box
temperature conditions in Taiwan. Taiwan has a geographical area of size was 10 cm × 20 cm x 30 cm, made from carton material. Selected
36,193 square kilometers. With a population of 23,476,640, its average boxes contained automatic temperature data loggers (HOBO 64 K
population density is 649 people per square kilometer (1680 per square Pendant®, Onset Computer Corporation, USA) to monitor the tempera-
mile). This makes it the 17th most densely populated the country in the ture of the shrimp during transportation. The recorders were placed
world. Consequently, door opening of the truck's container during inside the box and covered by the shrimp sample to record the core
loading/unloading occurs frequently. The annual average temperature temperature of the package. Prior to distribution, frozen shrimp
in Taiwan is 22 °C. While there is no severe cold in winter, the weather packages were in storage with temperature of −30 °C for 12 h.
in summer is brutally hot and highly humid. The hottest months are Thereafter, the frozen boxes were transported from Keelung to
from June to August with the highest temperature up to around 38 °C Kaohsiung using four different home delivery services.
(Central Weather Bureau of Taiwan, 2017).

2.3. Observed and predicted quality changes

2. Materials and method
2.3.1. Microbiological, chemical, and sensory analysis
2.1. Raw materials The microbiological, chemical, and sensory analyses were per-
formed to observe the quality changes under the given temperature
Live white shrimp (Litopenaeus vannamei) with an average body interval based on the obtained temperature from four home delivery
weight of 12–16 g and length of 12–15 cm were purchased on December services. The obtained temperature from a real condition was mimicked
8, 2016, from a local market in Keelung District, Taiwan. Live shrimps to allow for efficient kinetic analysis of quality changes. The mimicked
were transported to the laboratory by using an open container con- temperature sequence for each company was as follows: Company
taining seawater. The mean temperature of the seawater in Keelung was 1 = −11.6 °C for 10 h, −9.3 °C for 5 h, −7.7 °C for 7 h, −11.9 °C for 2,
19.7 °C (min = 12.5 °C, max = 28.8 °C) (Central Weather Bureau of -13.2 °C for 12 h, and −15.8 °C for 10 h, corresponding to the Teff of
Taiwan, 2017). The transportation time from local market to the la- −11.4 °C; Company 2 = −13.5 °C for 5 h, −7.1 °C for 18 h, −17.1 °C
boratory was approximately 10 min. To kill the shrimps, they were for 6 h, and −27.2 °C for 48 h, corresponding to the Teff of −13.4 °C;
washed with tap water and kept in the refrigerator at −30 °C for Company 3 = −13.4 °C for 5 h, −2.2 °C for 12 h, −4.6 °C for 12 h,
30 min. −5.6 °C for 36 h, and 8.9 °C for 5 h, corresponding to the Teff of −0.7 °C;
and Company 4 = −10.1 °C for 5, -1.6 °C for 12 h, 0.2 °C for 12 h,
2.2. Case companies and time-temperature recording −4.1 °C for 12 h, −8.1 °C for 12 h, −12.6 °C for 12 h, and −5.6 °C for
7 h, corresponding to the Teff of −4.0 °C. The mimicked temperature
We selected four major home delivery service companies in Taiwan was simulated by using a temperature programmable control cabinet
and collected the time-temperature data by putting temperature re- (Cherng Huei Co., Ltd., Taiwan). The analysis of quality changes was
corders in the delivering boxes. The four companies operate in all city performed by taking the shrimp samples in time intervals of 24 h and at
areas of Taiwan, and all of them offer shipment of low-temperature the end of simulation.
products, including food products. The companies were named A microbiological analysis was performed according to the method
Company 1, Company 2, Company 3, and Company 4 herein for con- previously described by Erkan et al. (2010). Aerobic plate counts (APC)
fidential purpose. Company 1 had revenues of more than NTD 15 bil- were determined using plate count agar (PCA, Difco™, SN 247940) after
lion with the worth of distributed packages amounting to NTD 20 bil- incubation for 24–48 h at 37 °C. The results are expressed as a logarithm
lion (∼USD 673,8 million). In the same year, Companies 2, 3, and 4 of colony forming units (log CFU) per gram of sample. The micro-
had revenues of NTD 147 million (∼USD 5.0 million), NTD 710.6 biological analyses were carried out in triplicate from three separated
million (∼USD 24.0 million), and NTD 4.7 billion (∼USD 158.6 shrimp in the same package. Total volatile basic nitrogen (TVBN) was

23
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

t
measured in duplicate by the Conway titration method following the ∑ log ⎛ t
predicted ⎞
⎝ observed ⎠
recommendation of Chinese National Standard (CNS, 1997). TVBN Bf = 10 n (3)
contents were expressed as mg TVBN/100 g shrimp meat, calculated
t
predicted ⎞
using Eq. (1). ∑ log ⎛ t
⎝ observed ⎠
Af = 10 n (4)
⎧ (V1 − V2) ×C×A ⎫
X= 5
×100 where Bf is the bias factor, Af is the accuracy factor, tobserved is the time
⎨ m× 100 ⎬
⎩ ⎭ (1) to each observed quality index value experimentally observed, tpredicted is
the time predicted to reach the same quality value as that observed, and
where X is the TVBN of the samples (mg/100 g), V1 is the amount of n is the number of observations. To evaluate the validation, Betts and
hydrochloric acid by the titrated boric acid absorbing liquid (mL), V2 is Walker (2004) categorized the Bf value of 0.9–1.1 as a good model, 0.7
the consumption amount of hydrochloric acid by the titrated blank to 0.9 as an acceptable model, 1.1 to 1.2 as model to use with caution,
absorbing liquid (mL), c is the concentration of the hydrochloric acid and less than 0.7 or higher than 1.2 as an unacceptable model.
(mol/L), and A is the mass of the nitrogen amount with 1 mL hydro-
chloric acid standard titration solution (1 mol/L) (mg). In this equation, 2.4. Remaining shelf life estimation model
A = 14, and m is the mass of the sample (mg) being measured.
In this study, affective test was used to measure the sensory attri- The remaining shelf life of frozen shrimp was calculated by em-
butes (overall acceptability, appearance, odor, taste, and texture) of ploying the kinetic values (Ea and kref ) obtained from Tsironi et al.
cooked frozen shrimp product, following the method described in (2009). Temperature dependence on shelf life could then be evaluated
Tsironi et al. (2016). To perform the sensory evaluation, six to nine by the changes of its quality indices (TVBN and sensory attribute)
panelists were recruited on the basis of their interest, availability, and (Giannoglou et al., 2014; Tsironi et al., 2016) as follows:
consumption of shrimp. Prior to the data collection, participants re-
ln CTVB − N1 − ln CTVB − N0
ceived regular training to assess the shrimp sensory attributes and were tSLTVBN =
given an explanation of the project. Frozen shrimp samples were
thawed under cold running water. They were placed into a container,
kref . TVB − N exp ⎡
⎣
( )(
−Ea
R
1
Teff
−
1
Tref ) ⎤⎦ (5)
held in a refrigerator (4 °C) for less than 1 h, and cooked in boiling s0 − s1
water for 3 min. The rating was assigned separately for each parameter tSLsensory =
on a 1–9 descriptive hedonic scale (9 being the highest quality score kref . sensory exp ⎡
⎣
( )(
−Ea
R
1
Teff
−
1
Tref ) ⎤⎦ (6)
and 1 the lowest). A score of five was considered as the average score
for minimum acceptability. where tSLTVBN is the shelf life of frozen shrimp based on TVBN, tSLsensory is
Furthermore, the quality changes of frozen shrimp under dynamic the shelf life of frozen shrimp based on sensory, CTVBN1 and s1 are the
temperature were validated by comparing the value obtained from limits of TVBN (CTVBN1 = 25 mg/100 g) and an overall impression (s1
observation and prediction (Tsironi et al., 2009). In this study, the ef- = 5), respectively (Kyrana and Lougovois, 2002; Limbo et al., 2009;
fective temperature (Teff ) was introduced to allow the prediction of the Tsironi et al., 2009). The ln CTVBN0 and s0 are the initial value.
value of quality parameters under dynamic temperature T(t) at dif-
ferent time t. As described by Taoukis et al. (2016), the Teff was defined 2.5. Simulation and what-if scenarios
as the “constant temperature that results in the same quality value as the
variable temperature distribution over the same time period” (p. 288). This Table 2 presents what-if scenarios discussed in this study. Various
approach was based on the Arrhenius model (Taoukis et al., 1999) and temperature scenarios were developed in this study to evaluate the
integrates into a single value the effect of the variable temperature impact of temperature variability during transportation by the home
profile, as given in Eq. (2), delivery cold chain. The first three scenarios were designed based on
the selection of maximum temperatures allowed during delivery, and
t the last three were related to temperature variation.
Ea⋅Tref
Teff = ∫ k (t )
⋅dt Scenario I was developed to simulate the actual temperature during
0 R⋅Tref ⋅ln − Ea
kref (2) delivery. Scenario II was simulated to reflect the recommendation of
Taiwan FDA that proposes the temperature in loading/unloading area
where Teff is the effective temperature (K), Tref is the reference tem- to not be higher than 15 °C (TFDA, 2017). The maximum temperature
perature for frozen food (Tref = 255.15 K), Ea is the activation energy of of 7 °C was selected in Scenario III as this temperature level was con-
the respective quality index (kJ/mol), k (t ) is the deterioration rate sidered as the barrier for the growth of pathogenic Vibrio para-
constant at different time, kref is the rate constant of change of each haemolyticus in seafood (Ma et al., 2016; Miles et al., 1997). Further-
quality index at Tref = 255.15 K, and R is the universal gas constant more, the actual temperature data obtained from the four companies
(8.314 J/mol·K). were fitted by a feature of @risk software (version 5.7) (Jankauskas and
Moreover, the observed and predicted values were validated based McLafferty, 1996) based on the given condition in Scenarios I, II, and III
on the bias and accuracy factors (Ross, 1996; Steele, 2004; Taoukis and (see Table 2). The distribution of temperature was best described by the
Giannakourou, 2004) as described in Eqs. (5) and (6). Logistic function.

Table 2
Scenarios tested in this study.
Scenario Content Fitted distribution of temperature

I Actual temperature profile recorded from four home delivery services Logistic (−8.5, 4.3) + 273.15
II The maximum temperature fluctuation in actual observation was reduced to 15 °C during distribution Logistic (−8.4, 4.3) + 273.15
III The maximum temperature fluctuation in actual observation was reduced to 7 °C during distribution Logistic (−8.5, 4.2) + 273.15
IV The range of the actual temperature was narrowed to −18 ± 12 °C (−8 °C to −28 °C) Logistic (−18, 12) + 273.15
V The range of the actual temperature was narrowed to −18 ± 9 °C (−10 °C to −26 °C) Logistic (−18, 9) + 273.15
VI The range of the actual temperature was narrowed to −18 ± 6 °C (−12 °C to −24 °C) Logistic (−18, 6) + 273.15
VII The range of the actual temperature was narrowed to −18 ± 3 °C (−14 °C to −22 °C) Logistic (−18, 3) + 273.15

24
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

Table 3
Input variables for remaining shelf life simulation in frozen shrimps.
Variable Description Input values Units Source/reference

Ea Activation energy TVBN 119 kJ/mol (Tsironi et al., 2009)

Taste 124
Overall acceptability 111
R Universal gas constant 8.314 J/mol·K (Tsironi et al., 2009)
Tref Reference temperature of frozen 255.15 Kelvin (Tsironi et al., 2009)
seafood
kref Deterioration rate at reference TVBN 0.002 day-1 (Tsironi et al., 2009)
temperature Taste 0.0056
Overall acceptability 0.0062
Ao Initial value of quality parameter in TVBN 5.51 mg/100 g This study
harvest Taste 8.4
Overall acceptability 8.29
Teff Temperature in transportation during Table 1 Kelvin This study
distribution
ti Time spent in transportation Uniform (46, 77)/24 day-1 This study
ki Quality deterioration rate (Tsironi et al., 2016)
kref exp ⎡ ( )
−Ea ⎛ 1 1 ⎞ ⎤
⎜
⎢ R ⎝ Teff Tref ⎠ ⎥
⎣ ⎦
⎟

Ai Maximum value of quality indices after TVBN A0 exp(ki × ti) mg/100 g (Tsironi et al., 2009)
distribution Taste and Overall A0 exp(−ki × ti)
acceptability
CTVBN or s0 Limit of quality indices TVBN 25 mg/100 g (Kyrana and Lougovois, 2002; Limbo
et al., 2009)
Taste and overall 5 point (Tsironi et al., 2009)
acceptability

SLday Remaining shelf-life at dynamic TVBN ln CTVB − N1 − ln CTVB − N0 day (Taoukis et al., 2016)
tSLTVBN =
temperature ⎡ −E ⎛ 1 1 ⎞⎤
kref . TVB − N exp ⎢ ⎛ a ⎞ ⎜ − ⎥
⎢ ⎝ R ⎠ ⎝ Teff Tref ⎟⎠ ⎥
⎣ ⎦
Taste and Overall acceptability s0 − s1
tSLsensory =
⎡ −E ⎛ 1 1 ⎞⎤
kref . sensory exp ⎢ ⎛ a ⎞ ⎜ − ⎥
⎢ ⎝ R ⎠ ⎝ Teff Tref ⎟⎠ ⎥
⎣ ⎦

Furthermore, we also evaluated the impact of temperature range home delivery in this study. This study found that nearly 66% of the
allowed during transportation. Available guidelines stipulate that the time during transportation in the home delivery services had a tem-
frozen temperature should be maintained at −18 °C or colder with perature higher than −18 °C. Temperature variability falls within
short upward fluctuations (AFGC, 2017; EC, 2004; US FDA, 2017). For −28.7 °C to 17.2 °C with an average of −11 °C. The temperature ranges
study purposes, therefore, the temperature ranges of −18 ± 12 °C, vary between the companies. Companies 1, 2, 3 and 4 have temperature
−18 ± 9 °C, −18 ± 6 °C, and −18 ± 3 °C were simulated to eval- ranges of −17.3 °C to −0.4 °C, −28.7 °C–7.8 °C, −19.6 °C–17.2 °C, and
uate their impact on the remaining shelf life and quality loss in frozen −16.3 °C to 4.1 °C, respectively, indicating that the temperature fluc-
shrimp in Scenarios IV, V, VI, and VII, respectively. tuations between the companies vary. The abusive temperature during
Next, the simulation of temperature levels developed in various transportation of another type of seafood product was previously re-
scenarios was then input into the developed model as shown in Table 3. ported. An example of this is the study performed by Tingman et al.
The simulation of developed scenarios was performed by using the (2010) who reported that the ambient temperature fluctuated between
Microsoft Office Excel (Office, 2013) with @risk add-in software (ver- −18.6 °C and 16.8 °C during transport of frozen tilapia fish in China,
sion 5.7) and simulated with Monte Carlo Simulation with 20,000 but the sample temperature only increased slightly to −17 °C. Another
iterations (Vose, 2008). To calculate the remaining shelf life, the kinetic example is the study carried out by Margeirsson et al. (2012) who
parameters obtained from Tsironi et al. (2009) and time-temperature showed that the temperature for cod fillets during transport exceed up
variability were input into the developed model. The quality loss was to 10.5 °C for 6.4 h. The author found that temperature abuse reduced
calculated based on Eq. (7). the storage life of cod fillets up to 1.5–3 days. Mai et al. (2011) reported
that the shelf life of chilled and super-chilled cod loins samples was
tSLref − tSLT (t )
Qloss (%) = × 100% expected to have shelf life longer than 10 days and 15 days, respec-
tSLref (7) tively, however, it was reduced to 8.7–8.8 days and 9.8–10.8 days due
to temperature abuse during transport. Another example, Mai et al.
where Qloss is the quality loss, tSLref is the remaining shelf life at the (2012) also found that the remaining shelf life of haddock fillets was
reference temperature, and tSLT (t ) is the remaining shelf life at the actual shorter because of longer distribution time at sea transport and abuse
temperature. temperature at air transport. Besides, the author observed that haddock
boxes have different temperature, depending on the box positions in the
3. Results and discussion container. The result in this study therefore indicates a serious tem-
perature fluctuation in the home delivery cold chain in Taiwan.
3.1. Time-temperature profile of frozen food home delivery Moreover, this study also showed that each company delivered the
packages with a different time interval. Under normal circumstances,
Better evaluation and interpretation of temperature data can be packages could be delivered within 48 h, given the fact that the distance
achieved when they are continuously recorded by using temperature from Keelung city to Kaohsiung city was only 378.7 km. In this study,
monitoring and/or a tracking tool such as electronic data loggers so that however, delivery time was found to be 46 h, 77 h, 70 h, and 72 h in
timely corrective actions can be taken. Table 4 and Fig. 2 show the Companies 1, 2, 3, and 4, respectively. This result may be due to dif-
summary of the time-temperature profile of frozen shrimp during food ferences in company strategies in delivering the packages. For example,

25
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

Table 4
Time-temperature profiles of frozen shrimp obtained from home delivery companies (n = 2).
Home delivery service Temperature (°C) Spending time (hour) Allocated time at different temperature ranges (%)

Max Min Average 0–7 °C −10 to 0 °C −18 to −10 °C ≤ −18 °C

Company 1 −0.4 −17.3 −12.0 46 42 16 42 0

Company 2 7.8 −28.7 −20.8 77 8 47 8 37
Company 3 17.2 −19.6 −3.9 52 19 69 9 3
Company 4 4.1 −16.3 −5.6 72 27 53 20 0
Summary 17.2 −28.7 −11.0 46–77 24 46.3 19.8 10.0

time spent during packages collection could vary between the compa- in most of the frozen shrimp samples, but lower than the ones reported
nies depending on the number of packages from customer/consumer or by Tsironi et al. (2009) and Jeyasekaran et al. (2006) who found the
drop-off point. In addition, delivery can be longer if the recipient is not initial total plate count was 5.7 log CFU/g and 6 log CFU/g, respec-
in place, so the package will be returned to the nearest warehouse, as tively. After simulation, it was observed that virtually no microbial
was the case with deliveries in this study. It can be assumed that the growth had occurred in this study. The number of APC after a simula-
more places to stop in order to collect and/or deliver the packages at tion in Companies 1, 2, 3 and 4 were found to be 4.2 log CFU/g, 4.5 log
different drop-off and/or consignee locations, the more temperature CFU/g, 4.4 log CFU/g, and 4.2 log CFU/g, respectively. Slow growth of
abuse could occur because of frequent door opening during loading/ APC in this study matches those observed in Tsironi et al. (2009), where
unloading packages from the truck container. Undelivered packages there was not growth of microbial at −8 °C or below. The author
will be delivered the next day, which must sometimes first be accom- confirmed that slow growth of microbial was observed at near sub-
panied by a phone call from the recipient. These findings enhance our freezing temperatures. Wu (2014) also reported that the formation of
understanding that temperature management control in the home de- icy crystal in shrimp at low temperature significantly reduced the total
livery cold chain remains a challenge due to its unique service char- viable count in 5 days of storage.
acteristics. Furthermore, frozen shrimps showed an initial TVBN value of 5.5
mg/100 g in this study (Fig. 3B). A slight change of TVBN value was
3.2. Deterioration of shrimp quality under dynamic temperature observed during simulation, accounted to 6.3 mg/100 g, 5.7 mg/100 g,
8.3 mg/100 g, and 9.1 mg/100 g observed in Companies 1, 2, 3 and 4,
According to the International Commission on Microbiological respectively.
Specification for Foods, the limit of acceptability of total bacterial load Table 5 shows the evaluation of sensory attributes of cooked frozen
on frozen raw crustaceans was 7 log CFU/g (ICMSF, 1986). In this shrimp. Initially, all the frozen shrimp samples were considered as
study, frozen shrimp samples showed an initial aerobic plate count having high sensory quality. The evaluation of sensory attributes in
(APC) of 4.3 log CFU/g (Fig. 3A), indicating that the shrimp was within cooked frozen shrimp showed that it was associated with the tem-
an acceptable level. This value is similar to the study performed by perature. The panelist judged that the sensory attributes in cooked
Hatha et al. (1998) who reported the initial APC was about 4 log CFU/g shrimp samples of Companies 3 and 4 had deteriorated more quickly

Fig. 2. Distribution of monitored temperature profile of frozen shrimp during transportation obtained from four home delivery service providers (n = 2).

26
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

5.2 10
Company 1 B
A Company 2
5.0
Company 3 Company 1
9
4.8 Company 4 Company 2
Company 3
4.6 8 Company 4

TVBN (mg/100 g)
log CFU/g

4.4
7
4.2

4.0 6

3.8
5
3.6

3.4 4
0 20 40 60 80 0 20 40 60 80

Time (hour) Time (hour)

Fig. 3. Observed aerobic plate count (A) and TVBN value (B) of shrimp product simulated with the time-temperature profile obtained from four home delivery
services (error bars indicate standard error of measurements of two different samples).

Table 5 and 4 were rejected by the panelists due to melanosis, color fading, and
Sensory evaluation of cooked frozen shrimp. unpleasant odor. The sample rejection in this study may correlate with
Time (h) Comp. 1 Comp. 2 Comp. 3 Comp. 4 the quality deterioration due to undergoing internal cell crystallization,
taking into consideration that the Teff of these two companies (3 and 4)
Overall acceptability was close to the threshold of freezing temperature (c.f. Connell, 1990;
0 8.3 ± 0.7 8.3 ± 0.7 8.3 ± 0.7 8.3 ± 0.7
Haard, 1998). The melanosis formation and color fading in shrimp
24 6.4 ± 1.0 7.3 ± 1.0 6.9 ± 0.3 7.0 ± 0.7
46 6.2 ± 1.3 – – –
during storage were previously reported by Nirmal and Benjakul (2011)
48 – 7.2 ± 1.0 5.3 ± 0.8 6.9 ± 1.2 and Chantarasuwan et al. (2011). While the melanosis formation oc-
70 – – 4.9 ± 1.0 – curred due to the biochemical process of polyphenol oxidase, the color
72 – 6.5 ± 0.9 – 4.2 ± 0.8 fading was considered as a result of leaching free carotenoids and
77 – 6.5 ± 1.0 – –
carotenoprotein (Nirmal and Benjakul, 2009). These results are in ac-
Appearance cord with those obtained by Zeng et al. (2005) who reported that the
0 8.5 ± 0.8 8.5 ± 0.8 8.5 ± 0.8 8.5 ± 0.8 higher temperatures accelerate sensory damage and reduce sensory
24 6.6 ± 1.2 7.4 ± 0.9 6.8 ± 0.6 7.1 ± 0.8 scores. Therefore, maintaining the seafood product at the recommended
46 6.4 ± 1.0 – – –
temperature is required to prevent the deterioration of sensory attri-
48 – 7.2 ± 1.1 5.4 ± 0.8 6.8 ± 0.9
70 – – 5.3 ± 0.9 –
butes.
72 – 6.5 ± 0.9 – 4.5 ± 1.0 Data validation performed in this study is shown in Table 6. Data
77 6.5 ± 1.1 validation was only performed based on the value of TVBN and sensory
attributes. The calculation of Af and Bf value based on APC was not
Odor
possible in this study due to slow microbial growth. Based on the TVBN
0 8.2 ± 0.7 8.2 ± 0.7 8.2 ± 0.7 8.2 ± 0.7
24 7.0 ± 0.8 7.3 ± 0.8 7.0 ± 0.4 7.2 ± 0.7 value in this study, the Bf value in Companies 1, 2, 3 and 4 was 1.1, 1.0,
46 6.6 ± 0.9 – – – 1.1, and 1.3, respectively, with an Af of 1.0–1.3 or an error up to 30%.
48 – 7.3 ± 0.9 5.5 ± 1.0 6.8 ± 1.1 Furthermore, the Bf value based on taste parameter in Companies 1, 2, 3
70 – – 5.5 ± 0.7 –
and 4 was 0.9, 0.8, 1.0, and 0.9, respectively, and it was found to be 0.9,
72 – 6.3 ± 1.0 – 4.2 ± 0.8
77 – 5.7 ± 1.4 – –
0.9, 0.9, and 09, respectively, based on the overall acceptability para-
meter. These results indicate that the use of the model developed by
Texture Tsironi et al. (2009) in this study should be used with caution (Betts and
0 8.2 ± 0.9 8.2 ± 0.9 8.2 ± 0.9 8.2 ± 0.9 Walker, 2004), except for TVBN value in Company 4. A possible ex-
24 7.1 ± 1.3 7.2 ± 0.9 7.0 ± 0.6 7.1 ± 0.8
planation for these results may be that the kinetic parameters obtained
46 5.7 ± 1.6
48 – 7.2 ± 1.2 5.3 ± 0.9 7.1 ± 1.0 from Tsironi et al. (2009) are not completely suitable in this study
70 – 5.3 ± 1.0 because of high-temperature fluctuation. The kinetic parameters ob-
72 – 6.5 ± 1.1 4.4 ± 1.1 tained from Tsironi et al. (2009) was developed on the temperature of
77 – 6.5 ± 0.8
−5 °C, −8 °C, −12 °C, and −15 °C, whereas this study has a tem-
Taste
perature range from −28.8 °C to 17.2 °C.
0 8.3 ± 0.7 8.3 ± 0.7 8.3 ± 0.7 8.3 ± 0.7
24 6.9 ± 1.0 7.1 ± 0.8 7.1 ± 0.5 7.2 ± 0.8
46 6.2 ± 1.4 – – – 3.3. The estimation of the remaining shelf life of frozen shrimp
48 – 7.4 ± 1.0 5.4 ± 1.0 6.5 ± 0.8
70 – – 5.3 ± 1.2 – Based on data validation, the estimation of remaining shelf life
72 – 6.5 ± 1.1 – 4.4 ± 1.0 based on APC was not reliable in this study due to the slow microbial
77 – 5.3 ± 1.6 – –
growth as mentioned earlier. In terms of the APC level, it can be as-
- not measured. sumed that the frozen shrimp would be acceptable after delivery. Based
on the TVBN value, the tSL after delivery transported by Company 1
than the samples of Companies 1 and 2. The slow deterioration of (Teff = −11.4 °C), Company 2 (Teff = −13.4 °C), Company 3
sensory quality was observed in Company 2, perhaps because it has (Teff = −0.4 °C), Company 4 (Teff = −4.0 °C) were estimated to be
lower Teff compared to the others. Conversely, the sensory attributes in 181.9 days, 276.8 days, 19.3 days, and 37.9 days, respectively.
Company 3 were found to have the fastest deterioration during the si- As suggested by Tsironi et al. (2009), the remaining shelf life of
mulation of distribution. Moreover, the sample shrimps in Companies 3 shrimp also could be predicted by the sensory attributes of taste and
overall acceptability, since these two sensory attributes are considered

27
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

Table 6
Validation of observed and predicted value of quality changes and estimated remaining shelf life of frozen shrimp.
Parameters Company 1 (Teff = −11.4 °C) Company 2 (Teff = −13.4 °C) Company 3 (Teff = −0.7 °C) Company 4 (Teff = −4.0 °C)

Af Bf tSL Af Bf tSL Af Bf tSL Af Bf tSL

TVBN 1.0 1.1 181.9 1.0 1.0 276.8 1.1 1.1 19.3 1.3 1.3 37.9
Taste 1.1 0.9 119.0 1.2 0.8 182.7 1.0 1.0 0.2 1.0 0.9 7.1
Overall acceptability 1.1 0.9 128.0 1.2 0.9 186.6 1.1 0.9 4.3 1.1 0.9 13.6

Mathematical model was obtained from Tsironi et al. (2009).

tSL = predicted remaining shelf life of frozen shrimp.
TVBN (limit = 25mg/100 g, kref(Tref=-18°C) = 0.002 day−1, Ea = 119 kJ/mol).
Taste (limit = 5 point, kref(Tref=-18°C) = 0.0056 day−1, Ea = 124 kJ/mol).
Overall acceptability (limit = 5 point, kref(Tref=-18°C) = 0.0062 day−1, Ea = 110 kJ/mol).

as good quality indices as they well correlated with evaluated chemical (18 ± 12 °C) was found to be 442.4 days, 345.5 days, and 313.2 days,
indices. While the estimation of remaining shelf life in Companies 1, 2, respectively, based on TVBN, taste, and overall acceptability para-
3, and 4 based on taste parameter in this study was predicted to be meters. Moreover, the estimated tSL in Scenario V was found to be 459.2
119.0 days, 182.7 days, 0.2 day, and 7.1 days, respectively, it was days, 358.4 days, and 326.4 days, respectively, and was found to be
found to be 128.0 days, 186.6 days, 4.3 day, and 13.6 days, respec- 489.8 days, 381.2 days, and 348.9 days, respectively, in Scenario VI.
tively, s evaluated by the overall acceptability parameter. Negative Longer tSL was observed at a narrow range of temperature simulated in
values indicate that samples were spoiled or rejected by the judges Scenario VII, and it was found to be 551.8, 429.4, and 393.1 days, re-
during sensory evaluation. spectively.
Nevertheless, it was observed that Scenarios I, II, and III would
3.4. Simulation outcomes and what-if scenarios result in over 70% food quality loss (Qloss ) in this study. The range of
Qloss in Scenarios IV, V, and VI was estimated to be 41.1%–43.1%,
Table 7 presents the overall outcome of simulation from different 38.5% to 41.0, and 34.3%–37.2%, respectively. Among other scenarios,
quality indicators for the seven scenarios. It has been previously re- better preservation of quality of frozen shrimp could be obtained in
ported that the shelf life of frozen shrimp could reach up to 677 days at Scenario VI, which could lower the Qloss to 25.9%–29.3%. These results
a reference temperature (−18 °C) (Tsironi et al., 2009); however, the demonstrate that maintaining the temperature of frozen food at −18 °C
temperature abuse may deteriorate the quality of frozen shrimp quickly with lower temperature fluctuation would better preserve the re-
resulting in a shorter shelf life. As reported in this study, the mean of maining shelf life and prevent the quality loss of frozen shrimp during
remaining shelf life (tSL) in Scenario I was in the range of 191.8 days, delivery. These results were also supported by evidence that tempera-
134.2 days, and 142.9 days from TVBN, taste, and overall acceptability ture fluctuation affected the quality of other type of seafood in-
perspectives, respectively. The mean of quality loss was in the range of vestigated by many researchers. Abusive temperature had a larger ef-
73.1%–77.9%. The value higher than 100% indicates the samples fect on peroxide and fatty acid values (Gormley et al., 2002), increased
would be rejected or spoiled. Lowering the maximum temperature to gaping and peritoneum deterioration (Romotowska et al., 2017), and
15 °C or even to 7 °C in Scenarios II and III had almost no impact on promoted undesired recrystallization of ice inside the muscle of fish,
preserving the shelf life of frozen shrimp. This result can be explained especially when the temperature fluctuated above −18 °C (Jessen et al.,
by the facts that the shrimp temperature during transport were mostly 2014). Therefore, we recommend the temperature of −18 ± 3 °C or
lower than 7 °C. The ambient temperature limit of 15 °C for low-tem- colder as an allowable temperature range for frozen shrimp during
perature food in loading/unloading area proposed by the TFDA (2017) transportation. This recommendation should be applied as an integral
may not be effective to preserve the quality of frozen shrimp as de- part of temperature management control in the home delivery cold
monstrated in this study. chain in Taiwan, taking into consideration that this intervention pro-
Next, the evaluation of temperature ranges in Scenarios IV, V, VI, vides better preservation of shrimp quality compared to the other sce-
and VII seems to have an impact on the tSL of frozen shrimp. It is ob- narios. This recommendation can be achieved by adequate training for
vious that a larger range of temperature fluctuations is rapidly short- cold chain personnel, the use of adequate equipment and application of
ening the tSL . The estimated tSL of frozen shrimp in Scenario IV real-time temperature monitoring along the home delivery cold chain.

Table 7
Estimated remaining shelf life and percentage loss of frozen shrimps in various scenarios.
Scenario TVBN Taste Overall acceptability

Remaining shelf life (day) Loss percentage (%) Remaining shelf life (day) Loss percentage (%) Remaining shelf life (day) Loss percentage (%)

a b
Mean Lower value Mean Upper value Mean Lower value Mean Upper value Mean Lower value Mean Upper value

I 191.8 5.4 74.6 99.3 134.2 −4.8 77.9 100.8 142.9 6.6 73.1 98.7
II 191.7 5.4 74.7 99.3 134.1 −4.7 77.9 100.8 142.8 6.7 73.1 98.7
III 190.0 5.6 74.9 99.3 132.6 −4.7 78.2 100.8 141.7 6.9 73.3 98.7
IV 442.4 −1.6 41.5 100.2 345.5 −4.3 43.1 100.7 313.2 0.1 41.0 100.0
V 459.2 1.1 39.3 99.9 358.4 −4.0 41.0 100.7 326.4 2.8 38.5 99.5
VI 489.8 16.3 35.2 97.8 381.2 −1.5 37.2 100.2 348.9 15.9 34.3 97.0
VII 551.8 105.0 27.0 86.1 429.4 58.8 29.3 90.3 393.1 85.2 25.9 83.9

Predicted remaining shelf life of frozen shrimp at constant temperature of −18 °C was 756.16, 607.14, 530.65 days estimated by TVBN, taste and overall accept-
ability, respectively.
a
Lower value at 5 percent data.
b
Upper value at 95 percent data.

28
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

The equipment used should be specifically designed and tested for the the truck storage, temperature and starting quality) is necessary.
transport of frozen food products. Furthermore, it is important to em- However, this study provides baseline information about the tempera-
phasize that the temperature of frozen food should be measured based ture condition allowed for transportation in the home delivery cold
on the food product temperature. This is critical because previous stu- chain in Taiwan.
dies have shown that ambient temperature in vehicle storage and/or Thirdly, this study is limited by the lack of information on the
freezer/refrigerator temperatures might be different with food product number of stops and openings of truck container/trailer during de-
temperatures (Baldera Zubeldia et al., 2016; Lundén et al., 2014; livery, the pre-cooling temperature in the truck container prior to dis-
Morelli et al., 2012). Interestingly, the required temperature of −18 °C tribution, and the position of the package in the truck's container.
or colder with short upward fluctuations for frozen food has been Further experimental and modeling research on these areas would be
previously suggested, for example, in the European Union, the United interesting.
States, and Australia (AFGC, 2017; EC, 2004; US FDA, 2017). In these
countries, the frozen food temperature is required to be measured based 4. Conclusion
on the food product temperature.
The study case used in this research aims to evaluate and interpret
3.5. Potential application/integration of this study the impact of temperature fluctuation on remaining shelf-life based on
the changes of quality parameters, and recommends the allowed tem-
Several decision-support frameworks for the management of food perature ranges in the home delivery cold chain in Taiwan. This study
distribution operations have been developed with consideration to shows that large temperature abuses occur during the distribution of
time-temperature control. For example, mixed-integer linear pro- frozen shrimp in Taiwan, which may accelerate the food quality dete-
graming model integrated with food quality was used by Rong et al. rioration. The impact of the temperature abuse was evaluated through
(2011) in decision-making on production and distribution of bell pep- various scenarios in this study, and it was concluded that the tem-
pers. In their case study, this approach was not only able to model the perature of −18 ± 3 °C could better preserve the quality and prolong
degradation of bell peppers quality, but it was also allow the cost es- the shelf life of frozen shrimp during distribution in the home delivery
timation used for managing the storage and distribution temperature. cold chain. Therefore, we suggest the TFDA to re-evaluate the tem-
Similar approach was also employed by Accorsi et al. (2017) who in- perature recommendation for low-temperature food product developed
tegrated the weather conditions in the mixed-integer linear programing in Taiwan. We recommend the temperature of −18 ± 3 °C or colder as
model on production, storage and distribution operations. In their case an allowable temperature range for frozen shrimp during transport and
study of a cold chain for cherries, the impact of weather condition on it should be measured based on the food product temperature. Proposed
the energy costs was evaluated. Another example is the study carried temperature of 15 °C in loading/unloading area in Taiwan was in-
out by Tromp et al. (2016) in which the author suggested to consider effective to preserve the shelf life of frozen shrimp simulated in this
the technical, logistical, and marketing interventions in order to control study. Additionally, we also recommend the TFDA to determine the
food waste. Technical intervention can be done by controlling the maximum duration and necessary action to be taken if low-temperature
temperature to delay the product's natural decay, adjusting the food food products are exposed to temperatures higher than allowed.
waste criteria, and maybe by increasing the use-by date of the food Furthermore, the additional panelists and better training for sensory
product. Logistical intervention can be enabled by decreasing the sto- panelist in sensory evaluation are recommended for future study.
rage times in the chain and managerial intervention can be done by Finally, the finding in this study may help the food business manager in
influencing the consumer demand. However, food safety aspect was not food cold chain industry to better manage the shelf-life of their food
considered in the approach of Tromp et al. (2016) that might cause product.
problem to customers/consumers. A common consideration for these
approaches was the temperature control. The approach used in this Acknowledgement
study therefore can be potentially applied or integrated with the ex-
isting approaches that have been previously proposed. This is important This study was made possible with financial support (106 TFDA-FS-
to support the management decision in food supply chain, taking into 507) from the Taiwan Food and Drug Administration. The authors wish
consideration that food business operators are not only dealing with to express their gratitude to this organization.
time-temperature aspect in practical term, but also need to consider the
broader aspects. References

3.6. Limitations of the study Abad, E., Palacio, F., Nuin, M., Zárate, A.G., de Juarros, A., Gómez, J.M., Marco, S., 2009.
RFID smart tag for traceability and cold chain monitoring of foods: Demonstration in
an intercontinental fresh fish logistic chain. J. Food Eng. 93, 394–399.
Firstly, the case study used in this research was evaluated by using Accorsi, R., Gallo, A., Manzini, R., 2017. A climate driven decision-support model for the
the model developed by Tsironi et al. (2009). Although this model was distribution of perishable products. J. Clean. Prod. 165, 917–929.
considered applicable in this study, we noticed that the model may not AFGC [Australian Food and Grocery Council], 2017. Australian Cold Chain Guidelines
2017. 5.13.18. https://www.afgc.org.au/wp-content/uploads/Australian-Cold-
describe the changes of quality parameters appropriately, especially Chain-Guidelines-2017.pdf.
when dealing with subjective assessments (sensory evaluation), which Ashby, B.H., 2006. Protecting Perishable Foods During Transport by Truck. US
may correspond to the number of panelists. Therefore, additional pa- Department of Agriculture, Washington, DC.
Baldera Zubeldia, B., Nieto Jiménez, M., Valenzuela Claros, M.T., Mariscal Andrés, J.L.,
nelists and better training for them may improve the consistency of Martin-Olmedo, P., 2016. Effectiveness of the cold chain control procedure in the
sensory evaluation results. Additionally, development of the specific retail sector in Southern Spain. Food Contr. 59, 614–618.
model for sensory evaluation of frozen shrimp transported in the home Betts, G.D., Walker, S.J., 2004. Verification and validation of food spoilage models. In:
Understanding and Measuring Shelf Life of Food. CRC Press, Boca Raton, FL, pp.
delivery cold chain in Taiwan is suggested.
184–217.
Secondly, taking into consideration that different companies have Bogataj, M., Bogataj, L., Vodopivec, R., 2005. Stability of perishable goods in cold logistic
different delivery strategies and temperature fluctuation ranges, the chains. Int. J. Prod. Econ. 93–94, 345–356.
results of this study case may not reflect all the situations of real tem- CAC [Codex Alimentarius Commission], 2003. General Principles of Food Hygiene. Joint
FAO/WHO Food Standards Programme, Rome, Italy.
perature during transportation in the home delivery cold chain in Central Weather Bureau of Taiwan, 2017. Climate Statistics. 6.7.17. http://www.cwb.
Taiwan as a whole. In addition, this study was performed without gov.tw/V7e/climate/monthlyData/mD.htm.
consideration of season variability. Further evaluation of the whole Chou, P., Lu, C., 2009. Assessing service quality, switching costs and customer loyalty in
home-delivery services in Taiwan. Transport Rev. 29, 741–758.
supply chain with various scenarios (i.e. number of samples, position in

29
N. Ndraha et al. Journal of Food Engineering 242 (2019) 21–30

Chantarasuwan, C., Benjakul, S., Visessanguan, W., 2011. Effects of sodium carbonate and Miles, D.W., Ross, T., Olley, J., McMeekin, T.A., 1997. Development and evaluation of a
sodium bicarbonate on yield and characteristics of Pacific white shrimp (Litopenaeus predictive model for the effect of temperature and water activity on the growth rate
vannamei). Food Sci. Technol. Int. 17, 403–414. of Vibrio parahaemolyticus. Int. J. Food Microbiol. 38, 133–142.
CNS (Chinese National Standard), 1997. Frozen Fish Inspection Act: the Use of Conway's Miller, J., 2016. 2016 Top Markets Report Cold Supply Chain. A Market Assessment Tool
Micro-diffusion Method. for U.S. Exporters. U.S. Department of Commerce, Washington, DC.
Connell, J., 1990. Quality deterioration and defects in products. In: Control of Fish MOEA [Ministry of Economic Affairs], 2013. Taiwan's cold-chain logistics ready to soar.
Quality. Fishing News Books, West Byfleet. 4.30.18. https://english.ey.gov.tw/News_Content2.aspx?n=8262ED7A25916ABF&
EC [European Commission], 2004. Regulation (EC) No 853/2004 on laying down specific sms=DD07AA2ECD4290A6&s=26B623A71F606A94.
hygiene rules for on the hygiene of foodstuffs. Off. J. Eur. Communities 5.1.18. Morelli, E., Noel, V., Rosset, P., Poumeyrol, G., 2012. Performance and conditions of use
http://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32004R0853& of refrigerated display cabinets among producer/vendors of foodstuffs. Food Contr.
qid=1525179229679&from=EN. 26, 363–368.
Erkan, N., Üretener, G., Alpas, H., 2010. Effect of high pressure (HP) on the quality and Ndraha, N., Hsiao, H.-I., Vlajic, J., Yang, M.-F., Lin, H.-T.V., 2018. Time-temperature
shelf life of red mullet (Mullus surmelutus). Innovat. Food Sci. Emerg. Technol. 11, abuse in the food cold chain: Review of issues, challenges, and recommendations.
259–264. Food Contr. 89, 12–21.
Giannoglou, M., Touli, A., Platakou, E., Tsironi, T., Taoukis, P.S., 2014. Predictive Nirmal, N.P., Benjakul, S., 2009. Melanosis and quality changes of Pacific white shrimp
modeling and selection of TTI smart labels for monitoring the quality and shelf-life of (Litopenaeus vannamei) treated with catechin during iced storage. J. Agric. Food
frozen seafood. Innovat. Food Sci. Emerg. Technol. 26, 294–301. Chem. 57, 3578–3586.
Göransson, M., Nilsson, F., Jevinger, Å., 2018. Temperature performance and food shelf- Nirmal, N.P., Benjakul, S., 2011. Inhibition of melanosis formation in Pacific white
life accuracy in cold food supply chains – insights from multiple field studies. Food shrimp by the extract of lead (Leucaena leucocephala) seed. Food Chem. 128,
Contr. 86, 332–341. 427–432.
Gormley, R., Walshe, T., Hussey, K., Butler, F., 2002. The effect of fluctuating vs. constant Romotowska, P.E., Gudjónsdóttir, M., Karlsdóttir, M.G., Kristinsson, H.G., Arason, S.,
frozen storage temperature regimes on some quality parameters of selected food 2017. Stability of frozen Atlantic mackerel (Scomber scombrus) as affected by tem-
products. LWT - Food Sci. Technol 35, 190–200. perature abuse during transportation. LWT - Food Sci. Technol 83, 275–282.
Gustavsson, J., Cederberg, C., Sonesson, U., Van Otterdijk, R., Meybeck, A., 2011. Global Rong, A., Akkerman, R., Grunow, M., 2011. An optimization approach for managing fresh
Food Losses and Food Waste. Food Agric. Organ, United Nations. food quality throughout the supply chain. Int. J. Prod. Econ. 131, 421–429.
Haard, N.F., 1998. Foods as cellular systems: impact on quality and preservation. CRC Ross, T., 1996. Indices for performance evaluation of predictive models in food micro-
Press, Boca Raton, FL. biology. J. Appl. Bacteriol. 81, 501–508.
Hatha, A.A.M., Paul, N., Rao, B., 1998. Bacteriological quality of individually quick- Steele, R., 2004. Understanding and Measuring the Shelf-life of Food. Woodhead
frozen (IQF) raw and cooked ready-to-eat shrimp produced from farm raised black Publishing, Abington.
tiger shrimp (Penaeus monodon). Food Microbiol. 15, 177–183. Taoukis, P.S., Giannakourou, M.C., 2004. Temperature and food stability: analysis and
ICMSF [International Commission on Microbiological Specifications for Foods], 1986. In: control. In: Understanding and Measuring the Shelf-life of Food. Elsevier, pp. 42–68.
Microorganisms in Foods 2: Sampling for Microbiological Analysis, Principles and Taoukis, P.S., Gogou, E., Tsironi, T., Giannoglou, M., Dermesonlouoglou, E., Katsaros, G.,
Specific Applications, second ed. Blackwell Scientific Publications, Oxford, England. 2016. Food cold chain management and optimization. In: Nedović, V., Raspor, P.,
Jankauskas, L., McLafferty, S., 1996. BestFit, distribution fitting software by Palisade Lević, J., Tumbas Šaponjac, V., Barbosa-Cánovas, G.V. (Eds.), Emerging and
Corporation. In: Simulation Conference, 1996. Proceedings. Winter, pp. 551–555. Traditional Technologies for Safe, Healthy and Quality Food, Food Engineering
Jessen, F., Nielsen, J., Larsen, E., 2014. Chilling and freezing of fish. In: Boziaris, I.S. Series. Springer International Publishing, Cham, pp. 285–309.
(Ed.), Seafood Processing: Technology, Quality and Safety. John Wiley & Sons, Ltd, Taoukis, P.S., Koutsoumanis, K., Nychas, G.J.E., 1999. Use of time–temperature in-
Volos, Greece, pp. 33–59. tegrators and predictive modelling for shelf life control of chilled fish under dynamic
Jeyasekaran, G., Ganesan, P., Anandaraj, R., Jeya Shakila, R., Sukumar, D., 2006. storage conditions. Int. J. Food Microbiol. 53, 21–31.
Quantitative and qualitative studies on the bacteriological quality of Indian white TFDA [Food and Drug Administration of Taiwan], 2017. Regulations on Good Hygiene
shrimp (Penaeus indicus) stored in dry ice. Food Microbiol. 23, 526–533. Practice for Food. cid=16&id=2870. 5.8.17. http://www.fda.gov.tw/EN/
Kuo, J.C., Chen, M.C., 2010. Developing an advanced multi-temperature joint distribution lawContent.aspx?.
system for the food cold chain. Food Contr. 21, 559–566. Tingman, W., Jian, Z., Xiaoshuan, Z., 2010. Fish product quality evaluation based on
Kyrana, V.R., Lougovois, V.P., 2002. Sensory, chemical and microbiological assessment of temperature monitoring in cold chain. Afr. J. Biotechnol. 9, 6146–6151.
farm-raised European sea bass (Dicentrarchus labrax) stored in melting ice. Int. J. Tromp, S.O., Haijema, R., Rijgersberg, H., van der Vorst, J.G.A.J., 2016. A systematic
Food Sci. Technol. 37, 319–328. approach to preventing chilled-food waste at the retail outlet. Int. J. Prod. Econ. 182,
Limbo, S., Sinelli, N., Torri, L., Riva, M., 2009. Freshness decay and shelf life predictive 508–518.
modelling of European sea bass (Dicentrarchus labrax) applying chemical methods Tsironi, T., Dermesonlouoglou, E., Giannakourou, M., Taoukis, P., 2009. Shelf life mod-
and electronic nose. LWT - Food Sci. Technol 42, 977–984. elling of frozen shrimp at variable temperature conditions. LWT - Food Sci. Technol
Lundén, J., Vanhanen, V., Myllymäki, T., Laamanen, E., Kotilainen, K., Hemminki, K., 42, 664–671.
2014. Temperature control efficacy of retail refrigeration equipment. Food Contr. 45, Tsironi, T., Giannoglou, M., Platakou, E., Taoukis, P., 2016. Evaluation of time tem-
109–114. perature integrators for shelf-life monitoring of frozen seafood under real cold chain
Ma, F., Liu, H., Wang, J., Zhang, Z., Sun, X., Pan, Y., Zhao, Y., 2016. Behavior of Vibrio conditions. Food Packag. Shelf Life 10, 46–53.
parahaemolyticus cocktail including pathogenic and nonpathogenic strains on cooked Tsironi, T., Gogou, E., Velliou, E., Taoukis, P.S., 2008. Application and validation of the
shrimp. Food Contr. 68, 124–132. TTI based chill chain management system SMAS (safety monitoring and assurance
Mai, N., Gudjónsdóttir, M., Lauzon, H.L., Sveinsdóttir, K., Martinsdóttir, E., Audorff, H., system) on shelf life optimization of vacuum packed chilled tuna. Int. J. Food
Reichstein, W., Haarer, D., Bogason, S.G., Arason, S., 2011. Continuous quality and Microbiol. 128, 108–115.
shelf life monitoring of retail-packed fresh cod loins in comparison with conventional Tsironi, T., Stamatiou, A., Giannoglou, M., Velliou, E., Taoukis, P.S., 2011. Predictive
methods. Food Contr. 22, 1000–1007. modelling and selection of time temperature integrators for monitoring the shelf life
Mai, N.T.T., Margeirsson, B., Margeirsson, S., Bogason, S.G., Sigurgísladóttir, S., Arason, of modified atmosphere packed gilthead seabream fillets. LWT - Food Sci. Technol 44,
S., 2012. Temperature mapping of fresh fish supply chains - Air and sea transport. J. 1156–1163.
Food Process. Eng. 35, 622–656. US FDA [US Food and Drug Administration], 2017. 21 CFR 600.15 temperatures during
Margeirsson, B., Lauzon, H.L., Pálsson, H., Popov, V., Gospavic, R., Jónsson, M.þ., shipment. Electron. Code Fed. Regul 5.1.18. https://www.accessdata.fda.gov/
Sigurgísladóttir, S., Arason, S., 2012. Temperature fluctuations and quality dete- scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm.
rioration of chilled cod (Gadus morhua) fillets packaged in different boxes stored on Vose, D., 2008. In: Risk Analysis: a Quantitative Guide, third ed. John Wiley & Sons, West
pallets under dynamic temperature conditions. Int. J. Refrig. 35, 187–201. Sussex, UK.
Marketandmarkets, 2018. Cold chain market by type, temperature range, technology, Wu, S., 2014. Effect of chitosan-based edible coating on preservation of white shrimp
application, and region - Global forecast to 2023. 4.29.18. https://www. during partially frozen storage. Int. J. Biol. Macromol. 65, 325–328.
marketsandmarkets.com/Market-Reports/cold-chains-frozen-food-market-811.html. Zeng, Q.Z., Thorarinsdottir, K.A., Olafsdottir, G., 2005. Quality changes of shrimp
Mercier, S., Villeneuve, S., Mondor, M., Uysal, I., 2017. Time-temperature management (Pandalus borealis) stored under different cooling conditions. J. Food Sci. 70,
along the food cold chain . A Rev. Recent. Dev. 16, 647–667. s459–s466.

30