Professional Documents
Culture Documents
Engineering in Agriculture, Environment and Food: Chao-Chin Chung, Ho-Hsien Chen, Ching-Hua Ting
Engineering in Agriculture, Environment and Food: Chao-Chin Chung, Ho-Hsien Chen, Ching-Hua Ting
Engineering in Agriculture, Environment and Food: Chao-Chin Chung, Ho-Hsien Chen, Ching-Hua Ting
Research paper
a r t i c l e i n f o a b s t r a c t
Article history: Autoclave thermal process is an important sterilisation procedure in the food canning industry. The
Received 30 November 2014 performance of sterilisation is determined by the heating temperature and duration and the pressure in
Received in revised form the autoclave, as well as by product's properties. In this study, a fuzzy logic controller is developed to
18 September 2015
maintain a small steady-state temperature error (121 ± 0.5 C) for an sterilising duration (F0) by adapting
Accepted 30 November 2015
Available online xxx
to process deviations in the autoclave. The precise temperature control associated with on-line F0
updation assures food stuff that can be heated at the specified temperature for an expected duration and
could lead therefore to successful sterilisation. The controls are implemented on a PLC (programmable
Keywords:
Fuzzy control
logic controller) for affordability, reliability, and robustness and the operations are supervised using a
F0 value web-based SCADA (supervisory control and data acquisition) system for remotely supervisory moni-
On-line updation toring and control.
PLC © 2015 Asian Agricultural and Biological Engineering Association. Published by Elsevier B.V. All rights
Sterilisation reserved.
Thermal process
http://dx.doi.org/10.1016/j.eaef.2015.11.003
1881-8366/© 2015 Asian Agricultural and Biological Engineering Association. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
2 C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8
heating, and incurs a hazardous pressure in the cooling stage which In food sterilisation by thermal treatment, the DT value of the C.
leads to can deformation (Farid and Abdul Ghani, 2004). This pro- botulinum in a foodstuff determines the suitability of putting the
cess deviation is detrimental to the performance of temperature foodstuff on the market. The DT value measures the duration that
control. Hence the air should be exhausted from the autoclave 90% of microorganisms are reduced at a specific temperature T. A
during sterilisation (Kumar et al., 2001; Akterian, 1999). However, microorganism with a larger DT is more bearable to heat. The effect
the incurred system uncertainty will lead to poor temperature of temperature on the rate of microbial destruction is described by
control by conventional control schemes. The literature demon- the above parameters, as (Teixeira and Sun, 2012; Singh and
strates, in temperature control of bio-process, that a fuzzy model Heldmon, 2001; Stoforos, 1995):
based control has better performance and less control effort than
conventional controls (Moelbak, 1999; Honda and Kobayashi, 2000; DT ¼ DTref 10ð121TÞ=Z (1)
Thompson and Dexter, 2005; Chung et al., 2010). Fuzzy control is
more intuitive and mainly relies on operator's expertise and where DTref is the 90% reduction duration at the reference tem-
knowledge (Lee and Kwon, 2007). This provides the engineer a perature 121 C and Z indicates the change in the death rate based
framework for modelling the non-linearity and complexity of a on the temperature T.
food process which normally comes with insufficient or imprecise The F value, known as the thermal death time or the safety in-
system information (Honda and Kobayashi, 2000; Chung et al., dex, determines the sterilising duration required to achieve a
2010; Lee and Kwon, 2007; O'Connor et al., 2002). However, certain inactivation ratio in the number of microorganisms at
fuzzy control is blind to process dynamics and its performance may temperature T (Teixeira and Sun, 2012):
be impaired by process deviations. Hence on-line system identifi-
cation of the controlled process solves the air exhausting problem N
F ¼ DT log 0 (2)
and provides a stable and secure operation in heating process (Lee N
et al., 2002). However, on-line system identification demands
tedious numerical computation which is a big challenge or even an where N0 is the number of microorganisms before sterilisation and
impossible mission for implementation on hardware with limited N is the number of surviving microorganisms after an F duration of
numerical computation capability. sterilisation at temperature T.
Existing sterilisers calculate the target sterilising time (F0 value) The sterilising duration is determined by F ¼ 12DT when the
using the monogram method with a thermal death curve (Farid and heating temperature is exactly maintained at 121 C (Singh and
Abdul Ghani, 2004; Stoforos, 1995). The conventional approach for Heldmon, 2001). In practice, it is impossible to maintain the steri-
calculating the F0 value is time-consuming for PLC (programmable lising temperature exactly and continually at 121 C in autoclave
logic controller) implementation and is unsuitable for on-line operation. Hence the F0 index which explicitly includes the steri-
adaptation to process deviations. The F0 value is sensitive to tem- lising duration is used as an alternative to assure adequate steri-
perature variation, so automatic control is always required during lisation. An F0 value defines the reference thermal death time
the heating cycle (Syafiie et al., 2011). An inadequate F0 value may within which, a certain population of microorganisms is eradicated
result in over or insufficient sterilisation. This problem is solved at 121 C with a death rate characteristic Z ¼ 10 C (Teixeira and
using on-line compensation for process deviations through Sun, 2012; Singh and Heldmon, 2001). The F0 value at a specific
updating the F0 value in real time. temperature T in C is calculated by:
The objectives of this study are (1) to develop a fuzzy logic
Zt
controller to control sterilization temperature and (2) to apply a
lookup table for calculating the sterilization duration F0. The F0 ¼ 10ðT121Þ=Z dt (3)
developed control scheme and F0 calculation strategy are to be 0
implemented using PLC which has limited numerical processing Inclusion of heating and cooling in the F0 index avoids inade-
capability but is popular in industries. The fuzzy logic controller quate sterilisation (Farid and Abdul Ghani, 2004). The F0 formula
performs precise temperature control and makes on-line correc- describes how thermal inactivation of microorganism populations
tions of unexpected process deviations in autoclave operation. Such can be quantified as a function of time and temperature.
a control scheme has the same effect of on-line system identifica-
tion but without the demand of delicate numerical computation. By
this way, processors can minimize, even prevent the occurrence of 2.2. Horizontal autoclave
deviant processes. Control data is organised as a lookup table
suitable for PLC implementation. Fig. 1 shows the horizontal autoclave developed in house for
study into canned food sterilisation. The autoclave has a capacity of
2. Thermal sterilisation 5000 kg and an interior size of 3.5 m long and 1.0 m in diameter. All
field devices are controlled and process responses are collected,
2.1. Fundamentals using a PLC (FX2N-4AD-PT, Mitsubishi, Japan). Process information
is displayed on a human machine interface (HMI) (UG220H-SC4,
The major objective of thermal processing in the food industry is Fuji, Japan). The PLC is supervised by a supervisory control and data
to guarantee food safety by killing microorganisms and inactivating acquisition (SCADA) system which was developed in house based
their enzymes or other metabolites. Most microorganisms can be on the LabVIEW software (Version 6i, National Instruments, USA).
inactivated by heat but some heat-bearable ones can survive in a The SCADA system performs F0 calculation, data logging, and pro-
certain temperature range (Datta and Teixeira, 1988; Kumar et al., cess monitoring. The internal status of the autoclave is monitored
1990). During microbial inactivation by heat, the right combina- with a surveillance camera (AXIS 2100 Video Server, Axis Com-
tion of temperature and time is important to allow two goals in a munications, Taiwan). Images from the camera together with
process: eliminate or reduce pathogenic or spoilage microorgan- operational information from the PLC can be monitored remotely
isms to an acceptable level and, at the same time, cause minimum via the Internet on a monitoring station.
effect on product quality and acceptability. The primary aim of the control task is to maintain the
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8 3
LEGENDS
pressure gauge
PLC HMI
thermometer
Pressurised air
safety valve
Cooling water
Exhaust gas
IPC RTD thermowell
signal recorder
SCADA
Steam distributor
CCD camera
proportional valve
temperature in the autoclave at a desired setpoint (121 C) through 2.3. Process control
regulating steam discharge from a boiler. Increasing steam
discharge increases the temperature and the pressure in the The sterilisation process is operated in accordance with the
autoclave. Cooling water is injected to reduce the internal tem- procedures described in Fig. 2. A complete sterilisation consists of
perature if necessary. The accumulated pressure must be relieved exhausting, heating, cooling, and draining processes. Heat transfer
once the pressure exceeds the allowable strength of the autoclave. within the autoclave has non-linear dynamics which cannot be
This operation introduces process deviation in the point of view of correctly described with deterministic models.
system control. The temperature control shall adapt to the process A new batch comes with cool foodstuffs and meanwhile,
deviation caused by occasional pressure relief. A temperature ambient air intrudes into the autoclave during can loading. The
transducer (SR301, Chancemore, Taiwan) and a pressure transducer residing air hinders heat convection which will lead to cans un-
(CAU-T, Jin Zon, Taiwan) provide the PLC with process response evenly heated and may result in incomplete sterilisation. This
measurements for feedback controls while a thermometer and a problem is alleviated by expelling the air with continual steam
pressure meter for operator's observation. injection before the start of batch processing. The air exhausting
Setup condition
1. Exhaust gas temperature and time
2. Heating temperature, pressure and time
3. Sterilization temperature, pressure and No Sterilization
time (F0) (Temp, Pressure, Time)
4. Cooling temperature, pressure and time
No Cooling
(Temp, Pressure, Time)
No Exhaust gas
(Temp, Time)
Yes: Temp. < Set point
Close: Pressurize valve &
Circulating pump
Yes: > Set point Open: Exhaust water valve &
Close: Exhaust gas valve Exhaust gas valve
Yes: Temp. > Set point Yes: Pressure < Set point
Open: Circulating pump
END
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
4 C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8
procedure warms up the cans to a temperature close to the desired and ZE for zero. Fig. 5 shows the five membership functions used to
sterilising temperature, and then the fuzzy control takes over fuzzify the two variables. Triangular membership functions were
temperature regulation, as shown in Fig. 3. Well controlled small chosen because of the efficiency in numerical computation.
temperature transition prevents the cans from thermal deforma- The FAM rule bank defines fuzzy rules that map the fuzzy input
tion. Steam condensate is collected and sprayed onto the cans. This sets to the fuzzy output set. FAM rules represent logical ”if-then”
increases the efficiency of heat exchange as water has a heat statements where the antecedent is the fuzzy input set, e[k], and
transfer coefficient larger than steam. Timing of sterilisation is the consequence is the valve opening, u[k]. For this controller the
triggered once the temperature reaches 121.5 C. Afterwards, the FAM rule bank can be conceptualised as a vector of input-output
temperature is maintained within 121 ± 0.5 C for a period deter- associations. A total of 5 rules are shown in Table 1. The Mamdani
mined by the F0 value, calculated by Eq. (3), for proper sterilisation. scheme is used to establish a fuzzy matrix equation R:
Once the sterilisation is finished, the system is forwarded to a
cooling procedure. The cooling procedure is operated in sequential R ¼ ðNBe NBu Þ∪ðNSe NSu Þ∪ðZEe ZEu Þ∪ðPSe PSu Þ∪ðPBe PBu Þ
mode. The water valve is completely open to inject cooling water (4)
into the autoclave, the exhaust water valve is regulated to discharge
The correlation-minimum inference method is used to obtain an
unnecessary water, and the exhaust gas valve is regulated to reduce
output fuzzy set, truncated by the degree of the input membership,
the inner pressure.
for each active FAM rule. The inferred result is passed to a centre
The cooling procedure reduces the inner pressure of the auto-
defuzzification algorithm, as:
clave when the sterilized cans are still at a high temperature and
with a high inner pressure. The cans may deform if cooling is not Z
well managed. A cooling phase starts at constant pressure with mR ðeÞede
decreasing temperature and then releases the pressure at constant u¼ Z (5)
temperature. Water spray into the autoclave decreases the tem- mR ðeÞde
perature and injection of clean pressurised air maintains the
pressure. Once the temperature is lowered to the room tempera- where mR(e) is the membership function. Here the truncated fuzzy
ture, the pressure is released gradually to the atmospheric level. A set membership functions are combined to form a single scalar
sterilisation cycle is now considered complete. output, the crisp valve opening u[k], with membership functions
defined in Fig. 5 (b). In implementation, an input-output mapping
3. Fuzzy logic control lookup table for the fuzzy control logic is coded into the PLC for
efficient, numerical computation.
3.1. Control system
T switch
+
-
sign(•)
100%
warmup and gas exhausting
feedback temperature control
+ u y
r e Fuzzy Logic Steam Valve Autoclave
+ Control -
-
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8 5
FAM
Bank
Temperature
Transducer
Table 2
Lookup table of control commands, u in mA, with valve opening scalings.
Scaling e ( C)
Fig. 5. Membership functions of the error variable e[k] and the controller output u[k].
4.3. On-line calculation of the F0 value
Table 1 Eq. (3) that calculates the F0 value is converted to a discrete form
FAM rule bank for the fuzzy sterilisation controller.
for PLC implementation, as:
Fuzzy set
X
n X
n
System error, e NB NS ZE PS PB F0 ¼ DFi ¼ 10ðTTref Þ=Z Dt (8)
Control command, u NB NS ZE PS PB
i¼1 i¼1
NB: negative big; NS: negative small; ZE: zero; PS: positive small; PB: positive big.
with T ¼ temperature of the canned foodstuff, Tref ¼ reference
temperature 121 C, the death rate characteristic Z ¼ 10 C, and Dt is
the time elapse between two calculations.
2 3 Inclusion of the heating time and the cooling time gives a
1:0 0:5 0 0 0 0 0 0 0
6 0:5 modified F0 (Simpson et al., 1993):
6 0:5 0:5 0:5 0 0 0 0 0 77
6 0 0:5 1:0 0:5 0:5 0:5 0 0 0 7
6 7
R¼6 0 7 F0;setting ¼ max F0;heating ; 0:8F0;heating þ 0:2F0;cooling (9)
6 0 0 0 0:5 1:0 0:5 0 0 7 (6)
6 0 0 0 0:5 0:5 0:5 1:0 0:5 0 7
6 7
4 0 0 0 0 0 0:5 0:5 0:5 0:5 5 The SCADA system calculates the F0 value and updates the PLC
0 0 0 0 0 0 0 0:5 1:0 control on line. Algorithm 1 describes the use of F0 in sterilisation
termination control. The temperature T is regulated by the fuzzy
The controller output can be approximated by: logic controller.
u0 ¼ e+R (7)
The lookup table gives a simple, efficient approach in the deri-
vation of controller output. The fuzzy controller has a control action
between 4 and 4 which is converted to 4e20 mA to drive the
steam proportional valve. To limit the maximum steam valve
opening, the derived fuzzy control command is scaled with a factor
before actuating the steam valve. Four scaling ratios, 100%, 50%, 30%
and 20%, were applied. Scaled resultants of fuzzy logic
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
6 C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8
Fig. 6. System responses of sterilising temperature controls with various valve opening scalings.
5. Results and discussion the other two have quite promising results. Hence the former two
controls are considered unacceptable because of offset responses
5.1. Temperature control of sterilisation and bigger oscillating control actions. Valve openings scaled by 30%
or 20% maintain temperature error within ±0.5 C and have less
Fig. 6 shows system responses of each scaling control. The oscillating control actions in comparison with the former two. Such
temperature setpoint is set at 105 C in the beginning of control and an accuracy is relatively good as in commercial heating/cooling
then increased to 121.5 C at 600 s. This procedure allows the applications an accuracy of 2e5% is considered acceptable (Delgado
autoclave to be filled with hot steam for warming up. Afterwards, et al., 2012). The 20% scaling incurs intolerable switching phe-
sterlisation at the nominal temperature proceeds in automatic nomenon in valve opening. Hence, a valve opening scaling of 30% is
fuzzy control mode. A control starting with 100% valve opening, chosen in system implementation for the interest of successful
arrives at the temperature setpoint quickly. While switching from sterilisation, control performance, energy saving, and actuator
open-loop to fuzzy control, fluctuating control action is inevitable. longevity.
The 100% and the 50% scaling controls result in offset responses and
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8 7
6. Conclusions
Acknowledgement
References
Akterian, S.G., 1999. On-line control strategy for compensating for arbitrary de-
viations in heating-medium temperature during batch thermal sterilization
Fig. 7. Comparisons of sterilising times calculated by off-line Eq. (3) and by on-line processes. J. Food Eng. 39, 1e7.
lookup table. Chung, C.C., Chen, H.H., Ting, C.H., 2010. Grey prediction fuzzy control for pH pro-
cesses in the food industry. J. Food Eng. 96, 575e582.
Datta, A.K., Teixeira, A.A., 1988. Numerically predicted transient temperature and
5.2. On-line calculation of F0 values velocity profile during natural convection heating of canned liquid foods. J. Food
Sci. 53, 191e195.
Delgado, A.E., Sun, D., Rubiolo, A.C., 2012. Thermal physical properties of foods. In:
The ability of the on-line lookup table method as an alternative
Sun, D.-W. (Ed.), Thermal Food Processing: New Technologies and Quality Is-
to the traditional off-line equation, Eq. (3), for calculating the F0 sues, second ed. CRC Press, Boca Raton, pp. 3e34. Ch. 1.
value was validated by sterilising cans with corn starch (CH-20) of Farid, M., Abdul Ghani, A.G., 2004. A new computational technique for the esti-
1% (w/v) and 3% (w/v). The preheating temperature was set at mation of sterilization time in canned food. Chem. Eng. Process 43, 523e531.
Honda, H., Kobayashi, T., 2000. Fuzzy control of bioprocess. J. Biosci. Bioeng. 89,
105 C, sterilising temperature at 121.5 C, stream pressure at 1.0 401e408.
bar, and cooling time at 4 min. Fig. 7 illustrates that the two Kumar, A., Bhattacharya, M., Blaylock, J., 1990. Numerical simulation of natural
methods do not have significant difference (p < 0.01) in between. convection heating of canned thick viscous liquid food products. J. Food Sci. 55,
1403e1411.
This means that the proposed study using the lookup table method Kumar, M.A., Ramesh, M.N., Nagaraja Rao, S., 2001. Retrofitting of a vertical retort
is feasible. In other words, the PLC-based online method calculated for on-line control of the sterilization process. J. Food Eng. 47, 89e96.
the F0 value directly and efficiently and hence the quality of a batch Lee, S.J., Kwon, Y.A., 2007. Study on fuzzy reasoning application for sensory evalu-
ation of sausages. Food Control 18, 811e816.
production can be assured. Lee, S.J., Hong, C.G., Han, T.S., Kang, J.Y., Kwon, Y.A., 2002. Application of fuzzy
control to start-up of twin screw extruder. Food Control 13, 301e306.
Moelbak, T., 1999. Advanced control of superheater steam temperatures e an
5.3. Reliability of the on-line method evaluation based on practical applications. Control Eng. Prac. 7, 1e10.
Olaimat, A.N., Holley, R.A., 2012. Factors influencing the microbial safety of fresh
produce: a review. Food Microbiol. 32, 1e19.
In this system, the sterilising time (F0 value) must be no less O'Connor, B., Riverol, C., Kelleher, P., Plant, N., Bevan, R., Hinchy, E., D'Arcy, J., 2002.
than 6 minutes to reduce pathogenic micro-organisms and spoil- Integration of fuzzy logic based control procedures in brewing. Food Control 13,
ages to an acceptable level. The traditional off-line method, Eq. (3), 23e31.
Simpson, R., Almonacid-Merino, S.F., Torres, J.A., 1993. Mathematical models and
is the most popular F0 calculation engine used in commercial logic for the computer control of batch retorts: conduction-heated foods. J. Food
sterilisers. Table 3 lists the demands of steam energy associated Eng. 20, 283e295.
with the two F0 calculation methods. In this system, the autoclave Simpson, R., Figueroa, I., Teixeira, A., 2007. Simple, practical, and efficient on-line
correction of process deviations in batch retort through simulation. Food
consumes 16.7 kg/min of steam. The steam releases 2200 KJ/kg
Control 18, 458e465.
when it condensates from 121 C. Hence in the case of 1% w/v corn Simpson, R., Teixeira, A., Almonacid, S., 2007. Advances with intelligent on-line
starch, the off-line F0 method took 8.15 min and 2.99 105 KJ to retort control and automation in thermal processing of canned foods. Food
complete a sterilisation batch. Clearly, the on-line method results in Control 18, 821e833.
Singh, R.P., Heldmon, D.R., 2001. Introduction to Food Eng, third ed. Academic Press,
smaller F0 values and therefore less steam consumptions. This is San Diego.
arrived at as the on-line method calculates the F0 value in a direct Siriwattanayotin, S., Yoovidhya, T., Meepadung, T., Ruenglertpanyakul, W., 2006.
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003
8 C.-C. Chung et al. / Engineering in Agriculture, Environment and Food xxx (2015) 1e8
Simulation of sterilization of canned liquid food using sucrose degradation as In: Sun, D.-W. (Ed.), Thermal Food Processing: New Technologies and Quality
an indicator. J. Food Eng. 73, 307e312. Issues, second ed. CRC Press, Boca Raton, pp. 81e110. Ch. 4.
Stoforos, N.G., 1995. Thermal process design. Food Control 6, 81e94. Thompson, R., Dexter, A., 2005. A fuzzy decision-making approach to temperature
Syafiie, S., Tadeo, F., Villafin, M., Alonso, A.A., 2011. Learning control for batch control in air-conditioning systems. Control Eng. Prac. 13, 689e698.
thermal sterilization of canned foods. ISA Trans. 50, 82e90. Zanoni, B., Pagliarini, E., Giovanelli, G., Lavelli, V., 2003. Modelling the effects of
Teixeira, A.A., 2012. Simulating thermal food processing using deterministic models. thermal sterilization on the quality of tomato puree. J. Food Eng. 56, 203e206.
Please cite this article in press as: Chung, C.-C., et al., Fuzzy logic for accurate control of heating temperature and duration in canned food
sterilisation, Engineering in Agriculture, Environment and Food (2015), http://dx.doi.org/10.1016/j.eaef.2015.11.003