These control charts may be used to maintain the inevitable quality loss within acceptable limits, and make it possible to readjust the storage conditions if the.
Z LebensmUnters Forsch (1994) 199:201-205
Zeitschrift for
9 Springer-Verlag 1994
Original paper Quality control charts for storage of apricots Giiliim ~umnu, Levent Baylndlrh, Mustafa l)zilgen Food EngineeringDepartment,MiddleEast TechnicalUniversity,TR-06531Ankara, Turkey ReceivedJanuary 31, 1994
Qualit~itskontrolldiagramme fiir die Lagerung von Aprikosen Zusammenfassung. Die Darstellung yon Qualitgtsver~nderungen in modifizierten Shewhart-Diagrammen wurde am Beispiel der Lagerung von Aprikosen, die mit Saccharosepolyestern konserviert sind, erprobt. Als Qualit~itsparameter dienten Ver~nderungen des Gewichts, der 15slichen Feststoffe, des pH-Wertes, des Ascorbins~iuregehaltes und der Farbskala nach Hunter. Die Kontrolldiagramme krnnen helfen, unvermeidbare Qualit~tsverluste in akzeptablen Grenzen zu halten und ermtiglichen es, die Lagerbedingungen zu 0ptimieren. Abstract. Construction of modified Shewhart charts to monitor quality variations of apricots coated with edible sucrose polyesters in storage are described by using data based on changes in weight, soluble solids, pH, ascorbic acid, and Hunter colour scale. These control charts may be used to maintain the inevitable quality loss within acceptable limits, and make it possible to readjust the storage conditions if the acceptable limits should be violated.
Introduction Coating fruit with an edible film may provide a barrier to gas exchange between fruit and the environment and extend the shelf life during marketing and consumption. Sucrose polymer coatings are among the edible coatings extensively used to extend the shelf life of various fruit, i.e. apples [1-3], bananas [4], mangoes [5], etc. Semperfresh is a commercial edible sucrose-polyester coating improved by incorporation of a high proportion of short-chain unsaturated fatty acid esters. Most of the research to extend the shelf life of fruit with edible coatings has been made with apples, and variations in skin colour [1, 6-8], firmness [1, 6, 7, 9], titrable acidity [1, 6, 8, 9), respiratory activity [9], pectins [9], weight loss [9], Correspondence to: M. Ozilgen
soluble solids [8, 9], ascorbid acid [9], composition of the internal atmosphere [2], sensory properties [8, 9], and reflectance of the surface [2] were among the measurable parameters used to evaluate quality during storage. Quality control techniques are generally applied to commodities consisting of a large number of individuals. It is not usually feasible to measure the quality attributes of each item, therefore the tests are applied on samples. The samples may be destroyed, i.e. pressed, and the juice analysed for some tests. The sample size is usually chosen to be much smaller than the total size of the commodities in order to reduce the cost of quality control. The Shewhart control charts consist of means and range charts and are used in quality control with measured properties [10-13]. The mean of the replicate measurements may be calculated as: 2 = -
n
x~.
(1)
i=1
Where x is the sample mean, xis are the values of the individual items constituting the sample, and n is the number of the items in the sample set. The means charts are constructed to control the means of the replicate samples within the required upper (UCL) and lower (LCL) control limits. The range (Ri) of a sample is defined as: R i = Xi,max -- Xi, min,
(2)
where x~.max and x~. raincorrespond to the maximum and minimum measurements of the quality factor in the i th sample set, respectively. The mean value of the ranges may be calculated as:
k
/~=~1 ~
R~
(3)
i=l
where k is the number of the sample sets. The range charts are constructed to assure that the range, i.e. the variance, of the individual samples are confined within pre-determined limits. In such a quality control plan the central line (CL~) and the upper (UCLx) and lower (LCLx) control limits of the means chart may be calculated as [10]:
202 k
1 CL x = ; ~
~,
(4)
4=1
3/~ UCL~ = CL~ + d2---~,
(5)
3/~ dav~ ,
(6)
LCL~ = CL~
where d 2 is a constant depending on the sample size n [10]. The central line (CLR) and the upper (UCLR) and lower (LCLR) control limits of the means chart may be calculated as
[10]: CL R = / ~ ,
(7)
UCLR = D4~,
(8)
LCL R = D3/~ ,
(9)
where d 2, D 3 and D e are constant, their values depending on the sample size and listed in statistical quality control tables [101. A storage process may be controlled by directly measuring the properties of the store room, i.e. temperature, or those of the apricots, i.e. soluble solids, pH, etc. With the latter it is possible to determine the rate of quality loss. In any statistical quality control plan it is attempted to maintain the ranges and the means of the samples within pre-determined control limits. Violation of these limits, especially in the early stages of storage, warns the operators that the apricots will deteriorate if the storage conditions are not improved. Then the operators may readjust the storage conditions, i.e. may lower the store room temperature, etc., to slow down further quality loss. Deterioration of apricots is irreversible; changing the storage conditions after the fruits have deteriorated cannot improve the quality. The parameters to be used in construction of the modified Shewhart charts should show a linear trend during the storage period and describe a quality attribute with reasonable sensitivity. Equations (4)-(9) are used in processes where the quality factors are maintained within the pre-specified constant limits, and are not suitable for use in quality control of the apricots in storage, where deterioration is inevitable and acceptable at a reasonably slow rate. In the present study we have modified Eqs. (4)-(9) to make them suitable for controlling apricot quality in storage, then have chosen appropriate quality factors to be used with these equations under different coating applications.
method [14]. Soluble solids and the pH values were measured with pressed apricot juice using a digital refractometer (Atago, RX 100, Japan), and a pH meter (Coming-EEL, Model 12, UK), respectively. Hunter colour chart parameters (lightness, a and b) were measured with an ultraviolet-visible spectrophotometer (Shimadzu UV 2100, Japan) from two opposite cheeks of each apricot. Each measurement was replicated eight times. The pH, soluble solids and ascorbic acid contents of the fruits were determined in three replications.
Results and discussion
Weight loss It was observed that the coatings generally improved quality retention during storage. Weight loss slowed down with increasing Semperfresh concentration (Table 1A). These results contradict those of Smith and Stow [2], who observed no significant effect of the coatings on the weight loss of Cox's Orange Pippin apples. Banks [4] reported only a slight reduction in weight loss of bananas after coating with sucrose polymers. These results may imply that the effectiveness of the sucrose polyesters in preventing weight loss greatly depends on the kind and the variety of the fruit. The average weight of the apricots decreased linearly with time. The central line of the means chart was: CL z = CLx0 - a t ,
(10)
where CLxrepresents the fitted sample means, CLx0 is the initial average weight of the sample, and c~ is the coefficient of the decrease in weight with time t. Values of the sample mean weight were calculated using Eq. (1). Constants CLx0 and c~ were calculated with linear regression to minimize the sum of squares error. It was also seen that the range of weight loss was proportional to the average weight loss of the samples during the storage period: R = a + b~,
(11)
where x represents the time-dependent sample means, a is the initial average range of the samples, and b is the coefficient of increase in range in proportion to the sample mean. Equation 10 was substituted in Eq. (11) forx, and then a timedependent expression was obtained for the range of samples: CL R = CLR0 - fit,
(12)
where CLn represents the fitted sample ranges, CLR0 is the initial range of the samples, and 13is the coefficient of the increase of the ranges with time. The values of the constants CLRB and/3 were calculated with linear regression to miniMaterials and methods mize the sum of squares error; Eq. (12) is given in Table 1A for the specific experimental conditions. Equations (10) and The apricots were harvested from a local orchard just before ripening (12) were substituted in Eqs. (5) and (6) for CLx and CL R, rewhile their surfaces were yellowishgreen. They were coated with 0.5, spectively, to calculate UCL x and LCLx (Table 1A). Equation t.0 or 1.5 (w/v) of Semperfresh (Samper BioTechnology,UK). The con(12) was substituted in Eqs. (8) and (9) to determine UCL R trol apricots were dipped into water only. The coated apricots were and LCL R, respectively. The means and range charts for washed and their natural wax was removed before coating. The control apricots were dipped into water only. The coated or control apricots weight loss are exemplified in Fig. la. Since the ranges of the were stored at room temperature (20~ C) during the experiments. samples changed with time, different values were observed Weight loss with each treatment was determined with a sample of with the slopes of the means chart parameters pertaining to ten apricots. A laboratory balance (Mettler, Model P 163, Germany, senthe same experimental conditions and LCL x, CLx and UCL x sitive to 0.001 g) was used for weight loss determination.Ascorbic acid concentrations were measured with the 2,6-dichlorophenol-indophenol were not parallel (Fig. la).
203 A Controlchart parametersfor weightloss (%)
Table 1. Controlchart parametersfor the storage of the appricots
Coating
LCL~
CL~
UCL
LCLn
CLR
UCLR
Control 0.5% 1.0% 1.5%
4.38+4.52t 2.90+4.12t 2.61+3.51t 3.05+3.40t
5.48+4.99t 3.64+4.69t 3.23+4.04t 3.17+4.19t
6.72+5A5t 4.37+5.27t 3.85+4.54t 3.29+4.74t
0.85+0.34t 0.53+0.42t 0.45+0.36t 0.09+0.40t
3.80+1.5lt 2.39+1.88t 2.00+1.61t 0.39+1.78t
6.76+2.68t 4.25+3.34t 3.56+2.87t 0.69+3.16t
B Controlchart parametersfor solublesolids(~ Coating
LCL~
CL~
UCL
LCLe
CLR
UCL~
Control 0.5% 1.0% 1.5%
9.8+0.8t 10.1+0.6t 9.9+0.6t 9.9+0.6t
10.2+0.8t 10.4+0.6t 10.3+0.6t 10.2+0.6t
10.5+0.8t 11.0+0.6t 10.7+0.6t 10.6+0.6t
0 0 0 0
0.46 0.40 0.58 0.50
1.05 0.91 1.32 1.14
C Controlchart parametersfor pH Coating
LCL~
CL~
UCL~
LCLR
Control 0.5% 1.0% 1.5%
3.06+0.03t 3.05+0.02t 3.05+0.02t 3.05+0.02t
3.07+0.03t 3.06+0.02t 3.55+0.02t 3.06+0.02t
3.07+0.04t 3.07+0.03t 3.06+0.03t 3.07+0.02t
0 00 0
CLR 0.004+0.007t -0.126+0.005t 0.089+0.004t 0.008+0.003t
UCLR 0.010+0.018t 0.031+0.016t 0.018+0.010t 0.021+0.009t
D Controlchart parametersfor asc0rbicacid loss (X=ln[mgascorbicacid/100g apricottissue])
LCL, lower controllimit;CL, central line; UCL, upper controllimit;2-, samplemean; R, range; t, time
= o
Z
Coating
LCL~
CL
UCL~
Control 0.5% 1.0% 1.5%
2.68-0.06t 2.69-0.05t 2.69-0.05t 2.69-0.04t
2.73-0.04t 2.72-0.04t 2.73-0.03t 2.73-0.02t
2.78-0.028t 0 2.80-0.025t 0 2.80-0.003t 0 2.78-0.001t 0
60[ 9
0
2.5
5
"
7.5
10
0
2.5
5
0
t
li ~ ;
2.5
r/l
5
7.5
10
2.5
I
I ~ I / _t~
5
7.5
.-Jl~
~ .
9
9
2.5
5
7.5
10
0
10
2.5
5
7.5
t~ io.o4
~
10
1` 2.7
10-4 ~ 1`
~
ja3
23 0
1`
Io-~ I0
'r
0
" 2.5
5
7.5
i00, 10
0
2_5
5
7.5
10
Time (days) -->
0.13+0.042t 0.14+0.037t 0.10+0.055t 0.12+0.050t
The means charts for the soluble solids contents of the apricots were constructed similarly to those of the weights. The range of soluble solids measurements neither increased nor decreased consistently during the storage period, therefore Eqs. (7)-(9) were used to construct the range charts. Equations (5) and (6) were used for the UCL x and LCL x charts, respectively, aft_er substituting Eq. (10) for CL x and the numerical values of R. The means and range charts for soluble solids are demonstrated in Fig. lb, and their equations are tabulated in Table lB. Since the ranges of the samples were not a function of tirne, the same value of a was observed with the slopes of all the means chart parameters, i.e. LCL x, CL x, and UCLx, pertaining to the same experimental conditions (Table 1B).
1.2 ~ 1'
?,I* s
0.05+0.016t 0.05+0.014t 0.05+0.021t 0.05+0.020t
2O
"1"18r
"
UCL~
Soluble solids
o 10
7.5
CLR
40
~
1`45 . 30
LCLR
pH variations D u r i n g r i p e n i n g the a c i d s o f a p r i c o t s d o n o t c h a n g e m u c h [ 1 5 ] . T h i s is a l s o c o n f i r m e d w i t h the m e a s u r e m e n t s s u m m a -
Fig.1 a - d . Sample qualitycontrol charts. Experimentaldata are shown in symbols( I ) ; solid lines describethe lower and the upper controllimits: 2-, sample mean; R, range a Weight loss with 1.5% Semperfresh coating,b Variationin solublesolidswith 1% Semperfreshcoating,e pH changes with 0.5% Semperfreshcoating,d Ascorbicacid loss with control(0% Semperfreshcoating)
204 rized in Table 1C. It was also shown in Table 1C that 1.0 and 1.5% Semperfresh treatments were effective in slowing down pH change. The effect of Semperfresh coating in slowing down changes in acidity was also shown by Chai et al. [3] with cold stored apples. Dhalla and Hanson [5] reported an increase in the ascorbic acid retention in mangoes treated with sucrose polyesters. Both the means and the range charts for variation of the pH were constructed similarly to those of the weights.
Loss of ascorbic acid Our results show that the rate of ascorbic acid loss decreases with increasing Semperfresh concentrations (Table 1D). The loss in average ascorbic acid of the samples was described with Eq. 1 where parameters CLx and CLx0 are the average of the natural logarithm of the measurements, and their initial value, respectively; parameter a is the coefficient of the loss of ascorbic acid with time. Loss of ascorbic acid generally agrees with a first-order irreversible, unimolecular kinetic rate expression, therefore working with the natural logarithm of the measurements was preferred and parameter X was calculated as ln[mg ascorbic acid/100 g apricot tissue]. Constants CLx0 and a were calculated by linear regression. The range of the ascorbic acid measurements increased consistently during the storage period, therefore the same procedure was followed as that of weight loss to construct the means and the range charts for the loss of ascorbic acid. The means and range charts for ascorbic acid loss are exemplified in Fig. ld, and their equations are tabulated in Table 1D. Table 2. Control chart parameters for Hunter colour scale parameters
Variations
in the Hunter colour scale parameters
The results summarized in Table 2 shows that variation of lightness and parameter b of the Hunter colour scale slowed down with increasing Semperfresh concentrations (Table 2). The reduction in colour change of apricots by means of the coating is attributed to the reduction in respiration rate. Coatings affect colour change either by affecting the chloroplast structure or the chloroplast degradation process [16]. The inverse of the Hunter a value is a measure of the green colour. A substantial reduction has been observed in the variation of the Hunter a value with 1.5% Semperfresh coating, implying the slowing down of the colour change process (Table 2C). These observations agree with those of Santerre et al. [8] and Chai et al. [3], who reported improved colour retention with apples upon coating with sucrose polyesters. The same procedure as weight loss was used to construct the quality control charts with lightness and parameter b. Variation in parameter a with time shows a sigmoidal curve. The values of parameter a were negative at the beginning of the experiments; they became positive after a few days of storage. Parameter a was transformed into a new variable:
X=In[( '-+--3'-~/(1a+_7"~]. amax + ")//
(13)
arnax + 3 ' / J
Parameter y is a constant; its value was found by the trial and error procedure as 3 with control, 0.5% and 1% Semperfreshcoated samples. The value of parameter y was 4 with 1.5% Semperfresh-coated samples. Parameter y was included in Eq. (13) to eliminate the problems caused by the negative values of
A Lightness Coating
LCLx
CLx
UCLx
LCL R
Control 0.5% 1.0% 1.5%
38.63-0.50t 38.65-0.38t 38.73-0.31t 38.62-0.27t
39.13-0.34t 39.21-0.25t 39.27-0.19t 39.16-0.18t
39.62-0.18t 39.77-0.11t 39.81-0.06t 39.67-0.08t
0.18+0.06t 1.32+0.43t 2.47+0.80t 0.21+0.05t 1.53+0.36t 2.84+0.67t 0.19+0.05t 1.43+0.34t 2.67+0.63t 0.19+0.03t 1.42+0.25t 2.64+0.47t
CL R
UCL R
B Parameter b Coating
LCLx
CL~
UCLx
LCLR
CLR
UCLR
Control 0.5% 1.0% 1.5%
17.54-0.41t 17.67-0.38t 17.77-0.34t 17.65-0.29t
17.91-0.32t 18.13-0.30t 18.14-0.24t 17.97-0.18t
18.28-0.23t 18.57-0.22t 18.51-0.14t 18.26-0.88t
0.14+0.04t 1.01+0.24t 0.16+0.03t 1.21+0.22t 0.14+0.04t 1.00+0.26t 0.11+0.04t 0.84+0.29t
1.89+0.45t 2.25+0.40t 1.86+0.49t 1.56+0.53t
CLx
UCLx
LCLx
CLR
UCLR
1.49 1.64 1.99
2.68 2.95 3.58
C Parameter a Coating
LCL~
Control 0.5% 1.0%
-3.99+0.62t -3.43+0.62t -2.87+0.62t 0.20 -4.60+0.62t -3.67+0.62t -3.06+0.62t 0.23 -4.60+0.63t -3.86+0.63t -3.12+0.63t 0.27
Parameter a with 1.5%Sernperfresh LCLx=-2.67+0.319t-0.244exp(0.145t) CLx=-2.67+0.319t UCLx=-2.67+0.319t+0.244exp(0.145t)
LCLR=0.09exp(0.1450 CLR=0.66exp(0.145t) UCLR=1.18exp(0.145t)
205 parameter a at the beginning of the experiments. Adding a constants to the data for transformation into more convenient forms is among the tools for analysing especially the non normal data [17]. The transformation described in Eq. (13) is used to convert sigmoidal plots into linear form [ 18]. Parameter am~x was the maximum attainable value of a. Quality control charts for Hunter scale parameter a with 0%, 0.5% and 1.0% Semperfresh-coated samples were prepared similarly to those of the soluble solids data. With the 1.5% Semperfresh-coated apricots, the ranges of the samples changed with time as: CL R = R 0 e~ ,
(14)
where R 0 was the range at the beginning of the experiments and/3 was a constant. Parameters R0 and/3 were obtained from the experimental data with regression after linearizing Eq. (14) and their values are given in Table 2. Quality control charts for variations in parameter a were constructed with 1.5% Semperfresh-coated apricots similarly to those of weight loss, but Eq. (14) was used instead of Eq. (12). While studying the data obtained with the soluble solids, pH and ascorbic acid loss, we recognized that the LCL Rwas negligibly small even at the end of the storage period, therefore we preferred to set it zero (Table 1B-D). In Figs. 1 and 2, construction of the quality control charts were exemplified by storage of the apricots coated after harvest with different Semperfresh treatments. Equations of the quality control chart parameters for the other coating applications are given in Tables 1 and 2. Distribution of the data points on these quality control charts were similar to those of the examples (Figs. 1 and 2), therefore those figures are not given. In the present study, variations in weight, pH, soluble solids, ascorbic acid and Hunter colour scale parameters were found to be appropriate quality factors for construction of the modified Shewhart charts since their variation with time had a linear trend. With ascorbic acid and Hunter scale parameter a, linear behaviour was obtained after appropriate transformations. These control charts may be used to maintain the inevitable quality loss within acceptable limits, and make it possible to readjust the storage conditions if the acceptable limits should be violated. Choosing these parameters for the construction of the quality control charts does not imply that other quality factors should be neglected. Different quality factors may be needed to monitor storage of the other fruits or varieties. It was shown in Fig. 1 that a large scatter was obtained with weight, pH, soluble solids and ascorbic acid means charts. We may conclude from these results that the possibility of detecting a violation from acceptable quality loss is higher with these charts in comparison with the others. Shewhart charts are extensively used in quality control in the manufacturing processes, but they are not suitable for use with the storage of the fruits due to the continuously changing structure of biological materials. We have suggested a procedure for construction of modified Shewhart charts to keep the inevitable quality loss in storage within acceptable limits. Variations in weight, soluble solids, pH, ascorbic acid and Hunter colour scale parameters of apricots were found
=
.5
=
36 0
2
4
6
8
10
0
2
4
6
I~ 14 0
2
4
6
8
10
0
2
4
6
8
10
18 17
15 8
10
5 T
o
2
4
6
8
lO
0
2
4
6
8
10
Time (days)--~
Fig.2a-b. Sample quality control charts prepared with the Hunter colour scale data. Experimentaldata are shown in symbols (I); solid lines describe the lower and the upper control limits, a Variationof lightness
with 0.5% Semperfreshcoating, b Variation of parameter b with 1% Semperfreshcoating, c Variationof parameter a with 1.5% Semperfresh coating
appropriate for the construction of these charts, but there is still a need for more data obtained with other cultivars and fruits.
References
1. DrakeSR, FellmanJK, Nelson JW (1987)J Food Sci 52:1283-1285 2. Smith SM, Stow JR (1984)Ann Appl Biol 104:383-391 3. Chai YL, Ott DB, Cash JN (1991) J Food Process Preserv 15:197-214 4. BanksNH (1984)J Exp Bot 35:127-137 5. DhallaR, Hanson SW (1988)Int J Food Sci & Techno123:107-112 6. Lau OL, YastremskiR (1991)HortSci 26:564-566 7. Chu CL (1986)HortScience 21:267-268 8. Santree CR, Leach TF, Cash JN (1989) J Food Process Preserv 13:293-305 9. TewariJD, Pandley N, Rai RM, Ram CB (1980)Indian J Plant Physiol (New Delhi) 23:257-265 10. Miller I, Freund JE (1985) Probability and statisticsfor engineers, 3rd edn. Prentice-Hall, Englewood Clifs, pp 426-431 11. HubbardRH (1990)Statisticalquality control for the food industry. Van Nostrand Reinhold, New York, p 29-51 12. MontgomertyCC (1991) Introduction to statistical quality control, 2nd edn. Wiley, Singapore, p 101-146 13. Grant EL, Leavenworth RS (1980) Statistical quality control, 5th edn. McGraw-Hill, Singapore, p 1-29 14. AOAC(1975) Officialmethods of analysis, 12th edn. Washington DC, Associationof OfficialAnalyticalChemists 15. Kara.9.alaI (1990) Bah~e i~riinlerininMuhafazasave PazarlanmasL Ege Universitesi Baslmevi, Izmir (Turkish) 16. Smith SM, Geeson J, Stow JR (1987)HortSci 22:772-776 17. Jacobs DC (1990)ChemEng Prog 86:19-27 18. WeissRM, Ollis DF (1980)Biotech Bioeng 22:859-873
