ORGANOLEPTIC ASSESSMENT
Food is made for eating, and the controlled eating of food to validate its acceptability is an important part of quality assurance. Assessment may include appeal to several senses, as colour, texture, smell and taste can all be important attributes.
In some manufacture, informal tasting on the line can be a valuable quality check, but the circum- stances and method of such tasting should be carefully controlled to maintain hygiene standards.
Off-line tasting should always be confined to a designated tasting area, where samples can be judged safely, impartially and free from external pressures, by informed tasters.
Buying samples need to be assessed by taste for suitability as replacements for existing supplies or for continuity of desired character from season to season. Organoleptic assessment may then be needed for delivery samples to be compared with buying samples.
Intermediates and finished products will be judged for continuity of character.
In some cases, the tasting will be of the finished product prepared in the same way as the consumer would for serving, but in other cases a simpler taste medium will be used; for example, spices may be tasted in a boiled starch solution. Analytical tastings will also be used to assess shelf life and to assist performance assessment of different packag- ing systems and changed processes. For this purpose, a trained panel sufficiently large to allow calculation of the statistical significance of the results is desirable.
Routine daily tasting of a range of factory products by a selected panel is a useful way of maintaining awareness of quality throughout the factory. These tastings should be carefully super- vised and structured if they are to continue to achieve their objective.
For certain types of product there exist special methods of tasting and assessment which have become industry standards. Examples of these are the Campden Food and Drink Research Associa- tion's methods for the assessment of standards of quality of canned fruits and vegetables, frozen fruits and vegetables and so on, and the Torry Taste Panel system for assessing freshness in fish.
In the case of the Campden standards the methods include tasting as well as a complete list of other standards, e.g. size grade, defects, foreign material, which enable the assessor to arrive at a numerical
QUALITY ASSURANCE AND CONTROL OPERATIONS 555
value of quality which is translated into alphabetical grades (e.g. Grade A).
Tasting is an important part of product innova- tion with a need to taste competitive products and numbers of alternative recipes. It is very important that these tastings are controlled and recorded without limiting innovative flair and generation of ideas.
Clearly, due regard must be given to safe hygienic preparation of food for tasting, and samples must be fit for consumption.
There are many techniques and applications of organoleptic assessment, both for research and development and for market research purposes, some of which, e.g. flavour profile analysis, may be very complex both in use and analysis, and require particular expertise. They are not generally applicable in day-to-day quality assurance and so are outside the scope of this chapter.
REFRACTOMETRIC SOLIDS
The normal method for measuring the soluble solids present in liquid foods or those which are more solid, such as preserves is by means of a refract- ometer.
Refractometers are available calibrated with refractive index and/or % sugar scales. The style varies from laboratory instruments with which readings accurate to 0.1 % sugar can be obtained if the temperature is accurately controlled, to hand- held instruments which can be readily used in a factory environment and will enable readings to be taken accurate to 0.5%. In addition, instruments are also available which can be fitted to processing plant, with or without feedback control features.
Whilst the reading on a sugar scale is true only for a pure sucrose solution, the advantages of using a reliable, speedy and simple method outweigh the disadvantages of the inaccuracies which occur due to the presence of other soluble materials. The sugar scale is used and the results are expressed as 'refrac- tometric solids'. Current legislation specifies that soluble solids should be determined by refract- ometer at 20°C with limits of ± 3% of the refract- ometer reading.
Cut-out
When examining cans of fruits or vegetables to determine their quality, measurements are made of the weight of the drained solids and the weight and density of the syrup or covering liquid. This exam- ination is known as the 'cut-out'.
Syrup which is added to fruit at a density of 45°
Brix will be diluted by the water present in the fruit, and will generally cut-out in the region of 25-30°
Brix. Similarly, an original syrup strength of 40°
Brix usually cuts out at 21-26° Brix. The final density depends on the particular fruit, its variety, and state of ripeness, and on the ratio of fruit to syrup in the can.
If the weight of the original fruit packed in the cans is known it is possible to calculate with a fair degree of accuracy the strength of the syrup used.
The weight of the drained solids in any particular fruit does not bear a constant relationship to the original weight, as many factors-such as the time and temperature of exhaust, condition of the fruit, and strength of syrup-all influence the final cut- out weight of the solids.
Canned fruits packed in a syrup of 40-45° Brix generally show the following drained weights (expressed as percentages of the filled weights): goo- seberries, 86-100; strawberries, 54-82; raspberries, 62-95; loganberries, 70-91; blackcurrant, 75-95;
blackberries, 66-90; cherries, 83-100; plums, 72- 97; damsons, 83-96 per cent. The wide variations are occasioned by differences in texture. Fruits in a syrup of 30° Brix show drained weights 2 per cent less than the above. The drained weights of vegeta- bles are: dwarf or runner beans (whole) 95-109;
(sliced) 101-110; beetroot (whole) 97-104; (sliced) 94-98; (diced) 92-96; broad beans 96-106; carrots (whole) 95-108; (sliced) 98-105; (diced) 98-104;
cauliflower 94-100; celery 88-100; peas (fresh green) 98-114; potatoes (new) 100-119; turnips 98-102 per cent of the original filled weights.
QUANTITY CONTROL
Two basic approaches to the control of weight or of volume are found in national and international legislation, one where the declared quantity is the minimum to be expected by the purchaser, and the other where the quantity is, within certain limits, the average amount packed by the producer. There are also circumstances in which products are sold simply by count, and derogations exist for very small packs and for various sorts of portion control packs. As the situation in these sectors is relatively straightforward, they are not covered in any detail here.
Minimum quantity
The minimum quantity requirement, such as formed the basis of UK legislation until the implementation of the EC Directive 76/2ll/EEC, really means that the product as sold must contain not less than the
contents declared on the label, and in theory this could be literally enforced, e.g. one grain under the declared ounce is an offence. In practice most authorities enforcing legislation of this type are aware of the statistical facts of life, and, assuming a normal distribution, are likely to regard 'minimum' as something akin to two standard deviations below the mean and regard such distribu- tions as acceptable, gross defects excepted. Hence the manufacturer should normally set his target fill quantity so that it is at least two standard deviations above his contents declaration in the worst operating circumstances. Clearly appropriate adjustments to this must be made if the fill distribution is not statistically normal, and to take account of any checkweighing and sorting operations which are undertaken after filling. With some products and packaging materials, it will also be necessary to consider any quantity changes which take place during storage and distribution.
Average quantity
For manufacturers operating in the European Community and many other markets, it is the 'average quantity' concept which must be complied with. In the UK, most packaged goods are subjected to the average system prescribed in the Weights and Measures Act 1979, as amended, and its ancillary regulations, which embody the requirements of the EC Directive.
This legislation requires that when a group of packages is examined in a defined manner by an authorized inspector, then the average quantity in the group must be the same as or greater than the declared quantity on the individual packages.
Whereas the minimum quantity legislation was primarily enforced by checks made at the point of sale, the Average Quantity law is primarily enforced by inspection at the point of production or importa- tion. Certain limits and tolerances are defined which must be complied with, and the legislation is enforced by the requirement that when formally tested by an inspector (a reference test), the batch or consignment in question does, in fact, pass that test. If it fails the test, then the sale of that batch is prohibited unless or until correction takes place.
The reference test is carried out on a sample of packages taken either from one batch of product or from one hour's production from a continuous line.
The average content of the packages is checked, after allowing for sampling error, and the number of packages which fall short of the declared quantity by more than a certain level, the tolerable negative error (TNE) , is noted. The latter packages are referred to as 'non-standard' and their number
must not exceed 2 Y2 % of the sample. Packages with a shortfall exceeding twice the TNE are known as 'inadequate', and if any are found in the sample, then the reference test is failed. To ensure that a reference test will not be failed, a packer may opt to measure the quantity in every package to see that no package contains less than the declared quantity. In most cases this is likely to be unaccep- tably expensive, since the quantity in the packages will inevitably be significantly greater than the declared quantity. The more usual procedure for packers is to install a control system which satisfies what are known as the 'Three Packer's Rules':
(i) The actual contents of the packages shall be not less, on average, than the nominal quantity
(ii) Not more than 2 Y2 % of the packages may be non-standard, i.e., have negative errors larger than the TNE specified for the nominal quantity
(iii) No package may be inadequate, i.e., have a negative error larger than twice the specified TNE.
The TNEs are specified in the Regulations, and range from 9% of the declared quantity of packages between 5 and 50 g or ml, to I % of the declared quantity of packages larger than 15 kg or litres. The higher percentage tolerances for packages contain- ing small amounts recognize the limits of the precision obtainable with normal production equipment.
The average weight legislation is designed to fit the observation that most popUlations of pre- packaged goods have a normal frequency of distri- bution in statistical terms (see Figure 16:3). Also the Tolerable Negative Errors given in the Regulations are twice the standard deviations to be expected in products which are 'difficult to fill'. So long as the population is more or less normally distributed, with the standard deviation approximately half the appropriate TNE, then 2.5% of the population will fall between T 1 and T 2 (the area hatched in Figure 16.3 between I and 2 TNEs below the average).
Hence when the population is normal and the standard deviation is not more than half of the TNE, when Packer's Rule I is satisfied then Packer's Rules 2 and 3 will also be satisfied. If either of these conditions is not met, a packer will need to take steps to ensure that each of the three rules is satisfied independently.
Unless the packer chooses to measure the quantity of each individual package, the Act and Regulations require that records of the checks are kept which are carried out to maintain quantity
QUALITY ASSURANCE AND CONTROL OPERATIONS 557
>-
"
c Q) ::J CT Q)
G::
3.728
X Quantity in package
Figure 16.3 Normal frequency distribution. Qn, nominal quantity, TJ = Qn - TNE; T2 = Qn - (2 x TNE); S = standard deviation; x = mean quantity.
control. The form of the records is not specified and is subject to agreement between the packer and the local Inspector. It is specified, however, that they must be held for a year and be available for inspec- tion as evidence that checks have been carried out during all production runs.
Packers of any goods in quantities between 5 g or 5 ml and 10 kg or 10 litres may voluntarily come under the average system by 'e'-marking the packages and thereby choosing to be controlled by EC Directive 76j211jEEC. Another benefit of 'e'- marking packages is that they can be exported for sale in other Member States without any further metrological control than that exercised in the packing factory.
Although in the UK nearly all pre-packed foods are subject to the Weights and Measures Act, 1979, there are exceptions, such as for very small quantity packages. These exempted products must satisfy the requirements of the Weights and Measures Act, 1963, which requires that each package must contain at least the declared quantity.
Control methods
In considering a quantity control system, there are three acceptable methods if it is impractical to fill every package to the declared quantity under manual control: (i) control by sampling; (ii) control by automatic checkweighing machines; and (iii) for liquids, control by using measuring container bottles.
(1) Control by sampling. If a packer opts to control the quantities in his packages by measuring samples of product, he must consider the variability of the performance of the filling equipment. The first question should be whether the frequency distribution is approximately normal, with a standard deviation less than half of the appropriate TNE. If not, simply ensuring that
the average quantity of the samples is above the declared quantity will not guarantee against excessive numbers of nonstandard and inadequate packs being produced. In these cases the packer will need to increase the target quantity until the second and third Packer's Rules are satisfied. Alternatively, a check-weigher may be used to remove the under- filled packages, thus shifting the distribution upwards and increasing the average quantity in the remaining packages.
A recommended method of assessing the variabil- ity of the performance of filling equipment is given in the UK Department of Trade's Code of Practical Guidance for Packers and Importers. This procedure will identify the sources of some of the variability. It also gives the method to be used to calculate the minimum amount to which the target quantity needs to be increased to guarantee compli- ance with Packer's Rules 2 and 3.
When the underlying pattern of filling variability has been established, the minimum target quantity which will satisfy the legal requirements may be calculated. A sampling allowance is almost always necessary when a process is being controlled by sampling so that the quantity control system does not give signals which are too early, so that unne- cessary adjustments are made, nor too late so that large numbers of unsatisfactory packages are produced before a problem is detected. Appropri- ate allowances are given in the DoT Code and these help determine the level of the target quantity above the nominal quantity and the amount of checking and control which will be necessary to ensure compliance with the law.
Popular quantity control systems involve plotting results on charts bearing pre-determined control limits and taking action when these control limits are exceeded. Three main types of control charts have evolved: Original Value plots; Shewart charts and Cumulative Sum charts (Cusum), and further information on their use is to be found in many books on statistical quality control.
(2) Control by automatic checkweighing. Automatic checkweighers measure the gross weight of indivi- dual packages on a production line. They check whether the packages are below a certain weight, the set point, and if so, automatically remove them. Such devices usually provide for the display of information about the characteristics of the pro- duction. Simple equipment may provide a digital display of the total throughput and the numbers of packages below or above each set point, and record the data. It is usually possible to count the packages outside any set point without necessarily removing them.
Because it is inevitable on a checkweigher that packages are weighed gross, an allowance for the variability of the weight of the packaging material must be included in setting the target weight. A further allowance is required for the 'zone of indeci- sion' of the checkweigher, i.e. the weights at which the checkweigher does not always give the same response. The determinations of both of these allow- ances are described in the DoT Code.
Some patterns of checkweigher send a feed-back electrical signal to an adjustment mechanism, e.g. a servo motor on the filling machine, in response to the data it records. This can be very effective in controlling average weight where gradual drift occurs. Other more sophisticated equipment allows full analysis on a continuous basis and provides full compliance records of the line operation.
(3) Control by measuring container. The Measur- ing Container Bottles (EEC Requirements) Regula- tions 1977 made it legal for bottle manufacturers to sell officially volume controlled containers to packers. These measuring container bottles are specified as having a capacity equal to the nominal capacity at a specified level from the brim. The use of measuring container bottles allows production to be checked in an adequate manner by reference to the liquid level using a template, without having to check the actual volume of the contents.
DATE MARKING AND PRODUCT CODING In the UK it is now a legal requirement that all foods be date or lot marked in some way The Food (Lot Marking) Regulations 1992. Effectively, since the coming into force of the Food Labelling Regula- tions 1984, most foods have been marked with an indication of minimum durability in the forms of Best before ... Best before end ... or Use by ...
dates. There were originally some exemptions to this which have recently come within this requirement such as frozen foods. There still are some exemp- tions to this requirement, such as vinegar, salt, chewing gum and so on. All these will now have to bear some indication of production date or batch number even if in encoded form.
There are several ways of marking products.
Marks may be embossed into the surface of the pack as in canned goods, printed on to labels or packaging film and labels may be edge coded by cutting slots into them. However in recent years the most significant development has been the ink jet coder which marks packs with charged particles of ink and is very versatile in what can be printed.
Product coding should be looked upon as a useful tool in any quality system and it is to the advantage of the producer that the product is coded for the following reasons.
Should any complaint arise the producer will be able to identify the production day and possibly the line and operative responsible. Ink jet coding even makes it possible to mark with a precise production time from which it may be possible to investigate fully the cause of the problem if genuine or be absolved from responsibility if the records indicate that such a problem could not have occurred during production. Where a problem is genuine the date marking system should also facilitate traceability of the product from the finished item back through to the ingredients supplier. Where a problem results in a product withdrawal a good coding system can mean the number withdrawn can be narrowed to a much lower number than would otherwise be the case. See also EMERGENCY PROCEDURES.
CONTROL CHARTS AND SAMPLING SCHEMES
When sufficient data are available it is possible to carry out statistical checks to balance the risk of passing a consignment of material containing too many defectives against the number of samples taken. A double sampling scheme which makes use of a low level of sampling initially, but a higher level of sampling when the proportion of defects rises, can be an economic way of ensuring a satisfactory quality level.
When sufficient data are not available for sound statistical sampling plans to be used, it is useful to plot information as it becomes available in order to identify adverse trends. Sampling rates should then be increased as necessary until normal performance is restored.
In the control of a process or manufacturing operation, the use of control charts to signal the need for change or adjustment is common prac- tice. Such charts require a knowledge of the opera- tion and its variability to allow a target, minimum requirement and range of acceptable variation to be prescribed. These may then be used to control the process within straightforward limits as indicated in Figure 16.4, adjustments being effected when results fall within the action zones.
Cumulative Sum charts (Cusum) are valuable for controlling situations where events are likely to alter relatively gradually. They allow adjustments to be made in response to a change in trend, rather than to one or two individual results, and they constantly