The tool aims to locate the source and the root causes of a particular problem of quality or safety, by the information recorded from a particular product, regardless of the stage of production where it is - whether raw material, in-process product or finished product. Through the traceability of products is possible to develop prevention and improvement actions, so that a specific problem does not occur again. Traceability can cover only internal actions of the company, or otherwise, may be complete, when it involves the entire chain of production, allowing identifying even basic raw material that led to the final product and locations outside the company where finished products are stored.
Consideration as the consumer safety, as the demands of the institutional environment and the costs of implementation of the traceability system will define the scope more suited to be deployed by the company. The Statistical Quality Control uses statistical tools to control a product or process.
To do this, it works with data collection and the interpretation thereof, acting as a fundamental tool to solve problems in critical product and process.
Thus, ensures the quality sector the product conformity with the specifications defined as ensures the production sector the information needed for effective control of manufacturing processes providing subsidies to decision making in purchasing processes, receiving raw materials and shipment of products and also in reducing cost and waste. From the identification of the market requirements it is collected sufficient statistical information necessary for the development of new products and assists in monitoring the quality profile of competing products.
Although not a mandatory requirement in the food industry, statistical quality control can prove beneficial to organizations in the sector regardless of their particular specialism and size [ 9 ]. According Grigg, the initiatives of training of new graduates entering the industry in the principles of quality assurance and statistical methods and training the existing workforce and management in applying statistical control procedures to processes will make this methods more use of it than they are [ 9 , 10 ].
The industrial statistic includes descriptive statistics, process capability analysis, measurement system analysis, basic graphics as histogram, scatter, box-plot, Pareto diagram, cause and effect, design of experiments, linear regression and correlation, multiple regression, hypothesis testing, confidence intervals, analysis of variance, analysis of process capability, among other tools [ 8 ]. It also covers the sampling techniques and control charts that will be described below, to be very useful to inspection and process control. The inspection process is to analyze or examine units of a product in order to verify with its quality characteristics are in accordance with technical or contractual specifications.
Upon inspection of the product by sampling units are randomly selected to compose the sample batch. Depending on the number of defectives in the sample or the level of quality, that lot is accepted or rejected. Thus, sampling allows, by analysis of a small part of the whole or lot it is possible to draw conclusions about the rest not inspected. Therefore, in the sampling inspection an absolute conclusion about the quality of the lot will never be achieved, there is always a risk rate inherent in the sampling plan and dependent on its discriminatory power.
The current continuous improvement programs that evolve throughout the production chain, call for reducing the use of inspection techniques for the evaluation of the product or process, based on the idea that efforts should focus on "getting it right" in the first time and not in check it, then add value to the product, if it was done properly. However, these inspection techniques for acceptance have restored the importance of quality of audits.
There are two types of sampling plans, sampling plans by attributes and sampling plans by variables. The sampling rate by attributes consists in classifying units of a product just as acceptable or unacceptable based on the presence or absence of a particular feature in each unit qualitative inspected. In the inspection by variable the characteristics or indicators of quality of the product unit are analyzed and the results are expressed by some continuous numeric scale.
While inspection by attributes takes values from the set of integers, inspection by variable takes values in the set of real numbers [ 11 , 12 ]. Upon inspection by attributes the probability of acceptance of the lot is based on Poisson Probability Distribution. The Poisson Probability Distribution is sometimes used to approximate the binomial distribution when the sample size n is too large and the proportion of defectives p is small.
Otherwise, the use of sampling plans by variable assumes that the Normal Probability Distribution fits well with the distribution of the values of the quality characteristic under study. Inspections by sampling can be used in finished products, raw materials, manufacturing operations, products in intermediate stages of processing, stored materials, among others.
There are situations when only one plan by variable applies, for example, when the buyer will accept the product, but will pay different prices depending on the level of product quality. Also when the analysis result of the product will be expressed as quantitative values. For example, in the determination of chemical composition, weight, volume, and physical and rheological measurements. Therefore, measures such as pH, acidity by titration, soluble solids, fat, objective measurements of color and texture, among others, are typical of the sampling variable.
After a sordid litany of recalls courtesy of the food industry, consumers are pointing the finger at companies that have failed to institute proper recall prevention techniques. Authored by a year food management specialist, Food Industry Quality Control Systems is the only single-source summary of basic quality.
The sampling by attributes can be implemented when it wanted to analyze a quality parameter in qualitative terms. Thus they are quite applied, for example, in visual analysis of packaging, the presence of dirt and physical damage in fruit and vegetables. If the process is in control properly, this ratio is around p 0 hypothesis H 0 true.
A single sampling plan by attributes is defined by two parameters: sample size and acceptance number. The likelihood of acceptance of batches relates to the sample size, the severity in the acceptance criterion and the quality level of the products being analyzed in relation to the predetermined quality parameter [ 11 ]. In the sampling plans by variables, the probability of acceptance is related to the quality level of the product under examination and depends on the average of the quality parameter in question and its variability. It also depends on the severity criterion for acceptance of the lot [ 12 ].
Finally, it is worth noting that the Codex Alimentarius recommends the use of the ISO series relating to the procedures for sampling by attributes and the ISO series for the procedures for sampling by variables [ 14 ].
The formal start of statistical process control occurred around , when Shewhart developed and applied control charts at Bell Telephone Laboratories , a telephone company in the United States [ 1 , 7 , 13 ]. As in the entire production process variability occurs, Chart Control or Control Chart, or Map Control, aims to monitor these changes in processes, as well as to evaluate the stability of this process and eliminate or control the causes of variations.
If these values are within limits, without any particular trend, the process is considered under control. But if the points relate outside the control limits or submit an atypical arrangement, the process is judged out of control.
Variability in process may be classified into two types: the variability caused by random or common cause, which are inherent in the process and will be present even considered that this process is fully standardized. If only this kind of cause is acting in the process, it is said that the manufacturing process remains in statistical control. The other type of variability is caused by remarkable and special causes that arise sporadically due to a particular situation which causes the process to behave in a completely different way than usual, which can result in a displacement of the quality level.
Thus, it is said that the process is out of statistical control. The manufacturing control is exercised by the manufacturer during the industrialization process. The goal is to maintain the quality of the product satisfactorily uniform, preventing the production of items outside specification.
The proofing that the process is in control or not is, made by examining unit samples taken periodically out of the production line. If the process is under control, samples that present variability corresponding to samples taken from a normal population, i.
The "under control process" supposes, therefore, that the quality characteristic of all units produced has Normal Probability Distribution Figure 3. Moreover, it also implies that this distribution remains stable, i. So it is said that in a process under statistical control, the variability is attributed solely to random causes.
Shipped from UK. Furthermore, policy and business strategy principles are discussed in the framework of food quality management and the concept of TQM receives special attention. Institutional Offers For institutional orders, please contact sales wageningenacademic. We integrate Human Capital Management HCM systems with apps for tracking employees' health screenings, hygiene compliance, relevant medical histories and compliance training. After a sordid litany of recalls courtesy of the food industry, consumers are pointing the finger at companies that have failed to institute proper recall prevention techniques. Food Service. Other Indicators 4.
These causes of variation do not cause appreciable variation in product quality; its elimination is impossible or anti-economical, and therefore, random causes are considered a natural part of the manufacturing process [ 8 ]. The Normal Distribution consists of an essential notion in statistical quality control rational. When the variability becomes "abnormal" changes in the quality characteristics of the product are sensitive. The causes of modification can be discovered and are therefore called "identifiable causes".
These causes require prompt corrective action, in order to eliminate them. In these situations the samples indicate that the manufacturing process has changed and that the units were produced out of control. Some typical situations in process out of control occur when can be seen points outside the control limits. This is the clearest indication of lack of control of a process, which requires an immediate investigation of the cause of variation.
Also can happened of points of the chart represent a trend, which consists of a continuous motion of the points of the control chart in one direction ascending or descending. Also there is a configuration in sequence in several successive points of the control chart shown in only one side of the center line eight or more consecutive points on one side of the center line. The food industry use control charts in different ways depending upon their level of maturity in statistical thinking [ 15 ]. In a survey conducted in UK food industry, revealed that while there are large differences in process types, quality priorities and key measures among different sub-sectors of the industry, the use of control charts was broadly similar.
This generally extended to the use of control charts for recording or monitoring product net weight and volume data [ 15 ]. There are two types of quality control charts: control charts for variables and control charts for attributes, which will be described below. Control charts for variables are named due to the fact that the quality characteristic being analyzed is expressed by a number on a continuous scale measures.
Some examples of control charts are to yield a formulation, to verify the volume of a drink during their bottling, the soluble solids of a sweet after its cooking and the time to deliver a product to the customer. Some control charts for variables most commonly used are: chart of the average x , chart of amplitude R , chart of standard deviation s.
When a quality characteristic of interest is expressed by a number on a continuous scale of measurement, the two control charts most used are the chart of the average x and a chart of variability R or s. The two charts should be employed simultaneously. Although the benefits of the application of control charts can be obtained in various situations of the food industry, the construction of the charts by variables will be exemplified by a typical situation of the food industry, in a packing operation. Imagine that a poultry slaughterhouse want to control the process of packaging of poultry cuts.
These samples, known as rational subgroup should be taken when one believes that the process is under control and the operating conditions kept as uniform as possible. The sections were collected when the machine was operating within normal procedure, i. See the values of R i in Table 1. The values of D 4 and D 3 are tabulated [ 7 , 8 ]. Build the chart of amplitude Figure 4. Analyze the behavior of the points on the chart of amplitude and verify if the process is in statistical control.
If necessary, recalculate the chart boundaries after the abandonment of the points there are out of control. Repeat this procedure until the control state is reached.