Advanced

En kvalitetskontroll - Snustillverkaren Fiedler & Lundgren kvalitetstestas

Timglas, Andreas (2008)
Department of Statistics
Abstract
This paper aims to describe the variation and develop a method to control the production for the product Metropol Kaktus from the snuff developer Fiedler & Lundgren. We are going to use the requirements from the product specifications as limits of the expected quality when we present methods to control the process. One method attempts to control during the process and the other attempts to maximize the number of products with strived quality before the process starts. Metropol Kaktus is a product produced with high quality, the mean weight is 20,023 gram and this is very close to the weight, 20 gram, which is reported in the product specifications. The variation in weight variable is less than what is reported in the specifications and the... (More)
This paper aims to describe the variation and develop a method to control the production for the product Metropol Kaktus from the snuff developer Fiedler & Lundgren. We are going to use the requirements from the product specifications as limits of the expected quality when we present methods to control the process. One method attempts to control during the process and the other attempts to maximize the number of products with strived quality before the process starts. Metropol Kaktus is a product produced with high quality, the mean weight is 20,023 gram and this is very close to the weight, 20 gram, which is reported in the product specifications. The variation in weight variable is less than what is reported in the specifications and the minimum weight is over the limit set by Livsmedelsverket. The mean moisture is one point lower then the moisture reported in the specification. The variation in moisture decreases if the batch moisture increases. We want to take this variation in consideration, therefore we use a weighted regression model to estimate variance and standard deviation. A control diagram, aims to control the production during the process. This could be done with either the mean value or the absolute value of the difference between the extreme values, R. A mean diagram uses the mean value for controlling the process. An upper and a lower limit are calculated with the standard deviation. The weight variable gets a 20,278 gram upper limit and a 19,796 gram lower limit. The moisture variable gets a gets a 48 % upper limit and a 44 % lower limit. An R-diagram uses d2σ to control the process, where d2 is a constant whose value is determined by the size of the random sample. Unfortunately these limits can’t be calculated because of the random samples differs in size. A control plan aims to control the process by using a starting value that maximizes the number of products with acceptable quality. We use the weighted regression model to predict the moisture in a product ready for sale by setting the moisture of the batch to a specific percentage. The model shows that if we set the batch moisture to 46,5 %, 99,76 % of products ready for sale are going to be within the right moisture percentage. In a control plan for the variable weight we want the probability that a box contains a smaller amount of snuff to be small, this is called consumer risk, and the probability that a box contains too much snuff should also be small, this is called producer risk. We decide what size the random sample should be and the model show 19 observations and that we should discard the batch if the weight is less than 20,15 gram. (Less)
Please use this url to cite or link to this publication:
@misc{1337729,
  abstract     = {This paper aims to describe the variation and develop a method to control the production for the product Metropol Kaktus from the snuff developer Fiedler & Lundgren. We are going to use the requirements from the product specifications as limits of the expected quality when we present methods to control the process. One method attempts to control during the process and the other attempts to maximize the number of products with strived quality before the process starts. Metropol Kaktus is a product produced with high quality, the mean weight is 20,023 gram and this is very close to the weight, 20 gram, which is reported in the product specifications. The variation in weight variable is less than what is reported in the specifications and the minimum weight is over the limit set by Livsmedelsverket. The mean moisture is one point lower then the moisture reported in the specification. The variation in moisture decreases if the batch moisture increases. We want to take this variation in consideration, therefore we use a weighted regression model to estimate variance and standard deviation. A control diagram, aims to control the production during the process. This could be done with either the mean value or the absolute value of the difference between the extreme values, R. A mean diagram uses the mean value for controlling the process. An upper and a lower limit are calculated with the standard deviation. The weight variable gets a 20,278 gram upper limit and a 19,796 gram lower limit. The moisture variable gets a gets a 48 % upper limit and a 44 % lower limit. An R-diagram uses d2σ to control the process, where d2 is a constant whose value is determined by the size of the random sample. Unfortunately these limits can’t be calculated because of the random samples differs in size. A control plan aims to control the process by using a starting value that maximizes the number of products with acceptable quality. We use the weighted regression model to predict the moisture in a product ready for sale by setting the moisture of the batch to a specific percentage. The model shows that if we set the batch moisture to 46,5 %, 99,76 % of products ready for sale are going to be within the right moisture percentage. In a control plan for the variable weight we want the probability that a box contains a smaller amount of snuff to be small, this is called consumer risk, and the probability that a box contains too much snuff should also be small, this is called producer risk. We decide what size the random sample should be and the model show 19 observations and that we should discard the batch if the weight is less than 20,15 gram.},
  author       = {Timglas, Andreas},
  keyword      = {Kvalitetskontroll,snustillverkning,styrdiagram,medelvärdesdiagram,linjär regression,kontrollplan,Statistics, operations research, programming, actuarial mathematics,Statistik, operationsanalys, programmering, aktuariematematik},
  language     = {swe},
  note         = {Student Paper},
  title        = {En kvalitetskontroll - Snustillverkaren Fiedler & Lundgren kvalitetstestas},
  year         = {2008},
}