Manual vs. Automated Statistical Process Control in the Food Industries
Israel Ortega-Ramos
The Prime Example
Our recent visit to a food packaging plant in New Jersey highlighted the inconsistent results of statistical process control routinely faced by Quality Control Managers. Product weight readings were taken from the manufacturing floor, entered into an Excel spreadsheet and analyzed. The results produced no predictable under or over filling trend despite the fact that the same people used the same scales at the same time of day. The problem is simple and fundamental. Human error is an inevitable part of the process of collecting statistical data. This is consistently overlooked in companies that utilize manual SPC[1] (statistical process control) for their manufactured goods. To ensure the human error factor is eliminated, resulting in lower costs and increased profitability, manufactures must begin utilizing more “high-tech” means of collecting, analyzing, and storing SPC data.
The Hidden Problems of the Current Manual SPC Process
To better understand the core problem, and find a solution, it is pivotal to understand how this food packaging plant utilizes manual SPC. Generally, several samples are taken from a product line at different times of day, usually 15 or 20 samples at a time. These samples are then individually weighed; a line worker records the results on a clipboard for analysis. The individual weight readings are entered into a computer and various statistical calculations are derived from the weighing results, including frequency distribution charts and Pareto charts that are used to adjust the actual filling machines to deliver a consistent result. The Quality Manager must then resolve any conflict between under filling a package, which breaks government laws and overfilling a package which causes lost revenue.
Using the diagram to get a better understanding of SPC will make it easier to locate the fundamental problems with the manual SPC system in use. The
The Prime Example
Our recent visit to a food packaging plant in New Jersey highlighted the inconsistent results of statistical process control routinely faced by Quality Control Managers. Product weight readings were taken from the manufacturing floor, entered into an Excel spreadsheet and analyzed. The results produced no predictable under or over filling trend despite the fact that the same people used the same scales at the same time of day. The problem is simple and fundamental. Human error is an inevitable part of the process of collecting statistical data. This is consistently overlooked in companies that utilize manual SPC[1] (statistical process control) for their manufactured goods. To ensure the human error factor is eliminated, resulting in lower costs and increased profitability, manufactures must begin utilizing more “high-tech” means of collecting, analyzing, and storing SPC data.
The Hidden Problems of the Current Manual SPC Process
To better understand the core problem, and find a solution, it is pivotal to understand how this food packaging plant utilizes manual SPC. Generally, several samples are taken from a product line at different times of day, usually 15 or 20 samples at a time. These samples are then individually weighed; a line worker records the results on a clipboard for analysis. The individual weight readings are entered into a computer and various statistical calculations are derived from the weighing results, including frequency distribution charts and Pareto charts that are used to adjust the actual filling machines to deliver a consistent result. The Quality Manager must then resolve any conflict between under filling a package, which breaks government laws and overfilling a package which causes lost revenue.
Using the diagram to get a better understanding of SPC will make it easier to locate the fundamental problems with the manual SPC system in use. The