# SPC orchestrates better process control.

SPC Orchestrates Better Process Control

It is impossible to adequately inspect 100% of any production, whether the product is aspirins....or zinc electrodes. To be both economically feasible and effective, inspection must rely on statistical sampling. Perhaps no industry, more than automotive, has jumped onto the bandwagon of statistical process control or SPC. In doing so, the number of automobiles actually inspected may be far fewer than in the past, but the quality of today's product is far higher. Many chemical producers have now implemented SPC to control their production.

Traditional testing methods ask the question, "Is the product in spec?" That is not good enough. We must find trends and anticipate changes before they ruin a production batch. The clues are there; we just need the tools to look for them. SPC Orchestra provides many of these tools. It is written as a series of spreadsheet templates and comes with an excellent manual that demonstrates both its operation and the fundamentals of SPC. While Lotus 1-2-3 version 2.01 was used for the review, other spreadsheets may also function, but anyone considering a purchase should check first. There is an extensive macro-driven menu system and even trivial differences may allow one so-called compatible spreadsheet to function and not another.

Data is entered into a spreadsheet. Two columns are needed. The first supplies the X value and the other is the data to be analyzed or Y value. One space is needed between the title (which is used as a label) and the data and one at the bottom to indicate the end of the data. Any missing data will be interpreted as the end of the data set. Several sets of data can run together with Runbatch.

Macros transfer the data to SPC.WK1 where the statistical calculations are performed. Printouts include a table of capability statistics and several graphs, including histograms and control charts for variables, attributes and CuSums. Individual points can be plotted or joined as either discrete subgroups or moving averages. You can go back and regenerate the graphs with other choices to see whether or not hidden trends can be made more visible, eg. by switching from a 3-point subgroup to a 10-point moving average. A useful feature is the ability to plot two control ranges on the same graph. A single graph might dramatically demonstrate the effects of some improvement to the system.

The cumulative sum or CuSum charts do not include a V-mask. Nevertheless, they make it easy to emphasize small differences. The pH example clearly indicated some form of cycling of the data.

When a system is in statistical control, all data points (or precisely 99.73% of them) must fall within the bounds of [+ or -] 3 [sigma] about the mean ([Mu]). The Lower and Upper Control Limits are defined on that basis:

LCL = [Mu] - 3 [Sigma] and UCL = [Mu] + 3 [Sigma]

Any point outside this range can not result from statistical fluctuations, but must result from some assignable cause. Once that cause has been diagnosed, provisions are provided for that point to be noted and deleted as a single way-out-of-spec point can contribute much more than its share to the statistical calculations. If there are several points beyond the limits and/or causes are unknown, the process is not in statistical control and requires better management.

SPC Orchestra is a Lotus spreadsheet and can use all the features of Lotus. Anyone who works with spreadsheets should find the operation both simple and familiar. The macro menu handles most graph scaling, labeling and saving options except for fixing the number of decimal points. I'm not sure why this last feature was not included, but many programs tend to overlook it. The number of decimal places can be set by leaving the macro system and going directly to the Lotus graph-options menu. The saved graphs can be plotted with Printgraph.

There are a few negatives. The X-axis scale varies from one option to another. This is only a minor nuisance when comparing plots. Control limits, specifications and means are plotted as data labels using lines made from "=", "[is less than]" & "[is greater than]" and "-" respectively. With displays other than CGA, the lines appear too short on the screen, but tend to be reasonable on the final printout. Some fine tuning of their length may be necessary as the various fonts for Printgraph and Freelance use characters of different widths and some use proportional spacing. The figures were all plotted with Printgraph using the standard Block 1 character set.

This system is not designed for realtime analysis; it does not automatically update each time you add a point. I've written programs that do, but then they don't have anywhere near the number of options available with SPC Orchestra. `Playing' with the various options provides the opportunity to make the analysis meet your needs. SPC rarely requires split-second decisions.

Who can use it? Anyone involved with QC in any way, shape or form who has data from monitoring some type of process. That could include pharmaceutical manufacturers, analytical laboratories, plant engineers and a whole spectrum of people who wish to improve their process. As it can do many things with the same data, it is also an excellent tool for training people who will use SPC. By designing SPC Orchestra as a spreadsheet, a simple approach is made available to anyone now using a computer system and familiar with spreadsheets. This also keeps the cost down and ensures the user that interfacing with a wide variety of video adapters and printers is available and updated.

It is impossible to adequately inspect 100% of any production, whether the product is aspirins....or zinc electrodes. To be both economically feasible and effective, inspection must rely on statistical sampling. Perhaps no industry, more than automotive, has jumped onto the bandwagon of statistical process control or SPC. In doing so, the number of automobiles actually inspected may be far fewer than in the past, but the quality of today's product is far higher. Many chemical producers have now implemented SPC to control their production.

Traditional testing methods ask the question, "Is the product in spec?" That is not good enough. We must find trends and anticipate changes before they ruin a production batch. The clues are there; we just need the tools to look for them. SPC Orchestra provides many of these tools. It is written as a series of spreadsheet templates and comes with an excellent manual that demonstrates both its operation and the fundamentals of SPC. While Lotus 1-2-3 version 2.01 was used for the review, other spreadsheets may also function, but anyone considering a purchase should check first. There is an extensive macro-driven menu system and even trivial differences may allow one so-called compatible spreadsheet to function and not another.

Data is entered into a spreadsheet. Two columns are needed. The first supplies the X value and the other is the data to be analyzed or Y value. One space is needed between the title (which is used as a label) and the data and one at the bottom to indicate the end of the data. Any missing data will be interpreted as the end of the data set. Several sets of data can run together with Runbatch.

Macros transfer the data to SPC.WK1 where the statistical calculations are performed. Printouts include a table of capability statistics and several graphs, including histograms and control charts for variables, attributes and CuSums. Individual points can be plotted or joined as either discrete subgroups or moving averages. You can go back and regenerate the graphs with other choices to see whether or not hidden trends can be made more visible, eg. by switching from a 3-point subgroup to a 10-point moving average. A useful feature is the ability to plot two control ranges on the same graph. A single graph might dramatically demonstrate the effects of some improvement to the system.

The cumulative sum or CuSum charts do not include a V-mask. Nevertheless, they make it easy to emphasize small differences. The pH example clearly indicated some form of cycling of the data.

When a system is in statistical control, all data points (or precisely 99.73% of them) must fall within the bounds of [+ or -] 3 [sigma] about the mean ([Mu]). The Lower and Upper Control Limits are defined on that basis:

LCL = [Mu] - 3 [Sigma] and UCL = [Mu] + 3 [Sigma]

Any point outside this range can not result from statistical fluctuations, but must result from some assignable cause. Once that cause has been diagnosed, provisions are provided for that point to be noted and deleted as a single way-out-of-spec point can contribute much more than its share to the statistical calculations. If there are several points beyond the limits and/or causes are unknown, the process is not in statistical control and requires better management.

SPC Orchestra is a Lotus spreadsheet and can use all the features of Lotus. Anyone who works with spreadsheets should find the operation both simple and familiar. The macro menu handles most graph scaling, labeling and saving options except for fixing the number of decimal points. I'm not sure why this last feature was not included, but many programs tend to overlook it. The number of decimal places can be set by leaving the macro system and going directly to the Lotus graph-options menu. The saved graphs can be plotted with Printgraph.

There are a few negatives. The X-axis scale varies from one option to another. This is only a minor nuisance when comparing plots. Control limits, specifications and means are plotted as data labels using lines made from "=", "[is less than]" & "[is greater than]" and "-" respectively. With displays other than CGA, the lines appear too short on the screen, but tend to be reasonable on the final printout. Some fine tuning of their length may be necessary as the various fonts for Printgraph and Freelance use characters of different widths and some use proportional spacing. The figures were all plotted with Printgraph using the standard Block 1 character set.

This system is not designed for realtime analysis; it does not automatically update each time you add a point. I've written programs that do, but then they don't have anywhere near the number of options available with SPC Orchestra. `Playing' with the various options provides the opportunity to make the analysis meet your needs. SPC rarely requires split-second decisions.

Who can use it? Anyone involved with QC in any way, shape or form who has data from monitoring some type of process. That could include pharmaceutical manufacturers, analytical laboratories, plant engineers and a whole spectrum of people who wish to improve their process. As it can do many things with the same data, it is also an excellent tool for training people who will use SPC. By designing SPC Orchestra as a spreadsheet, a simple approach is made available to anyone now using a computer system and familiar with spreadsheets. This also keeps the cost down and ensures the user that interfacing with a wide variety of video adapters and printers is available and updated.

Printer friendly Cite/link Email Feedback | |

Title Annotation: | statistical process control spreadsheet software |
---|---|

Author: | Silbert, Marvin D. |

Publication: | Canadian Chemical News |

Date: | Sep 1, 1989 |

Words: | 930 |

Previous Article: | Low-priced package for simple flowsheets. |

Next Article: | A school needs you. |

Topics: |