Statistical control charts, safety

Miner s safety performance recognition, circa 1920s 20 The systems model of health and safety management 24 The size of mobile mine equipment continues to increase 27 Simple model of outcomes dependent on culture and systems 32 Statistical control chart with incident rate variation 33 Universal Copper and Metals Mine accident frequency rate control chart 55... 

In industrial plants, large numbers of process variables must be maintained within specified limits in order for the plant to operate properly. Excursions of key variables beyond these limits can have significant consequences for plant safety, the environment, product quality and plant profitability. Statistical process control (SPC), also called statistical quality control (SQC), involves the application of statistical techniques to determine whether a process is operating normally or abnormally. Thus, SPC is a process monitoring technique that relies on quality control charts to monitor measured variables, especially product quality. 

Statistical process controls (cause-and-effect diagrams, control charts, etc.), as applied in quality management, can serve as performance measures for safety, if they are used prudently and with caution. 

It replaces the American National Standard for Uniform Record Keeping for Occupational Injuries and Illnesses, ANSI Z16.4-1977, and USA Standard Method of Recordkeeping and Measuring Work Injury Experience, ANSI Z16.1-1967. This standard is a development of the old Z16.1, which had been in use since 1937, before it was replaced by the (for injury and iUness statistical research) Occupational Safety and Health Administration (OSHA) system. The standard is useful in determining what kinds of events to evaluate. It includes statistical tools, including control charts, for data analysis. 

A control chart is a statistical device used for the study and control of safety performance in the workplace. The basis of control chart analysis is the knowledge of chance variations in the data (Duncan 1974,375). If a series of safety measurements are plotted and the obtained measurements are truly random, the distribution would approximate the normal bell-shaped curve. Plotting the data on a control chart, one would obtain measurements over time that fall into the ranges depicted in Figure 4.1, with more measures occurring at or near the average more frequently and readings at the extreme ends of the possible measurements infrequently. 

Because of the underlying statistical processes used with control charts, the safety manager can identify safety performance problems when they become statistically significant. The safety manager can also use control charts to implement continual improvement programs related to safety activities. 

Valid statistical measures, such as control charts, are convincing. In some situations, such as initiating the cultural change necessary to call attention to serious injury prevention, the frequency of occurrence data on such incidents that would be placed on a control chart will not be available since the subject is low-probability/severe-consequence events that do not occur often. Cost data for such events can be influential. For an additional reference, see Measurement of Safety Performance in On The Practice Of Safety. 

Petersen (2005) included an 18-page appendix on Statistical Process Control in Measurement of Safety Performance. Janicak (2010) has a 30-page chapter on Run Charts and Control Charts in Safety Metrics Tools and Techniques for Measuring Safety Performance. Other authors in years past have suggested the use of control charts to track safety performance. Yet, the subject does not appear in the current safety-related literature. 

In most of the plants we have studied, the safety process is stable. Although accident rates vary from month to month, there is usually no statistical trend. Trends, or their absence, must be determined through the use of control charts, which we will describe further in Chapter 4. 

The U chart is a critical tool for avoiding the overmanagement of a system that is in statistical control. It has been our experience that the U chart is frequently successful in helping traditional managers understand that accidents are produced by the process not by incompetent or unmotivated employees. These managers can then begin to look at safety improvement with a longer-term view and a systems perspective. 

In this appendix, basic probability and statistics concepts are reviewed that are considered for the safety analysis of Chapter 10 and the quality control charts of Chapter 21. 

However, risk in the process world has an even more fundamental role that is far more just to fulfill the Agency s expectation about process understanding and product safety it is the basis and rational for MSPC. In fact, it is what the original intent of Walter Andrew Shewhart (1891-1967) had in mind when he invented the notion of statistical process control (SPC) and the control Shewhart chart. Although not couched in precisely the language of risk, it was at the heart of what he was trying to do at Bell labs at the time [11]. 

---