Incidents data analysis

The results of the CSB incident data analysis are acknowledged as representing only a sampling of recent reactive incident data. This limitation precludes CSB from drawing statistical conclusions on incidence rates or inferring trends in the number or severity of incidents. However, despite these limitations, the data can be used to illustrate the profile and causes of reactive incidents. 

SIMS, and SNMS in rare cases, such as for HgCdJTei samples or some polymers, the sample structure can be modified by the incident ion beam. These effects can often be eliminated or minimized by limitii the total number of particles incident on the sample, increasing the analytical area, or by cooling the sample. Also, if channeling of the ion beam occurs in a crystal sample, this must be included in the data analysis or serious inaccuracies can result. To avoid unwanted channelii, samples are often manipulated during the analysis to present an average or random crystal orientation. 

The personnel responsible for the collection and analysis of incident data vary in different organizations. One common practice is to assign the responsibility to an investigation team which includes the first line supervisor, a safety specialist and a plant worker or staff representative. Depending on the severity of an incident, other management or corporate level investigation teams may become involved. 

This report is by Battelle Columbus Division to the Line Pipe Research Supervisory Committee of the American Gas Association. It presents an analysis of statistical data obtained from reports of lea)c or rupture (service) incidents and test failures in natural gas transmission and gathering lines over the 14.5 year period from 1970 through June, 1984. All gas transmission companies were required to notify the Office of Pipeline Safety Operations in the event of a "reportable" incident, as defined by the Code of Federal Regulations. The purpose of the study is to organize the reportable incident data into a meaningful format from which the safety record of the industry can be assessed. 

To determine maximum individual risk, generic frequency data are required for explosion events for Process Units 1 and 2. For Process Unit 1, incident data were available from the unit licenser identifying three explosions in approximately 15,000 operating years, for an explosion frequency of 2.0 x 10-4 per year. For Process Unit 2, a fault tree analysis of the nitrogen vessel brittle fracture event had been conducted as part of an unrelated project. That study concluded that the frequency of brittle fracture failure of the nitrogen vapor storage vessel was 5x10"4 per year. 

Data analysis is based on the proper tracking of the key environmental parameters of the experiment. Apart from the obvious parameters (e.g., the elongation in a straining experiment) there are some essential parameters that are related to the scattering apparatus itself (e.g., the intensity of the incident beam). 

The ellipsometer used in this study is described elsewhere(3). It consists of a Xenon light source, a monochromator, a polarizer, a sample holder, a rotating analyzer and a photomultiplier detector (Figure 1). An electrochemical cell with two windows is mounted at the center. The windows, being 120° apart, provide a 60° angle of incidence for the ellipsometer. A copper substrate and a platinum electrode function as anode and cathode respectively. Both are connected to a DC power supply. The system is automated with a personal computer to collect all experimental data during the deposition. Data analysis is carried out by a Fortran program run on a personal computer. 

In 1985, the United States Federal Register recommended that the analysis of tumor incidence data be carried out with a Cochran-Armitage (Armitage, 1955 Cochran, 1954) trend test. The test statistic of the Cochran-Armitage test is defined as this term ... 

The data analysis included evaluating the number, impact, profile, and causes of reactive incidents. CSB examined more than 40 data sources (e.g., industry and governmental databases and guidance documents safety/loss prevention texts and journals and industry association, professional society, insurance, and academic newsletters), focusing on incidents where the primary cause was related to chemical reactivity. 

Although the statistics provided in Section 3.3 concerning the number and severity of reactive incidents are grave, existing sources of incident data are inadequate to identify the number, severity, frequency, and causes of reactive incidents. The following limitations affected CSB analysis of incident data ... 

The CSB data analysis shows that reactive incidents are not limited to any one chemical or to a few classes of chemicals. Table 3 lists common chemical classes involved in the 167 incidents. None of these classes represent a majority of incidents in the CSB data... 

A common perception is that reactive incidents are primarily the result of runaway reactions. In fact, analysis of data from the 167 incidents suggests that other types of reactive hazards should also be of concern. CSB data analysis identified three common types of reactive hazards (see Appendix A for definitions) ... 

Figure 15. NFPA instability rating analysis (formerly reactivity rating) of incident data, 1980-2001.

As supported by the CSB incident data, two elements are particularly relevant to reactive hazards-Process Safety Information (PSI 29 CFR 1910.119 [d]) and Process Hazard Analysis (PHA 29 CFR 1910.119... 

To be effective the investigation must apply an approach which is based on basic incident causation theories and use tested data analysis techniques. Investigating incidents to determine root causes and make recommendations can be as much an art as a science. Within the industry, best practices in incident investigation have evolved substantially in the last 20 years. This chapter provides a brief overview of some of the more relevant causation theories. 

Some items and components may require examination that is more detailed. Specific techniques for data analysis are beyond the scope of this guidebook. Entire volumes have been written on specific issues such as the fracture patterns of alloys and the corresponding clues for determining the actual cause and mechanism for the failure. Known specific materials and alloys perform and fail in consistent and predictable ways. This area of expertise is normally supplied to the incident investigation team via the use of specialists, from either within the parent organization or from outside experts or labs engaged specifically for the task. 

In addition to the physical data analysis methods, traditional engineering analysis tools and methods are also useful during incident investigations. Traditional analysis tools can he used to determine the following. 

A total of 1118 drugs with ADR information were included in the analysis. Incidence data ( common, frequent, infrequent, rare, not known ) associated with each COSTART term for each compound were transformed to binary data. Each ADR for a specific drug was scored as 1 if the COSTART term was listed in its label at any clinical frequency including not known, and as 0 if the term was not listed. Thus, each occurrence of the ADR in the label information is counted as a positive response. 

Blair et al. (1998) performed a retrospective cohort mortality study of 14 457 workers employed for at least one year between 1952 and 1956 at an aircraft maintenance facility in the United States. Among this cohort were 6737 workers who had been exposed to carbon tetrachloride (Stewart et al., 1991). The methods used for this study are described in greater detail in the monograph on dichloromethane. An extensive exposure assessment was performed to classify exposure to trichloroethylene quantitatively and to classify exposure (ever/never) to other chemicals qualitatively (Stewart et al., 1991). Risks from chemicals other than trichloroethylene w ere examined in a Poisson regression analysis of cancer incidence data. Among women, exposure to carbon tetrachloride was associated with an increased risk of non-Hodgkin lymphoma (relative risk (RR), 3.3 95% CI,... 

---