About
As information of various incidents become available over time, it becomes pertinent to utilize this data to understand the underlying causes or contributors behind the incidents, identify their trend and analyze means of taking proactive measures to prevent such incidents in the future. In order to learn from incidents, good investigation is required although this is not done in most cases. Available investigation reports can still shed light on the various frequently occurring issues that lead to incidents. Depending on the depth of the studied investigation reports, the analysis can identify technical, operational, human and organizational contributors to incidents and enable understanding of proactive measures that need to be taken on time. Properly maintained database can help identify incident trends and provide good predictive models for future incidents. At the MKOPSC, various approaches have been undertaken to provide an analysis of contributing factors behind various kinds of incidents and develop time trend analysis of incidents. In that context, it is also important to have properly developed incident reporting systems to learn from them.
At MKOPSC, incident databases and investigation reports prepared by Bureau of Safety and Environmental Enforcement (BSEE) and Pipeline and Hazardous Materials Safety Administration (PHMSA) has been utilized to identify contributors behind incidents, and use then various data to develop causal and predictive models. See Offshore Safety, Refinery Safety and Pipeline Safety section for more information.
State of the art technologies using machine learning algorithm and artificial intelligence are also being used to handle the large information. See Big Data section for more information.
In addition, database on process safety incidents from publicly available information are also being developed.
In order to gain an overview of relevant risks across all industries, government facilities, fixed facilities, transportation modes, toxic, reactive, flammable and otherwise dangerous chemicals, an integrated database is required (Mannan, 2018 (https://www.sciencedirect.com/science/article/abs/pii/S0950423018304169). Realization of this is, however, outside MKO’s capability.
Chemical Incident Data Systems
A national repository database for chemical accidents
A national repository of information on chemical accidents has been identified as the top priority of National Chemical Safety Program Roundtable (MKOPSC Report). The deficiencies of current data systems have been identified in a number of studies. Although data is increasingly available on-line much is not and the underlying consistency and quality of the data is still lacking. Currently there are many federal, state and local agencies collecting and reporting chemical incidents. However, each agency can only collect information on incidents within their legislative authority. Rulemaking further limits the scope in many cases. Also, because of the varying focus amongst agencies the data collected and the terminology employed vary widely. The agencies are generally limited to certain chemicals, above threshold quantities, in particular facilities or transportation modes. These limitations stymie efforts to gain an overview of all chemical incidents.
The development of an integrated system must address this problem in depth. Additionally, it is also desirable to eliminate redundant reporting to multiple agencies. Currently, there is presently no reliable means for evaluating the performance of industry in limiting the number and severity of accidental chemical releases. In order to gain an overview of relevant risks across all industries, government facilities, fixed facilities, transportation modes, toxic, reactive, flammable and otherwise dangerous chemicals an integrated database is required (Mannan, 1998). The database could either be a series of databases or one all-inclusive one. Either way, this would allow an analysis starting with the undesirable consequences of chemical releases, whether injury, ecological or monetary back through the technical, management and social causes of the releases.
Integration of Existing Sources into the national incident repository database
While it is easy to be critical of the existing systems it should be recognized that substantial resources are involved in data collection, vetting, investigation and reporting. These incident investigations and reporting efforts are often integral to the agency’s function. This content cannot be captured by a new reporting system, it has to be integrated into the system. Because of the close coupling of the data gathering and investigation to the agency’s function it appears impractical to replace the existing systems with one new system. Integrating the systems should allow for effective use of these existing resources.
Technical advancement is also making it easier to integrate data from diverse resources. This data can then be readily transferred by electronic means to the integrated system. This proposed plan seeks to capitalize on the existing data and systems while bringing them together in an integrated system, making them more complete, consistent, and accessible. The plan will also address the flow, tracking, filtering, and vetting of information from initial reporting, through the relevant agencies and into the repository. It will provide a check to see if all significant incidents are being reported to the NRC. It can also provide additional information that is not otherwise being collected, especially incidents that do not fall within the jurisdiction of any particular agency. It may also provide some insight into public and media perception of chemical incidents.
Incident Reporting System
Initial federal reporting of incidents is primarily through the NRC. Approximately 25,000 cases per year are reported in this manner. The NRC maintains these initial reports in a database called the IRIS Incident Reporting Information System. The NRC selectively forwards this initial information to approximately 16 other federal agencies and to state agencies. Some agencies such as ATSDR employ a number of additional sources such as local fire departments, industry, medical providers and news media. The various agencies process portions of this information, conduct investigations, and make reports within the scope of the agency.
A consistent reporting method should also allow tracking of trends for various industries, processes, chemical types, and consequences. This knowledge would allow prioritization of efforts by government, industry, researchers and others to promote progress in chemical process safety. Such an effort would dovetail with other activities such as investigations by the CSB which focus on an in depth analysis of the most severe accidents. A full understanding of chemical accidents is probably best served by having an overview of all accidents as well as in-depth understanding of severe accidents. The Center would develop a taxonomy into which the existing data could be mapped. The resulting repository would then be made available on-line in a fully searchable system and for download or distribution on CD ROM in standard formats. The publicly available system would have identifying information redacted to be consistent with the requirements for HSEES data.
