The Continuous Hazard Tracking and Failure Prediction Methodology (CHTFPM) is a proactive methodology for gathering and analyzing information about a system in order to prevent accidents and system failures. The proactivity of the CHTFPM places it in contrast to conventional formal inductive and deductive hazard-analysis methodologies, which are limited in their effectiveness, in that they do not provide realtime information on whether the conditions in a system are becoming hazardous and could lead to a system malfunction: the conventional methodologies basically provide feedback on hazards after accidents have happened. The CHTFPM could be applied to advantage in almost all industries.

Information About a System is gathered by sampling in a prescribed manner, the information is analyzed to detect hazards, and corrective action is taken.

The CHTFPM involves the use of techniques from the arts of work sampling and control charting. With respect to a given system, random sampling is performed in order to detect and observe conditions, called “dendritics,” that could result in accidents or unacceptable risks. The collected data are then used to generate an abstract data control chart. On the basis of the pattern of the control chart, the system is said to be “under control” (in essence, safe and thus not to be disturbed) or else “out of control,” in which case it is investigated for the potential emergence of hazards from the observed conditions. The results of the investigation can provide guidance for steps to be taken to ameliorate the hazardous conditions to keep the system safe.

The steps in the evaluation and correction of a system according to the CHTFPM (see figure) are the following:

  1. Define the dendritics by use of the established techniques of preliminary hazard analysis (PHA), fault-tree analysis (FTA), system safety analysis (SAA), system hazard analysis (SHA), failure mode and effects analysis/ critical items list (FMEA/CIL), Pareto analysis, problem reporting and corrective action (PRACA), and/or safety checklists.
  2. Develop the random-sampling scheme.
  3. Construct the control chart from the samples.
  4. Develop a mathematical model.
  5. Test the control-chart observations for actual or potential “out of control” conditions.
  6. Take appropriate action to prevent or eliminate any “out of control” condition.

Step 1 is the fundamental to everything else in the CHTFPM: the effectiveness of the CHTFPM depends on the identification of the dendritics for sampling in a given system. Regarding the techniques used in step 1:

  • PHA can give the analyst both the necessary introduction of the system and an initial assessment of risk, identification of safety-critical areas, and evaluation of hazards.
  • The emphasis on conditions, instead of events, is the single most compelling feature of the FMEA for dendritic construction.
  • The CIL, derived from the FMEA, is useful for ensuring that the most critical items are factored into dendritic construction.
  • FTA can contribute to understanding of interactions among components in the system. The single most important tool for dendritic construction is the FMEA.

The sampling feature makes CHTFPM a cost-effective tool for dynamically analyzing and tracking a system for dendritics. The sampling can be performed in conjunction with other routine job functions. Sample data are used to design a control chart for use in ensuring that the system remains within an acceptable range (between the control limits); or, in other words, that it does not shift “out of control.” The more representative the sample is of the process in the system, the easier it is to detect shifts.

A major problem in the CHTFPM is to choose the size and frequency of samples. In general terms, the solution must be a compromise among considerations of the cost of sampling, the loss associated with running the process out of control, and the occurrence of various system shifts attributable to natural variability (essentially, background noise) or to assignable causes (e.g., machinery malfunctions or operator errors). A process that is operating with only natural variability is said to be in statistical control. A system operating in the presence of assignable causes is said to be “out of control.” The control chart provides an effective means of detecting assignable causes, identifying currently problematic areas, and predicting future problematic areas. By providing information regarding the tendency of the system, the control chart indicates when the system tends to become hazardous, facilitating the implementation of corrective steps.

This work was done by Rolando Quintana of the University of Texas for Kennedy Space Center. For further information, access the Technical Support Package (TSP) free on-line at www.nasatech.com/tsp  under the Information Sciences category. KSC-12066