Enhancing System Safety and Reliability: The Dynamic and Dependent Tree Theory Framework for Accurate System Analysis

Since its inception in the 1960s, fault tree analysis has been extensively utilized to determine the root causes of system failures. However, traditional fault tree analysis has limitations that reduce its accuracy when evaluating contemporary engineering systems. The assumptions that each fundamental event occurs independently and that component defects follow constant failure and repair rates are two examples of these limitations.

The design, operation, and maintenance of complex engineering systems necessitate more sophisticated analysis techniques because of technological advances and changes in business practices. For example, complex asset management strategies employed in modern systems cannot be adequately represented in conventional fault tree models. Interdependencies between components, such as in standby systems, common cause failures, and opportunistic maintenance, also introduce inaccuracies. In addition, the sequence of failures and the increasing failure rates of components in systems operating beyond their expected design life are not adequately addressed.

In a new study published in the peer-reviewed Journal Reliability Engineering and System Safety, Professor John Andrews and Dr. Silvia Tolo from the University of Nottingham in England, presented a novel fault tree analysis framework referred to as Dynamic and Dependent Tree Theory (D²T²). The new framework incorporates Binary Decision Diagrams, Stochastic Petri Nets, and Markov methods to address the shortcomings of conventional fault tree analysis and facilitate more precise prognostication of system failures.

For those systems which experience dynamic and dependent features, alternative methods such as Petri nets, Monte Carlo Simulation, and Markov models were investigated to depict the whole system performance. While these methods provided a more realistic analysis, they presented obstacles such as computational resource requirements and the need for specialized knowledge. The concept of Dynamic Fault Trees had been previously introduced to overcome these problems, incorporating specialized gate types to model different dependency types. However, it was acknowledged that this approach could become inefficient as dependencies spread across the fault tree structure.

The authors introduced the dynamic and dependent tree theory framework as a solution to the limitations of traditional fault tree analysis. They presented several objectives derived from discussions with industrial stakeholders in the nuclear, aerospace and railway sectors. The first objective was to enable the representation of component failure and repair durations using any probability distribution, allowing for more realistic component behavior modeling. The framework also incorporated diverse forms of interdependencies between components or subsystems, taking system structure, operation, and maintenance activities into account. Event sequences were incorporated into the fault tree structure to effectively represent complex maintenance processes.

The framework for the dynamic and dependent tree theory integrates multiple modeling techniques to ensure transparency and facilitate regulatory review.  In retaining the fault tree structure to represent the system failure causality, it gives a framework which is familiar to the engineering community and retains the relevance of historical models evolved over time.  The authors employed Petri net or Markov models to efficiently manage complex aspects, thereby minimizing the computational resources required for a comprehensive system analysis. Binary Decision Diagrams were utilized by the dynamic and dependent tree theory framework to quantify the fault tree structure and provide an effective means of integrating results from the other modelling techniques.  These models accounted for both transient and steady-state conditions, thus incorporating time-dependent performance characteristics.

Modularization is an essential element in the methodology as it disassembled the system into independent sub-models for efficient analysis. In addition, the integration of dependency models into the structure of the fault tree ensured that system behavior was appropriately considered. To demonstrate the efficiency of the dynamic and dependent tree theory framework, the research team conducted a case study using a simple pressure vessel cooling system as an illustration. The case study showed, step by step, how the framework surmounted the limitations of traditional fault tree analysis and provided a more comprehensive assessment of system failures. By applying the framework, the authors were able to calculate the probability and frequency of system failure.  It is then possible to extend these results to evaluate component importance measures, for a system featuring varying component failure and repair rates, and opportunistic and standby system dependencies.

In a nutshell, Professor Andrews and Dr. Tolo introduced a novel framework for fault tree analysis called Dynamic and Dependent Tree Theory, which addresses the limitations of traditional fault tree analysis by incorporating various modeling techniques to accurately analyze system failures, with features, such as, interdependencies between components, varying failure rates, and complex maintenance strategies. The research on the dynamic and dependent tree theory framework brings advancements to fault tree analysis, offering engineers a more powerful and comprehensive toolset to analyze and manage complex engineering systems. Its importance lies in its ability to provide a more accurate understanding of system failures, facilitate proactive maintenance strategies, and enhance overall system reliability and safety.