Conference Paper

Multilevel flow modeling for nuclear power plant diagnostics

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

As complexity and safety requirements of current and future nuclear power plants increase, innovative methods are being investigated to perform accurate and reliable system diagnoses. Detecting malfunctions, identifying their causes and possibly predicting their consequences are major challenges, especially if extended to the whole plant. Monitoring plant performances by means of data reconciliation techniques has proved successful to detect anomalies during operation, provide early warnings and eventually schedule maintenance. At the same time, the large amount of information provided by large-scale monitoring systems is hard to handle manually. In this paper, a function-oriented modeling approach, called Multilevel Flow Modeling, is proposed for performing an automatic analysis of the outcomes of the monitoring systems with the aim of identifying the root causes of the possibly detected anomalies. The novel combination of a data reconciliation system and the Multilevel Flow Modeling approach is illustrated with regard to the secondary loop of the Loviisa-2 Pressurized Water Reactor located in Finland.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... [15]. MFM has been used for a range of industrial processes including power and chemical engineering plants [16] [17]. Recent applications in intelligent control of and design of control architectures for power systems are reported in [18] [9]. ...
... The model is a powerful tool for analysing fault situations and for controlling the water mill. A MFM model of the nuclear power plant Loviisa is presented in [16]. ...
Technical Report
This report demonstrates how a generic agent-oriented framework has been enhanced for supporting qualitative and function-oriented systems modelling and analysis, so that it can be utilized for Supervision, Diagnosis and Prognosis (SDP) applications in advanced automation environments. The Multilevel Flow Modelling method (MFM) has been particularly chosen as subject for support in the framework, due to its proven strength in qualitative planning, modelling and diagnosis activities in process control applications, particularly involving mass-, energy-, and control-related functions. The report describes the generic framework ShapeShifter’s background, main elements and main application areas, the latter by means of its different supported tools. A description of the recent enhancements to the framework, in terms of MFM-supporting features for fault-tolerant design and early fault detection, as a part of the framework’s support for SDP-related activities, are also included. Furthermore, the description of a tool for qualitative modelling and analysis, called “the MFM Editor”, has been provided. In that respect, it has been shown how the tool uses ShapeShifter’s generic features for MFM-modelling and how it communicates with a qualitative reasoning tool, called “the MFM Reasoning system”, for analysis purposes related to fault-tolerant design as well as early fault detection in SDP applications. As far as the plan for further work is concerned, one central issue is to equip the framework and its supporting tools, including the MFM Editor, with as many relevant features as possible, so that they together can provide more support for various challenges in modelling and analysis of particularly complex automation environments, with not only their automation- (technology-) oriented properties but also their human factors involved in how they are operated, maintained, analysed and altered. This includes the description and visualisation of the one-to-many and many-to-one relationships between different physical process components and their MFM function representations by including a Process Modeller component in the MFM Editor.
... MFM has been used to represent a variety of complex dynamic processes including energy conversion systems like fossil power plants [17], nuclear power generation [9,12,[34][35][36][37], gas turbines [48] and ship engines [15]. MFM has also been used to model power transmission and distribution systems [13,14,42] and for chemical engineering systems such as oil refineries [8], distillation columns [41] and biochemical processes [15]. ...
... In this paper, the function-oriented modelling technique, Multilevel Flow Modelling (MFM) [1] is used within an innovative diagnostic scheme for automating residual analysis in nuclear power plants. MFMbased approaches have been successfully applied for early/hidden fault detection [2,3], diagnostics [4][5][6][7][8], on-line alarm analysis and filtering [9,10], to analyse faulty scenarios in HAZOP studies [11], modelling control purposes in power systems [12] and fault tree generation and risk analysis [13]. ...
Technical Report
As complexity and safety requirements of current and future nuclear power plants increase, innovative methods are being investigated to perform accurate and reliable system diagnoses. Detecting malfunctions, identifying their causes and possibly predicting their consequences are major challenges, especially if extended to the whole plant. Monitoring plant performances by means of data reconciliation techniques has proved successful to detect anomalies during operation, provide early warnings and eventually schedule maintenance. At the same time, the large amount of information provided by large-scale monitoring systems is hard to handle manually. In this HWR, a function-oriented modelling approach, called Multilevel Flow Modelling (MFM), is proposed for performing an automatic analysis of the outcomes of the monitoring systems with the aim of identifying the root causes of the possibly detected anomalies. Reasoning in MFM models derives its power from representation of process knowledge on several levels of specification. The detailed specification is the basis for implementation of automated model-based reasoning functions, whereas the more abstract specifications provides generic process knowledge for formulation of reasoning strategies and for giving explanations. Reasoning strategies and explanations can be directly visualized in terms of the means-end topology of the multilevel flow models and may be used for design of human machine interfaces supporting diagrammatic reasoning about spatial-temporal aspects of dynamic situations. The principles described in the HWR have been used in the implementation of a model based reasoning system. The purposes of the HWR are 1) to describe how MFM is used for reasoning about causes and consequences in complex dynamic processes and 2) to illustrate how MFM can be combined with a plant monitoring system, in this case the TEMPO, for automating the residual analysis. In this respect, a case study is illustrated with regards to the secondary loop of the Finnish Loviisa-2 VVER.
... An MFM model representing the functions of the water steam cycle in a nuclear power plant in more detail is presented by Gola et.al. [8] . ...
... MFM has been used to represent a variety of complex dynamic processes including energy conversion systems like fossil power plants [17], nuclear power generation [9, 12,34353637, gas turbines [48] and ship engines [15]. MFM has also been used to model power transmission and distribution systems [13, 14, 42] and for chemical engineering systems such as oil refineries [8], distillation columns [41] and biochemical processes [15]. ...
Chapter
Full-text available
A modeling environment and methodology are necessary to ensure quality and reusability of models in any domain. For MFM in particular, as a tool for modeling complex systems, awareness has been increasing for this need. Introducing the context of modeling support functions, this paper provides a review of MFM applications, and contextualizes the model development with respect to process design and operation knowledge. Developing a perspective for an environment for MFM-oriented model-and application-development a tool-chain is outlined and relevant software functions are discussed. With a perspective on MFM-modeling for existing processes and automation design, modeling stages and corresponding formal model properties are identified. Finally, practically feasible support functions and model-checks to support the model-development are suggested.
Conference Paper
In control design, fault-identification and fault tolerant control, the controlled process is usually perceived as a dynamical process, captured in a mathematical model. The design of a control system for a complex process, however, begins typically long before these mathematical models become relevant and available. To consider the role of control functions in process design, a good qualitative understanding of the process as well as of control functions is required. As the purpose of a control function is closely tied to the process functions, its failure has a direct effects on the process behaviour and its function. This paper presents a formal methodology for the qualitative representation of control functions in relation to their process context. Different types of relevant process and control abstractions are introduced and their application to formal analysis of control failure modes from a process perspective is presented. Finally anticipated applications in context of offline analysis and online supervisory control are discussed.
Conference Paper
Full-text available
Many new technologies with novel control capabilities have been developed in the context of ldquosmart gridrdquo research. However, often it is not clear how these capabilities should best be integrated in the overall system operation. New operation paradigms change the traditional control architecture of power systems and it is necessary to identify requirements and functions. How does new control architecture fit with the old architecture? How can power system functions be specified independent of technology? What is the purpose of control in power systems? In this paper, a method suitable for semantically consistent modeling of control architecture is presented. The method, called multilevel flow modeling (MFM), is applied to the case of system balancing. It was found that MFM is capable of capturing implicit control knowledge, which is otherwise difficult to formalize. The method has possible future applications in agent-based intelligent grids.
Chapter
Automatic, computerized diagnosis can be based on several different search strategies, e.g. a search for a match between a pattern of measured data and some stored symptom patterns, or a search to locate a change in the system’s functional state with reference to a stored model of normal or specified plant state. The latter strategy has a number of basic advantages: it is independent of the prediction and analysis of specific faults and events; the reference for search, the normal state, can be derived from actual plant operation by the computer; the strategy can be based on invariate relations such as conservation laws; etc.
Article
This paper describes three diagnostic methods for use with industrial processes. They are measurement validation, alarm analysis and fault diagnosis. Measurement validation means consistency checking of sensor and measurement values using any redundancy of instrumentation. Alarm analysis is the analysis of multiple alarm situations to find which alarms are directly connected to primary faults and which alarms are consequential effects of the primary ones. Finally, fault diagnosis is a search for the causes of and remedies for faults. The three methods use multilevel flow models (MFM), to describe the target process. They have been implemented in the programming tool G2, and successfully tested on simulations of two processes.
Article
There seems to be a great potential for using Multilevel Flow Modeling as a framework for reasoning in supervisory control of complex systems. It is a precondition, however, that knowledge about the causal relations between flow functions is represented. Previous attempts have used generic causation rules, specifying possible influences between specific types of flow functions. The problem with this approach is that the generic nature of such rules sometimes leads to invalid reasoning results. This paper presents a method for representing more precisely the actual causal structure of the system being modeled, directly in MFM. The method is based on a set of generic relations, which can be used to make explicit the causal relations hidden in the MFM connection relation. Implications of the method are illustrated by means of simple examples. 1. Introduction In the field of Cognitive Systems Engineering a great deal of research has been concerned with the development of operator support...
Model-based reasoning using MFM
  • M Fang
  • M Lind
Fang, M. & Lind, M. 1995. Model-based reasoning using MFM, Proc. PACES, May 16-18, Huangshan, China.
A Functional HAZOP methodology. Computers in Chemical Engineering
  • N L Rossing
  • M Lind
  • N Jensen
  • S B Jørgensen
Rossing, n.L. & Lind, M. and Jensen, n. & Jørgensen, S.B. 2010. A Functional HAZOP methodology. Computers in Chemical Engineering, 34(2), pp.244-253.