No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
The development of the MFM Editor and its applicability for supervision, diagnosis and prognosis
Abstract
This paper summarizes the development to date of the Java-based MFM Editor, a graphical editor supporting the Multilevel Flow Modeling (MFM) method, to be used for Supervision, Diagnosis and Prognosis (SDP) applications in advanced automation environments. The editor builds on the ShapeShifter graphical framework, previously described in the proceedings of the ESREL 2010, 2011 and 2012 conferences. The paper focuses on the editing capabilities and the recently added reasoning functionality (provided by the Technical University of Denmark) required to perform cause and consequence analyses (diagnosis and prognosis) of MFM models. Also, a recently developed process design module is described in some detail.
... For several years such a dedicated software system, the MFM Suite [1] [2], has been under development at the Institute for Energy Technology. The system will allow a user to graphically design and verify the semantic correctness of MFM models, in addition to creating graphical models of industrial processes. ...
This paper presents the status of a software system, the MFM Suite, dedicated to the design and analysis of MFM models related to diagnostic and prognostic analysis of physical processes. New and updated features of the system are described, as well as some examples of its practical use. The paper also briefly describes how the system facilitates the collaboration between control room and field operators via the Android-based MFM Viewer app.
... For several years such a dedicated software system, the MFM Suite, has been under development at the Institute for Energy Technology (Thunem, 2013, Thunem andZhang, 2015). The system will allow a user to graphically design and verify the semantic correctness of MFM models, in addition to creating graphical models of the industrial process. ...
This paper presents the status of a software system, the Multilevel Flow Modeling (MFM) Suite, dedicated to the design and analysis of MFM models related to diagnostic and prognostic analysis of physical processes. New and updated features of the system are described, as well as some examples of its practical use. The paper also briefly describes how the system facilitates the collaboration between control room and field operators via the Android-based MFM Viewer app.
... itis built on the 2nd generation of the Shape Shifter framework (Thunem et al, 2011). The background for and initial development of the tool has been described in Thunem (2012Thunem ( , 2013. in addition, the paper describes how the MFM Suite is used to develop an MFM model of the primary side of a PWR. Preliminary results of performing MFM reasoning on a LOCA situation are also included. ...
This paper reports on the results from the practical application of the ShapeShifter framework on the continued development of a graphical editing suite, the MFM Suite, for MFM and process model design and analysis. The primary use of the MFM Suite is diagnosis and prognosis of anomalies in physical processes. One of the Halden Reactor Project’s advanced NPP simulators based on a PWR is used to demonstrate the applicability of the suite in realistic situations. The paper presents a summary and suggests some plans for future research and development.
... In this study, a graphical editor based on the Java-based ShapeShifter framework supporting MFM modeling (called the MFM Editor) shown in Fig.2.14 developed by the Institute for Energy Technology in Norway is used [110][111] . The editor will be used to design MFM model and to set the values of selected functions, and subsequently to visualize the possible function paths that could cause these function values(cause paths) , and/or the possible function paths that could be the consequence of these function values(consequence paths). ...
Process safety is of considerable concern for society, in order to reduce the risk for major accidents with severe consequences for human lives and economy. The accidents also demonstrated process complexity as a major challenge, for process safety. Presently process safety is evaluated using qualitative methods which rely upon careful bookkeeping for reevaluation when process modifications and improvements are considered. Consequently it is desirable to develop a more systematic modeling methodology which may be applied for safety assessment and which conveniently may be reused when necessary.
A representative qualitative modeling framework is Multilevel Flow Modeling (MFM) which is based on functional modeling. It has been suggested that MFM can deal with the complexity of design and operation of process engineering systems with a promising application future. The purpose of the PhD project is to develop innovative modeling methods for automated analysis and evaluation of safety in industrial processes, especially in oil and gas industry. Validation of functional models is a key issue dealt with in the thesis. The thesis conducts in-depth research on modeling, reasoning, validation and safety analysis applications in the MFM modeling framework.
On the basis of an abstraction hierarchy theory, the foundation of MFM theory is introduced and an MFM modeling procedure is proposed for preventing the modeler from making errors. Dynamic simulator of a three phase separation process is established. By following the proposed modeling procedure, an MFM model is built for the first time. The modeling of the assumed thermodynamic gas-liquid phase equilibrium in the separator is discussed as well. The case study demonstrates the applied modelling procedure and also the strength of MFM for modeling of a real technical oil and gas process.
Based on the existing research of reasoning rules of MFM, a new reasoning strategy of extended MFM based on roles is proposed. The reasoning strategy is applied for an “untraditional” HAZOP study. The case study shows that the study extends the MFM model expression and reasoning ability. By including roles the discrimination between different types of causes for failure is improved.
To deal with the MFM model validation problem, a scientific-based validation method is proposed. With the application of the method for validation of the proposed MFM model for the three phase separation process, the qualitative confidence of the model is assured.
To systematically identify cause and evaluate the potential effect of a failure, an integrated qualitative and quantitative modeling framework for HAZOP studies that uses MFM with a knowledge-based reasoning system, together with a risk matrix, and quantitative dynamic simulation for verification and validation risks has been proposed. The integrated framework is successfully applied to a realistic three-phase separation process system. The results demonstrate the importance of the formulation of MFM models to represent the physical system for acquisition of HAZOP knowledge in the qualitative part of the overall methodology. From this point of view, the quantitative analysis based on the dynamic simulation complements and enhances the MFM model based process safety analysis of the system in particular with regard to the transient dynamics of the system. The integrated methodology could be best suitable for FEED (Front End Engineering Design) stage of process development.
... 21,22 In this study, a graphical editor based on the Java-based ShapeShifter framework supporting MFM modeling (called the MFM Editor) developed by the Institute for Energy Technology in Norway is used. 23,24 This MFM Editor integrated with the MFM reasoning system developed by DTU can generate cause and consequence trees for a given deviation in a system function. ...
The paper proposes a novel practical framework for computer assisted Hazard and Operability (HAZOP) that integrates qualitative reasoning about system function with quantitative dynamic simulation in order to facilitate detailed specific hazard and operability analysis. The practical framework is demonstrated and validated on a case study concerning a three phase separation process. The multilevel flow modeling (MFM) methodology is used to represent the plant goals and functions. First, means-end analysis is used to identify and formulate the intention of the process design in terms of components, functions, objectives and goals on different abstraction levels. Based on this abstraction, qualitative functional models are constructed for the process. Next MFM specified causal rules are extended with systems specific features to enable proper reasoning. Finally systematic hazard and operability analysis is performed to identify safety critical operations, its causes and consequences. The outcome is a qualitative hazard analysis of selected process deviations from normal operations and their consequences as input to a traditional HAZOP table. The list of unacceptable high risk deviations identified by the qualitative HAZOP analysis is used as input for rigorous analysis and evaluation by the quantitative analysis part of the framework. To this end, dynamic first-principles modeling is used to simulate the system behavior and thereby complement the results of the qualitative analysis part. The practical framework for computer assisted HAZOP studies introduced in this paper allows the HAZOP team to devote more attention to high consequence hazards.
Decision support systems are a key focus of research on developing control rooms to aid operators in making reliable decisions, and reducing incidents caused by human errors. For this purpose, models of complex systems can be developed to diagnose causes or consequences for specific alarms. Models applied in safety systems of complex and safety-critical systems require rigorous and reliable model building and testing. Multilevel Flow Modelling is a qualitative and discrete method for diagnosing faults and has previously only been validated by subjective and qualitative means. To ensure reliability during operation, this work aims to synthesize a procedure to measure model performance according to diagnostic requirements. A simple procedure is proposed for validating and evaluating the concept of Multilevel Flow Modelling. For this purpose, expert statements, dynamic process simulations, and pilot plant experiments are used for validation of simple Multilevel Flow Modelling models of a hydrocyclone unit for oil removal from produced water.
The chapter goes into detail of some new methods. First is the system-theoretic process analysis for scenario identification followed by the blended hazard identification (Hazid) method. Then, an overview is given of all trials thus far to automate hazard and operability study. Next, the principles of Bayesian data analysis are explained and the Bayesian network technique introduced. Because in risk analysis we have to deal with various types of uncertainty, this is given much attention. Fuzzy set approach is briefly explained. Examples are given of Bayesian network solutions of, for example, layers of protection cost-benefit. This chapter also shows how data of performance indicators can be used to measure effect on management effectiveness with, in turn, the state of workers and the relation with human error. It is further shown how Petri net and agent-based modeling can be applied in risk analysis. Finally, methods are discussed that can be used in resilience engineering.
Process safety is of considerable concern for society, in order to reduce the risk for major
accidents with severe consequences for human lives and economy. The accidents also indicated
process complexity as a major challenge for process safety. Presently process safety is evaluated using
qualitative methods which rely upon careful bookkeeping for reevaluation when process modifications
and improvements are considered. Consequently it is desirable to develop a more systematic
modeling methodology which may be applied for safety assessment and which conveniently may be
reused when necessary.
An established qualitative modeling framework is Multilevel Flow Modeling (MFM) which is
based on functional modeling. It has been suggested that MFM can deal with the complexity of design
and operation of process engineering systems with a promising application future. Qualitative
modeling and reasoning as implemented with MFM can with advantage be combined with quantitative
methods in order to automate analysis and evaluation of safety in industrial processes, especially in
oil and gas industry with increased coverage of the analysis or for validation purpose.
The paper will point out the difference and connections between the qualitative modeling (e.g.
functional modeling) and quantitative modeling (e.g. differential and algebraic equations, DAEs) in
the process safety context. Then the MFM method will be introduced. A recent HAZOP study of an oil
and gas separation plant is summarized. It is shown that validation is a key issue here. It has been
investigated how the reasoning results from an MFM model could be validated by comparing it with
simulation using a quantitative model. However, due to the complexity of advanced industrial process
system, MFM still faces many challenges in industrial process safety application. Finally, the
suggested future work within the aspects of supporting for MFM modeling construction, reasoning,
validation and counteraction planning are discussed.
Functional models are increasingly being used in industries, such as oil and gas, for decision making problems. However, the results obtained from the models that decision makers rely on are dependent on whether the model of the system is suitable for the specified purpose. Hence, it is desirable to check the consistency and suitableness before the model is applied in problem solving.Consequently the questions of model verification and validation need to be addressed on a functional modeling background to develop a systematic methodology on a scientific basis to lead to a practically applicable procedure.
The Multilevel Flow Modeling (MFM) methodology is adopted in the paper as a formalized qualitative functional modeling methodology for dynamic process systems. A procedure for a functional model validation is proposed. A simple engineering system, i.e. a central heating system is presented to illustrate the proposed functional model validation procedure.
This paper demonstrates how a generic agent-oriented framework can be used in advanced automation environments, for systems analysis in general and supervision, diagnosis and prognosis purposes in particular. The framework’s background and main application areas are briefly described. Next, the function-oriented method Multilevel Flow Modeling (MFM) and its reasoning mechanisms that have proven strength in qualitative planning, modeling and diagnosis activities are introduced. The main enhancements of the framework, as well as an MFM editor based on the framework and towards function-oriented supervision, diagnosis and prognosis purposes are equally explained. Finally, the paper sums up by also addressing plans for further enhancement and in that respect integration with other tailor-made tools for joint treatment of various modeling and analysis activities upon advanced automation environments.
This paper is the first publication of the generic application framework ShapeShifter, including its background and preliminary development. The framework provides support for the enhancement of existing and development of new applications where system objects or elements containing various types of information are in some way connected. Some current and potential applications of the framework are briefly described. In that regard, plans for enhancement and new features are also presented.
The paper describes the continued development of the Java-based ShapeShifter framework, previously described in papers published in the ESREL 2010 and 2011 proceedings, including updated and new base classes, visual components and essential functionality. The framework is based on the Capability-Oriented Agent Theory, advocating a multi-purpose view of socio-technical systems as made of human, organizational and technical agents and their assets. The paper further reports on how the framework is being used in the development of a graphical editor supporting the Multilevel Flow Modeling (MFM) method, to be utilized for Supervision, Diagnosis, and Prognosis (SDP) applications in advanced automation environments. The recently started development of the ShapeShifter-based Process Modeler, which in combination with the MFM Editor and the reasoning capabilities of the MFM Workbench developed by the Technical University of Denmark will constitute the MFM Suite, is also described.
We present a new theory and claims that it plays the core role in better management of the complexity in knowledge intensive computerised systems. In systems modelling based on capability-oriented agent theory, the concept of multipurpose is regarded as the central and indisputable nature of all systems. The introduction part provides the scientific background and initiating research behind the capability-oriented agent theory. Next, the rationale for its basic elements along with their description is presented. The application of the principle is then explained and illustrated.
The MFM Editor-A ShapeShifter-based Tool for Supervision, Diagnosis and Prognosis in Advanced Automation Environments
- Harald P-J Thunem
Thunem, Harald P-J (2012b), The MFM Editor-A
ShapeShifter-based Tool for Supervision, Diagnosis
and Prognosis in Advanced Automation Environments, International Workshop on Functional Modeling
(IWFM2012), november 6-7, Lyngby, Denmark.