You are on page 1of 4

Research on An Operation Monitoring and

Evaluation Technology for Advanced Main


Control Room of Nuclear Power Plant
Yang ZHAO, Wei LI, Ling ZHOU, Yan XIONG, Yifan JIAN and Minghu CHENG

 operating procedures presentation. Some mature technologies


Abstract—Human error is still a major cause contributing to have been introduced into the main control rooms. However,
the incidents and accidents at nuclear power plant. Especially the prevention of omission errors, inadequate or improper
operators try to flexibly operate the plant system according to actions of operators still greatly depends on strictly abiding by
their experience or encounter unfamiliar situations, human errors
may occur. So far, the prevention of omission errors, inadequate
procedures and cross-checking between operators.
or improper actions of operators still greatly depends on strictly The target of operation evaluation is to analyze the
abiding by procedures and cross-checking between operators. applicability of one or a group of operational actions by
This paper presents an operation monitoring and evaluation operators to the operation of nuclear power plant. According to
system for the advanced main control room of nuclear power the different evaluation criteria, the applicability of an
plant. The proposed system provides several functions including operation action can be classified into the following two types:
condition monitoring, action monitoring, operational goal
analysis, system reliability analysis and risk alert, which can
(1) Special applicability: the correctness of an operational
enable flexible operation by operators and therefore will be action is evaluated in a narrow sense using specified rules such
significant to improve the efficiency of nuclear power plant. as laws, standards, technical specifications and operating
procedures.
Index Terms—Process modeling, cause effect analysis, risk (2) Generalized applicability: the correctness of an
analysis, nuclear power generation. operational action is judged in a broad sense using general
criteria such as risk, reliability, availability and hazard.
Accordingly, there are two types of methodologies for
I. INTRODUCTION operation evaluation:

A LTHOUGH THE Operators in the main control room (MCR)


of nuclear power plant are well-trained, human error is
still a major cause contributing to the incidents and accidents at
(1) Diagnosis: evaluate whether an operational action is
consistent with the specific rules.
(2) Prognosis: evaluate whether an operational action will
nuclear power plant. Especially operators try to flexibly operate lead to undesirable consequences.
the plant system according to their experience or encounter This paper presents an operation monitoring and evaluation
unfamiliar situations, human errors may occur. Adoption of system for the advanced main control room of nuclear power
operator support systems for reducing the work and cognitive plant. The proposed system provides both diagnosis and
load of operators by artificial intelligence therefore becomes a prognosis functions for evaluating an action of operator both in
trend of advanced main control room where the operators will the narrow sense and the broad sense of applicability.
monitor and implement plant operation and controls by In the following parts of this paper, the framework of the
computers. So far, various operator support technologies have proposed operation monitoring and evaluation system will be
been developed. These technologies range from safety given and the key technologies will be introduced.
parameters monitoring, operational condition monitoring,
alarm analysis, to fault diagnosis and computerized emergency II. FRAMEWORK DESIGN FOR OPERATION MONITORING AND
EVALUATION
Manuscript received August 6, 2018. The framework of the proposed operation monitoring and
Yang ZHAO is with the the Science and Technology on Reactor System evaluation system is shown in Fig.1.
Design Technology Laboratory, Nuclear Power Institute of China, Chengdu,
China (e-mail: 30726091@qq.com).
Wei LI is with the the Science and Technology on Reactor System Design
Technology Laboratory, Nuclear Power Institute of China, Chengdu, China
(corresponding author, phone: 0086-28-85904114; e-mail: lcyccr@163.com).

POWERCON2018 Paper NO. 201805280000013 Page1/4 4794


978-1-5386-6461-2/18/$31.00 ©2018 IEEE
operators on the objectives of each system qualitatively. The
impact of an action on the goals, whether they are expected or
not, will be prompted in order to attract the attention of
operators.
E. Reliability Monitoring System
The components are configured to form structures and
systems to fulfill the desired functions and goals. Once an
action taken by operators changes the state of a component,
Reliability Monitoring System (RMS) is used for predicting the
availability of the system which involves the component in
realizing its goals quantitatively.
F. Risk Alert System
Alerts will be given in case operator action violates technical
regulations and operating procedures, as well as threatens the
goals of systems.
In the following section, the methodologies for some key
technologies including operating procedure implementation
supervision, goal analysis and reliability monitoring will be
Fig. 1. Framework of operation monitoring and evaluation system. briefly summarized.

The proposed operation monitoring and evaluation system III. KEY TECHNOLOGIES FOR OPERATION EVALUATION
include the following subsystems:
A. Operating Procedure Implementation Supervision
A. Condition Monitoring System The OPIS utilizes the hierarchical structure of operating
Condition Monitoring System (CMS) is for identifying the procedures as shown in Fig.2.
actual state of each system, structure and component (SSC)
such as being in operation, standby, failure or maintenance, as
well as the operational modes of the plant system such as being
in power operation, startup, hot shutdown, cold shutdown,
refueling, and accidental conditions.
B. Action Detection System
In the advanced main control room, both automatic control
signals and operator commands can be available though digital
I&C system. Action Detection System (ACS) is used for
distinguishing the soft control commands by operators from
automatic control signals in order to detect any action taken by
operators on SSCs.
C. Operating Procedure Implementation Supervision
System
In case an action by operators corresponds to a certain
operating procedure under the detected operational condition,
the Operating Procedure Implementation Supervision (OPIS) is
used for monitoring operators whether implement operating
procedures correctly. Any violation of the operating procedures
will be alerted.
D. Goal Analysis System Fig. 2. Hierarchical structure of operating procedures for implementation
supervision.
Under a given operational condition, each system has its own
design purposes (objectives) and desired state. Goal Analysis
OPIS starts with a check of entrance conditions of
System (GAS) is used for predicting the effects of an action by

POWERCON2018 Paper NO. 201805280000013 Page2/4 4795


procedures to identify whether there exists any operating The mass flow structure M1 realizes two objectives
procedure corresponding to the actual operational conditions including maintaining Steam Generator (SG) level (G1) and
detected by CMS. An operating procedure is organized at three steam generation (G2). Five flow functions include:
layers. (1) F1: water supply by feed water tank;
(1) Task layer: consists of the items of an operating (2) F2: feed water transport to SG;
procedure. (3) F3: storage of feed water in SG for steam generation;
(2) Sub-task layer: consists of the sub-items of an item. (4) F4: steam transport to turbine;
(3) Step layer: consists of the steps of a sub-item. A step (5) F5: steam drives turbine.
corresponds to an action that the operators can really Auxiliary feed water is provided for maintaining the SG level
implement. (G3) in case of loss of feed water.
Once an action is detected by ADS, the OPIS will identify The causalities of MFM can be expressed by causal
the step in the operating procedure corresponding to the action. dependency graph as shown in Fig.4. Once an action is detected
If the step is identified, OPIS will check the step requirements by ADS, MFM can be used for prognosis. As shown in Fig.5, if
including the prerequisites and the success criteria for F2 is detected to be loflo (flow rate of feed water is low because
execution of the step. The prerequisites include the conditions the operator reduces the opening of feed water valve) , it can be
and timing of the execution of the step and the completion of predicted that F3 (water level of SG is low) and F4 (flow rate of
the previous step. A task is fulfilled when all of its subtasks and main steam) will be also loflo according to the causalities in
steps are fulfilled and only when operators execute the MFM in Fig.4.
procedures step by step in sequence without any violation, their
actions will be recognized as correct. If any prerequisite of a
step is not met, it will trigger an alert, for example, omission of
step if the previous step has not been executed.
B. Goal Analysis
Multilevel Flow Modeling (MFM)[1,2] is used for goal
analysis. MFM is a functional modeling approach for
describing the complex process system with interactions. MFM
organizes plant information on different level of abstractions.
Fig. 4. Causalities of MFM expressed by causal dependency graph.
Two adjacent levels are connected by means-end relations. For
two flow functions at two adjacent levels connected with a
certain means-end relation, the flow function at the lower level
is the means of the flow function at the higher level, while the
flow function at the higher level is the purpose (end). With
means-end relations, MFM provides support to the goal based
problem solutions.
Taking mass flow of the secondary loop of PWR as an
example, a simplified MFM model extracted from a MFM for
normal operation is shown in Fig.3.

Fig. 5. Prognosis using causalities of MFM.

The predicted state of each MFM function will be further


compared with its actual state to evaluate the consequences of
action on goals. Fig.6 presents an example on goal analysis (G1
in Fig.3) by MFM. It is clear that if operator reduces the
opening of feed water valve, the actual high water level of SG
will be mitigated, the low water level of SG can not be
mitigated, the normal level of SG will decrease.

Fig. 3. A simplified MFM of the mass flow for the secondary loop of PWR.

POWERCON2018 Paper NO. 201805280000013 Page3/4 4796


objectives is also known as living PSA (Probabilistic Safety
Analysis).

IV. CONCLUSION
Human error is a major contributor of the incidents and
accidents in nuclear power plant. In order to prevent the human
errors, an operation monitoring and evaluation system that
integrates condition monitoring, as well as rule-based,
goal-based and risk-based approaches for operation evaluation
is proposed in this paper. The proposed system can not only
monitor the implementation of operating procedures by
operators, but also predict the effects of an action on the
objectives of systems qualitatively and quantitatively in both
narrow and broad sense of applicability.
This paper introduces the framework of the proposed
operation monitoring and evaluation system and the
methodologies of some key technologies. The prototype system
for a PWR system is now under design and development.

Fig. 6. An example of goal analysis by MFM.


REFERENCES
C. Reliability Monitoring [1] M. Lind, “Modelling goals and functions of complex industrial plant,”
Journal of Applied Artificial Intelligence, vol.8, pp. 259-283, 1994.
Reliability monitoring is performed by GO-FLOW [2] M. Lind, “An introduction to multilevel flow modeling, ” International
program[3]. A GO-FLOW model is a success-oriented and Journal of Nuclear Safety and Simulation, Vol.2(1), pp.22-32, 2011.
directed acyclic graph. Operator and signal are two basic [3] T. Matsuoka, M. Kobayashi, “GO-FLOW: A New Reliability Analysis
elements of a GO-FLOW model. Each operator represents a Methodology”, Nuclear Science and Engineering, Vol.98, No.1,
pp.64-78, 1988.
fundamental reliability logical relationship and a signal may
represent a mass flow, energy flow or control command.
A generic GO-FLOW model of component is shown in Fig.7.
The generic model combines 5 basic models (indicated in
different colors) each of which represents the reliability Yang Zhao is a senior engineer at Nuclear Power Institute of China (NPIC). He
characteristics of a component in operation, standby, is now the vice director of instrumentation and control department.
maintenance, test and failure, respectively. During reliability Wei LI is an engineer at Nuclear Power Institute of China (NPIC). He is
monitoring, the state of model will be switched to the actual majored in main control room design and operator support technologies by
state of component detected by CMS by turning demand signals artificial intelligence.
between 1 and 0 to open or block the relevant logic branches.

Fig. 7. Generic model of equipment with living PSA conception.

The technology for updating model to reflect the changes


of component state on the ability of system in realizing its

POWERCON2018 Paper NO. 201805280000013 Page4/4 4797

You might also like