Fuzzy business

3 August 2002

The Halden Reactor Project has developed two systems to investigate how signal validation and thermal performance monitoring techniques can be improved. By Øivind Berg, Paolo Fantoni, Svein Sunde and Fridtjov Øwre

PEANO is an online calibration monitoring system that makes use of artificial intelligence techniques. The system has been tested in cooperation with EPRI and Edan Engineering, using real data from a US PWR plant. These tests showed that PEANO could reliably assess the performance of the process instrumentation at different plant conditions. Real cases of zero and span drifts were successfully detected by the system.

TEMPO is a system for thermal performance monitoring and optimisation, which relies on plant-wide first principle models. The system has been installed on a Swedish BWR plant. Results obtained show an overall rms deviation from measured values of a few tenths of a percent, and giving goodness-of-fits in the order of 95%. The high accuracy demonstrated is a good basis for detecting possible faults and efficiency losses in steam turbine cycles.

Signal validation

To ensure safe, efficient, and economical operation of nuclear power plants one needs to calibrate the safety-related instrument channels. Current practice of the calibration processes is to periodically determine the performance characteristics of an instrument and make adjustments if necessary. Typically, this is done once every fuel cycle, irrespective if the instrument is in need of calibration. This is a costly and labour-intensive approach and does not guarantee detection of instrument failure during plant operation, when it could contribute to optimisation.

Online monitoring of instrument channels provides information about the condition of the monitored channels through accurate, more frequent monitoring of each channel's performance over time. This type of performance monitoring is a methodology that offers an alternative approach to traditional time-directed calibration. The online monitoring approach is based on evaluating the instrument channel performance by assessing its consistency with the parameters estimate, which is calculated from related plant indications. A number of different online monitoring implementations have been developed over the last few years and some plants already use it in addition to the time-based calibration programme to obtain additional information for plant maintenance.

A study has been performed with the PEANO system on data collected from a US nuclear power plant during normal operation. The aim was to assess how well some of the online monitoring implementations are suited for the task they are designed for. One of implementations selected was the PEANO system, which makes use of artificial intelligence techniques for its purpose.

PEANO system

Artificial neural networks (ANN) and fuzzy logic can be combined to exploit the learning and generalisation capability of the former technique with the approximate reasoning embedded in the latter approach. Real-time process signal validation is an application field where the use of this technique can improve the diagnosis of faulty sensors and the identification of outliers in a robust and reliable way.

PEANO implements a fuzzy-possibilistic clustering algorithm to classify the operating region in which the validation process has to be performed. The possibilistic approach (rather than probabilistic) allows a "don't know" classification that results in a fast detection of unforeseen plant conditions or outliers.

The fuzzy classifier identifies the incoming signal pattern (a set of reactor process signals) as a member of one of the clusters covering the entire universe of discourse represented by the possible combinations of steady state and transient values of the input set in the n-dimensional input world. Each cluster is associated with one ANN that was previously trained only with data belonging to that cluster, for the input set validation process.

During the operation, while the input point moved in an n-dimensional world (because of process state changes or transients), the classifier provides an automatic switching mechanism to allow the best tuned ANN to do the job.

There are two main advantages in using this architecture: the accuracy and generalisation capability is increased compared to the case of a single network working in the entire operating region; and the ability to identify abnormal conditions - where the system is not capable of operating with a satisfactory accuracy - is improved.

PEANO has a client-server architecture. The server is connected to the process through a TCP/IP communication protocol and the results of the validation activity are transferred to the client programs, also using TCP/IP. (figures currently not available) Figure 1 shows the display of a PEANO client during an online validation test.

The error bands in the mismatch plots are calculated by PEANO during the training according to the expected error of prediction for each particular cluster and signal.

Test results using PEANO

Figures 2 to 4 describe some results achieved with PEANO, using real data from a US PWR nuclear plant (data provided by EPRI). PEANO was used to monitor 55 process signals up to the secondary side of the steam generators. The data was taken from three months of operation (March to May 2000) at different operation conditions.

Figure 2 shows the recalibration of a steam flow sensor while the plant was in operation. It can be seen that after this intervention, the mismatch was well within the two standard deviation limits.

Figure 3 shows the monitoring of a feedwater flow sensor during the power-up ramp. While the mismatch between PEANO and the instrument is always within the limits, it is interesting to note that the tolerance band width changes with the plant operating conditions, because the accuracy of the system cannot be the same in all the possible conditions.

Figure 4 illustrates an interesting example of span drift of a steam flow sensor. Span drifts are difficult to detect because they show up only at some location of the instrument range. In this real-life example, the instrument was perfectly inside the calibration range until the power level came close to the rated level (high end of the instrument range). At this point the instrument started to drift and eventually finished outside the allowed tolerance band. At the plant, this drift was discovered only one month later.

Thermal performance monitoring

An important task for the operator during normal operation is to optimise plant performance with respect to economy and safety. Monitoring and optimisation of the thermal efficiency is getting increasingly important as energy market liberalisation exposes the plants to ever-stricter availability requirements and fiercer competition. Typical problems are fouling and leakages in heat exchangers, leaking valves, degradations of pumps and compressors, as well as drift in sensors, in particular flowmeters. These faults may not be severe enough to cause major operational problems, but will affect the economic performance of the plant if not identified and rectified at an early stage.

TEMPO system

TEMPO is a computer system that aims at satisfying information needs associated with efficiency monitoring, online calibration of plant measurements, fault detection, diagnosis, and "what-if" analysis of steam turbine cycles and similar systems. TEMPO relies on plant-wide first principle models, which are assembled from a menu of standard components in a graphical user interface. The models are configured by a click-and-drop procedure in a configuration screen, allowing for complete customisation.

TEMPO receives data from the process online, and performs steady-state calculations in three different modes.

In the simulation mode TEMPO calculates all states in a turbine cycle for a given input. For example, the state at the steam generator output to the high pressure turbine, or its control valves, and the state of the coolant input to the condenser may be specified. In the simulation mode, the user may freely study the consequences of changes in parameters or configurations, for example introducing leaks or fouling in heat exchangers.

In the optimisation mode, TEMPO varies user-selected parameters of the components of the plant in order to maximise efficiency or power. For example, the user may choose to optimise power by letting TEMPO adjust simultaneously the speed of selected pumps and positions of selected valves in order to find the combination of these that makes the plant attain the highest power. All parameters are varied within limits set by the user. In addition, other constraints such as temperature, pressure and steam quality may be imposed.

In the monitoring mode TEMPO adjusts user-selected parameters of components in order to minimise the difference between measured values of user-selected temperatures, flows and pressures and the corresponding calculated values of the plant model. Usually the input states to the system are varied if these are not known with high accuracy. The points that emerge from this are: an overall value for the probability that the process is free of (unknown) faults, the so-called goodness-of-fit Q; adjusted values for the measurements; process states for all streams included in the model; and estimates for component parameters.

The goodness-of-fit represents an overall assessment of whether there are faults or not in the process, and is usually the most important single number for fault detection and alarms. For tracking down the cause of an assumed fault, the values of the component parameters and process states will be of significant diagnostic value. In addition to values themselves, TEMPO calculates the uncertainty of all its estimates.

Test results using TEMPO

Recent tests at a BWR give an overall rms deviation from measured values of a few tenths of a percent, and giving goodness-of-fits Q in the order of 95%. Also, TEMPO had no difficulties detecting a known case of a faulty sensor.

In Figure 5, the calculated and measured feedwater flow (nearly overlapping curves) is shown. The typical deviation in these were 1-2kg/s, corresponding to 0.05-0.1%.

Figure 6 illustrates the use of TEMPO for calculating the consequences of various separator plate leaks in feedwater heaters in terms of their inlet and outlet temperatures, which may be of value in fault diagnosis.

Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.