Support from the centre20 February 2001
In an increasingly competitive generation business, use of remote monitoring, analysis and diagnostics is a cost effective way of bringing heavyweight, specialised, and therefore potentially expensive, expertise to bear on power plant operating issues and problems. The result is optimisation of efficiencies, availabilities and costs.
With the opening up of electricity markets in many countries, new competitors have entered the business and are challenging the national power utilities. Electricity generation has become a question of investment, where each project is evaluated extremely carefully to ensure profitability with reasonable risk. Profitability is achieved through high plant performance, availability and efficiency, and optimum investment and operational costs, ie by optimising the plant life-cycle profits (LCP).
Continuous development is obviously crucial for creating winning concepts and gaining success in this highly competitive market. Reliability-centred maintenance (RCM) and condition-based maintenance (CBM) offer new tools to manage the risks. Knowledge management, ie creating, refining and sharing crucial information and experience, is one of the most important core competencies of the new generation of energy companies.
Remote support is nowadays one of the strategies available to facilitate efficient co-operation between plant personnel and highly specialist experts. With effective co-operation, any problem areas can quickly be identified and addressed, while plant operating strategies can also be developed. In addition a systematic remote performance and condition monitoring programme is a valuable tool that can help the plant owners and operators meet their major objectives, ie high availability and efficiency with optimised operation and maintenance costs.
The use of performance and condition monitoring is expected to achieve total cost savings through improved machinery reliability and operation. Condition monitoring information is used to focus maintenance on the right components at the right time thus reducing the risk of additional damage. Remote support makes the early detection of incipient failures possible and provides decision-making support at short notice. The combination of local site knowledge with technical and expert support ensures that the right conclusions are drawn from all the available data, and that this is done in a well-planned and cost-effective way.
These concepts are embodied in a concept developed by Fortum called TOP (Totally Optimised Performance). This aims to maximise life cycle profits for the plant owner, and a key element of it is linkage of the plant to a network connecting to a centralised Performance Centre (Figure 1) in Myyrmäki, Vantaa, Finland, giving the power station access to advanced remote monitoring and optimisation applications.
Remote support and other performance services are currently being provided in Finland (to the Meri-Pori, Naantali and Inkoo coal -fired plants, the Jyväskylä, Joensuu, Haapavesi and Kokkola peat-fired power plants, the Hämeenlinna CCGT, the Imatra and Oulujoki hydro plants and the Loviisa nuclear plant) and abroad (to the Peterborough, Glanford Brigg and South Humber Bank CCGTs in the UK, the Edenderry peat-fired plant in Ireland, the Grangemouth CCGT in Scotland, the Burghausen CCGT in Germany, Laem Chabang CCGT in Thailand, Teluk Gong CCGT in Malaysia and the Ujpest CCGT in Hungary). Several new agreements are under negotiation.
The Performance Centre
The main goal of the TOP concept is to create an environment for efficient co-operation between the plant personnel and particular shared resources and highly specialised experts, enabling them to solve plant problems together. The remote support system include modules at the plant site to gather the essential information, a telecommunications network to transmit the data, and intelligent analysis and decision making modules to support the experts at the Performance Centre. The newest information and telecommunication technologies provide the tools and infrastructure for this type of service. The technical basis for remote support is the use of computer networks with applications including plant models, simulators, experience data and other support systems. For troubleshooting this enables real-time co-operation based on real-time data.
Systems at the Performance Centre are especially designed for supporting experts in their decision making. Because systems are centralised, they can be as highly sophisticated and relatively expensive, but the shared cost on a per-plant basis is kept low. In particular there are of course savings in terms of the experts’ travelling time and costs. Furthermore, with a centralised system it is easier to gather case studies and experience from a variety of power plants.
The effectiveness of incipient-failure detection techniques depends on how well they can convert the mass of acquired data to useful information that relates to machinery condition and give insights into it. The task of the analysis systems is to utilise all the data measured at plants as effectively as possible. That way, failures are detected in time, and at best they can be predicted. The analysis systems have been implemented using the latest technologies for expressing deep knowledge, such as modelling and intelligent or knowledge-based systems.
An intelligent, object-oriented environment and new remote monitoring and diagnostics applications have been built with Gensym’s G2 development environment. With this environment, the experts can build their own personal alarm systems using automated performance monitoring “agents”, which specify the tests to be carried out, their timing, warning and alarm limits etc. The “agents” carry out routine tasks automatically, giving human experts more time to concentrate on the most challenging analysis tasks. The applications are in continuous development as new detection methods are added and as all new cases are added into the experience database to support trouble-shooting.
The TOP concept and systems help individuals to learn from different failure cases and each others’ experience, and are thus an essential part of a learning organisation. The effective analysis of all abnormal or failure situations is part of learning to make the right decisions and even to take predictive actions. The cases can also be input into the training simulator to be taught to others. Furthermore, through this tight and systematic co-operation, new ideas have been generated leading to successful R&D projects ensuring continuous development.
New performance monitoring modules are being continuously developed in response to experience reported from the sites. Existing modules include the following:
• Operational economics
Operational economics can be monitored using the Solvo modelling and analysis package developed by Fortum. On-line Solvo is used to calculate thermal efficiencies etc, while the off-line version can be employed to simulate various operational scenarios. Solvo also calculates and monitors individual component performance indicators, such as compressor isentropic efficiency and turbine flow constants. The data are transmitted to the Performance Centre and analysed periodically, including comparison with historical data and also with data coming from other similar types of power plants.
Solvo calculations are based on mass and energy balances and specific equations describing the functions and operations of particular pieces of equipment.
In the on-line applications, Solvo acts as a tool for monitoring operational economics and also performs condition monitoring, while producing real efficiency values and values corrected to ISO conditions. This enables real condition monitoring to be performed on various components, making allowances for ambient conditions.
Solvo allows components to be subdivided into blocks, so, for example, the performance of the different parts of a boiler, such as evaporator, superheater or economiser, can be monitored individually.
• Vibration monitoring of rotating machinery
Vibration levels and spectra are measured at site either using a portable or a fixed measurement system. The data are transmitted to the Performance Centre where the experts analyse them using analysis and decision support systems. Where there are fixed measurement systems continuous monitoring at the Centre is possible. But generally the analysis is periodic and a regular report with suggested actions is given to the plant personnel.
• Process control
Process control loops are monitored continuously using selected performance indicators (Figure 2). Based on the information obtained the experts at the Performance Centre can tune and optimise the control parameters.
• Water chemistry
The quality of water chemistry is analysed based on the relevant plant measurements. The data are monitored by the analysis system at site and the expert at the Performance Centre. If abnormal results appear, the support system suggests corrective actions, and the expert checks and confirms the results.
For all these areas, the remote monitoring can be continuous or done in the form of periodic check-ups.
The real measure of success of these remote monitoring activities is not merely the development and implementation of technical solutions and systems, but cultivation of real team work involving the plant personnel and the experts at the Performance Centre, making full use of the know-how and experience that exists on both sides.
Historical information, eg operational and maintenance history and reports from previous analyses, is essential for the experts to draw conclusions. Databases containing information on similar equipment or incidents at other power plants are also continuously updated to help the experts in their decision making. These databases can also be used by plant personnel through the intranet and extranet.
As a further extension of the approach, going well beyond simply monitoring, applications are being developed in which operators are given advice about optimal setpoints, on the basis of analysis by specialists at the Performance Centre.
The optimiser operates in an advisory mode allowing the operators to acquire optimal setpoints for control variables. It can optimise any of the outputs while specifying limits for the other outputs and allowable values for the control variables. This optimisation is performed so that the resulting control recommendations are always in the operating region well-known to the operator. The module includes a what-if simulation, which the operator can use to study the effect of changes in the control variables in advance.
The Optimiser concept can also be applied to providing tools for optimising operator actions. At one plant such a module is in use for optimal timing of compressor washing offline and the replacement of filters (see below).
Remote support – some examples
The practical benefits of remote support are illustrated in the following examples, which cover gas turbine blade vibration monitoring, steam temperature control in a once-through boiler, boiler leak detection, operation point optimisation at a coal-fired plant, as well as the compressor washing case.
Gas turbine blade failure
At one plant a step change was noted in gas turbine vibration readings (Figure 3). The change was small, the vibration amplitude was small even after the change, which meant that there were no alarms at site.
The change was however captured in the regular check-ups, which were part of the plant’s vibration monitoring support agreement with the Performance Centre. It was found that the step change was seen in most of the vibration readings. The amplitude change, mainly in 1xN (machine rotation frequency) steps, indicated, that the cause was a sudden increase of unbalance. The most likely explanation for this kind of situation is a blade failure.
After this analysis by the vibration experts, the plant manager decided to run the machine down for endoscopic inspection, and the endoscopic inspection did indeed reveal a blade failure. This way any further damage to the gas turbine was avoided, for the risk of major breakdown is clearly high when running with a damaged machine. Furthermore, the material experts at the Performance Centre were able to analyse possible reasons for the blade failure.
Steam temperature control
In one process control loop monitoring case, occasional overheat and abnormally large temperature changes were detected in superheaters. Their continuation would have had a major impact on the remaining lifetime of the units. The control loops were analysed (using displays of the type shown in Figure 2) and faults in the structure of two control loops were found. The loops were fixed and instances of overheating were practically eliminated, while temperature fluctuations were halved or better.
Boiler leak detection
In a combined cycle gas turbine plant, with two identical units, continuous monitoring revealed a gradually increasing divergence in plant performance between the two. Both plant and boiler efficiencies showed that unit 2 efficiency was decreasing relative to unit 1. Calculations with the plant thermal model showed clear performance degradation in unit 2.
In addition to the regular upper-level measurements and their differences, the subprocesses and their measurements were monitored with the G2-based expert support systems at the Performance Centre. For example, the low-pressure circuit measurements from feedwater and steam flow rates and the differences between those parameters were monitored for both units. The normal difference between those parameters was about 0.2 kg/s due to blow-down and uncertainty of steam mass flow measurement. But measurements showed the difference to be as high as 1.5 kg/s. This was clearly due to a boiler tube leak. Although the difference had increased markedly, the plant as a whole consists of four similar units so total make-up water consumption had remained within its normal variation. Thus it was very difficult for the plant operators to notice the change, or locate the failure.
The boiler leak was detected faster and more easily through monitoring differences between the low-pressure circuits than from measuring plant or boiler efficiencies because it revealed the original failure cause and its location. In this case, the boiler leak was detected from the low-pressure circuit, via the G2 expert system, as much as two weeks before the changes could be observed from the plant or boiler efficiency levels (see Figure 4). In addition to giving the plant personnel knowledge of the real condition of the boiler, this approach gives them more time to plan and prepare for repairs.
With such model-based methods, failures can even be located simultaneously with being detected. In some cases, the failures can be predicted using the submodels and lower-level measurements before the symptoms can even be detected in the upper-level measurements.
Operation point optimiser
The Operation Point Optimiser has been applied to the Naantali coal-fired plant in Finland (Figure 5). In this case the Optimiser uses a model of the steady state behaviour of the coal-fired boiler. Due the complexity and severe non-linearity of the process, a neural network approach was chosen. The data required for constructing the model were gathered in several tests, but without disturbing the normal operation of the plant. The control inputs to the boiler model include oxygen in the flue gas, the percentage of burning air fed as over fire air (OFA), the coal/air ratio and the tilting angle of the burners. The outputs of the model are the boiler efficiency, the level of NOx emissions, the unburnt carbon in the fly ash and the maximum temperature in the boiler.
In addition to setpoints, the system has been used for analysing what-if scenarios and it also provides a continuous prediction of the current level of unburnt carbon in fly ash, that otherwise cannot be measured on-line.
The purpose of the optimisation is to maximise the boiler efficiency, while simultaneously keeping the plant emissions as low as possible, assuring the quality of the fly ash and preventing too high metal temperatures in the boiler tubes. The user interface of the optimisation is integrated into the plant information system, while the actual processing is performed on a separate workstation running a G2 environment. The system can be used by plant personnel, as well as by experts at the Performance Centre, with the aim of supporting plant staff in their use of the system and performing what-if and other analyses.
Optimisation tool for operational actions
Compressor and inlet air filter cleanliness has a significant effect on the thermal performance of gas turbines. To ensure optimum gas turbine performance, a new software tool has been developed for economically optimal scheduling of compressor offline-washing and filter replacement.
At most power plants, filters are replaced only on the basis of measured pressure drop, and offline-washes are done according to a predetermined advance planning schedule. Because the level of dirt in the inlet air is not constant through the seasons, this cannot be the economically optimum solution. Also electricity price changes move the optimum date for these operational actions. Accordingly an optimisation tool (Figure 6) has been developed which takes into account electricity and fuel prices, the manufacturer’s correction curves for pressure drop and measured performance changes. The optimisation tool can be used on a web-basis through the extranet. The software tool consists of four different parts:
• Process parameter transfer and filtering with the criteria needed for the calculation. Filtering enables, for example, the elimination of different load changes.
• Optimisation tool, which suggests the next dates for offline-washing and filter replacement.
• What-if analysis, where the user can study how the postponed offline-washing effects total costs, etc.
• History database, where the user can archive all the information about offline-washing and filter replacement, such as the type of filters or detergent used.
Looking to the future
The key to successful condition monitoring or condition based maintenance is to make better use of the existing measurement data. Automated systems are proving increasingly effective for routine O&M, but the lack of people, in particular experienced people, still calls for additional support in abnormal situations. This requires networking of the organisations, and raises the question of how to get the best possible co-operation between the experts at the Centre and the people at the plant. Advances in communications technology leave imagination as the only limit. The challenge is to harness the innovative ideas, knowledge and expertise of the most experienced people and target them better to improving power plant performance.