Ana Isabel Galvez & Emilio Moralo, Indra, Spain
Maintenance costs are some of the most significant in power generation, apart from those concerning investment. Use of the right tools and the following of good practices can reduce these costs. SmartSignal’s EPI*Center online advanced analytic, condition-based monitoring tool helps to achieve this, and improves plant availability and efficiency.
EPI Center’s remote monitoring capacity
EPI*Center can be integrated into a power generation management strategy, either within the plant organization or within a centralized monitoring unit. The implementation process includes a customized approach that lets the potential customer experience the system and its fundamentals in its own installation.
The system continuously monitors the real values produced by the most important assets in a plant and compares them with expected values to detect subtle deviations much earlier than traditional monitoring is able to. This allows staff to analyze a problem and decide how to address it with the lowest possible impact on the availability of the plant.
A plant operator can effectively integrate the software into a global monitoring and diagnosis strategy that plays a part in the process of managing power generation plant.
Every piece of equipment in a power plant is unique. Each piece of equipment is also operated and maintained differently and run under differing ambient conditions. EPI*Center understands the unique operating profile of each piece of equipment across all known loads and ambient conditions.
It looks at each set of readings of each piece of equipment in the context of its unique operating profile. Based on load, ambient condition and operating profile, the software continuously monitors the difference between actual and expected plant data. The difference between an actual value and an estimated value is called a residual. An admissible threshold for a residual is based on expert experience.
Identification of alerts and incidents
EPI*Center understands what the unique operating characteristic are across loads and ambient conditions of all the correlated sensors that are part of a plant model. It defines a tight dynamic band of normal behaviour within which the deviations it detects will not cause alarm.
EPI*Center has more than 40 patents behind it. It develops models by using similarity-based modelling (SBM), which was originally developed for the nuclear industry at the Argonne National Laboratory and the University of Chicago in the USA. It can use the empirical data for any asset in a power plant to predict in real time the expected value for each parameter after taking into account the operating conditions of the asset. It carries this out without the manufacturer’s data or physical equations.
The system generates the so-called residual by comparing actual data with estimated data while taking into account the load, ambient conditions and the unique operating characteristics of equipment. When the residual breaches defined thresholds, the system generates an alert. This is a type of first-level alarm that does not necessarily mean that the equipment is in a critical situation. If an alert appears repeatedly or a diagnostic rule is violated, the system generates an incident. Diagnostic rules are defined for individual parameters and can involve the behaviour of different signals in a model. Residuals have so-called fingerprints that the software uses to diagnose and prioritize impending failures).
EPI*Center comprises four main modules:
- EPI Center Run-time engine
- EPI Center Database
- Watchlist Manager
Workbench is the application that trains and maintains the models. It allows the selection of friendly data to build the models, test them and define thresholds and diagnosis rules. Runtime Engine is the calculation engine for the online monitoring. All the input and output data are stored in Database, and Watchlist Manager displays the information about incidents. Any user can access Watchlist Manager through the internet or an intranet.
Plant data can be acquired from databases such as SQL and ODBC, the most common solution being an interface with the PI system by software maker OSIsoft. The system can be also integrated with a maintenance management system to coordinate inspection orders or work orders with the incidents generated by EPI*Center. The system architecture allows effective remote monitoring, which gives the tool great flexibility in its choice of operating model.
Test drive prior to implementation
Successful implementations of the system are broad, but each customer must evaluate and understand how EPI*Center contributes toward corporate targets. For example, each customer must ask whether it is able to detect the most frequent and costly failure modes with the actual instrumentation and information infrastructures, what the expected revenue is, and how it can be integrated into operations?
Solution Test Drive has been designed with this in mind. It is a programme that allows the assessment of EPI*Center’s potential value to a company without significant investment in time or money. It consists of three phases:
- Discovery Analysis: The first step is to know more about the SmartSignal solution and to gather information about the plant, assets and business processes before taking the decision to go further with the evaluation.
- Solution Value Analysis: Based on the experiences of EPI*Center’s maker, an initial value analysis is developed for a certain asset base. The methodology identifies critical assets and quantifies expected value based on annual savings and revenue improvement.
- Solution Test Drive: The aim of this phase is to allow prospective clients to gain experience of the SmartSignal solution to be able to make a decision about implementing it. The Solution Test Drive consists of a customized study that helps a potential customer to better know EPI*Center and the potential benefits for their company. Over several meetings it covers system fundamentals and the application of EPI*Center to evaluate a historical catastrophic failure at the company. The aim is to determine what the maintenance cost savings and revenue improvement would be with the early detection that predictive monitoring provides. Offline analysis of actual data can also demonstrate the system’s capacity.
Besides the evaluation of the specific case, the Test Drive includes several workshops on the main aspects of the predictive monitoring tool. These include the foundations of the technology; how to sustain and maintain EPI*Center over a certain asset base; and the infrastructure assessment, which provides technical details about the hardware requirements, system architecture, installation and configuration, and SQL database.
An evaluation is then made of the approach to take to implement the system. Finally, a ‘path forward’ meeting is held with the decision-makers in the company. This meeting includes a summary and conclusions of the previous meetings. A discussion takes place about any barriers to a broad implementation, and the path forward is drawn.
An option that allows users to gain good knowledge of EPI*Center and to experience how it would fit within a company is the pilot project. Based on a temporary licence of, typically, six months, EPI*Center is implemented using representative assets and is kept running online for several months to prove its detection capacity and show how it fits into the organization. This allows decisions to be taken based on a company’s own experience and corporate data analysis.
Determining the value of early warning, however, requires more than just simple calculations. To summarize, the methodology necessary for calculating the benefit of early warning includes:
- Value analysis: This ensures that the project aligns with business goals. It is an analysis ranking of the business and economic impact of major equipment failure modes by value of failure, frequency of failure and existing sensor availability.
- System analysis: This provides a systematic way to identify a suite of models that can most accurately represent the equipment. It finds early warning signs of the key failure modes, groups into assets the models based on failure mode and captures them as fingerprint charts.
- Data analysis: This not only ensures that models are developed with data that is truly representative of normal operating conditions, it also ensures that the models accurately reflect the behavior of the process. It gathers historic failure data and certain economic data to produce deliverables that include the annual expected value of early warning of failure and the scope of the implementation of the solution.
- Availability and reliability analysis: The value case analysis uses certain operational and maintenance information to calculate an estimate of expected annual benefits for the target of the study.
The calculation of an estimate of expected annual benefits for the target of the study depends on a number of assumptions:
- Interval extension value: Performing maintenance at intervals based on equipment condition rather than the calendar means that the time between overhauls has the potential to be increased. The revenue improvement that is possible if intervals are extended is determined at this point.
- Value of reducing the duration of maintenance: The early warning of equipment problems allows repairs to be planned days, weeks or months ahead. This extra preparation can be used to reduce overall maintenance time by doubling up on repairs. The effect of reducing maintenance duration cuts the overall maintenance hours expended on repair, which means that the fully loaded labour cost per hour can be used to calculate the possible revenue improvement.
- Value of heat rate reduction: When key assets across a unit, plant or fleet are operating consistently at their normal levels, the overall heat rate will improve. This gain is achieved by minimizing low-performance variation. An estimate of this value is calculated by using the capacity factor, marginal fuel cost and an estimate of heat rate.
- Shift of unplanned maintenance to planned maintenance: The early detection of problems in performance and equipment can reduce maintenance costs and recover revenue-making opportunities that have been lost. This is because the necessary parts and maintenance skill sets are at hand, damage to equipment before repair is reduced and maintenance can be carried out at periods of low demand.
Values for the cost of a catastrophic, major or minor failure are fixed using values from the EPRI AP-2071 topical report of October 1981 entitled ‘Component Failure and Repair Data for Coal Fired Power Units’. Economic evaluation shows that the investment in EPI*Center is typically recovered in months but always in less than one year. Any early detection of a catastrophic failure may raise profitability enormously over a whole fleet.
Companies follow different schemes of system operation. Selection of a particular scheme depends on factors, such as the number of plants to be monitored, the availability of plant personnel who can follow and maintain the system, and the level of integration within a global strategy.
Annual benefit distribution in a coal fired power plant
The first option is to externalize the operation of the system. In this situation the supplier can host the system, supervise it on 24/7 basis and maintain the models. This option is very effective in case where a company wants to benefit from effective monitoring, but does not want to put resources into a specific scheme of system operation.
A company could follow this stage by involving its own staff in monitoring, but the models maintenance is provided by the system supplier. This avoids the need for special training and spending lengthy amounts of time on model training and adjustment.
Generators that want to monitoring plants can choose between two schemes. Either the company’s own operation and maintenance staff supervise EPI*Center or the plants are remotely monitored from a centralized unit. The latter option is the most widely applied for the monitoring of fleets.
A generating company’s greatest involvement is in the training of staff to develop and maintain models, as occurs in companies that have to monitor a large fleet. Any transient or intermediate options are also possible according to the actual needs of the customer.
Integration with core tools
Several acceptable schemes provide effective management of power generation, the specific scheme depending on factors such as corporate targets and size of utility size. Indra has been working for many years on supporting the power generation industry throughout the business cycle.
Several tools to support plant operation and management have been successfully implemented, such as plant information management systems, performance monitoring, combustion optimization for pulverized coal plants, advanced analytic à‚— condition-based monitoring (EPI*Center), training and engineering simulation, and generation management.
As technologies advance and power plant operations change, for as they require fewer staff, rotate personnel and have a greater availability of information, a solution that is emerging is remote and centralized monitoring and diagnosis. This groups O&M experts who can continuously supervise a fleet of power plants, which allows knowledge to be shared, the creation of benchmarks through the use of best practices and the standardization of procedures across a company that runs several power stations.
Power plants covering more than 350 GW of generation capacity are using SmartSignal solutions to improve their availability, profitability and environmental health and safety outcomes.
This article is based on the paper ‘SmartSignal EPI*Center: Implementation Options and Value Analysiss’ presented at POWER-GEN Europe 2007 in Madrid, Spain