New software from Power Technology is geared towards reducing NOx, CO2 and carbon-in-ash by automatically calculating optimized settings for coal fired boilers.

Gero Di Piazza

Stringent emission cutting policies, which is at the forefront of state targets around the world, have led to companies developing products to lower such nitrogen oxides and carbon oxide gasses. The Generic Neural Optimization Control Intelligent System (GNOCIS) Plus is the latest software package development from Powergen’s Power Technology division (now owned by Germany’s E.ON). The product is geared to reducing NOx and CO2 emissions by automatically calculating optimized settings for coal-fired boilers.

The GNOCIS Plus updated version was released in 2001, where it took on board experiences from the US and utilized them in plants in the UK. The data it stores is used to produce neural network maps from special model-building software developed by Power Technology. The software targets the key interactions between complex combustion settings, such as mill speeds, mill biasing and damper settings, and their effects.

It works by taking the key control parameters of a power station boiler, such as coal flows, excess oxygen (O2), burner tilts and load inputs and predicts the resultant emissions for NOx, boiler efficiency, and fly ash carbon levels.


Figure 1. A typical GNOCIS control display
Click here to enlarge image

null

The system can be used in advisory mode or in a direct closed-loop control mode for automatic optimal control. Operations staff can track the effect of the model using a GNOCIS Plus display screen in the control room. The on-line retraining capability allows changes resulting from different fuels and major outages to be accommodated without rebuilding the neural network maps. The software and screens are based on standard Windows functionality and appearance.

The benefits of GNOCIS Plus are site specific, but on average reductions in NOx of 10 – 15 per cent and efficiency improvements of 0.2 – 1 per cent can be expected. This on top of the estimated one year payback and enhanced ash sales contribute to a plant’s annual turnover.

Development

The initial testing of the GNOCIS Plus system was conducted during May 2001 on Unit 1 of the Greene County power plant in the US. Several tests were performed to investigate performance with respect to NOx reduction and efficiency improvement capabilities. It was installed as an enhancement to an earlier version of the software that was originally installed in 2000.

The Department of Energy’s (DoE) Clean Coal Technology Programme helped pioneer the use of artificial intelligence (computer-based decision-making that mimics the human brain) to optimize key coal utilization technologies. James Longanbachl, project manager, Coal Power Projects Division at the DoE said: “I think there is potential in the approach… to get additional efficiency and lower emissions at a small marginal cost. The utility can also minimize costs at the individual plant and perhaps eventually over their whole system.”

GNOCIS Plus is an enhancement to a plant’s digital control system targeted at improving utility boiler efficiency and reducing emissions. The GNOCIS development was funded by a consortium consisting of EPRI, Powergen, Southern Company, URS Corp. (which is licensed to supply GNOCIS Plus in North America), UK Department of Trade and Industry, and the US DoE. The team feel confident with this backing along with their history in consultancy. Roger Brandwood, an engineer at Power Technology said: “We are an engineering consultancy and not just a software house. This accumulated experience in plant optimization provides us with an in depth knowledge of the combustion process. We feel we are one of the few consultancies that can provide this amount of expertise in risk-based asset management in the utility sector.”

Technology

The GNOCIS, developed in the mid-1990s, was one of the first applications of neural network technology for real-world power plant applications in the mid 1990s. The revised model is GNOCIS Plus whichconsists of enhanced online learning capability, offline ‘what if’ analysis capability, and an optimiser performance that provides the functionality to recognise and compensate for changes in plant condition.

The system has several advanced features including:

  • Online retraining feature
  • Choice of optimization parameters
  • Robust optimization algorithms
  • Accommodates plant constraints
  • Flexibility in being configured for any plant design

It gives the user the option to be used in either open loop or closed loop control modes. The former will provide advisory information for the plant operators, requiring a response. The latter, will carry out functions automatically with no response required from the operator.


Figure 2. An overview of how the GNOCIS system operates
Click here to enlarge image

null

The online retraining feature means that the product will continually adapt to plant changes without the need to periodically rebuild models. It can be applied to optimize dust emission, to selective catalytic reduction (SCR) plant for the minimization of ammonia use, and to flue gas desulphurization plants.

Five days training will be provided for senior operators, enabling them to train other members of their staff as required. The installation process can be completed within a year. Several power plants across Europe have installed the GNOCIS Plus system. Italy’s Enel has installed the system at its 320 MW Vado Ligure power station. Powergen’s own 2000 MW Kingsnorth plant in the UK has achieved results that reveal a four per cent point reduction in loss on ignition (LOI) and a ten per cent reduction in NOx. GNOCIS Plus is also, installed at AEP’s 2000 MW Ferrybridge and Fiddler’s Ferry power plants and West Burton, owned by London Electricity.

Also, ‘start-up’ optimization using GNOCIS Plus is being developed to minimize plant damage and operator varia-bility under a multi-utility European programme.

West Burton

The West Burton, 4 x 500 MWe plant, ultimately owned by France’s state-owned Electricité de France (EDF), has installed GNOCIS Plus on Unit 2. The initial project began in 1999 as part of a demonstration of the technology by Powergen to TXU Europe, the plant’s previous owners. The project was to assess it with a view to deciding if it should be installed across all units. West Burton has gained sufficient confidence in GNOCIS Plus to purchase a software licence from Powergen to keep the system on the trial Unit. Further development is currently ongoing aimed at making the system more robust and user friendly for the operators.

West Burton has a closed loop installation including online retraining. This feature is particularly important as it allows for new operational data to be included in the optimizer – essential when there are thousands of possible set ups (some of which may occur only very infrequently). So far, GNOCIS Plus has proven successful during commissioning tests. Phil Ray, the performance engineer at West Burton said: “The system has shown good potential. We have seen an improvement of five to ten per cent in NOx reduction, that’s why we are continuing with it. The main issue we need to improve is its robustness.” Ray explained that currrently an engineer monitors the functioning of the system and takes GNOCIS Plus off-line in case of problems. A self-monitoring function will shortly be added to alert the operator to system problems. He notes that there are other areas that he would like to see slight improvements in. “For example, the target function, which GNOCIS aims to optimise, is a balance between NOx reduction and carbon in fly ash reduction. Currently this is coded into the software. We will shortly give the control of this directly to the operator through his control screen to enable him to use GNOCIS Plus to its best advantage.” Once trials are concluded, West Burton will decide whether to install the system on the remaining units.