DEGRADATION OF BOILER AND HEAT EXCHANGER MATERIALS: DATA GENERATION, DATABASES AND PREDICTIVE MODELLING

摘要

There is a need to generate and compile quantitative information on the degradation of boiler and heat exchanger materials in the combustion environments being encountered in power plants using new and/or dirtier fuels, as well as with components operating at higher metal temperatures. Within the European COST522 programme this topic is being addressed within the Plant Integration Group. The overall objective of the activity is to produce a database of materials performance and related exposure parameters for candidate materials in plant operating under novel environments and/or at higher metal temperatures. To generate the required data, the materials have been exposed in a range of combustion plant and simulated laboratory environments. The fuels used in the plants have included biomass (e.g. wood and straw), waste (e.g. RDF) and coal. The laboratory tests have been targeted at particular environments to investigate different specific degradation effects in more detail. The database is intended to facilitate the comparison of candidate materials and to enable some of their limitations to be identified, in terms of metal temperature and sensitivity to particular corrosive conditions. The database has been produced by NPL using data generated within the COST522 programme and available from the open literature. The most valuable data for inclusion in such a database are those reported in terms of metal loss (or metal loss distribution) and with a well characterised exposure environment. For the data to be readily incorporated into the database, it has been necessary to develop and apply standardised methods of data collection and presentation (in a spreadsheet). The preferred method of obtaining corrosion damage data has been by dimensional metrology on polished cross-sections, obtaining a distribution of damage measurements. The structured format of the data in the database is being used to produce models of materials performance as a function of environmental exposure parameters (e.g. metal temperature, SO{sub}x, HC1, exposure time). Both neural network and more conventional empirical modelling of these data are being investigated.