No abstract

No abstract

Knowledge acquisition has frequently been identified as the bottleneck in knowledge engineering6. Many techniques for knowledge acquisition have been developed, but this paper goes further to present a complete methodology. The use of general questions, specific questions, direct observation, the proper knowledge representation format, teachback techniques, and simulations form that methodology. An expert system created for advising the operators of a reactor on the condition of the reactor shows how each phase of the methodology can be applied to the knowledge acquisition process. BackgroundCreating an expert system that produces conclusions based on a known set of rules and facts is easy. Techniques and tools for storing rules and facts then doing inference using them are widespread and well known. The stage of expert system development that will make or break the end product is the knowledge acquisition. This paper discusses this aspect of expert system development by presenting a methodology for knowledge acquisition, showing how it was used in the development of an expert system assistant that is currently in field test on a fluid bed reactor, and how knowledge acquisition relates to other issues such as user interface design and design for maintenace.The chemical reactor for which the expert system was developed is a continuous process that is controlled by four shifts of control operators. The goal of the expert system is to help avoid upsets and produce a more uniform product. The inputs, process, and product created are proprietary, so they will not be discussed in this paper except in general terms. When a specific example is needed the names will be changed to generic terms. None of these restrictions affect the information presented in this paper. Application and ToolsThis section presents a brief introduction to the expert system that was created and to the expert system shell used to create it in order to provide an example for the discussion that follows. The expert system continually monitors sensor readings from the reactor and provides recommendalions on how to run the reactor according to the best practices that have been determined. Figure 1 shows how the expert system is integrated with other software. The expert system, everything in the dotted lines, reads sensor values from the database, performs inference on facts created from the sensor values, then writes any conclusions to the database. Any new recommendation added 196 /SPIE Vol. 2244 O-8194-1548-O/94/$6.OO Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/23/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

For cylindrical receivers with a capacity of about 400 MW/t, an aim-at-the-belt focusing strategy can produce average fluxes the order of 0.5 MW/m2 with peaks as high as 2 MW/m2. An absorber concept is described which uses liquid sodium coolant and a three-header configuration to efficiently capture this solar power. The mechanical design of this absorber is discussed and thermal performance estimates are presented showing the solar-capture efficiency over a range of solar intensities. The sodium-flow characteristics and some potential flow-control problems are also described. A thermal-stress analysis is presented which shows that a limiting factor on the flux capability may be tube-wall creep/fatigue failure and not the heat-transfer capability of sodium.

No abstract