Control of gasifier operations using neural networks, Task 7.11. Topical report, September 1992--December 1993 [electronic resource].
- Published
- Washington, D.C. : United States. Dept. of Energy, 1994.
Oak Ridge, Tenn. : Distributed by the Office of Scientific and Technical Information, U.S. Dept. of Energy. - Physical Description
- 64 pages : digital, PDF file
- Additional Creators
- University of North Dakota. Energy and Environmental Research Center, United States. Department of Energy, and United States. Department of Energy. Office of Scientific and Technical Information
Access Online
- Restrictions on Access
- Free-to-read Unrestricted online access
- Summary
- The incentive for this project was to investigate applicability of a new modeling tool -- a system of equations which emulate the human brain, called neural nets -- for chemical process modeling and control at the Dakota Gasification Company (DGC) coal gasification plant in Beulah, North Dakota. The specific goals were (1) to determine guidelines for identifying likely candidate processes for the use of neural nets, (2) to screen the various process areas of the Beulah plant, using these guidelines, (3) select a process area to study, and (4) apply neural nets to model that particular process area. The gasification area was the prominent candidate for the application of neural nets. Calculations showed that it was possible to save over $500,000 by increasing the steam-to-oxygen (S:O) ratio. The problem is that when the S:O ratio is increased, more gas liquor is produced, and there is a limit to how much gas liquor can be handled downstream. Thus, if a model were available to predict how much gas liquor would be produced, the S:O ratio could be increased to the operating limits. The performance of the neural net was apparently hindered by the type of data input used. Once the known effects of oxygen flow rate and S:O ratio were taken into account, there was no correlation left between any of the input data and the gas liquor flow rate. This result was verified using regression analysis, so there is likely no problem with the neutral net itself. Rather, the conclusion is that the data used did not give any information useful for predicting gas liquor flow. Therefore, in order to apply neural nets to go after the savings, more variables must be logged on a regular basis. The most important variable needed to continue this project is probably the volatiles content of the coal.
- Report Numbers
- E 1.99:doe/mc/10637--3742
doe/mc/10637--3742 - Subject(s)
- Other Subject(s)
- Note
- Published through SciTech Connect.
04/01/1994.
"doe/mc/10637--3742"
"DE94004116"
"400408000/AA0505000"
Rao, C.N.; Erjavec, J.; Cisney, S. - Type of Report and Period Covered Note
- Topical; 01/01/1992 - 12/31/1993
- Funding Information
- FC21-86MC10637
View MARC record | catkey: 13838823