Affiliation:
1. GE Energy, 1 River Road, Schenectady, NY 12345
Abstract
The ultimate proof of the soundness and viability of a novel technology is a full-scale demonstration test in which actual components are run successfully over the entire operating envelope. Consequently, the collection of accurate and meaningful test data is of utmost importance to the success of the test. An analysis of such data will validate the original design concepts and will lead to paths of further improvement for the next generations thereof. Statistical fundamentals to determine the accuracy and precision of measured data are amply documented and readily available in well-established standards. The yardstick that should be used for the “meaningfulness” of the measured test data is the satisfaction of the fundamental laws of conservation. While it is known that the “true” values of the sensor data when inserted into the governing equations for the tested component will result in perfect balances, “actual” measured values will always result in “imbalances.” Therefore, reconciliation of the individual measurements with the governing conservation equations is a must prior to the actual analysis of the data. Reconciliation in this context is an estimation of the true values of the sensor data from the actual sensor data by using statistical concepts. This paper describes the development of a data reconciliation concept that is universally applicable to any process or power plant system where sensor data are used. The usefulness and power of the technique are demonstrated by its application to a single-shaft combined cycle with both gas turbine and steam turbine driving a common generator. In the absence of a reliable and accurate measuring system to individually determine gas and steam turbine shaft outputs, data reconciliation is vital to an accurate analysis of the data.
Subject
Mechanical Engineering,Energy Engineering and Power Technology,Aerospace Engineering,Fuel Technology,Nuclear Energy and Engineering
Cited by
18 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献