Statistic evaluation of deviation between guaranteed and measured turbine efficiency

Transcription

1 Statistic evaluation of deviation between guaranteed and measured turbine efficiency Petr Ševčík Hydro Power Group Leading Engineer OSC, a.s. Brno, Czech Republic

2 Verification of Gibson method Discharge calculation by other program (by another independent expert) only comparison of algorithms Comparison with other primary method (current meters, thermodynamic method etc.) expensive, seldom Measurement on reversible turbine measurement in both operating directions reduce significantly systematic error comparison of whole measurement - seldom Statistic evaluation of deviation between guaranteed and measured turbine efficiency tests sufficient number of tests

3 Principle of measurement procedure used. Separated pressure records are used. Almost the whole penstock length is used as measuring section. The pressure sensors used are installed in exactly defined cross sections G and G. Measurement with free water upper level is used only for performance tests (not for guarantee measurement). Calculation algorithm according to standard IEC /99 is used (details see paper Verification of Gibson flow measurement from HYDRO 9). Measurement may be applied on various penstock s layouts (straight, curved, sections with different diameters, etc.) 7. There are no corrections for curved penstock applied 8. Turbine unit was always stopped by emergency shut down or by similar procedure 9. Leakage through guide vane evaluation is based on Gibson method if intake valve is installed or by other exact method (e.g. level drop behind stop log). Combination of Gibson and index flow measurement is always used

4 List of measured units See Tab in paper Statistic evaluation of deviation between guaranteed and measured turbine efficiency Brief overview: Only guarantee measurements 7 measured units, therefrom in pumping mode Penstock length m /.7 D Straight / nearly straight penstocks Penstocks with, or more bends Deviation between measured efficiency and guaranteed efficiency: η η = M η η Different types of guaranteed efficiency: Weighted efficiency Mean value from defined number of points Guarantee in one working point Guaranteed performance improvement M G η M = measured efficiency η G = guaranteed efficiency

5 Statistic evaluation - Gibson % % Occurence % % % % Deviation (η M - η G )/η M [%] Deviations were sorted into classes with bandwidth. % Similar to normal distribution with centre -.%

6 Curiosities explanation % Occurence % % % % % Deviation (η M - η G )/η M [%] Lower efficiency value for this case was confirmed by other independent measuring team using different method... In this case was guaranteed efficiency improvement for upgraded turbine, where turbine condition before upgrade was very bad. The efficiency improvement was significantly higher than expected value

7 Curiosities explanation % % Occurence % % % % Deviation (η M - η G )/η M [%] Excellent correlation between Gibson and ultrasonic transient time method ( Q =.%). One path ultrasonic flowmeter, D in front of measuring section, D after measuring section Flowmeter 7

8 Curiosities explanation Occurence % % % % % Reversible Francis turbine, pump mode Reversible Francis turbine, turbine mode % Deviation (η M - η G )/η M [%]... units, all of them measured in booth modes with similar results Measuring section for Gibson method 8

9 Curiosities explanation Double curved penstock Profile G Profile G % % Occurence % % % % Deviation (η M - η G )/η M [%] 9

10 Curiosities explanation Occurence % % % % % Curved penstock with differential penstock protection using ultrasonic measurement in two cross-sections. Excellent correlation with the ultrasonic measurement (four path) % Deviation (η M - η G )/η M [%]... y =.999x R =.9998 QUS [m /s] 8 y =.99x -.89 R =.9998 Qus - four paths, beginning of penstock Qus - one path, end of penstock 8 Q G [m /s]

11 Current meters % % Occurrence % % % % deviation (η M - η G )/η M [%].... Set of 8 acceptance test results was used for this evaluation Main part of measurements was carried out at low head HPPs The rate distribution corresponds very approximately with normal distribution with centre +.%

12 Comparison of Gibson x current meters % % Gibson Current meters Occurrence % % % % Deviation (η M - η G )/η M [%]

13 Conclusion Average deviation of measured efficiency from guaranteed efficiency value -.% corresponds very well with expected commercial efficiency increase often used for the bid. Deviation dispersion is done by measurement uncertainty and also by real deviation of measured efficiency against guaranteed one. Hydraulic design based on CFD method recently used is associated with improved reliability of guaranteed parameters. Outliers in dispersion graph were checked and fully explained. Very good reliability of Gibson flow measurement was confirmed as well as it s independency on penstock geometry. Influence of penstock irregularities (bends, cones etc.) on measured discharge accuracy wasn t confirmed. Gibson method was used as cost effective and easy to use flow measurement method with good accuracy. Very important is to verify provided data

14 OSC for Hydro Power Thank you for your attention Contact: Mr. Petr Ševčík Hydro Power Group Leading Engineer Division Control Systems OSC, a.s., Czech Republic phone: sevcikp@osc.cz