Online Flowrate Monitoring Experiences at Hydro-Québec Jonathan Nicolle & Gilles Proulx June 2012 9 th conference of the International Group on Hydraulic Efficiency Measurements Trondheim, Norway
Summary > Intro > Results with a 200 MW Francis > Results with a 110 MW Propeller > Details of the experimental setup > Conclusions 2 Groupe Technologie, Hydro-Québec
Intro > Hydro-Québec the largest hydro-utility in the world Government owned Operates 350 hydro turbines in 60 plants Located in Québec, Canada mostly Francis turbines, high flowrate + medium head > Dedicated division for flow and efficiency measurement (ESP) > Research center (Ireq) Electricity production, transport and distribution CFD expertise 3 Groupe Technologie, Hydro-Québec
Intro > Research project: dream of online monitoring system No such thing as too much flow measurements Information is the key Accurate, affordable, usable for various types of unit > Allow for better business decisions at various levels Daily production decision (which unit, which plant should I use?) How do I operate when the head changes? Reservoir management, river spate, how do I avoid spillway? When should I clean the trash rack? What kind of losses are introduced by that manifold/ collector? Is a cavitation repair justified based on efficiency losses? Does aeration, adjacent groups, tail water level, etc, have an impact??? 4 Groupe Technologie, Hydro-Québec
Intro > For existing powerhouse flow monitoring, index seems like the best (only) choice > Q? Why is it so difficult to use index testing with low head plants? > 2 problems from CFD investigation Signal amplitude problem: sensitivity to adjacent group operation Flow homology problem: the flow distribution change with the opening of the guide vane (streamlines are not the same so should not use an index) > WK coefficients are not constants 5 Groupe Technologie, Hydro-Québec
Intro > Our solution: angular index First presented at 8th IGHEM conference in India Pressure behind the guide vane Measure of the guide vane angle Experimental calibration from a curve Q K( opening)* P 0.5 Coefficient 300 250 200 150 100 50 WK p1-p3 WK p2-p3 Psup - p1 Pinf - p1 Psup - pinf R 2 = 0.9999 R 2 = 1 0 25 30 35 40 45 50 55 Opening angle best one Q is predicted for a given angle and P combination 6 Groupe Technologie, Hydro-Québec
Results for a 200 MW Francis > Long penstock > 8 paths acoustic time travel (ATT) flowmeter as a reference > Index Upstream: probes at the spiral casing inlet Downstream: 4 probes behind the guide vane 7 Groupe Technologie, Hydro-Québec
Results for a 200 MW Francis > Calibration curve smooth Kop = Q ref / P 0.5 2 nd order polynomial fit was enough 8 Groupe Technologie, Hydro-Québec
Results for a 200 MW Francis > 10 days of monitoring > Good news: 2 independant signals almost identical Normalised Normalised flowrate flowrate 130 130 125 125 120 120 115 115 110 110 105 105 100 100 95 95 Angular Index Sonic Angular Index Sonic 90 90 7 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 8 0 1 2 3 4 5 6 7 8 9 10 9 Groupe Technologie, Hydro-Québec Days Days
Results for a 200 MW Francis > A surprise After some time, a constant offset ( 1.2%) was detected between the relative and the reference method Automatic correction of the ATT method Angular index unaffected and lower dispersion 120 Flow ± 1 m 3 /s 120 Flow ± 0.5 m 3 /s Normalised flowrate 115 110 105 100 RMS 0.5 % Sonic series 1 Sonic series 2 Normalised flowrate 115 110 105 100 RMS 0.2 % angular index 95 22 23 24 25 26 27 28 29 30 Opening [deg] 10 Groupe Technologie, Hydro-Québec 95 22 23 24 25 26 27 28 29 30 Opening [deg]
Results for a 110 MW propeller > 4 pressure probes located between the guide vane and the runner. External WK for high pressure. > Semi-spiral casing > Again smooth calibration curve based on previous current meters measurements 70 65 Propeller index quadratic P 60 55 K coefficient 50 45 40 35 30 11 Groupe Technologie, Hydro-Québec 25 30 32 34 36 38 40 42 44 46 48 Angle [deg]
Results for a 110 MW propeller > Able to follow head change Index results taken close to the expected head Measurements are from a similar group (normalized) 1.1 angular index 25.5 m Current meters 25.5 angular index 27.5 m Current meters 27.5m Angular index 30 m Current meters 30 m 1.05 Qnorm 1 Flow ± 1 m 3 /s RMS 0.2-0.3 % 0.95 0.9 750 770 790 810 830 850 870 890 Opening [mm] 12 Groupe Technologie, Hydro-Québec
Results for a 110 MW propeller > Online monitoring: a real story Comparison between «optimal» (manual adjustments) and 2 standard automatic procedures Interface showing last 3 hours 13 Groupe Technologie, Hydro-Québec
Experimental setup > Pressure probes Must be able able to take the full head when the guide vanes are closed Averaged with a manifold 4 probes are enough for Francis, ok for propeller Two words: Automatic Bleed. > Guide vane Angular encoder is preferred to servomotor strokes before after > Signal acquisition Filter when the guide vanes move 14 Groupe Technologie, Hydro-Québec
Conclusion > Until now, not much attention has been given to online monitoring Pursue different goals than absolute measurements (complementary) More interesting & difficult for low head plants Demonstrated angular index usefulness for 2 very different turbines (random error 0.2-0.3 %) Captured phenomena that would be difficult to observe otherwise ATT flowmeter recalibration Automatic PFR procedures impact on efficiency What s next > Building a new experimental set More industrial Fully numerical signal Bleed valves controller > Starting to integrate flow data to the various HQ systems 15 Groupe Technologie, Hydro-Québec
If you think angular index is a good system, try it or contact us!!! 16 Groupe Technologie, Hydro-Québec
Good questions you might ask Why use an angular encoder instead of servomotor stroke? What happened with the ATT flowmeter on the Francis? What about CFD and experimental comparisons? Is the calibration transposable from one group to another? What if the turbine is replaced? What are the pitfalls for angular index? 17 Groupe Technologie, Hydro-Québec
Why prefer angular encoder > Servomotor strokes would work in theory > but as a CFD guy, I ve been struggling with servomotors stokes for my whole life Old powerhouse, modifications, extensions, not always on plan In-house system not enough accurate Pre-loading different for each unit Zero definition not always the same Servomotor is of value to the operator > Angular encoder is fresh start Accurate & return a value that has a meaning in terms of fluid flow Angular position is of value to the fluid flow analyst 18 Groupe Technologie, Hydro-Québec RETURN
ATT reference flowmeter > Not exactly sure Acquired the data and analysed them afterward > Seems like there was a problem with one of the path and the system recalibrated itself > Dont know if the flow value was good before or after the event > Important point is the random error or RMS value of both methods 19 Groupe Technologie, Hydro-Québec RETURN
CFD vs experiment > Not exactly there but not far either > Propeller Mean error 1.1% > Francis Mean error 0.9% 80 26 70 CFD exp 24 22 K coefficient 60 50 40 K Coefficient 20 18 16 30 20 25 30 35 40 45 50 55 Opening Angle CFD 14 exp 12 10 22 24 26 28 30 32 34 36 Opening Angle RETURN 20 Groupe Technologie, Hydro-Québec
Calibration generality > Question still open for debate > Is the calibration applicable to a similar unit? K would probably be very close but may be not identical > Is a new calibration curve necessary when the turbine change? Limited results from a kaplan turbine seem to suggest so Variation in the meridionnal flow distribution affects the index 21 Groupe Technologie, Hydro-Québec RETURN
Pitfalls > Be as accurate as you can for the guide vane opening, it is quite sensible > Calibration might not be valid for the whole operating range > If you use a low number (< 4) of downstream probes, you might have unwanted sensitivity > Bleed valves are essentials to survive de-watering of the unit and long term operation > It not just a change in technology, it requires a change in mentality, so plan data integration ahead. Flow values are of no use if they do not make their way through the compagny operating systems. Beware of admistrative inertia 22 Groupe Technologie, Hydro-Québec RETURN