TRANSIENT OPERATION

Page visibility: Public
License:         CC BY-NC-SA 4.0

CONTENTS

Organizing committee

1. Torbjørn Kristian Nielsen (NTNU), Chair
2. Michel Cervantes (LTU & NTNU), Co-chair
3. Ole Gunnar Dahlhaug (NTNU)
4. Chirag Trivedi (NTNU)
5. Carl W. Bergan(NTNU)
6. Rahul Goyal (IITR and LTU)

Background

Accurate numerical simulations of hydraulic turbines are challenging, time consuming and demand large computational resources. There is a need for optimizing the numerical techniques/procedures. The Francis-99 workshop series provides an open platform to industrial and academic researchers to enhance their capability in computational fluid dynamic techniques applied to simulate the hydraulic turbines. The second workshop of Francis-99 will continue with the same spirit as the first workshop. Detailed experimental studies have been conducted at different operating conditions. In addition, some experiments based on input received from the participant of the first workshop have been conducted. Both, steady state and transient, measurements are performed. Three steady state operating points are investigated: part load, best efficiency point, and high load. For the transient conditions, load variation and start-stop are investigated. Both pressure and velocity measurements are available for the detailed validation and investigation of the numerical model.

Scope

The second Francis-99 workshop is a continuation of the first workshop with additional challenges. It aims to further determine the state of the art in simulation of the Francis-99 model under steady state and transient operating conditions. Transient operation of the hydraulic turbines is increasing with the apparition of renewable energies, affecting significantly their life [1]. The motivations are multiple and reside in the availability of more powerful computers allowing the use of more advance turbulence models and accurate numerical schemes. Methods allowing uncertainty quantification of the simulations as well as faster result such as non-linear harmonic decomposition are appearing and need to be assessed.

The numerical results will be compared to time dependent experimental pressure and velocity measurements. Experimental data of pressure and velocity measurements are available on the website. Comparison of the experimental and numerical results will be performed by the participants themselves.

Expected contribution

The organizers expect from the participants to investigate one or several of the following topics with the provided numerical model:

• Steady state operation simulation (numerical and modelling influence).
  
• Transient operation simulation (numerical and modelling influence).
  
• Rotor-stator interaction, influence of the operating conditions.
  
• Rotating vortex rope simulation.
  
• Uncertainty quantification in numerical simulation.

Test case: Francis-99 turbine

Experiments have been conducted on the Francis-99 turbine model used for the first Francis-99 workshop. Both steady state (constant guide vane angle) and transient (time dependent guide vane angle) measurements were performed. Three operating points were selected for the steady state measurements; part load (PL), best efficiency point (BEP), and high load (HL). These operating points are different from the first workshop. The transient measurements include load acceptance from PL to BEP, load reduction from BEP to PL, turbine startup and shutdown. Pressure and velocity data at different locations of the turbine were acquired simultaneously during the measurements. Several repetitions of the measurements have been carried out to estimate the random uncertainty in the measurements. Below you will find:

• Locations of pressure and velocity measurements in the turbine which are same for all the operating points.
• Steady state test case which describes the investigated operating points and corresponding experimental data for numerical validation.
• Transient test case which describes the investigated operating conditions and corresponding experimental data for numerical validation. Additional data on variation of head, discharge, torque, runner angular speed, and guide vane movement in time domain are provided. This information will be needed to setup numerical model for the load variation and start-stop.

Locations of pressure and velocity measurements

For the pressure measurements, a pressure sensor VL2 was mounted at the vaneless space, and three pressure sensors DT5 and DT6 were mounted at the draft tube cone. Exact locations of the sensors and observed uncertainties are provided in Table 1. The reference coordinate system is shown in Figure 1 and Figure 2. The pressure data along with other flow variables were acquired at the sampling rate of 5 kHz and velocity data were acquired at the sampling rate of 40 Hz.

Table 1. Locations and uncertainty of pressure measuring sensors

Table 2. Locations of velocity measurement in the draft tube. Figure 2 shows the location of velocity measurement in the draft tube cone. Line-1 and line-2 are horizontal lines. Line-3 is the vertical line in the draft tube cone.

Figure 1. Global coordinates for the measurement locations, geometry, and mesh.

Figure 2. Sections of PIV measurement in the draft tube cone. Dimensions are not scaled and may not exactly match with with the dimensions in table above.

Steady state measurements

Table 3 shows the acquired flow parameters during PL, BEP and HL operating points. It should be noted that the uncertainties provided in the table are correspond to BEP condition. The uncertainty in hydraulic efficiency is the total uncertainty. This includes random and systematic uncertainty in the basic parameters used to compute hydraulic efficiency. Procedure available in IEC 60193 was followed to estimate the uncertainty and hydraulic efficiency. Cross sectional areas at the inlet and outlet pressure measurement are 0.0962 m² and 0.236 m², respectively. The outlet cross sectional area corresponds to the experimental pressure measurement section which is located at 1.58 m before the actual draft tube outlet in the numerical model. The height difference (z) between inlet and outlet pressure measurement is 1.0715 m. Uncertainty in the area measurements is ± 0.1%. Hydraulic torque output from the runner is the sum of toque to the generator and friction torque.

Table 3. Acquired flow parameters and other quantities during steady state measurements.

Net head is calculated using standard Bernoulli's equation

Transient measurements

Total four transient conditions have been investigated experimentally and several repetitions were performed at the same conditions to investigate the random uncertainty in the acquired data. The four transient conditions are:

1. Load acceptance: Increase turbine output power from BEP to HL by opening the guide vanes from 9.84°, i.e., increasing discharge.
2. Load reduction: Decrease turbine output power from BEP to PL by closing the guide vanes from 9.84°, i.e., decreasing discharge.
3. Turbine startup: Guide vane opening from 0.8° to BEP.
4. Turbine shutdown: Guide vane closing from BEP angle to 0.8°.

During the transient conditions the runner angular speed was constant, i.e., 333 rpm. For the load variation runner was operating at 333 rpm. For the startup, 333 rpm of the runner was achieved at 0.8° and it was synchronized to the corresponding load. During the shutdown, when guide vanes reached to 0.8°, the generator was decoupled from the load and set to speed-no-load condition. Data acquired during guide vane movement between 0.8° and 9.84° are provided. As a matter of fact, air bubbles were present in the model below 0.8° and the PIV data were inaccurate. Flow parameters observed at the load coupling/decoupling point are shown in Table 4. Further, Figure 3 and Figure 4 show variation of head discharge, torque and guide vane angle during the transient conditions. The minimum and maximum values are scaled between 0 and 1 to extract the trend of the corresponding parameters during the transients. At time t = 1 s, the guide vanes were set to open/close to perform transient conditions. After certain time, the guide vanes reached to set value of angle (i.e., 6.72° or 12.43°) then steady state condition was followed. Attached data provides all necessary information required to perform numerical simulations. Discharge values during the transient conditions of load acceptance, rejection and start-stop are not accurate since the flowmeter response time was low. However, discharge variation was the function of guide vane movement. The guide vanes movement was straight line therefore, initial and final discharge values can be taken from the steady state data at the corresponding guide vane position/ operating point and the linear variation can be considered. For the transient measurements, there is some delay in the flowmeter and inlet pressure transducer. Therefore, these values will not match exactly with the start/stop time of the guide vanes.

Table 4. Flow parameters observed at the load coupling/decoupling point during the start-stop conditions.

Figure 3. Variation of global parameters during load change from BEP to PL.

Figure 4. Variation of global parameters during load change from BEP to HL.

Numerical model

Geometry and mesh of the investigated model Francis turbine are provided (can be downloaded from Table 5) to perform numerical simulations. You are also most welcome to make your own mesh. The provided numerical model includes spiral casing, 14 stay vanes, 28 guide vanes, 15 blades, 15 splitters, and a draft tube. The global coordinate system is the same as the one provided for the experimental data (Figure 1). Geometry files are available in legacy format (igs and parasolid). The guide vane mesh is provided for the three steady operating points, i.e., PL, BEP and HL. The provided meshes are for high Reynolds number therefore y+ value would be above 30 at BEP. Mesh in the spiral casing is hybrid including 3 million cells, 65% hexahedral, 27% prism, 4.5% pyramid, and 3.5% tetra. Meshes in the other components are purely hexahedral or hex dominated. Mesh nodes in the distributor and runner passage are 0.3 and 0.8 million, respectively. Mesh nodes in the complete draft tube are one million. Maximum aspect ratio, maximum expansion ratio and minimum angle of mesh in the runner are 56.3, 1.7 and 36.8, respectively. The provided meshes are tested at BEP using ANSYS CFX and OpenFOAM solvers, and for all operating points using NUMECA solvers. However, not all the configurations are tested. Mesh in the distributor domain may be used to perform transient simulations by mesh deformation technique as long as solver supports.

Figure 5. Three-dimensional model of Frncis-99 turbine.

Author centre

Important dates

Last date for the abstract submission is April 1, 2016.

Accepted abstract will be notified on April 15, 2016.

Last date for the submission of full length paper is September 15, 2016.

Notification of the finally accepted full length papers is October 5, 2016.

Last date for the submission of final manuscript is November 10, 2016.

The second Francis-99 workshop will be organized on 14-15 December 2016.

Abstract and manuscript submission

Abstract and full length paper submission guidelines

Abstract and full length paper will be sent to Michel Cervantes.

Formatting the abstract: Submission file format should be pdf. Maximum numbers of words in abstract are 500. Maximum length of the paper title is 25 words, no special character, only "dash" can be used, if needed. The paper title should include a term "Francis-99" in order to link all Francis-99 papers with single keyword for citation.

Formatting the full length final paper: Please follow the guideline available for Journal of Physics: Conference Series.

Registration

The registration to the workshop is made by sending the registration form below to Debbie Koreman (debbie.w.koreman@ntnu.no , tel. +47-73593561) and paying the required fees. The registration fees include 25% VAT and are as follow:

- Author(s): Free

- MSc/PhD student who wants to attend the presentation sessions: Free.

- NTNU and LTU staff: Free.

- Participants without paper: 3500 NOK.

- Late Registration (01-14 December 2016): 4500 NOK.

- Registered participants that do not attend the workshop: 2000 NOK.

Registration is compulsory for all participants. It must be done before 1 December 2016.

Cancellations of the registration to the workshop must be made within December 7.

Venue

The workshop will be held at the Norwegian University of Science and Technology, Trondheim, Norway. The address is: Electric Power Engineering Building (room EL6), OS Bragstads plass 2E, N-7034 Trondheim. Map. This is preliminary.

Accommodation

A large number of hotels are available around the University (BEST WESTERN Chesterfield Rica Bakklandet, Clarion Grand Olav Clarion Hotel & Confress Trondheim, Comfort Hotel Park, Scandic Lerkendal, City Living, Trondheim Youth hostels ...).

Weather forecast

Francis-99 (II) workshop program

Wednesday 14/12-2016

1000-1130: Registration

1130-1230: Lunch

1230-1240: Prof. T. Nielsen: Introduction

1240-1310: C. Bergan: Francis-99 test case: experiments

1310-1340: R. Lestriez: Francis-99 test case: mesh

1340-1410: P. Mossinger, R. Jester-Zurker, A. Jung: Francis-99: Transient CFD simulation of load changes and turbine shutdown in a model sized high-head Francis turbine.

1410-1430: Coffee break

1430-1500: K-R Jakobsen, M. Aasved Holst: CFD simulations of transient load change on a high head Francis turbine

1500-1530: Y. Dewan, C. Custer, A. Ivashchenko: Simulation of the Francis-99 hydro turbine during steady and transient operation

1530-1600: A. Minakov, A. Sentyabov, D. Platonov: Numerical investigation of flow structure and pressure pulsation in the Francis-99 turbine during startup

1600-1630: A. Minakov, D. Platonov, A. Sentyabov, A. Gavrilov: Francis-99 turbine numerical flow simulation of steady state operation using RANS and RANS/LES turbulence model

1610-1630: A. Gavrilov, A. Dekterev, A. Minakov, D. Platonov, A. Sentyabov: Steady state operation simulation of the Francis-99 turbine by means of advanced turbulence models

1630-1700: E. Casartelli, L. Mangani, O. Ryan, A. Del Rio: Performance prediction of the high head Francis-99 turbine for steady operation points

1700-1900: (free time)

1900-2100: Dinner

Thursday 15/12-2014

0830-0900: Y. Zeng, L.X. Zhang, J.P. Guo, Y.K. Guo, Q.L. Pan, J. Qian: Efficiency limit factor analysis for the Francis-99 hydraulic turbine

0900-0930: N. Tonello, Y. Eude, B. de Laage de Meux, M. Ferrand : Frozen rotor and sliding mesh models applied to the 3D simulation of the Francis-99 Tokke turbine with Code_Saturne

0930-1000: P.Østby, J.T. Billdal, B. Haugen, O-G. Dahlhaug: On the relation between friction losses and pressure pulsations caused by rotor-stator interaction on the Francis-99 turbine

1000-1015: Coffee break

1015-1140: Group discussion

1140-1155: Prof. Ole-Gunnar Dahlhaug: Francis-99 III

1155-1200: Prof. T. Nielsen: Conclusion

1200-1300: Lunch

1300-1500: Visit Water Power Laboratory

Workshop summary

The second Francis-99 workshop was organized on 14 and 15 December 2016. Total 10 research papers were presented in the workshop. Around 35 researchers from different countries were participated in the workshop. Both steady and transient simulations were presented and discussed during the workshop. For the transient simulation, mesh deformation approach was used. Key concern was long simulation time during load variation and start-stop. There was alternative suggestion to use 1D-3D coupling that allow reduction of time. Main components of the turbines can be modelled using 3D and data from 1D analysis should be used as boundary condition. Further, size of the 3D CFD domain can be reduced by using passage modelling approaches as highlighted in the first workshop. Results presented from the hydropower industries indicated that modern passage-modelling techniques should be used which allow rough estimation of pressure loading on the blades during load variation and start-stop. Results presented using different techniques/solvers such as Ansys, star-ccm+, OpenFOAM have shown that newly implemented techniques can help to overcome the challenges in transient modelling and can provide results with improved accuracy.

Input to the third workshop

Simulation of entire turbine requires huge effort and time. To minimize the expensive simulation and accommodate fundamental research, participants have suggested two categories for the next workshop. Fluid-structure interaction in turbine and hydrofoil. Measurements on hydrofoil enable basic understanding between water and blade like material.

Publications

The second Francis-99 workshop was held on 14 and 15 December 2016. Total 10 research paper were presented in the workshop. All the papers were peer-reviewed before the publication in the workshop. The published papers in the workshop are available online at Journal of Physics: Conference Series. and the papers are freely available.

1. Peter Mössinger, Roland Jester-Zürker and Alexander Jung, 2017, "Francis-99: Transient CFD simulation of load changes and turbine shutdown in a model sized high-head Francis turbine," Journal of Physics: Conference Series, 782(1), p. 012001. doi:10.1088/1742-6596/782/1/012001.
2. Ken-Robert G. Jakobsen and Martin Aasved Holst, 2017, "CFD simulations of transient load change on a high head Francis turbine," Journal of Physics: Conference Series, 782(1), p. 012002. doi:10.1088/1742-6596/782/1/012002.
3. Yuvraj Dewan, Chad Custer and Artem Ivashchenko, 2017, "Simulation of the Francis-99 Hydro Turbine During Steady and Transient Operation," Journal of Physics: Conference Series, 782(1), p. 012003. doi:10.1088/1742-6596/782/1/012003.
4. A Minakov, A Sentyabov and D Platonov, 2017, "Numerical investigation of flow structure and pressure pulsation in the Francis-99 turbine during startup," Journal of Physics: Conference Series, 782(1), p. 012004. doi:10.1088/1742-6596/782/1/012004.
5. A Minakov, D Platonov, A Sentyabov and A Gavrilov, 2017, "Francis-99 turbine numerical flow simulation of steady state operation using RANS and RANS/LES turbulence model," Journal of Physics: Conference Series, 782(1), p. 012005. doi:10.1088/1742-6596/782/1/012005.
6. A Gavrilov, A Dekterev, A Minakov, D Platonov and A Sentyabov, 2017, "Steady state operation simulation of the Francis-99 turbine by means of advanced turbulence models," Journal of Physics: Conference Series, 782(1), p. 012006. doi:10.1088/1742-6596/782/1/012006.
7. E. Casartelli, L. Mangani, O. Ryan and A. Del Rio, 2017, "Performance prediction of the high head Francis-99 turbine for steady operation points," Journal of Physics: Conference Series, 782(1), p. 012007. doi:10.1088/1742-6596/782/1/012007.
8. Y Zeng, L X Zhang, J P Guo, Y K Guo, Q L Pan and J Qian, 2017, "Efficiency limit factor analysis for the Francis-99 hydraulic turbine," Journal of Physics: Conference Series, 782(1), p. 012008. doi:10.1088/1742-6596/782/1/012008.
9. N Tonello, Y Eude, B de Laage de Meux and M Ferrand, 2017, "Frozen Rotor and Sliding Mesh Models Applied to the 3D Simulation of the Francis-99 Tokke Turbine with Code_Saturne," Journal of Physics: Conference Series, 782(1), p. 012009. doi:10.1088/1742-6596/782/1/012009.
10. Petter T.K. Østby, Jan Tore Billdal, Bjørn Haugen and Ole Gunnar Dahlhaug, 2017, "On the relation between friction losses and pressure pulsations caused by Rotor Stator interaction on the Francis-99 turbine," Journal of Physics: Conference Series, 782(1), p. 012010. doi:10.1088/1742-6596/782/1/012010.

Group photograph

Acknowledgement

The workshop organizers would like to acknowledge the contribution made by Rémi Lestriez (NUMIBERICA) for providing the mesh; Jean-David Buron (Laval University, Canada) and Einar Agnalt (NTNU) for creating geometry of the Francis turbine and providing in different format; Jonathan Nicolle (Hydro-Québec, Canada) and Sebastien Houde (Laval University, Canada) for necessary valuable input during the geometry and mesh creation; Professor Håkan Nilsson (Chalmers University of Technology, Sweden) for his valuable inputs during OpenFOAM compatible mesh creation.

Download data

 Table 5. Click on the corresponding hyperlink to download the file.

Mesh specifications and available formats

Following nomenclatures shall be followed to understand the table:

SC+SV: Mesh in the entire spiral casing with 14 stay vanes.

GV_PL: Mesh in a guide vane passage at PL angle.

GV_BEP: Mesh in a guide vane passage at BEP angle.

GV_HL: Mesh in a guide vane passage at HL angle.

RU: Mesh in a runner blade passage.

DT: Mesh in the complete draft tube.

1SV+2GV: Mesh in one stay vane passage and two guide vane passage.

RU+Cone: Mesh in a runner blade passage with extended cone.

DT_uns: Unstructured mesh in the draft tube.

AUTOMESH files are provided for NUMECA software users, if any.

*.msh and *.cgns both formats are supported by ANSYS CFX and FLUENT.

Mesh for the OpenFOAM code is tested. Moreover, input files for OpenFOAM extend 3.2 are available in the same folder with example case.

@article{trivediEffectTransientsFrancis2013,   title = {Effect of transients on Francis turbine runner life: A review},   author = {Trivedi, Chirag and Gandhi, Bhupendra K and Cervantes, Michel J},   year = {2013},   journal = {Journal of Hydraulic Research},   volume = {51},   number = {2},   pages = {121--132},   url = {https://doi.org/10.1080/00221686.2012.732971} }