Results as of November 2016

In October 2015 a new calculational round was defined using the New Mexico measurements as a basis. The conditions for this calculational round were more or less comparable to the conditions for which calculations have been performed in Mexnext-I:, i.e. aligned flow at approximately 10, 15 and 24 m/s but in the present cases clean airfoil data are used for the outer part where tripped airfoil data were used in Mexnext-I. Moreover some participants took into account the tower and nacelle geometry. A huge amount of results were processed leading to hundreds (presentations of) parameters from 23 codes from 12 participants. The results were discussed at the 2nd meeting held in January 2016 at NREL where some encouraging conclusions could be drawn:
First the agreement between calculated and measured flow field data turned out to be very good. Moreover the calculated loads were, generally in better agreement with the measurements than the agreement which was found in Mexnext-I where the calculations were evenly spread around the measurements opposite to the situation in Mexnext-I where all loads were over predicted in particular at the important outer part of the blade and at the design wind speed of 15 m/s. Still the spread in calculational results was considered too large and moreover the difference between calculated and measured results at 15 m/s, 60% span was considered too large too. Several concentrated actions took place in order to understand this spread and differences between calculations and measurements. This was supported by additional calculations. These new results were discussed in June 2016 at an intermediate Mexnext meeting as held prior to the meeting of the EU project AVATAR at the University of Glasgow at which many Mexnext participants were present anyhow and where other participants could attend by web-conference. Generally speaking the spread in results has been reduced indeed but still the agreement was not considered satisfying and another effort was undertaken which was discussed at the third Mexnext meeting at ONERA Paris France.

It should also be known that airfoil data at the appropriate conditions (in terms of Reynolds number and tripping) are available for two out of the three airfoils on the Mexico blade (ie. NACA 644418 and the DU 91-W2-250 airfoil) but not for the RISOE airfoil which is the airfoil where the differences between calculations and measurements are found. A proposal for financing a wind tunnel experiment to measure the 2D airfoil data (and pressure distributions) on the remaining RISOE airfoil will be made to the EU project IRPWIND. Hopefully this experiment will shed light on the puzzling discrepancy.

Moreover a 2nd calculational round, based on New Mexico measurements at yawed conditions has been defined. The first results of this round have been discussed at the 3rd meeting at ONERA France in November 2016. 

Some comparisons between CFD calculations and measurements are reported in  http://iopscience.iop.org/article/10.1088/1742-6596/753/2/022054 and a validation of the actuator line and disc techniques using the New MEXICO measurements can be found in http://iopscience.iop.org/article/10.1088/1742-6596/753/3/032026  where more publications on the comparisons are expected in the near future.

It is noted that the Mexnext-III consortium agreed to release the measurements which have been used as a basis for the Mexnext-III calculational round to third parties. Thereto the coordinator can be approached, see http://www.mexnext.org/contact/

Apart from the comparison between calculations and measurements concentrated research is performed on the subjects mentioned in http://www.mexnext.org/approach/

Some results from this research can be found in papers which were presented at the Science of Making Torque conference in October 2016, where more results will be presented in the future.

1. Rotor experiments in controlled conditions continued: New Mexico http://iopscience.iop.org/article/10.1088/1742-6596/753/2/022004 a paper which describes the New Mexico experiment,
2. Momentum considerations on the New MEXICO experiment  http://iopscience.iop.org/article/10.1088/1742-6596/753/7/072001 a paper on the Momentum Balance from the New Mexico measurements of loads and velocities
3. Comparison of simulations on the NewMexico rotor operating in pitch fault conditions http://iopscience.iop.org/article/10.1088/1742-6596/753/2/022049 an article which analyses the New Mexico measurements at fault pitch conditions supported by calculations 

Results/Status as of September 2015

The first analysis of New Mexico results has been reported in http://www.ecn.nl/docs/library/report/2014/e14048.pdf

Mexnext-II is finished and reported in https://www.ecn.nl/publicaties/ECN-E--14-060 

In October 2014 the IEA ExCo approved an extension of Mexnext: i.e. Mexnext-III. The kick off meeting of that project has been held already in March 2015. The main aim of Mexnext-III is to analyse the measurements from the New Mexico experiment (not forgetting the ‘other than Mexico’ measurements).

Thereto the project starts with a very detailed quality check of New Mexico data. Although several data have been checked already in http://www.ecn.nl/docs/library/report/2014/e14048.pdf

Much work was needed to check the remaining data. Thereafter the measurements have been provided to the Mexnext project group in August 2015 which enables the start of the data analysis. Parallel to that a new calculational round is carried out based on New Mexico data. This calculational round will be defined in October 2015 after which the Mexnext participants start doing calculations.
Several  ‘other than Mexico data’  have and will be provided, e.g. measurements and the underlying information on the experiment carried out by FFA in the large CWEA tunnel in the end of the 1980’s.

Also interesting are PIV data on a 1/8 scaled model of the NREL Phase VI turbine which have been provided by CWEA/CARDC  in December 2014. These data will be compared  to  the original data from the NREL Phase VI experiment which were  heavily analysed in IEA Task 20.

The next plenary meeting of Mexnext will most likely be held in January 2015 at NREL USA. The meeting will be held in combination with an IEA expert meeting on aerodynamics as organized by IEA Task 11.

Results/Status as of August 2014

In June-July 2014 two weeks of New-Mexico measurements were carried out in the LLF of DNW. The model, instrumentation and data acquisition which was already tested before (e.g. in the TUDelft LST tunnel see below) was also working well in the DNW-LLF environment and many datapoints were collected  (1366 datapoints in New Mexico versus  950 datapoints in Mexico). As an illustration figure 1 shows a smoke visualisation of the tip vortices behind the rotor.

Figure 1: Tip vortex visualisation on New-Mexico

The New Mexico experiment started with a repetition of some Mexico measurements. It was then encouraging to see how little the results of the pressure distributions have changed after 8 years, see figure 2.

Figure 2: Pressure distribution at 82% span and design conditions: Mexico results compared with New Mexico results. Calculational results from a CFD code (DTU-Ellipsys) are added too.

Most interesting was that  the underlying flow measurements (carried out under responsibility of DNW) show some differences which at least partly explain the non-understood phenomena of Mexico, see figure 3.

Figure 3: PIV velocities measured by DNW as function of axial coordinate (x=0 is the rotor plane): Mexico results compared with New Mexico results. Calculational results from a CFD code (DTU-Ellipsys) are added too. 

However, the New Mexico experiment also aimed to improve the quality of the database by taking into account the lessons learned from the Mexico experiment and to benefit from the improvements in technical capabilities of DNW over the last 8 years. In this respect it should be known that the Mexico experiment in 2006 could not map the inner part of the flow field, partly because of limitations in the radial range of the PIV system at that time, partly because the nacelle of the turbine led to disturbing reflections at the inner part of the measurement field. In New Mexico the radial range of the PIV system was increased and the reflections were overcome by applying Rhodamine to the nacelle. Other lessons were that some experiments in Mexico turned out to be less succesful (e.g. standstill measurements). New Mexico offered the opportunity to repeat these measurements with more success. Moreover several additional measurements are done (e.g measurement with flow devices like guerney flaps and root spoilers and measurements at a pitch misalignment). Very important were acoustic measurements taken with an array of microphones near the turbine and with far field microphones where it should be realised that acoustics and aerodynamics are inextractibly connected. Finally flow vizualisation measurements have been carried out (see e.g. figure 1).

In summary the New Mexico experiment was very succesful and provided a huge amount of data. Further analysis is expected to take place in a third phase of Mexnext
 

Results/Status as of February 2014

In 2013 a calculation round on the NREL Phase VI (Nasa-Ames) experiment has been defined which consists of four cases in axial flow. Emphasis was put on measurements at a rotational speed of 90 rpm. Such rotational speed is higher than the commonly featured rotor speed of 72 rpm. This makes it interesting because it is expected to lead to different induction effects. When preparing these cases some improvements to the existing CAD file of the NREL Phase VI blade surface geometry turned out to be necessary in order to provide a reliable input to the CFD codes. The results of the calculations in comparison with the measurements will be presented at the next EWEA conference from 10 March - 13 March 2014 in Barcelona.

Also much attention was paid to the preparation of the New Mexico experiment. Thereto a large number of ‘lessons learned’ were summarized and presented to the Mexnext group in September 2013 at a plenary meeting. These lessons learned are now included in the preparation of the experiment and/or the test matrix. The preliminary test matrix includes pressure measurements, PIV measurements, load measurements (including the measurement of torque from the generator), microphone array measurements, and application of several flow visualization techniques. These measurements will be taken at several conditions including yaw and dynamic pitch. In addition to that, a test run is added where the blades will be equipped with Guerney flaps.
In order to prepare the New Mexico experiment the instrumented Mexico blades have been placed in the Low Speed Tunnel (LST) of the TU Delft, i.e. at quasi 2D conditions. In the LST the aerodynamic characteristics of the blades at standstill have been measured (including a flow visualization) where moreover the blade instrumentation and data acquisition could be tested and recalibrated. A first test in LST was done in November 2013 followed by a second experiment in January 2014. The experiments in the LST were very successful. The pressure sensors and data acquisition have been brought alive again and several improvements have been made to the instrumentation and data acquisition system. The analysis of aerodynamic characteristics at standstill is currently taking place. Some first results of this analysis will be reported at EWEA 2014 in Barcelona. In the figures below some representative oil flow visualizations from the LST experiment are shown.

Several analyses took place on measurements other than NREL Phase VI or Mexico. Amongst other things measurements from FFA/CARDC (as performed at the end of 1980’s) have been used to derive a 3D correction model on the drag coefficients. Also measurements on transition from Danaero, UAS-Kiel and TUDelft/ECN were found useful. The same holds for the very well documented IEA Task 18 measurements from the 90’s which could be used to assess tip effects and 3D drag effects. However, in some old experiments the information on quality and precise airfoil shape was found difficult to retrieve where it should be realized that the statement ‘everyone believes the measurements except the experimentalist’ is not very true within the Mexnext-II project team…

All of the above mentioned results and actions were discussed at the 3rd meeting of Mexnext-II (i.e. the 8th meeting of the overall Mexnext project) which has been held from September 25th to 28th 2013 at CENER Spain. At the meeting several technical results have been presented, together with the test plan for the New Mexico experiment where the last day was devoted to ‘aerodynamic free wheeling’, i.e. presentations on aerodynamic subjects not related to Mexnext. Some of these presentations can be found here.

Results/Status as of December 2012

The first phase of the project started in June 2008. The official end date of the first phase was in  June 2011. In October 2011 the IEA Executive Committee approved an extension of the project (Mexnext-II) for a three years period. In the extension of Mexnext it will not only be Mexico measurements which are analysed but a wide database of aerodynamic measurements on wind turbines (both in the wind tunnel as well as in the field) will be considered. Moreover in Mexnext-II additional measurements are performed on the Mexico rotor ('New Mexico') taking into account the lessons learnt in the first phase of the project.

The main results from the first Phase of Mexnext are described in many references, see e.g. [1] to [34]. 

The work performed in IEA Task 29 formed together with the work performed in IEA Task 20 and IEA Tasks 14/18 the basis for a PhD thesis, see [34].

Furthermore a public final report is completed. The final report discusses  a large variety of subjects, e.g. investigations on measurement quality and tunnel effects. Apart from the measurements at standstill, some outliers and a disturbance from nacelle  reflection at the inner part of the PIV measurement range, the data quality was generally found to be good. Furthermore the measured rotating airfoil characteristics were presented in different ways using different angle of attack methods (or even without angle of attack methods) and the rotating characteristics were compared with 2D measurements. PIV measurements in the wake have been processed and analysed, see below and a comparison was made between the observations from NASA-Ames and Mexico experiment. Moreover the blade geometry was measured and the effects from airfoil imperfections were assessed with CFD. Also effects from Reynolds number and Mach number were discussed etc. It is worthwhile mentioning that Kari and INTA even built scaled down models of the Mexico rotor and measured the performance of it in their own wind  tunnel sin comparison to the performance of the ‘real’ Mexico turbine (see the figure below). The results from some comparisons are presented in [19].

Furthermore comparisons are made between calculational and measured results. The first comparisons were made for non-yawed conditions at three different tip speed ratios: A low tip speed ratio (i.e. stalled conditions), the design tip speed ratio and a high tip speed ratio (i.e. turbulent wake state). Most interesting in this comparison is the fact that the comparison is not only made with load measurements but also with the underlying flow field measurements which drive these loads. This comparison has led to a striking observation. In the figures below the axial velocity decay at 82% span is presented for design conditions. Also indicated are the results from a cylindrical vortex wake method which is compatible to the momentum theory. The axial force coefficient in the cylindrical vortex sheet method was set to 0.89 which was beforehand the expected value in view of the fact that these measurements are performed at design conditions.
 
Apart from some  discrepancies near the rotor plane (which are explained in [10])  the measured velocity decay seems to agree reasonably well with the momentum theory if  CDax is set to the expected value of 0.89 but it was then very striking to note that the measured CDax at the design tip speed ratio of 6.66 obtained with two independant measurement techniques (pressure measurements and balance measurements) and for two tip speeds, is only 0.72, see the figure below.  Such low CDax yields much lower axial induced velocities and hence a higher velocity level.
The first logical thought would be that the CDax measurements are incorrect but this is difficult to believe in view of the fact that these values are determined with  two fully independent measurement techniques, the results of which agree very well at all datapoints at different tunnel speeds and different rotational speeds.
  
Another explanation for the anomalies could be tunnel effects. An extensive investigation on the impact of these effects has taken place within the Mexnext project on basis of CFD calculations.  Until now the impact of these effects seems to be  small at design conditions, and the limited effect which has been
found points in the  opposite direction: When the induction in the tunnel is the same to the induction in the  free stream,  the CDax is higher in the tunnel situation. This implies that the free stream CDax would even be lower than 0.72.

As such an explanation for the discrepancy between measurements and momentum theory has not been found yet but the logical consequence is that none of the calculations from the Mexnext group can predict both the velocities AND loads in a correct way: The  calculations overpredict the velocity (underpredict the induction) and/or overpredict the loads.
This is clearly visible in the figures below which show results from several CFD calculations for the normal force  distribution along the blade and the radial and axial traverses of the axial  velocity in comparison with  the measurements. As a matter of fact almost all codes overpredict both the loads and the velocities. Nevertheless the results of some codes approach the measurements very closely but this is only true for either the velocities or the loads. Despite these discrepancies it is very encouraging to see the good qualitative agreement between calculations and measurements also in terms of flow details. This was even found in yawed conditions. 

All graphs with comparisons between calculations and measurements are included in the final report. A request for high resolution graphs should be forwarded to the Operating Agent (http://www.mexnext.org/contact).

In November 2012 a Mexnext-II meeting was held which was attended by most participants.  At this meeting an inventory has been made of various experiments which will form the basis for the remaining activities in Mexnext-II. On basis of this inventory the usefullness of the experiments for several research topics was established. Moreover the first steps were made on the definition of the test matrix for the ''New Mexico' experiment.

More results on these subjects will be posted soon.

With regard to availability of the measurement data, it can be mentioned that a selection of data is made public. It concerns the measurements which have been used as basis for the comparison with calculational results. Also the information on the experimental set-up and the Mexico rotor can be made available under the condition that an NDA is signed on the confidentiality of one of the airfoils on the rotor. A request for the measurement data and the data of the Mexico rotor should be forwarded to the Operating Agent (http://www.mexnext.org/contact). The data are only supplied if a proper acknowledgement to the Mexico consortium in publications is guaranteed. Moreover feedback on the results should be given to the Mexico consortium.

 It is mentioned that the data from the 'New Mexico' experiment will only be supplied after a financial compensation.

  1. J. G. Schepers and H. Snel: ‘Model Experiments in Controlled Conditions, Final report’, ECN-E-07-042, Energy Research Center of the Netherlands, ECN, February 2007, http://www.ecn.nl/docs/library/report/2007/e07042.pdf
  2. H. Snel, J.G. Schepers, B. Montgomerie: ‘The MEXICO project (Model Experiments in Controlled Conditions): The database and first results of data processing and interpretation’, The Science of Making Torque from the Wind, 28–31 August 2007, Technical University of Denmark, http://iopscience.iop.org/1742-6596/75/1/012014/pdf?ejredirect=iopsciencetrial
  3. Kay, A. Investigating the Unsteady Aerodynamics associated with a horizontal axis wind turbine, with reference to the recent measurements gathered during the Mexico project. TU master project report
  4. L. Pascal Analysis of Mexico measurements,  ECN-Wind Memo-09-010, January 2009
  5. H. Snel, J.G. Schepers and A. Siccama, Mexico, the database and results of data processing and analysis 47th AIAA Aerospace Sciences meeting, Orlando, USA, January 2009
  6. A. Bechmann and N. Sørensen CFD simulation of the Mexico rotor wake, European Wind Energy Conference, March 2009, Marseille France
  7. A.K. Kuczaj Virtual Blade Model Simulations of the Mexico experiment, NRG-12810/09.97106, NRG Petten, The Netherlands
  8. Daniel Micallef, Menno Kloosterman, Carlos Ferreira, Tonio Sant , Gerard van Bussel Validating BEM, Direct and Inverse Free Wake Models with the MEXICO experiment, 48th AIAA Aerospace Sciences meeting, Orlando, USA, January 2010
  9. S Breton, C Sibuet, C Masson Using the Actuator Surface Method to Model the Three-Bladed MEXICO Wind Turbine 48th AIAA Aerospace Sciences meeting, January 2010
  10. J.G. Schepers, L. Pascal and H. Snel. First results from Mexnext: Analysis of detailed aerodynamic measurements on a 4.5 m diameter rotor placed in the large German Dutch Wind Tunnel DNW. European Wind Energy Conference, EWEC, April 2010, Warsaw Poland
  11. Wen Zhong Shen et al Validation of the Actuator Line / Navier Stokes technique using Mexico measurements, 'The Science of Making Torque from the Wind', June 2010
  12. Yang Hua et al  Determination of Aerofoil Data and Angle of Attack on the Mexico Rotor using Experimental Data  'The Science of Making Torque from the Wind', June 2010
  13. S. Schreck et al Rotational Augmentation Disparities in the UAE Phase VI and MEXICO Experiments 'The Science of Making Torque from the Wind', June 201
  14. S. Gomez-Iradi and X. Munduate: A CFD Investigation of the Influence of Trip-Tape on the MEXICO Wind Turbine Blade Sections 'The Science of Making Torque from the Wind', June 201
  15. B. Stoevesandt et al ''OpenFOAM:RANS-Simulation of a wind turbine and verification " 'The Science of Making Torque from the Wind', June 2011
  16. S. Breton, C. Sibuet, C. Masson, Analysis of the inflow conditions of the MEXICO Rotor : comparison between measurements and numerical simulations, 'The Science of Making Torque from the Wind', June 201
  17. J.G. Schepers, K. Boorsma, H. Snel,  IEA Task 29 Mexnext: Analysis of wind tunnel measurements from the EU project Mexico,  'The Science of Making Torque from the Wind', June 2010
  18. T. Lutz, Near Wake studies of the Mexico Rotor 'EWEA Annual Event', March 2011
  19. J.G. Schepers, K. Boorsma, C. Kim, T Cho,  Results from Mexnext: Analysis of detailed aerodynamic measurements on a 4.5 m diameter rotor placed in the large German Dutch Wind Tunnel DNW,  'EWEA Annual Event', March 2011
  20.  R. Pereira, J.G. Schepers, KM. Pavel,  VAlidation of the Beddoes Leishman Dynamic Stall model for Horizontal Axis Wind Turbines using Mexco data,  49th AIAA Aerospace Sciences Meeting Orlando USA, January 2011
  21.  S.K. Guntur, C. Bak and N.N. SorensenAnalysis of 3D stall models for wind turbine blades using data from the Mexico experiment13th International conference on Wind Engineering, ICWE, Amsterdam Holland, July 2011
  22. D. Micallef et alThe relevance of spanwise flows for yawed horizontal-axis wind turbines13th International conference on Wind Engineering, ICWE, Amsterdam Holland, July 2011
  23. Réthoré, P.-E., Sørensen, N.N., Zahle, F., Bechmann, A., Madsen, H.A., "CFD model of the MEXICO wind tunnel", 'EWEA Annual Event', March 2011
  24.  Réthoré, P.-E., Sørensen, N.N., Zahle, F., Bechmann, A., Madsen, H.A., "MEXICO Wind Tunnel and Wind Turbine modelled in CFD". AIAA Conference. Honolulu, Hawaii, USA. June 2011
  25. K. Meister "Grid dependency studies on tip vortex preservation in wind turbine CFD simulations". Wake Conference. Gotland University, Sweden. June 2011
  26. W.Z. Shen "Actuator Line / Navier Stokes Computations for Flows past the Yawed MEXICO Rotor". Wake Conference. Gotland University, Sweden. June 2011
  27. N. Sorensen "Near Wake Predictions Behind the MEXICO Rotor in Axial and Yawed Flow Conditions". Wake Conference. Gotland University, Sweden. June 2011
  28. R. Szasz "LES of the near wake of the MEXICO wind turbine". Wake Conference. Gotland University, Sweden. June 2011
  29. S. Breton "Numerical Analysis of the Vorticity Structure of the MEXICO Rotor in the Near Wake". Wake Conference. Gotland University, Sweden. June 2011
  30. K. Nillsen  "The wake behind the Mexico rotor". Wake Conference. Gotland University, Sweden. June 2011
  31. Yang Hua et al  "Extraction of airfoil data using PIV and pressure measurements” Journal of Wind Energy, 14, 2011, pages 539-556 .
  32. J.G. Schepers, K. Boorsma, X. Munduate, Final Results from Mexnext-I: "Analysis of detailed aerodynamic measurements on a 4.5 m diameter rotor placed in the large German Dutch Wind Tunnel DNW", 'The Science of Making Torque from the Wind', October 2012
  33. Iván Herráez "Aerodynamic Simulation of the MEXICO Rotor", 'The Science of Making Torque from the Wind', October 2012
  34. J.G. Schepers "Engineering models in wind energy aerodynamics, Development, Implementation and analysis using dedicated aerodynamic measurements", PhD Thesis, TUDelft, ISBN:978-94-6191-507-8, November 2012

 

 

 

Axial velocity traverse

Measured axial velocity decay at design conditions in comparison with calculated results at CDax = 0.89 (as expected at design conditions) and CDax = 0.72 (as expected from axial force measurements)

 

Radial velocity traverse

Axial velocity as function of radial coordinate just behind the wind turbine in the wake (tip of the blade at r=2.25 m), calculated and measured

Normal force along the blade

Normal force along the blade at design conditions: Calculated and measured

 

 

Contour plots of PIV measurements

Axial velocity traverse showing velocity decay from upstream to the wake and radial velocity traverse just after blade passage showing tip vortices (bottom)

Experiment performed at Kari

Comparison between Mexico (upper) and Kari(lower) set-up