The Power of Sharing Wind Energy Data and Code

“Sharing your data and code requires an ini­tial time invest­ment, but it’s worth it in the long run,” says TU Delft PhD researcher and FAIR data advo­cate, Mark Schel­ber­gen.

Mark’s inter­est in wind ener­gy piqued dur­ing this Master’s degree when he stud­ied the struc­tur­al opti­miza­tion of mul­ti-megawatt off­shore ver­ti­cal axis wind tur­bine rotors. 

Now, Mark’s doc­tor­al stud­ies aim to advance the wind ener­gy sec­tor one step fur­ther by devel­op­ing inno­v­a­tive and cost-effec­tive alter­na­tives to con­ven­tion­al wind tur­bines.

Kite power

In col­lab­o­ra­tion with lead­ing start-up and TU Delft spin-off com­pa­ny, Kitepow­er B.V., Mark inves­ti­gates ener­gy pro­duc­tion char­ac­ter­is­tics of pump­ing kite pow­er sys­tems.

“Air­borne wind ener­gy sys­tems use fly­ing devices to extract ener­gy from the wind,” says Mark. “The Kitepow­er sys­tem is a large kite attached to a gen­er­a­tor on the ground by a cable. The cable unwinds due to the pulling force gen­er­at­ed by the kite that’s fly­ing cross-wind.”

Mark explains the ben­e­fits of the pump­ing kite pow­er sys­tem over hor­i­zon­tal axis wind tur­bines. 

“Air­borne wind ener­gy sys­tems use up to 90% less mate­r­i­al than a con­ven­tion­al wind tur­bine which sig­nif­i­cant­ly reduces the man­u­fac­tur­ing costs. More­over, they’re mobile and easy to deploy at remote loca­tions.”

He adds that air­borne wind ener­gy sys­tems offer a lot of flex­i­bil­i­ty. “Kites can be oper­at­ed at heights above con­ven­tion­al wind tur­bines. This enables the har­ness of stronger and more per­sis­tent winds, and poten­tial­ly increas­es pow­er out­put.”

Performance estimation

As kite pow­er is rel­a­tive­ly new to the wind ener­gy sec­tor, Mark and his col­leagues with­in the Air­borne Wind Ener­gy (AWE) research group com­pare the per­for­mance of this nov­el tech­nol­o­gy with con­ven­tion­al wind tur­bines to assess its suit­abil­i­ty as a sus­tain­able green ener­gy solu­tion for the future.

To esti­mate the per­for­mance of kite pow­er, their study pub­lished in 2019 used his­tor­i­cal ERA5 netCDF cli­mate mod­el data cre­at­ed by the Euro­pean Cen­tre for Medi­um-Range Weath­er Fore­casts (ECMWF). The dataset includes hourly esti­mates of wind speed at more than 130 alti­tude lev­els over a sev­en year peri­od (from 2011 to 2017). 

In a fol­low-up study, Mark devel­oped Python soft­ware code to analyse the dataset, and pro­duce detailed wind sta­tis­tics and ener­gy yield esti­mates of kite pow­er for an on- and off-shore loca­tion in the Nether­lands. 

The soft­ware code uses infor­ma­tion about kite pow­er flight mechan­ics, plus infor­ma­tion about wind speed and direc­tion over a large height range to esti­mate the aver­age annu­al ener­gy yield. It has also been used as part of an AWE Master’s course, demon­strat­ing Mark’s tool­chain in edu­ca­tion. 

Sharing data and code 

Mark talks about his research group’s moti­va­tions for mak­ing the dataset under­ly­ing the jour­nal arti­cle open­ly avail­able in 4TU.ResearchData. 

“We pub­lished a data pack­age that com­pris­es the soft­ware code; the orig­i­nal ERA5 data used for the analy­sis; and, a detailed README file to pro­vide instruc­tions about how to down­load the ERA5 data and run the scripts.” 

“This not only allows future researchers to be able to repro­duce our analy­sis, but they’ll be able to use our data pack­age to pro­duce wind sta­tis­tics and ener­gy yield esti­mates for any loca­tion in the world. This infor­ma­tion is essen­tial for assess­ing the suit­abil­i­ty of deploy­ing kite pow­er sys­tems at spe­cif­ic loca­tions,” he adds.

Motivational mentorship

“We also appre­ci­ate that it can be chal­leng­ing to find suit­able data for wind ener­gy analy­sis. Plus, it’s not always easy to under­stand anoth­er researcher’s exper­i­ments by read­ing their jour­nal arti­cle alone, so hav­ing access to the under­ly­ing data and code is crit­i­cal to under­stand­ing and inter­pret­ing the pro­gramme of work cor­rect­ly,” he adds. 

Mark says that all mem­bers of the AWE group are men­tored to approach research data man­age­ment in a sim­i­lar way from the begin­ning of their project. A prime exam­ple is their col­lec­tive use of GitHub as a ver­sion con­trol man­age­ment tool for col­lab­o­ra­tive and trans­par­ent soft­ware code devel­op­ment. 

“I have my own GitHub repos­i­to­ries but the group also uses GitHub to share code with one anoth­er. This way, we learn good cod­ing and doc­u­men­ta­tion prac­tice from one anoth­er,” he says.

Personal development

In order to devel­op his cod­ing skills, Mark has par­tic­i­pat­ed in a Code Refin­ery work­shop where he learnt the val­ue of invest­ing time and effort to make his code robust, read­able, reusable and repro­ducible. 

“Some­times quick fix­es are need­ed, how­ev­er, these should not be kept for too long in order to pro­duce under­stand­able code that’s use­ful for oth­ers” explains Mark. 

He con­tin­ues, “You have to learn how to write and doc­u­ment your code prop­er­ly so that you can revis­it your scripts years lat­er and under­stand exact­ly what you did. It requires an ini­tial time invest­ment, but it’s worth it in the long run when it comes to mak­ing code FAIR and repro­ducible.”

“You have to learn how to write and doc­u­ment your code prop­er­ly so that you can revis­it your scripts years lat­er and exe­cute it with­out any dif­fi­cul­ties. It requires ini­tial time invest­ment, but it’s worth it in the long run!”

Mark’s good prac­tice and open way of work­ing has led to his code being reused and devel­oped by a Master’s degree stu­dent at the Uni­ver­si­ty of Bonn in Ger­many as part of a cross-insti­tu­tion­al research project.

In addi­tion, at the begin­ning of the year, he was nom­i­nat­ed for an Open Ini­tia­tives Tro­phy award and received a spe­cial men­tion dur­ing the Open Sci­ence Fes­ti­val for his invit­ing atti­tude toward Open Sci­ence and for demon­strat­ing how FAIR data prin­ci­ples can be prac­ti­cal­ly imple­ment­ed with­in his research field. 

We thank Mark for pub­lish­ing his dataset in 4TU.ResearchData and shar­ing his research sto­ry. We look for­ward to fol­low­ing his jour­ney toward more sus­tain­able air­borne wind ener­gy solu­tions.

Writ­ten by Con­nie Clare (4TU.ResearchData)
Edit­ed by Mark Schel­ber­gen and Roland Schmehl (Delft Uni­ver­si­ty of Tech­nol­o­gy)
Cov­er image snap­shot from thekitepower.com