SURVIVING THE HURRICANE SEASON:WIND-RESISTANT DESIGN OF SINGLE-AXIS PV TRACKING SYSTEMS Thorsten Kray, Ph.D. (Dr.-Ing.)

Daniel Markus, M.Sc.

WHY BOTHER?• 50% of utility-scale PV failure due to weather 1

• need to protect your investment• Detailed knowledge of wind loads on solar arrays is key.• Wind loads comprise not only static, but especially

dynamic components: resonance effects may double or triple the static loads!

• For the 1st time ever, we present results of a comprehensive and concise study with a broad range of parameters affecting wind loads on single-axis PV tracking systems.

1 https://www.solarpowerworldonline.com/2018/01/solar-industry-responding-increasing-intensity-natural-disasters/

I.F.I. Institut für Industrieaerodynamik GmbH(Institute for Industrial Aerodynamics)Aachen, Germany

LADBS Approved Laboratory forWind Tunnel Testing of Buildings and StructuresTesting Agency License Number TA 24830

• boundary layer wind tunnel testing of amodel array

Present Study:• model at a 1:50 scale• 8 model rows, each with a length of 84 meters at

full-scale and equipped with 250 pressure taps• test conditions: 5 tilt angles, 2 row spacings, 13

wind directions, 8 rows

⇒ more than 1000 tested configurations

STATIC vs DYNAMIC LOADSØ composed of a mean and

background fluctuating component

Ø distribution of loads over tracker area can be reconstructed

Ø often ignored by codes of practice

Ø caused by vortex shedding and gust buffeting

Ø excited by resonances in tracker structure

Data taken from „Cell, Interrupted “ (2016) by GCube Insurance

1,00

1,20

1,40

1,60

1,80

2,00

0,00 0,20 0,40 0,60 0,80 1,00 1,20 1,40

DAF

Natural Frequency x Chord Length / Velocity

0°10°20°30°45°

49.80%Weather

36.10%Fire

9%Electrical

Failure

2.30%Mechanical Breakdown

2.30%Lightning

0.50%Theft

PV CLAIMS IN NORTH AMERICA

PARAMETER TORQUE

Locationwithin Array

Perimeter +

Interior -

Row Spacing + +

Tilt Angle + -

• frequency domain analysis ofload effect time-series

• calculate power density spectra to determine dynamic response

• determine Dynamic Amplification Factors (DAFs) for easy application with static loads:

OVERALL LOAD = STATIC LOAD x DAF

• based on time-dependent 2D CFD simulation (Computational Fluid Dynamics)

• grid resolution should be increased near tracker surfaces

• modelling of many different parameters in incident flow and of the tracker’s structural characteristics is easily possible (wind speed, natural frequency, damping ratio, torsional stiffness etc.)

• simulation of single and multibody configurations

Only the combined informationmakes your system wind-resistant!

0%20%40%60%80%

100%

0 10 20 30 40Tilt Angle / deg

1st & 2nd row 3rd & 4th row Inner rows

and TRACKER INSTABILITY

• DAFs are determined for system- and site-specific reduced frequencies.• Resonances can easily lead to amplification of 100% (DAF = 2) or even higher!• highly dependent on tilt angle, load effect, row spacing and row location

within the array

• 1st and 2nd row are most exposed and feature highest loads.

• Inner rows are effectively shielded.

Top right: Pressure coefficients representing peak torque across an array with high values in red and low values in blue

Bottom left and bottom right:Effect of different parameters onpeak torque across an array

• multibody configurations: interior rows are less prone to torsional galloping (lower oscillation amplitudes, see diagram above)

#1 #2 #3

#4 #5 #6

Six row tracker array

“KNOWLEDGEIS POWER“

Sir Francis Bacon

Ø caused by torsional galloping due to vortex shedding or vortex lock-in

Ø can lead to catastrophic failure due to high amplitudes of oscillation at a critical wind speed

Ø study of onset of instability is necessary for definition ofsafe stowing policy

Ø combination of static + dynamic loads in terms of equivalent static wind loads

Ø structural failure in case of underestimation

Web www.ifi-ac.comwww.ifi-aachen.de

eMail kray@ifi-aachen.de

-40

-20

0

20

0 1 2 3 4 5

Angl

e (d

eg)

Time (s)Tracker multibody Tracker single body