estimating fatigue life of gears in halt test using...

Estimating Fatigue Life of Gears in HALT TEST

using KISSsoft for Wind turbine gearboxes

M. Mahendran,

Head -Engineering,

Shanthi Gears Ltd .,

Coimbatore.

Paper presented for KISSsoft User Conference

25th – 26th November, 2015. Pune, India

Wind turbine gearboxes designed for :

• Highest reliability with design life of 20 years

• Withstanding very diverse spectrum of dynamic loads

• Manage load transients during emergency stop /“grid loss”

• Compact, lightweight and minimum cost

• Adhering stringent certification norms (GL /IEC /..)

• Site specific loads

• Turbine Blade Specific loads

The calculated bending or pitting fatigue life of a gear is a measure

of its ability to accumulate discrete damage until failure occurs.

Calculated service life is based on the theory that:

• every load cycle (every revolution) is damaging to the gear.

• the amount of damage depends on the stress level

• damage can be considered as zero for lower stress levels.

The fatigue life calculation requires

• the stress spectrum

• material fatigue properties

• a damage accumulation method

NOTE: HALT method is not applicable for predicting Scuffing & Micropitting

failures.

Fatigue Life Calculation:

Fatigue Life Calculation (cont..):

• mainly based on linear damage accumulation method Palmgren-

Miner rule/ Corten-Dolan / Haibach.

• Failure could be expected U=1.0 or above 1

• according to the S-N curve established for the material and

stress spectra for each load level i acc. ISO 6336

• Simulates few /20 year life of gearbox components in testing

• Over-stress the gearbox components in controlled manner to

induce failures in the product very quickly

• Test results will expose the weakest elements of design

• Overloads, approx. closer to 150 % of nominal torque been

applied

Note: HALT is not a pass/fail test. It is a process of finding the failure

modes and do the design optimization.

HALT - Highly Accelerated Life Testing:

During Test, the following parameters

are monitored:

• Bearing and oil temperatures

• Oil pressures

• Stresses in critical components

• Power, torque, and speed

• Gear wear patterns

• Particle count

• Efficiency

• Stiffness and casing deflection

HALT - Highly Accelerated Life Testing(cont..):

Life consumed in HALT:

Damage due to HALT = U_HALT

Total permissible damage U_total = 1.00

Life Estimated under LDD = 20 years

Life consumed in HALT = U_HALT X 20 years

Life Remains after HALT U_rest = U_total - U_HALT

Damage is proportional to life, so, one unit of damage = 20 years

Hence, remaining life L_rest = 20years - life consumed in halt

= 20years - ( U_HALT x 20years)

= 20years x (1- U_HALT )

Gearbox Load Data:

Simulated load data collected from aero-elastic wind turbine codes / simulation

software / Wind turbine load measurements.

The data is compiled for all design/assumed range of operational speeds and torque

range, with estimated running hours for each bin.

Gearbox Load Data(cont..):

( * Customised / altered LDD is used here to explain concept. )

Rotor Nominal Power 400 kW

Rotor Speed 40 min^-1

Rotor Nominal Torque 95.5 kNm

Total Ratio 37.52

LSS IMS HSS

Type of Gearing HELICAL mn=7 HELICAL mn=5 HELICAL mn=3

CD 450 320 225

Teeth on Wheel z2 95 95 109

Teeth on Pinion z1 27 27 36

Ratio 3.52 3.52 3.02

Nominal Speeds (RPM)

Input side of stage 40 140.8 495.6

Output side of stage 140.8 495.6 1500

Nominal Torque (kNm)

Input side of stage 95.5 27.13 7.7

Output side of stage 27.13 7.7 2.54

Gearbox Spec. Technical Data (ref.*):

( *assumed tooth data is used here to explain concept. )

Time factor Torque Factor

0.000000003 1.413613

0.000000006 1.308901

0.000000008 1.151832

0.000000012 1.146597

0.000000029 1.115183

0.000000041 1.08377

0.000000059 1.052356

0.000002055 1.020942

0.000477330 0.989529

0.031492036 0.958115

0.080806231 0.926702

0.042286441 0.895288

0.028845230 0.863874

0.025736243 0.832461

0.023858955 0.801047

0.021419538 0.769634

0.020546210 0.73822

0.021120494 0.706806

0.017897556 0.675393

0.019375987 0.643979

0.018264337 0.612565

0.016037806 0.581152

0.016594262 0.549738

0.017235954 0.518325

0.016895333 0.486911

0.017325807 0.455497

0.018373122 0.424084

0.022560848 0.39267

0.031134675 0.361257

0.035597599 0.329843

0.039521123 0.298429

0.035452380 0.267016

0.034612283 0.235602

0.043324047 0.204188

0.040684805 0.172775

0.031202781 0.141361

0.039378345 0.109948

0.055551284 0.078534

0.046863152 0.04712

0.089525620 0.015707

T_NOM=95.5 kNm

@KissSoft

Load Spectrum:

A load spectrum “simulated” or “assumed field loads”

consists of the frequency, speed, and power or torque data.

( * Customised / altered spectrum is used here to explain concept. )

KISSsoft – Safety factor calculation with Load Spectrum:

KISSsoft -Damage calculation with LDD:

HALT - Highly Accelerated Life Testing:

Hours Torque (kNm)

80 116.35

197 96

37 9.6

36 0.955

total =350 Hrs

Hours Torque (kNm)

0.226316 1.218325

0.563158 1.005236

0.105263 0.100524

0.105263 0.01

T_NOM=95.5 kNm

KISSsoft (converted to Factors)

( * Customised / altered spectrum is used here to explain methodology. )

KISSsoft – Safety factor calculation with HALT Spectrum:

KISSsoft -Damage calculation with HALT :

The damage in HALT is different for root and flank. Also, the damage in HALT for each

stage of gearbox (LSS/IMS/HSS) is separate.

KISSsoft –HALT -Damage calculation for all Stages:

In KISSsoft, Damage also can be calculated on basis of system & individual

component’s service life

KISSsoft (HALT) Damage calc. –as system & component:

KISSsoft calculation & correlation with test results:

The contact stress & contact pattern computed by EES with optimal gear modifications

(for 100% nominal torque using KISSsoft software for 2MW gearbox ) and the full load

test results (conducted by one of EES customer in late 2012 / early 2013)had shown

finest correlation between computation and test results.

Courtesy : Mr. Dinner

• The load spectra is not being very informative in many areas

• The damages originate in very few spectrum steps for highly stressed gears

• The damage-relevant steps for the bearing lay somewhat lower than for the

gearing due to smaller S-N curve slope.

• Possible gaps in Material quality , process variations from design

considerations.

• Need several calculations with different spectrum for changed tower height,

blade length, different installation places.

• Exact prediction of field loads and simulation of dynamics in test bench is not

easily achievable.

LDD & HALT - Expert Opinions :

• For scuffing, the fatigue calculation method is not applicable (because

scuffing is not a fatigue effect) and that - if torque in HALT is high, scuffing

may be a limiting factor.

• For Micropitting, no HALT is possible as it is not yet clearly understood how

LDD affects Micropitting / how Micropitting depends on no. of cycles..

• Failure is at U=1 for gear rating standards but research suggest that U=0.4

may be more realistic.

• HALT is a useful technique to ensure the higher reliability of the product

and it enables faster time-to-market of multi-MW WTG gearboxes.

LDD & HALT - Expert Opinions :

1. ISO 6336:2006 - Part 6: Calculation of service life under variable load

2. Static and Fatigue Calculations of Wind Turbine Gearboxes -Dipl. Ing. ETH Hanspeter

Dinner

3. Dr. Kissling, & H. Dinner - Lecture Notes, India - Gear Rating and Wind Turbine

gearboxes

4. Mr.H.Dinner & Dr.Giger - Lecture Notes -Windmill gearbox Technologies & KissSoft

training ,CWET, Chennai, June-2007

5. KISSsoft software 03-2015, www.KISSsoft.ch

6. Germanisher Lloyd (GL) Guideline for the Certification of Wind Turbine Edition 2010

7. Advances in wind turbine and components testing- IEA R&D Wind Task 11 - Topical

Expert Meeting proceedings Feb. 2012

References: