probabilistic assessment of the potential creep damage at ...ø faster start-ups ... damage at...

Unrestricted © Siemens AG 2014 All rights reserved. siemens.com/answers

Probabilistic assessment of the potentialcreep damage at the surface of steam turbinerotor shafts

Ralf Voß, Henning Almstedt

2014-09-26

Unrestricted © Siemens AG 2014 All rights reserved.

Voß / E P SU R&D TEC COT 1

Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsTable of content

• Motivation

• Deterministic design rules

• Implementation strategies ofProbabilistic Design

• Exemplary application: Potentialcreep damage in steam turbinerotor shafts

• Resume

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shafts

Requirements from steam turbine customers:• Due to changes in the energy-mix increased flexibility

necessary:Ø Increasing number of start-upsØ Faster start-upsØ Part-load-operation

• Increasing cost-pressure on steam turbine manufacturers(e.g. US: low gas priceà efficiency less important; EU: powerplants are less time in load-operation à worse ROI)

Motivation

► Motivation

► Deterministicdesign rules

► Implementationstrategies ofProbabilisticDesign

► Exemplaryapplication

► Resume

à Increasing mechanical utilization of steam turbinecomponents required

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsDeterministic design rules

State-of-the-art in the design of steam turbines: Deterministicmechanical design methods using „Worst-Case“-assumptions:• Nominal steam temperature and pressure• Minimum material parameters (e.g. creep rupture strength)Judgement:

► Motivation




► Resume

• Is this really the limit?• What risk of failure belongs to this utilization?• How robust is the result when inputs variate?

SRSR

Em

CalcEmCalc

1

52

52 £Û£s

s

Inner of rotor: 1/S1/S

2014-09-26




Curves of equalrisk level

Implementation strategies of Probabilistic Design

► Motivation




► Resume

TurbineModel

Material

Geometry

Load regime

Turbine life timeconsumption(e.g. LCF-damage)

Inherent uncertainties(stochastic model)

Uncertainties in parameterestimation (depends on

available data)

PoF(Probabilityof Failure)

Numberof starts

Accepted PoF(e.g. 10%)

Approach 1: Development of Probabilistic Design methods thatare released in internal design rules or norms

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsImplementation strategies of Probabilistic Design

► Motivation




► Resume

Approach 1: Development of Probabilistic Design methods thatare released in internal design rules or norms

Necessary aspects for realization:• Update of design rule:Ø Calculation procedureØ Acceptable risk level

• Development of adequate tools• Knowledge building in affected areas (e.g. mechanical design

department)

à Challenging and time-consuming implementationprocess!

2014-09-26




Curves of equalrisk level

Implementation strategies of Probabilistic Design

► Motivation




► Resume

Approach 2: Usage of Probabilistic Design methods to determinerisk-based safety factors for deterministic design. The designrules are updated with those new safety factors.

PoF(Probabilityof Failure)

Numberof starts


Value calculated according todeterministic design rule

Sdet Sprob

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsImplementation strategies of Probabilistic Design

► Motivation




► Resume

Approach 2: Usage of Probabilistic Design methods to determinerisk-based safety factors for deterministic design. The designrules are updated with those new safety factors.

Aspects for realization:• Update of design rule:Ø Safety factor

• Some conservativities still need to be kept (due to simplifications)• Changes in boundary conditions of probabilistic calculations need

to be checked regularly

à Quick implementation possible (but potential for furtherimprovement)!

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsExemplary application: Introduction

► Motivation



► Exemplaryapplication► Introduction

► Mechanics

► Sensitivtiyanalysis

► Probability offailure

► Risk assessment

► Challenges

► Resume

det

1

SDDDD

c

cf

£

<+=

w P

q&q&

Background:High-temperature rotors are exposedto life-time consumption in the inletstress-relieve groove due to:• Dc: creeping (steady state effect)

and• Df: Temperature gradients during

start up and shut down limiting thenumber of start-ups (low-cycle-fatigue damage)

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsExemplary application: Mechanics

► Motivation




► Mechanics



► Risk assessment

► Challenges

► Resume

w PApproach: Simplificationby a cylinder

( ) ÷÷ø

öççè

æ××

×+×××= 322

2222 3max

tfV Rf

PRfp

prs

Centrifugal force(50/60Hz)

Torsion caused bytransmittable power

),,(max

TimeOperationTfD VC s=Þ

2014-09-26




Sensitivity analysis using Monte-Carlo-Simulation:

Exemplary application: Sensitivity analysis

► Motivation




► Mechanics



► Risk assessment

► Challenges

► Resume

Calcu-lationmodel

Rf (Geometry)

P (Power)

T (Temperature)

(MaterialProperty)

),(max

Tf Vs

CD

Obs

erve

dfre

quen

cy

CD

Normed share of rangcorrelation coefficientaccording to Spearman:

Material Property(Variation of creeprupture strength)

Temperature

2014-09-26




Application to a statistical load regime:

Values used fordeterministic

designPD

F

Design target1)

Creep Damage Fraction Dc

Exemplary application: Probability of Failure

► Motivation




► Mechanics



► Risk assessment

► Challenges

► Resume

Share of valuesabove Sdet (here:

1.3%)

Sig_min = 0.8*Sig_median (Mpa)T_nom = 550ºC1)

1): Siemens-internalvalues differ from thesevalues!

detS

2014-09-26




Application to a statistical load regime:

Values used fordeterministic

designPD

F

Design target1)



► Motivation




► Mechanics



► Risk assessment

► Challenges

► Resume

Share of valuesabove Sdet (here:

3.8%)

Sig_min = 0.8*Sig_median (Mpa)T_nom = 550ºC1)

1): Siemens-internalvalues differ from thesevalues!

detS probSSx =× det

2014-09-26




Curves of equal risk level 1)


► Motivation




► Mechanics



► Risk assessment

► Challenges

► Resume

1

2

3

Scenario

Scenario

Scenario

4Scenario

1

2

3 4

1): Siemens-internal value differs from this value!

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsExemplary application: Probability of Failure

► Motivation




► Mechanics



► Risk assessment

► Challenges

► Resume

Challenges:• Adequate determination of the input variations and frequent

verification of changes in the variations• Development of efficient and problem-specific simulation

methods• Determination of an acceptable risk level (Health&Safety,

financial)

2014-09-26



Probabilistic assessment of the potential creepdamage at the surface of steam turbine rotor shaftsResume

► Motivation




► Resume

Resume:

• Presentation of limitations ofdeterministic mechanicaldesign rules

• Introduction of two implementationstrategies of Probabilistic Design

• Exemplary application: Analysis of thepotential creep damage at steamturbine rotor shaftsØ Sensitivity analysisØ Probability of FailureØ Risk assessmentØ Challenges of the approach

w P

Turbine Model

Material

Geometry

Load regime

Turbine life timeconsumption(e.g. creep-damage)

PoF (Probabilityof Failure)

e.g. Operationtime


PDF

Design target


Inner of rotor: 1/S1/S

Sdet

Sprob

detS probSSx =× det

probabilistic assessment of the potential creep damage at ...ø faster start-ups ... damage at...

Documents