Feasibility study of large gas turbine outer casing bolts ...
Transcript of Feasibility study of large gas turbine outer casing bolts ...
Security classification © Siemens AG 2013 All rights reserved. http://siemens.com/energy/power-generation/gas-turbines
Intern © Siemens AG 2015 All rights reservedFeasibility study of large gas turbineouter casing bolts with optiSlang
November 2015
Uwe Lohse Siemens AG Large Gas Turbines PG GT LGT EN DI DO EUBurkhard Voss Siemens AG Large Gas Turbines PG GT LGT EN SC CASAnton Enns ITB Dortmund
Intern © Siemens AG 2015 All rights reserved
ConfidentialPage 2 WOST 2015
ECC
Technology not specified in the CCL, AL or ECL
Technical Classification: ECCN: EAR99 AL: N ECL: N US-Content:
This document is subject to U.S. and/or national export regulations. Unauthorized diversion may be prohibited. In cases of doubtcontact your Export Compliance Department.
Dieses Dokument unterliegt nationalen und/oder U.S. Ausfuhrvorschriften. Die Verbreitung kann eine Genehmigung benötigen. ImZweifel wenden Sie sich an Ihre ECC Abteilung.
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 3 WOST 2015
Introduction
Content
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
44
33
22
11
optiSlang model and results
Calibration of FE-Model by measurements
Robust design
55 Summary
ConfidentialPage 4 WOST 2015
Large scale Siemens Gas Turbines:The right engine for every power category
50H
z
SGT5-8000H
SGT5-4000F
SGT5-2000E
307 MW
172 MW
400 MW
SGT6-8000H
114 MW
SGT6-5000F
SGT6-2000E
296 MW
242 MW
60H
z
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 5 WOST 2015
Where we are – Gas Turbine Locations & JointVenture Partners
JV location
Gas Turbines location
Muelheim, Germany
SGTT,St. Petersburg, Russia
SEPG/SGTP,Shanghai, China
Erlangen, Germany
Charlottesville, USA
Gurgaon, India
Shanghai, China
Jupiter, USAOrlando, USA
Charlotte Plant, USA
Berlin Plant, Germany
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 6 WOST 2015
SGT5-8000H during assembly at Berlin plant
EfficiencyGT 40%GUD >60%With district heating~85%
PowerGT 400MWCCPP 600MW
Weight ~390tLength ~13,1 mDiameter ~4,9 m
Fleet > 74 units
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 7 WOST 2015
Key thermodynamic values for bolting design
-40°C +50°C
~440°C >10 bar
~580° to ~650°C
Compressor section
Combustion section
Turbine section
Exhaust section
SUCK
SQUIZEBANG
BLOW
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 8 WOST 2015
Introduction – Motivation
§ The estimation of the pretension loss due to bolt temperature shows astrong effect of the temperature difference between bolt / flange
§ The temperature difference is direct proportional to the low cyclefatigue of the bolts
§ To identify the major influences a study by use of optiSlang wasperformed
§ A second goal of this study was to show the feasibility to use optiSlangin the daily design process ( optimization, robust design )
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 9 WOST 2015
Introduction – Procedure
Following steps are performed
1. Develop a simplified parametric FE-model
2. Define design space for estimated parameter
3. Run a DOE to find a meta model
4. Evaluate the meta model to reduce the model to significant parameters
5. Perform a calibration of the FE-model based on measurements
6. Investigate the robustness of the design
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU.
ConfidentialPage 10 WOST 2015
Introduction
Content
44
33
22
11
optiSlang model and results
Calibration of FE-Model by measurements
Robust design
55 Summary
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 11 WOST 2015
Parameterization and OptimizationFE-Model
FE-Model
Temperature
optiSLang
Target Value
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
Geometry Contact
Temperature over time Material Properties
Input Parameter
HTC shaft to hole
ConfidentialPage 12 WOST 2015
ParameterizationGeometry
Parametricdimensions
Dimensions driven byparametric dimensions
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
• The parametric geometrical FE model can be defined by two parameters only
ConfidentialPage 13 WOST 2015
ParameterizationThermal Contact
HTC1 nut to bolt
HTC3 flange to washer
HTC2 nut to washer
HTC4 shaft to hole
HTC1 to HTC4 are parameters
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 14 WOST 2015
ParameterizationSummary FE-Model
Action Geometry ContactProperties
Boundaryconditions
Material properties
ValuesConstant
Valuesto depend on temperature
- -
Valuesto depend on time
-
Workbench function, low effort
Small APDL commands, low effort
APDL commands, high effort
§ In GeneralFor all input data used in this example a parameterization is possible.However, with different effort.
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 15 WOST 2015
Input Parameters for the Sensitivity Analysis
Number Description Variable RangeMinimum
RangeMaximum
Unit
1 radius R 20 60 mm
2 flange-Height H 100 400 mm
3 HTC bolt to nut HTC_bolt_nut 1 000 50 000 W/m² C
4 HTC nut to flange HTC_nut_flange 1 000 50 000 W/m² C
5 HTC washer to flange HTC_washer_flange 1 000 50 000 W/m² C
6 HTC shaft to hole HTC_shaft hole 10 50 000 W/m² C
7 thermal conductivitybolt material
lambda_bolt 25 40 W/mK
8 thermal conductivityflange material
lambda_flange 28 35 W/mK
9 specific heatbolt / flange material
sp_heat_flange 500 600 J/kgK
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU.
ConfidentialPage 16 WOST 2015
T1 T2
ResultsOutput Parameter
max. DT
§ Output Parameter:Maximum temperature difference DTmax between T1 (bolt)and T2 (flange) to be evaluated by APDL or Enhanced Tool Kit (ETK)
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 17 WOST 2015
ResultsDOE
§ The sensitivity analysis is driven by optiSlang§ 100 Design Points are calculated§ Every Design Point is one correlation of input parameters in the
given ranges§ With the calculated design points the meta-model is created
...
...
optiSLang – Input Parameters Workbench Design Points
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 18 WOST 2015
ResultsDOE
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
Significant influence of inputparameters on Output Parametermy_Delta_Temp:§ Bolt radius 55%§ HTC shaft to hole 34%§ Flange height 11%
§ Negligible important inputparameters were automaticallyfiltered by optiSlang and notconsidered in the CoP.
ConfidentialPage 19 WOST 2015
ResultsDOEDT (bolt radius; HTC shaft to hole)
Design Point (DP1) withDTmax = 280.1 °CR = 51.4 mmH = 350.5 mmHTC shaft to hole = 16 W/m²K
Design Point (DP85) withDTmin = 105.6 °CR = 21 mmH = 212.5 mmHTC shaft to hole = 76 W/m²K
• Surface plot shows also points outside of CoP 99%• Formula is inside of optiSlang
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 20 WOST 2015
• Surface plot shows also points outside of CoP 99%• Formula is inside of optiSlang
Design Point (DP1) withDTmax = 280,1°C
R = 51,4 mmH = 350,5 mm
HTC shaft to hole = 16W/m²K
Design Point (DP85) withDTmin = 105,6°C
R = 21 mmH = 212,5 mm
HTC shaft to hole= 76W/m²K
ResultsDOEDT (flange height; bolt radius)
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 21 WOST 2015
• Surface plot shows also points outside of CoP 99%• Formula is inside of optiSlang
Design Point (DP1) withDTmax = 280,1°C
R = 51,4 mmH = 350,5 mm
HTC shaft to hole = 16 W/m²K
Design Point (DP85) withDTmin = 105,6°C
R = 21 mmH = 212,5 mm
HTC shaft to hole = 76W/m²K
ResultsDOEDT (flange height; HTC shaft to hole)
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 22 WOST 2015
Introduction
Content
44
33
22
11
optiSlang model and results
Calibration of FE-Model by measurements
Robust design
55 Summary
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 23 WOST 2015
Detailed calibration of the HTC(variation of the HTC value over time)
• Instead of one HTC value over time, the time vector is divided in several sections• For every section one HTC value is defined (HTC over time)• Purpose is to minimize the deviations between the measurement and the analysis for
the bolt
HTC 1 HTC 2 HTC 3 = HTC (t)
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU.
ConfidentialPage 24 WOST 2015
Detailed calibration of the HTC in the bolt cavity(variation of the HTC value over time)
• Process chain for thecalibration of themeasurement data is shown
• This process will be donethree times, for every sectionof the bolt temperature - curve
• In the ETK one section is selected• On the next pages the results for every
section are presented
HTC
Bolt-temperatureover time
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 25 WOST 2015
Calibration of FE-ModelResults for BoltExample for overall calibration
Coefficient of Prognosis = 100 %The Approximation is very fine.
x-axis:HTC value
y-axis:difference betweensignal and signalreference
Best design
Approximation of results
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU.
ConfidentialPage 26 WOST 2015
Detailed calibration of the HTC in the bolt cavity(variation of the HTC value over time)
Summary• The variation of the HTC value
over time is possible withANSYS and opitSlang, bydividing the time vector inseveral sections
• the calibration of themeasurement date with HTC(t)shows improved results
• for a higher improvement a finerresolution of the time vector andmore sections are required
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 27 WOST 2015
Comparison of Calibration with Measurement
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 28 WOST 2015
Introduction
Content
44
33
22
11
optiSlang model and results
Calibration of FE-Model by measurements
Robust design
55 Summary
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 29 WOST 2015
Result of Robust Design Study / Sensitivity study
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
Distribution for HTC
Distribution for R
Distribution for H
Result of Distributions
ConfidentialPage 30 WOST 2015
Available plots of the Sensitivity study in optiSlang
Will be probably available inthe next optiSlang version
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 31 WOST 2015
Vision
§ Often the designer are asked during a meeting about the influence of parameter to thedesign. The second question deals with the release of budget and resources, if apotential change of a parameter makes sense.
§ Actual the designer has no tool for fast estimations, so he has to answer this questionlater on or makes statements based on his experiences.
§ This may lead to loss of time, budget and reputation of the designer.
Ø The vision is to have a simplified meta model of the significant parameter as Excel orMat lab available, e. g. for meetings, for a very fast and first estimation of the effect ofa proposed change of the parameter.
Ø This helps to save money and accelerates the development time.
Ø The final optimization will be made again with ANSYS and optiSlang!
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU.
ConfidentialPage 32 WOST 2015
MOP Solver in ExcelActual available solution in optiSlangfor the vision
Excel-file
Import 1Data Set from optiSLang:Input and OutputParameters
Import 2Meta model based on thedata setMoP solver required asMacro in EXCELMoP License required!
Table template
New input parameters• Values within chosen range• There is an option to
extrapolate (notrecommended due todeviations to the meta model)
New output parameters• are calculated from the
meta model• Small deviations to the
data set results due to theapproximation
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 33 WOST 2015
Introduction
Content
44
33
22
11
optiSlang model and results
Calibration of FE-Model by measurements
Robust design
55 Summary
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 34 WOST 2015
Summary I/II
§ The study shows the significant input parameter of the investigated designparameter and design space very fast. The definition of the parametric modeltakes the most time!
§ The results show a very good Coefficient of Prognosis (COP) For the choosenparameter.
§ In any case, if possible, the model should validated, by testes. The goal of thisto check that the chosen parameter covers all significant impacting parametersis considered in the DOE.
§ The data fit with the meta model to real measurements shows a goodagreement. This is the additional hint that all major parameters were consideredin the study.
§ A validation of the identified meta model due to measurements is stronglyrecommended.
§ Robust design study is easily possible with the meta model. Direct input of theparameter distribution in Workbench is possible.
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
ConfidentialPage 35 WOST 2015
Summary II/II
§ Major hurdles are the definition of the parametric FE-model.
§ For the nonlinear material properties and the transient temperature and heattransfer coefficients a script with APDL must be written.
§ A direct input in ANSYS Workbench is not feasible. APDL knowledge isrequired.
§ Distribution of the significant parameter have to be known for the robust designstudy.
§ To reduce the data fitting on an interesting section of the transient, ETK has tobe used.
§ In the transient case the average value over the whole transient is possible tofit. Perhaps calibration of several segments are required, if jumps or kinksfound in the measurement.
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 36 WOST 2015
Potential further projects in discussion
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
Optimization on differentcomponents regarding weight,stress, low cycle fatigue,deformation, creep
Probabilistic study of differentoperation behavior. Regarding lowlife cycle fatigue
Probabilistic study random casingtolerances regarding low life cyclefatigue
ConfidentialPage 37 WOST 2015
Uwe Lohse
Siemens AGLarge Gas TurbinesPG GT LGT EN DI DO EUMellinghofer Str. 5545473 Mülheim an der Ruhr, DeutschlandTel.: +49 208 456-4062Mobil: +49 1522 8874918mailto:[email protected]
Burkhard Voss
Siemens AGLarge Gas TurbinesPG GT LGT EN SC CASMellinghofer Str. 5545473 Mülheim an der Ruhr, DeutschlandTel.: +49 208 456-2214Mobil: ++49 (172) 1333677mailto:[email protected]
Anton Enns_________________________________________________________
ITBIngenieurgesellschaft für technische Berechnungen mbHEuropaplatz 7, 44269 Dortmund
Tel.: +49 (0)231 94 53 65 - 23Fax: +49 (0)231 94 53 65 - 11
Geschäftsführer: Dr. Frank BrehmerHandelsregister Dortmund: HRB 16440
www.itb-fem.de
AcknowledgmentsPaul Levitan ITBFrank Bremer ITBVeit Bayer Dynardo
Contact
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.
ConfidentialPage 38 WOST 2015
Thanks for your attention!
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU.
ConfidentialPage 39 WOST 2015
Disclaimer
This document contains forward-looking statements and information – that is, statements related to future, not past,events. These statements may be identified either orally or in writing by words as “expects”, “anticipates”, “intends”,“plans”, “believes”, “seeks”, “estimates”, “will” or words of similar meaning. Such statements are based on our currentexpectations and certain assumptions, and are, therefore, subject to certain risks and uncertainties. A variety of factors,many of which are beyond Siemens’ control, affect its operations, performance, business strategy and results and couldcause the actual results, performance or achievements of Siemens worldwide to be materially different from any futureresults, performance or achievements that may be expressed or implied by such forward-looking statements. For us,particular uncertainties arise, among others, from changes in general economic and business conditions, changes incurrency exchange rates and interest rates, introduction of competing products or technologies by other companies,lack of acceptance of new products or services by customers targeted by Siemens worldwide, changes in businessstrategy and various other factors. More detailed information about certain of these factors is contained in Siemens’filings with the SEC, which are available on the Siemens website, www.siemens.com and on the SEC’s website,www.sec.gov. Should one or more of these risks or uncertainties materialize, or should underlying assumptions proveincorrect, actual results may vary materially from those described in the relevant forward-looking statement asanticipated, believed, estimated, expected, intended, planned or projected. Siemens does not intend or assume anyobligation to update or revise these forward-looking statements in light of developments which differ from thoseanticipated.
Trademarks mentioned in this document are the property of Siemens AG, it's affiliates or their respective owners.
Intern © Siemens AG 2015 All rights reservedUwe Lohse PG GT EN LGT DI DO EU
.