CFDPAPERBHEL

8/7/2019 CFDPAPERBHEL

1/10

1

2nd

National Conference on CFD APPLICATIONS IN POWER AND INDUSTRY SECTORSCentre of Excellence for CFD, BHEL

28th

29th

January 2009, Hyderabad, INDIA

ARANIRD001

ESTIMATION OF EFFICIENCY OF LOW PRESSURE STEAM TURBINE BLADINGUSIND CFD TECHNIQUE

KCH PeraiahResearch & Development, ARANI POWER

SYSTEMS LTDHyderabad, A.P., India

Dr. Udai SinghUKS Consultants

Trendles, Ryme Intrinseca, Sherborne DT9 6JXEngland

Ramu MaddiResearch & Development, ARANI POWER


Sumalatha HResearch & Development, ARANI POWER


ABSTRACT

The flow in a Low Pressure Steam Turbine,carrying longer blade passages, is complexand involves understanding of energyconversion in three dimensional geometries.To develop better performing blades, it isessential to identify the losses generatingmechanism and study their influence andeffects on performance. This paper outlinesdesign considerations and the estimation ofefficiency of Low Pressure Steam Turbineusing CFD, thus aiding in optimizing thedesign and helps in integrating CFD into thedesign process itself. The CFD results are inconcurrence with the two dimensional meanline code & stream line curvature method.

INTRODUCTION

During the course of expansion of steam inturbines, the state path crosses the saturationline and hence subsequent turbine stages(mostly last stages i.e. Low Pressure stages)operate with wet steam. Since the pressureis low, the volumetric flow is more, to

accommndate increased volumetric flow[1],it is necessary to have larger area leading tolong blades and these stages have lowerthermodynamic efficiencies than thoseoperating in the superheated region. Modernturbomachinery designs aim to increaseblade loading and pressure ratio whilemaintaining the same high efficiency level.This results in a higher power density andlower part count and therefore lower cost. Inthis perspective, secondary flows and theinteractions between leakage, radialflows,mainstream flow seperation and flowreversals[2] contribute considerably to theoverall turbine losses. The recentdevelopments of the CFD code

performance in terms of accuracy,sensitivity and efficiency, enables to reducethe design cycle by coupling CFD codeswith optimization tools.

Need for 3D design

Most of the phenomena involoved in turbomachinery flow can be understood andpredicted on a two-dimensional or quasi


2/10

2

three-dimensional basis, but some aspects ofthe flow(refer Need for CFD indesign) mustbe considered as fully three-dimensional(3D). Mostly 2D and quasi 3D methods arevery useful in the preliminary designprocess. When it comes to the Low Pressure

blading, the blade heights are high the flowis complex and 3D in nature. Thisnecessiates 3D design for the Low pressuresteam turbine.

Need for Twist

In general, if the Blade Height to Hubdiameter is greater than 0.14, we may haveto go for the twist of the Blade. Acrossradial height of the blade the change inperipheral velocity changes the velocity

triangle along the span of the blade. Thus tocompensate this and to maintain optimumincidence and deviation angles at all radii, itmay be necessary to go for Twist of theblade.

Fig. 01 Schematic diagram of twisted profile

In deciding the twist additionallyboth the Radial Equilibrium and Free-Vortextheories are considered. The Figure 01shows the schematic diagram of twistedprofile from top view.

Basic Losses in Turbomachinery

It is assumed that the overall ioss in amachine blade row is the sum of a numberof basic loss components. It is usuallyassumed that these can be evaluatedindependently and then combined linearly to

obtain the total loss. The basic losscomponents are usually listed as follows:

Profile loss: loss due to theboundary layer growth on the blade pressureand suction surface in a uniform two-

dimensional flow plus the trailing-edge loss.

Secondary loss: The secondarylosses arise from complex three-dimensionalflows set up as a result of the rolling up ofthe endwall boundary layer, and associatedstreamwise vorticity and separation of theflow. This formation of losses is shown inFig 02. This interaction between the flowthrough the blade passage and the end wallboundary layer results in a total pressureloss which is often a significant fraction ofthe total loss for low aspect ratio turbines.

Tip clearance loss: The tip clearanceloss arises due to the required gap betweenthe end of a rotor blade and the adjacentwall. For the two types of blades generallyencountered, i.e. unshrouded and shroudedblades, the loss generating mechanisms arequite different. In the case of shrouded rotorblades, the clearance flow is separated fromthe flow through the blading. Losses arise

then from energy dissipation across theshroud and from mixing of the clearanceflow with the main flow downstream of theblade row. For unshrouded rotor blades, aclear separation between secondary andclearance losses is not possible. In this casethe secondary loss obtained at zeroclearance is generally assumed to remainunaltered as the clearance gap is increased.The tip clearance loss is then taken as thedifference between total losses with andwithout clearance. However, it is obvious

that secondary loss will be affectedsomewhat by the clearance. Other lossesinclude disc friction losses, windage lossesand wetness losses etc.

All the above correlations express losses asa function of key geometric and flowparameters.


3/10

3

Fig. 02 Formation of Secondary losses

NEED FOR CFD IN DESIGN

The goal of the aerodynamic designprocess, for turbomachinery components, isto minimize aerodynamic losses andmaximize performance, within thegeometric, physical, and economicconstraints placed on the component. Thisgoal is accomplished through a process thatconsists of two primary phases: preliminarydesign and detailed design. The preliminary

design phase establishes the overallcharacteristics of the component, such that itwill satisfy the requirements and constraintsof the overall turbine design. Basic flow-path configurations, blade counts, blade-rowspacings, and initial blade shapes are amongthe characteristics determined duringpreliminary design wherein structuralrequirement are also considered. Theprocess is highly iterative, due to the largenumber of component and flow-pathfeatures that must be optimized, through the

analysis of many configurations.In contrast, the detailed designprocess focuses on one or a small number ofdesign configurations that offer the optimumcombination of features, and the best matchwith aerodynamic performance objectives,based on the analyses of the preliminarydesign. The objective of the detailed designprocess is to predict, as realistically as

possible, those characteristics of the flowthat are critical to the aerodynamicperformance of the turbomachinerycomponent being analyzed. Suchcharacteristics would include tip clearanceflows, shockboundary-layer interactions,

bladeend-wall interactions, flow

separations, wakes, and any other regionsof high loss. The level of detail andcapability required of a particular flowmodel will be determined by the phase ofthe design process to which it is applied.During the preliminary design process,simplifying assumptions are typically madethat allow the flow to be modeled in lessdetail. Once the overall characteristics of aparticular design have been establishedusing these 2D and quasi 3D tools, then thedetailed behavior of the flow for thatconfiguration must be determined, using thefull capabilities of the available CFDanalysis tools.During the detailed design process, designfeatures that are specifically three-dimensional in nature are being evaluated.Characteristics that can significantly impactcomponent performance, such as bladetwist, blade metal angles, blade throat,

Leading and trailing edge radius, blade

position, and tip clearance height, must beoptimized, based on their influence on aviscous flow field. To evaluate such issuesin a test rig would be time-consuming.Therefore, the analytical tools of thedetailed design process must model the flowphysics with sufficient accuracy to assist inthe evaluation process. To achieve thenecessary level of accuracy, a 3D viscoustransonic flow model must be employed inthe analysis tools. The application of such amodel will result in solution times several

orders of magnitude longer than those forpreliminary design tools. However, theexpense in terms of computation time can beoffset by reductions in bothdesign/development cycle time and designrisk. By employing 3D viscous flow analysistools during the detailed design, thenecessity of performing one or more


4/10


5/10

5

Fig 03. Geometric model of LP Blading

The approach taken to describe the flow-path geometry will be determined, in part,by the complexity of the flow-path boundaryshapes. Since the bladed disks are axi-symmetric in nature, modeling of single

blade row with necessary boundaryconditions will be sufficient. The techniqueselected for the geometric description andmodeling of components should offersufficient flexibility to accommodate designchanges.Because component and flow-pathshapes constantly evolve during the designprocess, modification of the geometricmodel of the flow-path boundaries must alsobe easily accomplished, and the updatedmodel must be made rapidly available.These requirements necessiates to utilize the

geometric modeling capabilities of acomputer-aided design (CAD) system. Weused relatively automated scheme for thedescription and modification of componentgeometry. Parametric representations of themodel and tabular representation of the flowpath coordinates at a selected streamwisecalculations are used. This faciliates theflexibility for any flow path changes infuture. Another issue that must beconsidered in the description of flow-pathgeometry is that of the transfer of consistentinformation to other functional groupsduring the design process. Particularly in aconcurrent engineering [5] environment, itis essential that all functional organizationshave access to and work with the samemaster geometric model of an object orflow path.

GRID GENERATION

After the geometric model of theturbomachinery component has beenestablished, the next step in the process ofcommunicating this configuration to the

CFD analysis program is to define thecomputational grid within the physicaldomain. The boundaries of this region aretypically defined by the flow-path surfaces(end walls, blades, etc.) and by the periodicboundaries between blade passages, whereappropriate.

Fig. 04 Grid Generation of model

Inlet and exit boundaries areestablished at points upstream anddownstream, where the necessary flow

conditions are assumed to be known.Within this region, a three-

dimensional computational grid is applied,such that the governing equations will besolved at every point on the grid, or withinevery cell formed by the grid. The gridimposed on the physical domain mustconform to the boundaries of that domainand must provide adequate resolution in allareas of the flow field to permit accurateprediction of the flow behavior. The meshwas generated as shown in Fig. 04. And thegrid independnce check was done and theoptimum number of elements found forstator or rotor are aound 2,40,000, and they+ value [6] also considred. In our modelwe generated structured mesh. The typicalmesh pattern aound blade profile is shownin Fig.05.


6/10

6

Fig. 05 Mesh pattern aound the profileSufficient number of nodes are

taken around the blade profile and wall

boundaries to caputre wall boundary layereffect.

PRE and POSTPROCESSING

A significant percentage of the total timespent on a CFD analysis is involved in pre-and postprocessing activities. Both the setupfor an analysis and the evaluation of resultsrequire considerable effort on the part of thecomponent designer. Therefore, the use ofsoftware tools to automate or facilitate theseactivities has the potential to substantiallyreduce the time required for the analysis andimprove the overall efficiency of theprocess. Preprocessing involves thedefinition of the boundaries of meshelements and interfaces between rotor andstator etc. The basic boundary conditionsused for an element was shown in Fig 06.And for inflow boundary total temperature,total pressure and fiow angle are given.

Fig. 06 Boundaries on mesh volumeFor outlet flow boundary Static pressureoutlet model was slected. The basicboundaries and inputs given in CFD modelare given below.

SIMULATION TYPE : STEADY

HEAT TRANSFER MODEL: TOTAL ENERGY

TURBULENCE MODEL: k epsilon

INLET: TOTAL PRESSURE

TOTAL TEMPERATURE

TURBULENCE LEVELWETNESS FRACTION

OUTLET: STATIC PRESSURE

MATERIAL: STEAM

SPEED in rpm

And the above problem was run for theconvergence level of 10-4 in 450 iterations.Post processing function for a CFD analysisprovides the necessary information ofvariables such as pressure, temperature,velocity and enthalpy etc. in the flow

regime. And the important parameters arecalculated and tabulated in the Table 1.From postprocesing we caluculatedefficiences.The postprocessing function fora CFD analysis provides a numerical flowvisualization capability. This facility isessential for understanding and interpretinganalysis results that consist of number ofdependent variables at thousands of discrete


7/10

7

locations within the flow field. For theseCFDvisualization tools to be useful to thecomponent designer, they must be highlyinteractive, and also user-friendly, requiringonly minimal training. To obtain usefulinformation in a 3D environment, the

visualization tool must have the capabilityto display color contours of scalar propertieson flow-path and component surfaces, andalso on user- specified, arbitrarily orientedslices through the flow path. Vectorproperties at discrete grid nodes or cells arebest represented by arrows, oriented in thedirection of the vector, and scaled to itsmagnitude.

Additional scalar property information maybe communicated in a vector display, bysuperimposing a color scale on the vectorsthemselves. Particle traces may be generatedto visualize flow behavior, by positioningrakes at user- selected locations in theflow path. Particles are then released from

these rakes, and their paths are determinedby integrating through time, given the localvelocity vector distribution. The Figures 07and 08 shown the flow properties in thedomain.

Fig.07 Mach number distribution in stage


8/10

8

Fig.08 Streamlines shown in flow domain


9/10

9

Quantity LP BLADING Diff %

2-D CODE CFD

Inelt Temperature, (deg C) 108.75931 109.294 0.4916

Inlet Pressure, bar abs 1.3739412 1.404 2.1878

MassFlow Rate, kg/s 24.938417 24.85 -0.3545Rel. Exit Mach Number 1.2000378 1.167 -2.7531

Absolute Exit Pressure (bar) 0.098132595 0.0974593 -0.6861

Exit Temperature (deg C) 45.46455 44.752 -1.5673

Dryness Fraction 86.74959 88.2 1.6720

Stage Efficiency, % 84.5 83.68 -0.9836

Actual Total Enthalpy Drop, kJ/kg 218.87 218.46 -0.1873

Blade Row Work, KW 5456.6 5428.9 -0.5076

Exit Relative Velocity m/sec 453.23184 464.2 2.4200

Exit Axial Velocity, m/sec 232.75327 229.65 -1.3333

Exit Absolute tangential Velocity,m/sec -44.22278 -43.9 -0.7299

Exit Sonic Velocity, m/sec 377.6813 378.23 0.1453Exit Density, kg/m^3 0.077171996 0.07694 -0.3006

Table 01 : Comparison of CFD results with 2D code

RESULTS AND CONCLUSION

Subsequently the aerodynamic losses for CFD arecorelated with 2D code and quasi 3D streamlinecurvature method. It gives the validation of CFD model.The proceedure, boundary conditions and the CFD model

can be utilized for the Design Optimization process.

REFERENCES

[1] Earl Logan, Jr.Ramendra P. Roy, Handbook ofTurbomachinery, Second Edition.

[2] J.S.Rao, K. Ch Periah et. al., Estimation of DynamicStresses in Last Stage Steam Turbine Blades underReverse Flow Conditions, International Conference onVibration Engineering & Technology of Machinery,VETOMAC-IV, December 17-19, 2007.

[3] F. J. Malzacher et.al., Aerodesign and Testing of anAeromechanically Highly Loaded LP Turbine, Journal ofTurbomachinery, OCTOBER 2006 Vol 128 / 643,Transcations of ASME.

[4] Paul Traub et. al., NUMERICAL INVESTIGATIONSFOR OPTIMIZING THE AERO-ACOUSTICALDESIGN OF MODERN LP-TURBINES, ICSV13, July2-6, 2006, Vienna, Austria.

[5] Richard A. Layton et. al., Conceptual Basis for a NewApproach to Bladed-Disk Design, Journal of Engineeringfor Gas Turbines and Power, 324/ Vol.122, APRIL 2000,Transcations of ASME.

[6] Stephane Burguburu et. al., Numerical Optimizationof Turbomachinery Bladings Journal of TurbomachineryJANUARY 2004, Vol. 126 91, Transcations of ASME


10/10

10

Documents

CFDPAPERBHEL