7- Fundamentals of Turbine Design

62
7 Fundamentals of Turbine Design David M. Mathis Honeywell Aerospace, Tempe, Arizona, U.S.A. INTRODUCTION Turbines are used to convert the energy contained in a continuous flow of fluid into rotational mechanical energy of a shaft. Turbines are used in a wide range of applications, in a wide variety of sizes. Large single-stage turbines are used for power generation in hydroelectric dams, while large multistage turbines are used in steam power plants. Aircraft propulsion engines (turbofans, turbojets, and turboprops) use multistage turbines in their power and gas generator sections. Other, less obvious uses of turbines for aircraft are in auxiliary power units, ground power units, starters for main engines, turboexpanders in environmental controls, and constant- speed drives for electrical and hydraulic power generation. Rocket engines use turbines to power pumps to pressurize the propellants before they reach the combustion chamber. Two familiar consumer applications of turbines are turbochargers for passenger vehicles and wind turbines (windmills). Many excellent texts have been written regarding the design and analyses of turbines [1–3]. There is also a large institutional body of knowledge and practices for the design and performance prediction of Copyright © 2003 Marcel Dekker, Inc.

Transcript of 7- Fundamentals of Turbine Design

  • 5/22/2018 7- Fundamentals of Turbine Design

    1/62

    7

    Fundamentals of Turbine Design

    David M. Mathis

    Honeywell Aerospace, Tempe, Arizona, U.S.A.

    INTRODUCTION

    Turbines are used to convert the energy contained in a continuous flow of

    fluid into rotational mechanical energy of a shaft. Turbines are used in a

    wide range of applications, in a wide variety of sizes. Large single-stage

    turbines are used for power generation in hydroelectric dams, while large

    multistage turbines are used in steam power plants. Aircraft propulsion

    engines (turbofans, turbojets, and turboprops) use multistage turbines intheir power and gas generator sections. Other, less obvious uses of turbines

    for aircraft are in auxiliary power units, ground power units, starters for

    main engines, turboexpanders in environmental controls, and constant-

    speed drives for electrical and hydraulic power generation. Rocket engines

    use turbines to power pumps to pressurize the propellants before they reach

    the combustion chamber. Two familiar consumer applications of turbines

    are turbochargers for passenger vehicles and wind turbines (windmills).

    Many excellent texts have been written regarding the design and

    analyses of turbines [13]. There is also a large institutional body ofknowledge and practices for the design and performance prediction of

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    2/62

    power plant and aircraft propulsion engine turbines. Here we make no

    attempt to cover these areas. The purpose of this text is to familiarize a

    mechanical or aerospace engineer who does not specialize in turbines with

    basic turbine design and analysis. The emphasis will be on smaller turbines

    for applications other than propulsion or electrical power generation.Further restricting our emphasis, detailed design activities such as geometric

    specification of blades, vanes, etc. will not be covered. Our intent is to give

    the system engineer the necessary information to choose the correct type of

    turbine, estimate its performance, and determine its overall geometry

    (diameter, blade height, and chord).

    This chapter will first cover those equations and concepts that apply to

    all types of turbines. Subsequently, the two main turbine types will be

    discussed, specifically, axial-flow turbines and radial-inflow turbines.

    BASIC TURBINE CONCEPTS

    Flow Through a Turbine

    Figure 1shows cross sections of generic single-stage axial-flow and radial-

    inflow turbines. The figure shows the station notation used for subsequent

    analyses. The high-pressure flow enters the turbine at station in, passes

    through the inlet, and is guided to the stator inlet (station 0), where vanes

    turn the flow in the tangential direction. The flow leaves the stator vanes andenters the rotor blades (at station 1), which turn the flow back in the

    opposite direction, extracting energy from the flow. The flow leaving the

    rotor blades (station 2), now at a pressure lower than inlet, passes through a

    diffuser where a controlled increase in flow area converts dynamic head to

    static pressure. Following the diffuser, the flow exits to the discharge

    conditions (station dis).

    The purpose of the inlet is to guide the flow from the supply source to

    the stator vanes with a minimum loss in total pressure. Several types of inlets

    are shown in Fig. 2. Most auxiliary types of turbines such as starters anddrive units are supplied from ducts and typically have axial inlets such as

    that shown inFig. 2(a)or a tangential entry like that ofFig. 2(b). The axial

    inlet acts as a transition between the small diameter of the supply duct and

    the larger diameter of the turbine. No flow turning or significant

    acceleration is done in this type of inlet. In the tangential-entry scroll of

    Fig. 2(b), the flow is accelerated and turned tangentially before entering the

    stator, reducing the flow turning done by the stator. Another type of inlet

    for an auxiliary turbine is the plenum shown inFig. 2(c). For turbines that

    are part of an engine, the inlet is typically integrated with the combustor[Fig. 2(d)] or the discharge of a previous turbine stage [Fig. 2(e)].

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    3/62

    Figure 1 Cross sections of generic single-stage turbines: (a) axial-flow turbine, (b)

    radial-inflow turbine.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    4/62

    The sole purpose of the stator is to induce a swirl component to the

    flow so that a torque can be imparted to the rotor blades. Stators are

    typically equipped with numerous curved airfoils called vanes that turn the

    flow in the tangential direction. Cross sections of an axial-flow turbine statorand a radial-inflow turbine stator are shown in Fig. 3(a) and 3(b),

    Figure 2 Common types of turbine inlets: (a) in-line axial inlet, (b) tangential-entry

    scroll inlet for axial-flow turbine, (c) plenum inlet with radial or axial entry, (d)

    turbine stage downstream of combustor, (e) turbine stage in multistage turbine.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    5/62

    respectively. Radial-inflow turbines supplied from a scroll, such as

    turbocharger turbines, often have no vanes in the stator. For turbines

    that must operate efficiently over a wide range of inlet flow conditions,

    variable-geometry stators are used, typically with pivoting stator vanes. For

    high-temperature applications, the stator vanes are cooled using lower-temperature fluid, usually compressor bleed air.

    The purpose of the rotor is to extract energy from the flow, converting

    it to shaft power. The rotor blades are attached to a rotating disk that

    transfers the torque of the rotor blades to the turbine output shaft. Like the

    stator, the rotor has a number of individual curved airfoils called rotor or

    turbine blades. The blades are angled to accept the flow from the stator with

    minimum disturbance when the turbine is operating at design conditions.

    The flow from the stator is then turned back in the opposite direction in a

    controlled manner, causing a change in tangential momentum and a force tobe exerted on the blades.Figure 4shows cross sections of generic axial-flow

    and radial-inflow turbine blades. Axial-flow rotors have been constructed

    with blades integral with the disk and with blades individually inserted into

    the disk using a dovetail arrangement. Cooling is often used for rotors in

    high-temperature applications. Exotic materials are sometimes used for both

    rotors and stators to withstand the high temperatures encountered in high-

    performance applications.

    The flow leaving the turbine rotor can have a significant amount of

    kinetic energy. If this kinetic energy is converted to static pressure in anefficient manner, the turbine can be operated with a rotor discharge static

    pressure lower than the static pressure at diffuser discharge. This increases

    the turbine power output for given inlet and discharge conditions. Diffusers

    used with turbines are generally of the form shown in Fig. 5(a)and5(b)and

    increase the flow area gradually by changes in passage height, mean radius

    of the passage, or a combination of the two. Diffusers with a change in

    radius have the advantage of diffusing the swirl component of the rotor

    discharge velocity as well as the throughflow component.

    Turbine Energy Transfer

    The combined parts of the turbine allow energy to be extracted from the

    flow and converted to useful mechanical energy at the shaft. The amount of

    energy extraction is some fraction of the energy available to the turbine. The

    following describes the calculation of the available energy for a turbine and

    assumes familiarity with thermodynamics and compressible flow.

    Flow through a turbine is usually modeled as an adiabatic expansion.

    The process is considered adiabatic since the amount of energy transferredas heat is generally insignificant compared to the energy transferred as work.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    6/62

    In the ideal case, the expansion is isentropic, as shown in Fig. 6(a) in an

    enthalpyentropy (hs) diagram. The inlet to the turbine is at pressure p0inand the exit is at p0dis. The isentropic enthalpy change is the most specific

    Figure 3 Typical stator vane shapes: (a) axial-flow stator, (b) radial-inflow stator.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    7/62

    energy that can be extracted from the fluid. Thus, if the inlet pressure and

    temperature and the exit pressure from a turbine are known, the maximum

    specific energy extraction can be easily determined from a state diagram forthe turbine working fluid. For an ideal gas with constant specific heats, the

    Figure 4 Typical rotor blade shapes: (a) axial-flow rotor, (b) radial-inflow rotor.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    8/62

    isentropic enthalpy drop is calculated from

    DhisentropiccpT0in 1 pdisp0in g1=g" # 1

    Figure 5 Turbine diffuser configurations: (a) constant mean-diameter diffuser, (b)

    increasing mean-diameter curved diffuser.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    9/62

    Where

    T0ininlet absolute total temperature:

    Opspecific heat at constant pressure:gratio of specific heats:

    Figure 6 The expansion process across a turbine: (a) idealized isentropic

    expansion, (b) actual expansion.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    10/62

    The approximation of Eq. (1) is adequate for turbines operating on air and

    other common gases at moderate pressures and temperatures. Total

    conditions are normally used for both temperature and pressure at the

    inlet to the turbine, so that the inlet pressure is correctly referred to as p0in inEq. (1).

    As discussed earlier, the actual energy transfer in a turbine is smaller

    than the isentropic value due to irreversibilities in the flow. The actual

    process is marked by an increase in entropy and is represented in the hs

    diagram of Fig. 6(b) by a dotted line. The actual path is uncertain, as the

    details of the entropy changes within the turbine are usually not known. Due

    to the curvature of the isobars, the enthalpy change associated with an

    entropy increase is less than that for an isentropic process. The degree of

    entropy rise is usually described indirectly by the ratio of the actual enthalpy

    drop to the isentropic enthalpy drop. This quantity is referred to as theisentropic (sometimes adiabatic) efficiency, Z, and is calculated from

    ZOA hinhdisDhisentropic

    2

    The subscript OA indicates the overall efficiency, since the enthalpy drop is

    taken across the entire turbine. The efficiency is one of the critical

    parameters that describe turbine performance.

    So far we have not specified whether the total or static pressure shouldbe used at the turbine exit for calculating the isentropic enthalpy drop.

    (Note that this does not affect the actual enthalpy drop, just the ideal

    enthalpy drop.) Usage depends on application. For applications where the

    kinetic energy leaving the turbine rotor is useful, total pressure is used. Such

    cases include all but the last stage in a multistage turbine (the kinetic energy

    of the exhaust can be converted into useful work by the following stage) and

    cases where the turbine exhaust is used to generate thrust, such as in a

    turbojet. For most power-generating applications, the turbine is rated using

    static exit pressure, since the exit kinetic energy is usually dissipated in theatmosphere. Note that the total-to-static efficiency will be lower than the

    total-to-total efficiency since the static exit pressure is lower than the total.

    With the energy available to the turbine established by the inlet and

    exit conditions, lets take a closer look at the actual mechanism of energy

    transfer within the turbine.Figure 7shows a generalized turbine rotor. Flow

    enters the upstream side of the rotor at point 1 with velocity V!

    1 and exits

    from the downstream side at point 2 with velocity V!

    2. The rotor spins about

    its centerline coincident with the x-axis with rotational velocity o. The

    location of points 1 and 2 is arbitrary (as long as they are on the rotor), asare the two velocity vectors. The velocity vectors are assumed to represent

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    11/62

    the average for the gas flowing through the turbine. The net torque Gacting

    on the rotor can be represented as the difference of two torques on either

    side of the rotor:

    Gr1Fy1r2Fy2 3

    where Fy is the force in the tangential direction and r is the radius to the

    point. From Newtons second law, the tangential force is equal to the rate of

    Figure 7 Velocities at the inlet and exit of a turbine rotor.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    12/62

    change of angular momentum:

    FydmVydt

    4

    Performing the derivative, assuming constant Vy and mass flow rate _mm,

    results in

    G _mmVy1r1Vy2r2 5The energy transfer per time (power) is obtained by multiplying both sides

    of Eq. (5) by the rotational velocity o:

    PGoo _mmVy1r1Vy2r2 6

    The power P can also be calculated from the hs diagram for the actualprocess as

    P _mmhinhdis _mmDhactual 7Combining Eqs. (6) and (7) and defining the wheel speed Uas

    Uor 8results in the Euler equation for energy transfer in a turbomachine:

    DhactualU1Vy1U2Vy2 9

    The Euler equation, as derived here, assumes adiabatic flow through the

    turbine, since enthalpy change is allowed only across the rotor. The Euler

    equation relates the thermodynamic energy transfer to the change in

    velocities at the inlet and exit of the rotor. This leads us to examine these

    velocities more closely, since they determine the work extracted from the

    turbine.

    Velocity DiagramsEulers equation shows that energy transfer in a turbine is directly related to

    the velocities in the turbine. It is convenient to graphically display these

    velocities at the rotor inlet and exit in diagrams called velocity or vector

    diagrams. These diagrams are drawn in a single plane. For an axial-flow

    turbine, they are drawn in the xy plane at a specific value ofr. At the inlet

    of a radial-inflow turbine, where the flow is generally in the ry plane, the

    diagram is drawn in that plane at a specific value ofx. The exit diagram for

    a radial-inflow turbine is drawn in the xy plane at a specific value ofr.

    Figure 8(a) shows the velocity diagram at the inlet to an axial-flowrotor. The stator and rotor blade shapes are included to show the relation

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    13/62

    between the velocity diagram and the physical geometry of the turbine. The

    flow leaves the stator at an angle ofa1 from the axial direction. The velocity

    vector V!

    1 can be broken into two components, Vx1 in the axial direction

    andVy1 in the tangential direction. Note that the turbine work is controlled

    by the tangential component, while the turbine flow rate is controlled by theaxial component (for an axial-flow turbine). The vector V

    !1 is measured in

    an absolute, nonrotating reference frame and is referred to as the absolute

    rotor inlet velocity. Likewise, the anglea1is called the absolute flow angle at

    rotor inlet. A rotating reference frame can also be fixed to the rotor.

    Velocities in this reference are determined by subtracting the rotor velocity

    from the absolute velocity. Defining the relative velocity vector at the inlet

    to be W

    !

    1, we can write

    W!1 V!1U1 10The vector notation is not used for the rotor velocity U1as it is always in the

    tangential direction. The relative velocity vector is also shown in Fig. 8(a).

    The relative flow angleb1is defined as the angle between the relative velocity

    vector and the axial direction. Inspection of the diagram of Fig. 8(a) reveals

    Figure 8 Velocity diagrams for an axial-flow turbine: (a) rotor inlet velocity

    diagram, (b) rotor exit velocity diagram.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    14/62

    several relationships between the relative velocity and absolute velocity and

    their components:

    V21

    V2x1

    V2y1

    11

    W21 W2x1W2y1 12Wy1Vy1U1 13Wx1Vx1 14

    The sign convention used here is that tangential components in the direction

    of the wheel speed are positive. This implies that botha1 andb1 are positive

    angles.Figure 8(b)shows the vector diagram at the outlet of the rotor. Note

    that in this diagram, both the absolute and relative tangential components

    are opposite the direction of the blade speed and are referred to as negativevalues. The two angles are also negative.

    In addition to the relative velocities and flow angles, we can also define

    other relative quantities such as relative total temperature and relative total

    pressure. In the absolute frame of reference, the total temperature is defined

    as

    T0TV2

    2cp15

    In the relative frame of reference, the relative total temperatureT00is definedas

    T00TW2

    2cp16

    The static temperature is invariant with regard to reference frame.

    Combining Eqs. (15) and (16), we have

    T00T0W2 V22cp

    17

    The relative total temperature is the stagnation temperature in the rotating

    reference; hence it is the temperature that the rotor material is subjected to.

    Equation (17) shows that if the relative velocity is lower than the absolute

    velocity, the relative total temperature will be lower than the absolute. This

    is an important consideration to the mechanical integrity of the turbine.

    As with the static temperature, the static pressure is also invariant withreference frame. The relative total pressure can then be calculated from the

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    15/62

    gas dynamics relation

    p00

    p T

    00

    T g=g1

    18

    Types of Velocity Diagrams

    There are an infinite number of variations of the velocity diagrams shown in

    Fig. 8. To help distinguish and classify them, the vector diagrams are

    identified according to reaction, exit swirl, stage loading, and flow

    coefficient. The reaction is the ratio of the change in static enthalpy across

    the rotor to the change in total enthalpy across the stage. In terms ofvelocities, the change in total enthalpy is given by Eq. (9). The change in

    static enthalpy (denoted as hs) can be found from

    hs1hs2 h1V21

    2

    h2V

    22

    2

    U1Vy1U2Vy212V21V22 19

    Geometric manipulation of the vector diagram of Fig. 8 results in

    UVy12V2 U2 W2 20

    Applying to Eqs. (9) and (19), the stage reaction can be expressed as

    Rstg U21U22 W21W22

    V21V22 U21U22 W21W22 21

    Stage reaction is normally held to values greater than or equal to 0. For an

    axial-flow turbine with no change in mean radius between rotor inlet androtor outlet,U1U2 and the reaction is controlled by the change in relativevelocity across the rotor. Negative reaction implies that W1 >W2,indicating diffusion occurs in the rotor. Due to the increased boundary-

    layer losses and possible flow separation associated with diffusion, negative

    reaction is generally avoided. Diagrams with zero reaction (no change in

    magnitude of relative velocity across the rotor) are referred to as impulse

    diagrams and are used in turbines with large work extraction. Diagrams

    with reactions greater than 0 are referred to as reaction diagrams. Stage

    reaction is usually limited to about 0.5 due to exit kinetic energyconsiderations.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    16/62

    Exit swirl refers to the value of Vy2. For turbines discharging to

    ambient, the most efficient diagram has zero exit swirl. While a negative

    value of exit swirl increases the work extraction, the magnitude of the

    turbine discharge velocity increases, leading to a larger difference between

    the exit static and total pressures. For turbines rated on exit static pressure,the tradeoff between increased work and lower exit static pressure results in

    lower efficiency levels. Most turbines operating in air with pressure ratios of

    3:1 or less use zero exit swirl vector diagrams.

    The stage loading is measured by the loading coefficientl. The loading

    coefficient is defined here as

    l DhactualU2

    22

    which can also be written as

    l DVyU

    23

    for turbines with no change in U between inlet and outlet. The loading

    coefficient is usually calculated for an axial-flow turbine stage at either the

    hub or mean radius. For a radial-inflow turbine, the rotor tip speed is used

    in Eq. (23).

    The stage flow is controlled by the flow coefficient, defined as

    fVxU

    24

    These four parameters are related to each other through the vector diagram.

    Specification of three of them completely defines the vector diagram.

    Figure 9 presents examples of a variety of vector diagrams, with exit

    swirl, reaction, and loading coefficient tabulated. Figure 9(a) shows a vector

    diagram appropriate for an auxiliary turbine application, with relatively

    high loading (near impulse) and zero exit swirl. A diagram more typical of a

    stage in a multistage turbine is shown in Fig. 9(b), since the exit kinetic

    energy can be utilized in the following stage, the diagram does show

    significant exit swirl. Both Fig. 9(a) and 9(b) are for axial turbines; 9(c) is the

    vector diagram for a radial-inflow turbine. The major difference is the

    change in Ubetween the inlet and exit of the turbine.

    Turbine Losses

    The difference between the ideal turbine work and the actual turbine work ismade up of the losses in the turbine. The losses can be apportioned to each

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    17/62

    Figure 9 Variations in turbine velocity diagrams: (a) axial-flow diagram for single-

    stage auxiliary turbine, (b) axial-flow diagram for one stage in multistage turbine, (c)

    radial-inflow turbine velocity diagram.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    18/62

    component so that we may write

    Dhideal DhactualLinletLstatorLrotorLdiffuserLexit 25where the L terms represent losses in enthalpy in each component. Losses

    can also be looked at from a pressure viewpoint. An ideal exit pressure can

    be determined from

    DhactualcpT0in 1 pdisideal

    p0in

    g1=g" # 26

    The component losses are then represented as losses in total pressure, the

    sum of which is equal to the difference between the actual and ideal exit

    pressure:

    Pdis pdisideal Dp0inlet Dp0stator Dp00rotor Dp0diffuser Dp0exit 27Most loss models incorporate the pressure loss concept.

    Inlet Losses

    Losses in inlets are usually modeled with a total pressure loss coefficient Ktdefined as

    Dp0inlet Ktinlet1

    2rV2inlet

    28

    Where

    Vinlet velocity at the upstream end of the inlet:rdensity of the working fluid:

    The losses in an inlet primarily arise from frictional and turning effects.

    Within packaging constraints, the inlet should be made as large as possibleto reduce velocities and minimize losses. Axial inlets such as that ofFig. 2(a)

    have low frictional losses (due to their short length and relatively low

    velocities), but often suffer from turning losses due to flow separation along

    their outer diameter. Longer axial inlets with more gradual changes in outer

    diameter tend to reduce the turning losses and prevent separation, but

    adversely impact turbine envelope. Tangential entry inlets tend to have

    higher losses due to the tangential turning and acceleration of the flow. The

    spiral flow path also tends to be longer, increasing frictional losses.

    Typically, loss coefficients for practical axial inlets are in the range of 0.5 to2.0, while tangential inlets are in the range of 1.0 to 3.0. In terms of inlet

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    19/62

    pressure, inlet losses are usually on the order of 13% of the inlet total

    pressure. For turbines in engines, there is usually no real inlet, as they are

    closely coupled to the combustor or the preceding turbine stage. In this case,

    the duct losses are usually assessed to the upstream component.

    Stator Losses

    The stator losses arise primarily from friction within the vane row, the

    secondary flows caused by the flow turning, and exit losses due to blockage

    at the vane row trailing edge. The stator loss coefficient can be defined in

    several ways. Two popular definitions are

    Dp0statorYstator1

    2rV21 29

    or

    Dp0statorYstator1

    2r

    V20V212

    30

    In either case, the loss coefficient is made up of the sum of coefficients for

    each loss contributor:

    YstatorYprofileYsecondaryYtrailing edge 31Profile refers to frictional losses. There can be additional loss contributions

    due to incidence (the flow coming into the stator is not aligned with the

    leading edge), shock losses (when the stator exit velocity is supersonic), and

    others. Much work has been dedicated to determining the proper values for

    the coefficients, and several very satisfactory loss model systems have been

    developed. As loss models differ for axial-flow and radial-inflow turbines,

    these models will be discussed in the individual sections that follow.

    Rotor Losses

    Rotor losses are modeled in a manner similar to that for stators. However,

    the pressure loss is measured as a difference in relative total pressures and

    the kinetic energy is based on relative velocities. As with stators, the rotor

    loss is based on either the exit relative kinetic energy or the average of the

    inlet and exit relative kinetic energies:

    Dp00rotorXrotor 12 rW22

    32

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    20/62

    or

    Dp00rotorXrotor1

    2r

    W21W222

    33

    With rotors, incidence loss can be significant, so we include that contributor

    in the expression for the rotor loss coefficient:

    XrotorXprofileXsecondaryXtrailing edgeXincidence 34Other losses associated with the rotor are tip clearance and windage losses.

    Turbine rotors operate with a small clearance between the tips of the blades

    and the turbine housing. Flow leaks across this gap from the high-pressure

    side of the blade to the low-pressure side, causing a reduction in the pressure

    difference at the tip of the blade. This reduces the tangential force on theblade, decreasing the torque delivered to the shaft. Tip clearance effects can

    be reduced by shrouding the turbine blades with a ring, but this

    introduces manufacturing and mechanical integrity challenges. The loss

    associated with tip clearance can be modeled either using a pressure loss

    coefficient or directly as a reduction in the turbine efficiency. The specific

    models differ with turbine type and will be discussed in following sections.

    Windage losses arise from the drag of the turbine disk. As the disk

    spins in the housing, the no-slip condition on the rotating surface induces

    rotation of the neighboring fluid, establishing a radial pressure gradient inthe cavity. This is commonly referred to as disk pumping. For low-head

    turbines operating in dense fluids, the windage losses can be considerable.

    Windage effects are handled by calculating the windage torque from a disk

    moment coefficient defined as

    Gwindage 2Cm12ro2r5disk

    35

    The output torque of the turbine is reduced by the windage torque. Values

    of the moment coefficientCmdepend on the geometry of the disk cavity andthe speed of the disk. Nece and Daily [46] are reliable sources of moment

    coefficient data.

    Diffuser Losses

    Losses in the diffuser arise from sources similar to those in other flow

    passages, namely, friction and flow turning. The diffuser loss can be

    expressed in terms of a loss coefficient for accounting in turbineperformance, but diffuser performance is usually expressed in terms of

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    21/62

    diffuser recovery, defined as

    Rppdisp2p02

    p2

    36

    The diffuser recovery measures how much of the kinetic energy at diffuser

    inlet is converted to a rise in static pressure. Recovery is a function of area

    ratioAdis=A2, length, and curvature. For an ideal diffuser of infinite arearatio, the recovery is 1.0. Peak recovery of a real diffuser of given length

    takes place when the area ratio is set large enough so that the flow is on the

    verge of separating from the walls of the diffuser. When the flow separates

    within the diffuser, the diffuser is said to be stalled. Once stalled, diffuser

    recovery drops dramatically. Curvature of the mean radius of the diffuser

    tends to decrease the attainable recovery, since the boundary layer on one ofthe diffuser walls is subjected to a curvature-induced adverse pressure

    gradient in addition to the adverse pressure gradient caused by the increase

    in flow area.

    Even with the recent advances in general-use computational fluid

    dynamics (CFD) tools, analytical prediction of diffuser recovery is not

    normally performed as part of the preliminary turbine design. Diffuser

    performance is normally obtained from empirically derived plots such as

    that shown inFig. 10.Diffuser recovery is plotted as a function of area ratio

    and diffuser length. The curvature of the contours of recovery shows thelarge fall-off in diffuser recovery after the diffuser stall. The locus of

    maximum recovery is referred to as the line of impending stall. Diffusers

    should not be designed to operate above this line. Runstadler et al. [7, 8] and

    Sovran and Klomp [9] present charts of diffuser recovery as a function of

    inlet Mach number and blockage, as well as the three geometric factors

    noted earlier.

    The total pressure loss across a diffuser operating in incompressible

    flow can be calculated using continuity and the definition of diffuser

    recovery. The recovery for an ideal diffuser (no total pressure loss) is givenby

    Rpideal1 A2

    Adis

    237

    The total pressure loss for a nonideal diffuser in incompressible flow is given

    by

    p02p0dis12rV22

    Ktdiff RpidealRp 38

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    22/62

    This can be used to calculate the diffuser loss when compressibility is not

    important. If the Mach number at the inlet to the diffuser is above 0.20.3,

    this can be used as a starting guess, and the actual value can be determinedby iteration. The diffuser recovery is a function of the inlet Mach number,

    blockage, and geometry (straight, curved, conical, or annular); it is critical

    to use the correct diffuser performance chart when estimating diffuser

    recovery.

    Exit Losses

    Exit losses are quite simple. If the kinetic energy of the flow exiting the

    diffuser is used in following stages, or contributes to thrust, the exit losses

    are zero. If, however, the diffuser discharge energy is not utilized, the exit

    loss is the exit kinetic energy of the flow. For this case,

    Dp0exitp0dispdis 39

    Nondimensional Parameters

    Turbine performance is dependent on rotational speed, size, working fluid,enthalpy drop or head, and flow rate. To make comparisons between

    Figure 10 Conical diffuser performance chart. (Replotted from Ref. 8).

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    23/62

    different turbines easier, dimensional analysis leads to the formation of

    several dimensionless parameters that can be used to describe turbines.

    Specific Speed and Specific Diameter

    The specific speed of a turbine is defined as

    Ns offiffiffiffiffiffi

    Q2p

    Dhideal3=4 40

    where Q2 is the volumetric flow rate through the turbine at rotor exit. The

    specific speed is used to relate the performance of geometrically similar

    turbines of different size. In general, turbine efficiency for two turbines of

    the same specific speed will be the same, except for differences in tipclearance and Reynolds number. Maintaining specific speed of a turbine is a

    common approach to scaling of a turbine to different flow rates.

    The specific diameter is defined as

    DsdtipDhideal1=4ffiffiffiffiffiffi

    Q2p 41

    where dtip is the tip diameter of the turbine rotor, either radial in-flow or

    axial flow. Specific diameter and specific speed are used to correlate turbine

    performance. Balje [3] presents extensive analytical studies that result in

    maps of peak turbine efficiency versus specific speed and diameter for

    various types of turbines. These charts can be quite valuable during initial

    turbine sizing and performance estimation.

    Blade-Jet Speed Ratio

    Turbine performance can also be correlated against the blade-jet speed

    ratio, which is a measure of the blade speed relative to the ideal stator exitvelocity. Primarily used in impulse turbines, where the entire static enthalpy

    drop is taken across the stator, the ideal stator exit velocity,C0, is calculated

    assuming the entire ideal enthalpy drop is converted into kinetic energy:

    C0ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

    2Dhidealp

    42The blade-jet speed ratio is then calculated from

    U

    C0

    U

    ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2Dhidealp 43The value of blade speed at the mean turbine blade radius is typically used in

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    24/62

    Eq. (43) for axial turbines; for radial-inflow turbines, the rotor tip speed is

    used.

    Reynolds Number

    The Reynolds number for a turbine is usually defined as

    RerUtipdtipm

    44

    wheremis the viscosity of the working fluid. Sometimes odtip is substituted

    for Utip, resulting in a value twice that of Eq. (44). The Reynolds number

    relates the viscous and inertial effects in the fluid flow. For most

    turbomachinery operating on air, the Reynolds number is of secondaryimportance. However, when turbomachinery is scaled (either larger or

    smaller), the Reynolds number changes, resulting in a change in turbine

    efficiency. Glassman [1] suggests the following for adjusting turbine losses to

    account for Reynolds number changes:

    1Z0a1Z0b

    AB RebRea

    0:245

    whereZ0 indicates total-to-total efficiency and AandB sum to 1.0. That allthe loss is not scaled by the Reynolds number ratio reflects that not all losses

    are viscous in origin. Also, total-to-total efficiency is used since the kinetic

    energy of the exit loss is not affected by Reynolds number. Glassman [1]

    suggests values of 0.30.4 forA (the nonviscous loss) and from 0.7 to 0.6 for

    B (the viscous loss).

    Equivalent or Corrected Quantities

    In order to eliminate the dependence of turbine performance maps on the

    values of inlet temperature and pressure, corrected quantities such as

    corrected flow, corrected speed, corrected torque, and corrected power were

    developed. Using corrected quantities, turbine performance can be

    represented by just a few curves for a wide variety of operating conditions.

    Corrected quantities are not nondimensional. Glassman [1] provides a

    detailed derivation of the corrected quantities. The corrected flow is defined

    as

    wcorrwffiffiffiypd

    46

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    25/62

    where

    y T0in

    TSTD47

    and

    d p0in

    pSTD48

    The standard conditions are usually taken to be 518.7 R and 14.7 psia.

    Corrected speed is defined as

    Ncorr Nffiffiffiyp 49Equation (5) shows torque to be the product of flow rate and the change in

    tangential velocity across the rotor. Corrected flow is defined above;

    corrected velocities appear with y1=2 in the denominator from the corrected

    shaft speed. Therefore, corrected torque is defined as

    GcorrGd

    50

    The form of the corrected power is determined from the product of

    corrected torque and corrected speed:

    Pcorr Pdffiffiffiy

    p 51

    These corrected quantities are used to reduce turbine performance data to

    curves of constant-pressure ratio on two charts. Figure 11 presents typical

    turbine performance maps using the corrected quantities. Figure 11(a)

    presents corrected flow as a function of corrected speed and pressure ratio,

    whileFig. 11(b)shows corrected torque versus corrected speed and pressure

    ratio. Characteristics typical of both radial-inflow and axial-flow turbinesare presented in Fig. 11.

    AXIAL-FLOW TURBINE SIZING

    Axial-Flow Turbine Performance Prediction

    Prediction methods for axial-flow turbine performance methods can be

    roughly broken into two groups according to Sieverding [10]. The first

    group bases turbine stage performance on overall parameters such as workcoefficient and flow coefficient. These are most often used in preliminary

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    26/62

    Figure 11 Typical performance maps using corrected quantities for axial-flow andradial-inflow turbines: (a) corrected flow vs. pressure ratio and corrected speed; (b)

    corrected torque vs. pressure ratio and corrected speed.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    27/62

    sizing exercises where the details of the turbine design are unknown. Smith

    [11] and Soderberg [12] are both examples of this black box approach, as

    are Baljes [3] maps of turbine efficiency as a function of specific speed and

    specific diameter.

    The second grouping is based on the approach outlined earlier whereturbine losses are broken down to a much finer level. In these methods, a

    large number of individual losses are summed to arrive at the total loss.

    Each of these loss components is dependent on geometric and aerodynamic

    parameters. This requires more knowledge of the turbine configuration,

    such as flow path and blading geometry, before a performance estimate can

    be made. As such, these methods are better suited for more detailed turbine

    design studies.

    Among the loss component methods, Sieverding [10] gives an excellent

    review of the more popular component loss models. The progenitor of afamily of loss models is that developed by Ainley and Mathieson [13]. It has

    been modified and improved by Dunham and Came [14] and, more

    recently, by Kacker and Okapuu [15]. A somewhat different approach is

    taken by Craig and Cox [16]. All these methods are based on correlations of

    experimental data.

    An alternate approach is to analytically predict the major loss

    components such as profile or friction losses and trailing-edge thickness

    losses by computing the boundary layers along the blade surfaces. Profile

    losses are then computed from the momentum thickness of the boundarylayers on the pressure and suction surfaces of the blades or vanes. Glassman

    [1] gives a detailed explanation of this method. Note that this technique

    requires even more information on the turbine design; to calculate the

    boundary layer it is necessary to know both the surface contour and the

    velocities along the blade surface. Thus, this method cannot be used until

    blade geometries have been completely specified and detailed flow channel

    calculations have been made.

    In addition to the published prediction methods just noted, each of the

    major turbine design houses (such as AlliedSignal, Allison, General Electric,Lycoming, Pratt & Whitney, Sundstrand, and Williams) has its own

    proprietary models based on a large turbine performance database. Of

    course, it is not possible to report those here.

    For our purposes (determining the size and approximate performance

    of a turbine) we will concentrate on the overall performance prediction

    methods, specifically Smiths chart and Soderbergs correlation. Figure 12

    shows Smiths [11] chart, where contours of total-to-total efficiency are

    plotted versus flow coefficient and work factor [see Eqs. (23) and (24)]. Both

    the flow coefficient and stage work coefficient are defined using velocities atthe mean radius of the turbine. The efficiency contours are based on the

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    28/62

    measured efficiency for 70 turbines. All the turbines have a constant axial

    velocity across the stage, zero incidence at design point, and reactions

    ranging from 20% to 60%. Reynolds number for the turbines range from

    100,000 to 300,000. Aspect ratio (blade height to axial chord) for the tested

    turbines is in the range of 34. Smiths chart does not account for the effects

    of blade aspect ratio, Mach number effects, or trailing-edge thickness

    variations. The data have been corrected to reflect zero tip clearance, so theefficiencies must be adjusted for the tip clearance loss of the application.

    Sieverding [10] considers Soderbergs correlation to be outdated but

    still useful in preliminary design stages due to its simplicity. In Soderbergs

    [12] correlation, blade-row kinetic energy losses are calculated from

    Vo2idealV2oV2o

    x

    105Rth

    1=4 1xref 0:9750:075 cxh 1 h in o 52

    Figure 12 Smiths chart for stage zero-clearance total-to-total efficiency as

    function of mean-radius flow and loading coefficient. (Replotted from Ref. 11

    with permission of the Royal Aeronautical Society).

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    29/62

    where Rth is the Reynolds number based on the hydraulic diameter at the

    blade passage minimum area (referred to as the throat) defined as

    RthroVo

    mo

    2hs cos

    ao

    hs cosao 53

    where h is the blade height and s is the spacing between the blades at the

    mean radius. The blade axial chord is identified by cx. In both Eqs. (52) and

    (53), the subscript o refers to blade-row outlet conditions, either stator or

    rotor (for the rotor, the absolute velocity V is replaced by the relative

    velocity W, standard practice for all blade-row relations). The reference

    loss coefficient xrefis a function of blade turning and thickness and can be

    found inFig. 13. Compared to Smiths chart, this correlation requires moreknowledge of the turbine geometry, but no more than would be required in

    a conceptual turbine design. The losses predicted by this method are only

    valid for the optimum blade chord-to-spacing ratio and for zero incidence.

    Tip clearance losses must also be added in the final determination of turbine

    efficiency. Like Smiths chart, this correlation results in a total-to-total

    efficiency for the turbine.

    The optimum value of blade chord-to-spacing ratio can be found using

    the definition of the Zweifel coefficient [17]:

    z 2cx=s

    cos ao

    cos aisinaiao

    54

    where the subscript i refers to blade-row inlet. Zweifel [17] states that

    optimum soliditycx=s occurs when z0:8.Tip clearance losses are caused by flow leakage through the gap

    between the turbine blade and the stationary shroud. This flow does not get

    turned by the turbine blade; so it does not result in work extraction. In

    addition, the flow through the clearance region causes a reduction of thepressure loading across the blade tip, further reducing the turbine efficiency.

    The leakage flow is primarily controlled by the radial clearance, but is also

    affected by the geometry of the shroud and the blade reaction. Leakage

    effects can be reduced by attaching a shroud to the turbine blade tips, which

    eliminates the tip unloading phenomenon. For preliminary design purposes,

    the tip clearance loss for unshrouded turbine wheels can be approximated by

    Z

    Zzc 1

    Kc

    rtip

    rmean

    cr

    h 55

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    30/62

    Where

    Zzc zero clearance efficiency:orradial tip clearance:

    rtippassage tip radius:rmeanmean passage radius:

    Kc empirically derived constant:

    Based on measurements reported by Haas and Kofskey [18], the value ofKcis between 1.5 and 2.0, depending on geometric configuration. For

    preliminary design purposes, the conservative value should be used. When

    using Soderbergs correlation, the value of Kc should be taken as 1, since

    Soderberg corrected his data using that value for Kc.

    With the information above, the turbine efficiency (total-to-total) canbe determined from the stator inlet (station 0) to rotor exit (station 2). In

    Figure 13 Soderbergs loss coefficient as function of deflection angle and blade

    thickness. (Replotted from Ref. 12 with permission from Pergamon Press Ltd.)

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    31/62

    order to determine the overall turbine efficiency, it is necessary to include the

    inlet, diffuser, and exit losses. These losses do not affect the turbine work

    extraction, but result in the overall pressure ratio across the turbine being

    larger than the stage pressure ratio. The overall efficiency can be calculated

    from

    ZOA Z0020 1 p02=p00g1=g1 pdis=p0ing1=g

    56

    The pressure losses in the inlet, diffuser, and exit are calculated from the

    information presented earlier.

    Mechanical, Geometric, and Manufacturing Constraints

    Turbine design is as much or more affected by mechanical considerations as

    it is by aerodynamic considerations. Aerodynamic performance is normally

    constrained by the stress limitations of the turbine material. At this point in

    the history of turbine design, turbine performance at elevated temperatures

    is limited by materials, not aerodynamics. Material and manufacturing

    limitations affect both the geometry of the turbine wheel and its operating

    conditions.

    Turbine blade speed is limited by the centrifugal stresses in the diskand by the tensile stress at the blade root (where the blade attaches to the

    disk). The allowable stress limit is affected by the turbine material, turbine

    temperature, and turbine life requirements. Typical turbine materials for

    aircraft auxiliary turbines are titanium in moderate-temperature applica-

    tions (turbine relative temperatures below 1,000 8F) and superalloys for

    higher temperatures.

    Allowable blade-tip speed for axial-flow turbines is a complex function

    of inlet temperature, availability of cooling air, thermal cycling (low cycle

    fatigue damage), and desired operating life. In general, design point bladespeeds are held below 2,200 ft/sec, but higher blade speeds can be withstood

    for shorter lifetimes, if temperatures permit. For auxiliary turbine

    applications with inlet temperatures below 300 8F and pressure ratios of 3

    or below, blade speed limits are generally not a design driver.

    Both stress and manufacturing considerations limit the turbine blade

    hub-to-tip radius ratio to values greater than about 0.6. If the hub diameter

    is much smaller, it is difficult to physically accommodate the required

    number of blades on the hub. Also, the twist of the turbine blade increases,

    leading to sections at the tip not being directly supported by the hub section.This leads to high bending loads in the blade and higher stress levels. For

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    32/62

    performance reasons (secondary flow losses and tip clearance losses), it is

    desirable to keep the hub-to-tip radius ratio below 0.8.

    Manufacturing considerations limit blade angles on rotors to less than

    608 and stator vane exit angles to less than 758. Casting capabilities limit

    stator trailing-edge thicknesstteto no less than 0.015 in., restricting statorvane count. For performance reasons, the trailing-edge blockage should be

    kept less than 10% at all radii. The trailing-edge blockage is defined here as

    the ratio of the trailing-edge tangential thickness (b) to the blade or vane

    spacing (s):

    b

    stte= cos ate

    2pr=Z 57

    whereZis the blade or vane count. Rotor blades are usually machined, butfor stress and tolerance reasons the trailing-edge thickness is normally no

    less than 0.015 in. The 10% limitation on blockage is also valid for rotors.

    Auxiliary turbines often are required to survive free-run conditions.

    Free run occurs when the turbine load is removed but the air supply is not.

    This can happen if an output shaft fails or if an inlet control valve fails to

    close. Without any load, the turbine accelerates until the power output of

    the turbine is matched by the geartrain and aerodynamic losses. Free-run

    speed is roughly twice design-point speed for most aircraft auxiliary

    turbines. This restricts the allowable design-point speeds and stress levelsfurther, since the disk and blade may be required to survive free-run

    operation.

    Hub-to-Tip Variation in Vector Diagram

    Up to this point we have only considered the vector diagram at the mean

    radius of the turbine. For turbines with high hub-to-tip radius ratios (above

    0.85), the variation in vector diagram is not important. For a turbine with

    relatively tall blades, however, the variation is significant.The change in vector diagram with radius is due to the change in blade

    speed and the balance between pressure and body forces acting on the

    working fluid as it goes through the turbine. Examples of body forces

    include the centrifugal force acting on a fluid element that has a tangential

    velocity (such as between the stator and rotor), and the accelerations caused

    by a change in flow direction if the flow path is curved in the meridional

    plane. The balance of these forces (body and pressure) is referred to as radial

    equilibrium. Glassman [1] presents a detailed mathematical development of

    the equations that govern radial equilibrium. For our purposes, we willconcentrate on the conditions that satisfy radial equilibrium.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    33/62

    The classical approach to satisfying radial equilibrium is to use a free

    vortex variation in the vector diagram from the hub to the tip of the rotor

    blade. A free vortex variation is obtained by holding the product of the

    radius and tangential velocity constant

    rVy

    constant

    . When this is done,

    the axial velocity Vx is invariant with radius. Until the widespread use ofcomputers in turbine design, almost all turbines employed free vortex

    diagrams due to their simplicity. For preliminary design purposes, the free

    vortex diagram is more than satisfactory.

    Aside from its simplicity, the free vortex diagram has other

    advantages. Holding rVy constant implies that the work extraction is

    constant with radius. With Vx constant, the mass flow varies little with

    radius. This implies that the mean section vector diagram is an excellent

    representation of the entire turbine from both a work and mass flow

    standpoint.When using a free vortex distribution, there are two key items to

    examine in addition to the mean vector diagram. The hub diagram suffers

    from low reaction due to the increase in Vy and should be checked to ensure

    at least a zero value of reaction. From hub to tip, the reduction in Vy and

    increase inUcause a large change in the rotor inlet relative flow angle, with

    the rotor tip section tending to overhang the hub section. By choosing a

    moderate hub-to-tip radius ratio (if possible), both low hub reaction and

    excessive rotor blade twist can be avoided.

    For a zero exit swirl vector diagram, some simple relations can bedeveloped for the allowable mean radius work coefficient and the hub-to-tip

    twist of the rotor blade. For a zero exit swirl diagram, zero reaction occurs

    for a work coefficient of 2.0. Using this as an upper limit at the hub, the

    work coefficient at mean radius is found from

    lm2 rhrm

    258

    For a turbine with a hub-to-tip radius ratio of 0.7, the maximum work

    coefficient at mean radius for impulse conditions at the hub is 1.356. The

    deviation in inlet flow angle to the rotor from hub to tip for a free vortex

    distribution is given by

    Db1b1hb1t

    tan1 lmrm=rh2 1

    fm

    rm=rh

    " #tan1 lmrm=rt

    2 1fm

    rm=rt

    " # 59

    For a vector diagram with lm1:356; rh=rt0:7, and

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    34/62

    fm0:6;Db156:1, which is acceptable from a manufacturing viewpoint.Large negative inlet angles at the blade tip are to be avoided.

    An Example of Turbine Sizing

    In order to demonstrate the concepts described in this and preceding

    sections, an example is presented of the sizing of a typical auxiliary

    turbine for use in an aircraft application. The turbine is to be sized to meet

    the following requirements:

    1. Generates 100 hp at design point.

    2. Operates at an overall pressure ratio of 3:1 in air.

    3. Inlet pressure is 44.1 psia, and inlet temperature is 300 8F.

    The object of this exercise is to determine the turbine size, flow rate,

    and operating speed with a turbine design meeting the mechanical,

    geometric, and manufacturing constraints outlined earlier. The following

    procedure will be followed to perform this exercise:

    1. Determine available energy (isentropic enthalpy drop).

    2. Guesstimate overall efficiency to calculate flow rate.

    3. Select the vector diagram parameters.

    4. Calculate the vector diagram.

    5. Determine the rotor overall geometry.6. Calculate the overall efficiency based on Smiths chart both with

    and without a diffuser.

    The process is iterative in that the efficiency determined in step 6 is then used

    as the guess in step 2, with the process repeated until no change is found in

    the predicted efficiency. We will also predict the turbine efficiency using

    Soderbergs correlation.

    The first step is to calculate the energy available to the turbine using

    Eq. (1). For air, typical values for the specific heat and the ratio of specific

    heats are 0:24 Btu=lbmR and 1.4, respectively. It is also necessary toconvert the inlet temperature to the absolute scale. We then have

    Dhisentropic 0:24 BtulbmR

    760R 1 1

    3

    0:4=1:4" #49:14 Btu

    lbm

    Note that more digits are carried through the calculations than indicated, so

    exact agreement may not occur in all instances. The vector diagram is

    calculated using the work actually done by the blade row; therefore, we need

    to start with a guess to the overall efficiency of the turbine. A good startingpoint is usually an overall efficiency of 0.8, including the effects of tip

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    35/62

    clearance. Since tip clearance represents a loss at the tip of the blade, the rest

    of the blade does more than the average work. Therefore, the vector

    diagram is calculated using the zero-clearance efficiency. Since we do not

    know the turbine geometry at this point, we must make another assumption:

    we assume that the tip clearance loss is 5%, so that the overall zero-clearanceefficiency is 0.84. Note that the required flow rate is calculated using the

    overall efficiency with clearance, since that represents the energy available at

    the turbine shaft. Equation (2) is used to calculate the actual enthalpy drops:

    DhOA 0:8 49:14Btulbm

    39:31Btu

    lbm

    and

    DhOA ZC 0:84 49:14Btulbm

    41:28Btu

    lbm

    The required turbine flow is found using Eq. (7):

    _mm PDhOA

    100 hp:7069 Btu=sec=hp39:31 Btu=lbm

    1:798 lbm= sec

    The mass flow rate is needed to calculate turbine flow area and is also a

    system requirement.

    We specify the vector diagram by selecting values of the turbine work

    and flow coefficients. We also select a turbine hub-to-tip radius ratio of 0.7,

    restricting the choice of mean work coefficient to values less than 1.356 in

    order to avoid negative reaction at the hub. From Smiths chart(Fig. 12), we

    initially choose a work coefficient of 1.3 and a flow coefficient of 0.6 to result

    in a zero-clearance, stator inlet to rotor exit total-to-total efficiency of 0.94.

    We apply these coefficients at the mean radius of the turbine. From Eq. (22)

    we calculate the mean blade speed, Um:

    UmffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiDhOA ZC

    l

    r

    ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffi32:174 ftlbf=lbmsec2778:16ftlbf=Btu41:28 Btu=lbm

    1:3

    r891:6 ft= sec

    The axial velocity is calculated from Eq. (51):

    Vx2 0:6891:6 ft= sec 535:0 ft= sec

    In order to construct the vector diagram, we make two more assumptions:

    (1) there is zero swirl leaving the turbine stage in order to minimize the exitkinetic energy loss, and (2) the axial velocity is constant through the stage.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    36/62

    By assuming that Vy2 is zero, Eq. (23) reduces to

    Vy1lUm 1:3891:6 ft= sec 1159:1 ft= sec

    Using Eqs. (10) through (14) results in the vector diagram shown in Fig. 14.Note that the critical stator and rotor exit angles are within the guidelines

    presented earlier.

    The rotor blade height and mean radius are determined by the

    required rotor exit flow area and the hub-to-tip radius ratio. The rotor exit

    flow area is determined from continuity:

    A2r2Vx2_mm

    The mass flow rate and axial velocity have previously been calculated;

    the density is dependent on the rotor exit temperature and pressure. For a

    turbine without a diffuser, the rotor exit static pressure is the same as the

    discharge pressure, assuming the rotor exit annulus is not choked. For a

    Figure 14 Mean-radius velocity diagrams for first iteration of axial-flow turbine

    sizing example.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    37/62

    turbine with an effective diffuser, the rotor exit static pressure will be less

    than the discharge value. We will examine both cases.

    Turbine Without DiffuserFirst we consider the turbine without a diffuser. Assuming perfect gas

    behavior, the density is calculated from

    r2 p2

    RgasT2

    where the temperature and pressure are static values and Rgas is the gas

    constant. The rotor exit total temperature is determined from

    T02T00 DhOA ZCCp 760 R 41:28 Btu=lbm

    0:24 Btu=lbmR 588:0 R

    The zero-clearance enthalpy drop is used because the local tempera-

    ture over the majority of the blade will reflect the higher work (a higher

    discharge temperature will be measured downstream of the turbine after

    mixing of the tip clearance flow has occurred). Next we calculate the rotor

    exit critical Mach number to determine the static temperature. The critical

    sonic velocity is calculated from

    acrffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

    2g

    g1 gRgasT0

    s

    where g is a conversion factor. For air at low temperatures,

    Rgas53:34 ft-lbf=lbmR, resulting in

    acr2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

    21:411:4 32:174

    ftlbflbmsec2

    53:34

    ftlbflbmR

    588R

    s

    1085 ft=secThe static temperature is found from

    T2T02 1g1g1

    V2

    acr2

    2" #

    with zero exit swirl, V2Vx2 resulting in

    T2 588R 11:4111:4 535 ft=sec1085 ft=sec

    2" #564:2 R

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    38/62

    The density can now be determined:

    r2

    44:1 lbf=in2

    3

    144 in

    2

    ft2

    53:34

    ftlbf

    lbmR 564:2 R 0:0703lbm

    ft

    3

    and the required flow area:

    A2 1:798 lbm=sec0:0703 lbm=ft3535:0 ft=sec144

    in:2

    ft2

    6:882in:2

    The rotor exit hub and tip radii cannot be uniquely determined until

    either shaft speed, blade height, or hub-to-tip radius ratio is specified. Once

    one parameter is specified, the others are determined. For this example, we

    choose a hub-to-tip ratio of 0.7 as a compromise between performance and

    manufacturability. If the turbine shaft speed were restricted to a certain

    value or range of values, it would make more sense to specify the shaft

    speed. The turbine tip radius is determined from

    rt2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

    A2

    p1 rh=rt2

    s

    ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi6:882 in:2

    p1 0:72

    s 2:073 in:

    This results in a hub radius of 1.451 in., a mean radius of 1.762 in. and

    a blade height of 0.622 in. The shaft speed is found from Eq. (8):

    oUm=rm 891:6 ft=sec1:762in1 ft=12in6073 rad=sec

    or 57,600 rpm. The tip speed of the turbine is 1,049 ft/sec, well within our

    guidelines.

    The next step is to calculate the overall efficiency. From Smiths chart,

    a stator inlet to rotor exit total-to-total efficiency at zero clearance is

    available. We must correct this for tip clearance effects, the inlet loss, andthe exit kinetic energy loss. At l1:3 and f 0:6, Smiths chart predicts

    Z0020ZC0:94Assuming a tip clearance of 0.015 in., the total-to-total efficiency including

    the tip clearance loss is calculated from Eq. (55) using a value of 2 for Kc:

    Z0020 Z0020ZC 12rt

    rm

    d

    h

    0:94 12 2:073

    1:762

    0:015

    0:622

    0:8867

    Equation (56) is used to determine the overall efficiency including inlet andexit losses. From the problem statement, we know that the overall pressure

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    39/62

    ratiop0in=pdisis 3. The stator inlet to rotor exit total-to-total pressure ratiois calculated from

    p02

    p00 pdis

    p0in p02

    pdis p0in

    p00 Based on earlier discussions, we assume an inlet total pressure loss ratio of

    0.99. With no diffuser, the discharge and rotor exit stations are the same, so

    the ratio of static to total pressure is found from the rotor exit Mach

    number:

    pdis

    p02p2

    p02 1g1

    g1V2

    acr2

    2" # gg1 11

    6

    535

    1085

    2" #3:50:8652

    We can now calculate the total-to-total pressure ratio from stator inlet

    to rotor exit and the overall efficiency:

    p02p00

    13

    1

    0:8652

    1

    0:99

    0:3891

    and

    ZOA

    0:8867

    1 0:38910:4=1:4

    1 13 0:4=1:4 0:7779This completes the first iteration on the turbine size and performance for the

    case without a diffuser. To improve the accuracy of the result, the preceding

    calculations would be repeated using the new values of overall efficiency and

    tip clearance loss.

    Turbine with DiffuserFor an auxiliary type of turbine such as this, a diffuser recovery of 0.4 is

    reasonable to expect with a well-designed diffuser. The rotor exit total

    pressure is calculated from the definition of diffuser recovery given in Eq.

    (35):

    p02 pdis

    Rp1p2=p02 p2=p02 44:1 psia=3

    0:410:8652 0:865215:99 psia

    and the rotor exit static pressure is

    p2 15:99 psia0:8652 13:84 psia

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    40/62

    This is a considerable reduction compared to the discharge pressure of

    14.7 psia. From this point, the rotor exit geometry is calculated in the same

    way as that presented for the case without the diffuser. The following results

    are obtained:

    r20:0662 lbm=ft3A27:311 in:2rt22:136 in:rh21:495 in:

    rm21:816 in:h20:641 in:N56;270 rpm

    The tip speed is the same as the turbine without the diffuser, since the mean

    blade speed is unchanged, as is the hub-to-tip radius ratio of the rotor. The

    efficiency calculations also proceed in the same manner as the earlier case

    with the following results (using the same inlet pressure loss assumption):

    Z0020 0:8882

    p02p00

    0:3663ZOA0:8224

    Since this result differs from our original assumption for overall

    efficiency, further iterations would be performed to obtain a more accurate

    answer. Note the almost 6% increase in overall efficiency due to the

    inclusion of a diffuser. This indicates a large amount of energy is contained

    in the turbine exhaust. The efficiency gain associated with a diffuser is

    dependent on diffuser recovery, rotor exit Mach number, and overallpressure ratio and is easily calculated. Figure 15 shows the efficiency

    benefit associated with a diffuser for an overall turbine pressure ratio

    (total-to-static) of 3. Efficiency gains are plotted as a function of rotor exit

    critical Mach number and diffuser recovery. As rotor exit Mach number

    increases, the advantages of including a diffuser become larger. This

    tradeoff is important to consider when sizing the turbine. For a given flow

    or power level, turbine rotor diameter can be reduced by accepting high

    rotor exit velocities (high values of flow coefficient); however, turbine

    efficiency will suffer unless a diffuser is included, adversely impacting theaxial envelope.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    41/62

    Automation of Calculations and Trade Studies

    The calculations outlined in this example can be easily automated in either a

    computer program or a spreadsheet with iteration capability. An example of

    the latter is presented inFig. 16,which contains the iterated final results forthe example turbine when equipped with a diffuser. The advantage of

    automation is the capability to quickly perform trade studies to optimize the

    turbine preliminary design. Prospective variables for study include work and

    flow coefficients, diffuser recovery, shaft speed or hub-to-tip radius ratio,

    inlet loss, tip clearance, exit swirl, and others.

    Soderbergs Method

    We conclude this example by calculating the turbine performance usingSoderbergs correlation. We will use the iterated turbine design results

    shown in the spreadsheet of Fig. 16. Soderbergs correlation [Eq. (52)]

    requires the vane and blade chords in order to calculate the aspect ratio

    cx=h. We first determine the blade number by setting the blockage level atmean radius to 10% and the trailing-edge thickness for both the rotor and

    Figure 15 Effect of diffuser on turbine efficiency at an overall pressure ratio of 3.0.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    42/62

    stator at 0.020 in. These values are selected based on the guidelines given

    earlier in the chapter. Solving Eq. (57) for the blade number results in

    Zb=s2prmtte= cosate

    For the stator, the flow anglea1is used forate; for the rotor, the relative flow

    angle b2 is substituted for ate. The blade angle is slightly different from the

    flow angle due to blockage effects, but for preliminary sizing, the

    approximation is acceptable. For the stator, we have

    Zstator 0:12p1:773 in0:020 in= cos65:2223:35

    Figure 16 Spreadsheet for preliminary axial-flow turbine sizing showing iterated

    results for example turbine.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    43/62

    and for the rotor

    Zrotor 0:12p1:773 in

    0:020 in

    = cos

    59:04

    28:65

    Of course, only integral number of blades are allowed, so we choose 23

    vanes for the stator and 29 rotor blades, resulting in a blade spacing of

    0.484 in. for the stator and 0.384 in. for the rotor. Normal practice is to

    avoid even blade counts for both the rotor and stator to reduce rotor blade

    vibration response. The blade chord is now calculated from Zweifels

    relation given in Eq. (54) using the optimum value of 0.8 for the Zweifel

    coefficient:

    cx 2

    z=s

    cos

    ao

    cosai sinaiao For the stator,

    cxstator 2

    0:8=0:484in:cos65:22

    cos0 sin65:22

    0:460 in:

    and for the rotor,

    c

    xrotor 2

    0:8=0:384 in:cos

    59:04

    cos26:57 sin

    26

    :57

    59

    :04

    0:551 in:

    The Reynolds number for each blade row is calculated from Eq. (53).

    At the exit of each blade row, the static temperature and pressure are

    required to calculate the density. The viscosity is calculated using the total

    temperature to approximate the temperature in the boundary layers where

    viscous effects dominate. For the stator, the exit total temperature is the

    same as the inlet temperature. We assume a 1% total pressure loss across the

    stator. Using the stator exit Mach number, the static pressure is calculated:

    p144:1 psia0:990:99 1161:05172

    3:5 21:18 psia

    as is the static temperature:

    T1 760R 1161:05172

    619:9 R

    Using the perfect gas relation, the stator exit density r1 is calculated to be0:09225 lbm=ft

    3. The viscosity is determined using an expression derived

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    44/62

    from that presented by ASHRAE [19]:

    m ffiffiffiffiT

    p

    1:34103

    306:288=T

    13658:3=T2

    1; 239; 069=T3

    6106 lbm

    ft-sec

    The original expression was in SI units. For the stator, the viscosity

    m11:59986105 lbm=ft-sec. The Reynolds number is then calculated usingEq. (53):

    Rth

    stator

    0:09225 lbmft3

    1297:4 ftsec

    1:59986

    105 lbm

    ft-sec

    20:626 in:0:484 in: cos65:2212 ft

    in0:626 in: 0:484 in: cos65:22

    resulting in a Reynolds number of 1:91036105. A similar procedure is usedfor the rotor, except the relative velocity and angle at the rotor exit (station

    2) are used. The viscosity is calculated using the relative total temperature

    determined using Eq. (17). For the rotor, the Reynolds number is

    1:23566105.

    The reference value of the loss coefficient x is found fromFig. 13as afunction of the deflection across the blade row. The deflection is the

    difference between the inlet and outlet flow angles. For the stator, the

    deflection is 65.228, and for the rotor it is 85.618, resulting in xref s0:068andxref r0:083, assuming a blade thickness ratio of 0.2. The adjusted losscoefficients are calculated from Eq. (52):

    xstator 105

    1:91036105 1=4

    10:068 0:9750:075 0:4600:626 1

    0:0852

    and for the rotor

    xrotor 105

    1:23566105

    1=410:083 0:9750:075 0:551

    0:626

    1

    0:1209

    The stator inlet to rotor exit total-to-total efficiency is calculated from theratio of the energy extracted from the flowUDVy divided by the sum of

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    45/62

    the energy extracted and the rotor and stator losses:

    Z0020ZC UDVy

    UDVy

    12

    V22xstator

    12

    W22xrotor

    Numerically, we have

    Z0020ZC 1:3906:15 ft

    sec2

    1:3906:15 ftsec

    2 0:08522 1297:39 ft

    sec2 0:1209

    2 1056:75 ft

    sec2

    0:8846which is considerably lower than the 0.94 value from Smiths chart.

    Correcting for tip clearance using a value of 1 for Kc in Eq. (55) yields

    Z0020 0:8846 1 1

    0:85

    0:015

    0:626

    0:8597

    and correcting for overall pressure ratio using the total-to-total pressure

    ratio fromFig. 16results in the overall efficiency:

    ZOA0:85971 1=2:7201g1=g

    1 1=3:0g1=g 0:7935

    This value is 0.025 lower than the value of 0.8187 from Fig. 16 predictedusing Smiths chart. Sieverding [10] notes that Smiths chart was developed

    for blades with high aspect ratios (h=cxin the range of 34), which will resultin higher efficiency than lower aspect ratios, such as in this example. For

    preliminary sizing purposes, the conservative result should be used.

    Partial Admission Turbines

    For applications where the shaft speed is restricted to low values or the

    volumetric flow rate is very low, higher efficiency can sometimes be obtainedwith a turbine stator that only admits flow to the rotor over a portion of its

    circumference. Such a turbine is called a partial-admission turbine. Partial-

    admission turbines are indicated when the specific speed of the turbine is

    low. Balje [3] indicates partial admission to be desirable for specific speeds

    less than 0.1. Several conditions can contribute to low specific speed.

    Typically, drive turbines operate most efficiently at shaft speeds higher than

    the loads they are coupled to, such as generators, hydraulic pumps, and, in

    the case of an air turbine starter, the main engine of an aircraft. For low-cost

    applications, it may be desirable to eliminate the speed-reducing gearboxand couple the load directly to the turbine shaft. In order to attain adequate

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    46/62

    blade speed at the reduced shaft speed, it is necessary to increase the turbine

    diameter, which causes the blade height to decrease. The short blades cause

    an increase in secondary flow losses reducing turbine efficiency. With partial

    admission, the blade height can be increased, reducing secondary flow

    losses. In a low-flow-rate situation, maintaining a given hub-to-tip radiusratio results in an increase in the design shaft speed and a decrease in the

    overall size of the turbine. However, manufacturing limits restrict the radial

    tip clearance and blade thickness. With a small blade height, tip clearance

    losses are increased. With a limitation on how thin blades can be made, it is

    necessary to reduce blade count in order to keep trailing-edge blockage to a

    reasonable level. Fewer blades result in longer blade chord and reduced

    aspect ratio, leading to higher secondary flow losses. The taller blades

    associated with partial admission can increase turbine performance. For

    high-head applications a high blade speed is necessary for peak efficiency.With shaft speed restricted by bearing and manufacturing limitations, an

    increase in turbine diameter is required, resulting in a situation similar to the

    no-gearbox case discussed earlier. Here, too, partial admission can result in

    improved turbine efficiency.

    The penalty for partial admission is two additional losses not found in

    full-admission turbines. These are the pumping loss and sector loss. The

    pumping loss accounts for the drag of the rotor blades as they pass through

    the inactive arc, the portion of the circumference not supplied with flow

    from the stator. The sector loss arises from the decrease in momentumcaused by the mixing of the stator exit flow with the relatively stagnant fluid

    occupying the blade passage just as it enters the active arc. Instead of being

    converted into useful shaft work, the stator exit flow is used to accelerate

    this stagnant fluid up to the rotor exit velocity. An additional loss occurs at

    the other end of the active arc as the blade passages leave the active zone.

    Just as a blade passage is at the edge of the last active stator vane passage,

    the flow into the rotor blade passage is reduced. This reduced flow has the

    entire blade passage to expand into. The sudden expansion causes a loss in

    momentum resulting in decreased power output from the turbine. Lossmodels for partial-admission effects are not as well developed as those for

    conventional, full-admission turbines. As a historical basis, Glassman [1]

    presents Stodolas [20] pumping loss model and Stennings [21] sector loss

    model in an understandable form and discusses their use. More recently,

    Macchi and Lozza [22] have compiled a number of more modern loss

    models and exercised them during the design of partial-admission turbines.

    The reader is referred to those sources for detailed information regarding

    the estimation of partial-admission losses.

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    47/62

    RADIAL-INFLOW TURBINE SIZING

    Differences Between Radial-Inflow and Axial-Flow Turbines

    Radial-inflow turbines enjoy widespread use in automotive turbochargers

    and in small gas turbine engines (auxiliary power units, turboprops, andexpendable turbine engines). One advantage is their low cost relative to

    machined axial turbines, as most of these applications use integrally bladed

    cast radial-inflow turbine wheels.

    The obvious difference between radial-inflow and axial-flow turbines is

    easily seen inFig. 1; a radial-inflow turbine has a significant change in the

    mean radius between rotor inlet and rotor outlet, whereas an axial-flow

    turbine has only a minimal change in mean radius, if any. Because of this

    geometric difference, there are considerable differences in the performance

    characteristics of these two types of turbines. Referring to the typicalradial-inflow vector diagram of Fig. 9(c), the radius change causes a

    considerable decrease in wheel speed Ubetween rotor inlet and outlet. For

    zero exit swirl, this results in a reduced relative exit velocity compared to an

    axial turbine with the same inlet vector diagram (since U2&U1 for an axial

    rotor). Since frictional losses are proportional to the square of velocity, this

    results in higher rotor efficiency for the radial-inflow turbine. However, the

    effect of reduced velocity level is somewhat offset by the long, low-aspect-

    ratio blade passages of a radial-inflow rotor.

    Compared to the axial-flow diagram of Fig. 9(a), there is a much largerdifference between the rotor inlet relative and absolute velocities for the

    radial-inflow diagram. Referring to Eq. (17), this results in a lower relative

    inlet total temperature at design point for the radial-inflow turbine. In

    addition, due to the decrease in rotor speed with radius, the relative total

    temperature decreases toward the root of radial-inflow turbine blades (see

    Mathis [23]). This is a major advantage for high inlet temperature

    applications, since material properties are strongly temperature-dependent.

    The combination of radial blades at rotor inlet (eliminating bending stresses

    due to wheel rotation) and the decreased temperature in the high-stressblade root areas allows the radial-inflow turbine to operate at significantly

    higher wheel speeds than an axial-flow turbine, providing an appreciable

    increase in turbine efficiency for high-pressure-ratio, high-work applica-

    tions.

    For applications with moderate inlet temperatures (less than 500 8F)

    and pressure ratios (less than 4:1), the blade speed of an axial wheel is not

    constrained by stress considerations and the radial-inflow turbine is at a size

    disadvantage. Due to bending stress considerations in the rotor blades,

    radial blades are used at the inlet to eliminate bending loads. This limits theVy1=U1 ratio to 1 or less, meaning that the tip speed for an equal work

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    48/62

    radial-inflow turbine will be higher than that for an axial-flow turbine,

    which can have Vy1=U1 > 1 with only a small impact on efficiency. Thisassumes zero exit swirl. For a fixed shaft speed, this means that the radial-

    inflow turbine will be larger (and heavier) than an axial-flow turbine. Stage

    work can be increased by adding exit swirl; however, the radial-inflowturbine is again at a disadvantage. The lower wheel speed at exit for the

    radial-inflow turbine means that more Vy2 must be added for the same

    amount of work increase, resulting in higher exit absolute velocities

    compared to an axial-flow turbine. In addition, high values of exit swirl

    negatively impact obtainable diffuser recoveries.

    Packaging considerations may lead to the selection of a radial-inflow

    turbine. The outside diameter of a radial-inflow turbine is considerably

    larger than the rotor tip diameter, due to the stator and inlet scroll or torus.

    Compared to an axial-flow turbine, the radial-inflow package diameter maybe twice as large or more. However, the axial length of the package is

    typically considerably less than for an axial turbine when the inlet and

    diffuser are included. Thus, if the envelope is axially limited but large in

    diameter, a radial-inflow turbine may be best suited for the application,

    considering performance requirements can be met.

    For auxiliary turbine applications where free run may be encountered,

    radial-inflow turbines have the advantage of lower free-run speed than an

    axial turbine of comparable design-point performance. Figure 11shows the

    off-design performance characteristics of both radial-inflow and axial-flowturbines. At higher shaft speeds, the reduction in mass flow for the radial-

    inflow turbine leads to lower torque output and a lower free-run speed.

    Because of the change in radius in the rotor, the flow through the rotor must

    overcome a centrifugal pressure gradient caused by wheel rotation. As shaft

    speed increases, this pressure gradient becomes stronger. For a given overall

    pressure ratio, this increases the pressure ratio across the rotor and

    decreases the pressure ratio across the stator, leading to a reduced mass flow

    rate. A complete description of this phenomenon and its effect on relative

    temperature at free-run conditions is presented by Mathis [23]. However, therotor disk weight savings from the lower free-run speed of a radial-inflow

    turbine is offset by the heavier containment armor required due to the

    increased length of a radial-inflow turbine rotor compared to an axial

    turbine.

    Radial-Inflow Turbine Performance

    The literature on performance prediction and loss modeling for radial-

    inflow turbines is substantially less than that for axial-flow turbines. Wilson[2] states that most radial-inflow turbine designs are small extrapolations or

    Copyright 2003 Marcel Dekker, Inc.

  • 5/22/2018 7- Fundamentals of Turbine Design

    49/62

    interpolations from existing designs and that new designs are executed using

    a cut-and-try approach. Rodgers [24] says that minimal applicable

    cascade test information exists (such as that used to develop many of the

    axial-flow turbine loss models) and that exact analytical treatment of the

    flow within the rotor is difficult due to the strong three-dimensionalcharacter of the flow. Glassman [1] presents a description of radial-inflow

    turbine performance trends based on both analytical modeling and

    experimental results and also describes design methods for the rotor and

    stator blades. More recently, Rodgers [24] has published an empirically

    derived performance prediction method based on meanline quantities for

    radial-inflow turbines used in small gas turbines. Balje [3] presents analytical

    performance predictions in the form of efficiency versus specific speed and

    specific diameter maps.

    For our purposes, we will use the results of Kofskey and Nusbaum[25], who performed a systematic experimental study investigating the effect

    of specific speed on radial-inflow turbine performance. Kofskey and

    Nusbaum used five different stators of varying flow area to cover a wide

    range of specific speeds (0.2 to 0.8). Three rotors were used in conjunction

    with these stators in an attempt to attain optimum performance at both

    extremes of the specific speed range. Results of their testin