http://pia.sagepub.

com/
Energy Engineers, Part A: Journal of Power and
Proceedings of the Institution of Mechanical
http://pia.sagepub.com/content/224/2/261
The online version of this article can be found at:

DOI: 10.1243/09576509JPE818
2010 224: 261 Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy
A Guha and B Smiley
Experiment and analysis for an improved design of the inlet and nozzle in Tesla disc turbines

Published by:
http://www.sagepublications.com
On behalf of:

Institution of Mechanical Engineers

be found at:
can Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy Additional services and information for

http://pia.sagepub.com/cgi/alerts Email Alerts:

http://pia.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.nav Reprints:

http://www.sagepub.com/journalsPermissions.nav Permissions:

http://pia.sagepub.com/content/224/2/261.refs.html Citations:

at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from

261
Experiment and analysis for an improved design
of the inlet and nozzle inTesla disc turbines
A Guha
,
and B Smiley
Aerospace Engineering Department, University of Bristol, Bristol, UK
The manuscript was received on 19 May 2009 and was accepted after revision for publication on 4 September 2009.
DOI: 10.1243/09576509JPE818
Abstract: In this article, the performance of the inlet to a Tesla disc turbine has been studied.
The losses in the inlet and nozzle are known to be a major reason why the overall efciency of
disc turbines is not high. A new nozzle utilizing a plenum chamber inlet has been designed
and tested here. Experiments have demonstrated less than 1 per cent loss in total pressure
in the new design compared to losses in the range 1334 per cent for the original nozzle and
inlet. Other than the dramatic improvement in loss reduction, the newplenum-integrated nozzle
achieves a considerable enhancement in the uniformity of the jet. This has been demonstrated
here both by experimental traverses of Pitot tubes as well as computational uid dynamics
studies. The greater uniformity of the jet means that a single Pitot measurement approxi-
mately positioned at the centre of the jet would record a value close to the true centre-line
total pressure, and that calculations based on centre-line values of total pressure would give,
to a good accuracy, the average loss coefcient of the nozzleinlet assembly. The uniformity of
the jet also means that all disc passages would receive uniform inlet conditions; this should
improve the performance of the rotor thereby further enhancing the overall efciency of theTesla
turbine.
Keywords: Tesla turbine, efciency, power, nozzle, rotor, jet, pitot traverse, computational uid
dynamics
1 INTRODUCTION
In this article, we report design improvements to a
Tesla disc turbine and an experimental and theoret-
ical investigation into its operating characteristics. It
has previously been recognized [13] that the perfor-
mance of the nozzle and the inlet is a limiting factor
for the overall efciency of such turbines. The detailed
ow features of an existing nozzleinlet assembly
are analysed here and the sources of major losses
are identied. This then led to an improved design.
Experimental andcomputational results showthat the
performance of the new nozzleinlet assembly is sig-
nicantly improved. This would have a substantial

Corresponding author: Aerospace Engineering Department,

University of Bristol, University Walk, Bristol BS8 1TR, UK.
email: A.Guha@bristol.ac.uk; guha.a@rediffmail.com

Now at: Mechanical Engineering Department, Indian Institute of

Technology, Kharagpur 721302, India.
impact on the overall performance of Tesla disc tur-
bines. In reference [2], Rice writes: In general, it has
beenfoundthat theefciencyof rotor canbeveryhigh,
at least equal to that achieved by conventional rotors.
But it has proved very difcult to achieve efcient noz-
zles inthe case of turbines. [] As a result, only modest
machine efciencies have been demonstrated. This
article therefore addresses and solves a major issue
in the design of Tesla disc turbines that seems to
have seriously affected their development for over
50 years.
1.1 Operating principle
The Tesla disc turbine is a bladeless turbine invented
by NikolaTeslain1913[4, 5]. Fluidis injectedthrougha
nozzle nearly tangentially onto an array of co-axial at
discs. As the uidmoves throughthe gaps betweenthe
discs shear stresses arise due to the difference in tan-
gential velocities of the discs and the uid. The shear
stresses acting over the surface area of the discs give
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
262 A Guha and B Smiley
rise to a torque and allow work transfer from the uid
to the rotor. The uid exits the rotor through holes in
the discs near the shaft with a throughow velocity in
the axial direction.
1.2 Perspectives of development inTesla turbines
The disc turbine showed early promise. Initially Tesla,
with the help of Allis Chalmers Company built a
turbine with a power output of 500 kW [6]. After test-
ing, the discs were inspected and found to have been
stretched by radial stresses resulting from the high
angular velocity of the turbine. It is likely that the
stretching was due to the choice of steel used which
hada lowyieldpoint [6]. This setback andlack of funds
caused Allis Chalmers Company to stop development
of the Tesla turbine and focus instead on the Curtis
and Parsons type of Impulse and Impulse-Reaction
machines [6]. From the 1950s onwards there was a
resurgence of interest in the Tesla turbine and a num-
ber of disc turbines were built and tested [1, 3, 79].
In these experimental investigations, the efciencies
demonstrated by the Tesla turbine have been in the
range 14.635.5 per cent [10], which is low compared
to modern day gas turbines which has an efciency of
around 90 per cent [11].
Eventhoughthe efciency of theTesla turbine is low
comparedtoconventional turbomachineryit maynd
application in special areas where it has an advantage
over bladed turbines. The rst advantage is the sim-
plicity of design and manufacture. It is also relatively
inexpensive. The turbine could be useful in situations
where very viscous or non-Newtonian working u-
ids are used, or with non-conventional fuels such as
biomass. It is believed that Tesla turbines can cope
better with particle-laden two-phase ows because of
the self-cleaning nature of the discs (references [12] to
[16] describe the general aspects of two-phase ows).
Other than the nozzleinlet assembly, losses occur
in the disc rotor itself. Analytical investigations have
been published with the aim of predicting the max-
imum isentropic efciency of the disc rotor. A study
conducted by Lawn and Rice [17] concluded that
the maximum theoretical efciency of the disc rotor
was 81 per cent. This agrees with the ndings of
Allen [18], who predicted possible efciencies exceed-
ing 80 per cent. More recently, a paper by Rice [2]
predicts that the maximum efciency of the rotor
could exceed 95 per cent. The newdesign described in
this article paves the way for the efciency of the rotor
to be determined experimentally for the rst time.
The Tesla turbine, to date, has not been used com-
mercially. This is mainly due to the low turbine ef-
ciencies demonstrated. Since the gas turbine cycle has
a poor work ratio (thereby the component efcien-
cies having a strong impact on the cycle efciency),
the efciency of the Tesla turbine has to be improved
substantially over its current values for its use in place
of the conventional turbines even in niche application
areas.
1.3 Difference between experimental and
theoretical efciency
It can be seen that there is a large difference between
the theoretical maximumefciency of the disc turbine
and the efciencies demonstrated experimentally. In
his paper of 2003 [2], Rice states that there is little to
no literature devoted to the ows that cause the main
losses in the Tesla turbine.
Investigators have focused their attention on the
disc rotor and little attention has been paid to other
key components in the turbine such as the nozzle and
the exhaust. As a result, no attempt has been made to
measure losses in these components individually and
apparently no research has gone into improving their
design.
The discrepancy between theoretical and experi-
mentally demonstrated efciencies could also be due
to the difculty faced by investigators when trying
to optimize the turbine. There are many parameters
which can be changed and there is extensive cross-
coupling between them. After the success of Whittle
and von Ohain, the gas turbine became the centre
point of research and development, and the under-
standing of its performance and optimization has
reached quite a mature stage [11, 1923]. The under-
standing of the performance of Tesla turbines is not
nearly as thorough. This has resulted in lower than
optimum efciencies.
1.4 The scope of the present study
In this study, a detailed investigation into the inlet
and nozzle of the Tesla turbine is conducted. Rice [2]
states that the efciency of the disc rotor can be very
high, at least equal to that achieved by conventional
rotors, but that it has proved very difcult to achieve
efcient nozzles in the case of Tesla turbines. Rice [2]
goes ontosay that there are inherent losses as the uid
enters the rotor because the nozzles in Tesla turbines
are necessarily long and inefcient.
In an earlier phase of this project a Tesla turbine
was designed, built, and tested [3]. A thorough anal-
ysis of the turbine is presented in reference [3]. The
inlet and nozzle of this turbine were designed on
the basis of knowledge reported in the existing lit-
erature and therefore suffer from similar limitations.
The inlet was assembled from off-the-shelf pneu-
matic parts and was long and complex. However, the
inlet incorporated an innovation in that it allowed
nozzles with different geometries to be interchanged
without any modication of the turbine casing being
necessary.
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 263
The maximumefciency of this turbine was approx-
imately 25 per cent [3], which is similar to the ef-
ciencies found in the literature. A detailed study into
the performance of the inlet presented in this article,
which incorporates experimental and computational
uid dynamics (CFD) analysis, showed that the noz-
zle and inlet were responsible for losses in stagnation
pressure up to 35 per cent. This conrms that Rice [2]
was correct to suggest that the nozzle and inlet were
critical components in the Tesla turbine.
A new inlet was therefore designed, manufactured
and tested. The ndings are reported in this article.
The new design retained the exibility of the test rig
described in reference [3], but allowed losses in stag-
nation pressure due to the nozzle and inlet to be
reduced to below 1 per cent. This could be signicant
for achieving commercial viability of theTesla turbine.
Section 2 outlines methods for calculation of the
performance of the turbine nozzle and inlet, and
applies them to existing designs to assess and identify
problem areas. Section 3 then describes the ratio-
nale for the design of a new nozzle and new inlet
based on the knowledge gleaned from the calcula-
tions of section2. Section4 describes the experimental
set-up used for the present study, and compares the
experimental results for the performance of the new
designs with that for the existing designs of the tur-
bine nozzle and inlet. Appendix 2 describes a CFD
analysis of various nozzleinlet assemblies. Appendix
3 presents analytical formulation for designing and
analysingplenumchambers, together withthenumer-
ical scheme to solve the equations.
2 METHODOF ANALYSIS FORTURBINE
NOZZLE ANDINLET
2.1 Description of turbine inlet and nozzle
In this section, the turbine that was built in an earlier
phase of this project [3] is described with particular
attentionpaidtothenozzleandinlet. Methods arepre-
sentedwhichallowtotal pressure losses inanarbitrary
turbine inlet to be determined. Using these meth-
ods an estimate for the loss coefcient in the current
turbine inlet and nozzle is calculated.
2.1.1 Description of turbine
The main features of the turbine were designed
according to experience noted in references [2], [5],
[6], [8], and [9]. The disc diameter is 92 mm (3.6 in),
the thickness of each disc is 0.9 mm, and the rotor-
to-housing diametrical clearance is 0.3 mm. Anoverall
viewof the turbine canbe seeninFig. 1. The discs have
a single central outlet port, since this conguration
was found to be more efcient by Davydov and Sher-
styuk [9] and Rice [2]. In order to accommodate the
outlet of the uid, the shaft is supportedas a cantilever
by means of bearings inside the base-plate (Fig. 1(d)).
2.1.2 Description of nozzle
The overall design of the turbine is very exible allow-
ing parameters to be varied in order that their effect
on the performance of the turbine can be measured.
It is possible to change the number of discs, disc spac-
ing, and geometry of the nozzle without changing the
base-plate or housing. The geometry of the nozzle can
be changed using an interchangeable nozzle insert. As
a design compromise it was necessary that the nozzle
incorporated a 90

bend just before the exit plane. The

nozzle insert can be seen in its position in Fig. 1(c).
The nozzle outlet area is rectangular. Inthis article, the
major dimension of the nozzle outlet is called height
and the minor dimension is called width. The height
of the slot-shaped nozzle can be adapted to the vari-
able axial extent of the rotor (which depends on the
number of discs used) by using inserts with the shape
of the nozzle channel, this is so that the unused area of
the nozzle can be blocked and the jet can be directed
more efciently to the rotor avoiding major leakages.
(However, it is suspected that the inserts may not have
blockedthe intendedarea completely, allowing a small
amount of leakage.)
2.1.3 Description of inlet
The inlet and nozzle for the turbine are shown in
Figs 1(b) and (c) respectively. The working uid is
delivered by 8 mm hosing to a series of off-the-shelf
pneumatic ttings. First, there is a push-on-hose t-
ting which attaches via an adapter to an elbow which
channels the ow into the base-plate. The elbow con-
tains a static pressure tapping and a total pressure
probe. The duct drilled into the base-plate aligns with
the hole in the nozzle insert allowing the uid to
enter the nozzle. The seal between the two sections
is maintained by an O-ring.
2.1.4 Denition of nozzle and inlet
For the purposes of this article, the turbine inlet is
dened as the section of duct from the start of the
elbow tting until the beginning of the nozzle insert.
The nozzle is denedas the uidpathstarting fromthe
6 mm entrance to the nozzle insert up until the point
at which the uid exits the nozzle.
2.2 Analytical tools for predicting stagnation
pressure drop along ducts
Empirical methods for predicting total pressure losses
caused by the nozzle and inlet were used to estimate
the potential magnitude of losses in the inlet and
provide physical insight. It is suggested in Crane [24]
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
264 A Guha and B Smiley
Fig. 1 (a) Overall view of the test rig and turbine. Keys: (1) pressure regulator, (2) air cylin-
der, (3) supply line, (4) turbine housing, (5) base-plate, (6) spindle, and (7) data logger.
(b) Detailed viewof turbine inlet. Keys: (1) supply line, (2) base-plate, (3) pneumatic ttings,
(4) static pressure tapping, (5) superimposed image of internal diameter changes of inlet,
(6) elbow tting, and (7) Pitot probe. (c) Detailed view of turbine housing. Keys: (1) nozzle
insert, (2) entrance to nozzle, (3) nozzle exit, (4) exhaust hole, (5) static pressure tappings
(one of the nine labelled), (6) turbine housing, and (7) potential uid leak path. (d) Detailed
view of turbine rotor. Keys: (1) base-plate, (2) interface between the inlet and the entrance
to the nozzle insert, (3) bolts holding the discs (one of the three labelled), (4) locating pins
(one of the three labelled), (5) discs, and (6) bolt holes connecting base-plate to housing
(one of ve labelled)
that losses in ducts are caused by four main mecha-
nisms.
1. Friction against the walls of the duct.
2. Viscous eddies causedby abrupt or gradual changes
in duct cross-sectional area.
3. Pressure gradients and secondary ows caused by
changes in the duct direction.
4. Faulty seals leading to leakage of uid to the
atmosphere.
Methods for predictinglosses causedbythesemech-
anisms are examined in turn.
2.2.1 Friction against the walls
In any duct system, mechanical energy is lost through
viscous interaction between the uid and the walls.
The relationship between wall friction and pressure
drop can be expressed using Darcys equation [24, 25]
p
f
=
1
2
k
f
u
2
(1)
k
f
= F
l
d
(2)
2.2.2 Changes in duct cross-sectional area
When an abrupt enlargement in duct cross section
occurs, the ow cannot follow the contour of the
duct causing separation. Turbulent eddies form in the
region of separation causing energy to be dissipated
in the form of heat. An expression for the pressure
loss coefcient can be found analytically by com-
bining the continuity, steady ow momentum and
incompressible energy equations [25]
p
e
=
1
2
k
e
u
2
i
(3)
k
e
= (1
2
)
2
(4)
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 265
In equation (3), the velocity in the duct immediately
before the abrupt enlargement is used.
When uid ows through an abrupt contraction a
vena contracta is formed in the pipe directly down-
stream of the contraction. A region of separated ow
is formed between the vena contracta and the wall,
and most of the loss of total pressure takes place in
the downstream expansion of the ow cross-sectional
area from the vena contracta to that of the duct [25].
The pressure loss can be calculated from equation (5)
p
c
=
1
2
k
c
u
2
o
(5)
k
c
= 0.33
2
0.18
1
+0.5 (6)
In equation (5), the velocity in the duct immediately
after the abrupt contraction is used. In equation (6),
we have tted a quadratic polynomial to the empirical
data for k
c
given in reference [25].
2.2.3 Changes in the ow direction
When uid ows along a curved path a radial pressure
gradient forms to provide the necessary centripetal
acceleration. This results in an increase in pressure
towards the outside of the bendanda decrease inpres-
sure along the inside of the bend. Apressure maximum
occurs at the midpoint of the bend on the outer side
and a pressure minimum occurs on the inner side of
the bend adjacent to this. This state necessitates the
existence of two adverse pressure gradients one on
the outer side of the bendbefore the midpoint andone
onthe inside of the bend after the midpoint. If the cur-
vature of the bend is sufciently severe these adverse
pressure gradients can cause ow separation with the
associated total pressure losses.
The pressure gradients in the bends also cause sec-
ondary ows in the form of counter-rotating eddies
whose effect is superimposed onto the primary ow
eld. Theeffect of theseeddies canpersist downstream
for up to 75 times the pipe diameter [25]. Crawford
et al. [26] present a method to estimate the total pres-
sure losses caused by a bend if the bend radius, pipe
diameter, Reynolds number, andsurfacefrictionfactor
are known.
2.2.4 Faulty seals leading to leakage of uid to
the atmosphere
The static pressure inside the inlet duct is higher than
atmospheric pressure. This means that if the duct is
not properly sealed uid will leak to the atmosphere.
Any uid which escapes to the atmosphere bypasses
the nozzle and performs no useful work. Two major
potential leak paths exist in the turbine. First, uid
could leak at the interface of the nozzle insert and
the turbine housing (as shown in Fig. 1(c)); second,
a part of the jet may not enter between the discs, thus
bypassing the rotor.
Table 1 Calculated loss coefcients for inlet and nozzle
Loss mechanism
Loss
coefcient
Friction against wall 0.27
Changes in inlet duct diameter in three places
(all referred to the velocity in the 6 mm
diameter section)
0.86
90

bend in the elbow 1.65

90

bend in the nozzle 0.94

2.3 Relative magnitude of different loss
mechanisms in the inlet
Using the empirical methods presented in section 2.2
the pressure loss coefcients for the inlet duct and
nozzle are estimated. Table 1 shows the calculated
loss coefcients of the various sections and ow fea-
tures in the inlet and nozzle. The above formulae
are valid only for incompressible ow. For a nozzle
exit area of 1.5 6.3 mm, a simple one-dimensional
(1D) isentropic analysis estimates the Mach number
in the inlet duct to be within the incompressibility
limit. However, as the nozzle exit area is increased,
the Mach number in the duct increases and the ow
would need to be treated as compressible. There are
also unknown quantities of leakage, which increases
with increasing pressure in the system. (Direct mea-
surement of the mass owrate at the inlet and the
exit of the turbine can determine the overall leakage,
but such instrumentation was not available for the
present investigation.) Thus, it is difcult to estimate
the exact loss empirically. The empirical analysis, how-
ever, brings out the relative magnitudes of losses and
the qualitative physics well. It also shows that a sig-
nicant amount of loss can be incurred in a relatively
short inlet that is designed not keeping the loss mech-
anisms in perspective. At any rate, the actual overall
loss has been measured experimentally in this work.
2.4 Performance of the nozzle
2.4.1 Background
The purpose of a nozzle is to convert energy stored
in the form of pressure into directed kinetic energy.
For subsonic ow, this is achieved by reducing the
duct area in the direction of the ow [27]. This
induces a favourable pressure gradient causing the
uid to be accelerated through the nozzle. Separa-
tion is not expected because of the presence of the
favourable pressure gradient; as a result, the efciency
of nozzles is usually very high often exceeding 96 per
cent [28, 29].
For small-sized nozzles with throat widths of less
than 3 mm, the boundary layer can occupy a signif-
icant portion of the nozzle area [30]. Flows in these
nozzles are characterized by small Reynolds numbers
(Re < 10
5
) and are laminar [30]. The relative thickness
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
266 A Guha and B Smiley
of the boundary layer can result in reduced nozzle
performance because of increased viscous losses [31].
In nozzles of this kind it is found that total pressure
losses decrease as the Reynolds number of the ow
increases [30], since the thickness of the boundary
layer decreases with increasing Reynolds number.
2.4.2 Ratio of total pressures
The total pressure remains constant in isentropic ow
in the absence of work transfer. For a real adiabatic
owwithout work transfer losses due to friction mani-
fest themselves as a drop in total pressure. Benedict
et al. [32] states that the ratio of total pressures is a sig-
nicant but little discussed parameter which gives a
clear thermodynamic indication of the loss in terms of
entropy. Whenstudying nozzle performance the use of
this parameter is convenient when an accurate mea-
surement of stagnation temperature is not possible.

nozzle
=
p
0o
p
0i
(7)

nozzle
is calculated here by dividing the outlet gauge
total pressure by the inlet gauge total pressure. Gauge
total pressure is used because it represents the energy
available to do useful work (i.e. although uid with an
absolute pressure of 1 bar has energy in the form of
pressure, this energy cannot be usefully extracted).
2.4.3 Effects of 90

bend combined with a change

in ow cross-sectional area
The loss in total pressure is not the only relevant
parameter in assessing the performance of the noz-
zle, but we argue that there are also other important
features such as the ow uniformity over the nozzle
outlet area that characterize the overall performance
of the nozzle.
As stated in section 2.1.2, the exit plane of the noz-
zle is at 90

to the plane of the inlet. It is known that

a 90

bend in a duct will distort the velocity prole in

the jet causing higher velocity owaround the outside
of the bend and low velocity ow on the inside [28].
This is not desirable for the nozzle in a disc turbine
because it wouldcause some discs tosee a higher mass
owrate than others. However, although there is liter-
ature describing the ow characteristics of a pipe of
constant cross-section undergoing a 90

bend, it is not
known how the presence of a contraction (the nozzle
in the present case) and resulting favourable pressure
gradient directly after the 90

bend would affect the

ow eld.
2.4.4 Computational investigation of nozzle exit
ow uniformity
It was therefore decided to carry out a CFD simula-
tionof the oweld inthe nozzle. Analysis was carried
out using a NavierStokes code with turbulence mod-
elling (k) in FLUENT [33]. Geometry and Cartesian
meshes were generated using Gambit [34].
The cross section of the inlet portion of the nozzle is
circular but the nozzle exit cross section is rectangu-
lar. This meant that it was not possible to create a 2D
model by taking a 2D slice of the nozzle. It was nec-
essary to create a 2D model of the nozzle insert that
maintained the most important characteristics of the
real ow. The two most important parameters which
needed to be modelled were the geometry (i.e. the 90

bend, andthe reductionincross-sectional area caused

by the nozzle). The ratio of nozzle inlet to exit area was
representedinthe 2Dmodel as the ratio of nozzle inlet
toexit width. As far as possible the rest of the 2Dgeom-
etry was based on the dimensions of the real nozzle
insert.
Three cases were studied to replicate the conditions
found in different Tesla turbine nozzles and these are
summarized in Table 2:
Case (a): 1.5 6.3 mm plenum-integrated nozzle. The
uid froma stationary state inside aninnite reservoir
exits through the nozzle. It would be expected that the
ow in the jet would be very uniform with little loss in
total pressure. This case was used as a benchmark to
compare the other cases against.
Case (b): 1.5 6.3 mm original nozzle. This case repre-
sents a nozzle that has an inlet to exit area ratio of 3.0.
A nozzle of this type was used in most tests reported
by Hoya and Guha [3] as it allowed a rotor containing
six discs to be tested. This case should demonstrate
the effect of the pressure gradients caused by a 90

bend coinciding with the favourable pressure gradient

caused by an area contraction.
Case (c): 2.0 20 mm original nozzle. This case shows
the effect of using a nozzle where the exit area is larger
than the inlet area this kind of nozzle was used when
testing the rotor with a large number of discs. In this
case, the favourable pressure gradient associated with
the nozzle is removed.
Details of the CFD investigation and results for the
above cases are given in Appendix 2.
Table 2 Summary of CFD tests runs on ow eld in the
nozzle
Case Flow type
Width
(mm)
Length
(mm)
Inlet area
(mm
2
)
Exit area
(mm
2
)
(a) Reservoir 1.5 6.3 Innite 9.45
(b) Duct ow with 90

bend
1.5 6.3 28.3 9.45
(c) Duct ow with 90

bend
2.0 20 28.3 40.0
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 267
2.5 Notes on inlet performance
2.5.1 Choking upstream of the nozzle
Choking occurs whenthe ratioof inlet total pressure to
outlet static pressure across a converging owpassage
becomes higher than a certain value, the limiting ratio
is 1.89 for the isentropic ow of a perfect gas of isen-
tropic index 1.4 (air) [25]. At choked conditions, the
ow velocity at the minimum cross-section becomes
sonic and the non-dimensional mass owrate cannot
rise any further (if no variation in p
0
or T
0
occurs).
In the test rig, it was found that for some of the
nozzles, the inlet area (which was xed for all cases
to have a common connection to the compressed air
supply) was smaller than the nozzle outlet area (which
was varied according to how many discs were used in
the test Tesla turbine). This meant that choking would
occur in the duct upstreamof the nozzle rather than at
the throat of the nozzle. This is undesirable for several
reasons.
1. The ow velocity in the duct would be high. Since
pressure losses increase with the square of the
velocity, as shown by equation (1), this would lead
to high total pressure losses [24].
2. The losses in total pressure in small nozzles
decrease as the Reynolds number increases [30].
Thus, the best performance wouldbe reachedwhen
the ow velocity through the nozzle reached its
maximumvalue. If the owchokes upstreamof the
nozzle the maximum velocity through the nozzle
may not be reached.
3. The supposed nozzle would act as a diffuser.
Choking upstreamof the nozzle may also mean that
a convergentdivergent passage may be formed, and
hence supersonic ow may occur. This needs further
investigation.
2.5.2 Difculty of determining stagnation pressure
at nozzle inlet
Another problemfacedby previous investigators using
this turbine [3] was that, due to geometric constraints,
it was not possible to measure the total pressure
exactly at the inlet of the nozzle. The closest place
where the measurement could be made was in the
elbow tting upstream of the nozzle. There were a
few abrupt changes in the ow cross-sectional area
between the elbow tting and the nozzle. Hoya and
Guha [3] used empirical methods to estimate the
pressure drop, but accuracy of this method for a
compressible ow regime is not clear.
3 DESIGNOF NEWINLET
It has been shown that the Tesla turbines inlet and
nozzle can be responsible for signicant losses. Here
a design methodology has been formulated to reduce
these losses by integrating a plenumchamber withthe
inlet of the Tesla turbine.
3.1 Requirements of plenumchamber inlet for a
Tesla turbine
From studies reported in references [3] and [10], and
the available literature, a number of design require-
ments have been identied to which an improved
nozzle design must conform.
1. The nozzle must provide similar mass ow to each
disc passage.
2. The nozzle and inlet demonstrate low total pres-
sure losses over the inlet pressure range used by the
turbine.
3. Multiple different nozzle geometries must be able
to be tested without modifying the inlet or turbine.
4. It must be possible to test the efciency of the
nozzle separately to the efciency of the whole
turbine.
5. The inlet must allowthe total pressure and temper-
ature to be measured directly before the nozzle.
3.2 Working principles and design of plenum
chamber
A plenum chamber is a high pressure settling tank in
which a uid is brought to near-stagnation state. By
the principle of conservation of mass this necessitates
a large increase in ow cross-sectional area.
Although plenum chambers are common compo-
nents in many engineering systems [35] there is little
discussion in the literature devoted to them. This
fact is commented upon by Lau et al. [36]. There is
no guidance on how a plenum chamber should be
designed or how the design of a plenum chamber
would affect its performance. Several investigations,
however, havebeenmadeintothepressureloss caused
by cylindrical plenum chambers with aligned and
non-aligned inlets and exits [36, 37].
3.2.1 Character of ow in a plenum chamber
Harrison and Klemz [38] conducted a computational
study into the ow in plenum chambers by numer-
ically solving the 2D NavierStokes equations with
turbulence modelling using a nite difference scheme.
They concludedthat the owinside the plenumcham-
ber could be termed as re-circulating since the ow
had no dominant direction. Sparrow and Bosmans
[37] found experimentally that the pressure loss in the
plenum chamber was insensitive to Reynolds num-
ber from which they concluded that inertial losses
dominate frictional losses. They also found a large
swirling component in the uid in the plenum cham-
ber. When studying the pressure loss caused by a
cylindrical plenum chamber with axially aligned inlet
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
268 A Guha and B Smiley
andoutlet, Lauet al. [36] foundthat, intherangetested
l/d < 10, the pressure loss was always lower than the
loss expected for the combined effects of an abrupt
enlargement andanabrupt contractionoccurringsep-
arately. They also found that the pressure loss in the
plenum chamber increased with the length of the
plenum chamber. This suggests that, as the plenums
lengthincreases, the abrupt enlargement andcontrac-
tion assert their separate identities to a greater extent
(i.e. the ow loses the character given to it by the
abrupt enlargement and becomes more uniform (like
a fully developed pipe ow)).
Although the above details provide a description of
the ow structure inside a plenum chamber, they do
not provide the analytical tools necessary to design a
plenum chamber.
3.2.2 Description of new plenum-integrated nozzle
In order to address the paucity of information in
the literature, a new approach for the design of the
plenum chamber was formulated, which is described
inAppendix 3. Anite difference approachtosolve the
1D isentropic ow equations was used to predict the
time-variation and steady state values of all ow vari-
ables at different key points in the oweld, including
thevelocity of theowat theinlet totheplenumcham-
ber. The results from this analysis showed that for a
nozzle with a h = 6.3 mm and w 3 mm, the veloc-
ity of the ow in the plenum inlet was sufciently
small that the uid could be considered incompress-
ible. The diameter ratio required to produce a near
stagnant ow inside the plenum chamber was cal-
culated with two different approaches continuity
equation and pipe ow loss equations. This analy-
sis showed for a diameter ratio of 0.04 the difference
between the total and stagnation pressure would be of
the order of 0.1 per cent. The length required for reat-
tachment was predicted using experimental sudden
enlargement data [39]. Full details of this approach
can be found in Appendix 3. The nal dimensions of
the designed plenum chamber are as follows: inlet
diameter 6 mm, plenum diameter 25 mm, plenum
length 150 mm. Stress calculations were undertaken
to ensure that the plenum chamber would be able to
contain pressures of up to 10 bar safely.
The plenumchamber was manufactured and tested
withtheoldandnewnozzles. Theresults arepresented
in section 4.
3.2.3 Integration of plenum chamber into existing
turbine rig
The experimental results presented in this paper con-
centrate on the performance of various nozzleinlet
assemblies. The purpose of the present design and
testing was to prove the concept of using a plenum
chamber as the inlet of a Tesla turbine.
2
4
5
6
7
9
8
10
11
12
13
3
1
Fig. 2 Expanded view of the conceptual design for the
plenumchamber integrationwithturbine casing.
Keys: (1) inlet, (2) diffuser, (3) groove for O-ring,
(4) plenum chamber, (5) attachment to turbine
housing, (6) integrated nozzle, (7) spacer, (8) slot
in turbine casing, (9) end plate sealing plenum
chamber, (10) turbine housing, (11) disc rotor,
(12) spindle, and (13) axle
Figure 2 shows the expanded view of a conceptual
design assembly for integrating the plenum chamber
with the turbine casing. The turbine base-plate which
holds the rotor has been excluded from the diagram
to allow the rotor and spindle to be fully visible. It
would be necessary to modify the turbine casing for
the new plenum chamber to be attached. One edge of
the casing is planedat anda hole is cut intothe casing
to expose the discs (this could be achieved by a CNC
machine). The nozzle has a keyed section which slots
into the hole in the turbine casing allowing the uid to
be injected onto the turbine. Flanges extending from
theplenumchamber allowthechamber tobeboltedto
the turbine. The whole assembly would be sealed by a
custom-made gasket. Because the turbine is primarily
impulsive, the internal static pressure is not signif-
icantly elevated above atmospheric pressure. This
means that sealing between the plenum chamber and
the casing is important but not as critical as ensuring
that the inlet of the plenum chamber is sealed.
3.2.4 Estimating losses at the inlet of plenum
chamber integrated in a Tesla turbine
For the purpose of the present study which centres on
the design and performance of the nozzle, the inlet to
the plenum chamber was left as an abrupt enlarge-
ment. However, when building a real Tesla turbine
incorporating the plenum chamber, a diffuser should
be used at the inlet of the plenum chamber, as shown
in Fig. 2. This would allow the majority of the kinetic
energy of the ow to be conserved and converted into
static pressure. Zaryankin [40] provides a method to
determine the losses associated with a diffuser by way
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 269
of formulating a friction factor that takes into account
the twomainsources of loss indiffusers: frictionlosses
and expansion losses. Using this method, the maxi-
mum pressure loss which would be experienced in a
normal plenum operation is of the order of 0.03 bar.
3.3 Summary of the features on the designed
plenumchamber
A plenum chamber inlet with integrated nozzle has
been designed in such a way that the problems identi-
ed withthe original turbine inlet nozzle insection2.2
are mitigated. The plenum provides a near stagnant
ow directly before the nozzle which eliminates the
large pressure gradient identied in the original noz-
zles. This improves the uniformity of the ow across
the nozzle outlet area and reduces losses in total pres-
sure. This design also allows accurate measurement
of the stagnation pressure and temperature directly
before the nozzle allowing the percentage stagna-
tion pressure loss over the nozzle to be accurately
determined. The plenum chamber design allows the
efciency of the nozzle to be measured individually
which would allowthe efciency of the disc rotor to be
directly inferred.
4 RESULTS ANDDISCUSSION
Two principal experimental set-ups and results are
presented here on the performance of nozzles and
inlets. The rst line of investigation centred around
theloss of total pressureinvarious nozzleinlet assem-
blies. The test rig for this aspect of the study is shownin
Fig. 1. The second line of investigation centred around
the uniformity of the jet over the cross section of the
nozzle outlet. The experimental set-up for this part of
study is shown in Fig. 3.
4.1 Efcacy of various nozzleinlet assemblies
The losses associated with different nozzles and inlets
were tested in three scenarios to try to gain the best
insight into where the losses were occurring.
1. Original nozzle and turbine inlet assembly: in this
test, the efciency of the nozzle was tested inside
the turbine casing, replicating the conditions seen
by the nozzle in normal turbine operation. This test
demonstrated the extent to which poor nozzle per-
formance wouldimpact onthe measuredefciency
of the whole machine.
2. Original nozzle with plenuminlet: the original noz-
zles were bolted to the new plenum chamber inlet.
This allowed the performance of the old nozzles on
their own to be measured as the inlet nowprovided
negligible losses.
Fig. 3 Experimental set-up for performing total pres-
sure traverses. Keys: (1) clamp on traverse,
(2) travelling element, (3) bar supporting Pitot
probe, (4) Pitot probe, (5) tube transmitting noz-
zle exit total pressure, (6) nozzle, (7) controller
interfacing the traverse with the computer, (8) G
clamps stabilizing assembly, (9) plenum cham-
ber, (10) tube transmitting plenumtotal pressure,
(11) supply line fromcylinder, (12) traverse power
supply, and (13) Scanivalve
3. New plenum-integrated nozzle: the new plenum-
integrated nozzle was tested to determine the
extent of the efciency improvement that could be
achieved by using a plenum chamber.
Air was used as the working uid and was provided
by a high pressure cylinder with a maximum pressure
of 200 bar. This was reduced to the range 010 bar by a
two-stage pressure regulator. Air was delivered to the
turbine via 8 mm pneumatic tubing. For the turbine
inlet, stagnation and static pressure measurements
were taken at an elbow tting directly before the uid
entered the turbine casing. For the plenum chamber,
pressure measurements were taken by two pressure
tappings one pointing axially and the other radially.
The outlets from the pressure tappings were delivered
to a ZOC 22B Scanivalve via thin exible tubing. The
ZOC 22B Scanivalve has a full scale accuracy of 0.1 per
cent over the range 03.6 bar gauge pressure [41].
4.1.1 Comparison of inlet performances for
1.5 mm nozzle
Figure 4 shows a comparison of the 1.5 mm nozzle
for the three different testing procedures outlined in
section 4.1. It can be seen that, for the original nozzle,
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
270 A Guha and B Smiley
65
70
75
80
85
90
95
100
0.0 1.0 2.0 3.0
C
e
n
t
r
e
l
i
n
e

r
a
t
i
o

o
f

t
o
t
a
l

g
a
u
g
e

p
r
e
s
s
u
r
e
s

(
%
)

Nozzle exit total gauge pressure (bar)
Fig. 4 Comparisonof performance of 1.5 6.3 mmnoz-
zle with different inlets. Keys:
__________
(top) new
plenum-integratednozzle,
__ __ __ __
(middle) orig-
inal nozzle with plenuminlet, and
(bottom) original nozzle with turbine inlet
assembly
thetotal pressureratioacross thenozzleincreases with
increasing inlet pressure. This is because as the inlet
pressure increases, the Mach number of the ow in
the nozzle increases. This causes the boundary layers
to become thinner which reduces the viscous losses in
the nozzle [30, 31]. After the ow is choked, the static
pressure of the jet at the nozzle exit increases with
increasing nozzle inlet total pressure, which results in
an increased uid density in the nozzle. Consequently
the Reynolds number of the ow increases monotoni-
cally with increasing supply pressure when the nozzle
outlet area is kept xed. This manifests itself inthe loss
in total pressure decreasing monotonically with sup-
ply pressure. For the new plenum-integrated nozzle,
the loss in total pressure is very lowand there is scatter
in the data, making it difcult to rationalize any trend.
Comparingthe performance of the different inlets, it
canbeseenthat theoriginal nozzleandturbineassem-
bly has the poorest performance and the plenum-
integratednozzle has the highest performance. Table 3
summarizes the loss in total pressure, as percent of
nozzle inlet total pressure, for various inlet-nozzle
Table 3 Comparison of loss in total pressure, as per-
cent of nozzle inlet total pressure, for various
inletnozzle assemblies (1.5 6.3 mm nozzle
exit)
% loss in total pressure
Total
pressure at
nozzle inlet
(bar)
Original
nozzle and
turbine
assembly
Original
nozzle with
plenum
inlet
New
plenum-
integrated
nozzle
1.5 23 4 0.5
3.5 14.5 2.3 0.4
assemblies (1.5 6.3 mm nozzle exit). Results are tab-
ulatedat twoinlet total pressure values, 1.5and3.5 bar.
As comparedtothe original nozzle andturbine assem-
bly, the new plenum-integrated nozzle has reduced
the loss by a factor of about 4050 (an improvement
of 40005000 per cent).
By comparing the values in the second and the third
columns of Table 3, it can be seen that it is the original
inlet that is responsible for the majority of losses. This
shows that the original nozzle even with the 90

bend
has a reasonable efciency. This is consistent with
the prediction made in section 2.4 that the favourable
pressure gradient causedby the nozzle would, to some
extent, offset the adverse pressure gradients created
by the 90

bend. By comparing the values in the last

two columns of Table 3, it is seen that signicant fur-
ther reduction in total pressure loss is achieved by the
removal of the 90

bend within the nozzle.

4.1.2 Comparison of inlet performances with varying
nozzle width
Similar experiments as insection4.1.1wereconducted
for nozzles of different widths. The new plenum-
integrated nozzle maintained a total pressure loss
below 1 per cent for all nozzles tested up to 3 mm
width. For the original nozzle withturbine inlet assem-
bly, the loss in total pressure increased further by a
signicant amount as nozzle width increased. This
is because as the nozzle area gets closer to the duct
area the ow velocity in the duct gets closer to the
jet velocity. This signicantly increases losses due to
changes in the pipe direction and changes in the pipe
cross-sectional area because these losses scale with
the square of the velocity. Thus, the difference in total
pressure loss between the original inlet and the new
plenum-integratednozzle wouldbe muchhigher than
what is shownbetweenthesecondandfourthcolumns
of Table 3 when the nozzle outlet area increases.
4.1.3 Summary of results on efcacy of nozzleinlet
assemblies
The inlet and nozzle were responsible for signicant
pressure losses in the turbine. Most of this loss was
due to abrupt changes in area and the 90

elbowin the
inlet, and the 90

bend in the nozzle. This conclusively

shows that the inlet andnozzle totheTesla disc turbine
need not be long and inefcient as was concluded in
reference [2].
4.2 Pitot tube traverse results to assess the
uniformity of jets
So far the performance comparison between the noz-
zles has been made by comparing the ratio of centre-
line total pressures. In reality, the comparison should
be made betweenthe average total pressure across the
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 271
ow before and after the nozzle as this accounts for
the fact that the total pressure is not constant across
the jet or the inlet. In order to assess the extent of
ow non-uniformity and to ascertain the extent to
which a single Pitot tube measurement at an approxi-
mately centre-line position represents averaged value,
Pitot tube traverses were conducted in both directions
(width and height) over the nozzle outlet area. The
experimental set-up is shown in Fig. 3. This involved
using a high precision traverse mechanism, which is
digitally precision-controlled through a computer.
The non-uniformity in total pressure is caused by
two major effects. Bends in the inlet and within the
nozzle distorts the ow as explained in section 2. This
will be more prevalent in the height-wise direction in
the original nozzle and turbine assembly. The action
of viscosity (wall effect) also creates non-uniformity in
total pressure. As a percent of nozzle dimension this
will be more pronounced in the direction of nozzle
width (since the widths of the nozzles are smaller than
their respective heights).
Figure5andTable4showillustrativeresults for three
nozzleinlet assemblies. Table 4 shows howeffectively
a Pitot tube approximately positioned at the centre
would record a value close to the true centre-line total
pressure. FromFig. 5 it canbe seenthat the nozzle and
inlet has a very denite impact on the uniformity of
the ow from the nozzle. As would be expected, the
plenum chamber has a very symmetrical prole. The
original nozzles that incorporate the 90

bend tend
to have a higher total pressure towards the outside
of the bend. This is far more apparent on the noz-
zle with a 20 mm width. This is because the outlet
area of that nozzle is larger than the nozzle entrance,
this means that there is no contraction and therefore
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.5 1.0
T
o
t
a
l

g
a
u
g
e

p
r
e
s
s
u
r
e
s

r
a
t
i
o
Normalised distance along the major
dimension of the nozzle
Fig. 5 Results from experimental total pressure
traverses along the major dimension of
the nozzle outlet. Keys:
_______
1.5 6.3 mm
plenum-integrated nozzle, - - - - - 1.5 6.3 mm
original nozzle, and
__ __ __
2.0 20.0 mm original
nozzle (for these tests, the inlet centre-line total
gauge pressure is 0.5 bar)
Table 4 Comparison of area over which Pitot measure-
ment is approximately representative of actual
centre-line total pressure as a function allowable
maximum error
Central portion of jet as % of height over which the
variation in total pressure is within the error limit
Maximum
error (%)
Plenum-
integrated
nozzle
1.5 6.3 mm
Original
nozzle with
turbine
assembly
1.5 6.3 mm
Original
nozzle with
turbine
assembly
2 20 mm
1 81 32 10
2 87 45 20
3 87 54 35
4 87 61 52
5 87 74 52
no favourable pressure gradient. In contrast to this,
the 1.5 mm nozzle has a much less distorted prole
because the favourable pressure gradient caused by
the nozzle counteracts the adverse pressure gradients
caused by the sudden bend. These results match and
verify the 2D CFD results discussed in Appendix 2.
The greater the ow non-uniformity, the greater
would be the difference between the centre-line total
pressure and the area-weighted value. Since in quasi-
1D analysis of performance of turbines and nozzles
only one representative value is usedat eachsectionof
theowpath, efciencies calculatedonthebasis of the
centre-line values may be signicantly different from
their true values (if the ow non-uniformity is severe).
The effects of non-uniformity on area-weighted total
pressures, for various nozzleinlet assemblies, are dis-
cussed in Appendix 2. The results in Appendix 2 show
that for a badly designed nozzle the actual perfor-
mance would be much worse than the efciency
calculated using centre-line total pressure values.
5 CONCLUSION
Inthis work, the performance of the inlet toaTesla disc
turbine has been thoroughly studied and the causes of
loss have been established. It has been shown that the
nozzle and inlet are responsible for a large proportion
of the losses in traditional designs of a Tesla turbine.
The work reports on how to signicantly improve the
performance of the inlet andthe nozzle, inthe past the
poor performance of whichhas oftenbeenassumedto
be unavoidable [2].
A new nozzle utilizing a plenum chamber inlet
has been designed and tested. Experiments have
demonstrated less than 1 per cent loss in total pres-
sure in the newdesign compared to losses in the range
1334 per cent for the original nozzle and inlet. As
compared to the old inletnozzle assembly, the new
designreducedthe loss intotal pressure by a very large
factor (4050). This would signicantly improve the
overall efciency of the Tesla disc turbine. This result
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
272 A Guha and B Smiley
shows that the suggestion in the literature that the
nozzle and inlet were responsible for signicant losses
is true but that the nozzle and inlet can, with careful
design, perform very efciently.
Other thanthe dramatic improvement inloss reduc-
tion, the new plenum-integrated nozzle achieves a
considerable enhancement in the uniformity of the
jet. This has been demonstrated here both by experi-
mental traverses of Pitot tubes as well as CFD studies.
The greater uniformity of the jet means that a single
Pitot measurement approximately positioned at the
centre of the jet would record a value close to the
true centre-line total pressure, and that calculations
based on centre-line values of total pressure would
give, to a good accuracy, the area-averaged loss coef-
cient of the nozzleinlet assembly. The uniformity of
the jet also means that all disc passages would receive
uniforminlet conditions; this should improve the per-
formance of the rotor thereby further enhancing the
overall efciency of the Tesla turbine.
A design has been put forward which would allow
the plenum chamber and nozzle to be integrated with
the current turbine with only minimal changes to the
casing being required. By interchanging the nozzle
with other different geometries, it would be easy to
study how the way in which the uid is injected into
the rotor affects the overall machine efciency.
This design paves the way for the efciency of the
rotor to be determined experimentally for the rst
time. Up till now, experimenters have measured the
overall efciency of the whole turbine. Since this
gure includednozzle losses, rotor losses, andexhaust
losses, it was difcult to determine quantitatively
which component was responsible for the poor per-
formance of the turbine. With a nozzle that can be
removed fromthe turbine casing, the nozzle efciency
can be determined in isolation thus allowing the rotor
efciency to be inferred more accurately.
ACKNOWLEDGEMENTS
The authors would like to express their gratitude to
Renate Stump for her help with the CFD studies with
Fluent as giveninAppendix 2, andtoP. Margaris for his
help with the Pitot traverse experiments. The authors
are grateful to, among others, Keven Chappell, Sam
Beale (Rolls-Royce), Lindsay Clare, Steve Macqueen,
Sandy Mitchell, andthe technicians of the engineering
faculty of University of Bristol.
Authors 2010
REFERENCES
1 Rice, W. An analytical and experimental investigation
of multiple-disk turbines. J. Eng. Power, 1965, 87(1),
2936.
2 Rice,W. Tesla turbomachinery. In Handbook of turboma-
chinery (Ed. E. Logan), 2003, pp. 861874 (Marcel Dekker,
NewYork).
3 Hoya, G. P. and Guha, A. The designof a test rig andstudy
of the performance and efciency of a Tesla disc turbine.
Proc. IMechE, Part A: J. Power and Energy, 2009, 223(A4),
451465. DOI: 10.1243/09576509JPE664.
4 Tesla, N. Fluid propulsion. US Patent 1,061,142, May
1913.
5 Tesla, N. Turbine. US Patent 1,061,206, May 1913.
6 Cairns, W. M. J. The Tesla disc turbine, 2003 (Camden
Miniature Steam Services, Great Britain).
7 Armstrong, J. H. An investigation of the performance of
a modied Tesla turbine. MS Thesis, Georgia Institute of
Technology, June 1952.
8 Beans, E. W. Investigation into the performance char-
acteristics of a friction turbine. J. Spacecr., 1966, 3(1),
131134.
9 Davydov, A. B. and Sherstyuk, A. N. Experimental
researchona disc microturbine. Russ. Eng. J., 1980, 60(8),
1922.
10 McGarey, S. and Monson, P. Performance and efciency
of disk turbines. MEng research project report no. 1203,
University of Bristol, 2007.
11 Guha, A. Optimisation and design of aero gas turbine
engines. Aeronaut. J., 2001, 105(1049), 345358.
12 Guha, A. Auniedtheory of aerodynamic andcondensa-
tionshock waves invapour-droplet ows withor without
a carrier gas. Phys. Fluids, 1994, 6(5), 18931913.
13 Guha, A. Structure of partly dispersed normal shock
waves in vapour-droplet ows. Phys. Fluids A, 1992, 4(7),
15661578.
14 Guha, A. Jump conditions across normal shock waves
in pure vapour-droplet ows. J. Fluid Mech., 1992, 241,
349369.
15 Guha, A. A unied theory for the interpretation of total
pressure and temperature in two-phase ows at sub-
sonic and supersonic speeds. Proc. R. Soc., 1998, 454,
671695.
16 Guha, A. Computation, analysis andtheory of two-phase
ows. Aeronaut. J., 1998, 102(1012), 7182.
17 LawnJr, M. J. and Rice,W. Calculated design data for the
multiple-disk turbine using incompressible uid. Trans.
ASME, J. Fluids Eng., 1974, 96, 252258.
18 Allen, J. Examining the Tesla turbine. In Proceedings of
the 27th Aerospace Sciences Meeting, Reno, NV, 912
January 1989, AIAA Paper 890838.
19 Mattingly, J. D. Elements of gas turbine propulsion, 1996
(McGraw Hill International Editions, Singapore).
20 Cohen, H., Rogers, G. F. C., and Saravanamuttoo, H. I. H.
Gas turbine theory, 4th edition, 1996 (Longman Group
Ltd, Cornwall).
21 Guha, A. Performance and optimization of gas tur-
bines with real gas effects. Proc. IMechE, Part A:
J. Power and Energy, 2001, 215(A4), 507512. DOI:
10.1243/0957650011538631.
22 Guha, A. An efcient generic method for calculating the
properties of combustion products. Proc. IMechE, Part
A: J. Power and Energy, 2001, 215(A3), 375387. DOI:
10.1243/0957650011538596.
23 Guha, A. Optimum fan pressure ratio for bypass engines
with separate or mixed exhaust streams. AIAA J., 2001,
17(5), 11171122.
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 273
24 Crane. Flow of uids through valves, ttings and pipes,
1982 (Crane Co., NewYork).
25 Massey, B. Mechanics of uids, 8th edition, 2006 (Taylor
& Francis, Oxon).
26 Crawford, N. M., Cunningham, G., and Spedding, P. L.
Prediction of pressure drop for turbulent uid ow in
90

bends. J. Process Mech. Eng., 2003, 217, 153155.

27 Anderson Jr, J. D. Fundamentals of aerodynamics, 3rd
edition, 2001 (McGraw Hill, NewYork).
28 Shepherd, D. G. Principles of turbomachinery, 1956 (The
Macmillan Company, NewYork).
29 Shames, I. H. Mechanics of uids, 4th edition, 2002
(McGraw-Hill Professional, NewYork).
30 Butenko, V. A., Rylov, Y. P., and Chikov, V. P. Experi-
mental investigation of the characteristics of small-sized
nozzles. Izv. Akad. Naur SSSR, 1976, 6, 137140.
31 Whalen, V. M. Low Reynolds number nozzle ow study.
MS Thesis, NASA technical memoranum, 1987.
32 Benedict, R. P., Carlucci, N. A., and Swetz, S. D. Flow
losses in abrupt enlargements and contractions. J. Eng.
Power, 1966, 88, 7381.
33 Fluent Inc. FLUENT 6.3 users guide, 20032004 (Fluent
Inc., Lebanon).
34 Fluent Inc. Gambit users guide, 2009, available
fromhttp://web.njit.edu/topics/Prog_Lang_Docs/html/
FLUENT/uent/gambit/help/users_guide/ugtoc.htm.
35 Williams, G. H. Homebuilt hovercraft, Flight Interna-
tional Supplement, 28 January 1965.
36 Lau, S. C., Sparrow, E. M., and Ramsey, J. W. Effect of
plenum length and diameter on turbulent heat transfer
in a downstream tube and on plenum-related pressure
losses. J. Heat Transfer, 1981, 103, 415421.
37 Sparrow, E. M. and Bosmans, L. D. Heat transfer and
uid ow experiments with a tube fed by a plenum
having nonaligned inlet and exit. Trans. ASME, J. Heat
Transfer, 1983, 105, 5662.
38 Harrison, J. A. and Klemz, B. L. An investigation of the
steady uidowwithina simple plenumchamber. High-
Speed Surf. Craft, 1982, 3234.
39 So, R. M. C. Inlet centerline turbulence effects on
reattachment length. Exp. Fluids, 1987, 5, 424426.
40 Zaryankin, A. E. A note on calculating losses in diffuser
elements. Inz.-Fiz. Zh., 1965, 8(4), 531535.
41 Scanivalve Corp. ZOC 22B electronic pressure scanning
module, instruction and service manual, 2005, available
from Scanivalve site www.scanivalve.com.
APPENDIX 1
Notation
a speed of sound
A cross-sectional area
d diameter
F friction factor
h Nozzles major dimension (height)
H enlargement step size = (d
o
d
i
)/2
k pressure loss coefcient
l length
m mass owrate
M mach number
n time step
p absolute static pressure
p
0
absolute total pressure
p gauge pressure
p change in total pressure
R gas constant
t time
T static temperature
u velocity
w Nozzles minor dimension (width)
x
L
reattachment length
diameter ratio = d
i
/d
o
change in quantity
static to total pressure ratio = p/p
0
uid density
ratio of specic heats
ratio of total gauge pressures = p
0o
/ p
0i
Subscripts
c duct contraction loss
e duct expansion loss
f friction loss
i inlet
o outlet
APPENDIX 2 CFDSTUDY OFTHE FLOWIN
THE INLET ANDNOZZLE
As was described in section 2.4.4, CFD analysis with
FLUENT was undertaken to provide insight into the
oweld in three different nozzleinlet combinations.
Simulations were run with a nozzle inlet total pressure
of 1.56 bar, the jet discharging into atmosphere. The
total temperature was taken to be 288 K. The solution
was considered to have converged when the residuals
were below 1e-6.
Figure 6 shows the predictions of FLUENT for the
three geometries. The contours of total gauge pres-
sure are shown. The plenum-like nozzle produces a
very uniformjet with lowlosses in total pressure. Both
nozzles that incorporated the 90

bend at the outlet

had ow separation at the inside edge and a higher
total pressure on the outside than the inside. This ts
with the experimental results (section 4.2) and the
theorythat was presentedinsection2.2. Thetotal pres-
sure prole of the nozzle with a height of 20 mm was
more severely distorted than the nozzle with a 6.3 mm
height. It has been explained in section 2.2.3 that the
outlet area of the nozzle with a height of 20 mm is
greater than the area of the inlet duct and hence the
counterbalancing effect of a favourable pressure gra-
dient that is usually present in a nozzle is not present
here.
Figure 7 shows the variation of total pressure ratio
across the major dimension of the three nozzles as
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
274 A Guha and B Smiley
0.650
0.604
0.558
0.512
0.466
0.420
0.374
0.328
0.282
0.236
0.190
0.144
0.098
0.052
0.006
-0.040
(a)
(b) (c)
Fig. 6 Total gauge pressure contour plots for various nozzles, results of CFD analysis. Keys:
(a) 1.5 6.3 mm plenum-integrated nozzle (ow from top to bottom), (b) 1.5 6.3 mm
original nozzle (owfromtop to lower right), and (c) 2.0 20 mmoriginal nozzle (owfrom
top to right)
0
0.2
0.4
0.6
0.8
1
0 0.5 1
T
o
t
a
l

g
a
u
g
e

p
r
e
s
s
u
r
e
s

r
a
t
i
o
Normalised distance along the major
dimension of the nozzle
Fig. 7 Total pressure proles at nozzle exit, results
of CFD analysis. Keys:
_______
1.5 6.3 mm
plenum-integrated nozzle, - - - - 1.5 6.3 mm
original nozzle, and
__ __ __
2.0 20 mm original
nozzle
predicted by the CFD simulations. This can be com-
pared with Fig. 5, where the measured values of the
same quantities were plotted. There is goodqualitative
agreement between the experimental and computa-
tional results. Area-weighted average total pressure
was calculated from the CFD results for the three
geometries and is presented in Table 5.
Table 5 shows that there is very close agreement
between the CFD and experimental results for the
two nozzles with a height of 6.3 mm. This shows that
the 2D CFD model in these cases sufciently cap-
tures the most salient characteristics of the real 3D
ow. For the experiments with the original nozzle, the
measured values include the loss in the inlet section,
whereas the CFD simulation was applied only to the
nozzle inserts. For the 2 20 mm nozzle, the exit area
is larger than the cross-sectional area of the inlet duct.
This meant a high velocity in the inlet, with corre-
spondingly higher losses there this accounts for the
Table 5 Comparison of area weighted total gauge
pressure ratios ( p
0o,average
/ p
0i
) obtained from
CFD and experiment
Nozzleinlet
Ratio of
total
pressures
from CFD
Ratio of
total
pressures
from
experiment
Percentage
difference
Plenum-integrated nozzle
1.5 6.3 mm
0.95 0.94 1
Original nozzle with
turbine assembly
1.5 6.3 mm
0.66 0.64 3
Original nozzle with
turbine assembly
2 20 mm
0.47 0.41 15
slightly greater difference between the experiments
and CFD results for this case.
APPENDIX 3 DESIGNANDANALYSIS OFTHE
PLENUMCHAMBER ANDTHE NEW
INLETNOZZLE ASSEMBLY
Owing to the lack of information available in the lit-
erature about how to size a plenum chamber, it was
necessary to develop analytical and empirical tools to
predict the ow characteristics in a plenum chamber.
Figure 8 gives an idealized description of the reser-
voir (which represents the cylinder condition after the
pressure regulator) and plenum, in which the impor-
tant locations are numbered. These numbers are used
as subscripts for all variables in the following analysis.
Determination of owvariables: equations and
a numerical scheme
It is assumed that the plenum and the nozzle are ini-
tially at atmospheric pressure. The equations and the
numerical scheme developed belowcalculate howthe
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 275
1 2 3 4
Reservoir Plenum Chamber
Fig. 8 Schematic diagram of plenum chamber used for
unsteady analysis
different ow variables at the numbered locations in
Fig. 8 develop with time as the valve in the high pres-
sure cylinder is suddenly switched on. The following
computation therefore gives an unsteady analysis, the
steady solution is obtained as the asymptotic limit of
the same computations.
For isentropic ow, the mass ow through an ori-
ce can be determined if the inlet stagnation pressure,
the outlet static pressure, the area of the orice,
and stagnation temperature are known. It is evident
that initially when the valve from the cylinder to the
plenum chamber is rst opened, this is an unsteady
problem. The pressure in the plenum chamber would
rise because the mass ow at 2 would be larger than
the mass ow at 4. Eventually, however, steady state
conditions would be reached where the mass owrate
at 2 is equal to the mass owrate at 4.
Writing the conservation of mass for a compressible
uid
m = Au (8)
The speed of sound in an ideal gas and the equation
relating Mach number to velocity can be written as
follows
a =
_
RT (9)
M =
u
a
(10)
Combining equations (8) to (10) results in
m = AM
_
RT (11)
Using the standard relations for isentropic ow [27]
p
p
0
=
_

0
_

=
_
T
T
0
_
/( 1)
=
_
1 +
1
2
M
2
_
/( 1)
(12)
Equation (11) can be written purely in terms of
stagnation quantities and pressure
m =
0
_
p
p
0
_
1/
A

_
2 RT
0
1
_
1
_
p
p
0
_
( 1)/
_
(13)
It is astandardpracticewhenformulatingnumerical
schemes to solve for a steady state problem to formu-
late an unsteady equation where the time derivative
represents the level of error present at that time step.
This allows the solution to be iterated. Convergence
is achieved when the time derivative reaches zero.
When the system reaches steady conditions, the mass
owrate at 2 will equal the mass owrate at 4. This
implies that themass of air insidetheplenumchamber
(point 3) will not change with time
dm
3
dt
= m
2
m
4
= Error in solution (14)
This equation can be represented using a rst-order
nite difference approximation
m
n
3
= m
n1
3
+t( m
n1
2
m
n1
4
) (15)
where
m
2
= f
_
p
3
p
0
, A
2
_
, m
4
= f
_
p
a
p
0
, A
4
_
(16)
Since the cross-sectional area of the plenum cham-
ber is large, it is assumed that the total and static
pressure in the plenumchamber are nearly equal. The
calculation procedure is as follows.
1. Calculate inlet mass owrate using p
01
and p
2
.
2. As a simple approximation, assume that the kinetic
energy at point 2 is lost as the jet enters into the
plenum chamber. As a result p
2
= p
3
= p
03
.
3. Calculate exit mass owrate using p
04
and p
4
. p
04
is
assumed to be the same as p
03
which is calculated
using equation (12). The plenum Mach number is
calculated from equation (11).
4. Calculate the net mass ow out of the plenum
chamber and hence the updated mass of uid
inside the plenum chamber.
5. Calculate the updated pressure in the plenum
chamber using the equation of state. The tempera-
ture is taken to be the temperature at the exit of the
inlet duct.
6. Repeat until the mass of uid in the plenum
chamber reaches a steady state.
Equation (15) was iterated using Matlab for a num-
ber of different conditions: cylinder absolute pressure
and nozzle width were varied.
Figure 9 shows the time evolution of the plenum
inlet and exit Mach numbers. It shows that initially
the inlet duct is choked and the mass owrate through
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
276 A Guha and B Smiley
Fig. 9 Development with time of plenum inlet and exit
Mach numbers. Keys: - - - - - M
2
,
______
M
4
(for an
inlet gauge pressure of 3.51 bar)
the exit duct is zero. As time passes, the mass owrate
through the exit duct increases until it reaches choked
condition. By this time, the inlet Mach number (M
2
)
has decreased signicantly (since the pressure in the
plenum chamber has risen).
This graph shows that it takes up to a second for the
plenumchamber toreachsteadystateconditions. This
was taken into account during experimental testing.
Determination of the diameter and length of the
plenumchamber
The diameter of the plenumchamber is determinedby
two methods: (i) simple continuity consideration and
(ii) model to include viscous losses.
Continuity consideration
The incompressible continuity equation can be re-
arranged to yield the expected plenum velocity for a
given diameter ratio
u
3
= u
2

2
(17)
Prediction of total to static pressure ratio including
viscous losses
To complement the simple analysis above it was
decided to perform an analysis where viscous losses
were considered. A ow can be considered stagnant
if the ratio of the total pressure to the static pressure
is equal to 1. This means that the ow has no bulk-
directed kinetic energy. Benedict et al. [32] performed
analysis on sudden enlargements in pipes for steady
incompressible ow and provide a means of calculat-
ing the change in stagnation pressure and change in
static pressure over an abrupt enlargement
p
03
p
02
= 1 (1
2
)(1 2
2
+
4
) (18)
p
3
p
2
= 1 +2
2
_
1
2

2
_
(1
2
) (19)
These two equations can be combined to form
a relationship for the total to static pressure after
the abrupt enlargement (after ow reattachment has
occurred) as a function of the total to static pressure
before the abrupt enlargement

3
=
2
_
1 +2
2
(1
2
/
2
) (1
2
)
1 (1
2
)(
2
1)
2
_
(20)
Equation (20) is used to determine an appropriate
diameter ratio to produce a near stagnant ow in the
plenum chamber, and is plotted in Fig. 10.
Figure 10 shows that, for diameter ratios lower than
0.25, the ow nearly stagnates inside the plenum
chamber. As an example, equation (17) predicts the
plenumMach number is 0.019 for an inlet Mach num-
ber 0.3 and a diameter ratio of 0.25. Equation (20)
would predict a plenum Mach number of 0.02 for
these owconditions. Thus, the twomethods give very
nearly the same answer.
The other important dimension of the plenum
chamber is the axial length. Lau et al. [36] experimen-
tally showed that for a plenum chamber with a length
to diameter ratio of 10 the owdid not completely lose
its history as the loss coefcient was still lower than
what would be expected from the combined effects
of an abrupt enlargement and an abrupt contraction
occurring separately. This suggests that the ow still
had characteristics of a jet and might not have fully
re-attached. The analysis carried out by Benedict et al.
[32] is only valid after the ow has re-attached so it is
necessary to nd a method to determine the length it
takes for the ow to re-attach.
As no analytical tool to make this prediction could
be found in the literature it was necessary to use
experimental data. So [39] carried out a survey of the
Fig. 10 Ratio of static to total pressure in the plenum
with varying diameter ratios (for various p
2
/p
02
,
refer to Fig. 8 for locations of points)
Proc. IMechE Vol. 224 Part A: J. Power and Energy JPE818
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from
Experiment and analysis for an improved design of the inlet and nozzle inTesla disc turbines 277
Fig. 11 A plot of x
L
/H versus d
3
/d
2
for axisymmetric
sudden-expansion ows, data from reference
[39]
literature relating to reattachment lengths for axisym-
metric sudden enlargement ows. Figure 11 shows
a compilation of the reattachment lengths found by
So. The compiled data does not show any systematic
pattern for the variation of x
L
/H with d
2
/d
1
. However,
this shows that there is approximately a linear rela-
tion between the reattachment length x
L
and the step
height H. The maximum x
L
/H value found by So was
just under 10.
The following empirical method was used to deter-
mine the maximum reattachment length for a chosen
diameter ratio
H =
(d
2
d
1
)
2
=
d
1
2
(
1
1) (21)
It was noted previously that in order to achieve a
nearly stagnant condition inside the plenumchamber
a diameter ratio of 0.25 was necessary. Equation (21)
gives the required value of H as equal to (3/2)d
1
. Since
Fig. 11 shows that the maximum value for x
L
/H is 10.
Therefore, maximum reattachment length would be
15 d
1
.
This shows that if the plenum chamber is 15 times
the inlet diameter then the ow will reattach. It was
decided to design the plenum chamber conserva-
tively so the length was set to 25 times the inlet
diameter.
JPE818 Proc. IMechE Vol. 224 Part A: J. Power and Energy
at INDIAN INSTITUTE OF TECH on September 14, 2011 pia.sagepub.com Downloaded from