Professional Documents
Culture Documents
AND TESTING
URN 03/1246
Contractor
Hydroplan UK
Gilbert Gilkes & Gordon Ltd.
i
ii
EXECUTIVE SUMMARY
E1 Background to the Project
It is recognised that Pelton turbine technology may be used effectively over a relatively wide
range head and flow conditions when compared to other turbine categories and is suitable for
many medium and high head sites. The technology is proven and the overall concept is
reasonably simple.
The purpose of this project has been to research, design and partially test two key components
of a low cost Pelton turbine suitable for smaller UK sites and the export market. These
components are the concentric casing surrounding the turbine and the turbine runner,
specifically investigating bolted runner bucket technology.
From a previous assessment of the Gilkes range of Pelton turbines, two key facts have
emerged from this work:-
• The vertical shaft Pelton turbine configuration offers the most flexibility (and potential
least cost).
• The runner represents 40% of the turbine cost.
This project has concentrated on both of these facts by investigating and developing:-
(i) The Concentric Case Vertical Shaft Pelton (CCVSP).
(ii) Bolted Runner Bucket (BRB) technology.
A thorough testing program has shown the BRB will reduce manufacturing time and improve
production control without compromising on operating performance.
iii
Cost / Benefits Assessment
The new BRB design has the potential to lead to considerable savings in turbine manufacture
while potentially leading to improvements in performance.
Manufacturing time
A significant part of finishing a conventional runner is the bucket grinding and
polishing process. BRB technology will dramatically reduce this time by allowing easy
access to all the surfaces to be machined and the new process eliminates all hand
finishing operations. The lead-time to manufacture a runner is reduced considerably. If
the runner is entirely of a new design and solid modelling of the buckets and tooling are
required, a reduction in time of 68% can be expected. If new tooling only is required, a
reduction in lead-time of 79% can be achieved. A replacement runner can be made to an
existing design with an 89% saving in lead-time.
Hydraulic performance
It is envisaged that hydraulic performance will improve due to greater control over the
bucket surface geometry and finish.
Fatigue considerations
Historically, the problem with bolt-on buckets has been cyclic fatigue and corrosion
exacerbating stress concentrations around the bolt leading to premature failure. The
massive number of applied stress cycles, as high as high as 5x1010 over an operating
lifetime, has been found to make fatigue failure the most likely failure mechanism. The
FEM computer program has the capability to predict fatigue performance and reliable
texts on the subject have been consulted, however the true test for the runner will be field
trials.
iv
Figure E1 - Assembled Test runner
An analysis was undertaken based on the layout of an existing 4 jet vertical shaft Pelton
turbine installed by Gilkes in 1986. The conventional manifolds construction cost,
performance and footprint area were then compared to that of a option 1 type manifold of the
same capacity. CFD was used to model the existing and theoretical cases.
Cost comparison
Prices were sought to manufacture the existing concentric case in the option 1 type design.
The CCVSP was increased in capacity so that velocities in both manifolds were the same.
Figure E2 below shows the existing manifold layout with the CCVSP (in blue) superimposed
over it for comparison. Table E3 provides a comparison of performance.
v
Figure E2 Footprint comparison
vi
Cost / Benefits
The CCVSP design seems to have improved the turbine in three areas.
• A reduction in powerhouse construction costs of 2 to 3% may be possible due to the
reduced footprint area required by the CCVSP.
• Manifold costs may be reduced by 14%
• A modest reduction in headloss may be possible and may improve further if flow baffles
were incorporated into the branch pipe transitions where the major flow separation
occurred. Baffles have successfully used where vortex formation has been a problem.
E5 Conclusions
The partnership between the leading UK turbine manufacturer and a leading UK small Hydro
Consultancy has lead to the completion of a highly successful project . The BRB and CCVSP
technologies are both novel and practical solutions to old problems. Further testing of both
technologies will result in increased confidence which will in turn lead to a competitive
market advantage for UK companies.
vii
viii
1 INTRODUCTION
1.1 Background to the Project
It is recognised that Pelton turbine technology may be used effectively over a relatively wide
range head and flow conditions when compared to other turbine categories and is suitable for
many medium and high head sites. The technology is proven and the overall concept is
reasonably simple.
The purpose of this project has been to research, design and partially test two key components
of a low cost Pelton turbine suitable for smaller UK sites and the export market. These
components are the concentric casing surrounding the turbine and the turbine runner,
specifically investigating bolted runner bucket technology.
From a previous assessment of the Gilkes range of Pelton turbines, two key facts have
emerged from this work:-
• The vertical shaft Pelton turbine configuration offers the most flexibility (and potential
least cost).
• The runner represents 40% of the turbine cost.
This project has concentrated on both of these facts by investigating and developing:-
• The Concentric Case Vertical Shaft Pelton (CCVSP).
• Bolted Runner Bucket (BRB) technology.
Solid modelling
Solid modelling was used extensively to develop the models for finite element analysis and to
produce the machining data for the hard tooling required for wax pattern production.
Design Requirements
Pelton runners are subject to a combination of stresses caused by centrifugal force and cyclic
loads. The centrifugal force is induced by the by the fast rotating body and is related to the
runner speed and mass. The cyclic load is induced each time the bucket meets the jet. On
high-speed multi jet runners, the individual buckets can undergo as much as 5 x 1010
repetitive cycles over its lifetime. Therefore, runner susceptibility to fatigue failure must be a
major consideration in the design process.
Designs where interchangeable buckets are individually attached to the central runner disc are
also desirable but is not widely used for a variety of reasons, but is mainly due to difficulties
in avoiding premature fatigue failure. Several construction methods are currently employed
in the manufacturing industry which provide an alternative to one piece castings.
Recent Developments
Maximum stress in the runner was set to about 20 – 30 MN/m2. For reference, Nechleba gives
30 MN/m2 normal stress, though runner materials are not specified. A value of 45 MN/m2
peak is recommended by Brekke. To use Brekke’s figure the surface defects must be no
greater than 2mm by 1mm and subsurface defects no greater than 2mm by 2mm.
The starting point was the need to attach a ring of buckets to the shaft. Previous work
suggested that buckets could be attached to a runner using two side plates with bolts passing
through wedge shaped extensions to the buckets, roots. Considering the physical aspects it
was obvious that the roots were too narrow (about 0.031m) to carry reasonable sized bolts. It
was decided to use wedge shaped roots, locked in the hub with tapered shoulders. The hub
was modelled by the finite element method. Axisymmetric elements were selected to reduce
the 3D problem to 2D. The ring of buckets were modelled as a solid disc because, at runaway,
it is not expected that the joints between adjacent buckets will open. After many calculations,
the optimum geometry appears to be as illustrated in figure 2.4.
.
Figure 2.4. Section showing selected hub and root geometry, stress due to runaway speed,
2776 rpm, 159 radians / sec
The maximum principal stress in the model is 15.3 MN/m2 This occurs in the centre of the
root and is caused by the centrifugal force trying to bend the shoulders back.
For production, it is intended to use lost wax casting techniques to produce the individual
buckets. Reproducibility of the process is reported to be excellent and rework of the pattern
after a test cast and dimensional check will reduce the deviation from drawing (Reference
ASM Handbook Volume 15 page 248, 1998 [7] ). The prototype bucket was then
dimensionally examined in detail at Gilkes’ factory by a co-ordinate measuring machine and
checked against the drawing. Deviations were added to the pattern solid model to correct the
geometry and a new, updated, pattern produced, this achieved a near net shape casting with
the optimum hydraulic surface geometry and the minimum machining allowance.
Several methods of connecting the hub to the shaft were investigated. Keying, shrinking and
splining were rejected on cost or performance grounds. Clamping was investigated due to
previous good experience with these devices. Eventually a technically acceptable design was
evolved between Gilbert Gilkes and Gordon with Ringfeeder, the manufacturers of shaft
fixing components.
2.6 Testing program
The test programme tested the integrity of the connection between the novel buckets and hub.
It also indicated the extreme operating envelope of runaway speed and locked rotor torque.
From these it was possible to estimate a safe maximum operating head.
The central disk, which represents the assembly of runner buckets, was clamped to a suitable
rigid mounting. Drive was by hydraulic or pneumatic wrench bearing on the 110 mm A/F
hexagon. A 60 mm A/F hexagon was machined on the central shaft - this was intended to
augment the drive to the 110 mm hexagon if required.
Figure 2.5 - Assembly of test fixture, visualised by the solid modelling software
Figure 2.6 - Assembly prior to testing.
The following tests were performed at the testing facilities at Gilkes and Lancaster University.
four jets and the runner at its point of best efficiency. The resonant frequency of the bucket is 2,163
Hz. near the 36th harmonic of the bucket passing frequency it is very unlikely that resonance of the
bucket will be a problem. If the runner speed is 1500 rpm, as designed, the bucket passing
frequency is 50 Hz. for four-jet operation and the resonant frequency of the bucket is near the 43rd
harmonic.
Figure 2.7 - Test 8, Bucket bending test to failure in the 100 tonne universal testing machine
2.7 Cost / Benefit & Conclusions
The development of the BRB during the project has resulted in an innovation that has the
potential to lead to substantial cost savings in runner manufacture. A UK patent has been
applied for to protect the invention. This will be extended to cover each country where a
competitor operates.
A thorough testing program has shown the BRB will reduce manufacturing time and improve
production control without compromising on operating performance.
Manufacturing time
A significant part of finishing a conventional runner is the bucket grinding and
polishing process. BRB technology will dramatically reduce this time by allowing easy
access to all the surfaces to be machined and the new process eliminates all hand
finishing operations. The lead-time to manufacture a runner is reduced considerably. If
the runner is entirely of a new design and solid modelling of the buckets and tooling are
required, a reduction in time of 68% can be expected. If new tooling only is required, a
reduction in lead-time of 79% can be achieved. A replacement runner can be made to an
existing design with an 89% saving in lead-time.
Hydraulic performance
It is envisaged that hydraulic performance will improve due to greater control over the
bucket surface geometry and finish.
Fatigue considerations
Historically, the problem with bolt-on buckets has been cyclic fatigue and corrosion
exacerbating stress concentrations around the bolt leading to premature failure. The
massive number of applied stress cycles, as high as high as 5x1010 over an operating
lifetime, has been found to make fatigue failure the most likely failure mechanism. The
FEM computer program has the capability to predict fatigue performance and reliable
texts on the subject have been consulted, however the true test for the runner will be field
trials.
Model Parameters
A number of different turbine manifold configurations were conceptualised in the earlier stage
of the project that may have been appropriate to the requirements of the project. Five final
options were short listed to be modelled after analysis with the CFD (Computational Fluids
Dynamics) program CFX 5.51.
The basic turbine parameters for the prototype turbine were selected after considering the
casing dimensions from several existing vertical shaft pelton turbines in successful operation.
The selected basic parameters chosen for the ccvsp prototype can be summarised as:-
• All jets 82.6mm Diameter
• Each jet 28.5 l/s (114 l/s total flow divided by 4 jets)
• Internal wall roughness 0.05mm
• Casing internal diameter 1323mm
• Average Velocity in manifold 5m/s
Headloss Comparison
An indication of the hydraulic performance of each option is the hydraulic pressure loss from
the model inlet to the end of the individual jets. The hydraulic headloss was calculated for
each option by measuring the pressure at the inlet and jets using a tool within CFX. Figure 3.2
below show the pressure at each jet and hence the hydraulic headloss.
3.5
3.0
2.5
Headloss (m)
2.0
1.5
1.0
0.5
0.0
0 1 2 3 4 5
Jet Num ber
Options 1 & 5 were selected for further development based on their hydraulic performance
following analysis with CFX 5.51 and their likely ease of manufacture. Construction of the
test rig for physical modelling proceeded using a modelling scale of 0.5:1.0. The construction
was undertaken by York Plastics of Yorkshire and the rig was delivered to Gilkes mid
December 2002.
Figure 3.3 - The test rig set up at Gilbert Gilkes and Gordon’s test bay
The model was supplied with two 3-inch flexible pipe tails for connection to the Gilkes test
system. The four outlet valves were threaded 2 inch BSP. The schematic below illustrates the
test rig configuration.
After connection to the external pipework, the following procedure to examine the flow
distribution and headloss took place for each possible jet configuration:
Headloss results
The testing procedure for option 1 and option 5 was undertaken for every jet flow
combination. Headloss from the inlet to the relative jet was recorded. Any notable flow
occurrence during the test was noted. Table 3.5 below summarises the results.
Average
CFX Miller
Jet Measured
∆h ∆h
∆h
Table 3.5 - Comparison of the calculated and measured head loss through option 1
manifold, all four jets in operation, including CFX Pressure Contour plot.
In this case the values calculated by CFX and the Miller methodology are reasonably close.
The values measured during testing however are consistently lower. This may have been
caused by the proximity of the manifold pressure tapping to jet 1, which according to further
CFD analysis, was subject to severe turbulence, which may have skewed these results. In fact
this pressure tapping was recording a head gain rather than a head loss for some jet
configurations.
Jet CFX Miller Measured
∆h ∆h ∆h
Table 3.6 - Comparison of the calculated and measured head loss through option 5
manifold, all four jets in operation, including CFX Pressure Contour plot.
For option 5, the calculated head losses using CFX agree with the measured headlosses
reasonably well. The exception is jet 3 where the calculated values appear rather high,
though the geometry of the path leading to jet 3 is particularly irregular and difficult to
calculate by theoretical means.
Test observations
Whilst testing the option 5 model, a vortex was observed to form occasionally in jet 2. This
indicates that a low pressure region had developed though the effect on the hydraulic headloss
was not noticeable.
Vortex
Figure 3.8 - showing mixing and consequent losses at jet 1 in the option 1 manifold - jets 1 and 4
open, 0.005 m3/sec.
In the region to the right of figure 3.9 there is no fluid flow; the green dye is in a stagnant
region. It is thought that the region containing the dye is the cause of most of the losses
observed. Figure 3.9 shows the stagnation and mixing from above.
Figure 3.9 - View from above of mixing occurring at the manifold, jet 1 junction, jet 1 only open.
The purpose of the scale modelling was to help predict the hydraulic performance of the
prototype manifold without the expense of full scale construction. The well understood laws of
similarity mechanics were used to transpose the model performance to the prototype. The CFD
software was used to estimate the prototype performance and the empirical hydraulic method
was also included as a reference.
The Euler scaling laws were able to be applied to the CCVSP test rig because of its turbulent
fluid flows and corresponding high Reynolds numbers (>10^6). Equation 3.10 was derived
for use in calculating the prototype casing results, which compares model and prototype unit
flow, unit area and change in pressure.
Figure 3.11
Summary of Headlosses
To summarise, the headlosses in the prototype were calculated in two ways:-
1. Using the similarity mechanics equation 3.10, based on the model headlosses
generated by CFX for scaling comparison.
2. A CFX model of the prototype was run using full-scale dimensions and a wall
roughness of 0.05mm (equivalent to steel).
The results are presented in table 3.12 both option 1 and option 5 prototypes.
Based on Similarity Mechanics CFX results
equation 3.10
An analysis was undertaken based on the layout of an existing 4 jet vertical shaft Pelton
turbine installed by Gilkes in 1986. The existing installation was a very large capacity
however, well beyond the expected operating range of the CCVSP, so the existing
configuration was scaled down so a fair comparison could be made.
The conventional manifolds construction cost, performance and footprint area were then
compared to that of a option 1 type manifold of the same capacity. CFD was used to model
the existing and theoretical cases.
Cost comparison
Prices were sought to manufacture the existing concentric case in the option 1 type design.
The CCVSP was increased in capacity so that velocities in both manifolds were the same.
Figure 3.13 below shows the existing Hatchet Creek manifold layout with the CCVSP (in
blue) superimposed over it for comparison. The footprint area of the new manifold was 4.29m
x 4.03m (17.3m²) compared to 5.1m x 5.1m (26.0m²) for the existing (excluding spear jet
actuators in both cases). A typical powerhouse cost of £110k for a 100m² area was used in
the analysis. Table 3.14 provides a cost comparison.
Figure 3.13
Hydraulic performance
The following operating parameters were used in the hydraulic comparison between the
existing and CCVSP manifolds:
• All jet branch pipes 290mm diameter
• Each jet 330l/s (1.32 m³/s total flow divided by 4 jets)
• Internal wall roughness 0.05mm
• Average velocity in manifold 5 m/s
Both manifolds were modelled using CFX to estimate headloss loss and hydraulic
performance. The results are illustrated in table 3.15
Table 3.15
No onerous turbulence was observed in either model, probably due to the relatively low
manifold velocities. The headloss experienced in the existing manifold is reasonably high in
proportion to the schemes gross head – up to 2.5% of total energy, though this percentage loss
would be less for higher operating heads.
The average reduction in jet headloss is therefore 1.84m (2.8m average headloss for existing
less 0.975m average for CCVSP). To calculate the additional annual energy production gained
from the saving in headloss achieved, an energy value of 5.5p/kW, and a plant load factor of
0.5 was used. This resulted in an additional £819 generation per annum for a typically sized
scheme.
4 Conclusions
Model testing
It would have been more reassuring to have closer correlation between the theoretical and
measured results for both options, however when taken in the context of the operating
pressures of working manifolds (typically 150m+), the headlosses in table 5.3 above represent
less than 1.0% of the total head within the system. Generally 1-3% is considered acceptable.
The CFX results were verified by the Miller calculations and generally by the physical testing.
CFX was established as an accurate tool that can determine manifold hydraulic performance.
CCVSP design
Both prototype manifolds had relatively low headlosses, though option 1 was considered to be
better (lower cost, ease of construction and lower headloss.
Cost / Benefits
The CCVSP design seems to have improved the turbine in three areas.
• A reduction in powerhouse construction costs of 2 to 3% may be possible due to the
reduced footprint area required by the CCVSP.
• Manifold costs may be reduced by 14%
• A modest reduction in headloss may be possible and may improve further if flow
baffles were incorporated into the branch pipe transitions where the major flow
separation occurred. Baffles have successfully used where vortex formation has been
a problem.