lable at ScienceDirect

Renewable Energy 71 (2014) 433e441

Contents lists avai

Renewable Energy

journal homepage: www.elsevier .com/locate/renene

Hydro turbine prototype testing and generation of performancecurves: Fully automated approach

George A. Aggidis*, Audrius �ZidonisLancaster University Renewable Energy Group and Fluid Machinery Group, Engineering Department, Lancaster LA1 4YR, UK

a r t i c l e i n f o

Article history:Received 1 August 2012Accepted 20 May 2014Available online

Keywords:Renewable energyHydropowerTurbinesTurbine testingHill chartsAutomation

* Corresponding author.E-mail address: g.aggidis@lancaster.ac.uk (G.A. Ag

http://dx.doi.org/10.1016/j.renene.2014.05.0430960-1481/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

This paper presents a technology that can accelerate the development of hydro turbines by fully auto-mating the initial testing process of prototype turbine models and automatically converting the acquireddata into efficiency hill charts that allow straight forward comparison of prototypes' performance. Thetesting procedure of both reaction and impulse turbines is illustrated using models of Francis and Peltonturbines respectively. For the development of an appropriate hill chart containing no less than 780 pointsthe average duration of the fully automated test is 4 h while the acquired data files can be processed intodescriptive standard efficiency hill charts within less than a minute. These hill charts can then be used inresearch and development to quickly evaluate and compare the performance of initial turbine prototypedesigns before proceeding to much lengthier and more expensive development stage of the chosendesign.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

In a number of countries around the world including the UK, theconstant increase in fossil fuel energy prices together with a needto improve their energy security by reducing the dependency onimported fuel supplies is boosting the development of renewableenergy technologies [1e3]. The UK has an estimated untappedgreen-field small scale (capped to 10 MW) hydropower capacity of1.5 GWs [4]. This paper describes the automation of a manual tur-bine prototype testing facility (Fig. 1) manufactured by GilbertGilkes & Gordon Ltd to enable very fast data acquisition and pro-cessing into turbine efficiency hill charts.

Ability to quickly assess the performance of a prototype turbineat the initial stage of development is very important before movingtowards more accurate but much more time consuming andexpensive development phase. Depending on the conditions of aparticular application, different types of turbines are used [5] andtherefore different issues are to be addressed. Even though formodelling of reaction turbines Computational Fluid Dynamics hasreached a feasible stage [6], numerical modelling of impulse tur-bines (like Pelton [7] or Turgo [8]) is still a challenge. Whenmodelling a full geometry, time durations of up to 5 days persimulation of one data point are reported [9,10]. Alternatively to

gidis).

reduce the time duration severe simplifications of geometry[11e15] or turbine working principle [16e19] are implied. On theother hand, experimental model tests that use runner dimensionsand flow conditions that allow full scalability and ensure very highaccuracy are usually performed only as the last stage of develop-ment because of its complexity and cost. That is why quickly pre-testing of prototype turbine models might aid the research anddevelopment process overall.

2. Background

The guaranteed efficiency of a turbine (Eq. (1)) as defined bythe International Code for Model Acceptance Tests IEC 60193:1999[20] is the ratio of the mechanical power provided by a shaft of aturbine (Eq. (2)) to the power generator divided by the hydraulicpower (Eq. (3)):

h ¼ PmPh

(1)

where h is the guaranteed efficiency of a turbine, Pm is the me-chanical power [J], Ph is the hydraulic power [J],

Pm ¼ u$T (2)

where u is the rotational speed of a turbine shaft [rad/s], T is thetorque provided by the turbine shaft [Nm],

Fig. 1. Fully automated turbine prototype tester.

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441434

Ph ¼ rQ$gDH (3)

where r is the density of water [kg/m3], Q is the volumetric flowrate [m3/s], g is the gravitational acceleration constant taken as9.81 m/s2, DH is the net pressure head [m].

Calculating the mechanical power provided by the shaft istrivial, however finding the hydraulic power is more complicated asthe net pressure head DH consists of more than one component[21], i.e. the gross pressure head, the head of pressure loss in apenstock and the velocity head. Moreover, the components differwhen calculating the net head for impulse or reaction type tur-bines. Fig. 2 presents general schematics of a hydropower plant.

First of all, it is important to understand how the gross pressurehead is measured in each case. The gross pressure head for reactionturbines is simply the difference between the upstream and thedownstream water levels. However, for impulse turbines the grosspressure head is measured as the distance between the upstreamwater level and the level of a jet impact point which is alwayshigher than the downstream water level. Equations (4) and (5)show how the net pressure head is calculated for impulse and re-action turbines respectively and what terms are important for eachof them:

Impulse turbines DH ¼ HG � HL (4)

Reaction turbines DH ¼ HG � HL � HV (5)

where HG is the gross pressure head [m], HL is the friction loss head[m] and HV is the velocity head [m].

Fig. 2. Schematics of a hydropower station.

Velocity pressure head is used only when calculating the netpressure head on reaction turbines as the downstream waterflow does not affect the performance of the impulse turbines.Physically the velocity head and the friction loss head are notdistances measured from water levels; however they can beconverted into adequate quantities expressed in metres (Eq. (6))and then for the sake of convenience sketched on the diagram asshown in Fig. 2.

HV ¼ v2

2g(6)

where v is the mean velocity of downstream water flow [m/s].

3. Affinity laws

When designing a hydropower station it is possible to calculatethe performance of a known turbine design if the performance ofits model is known. Performance scaling can be done by using theaffinity or so called similarity laws [22e24]. The affinity lawsmathematically relate the same turbine at different speeds orgeometrically similar turbines at the same speed. Equations (7)e(9)show the relationships when the diameter of the runner is keptconstant, whereas Equations (10)e(12) are used when the rota-tional speed is constant:

if D ¼ const:Q1

Q2¼ n1

n2(7)

DH1

DH2¼

�n1n2

�2(8)

P1P2

¼�n1n2

�3

(9)

where n is the rotational speed [rpm],

if n ¼ const:Q1

Q2¼ D1

D2(10)

DH1

DH2¼

�D1

D2

�2

(11)

P1P2

¼�D1

D2

�3(12)

where D is the runner diameter [m].In general, these laws are expressed as:

Qt

Qm¼

ffiffiffiffiffiffiffiffiffiDHt

pffiffiffiffiffiffiffiffiffiffiffiDHm

p $D2t

D2m

(13)

ntnm

¼ffiffiffiffiffiffiffiffiffiDHt

pffiffiffiffiffiffiffiffiffiffiffiDHm

p $Dm

Dt(14)

where indexes t correspond to the industrial turbine and m to thelaboratory model. Therefore by choosing DHt ¼ 1 m and Dt ¼ 1 mand rearranging Equations (13) and (14) to express Qm andnm, the formulae to calculate quantities known as the unitspeed n11 [rpm] and the unit (specific) discharge Q11 [m3/s] arederived:

Fig. 3. Schematics of the turbine tester (impulse turbines).

Fig. 5. Projections of a 3D efficiency curve: turbine efficiency vs n11 (left) and Q11 vs n11(right).

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441 435

n11 ¼ n$DffiffiffiffiffiffiffiDH

p (15)

Q11 ¼ Q

D2$ffiffiffiffiffiffiffiDH

p (16)

Combinations of n11 and Q11 providing constant efficiency h

form iso-efficiency curve. Set of such iso-efficiency curves form ahill chart (example shown in Fig. 15).

For scaling to be possible, care must be taken to ensure that themodel dimensions and test conditions meet the criteria provided inthe IEC 60193:1999 testing standards [20].

4. Methodology

The Gilkes turbine tester can be treated as a small scale model ofa real hydropower station which has all the corresponding featuresas the real power plant would have. Fig. 3 shows a schematic dia-gram of the tester with an impulse turbine mounted on. A cen-trifugal pump powered by a motor provides the flow and thepressure, which can be converted into the pressure head, andtherefore they can be treated as the gross pressure head and theriver flow in the real power stationwhichwas described in previoussection. The tester has a turbine which is coupled to a load tosimulate the load applied by the generator.

Fig. 4. Schematics of the turbine tester (reaction turbines).

4.1. Impulse turbines

To calculate the range of efficiencies of an impulse turbinemodel the quantities shown in the diagram (Fig. 3) are measured.

The readings for both u and T can be taken directly off the shaft[25] of a turbine as it would be in a real power station (the rota-tional speed and the torque supplied to the generator). However, tobe able to use the hydraulic power formula (Eq. (3)), indirectreadings are taken and then converted into Q and DH. The flow ismeasured by using a triangular profile weir [26,27], measuring thehead on the weir, Z:

90�triangular profile weir Q ¼ 8

15Cd

ffiffiffiffiffiffi2g

ptan

�90

�

2

�Z2:5 (17)

where Cd is a coefficient of discharge and equals to 0.6.The conventional way of measuring the net pressure head DH

for impulse turbines and reaction turbines is explained in Section 2Background. Therefore, for impulse turbines mounted on the tester(Fig. 3), the gross head can be calculated from the pressure dropbetween the nozzle inlet pressure and the atmospheric pressure:

HGI ¼pr$g

� ZI (18)

where HGI is the gross head of an impulse turbine [m], ZI is thedistance of a pressure measurement point below the level of jetimpact [m] as shown in the diagram (Fig. 3).

Since p is measured below the point of the jet impact, the headseen by the nozzle is lower and has to be corrected by subtractingZI. For simplicity pipe friction loss head is neglected in this tester asit is expected that the prototype designs would be compared atidentical operating conditions (or performance envelopes) there-fore the losses in the pipeline would cancel out. Then DH ¼ HGI andthe equation for impulse turbines mounted on the tester (Eq. (19))can be derived from Eq. (4):

DHI ¼pr$g

� ZI (19)

Fig. 6. Projections of 3D efficiency curves: turbine efficiency vs n11 (left) and Q11 vs n11(right).

Fig. 7. 3D data covered with a mesh of tetragon elements.

Fig. 8. Stepper motor coupled to the inlet nozzle spear of the Pelton turbine model.

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441436

Temperature fluctuations can induce unwanted errors thereforeit is suggested to monitor the temperature during the testing sothat the density value can be corrected.

4.2. Reaction turbines

A method to calculate the net pressure head on reaction tur-bines is slightly different because of the gross head being measuredas the total distance from the upstream level to the downstreamlevel. Again, the pressure sensor p is lower than the downstreamlevel and has to be accounted for in a similar fashion as with theimpulse turbine. However, this time the level difference betweenthe pressure sensor and the downstream level is combined of twocomponents (Fig. 4) ZR and Z.

ZR is the distance from the level of the pressure sensor to thelevel of a tip of the triangular profile weir, i.e. the distance to thedownstream level when no flow is present. The component ZR is aconstant figure once the pressure sensor is mounted. However,there is a varying component Z, which is varying with the flow.Therefore, the gross head on reaction turbine mounted on thetester as shown in the schematics can be expressed as:

HGR ¼ pr$g

� ðZR þ ZÞ (20)

As explained in Section 2 Background, the velocity head is to besubtracted from the gross head for the reaction turbines. However,for simplicity the velocity head and pipe friction losses are

Table 1List of sensors used in the automation of the turbine tester.

Sensor Manufacturer (model) Total error Variable

Pressure sensor Keller (Series 21 SR/MR) ±0.50% POptical speed sensor Compact Instruments

(VLS/DA1)±0.75% n, ua

Ultrasonic sensor PepperþFuchs (UB500-F42S-I-V15)

±1.00% Z, Qb

Single point load cell Tedea-Huntleigh (1042e3 kg) ±0.05% Tc

a n is acquired in rpm, whereas u has to be in rad/s so that it could be used in Eq.(2). Therefore unit conversion takes place.

b The ultrasonic sensor is measuring the water head Z on the triangular profileweir, which is then converted to the flow Q (Eq. (17)).

c Applied brake torque T on the turbine shaft is calculated by multiplying thebrake arm radius times the load force measured by the Single Point Load Cell.

neglected because of analogous reasons to the explained beforethat these losses would cancel out if the prototype turbine modelsare compared at identical operating conditions. Therefore, bysubstituting the assumptions listed above and the reaction turbinegross head expression (Eq. (20)) into Eq. (5), formula for the netpressure head seen by a reaction turbine mounted on the tester isderived:

DHR ¼ pr$g

� ðZR þ ZÞ (21)

Again, monitoring the temperature throughout the testing tocorrect the density is suggested.

Since all the quantities required for calculating n11, Q11 and h aremeasured or calculated, it is possible to process the acquired datainto a three-dimensional efficiency hill chart.

4.3. Test control

The formulae provided above describe methodology ofmeasuring turbines' performance under various conditions. How-ever, to be able to acquire all the efficiencies in the range of n11 andQ11 available by the turbine prototype model the inlet flow and theload on the turbine have to be controlled. By increasing the load onthe turbine output shaft a rotational speed is decreased. By varyingthe rotational speed, the unit speed n11 is varied proportionally ifDH and D are constant (Eq. (15)). Therefore, by varying load, effi-ciency vs. n11 curve can be acquired. However this is not entirelytrue for reaction turbines as DH is slightly varying because ofvariation in Z. Moreover, in the setting of the tester as described

Fig. 9. Stepper motor coupled to a guide vane control of the Francis turbine model.

Fig. 10. Schematics of the turbine shaft torque applying arm.

Table 2Test plan for the Francis turbine model.

Angular velocity of the stepper motors 5 rpmTotal revolutions on the flow valve 3.5Number of the flow rate samples 15Diameter of the turbine runner 0.08 mTotal revolutions on the load control nut 1.5Number of the load samples 50Extra revolutions on the load control nut (to reach

a complete stop)3

Number of samples per channel 500Delay time before taking readings 5 sBrake (load) arm radius 0.17 mZR �0.175 m

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441 437

above, for a reaction turbine there is a relation of head and load onthe turbine because higher load means more resistance to the flowand provided that the pump is working at constant power, thepressure head increases with the load being applied to the turbineshaft. This is where the affinity laws become extremely useful andallow the control of the test conditions to be simplified and the testduration reduced. Moreover, it corrects any unwanted instabilitiesif all the readings are taken at the same time because the data iscollected in 3D, i.e. ‘x’, 'y' and 'z' axes being n11, Q11 and turbineefficiency respectively. This way a result of variation in the loadwith a flow control position being constant (i.e. angle of guidevanes, nozzle opening distance, etc. is not moved) is a turbine ef-ficiency curve ranging from n11 min to n11 max and being slightlyinclined in a direction of Q11 (Fig. 5).

To be able to acquire more of such efficiency curves, preferablyparallel or close to parallel ones, so that a surface of turbine's per-formance is completely covered and then standard 2D hill chartscan be produced, inlet flow has to be controlled. Q11 is variedproportionally by varying the inlet flow Q, provided that DH isconstant (Eq. (16)). Again, DH is varying with restrictions beingapplied to the flow but this is not a problem as the 3D curve doeskeep all the information of the variations. In general, even if thelines would be not parallel but intersecting, they would be stilllying on the surface of the performance hill, which is the data that isessential. Therefore changing the inlet flow after a range of loads(free spin to a complete halt) has been tested, allows acquisition of

Fig. 11. Francis turbine prototype model.

another efficiency curve that has beenmoved in the direction of Q11.By repeating the acquisition of efficiency curves at different flowrates varying from themaximum flow rate to 0 flow rate a completeset of efficiency curves is acquired (Fig. 6).

When the data is acquired in the form of multiple curves lyingon the surface of a turbine efficiency hill, data processing to stan-dard 2D hill charts takes place. Firstly, 5th order polynomials arefitted on every single efficiency curve [28]. Then data points areevenly spaced by interpolating the efficiency and Q11 within thecurves in equally spaced steps of n11. When every efficiency curve isprocessed to have the same amount of evenly spaced data points,the points are connected to the same index data point of theneighbouring curve. This procedure provides a surface mesh oftetragon elements (Fig. 7). Finally, when the surface is meshed, itcan be ‘sliced’ at chosen ‘heights’ like in an isobaric map by inter-polating the borders of the quadrants that intersect with chosenefficiency plane. The intersect points are then connected by a closedcurve (in some cases open, if the area within that curve is outsidethe tested range of the turbine model) and represents single effi-ciency in the hill chart, i.e. Q11 vs. n11 graph.

5. Implementation

The methodology described in the previous section wasimplemented by fully automating the turbine tester which wasoriginally designed andmanufactured by Gilbert Gilkes and GordonLtd. Both functions: data acquisition and test control is automatedand operated by a Virtual Instrument (V.I.) that is programmed inLabVIEW. A separate V.I. is programmed to process acquired data

Table 3Example of Francis turbine results.

Flow index Load index P [bar] N [rpm] Z [mm] F0 [g]

0 0 2.34 4657 80 680 1 2.37 4678 81 700 2 2.35 4670 81 700 3 2.36 4686 80 730 4 2.34 4660 81 650 5 2.33 4657 81 720 6 2.34 4656 81 69…

0 49 1.35 1742 100 13340 50 1.35 1713 100 13380 51 1.37 0 100 13961 0 2.36 4694 81 571 1 2.35 4677 81 671 2 2.36 4667 82 661 3 2.36 4685 81 681 4 2.37 4682 81 691 5 2.36 4683 81 721 6 2.36 4679 81 72…

Fig. 12. Raw results of the Francis turbine prototype model test in 3D.

Fig. 14. Francis turbine efficiency hill chart with the mesh.

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441438

into graphs and efficiency hill charts shown in Figs. 5e7. Fourelectronic sensors are installed on the turbine tester and connectedto a data acquisition card made by National Instruments (NI USB-6008) to allow instant acquisition of all readings required. Table 1presents all the sensors installed and links them with a variablethat is measured by that sensor.

u ¼ n,2P60

(22)

By using all the sensors provided in Table 1, a single data point isacquired, provided that the testing facility is at a steady state (i.e.turbine shaft brake torque T and flow Q are kept constant). How-ever, as explained inMethodology section, T andQ have to be variedto test the entire performance range of a turbine and construct thecomplete efficiency hill chart.

The control of the flow valves and the torque applying brake armis automated by installing two unipolar stepper motors with aninternal step down gearing ratio of 25:1, providing 1 Nm holdingtorque and 0.3� step angle. The motors are connected to thestandalone PC driver boards controlled by the LabVIEW V.I. via thedata acquisition card. Figs. 8 and 9 present photographs of the flowcontrolling stepper motor coupled to the Pelton and Francis flowvalves respectively.

The schematics of the brake arm used to apply the torque loadon the turbine shaft and measure it is shown in Fig. 10. A rotatingdisc is mounted on the shaft to increase the diameter of the shaft atthe place where the load is applied. The brake arm is causing fric-tion on the surface of the rotating disc. It is done by pushing twoself-lubricating pads mounted on the brake arm to the disc. The

Fig. 13. Processed results of the Francis turbine prototype model test in 3D: turbineefficiency hill surface meshed with tetragonal elements.

brake force is adjusted by tightening the control nut which isdirectly coupled to the stepper motor and hence driven by it. TheSingle Point Load Cell is measuring the load F applied by the brakearm which multiplied by the brake arm radius r gives the torque Tapplied to the turbine shaft:

T ¼ F$r (23)

where F is the force [N] applied to the load cell by the brake arm, r isthe brake arm radius [m].

6. Test results

Results of the automated tests performed on the horizontal axisFrancis and Pelton turbine prototype models are provided in thissection.

6.1. Francis turbine (reaction)

A photograph of the horizontal axis Francis turbine model thatwas tested is shown in Fig. 11. The transparent pipe seen in thepicture is the draft tube. Six guide vanes that control the flow rateand the feed angle to the runner can be seen behind the transparentwall. The guide vanes are controlled by the flow controlling steppermotor as described in Section 5 Implementation.

Table 2 contains the information about the chosen test plan andimportant runner dimensions.

A large volume of numerical data (780 rows containing 6readings) is collected. An example of the format of the acquirednumerical data is shown in Table 3.

Fig. 15. Normalised to 100% 2D efficiency hill chart of the Francis turbine model.

Fig. 16. Pelton turbine prototype model.

Table 5Example of Pelton turbine results.

Flow index Load index p [bar] n [rpm] Z [mm] F0 [g]

0 0 1.71 2438 93 320 1 1.71 2445 95 280 2 1.71 2435 93 300 3 1.71 2444 92 300 4 1.71 2380 92 380 5 1.71 2435 92 330 6 1.70 2431 93 29…

0 48 1.71 469 94 26550 49 1.71 398 93 26830 50 1.72 333 93 26940 51 1.72 0 93 28381 0 1.73 2525 91 391 1 1.74 2536 90 351 2 1.72 2537 93 311 3 1.72 2530 90 421 4 1.73 2526 92 45…

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441 439

These numerical results are automatically processed into graphsby the V.I. programmed in LabVIEW and mentioned in Section 5Implementation. The formulae used for data processing are givenin Sections 2, 3 and 4. The 3D graphs of the raw generic resultsplotted by the V.I. are shown in Fig. 12.

The raw results are automatically meshed as described in Sec-tion 4Methodology. The 3D graph of the processed results is shownin Fig. 13.

The 2D efficiency hill chart is automatically constructed byinterpolating the mesh at chosen efficiencies and then plotting theacquired iso-efficiency curves on the n11eQ11 plane. More detaileddescription of this technique is provided in Section 4 Methodology.Fig. 14 presents the efficiency curves together with the top pro-jection of the mesh.

The absolute best efficiency of the automatically tested Francisprototype model was calculated and the final generic efficiency hillchart normalised to 100% is shown in Fig. 15.

6.2. Pelton turbine (impulse)

A photograph of the horizontal axis Pelton turbine prototypemodel that was tested is shown in Fig. 16. The runner with a total of16 buckets can be seen in the centre through a transparentwall. Theinlet nozzle can be seen in the bottom right corner mounted hor-izontally. The flow control spear is directly coupled to the steppermotor as described in Section 5 Implementation.

Table 4 contains the information about the chosen test plan andimportant rig dimensions.

A large volume of numerical data (936 rows containing 6readings) is collected. An example of the numerical data acquired isshown in Table 5.

Table 4Test plan for the Pelton turbine model.

Angular velocity of the stepper motors 10 rpmTotal revolutions on the flow valve 12Number of the flow rate samples 18Diameter of the turbine runner 0.09 mTotal revolutions on the load control nut 5Number of the load samples 50Extra revolutions on the load control nut (to reach

a complete stop)5

Number of samples per channel 500Delay time before taking readings 5 sBrake (load) arm radius 0.17 mZI �0.175 m

Following the same steps as for the Francis results, the numer-ical Pelton results are automatically processed into the graphicalformat. The 3D graphs of the raw results plotted by the V.I. areshown in Fig. 17.

Again, the surface of the raw results is automatically meshed.The 3D graph of the processed results is shown in Fig. 18.

Same as with the Francis turbine results, the 2D efficiency hillchart is automatically constructed by interpolating the mesh atchosen efficiencies and then plotting the acquired iso-efficiencycurves on the n11eQ11 plane. Fig. 19 presents the efficiency curvestogether with the top projection of the mesh.

The absolute best efficiency of the automatically tested Peltonprototype model was calculated and the final generic efficiency hillchart normalised to 100% is shown in Fig. 20.

6.3. Comparison of the data

The automatically obtained results are compared to the originalresults provided in the operating manual of the Gilkes Tutor GH51967 [29]. However, since the Gilkes turbine tester and the proto-type turbine runners were manufactured in 1960s as an academicteaching tool rather than an industrial turbine testing facility therunners used for the results provided in the manual are not thesame ones as used in this project. Even though it is expected thatthey are of the same design and dimensions the manufacturingquality of the prototype runners is quite poor compared to the realindustrial turbines. Hence, two different runners of the same

Fig. 17. Raw results of the Pelton turbine prototype model test in 3D.

Fig. 18. Processed results of the Pelton turbine prototype model test in 3D: turbineefficiency hill surface meshed with tetragonal elements.

Fig. 19. Pelton turbine efficiency hill chart with the mesh.

Fig. 21. Manufacturing quality of the academic Pelton prototype runner.

Fig. 22. The original normalised hill chart of the Francis turbine model.

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441440

design are expected to have different efficiencies or producedifferent shape iso-efficiency curves. Fig. 21 shows a photograph ofthe Pelton turbine prototype model surface quality to illustrate theproblem.

However, despite the limitations listed above the results can stillbe compared and close resemblance can be seen. Fig. 22 presentsthe original results of the Francis turbine prototype model nor-malised to 100%. The absolute best efficiency of the model testedoriginally was 3% lower than the efficiency of the model used forautomatic testing. It can be seen that the original results match theresults acquired automatically (Fig. 15) very closely in terms of theoperational range and the location of the best efficiency point.

Fig. 20. Normalised to 100% 2D efficiency hill chart of the Pelton turbine prototypemodel.

Fig. 23 presents the original results of the Pelton turbine pro-totype model normalised to 100%. The absolute best efficiency ofthe model tested originally was 10% lower than the efficiency of themodel used for automatic testing. Again, the original results matchthe automatically acquired results (Fig. 20) very closely in terms ofthe operational range and the location of the best efficiency point.Moreover, the automated test has tested the turbine at widerconditions (i.e. the maximum tested unit flow of the original testwas ~0.09 m3/s whereas the maximum tested unit flow of theautomated test was ~0.11 m3/s).

To conclude the comparison of data, the chosen algorithms forautomated control of the testing conditions and the chosen data

Fig. 23. The original normalised hill chart of the Pelton turbine model.

G.A. Aggidis, A. �Zidonis / Renewable Energy 71 (2014) 433e441 441

acquisition technique are proved to be valid. The only possibleproblem might be in the implementation of the turbine outputtorque reading technique as it is based on the mechanical brake.The weakness of the mechanical brake is its instability at low loads.The brake arm starts to vibrate when it is griping the rotating discwith low force and these unwanted vibrations are disturbing thereadings. The reading quality of the output torque is acceptable forthe prototype models used to date, however when testing modelswith accurate and fine surface finishes it is suggested to replace themechanical brake with hydraulic or electromagnetic brakes.

7. Conclusions

The technology for fast fully automated initial testing of hydroturbines was developed and implemented on the originally manualturbine testing facility manufactured by Gilbert Gilkes and GordonLtd. The testing facility was upgraded to a stage where test control,data acquisition and processing are fully automated. The operator isonly required to mount the prototype model to be tested, calibratethe sensors and specify the test plan. After the test is started, thefacility can work for a number of hours and acquire hundreds ofdata points without any external action required. A typical testwould acquire approximately 800 data points in 4 h. Aftercompletion of a test, the data file can be loaded on the processingsoftware and a hill chart characterising the performance of thetested turbine design can be produced within less than one minute.

The implemented technology was illustrated by successfullyperforming tests on both reaction (Francis) and impulse (Pelton)turbine models and processing the acquired data into meaningfulresults of the standard form. Moreover, if the care is taken to ensurethat the model dimensions and the test conditions are in agree-ment with the limits provided in the testing standard [20] the af-finity laws provided in this paper allow scaling andmodelling of theturbine's performance by using the formulae (Eq. (15), Eq. (16)).Scaling is very important for the industry when hydropower plantis designed.

Finally, the described technique for fully automated turbinetesting can be very useful in further developing hydropowertechnologies. It can reduce the duration of initial research anddevelopment phase drastically by enabling quick testing of newprototype designs.

Acknowledgements

The authors would like to thank the Lancaster UniversityRenewable Energy Group and Fluid Machinery Group, Dr. AntoniosTourlidakis, Andrew Gavriluk, Barry Noble and Ian Nickson for theircomments, suggestions and technical support. The authors grate-fully acknowledge the contribution of the turbine manufacturingcompany Gilbert Gilkes and Gordon Ltd. which supplied themanual turbine testing facility.

References

[1] European Renewable Energy Council (EREC). Renewable energy in Europe:markets, trends and technologies. 2nd ed. London: Earthscan; 2010.

[2] European Commission. Promotion of the use of energy from renewablesources [Internet] [Cited 2014 April 18]. Available from:, http://europa.eu/legislation_summaries/energy/renewable_energy/en0009_en.htm; 2009.

[3] Aalto P, Temel DK. European energy security: natural gas and the integrationprocess. J Common Mark Stud; 2014.

[4] British Hydropower Association (BHA). Hydropower: opportunities, chal-lenges & sharing best practice [Internet]. The British Hydropower Associationperspective; 2010 [Cited 2014 April 18]. Available from:, http://www.ciwem.org/media/142437/The%20British%20Hydropower%20Association%20Perspective.pdf.

[5] Lancaster University Renewable Energy Group (LUREG). Hydro resourceevaluation tool: engineering options [Internet]. [Cited 2014 April 18]. Avail-able from: http://www.engineering.lancs.ac.uk/lureg/nwhrm/engineering/index.php?#tab.

[6] Keck H, Sick M. Thirty years of numerical flow simulation in hydraulic tur-bomachines. Acta Mech 2008;201:211e29.

[7] Pelton LA. 1880, US Patent 223,692, October 26, 1880.[8] Crewdson E. 1920, UK Patent 155, 175, May 10, 1920.[9] Jo�st D, Lipej A, Me�znar P. Numerical prediction of efficiency, cavitation and

unsteady phenomena in water turbines. In: ASME 2008 9th biennial confer-ence on engineering systems design and analysis; 2008. pp. 157e66.

[10] Jo�st D, Me�znar P, Lipej A. Numerical prediction of Pelton turbine efficiency. In:IOP conference series: earth and environmental sciencevol. 12(1); 2010.p. 012080.

[11] Perrig A. Hydrodynamics of the free surface flow in pelton turbine buckets[Ph.D. Thesis, EPFL_TH3715]; 2007.

[12] Koukouvinis PK, Anagnostopoulos JS, Papantonis DE. SPH method used forflow predictions at a Turgo impulse turbine: comparison with Fluent. WorldAcad Sci Eng Tech 2011;79:659e66.

[13] Anagnostopoulos JS, Koukouvinis PK, Stamatelos FG, Papantonis DE. Optimaldesign and experimental validation of a Turgo model Hydro turbine. In: ASME2012 11th biennial conference on engineering systems design and analysis;2012. pp. 157e66.

[14] Barstad LF. CFD analysis of a pelton turbine. MSc Thesis from the NorwegianUniversity of Science and Technology; 2012.

[15] �Zidonis A, Panagiotopoulos A, Aggidis GA, Anagnostopoulos JS, Papantonis DE.Parametric Optimisation of two pelton turbine runner designs usingcomputational fluid dynamics. J Hydrodyn Ser B; 2014 [in press].

[16] Zopp�e B, Pellone C, MaItre T, Leroy P. Flow analysis inside a Pelton turbinebucket. J Turbomach 2006;128(3):500e11.

[17] Klemensten LA. An experimental and numerical study of the free surfacePelton bucket flow [MSc Thesis]. Norwegian University of Science and Tech-nology; 2010.

[18] Correa JLC, De Andrade J, Asuaje M. A preliminary analysis of a turgo typeturbine CFD simulation designed with an integrated dimensional methodol-ogy. In: ASME 2012 fluids engineering division summer meeting collocatedwith the ASME 2012 heat transfer summer conference and the ASME 201210th international conference on nanochannels, microchannels, and mini-channels; 2012. pp. 327e37.

[19] Correa JLC, De Andrade J, Noguera R, Croquer S, Jeanty F, Asuaje M. DesignProcedure for a Turgo type turbine using a three-dimensional potential flow.In: ASME turbo expo 2012: turbine technical conference and exposition;2012. pp. 2039e52.

[20] International Electrotechnical Commission (IEC). 60193. Hydraulic turbines,storage pumps and pump-turbines: model acceptance tests. 1999; 1999.

[21] European Small Hydropower Association (ESHA). Guide on how to develop asmall hydropower plant; 2004.

[22] Munson BR, Young DF, Okiishi TH. Fundamentals of fluid mechanics. 5th ed.John Willey & Sons, Inc; 2005.

[23] Mosonyi E. Water power development, volume one: low-head power plants.3rd Eng. ed. Budapest: Akad�emiai Kiad�o; 1987.

[24] Nechleba M. Hydraulic turbines: their design and equipment. Prague: Artia;1957.

[25] Doebelin EO. Measurement systems: application and design. 5th ed. Colum-bus: The Ohio State University, US; 1990.

[26] Boiten W. Flow-measuring structures. Wageningen: Wageningen AgriculturalUniversity; 1993.

[27] Cusick CF. Sewage and industrial wastes, volume 29: open channel flowmeasurement. Water Environment Federation; 1957.

[28] George GP. Numerical methods of curve fitting. Cambridge U.P.; 1961.[29] Gilbert Gilkes and Gordon Ltd.. Gilkes Tutor GH5: operating and maintenance.

Kendal: Gilkes; 1967.