PDF Viewing Options

Not Your Device? Search For Manuals or Datasheets below:

File Info : application/pdf, 120 Pages, 6.59MB

Document DEVICE REPORTPu350-Pu500 Windmills Technical Report
A project of Volunteers in Asia
Fchnical Report 1982 by Niek Vande Ven
Published by: Stichting TOOL
Entrepotdok 68A/69A 1018 AD Amsterdam
Available from:
same as above
Reproduced by permission.
Reproductionof this microfiche documentin any form is subject to the samerestrictions as those of the original document.

\ -1`

dealing with the
over the period


The Netherlands assistance in the windmill projects in Ghazipur and Allahabad officially

stopped at the end of March 1981. It was recognized by the project management that the experiences of the experts in these projects should be recorded.

committee -- _

Finally it was agreed that the financing of this activity took place through the Steering

Committee Windenergy Developing countries (SWD). This is a government financed organi-

sation promoting the interest for wind energy in developing countries and aims at helping

governments, institutions and private parties in the Third World with their efforts to utilize

wind energy.

The author is very gmteful to all parties concerned that the experiences could be materialized

in this publication.

It is hoped that it will contribute to the extention of development in

many countries


TOOL is a Dutch foundation development and application


in the global process of renewable

of socially appropriate technologies.

The broad obiec tive is to promote greater freedom for groups which are deprived

of fuil opportunities for loco!, self-progmmmed

and self-sustained development.

The strategy is to provide and support information resource links among the


users and generators of appropriate technology for development.

The operations of TOOL are designed

- technical advice and support

- research and development

- publications

- documentation systems and services

- appl icstion projects

- education and training

- organisational


to match

need and resource:

WOT is a non-profit organisation at the Twente University of Technology and

gives technical advice in the field of wind energy, solar energy and water

supply. The broad objective is to improve the position of the weaker sections

in society and the advice should be appropriate to the local situations and

circumstances. To support the technical advice the WOT has a testing field

where various designs are being developed and tested, The WOT is mainly a

voluntary organisation and consists mainly of students of the university. Several

staffmembers take care for the administrative

and technical support.




Short II&tory.


The design criteria and first developments.


The construction of the first prototype



The first experiments in the Netherlands



Technical developments of the windmills.


The first experiences with the 5 m dia





The prototypes under development


2.2.1 The tower constrcution


2.2.2 The head construction


2.2.3 The moving parts: Crank mechanism,

bearings etc.


2.2.4 The rotor


2.2.5 The piston pump


2.2.6 The pump suspension



The 4 PU 350 windmill for low water




Test - Rig for 12 PU 500 windmill.


Demanded features of the Test - Rig


3.1.1 Drive of the Test - Rig


3.1.2 Self-moving head construction


3.193 Simulation of the elevation head


3.1.4 Instruments


3.1.5 Way of testing


3e2 Pumps to be tested


3.2.1 The 4" S 500


3.2.2 The 5" SS 500


3.2.3 !lhe 6" PVC 500


3.2.4 The 6" S 500






Volumetric efficiencies



Conclusions with regard to life-time

features of the test pumps



Properties and characteristics

of the

12 PU 500.


Input-ouput relations, volumetric and




The power characteristics

of the windmill



Analysis of the windmill-pistonpump




Defining optimal starting windspeed,

windmill- and pump-dimensions.


Deriving the optimal starting windspeed



Deriving required rotor dia



Pump dimensions



Numerical eTample


Short History

In 1975 the Organization of the Rural POOR (ORP) in Ghazipur

(India, U.P.) wanted to support small and marginal fargers

in growing an extra crop during summer season by means of


As rather high wind speeds occur during summers

the idea was born to use wind energy as the power source

for pumping the irrigation water from the wells.

In cooperation with the TOOl-foundation in the Netherlands

some feasibility

studies have been carried out and eventu-

ally a proposal for a pilot windmill project has been gran-

ted bothby the Government of mdia and the Netherlands.

The latter sponsored the pilot'project

which started in

August of 1977.

The puoject was extended year by year while the 12 PU 500

Windmill design improved step by step. Expansion of the

project occurred in 1979 when the newly built ORP-TOOL-

Windmill workshop was inaugurated and the Windmill Research

and Developments activities were taken up at the Allahabad

Polytechnic Allahabad (APA). In the course of 1980 another

Windmill project started in Trichy (South India) where the

Workshop for Rehabilitation

and Training of Handicapped

(WORTH) also adopted the same windmills for irrigation as

one of its products.(The latter has been financed by HIVOS.)

In January of 1981 the Allahabad Polytechnic Allahabad started the "National Windmill Demonstration Project" which was initiated and sponsored by the Department of Science and Technology (DST) of the Government of India. During the first three months thirty 12 PU 500 windmills have been succesfully distributed all over India attended with trainingcourses for mechanics and engineers,after which the TOOL-project ran to its end.

-I ,

The following

year three hundred 12 PU 500 windmills have been :,, ,a;

_"_ - b

:' ";<iL.i,:.-:

built by the Allahabad Polytechnic by order of the Department of Scienceand Technology while plans exist regarding a windmill production of several thousands during the next year.


I. The design criteria and first developments.

Previous to the introduction of the new waterlifting


vice sufficient knowledge concerning its feasibility


acceptance should be gathered in order to obtain relevant

input for a proper set up of a pilot project.

Therefore short term missions were undertaken to Ghazipur

to gather information about materials, equipments and ex-

pertise. Also information on winddata of the eastern

part of Uttar Pradesh gave important information in order

to calculate the size of the windmill which had to be de-


Since the windspeeds in the Ghazipur district are rather low, rough calculations pointed out that even in the windy season a windmill applied for irrigation of a command area of two and a half acres (I ha) needs a rotor diameter of at least 7 metres. This calculation was based on the assumption that the average water requirement of the irrigated plots amounts to 2 7 mm a day, the total pumping head increases to I4 metres (dry season) and that the windmill starts pumping with a total efficiency of 0.10 (if the windspeed exceeds 2.5 m/s.).

Other calculations also led to smaller windwheel diametres, very high starting windspeeds and extremely huge storage tanks on the one hand and a large rotor diameter, moderate tankcapacities and low starting windspeeds on the other. An optimum must exist depending on material- and labourcosts and the loss of agricultural area in order to realize the needed storage tank. All these matters were taken into account to approach this unknown ideal of windmill size and storage capacity, but still a lot of uncertain aspects and doubts remained.

Moreover in 1977 the availability

of expertise and successful

designs of cheap, reliable and appropriate windmills were

rather limited in spite of the fact that many efforts had been undertaken already to develop windmills in other developing countries. Such experiences always showed that the introduction of the windmill is easily underrated. First the technology has to be developed and implemented successfully under the local circumstances after which a process of transfer of technology should take place.

Due to the above mentioned lack of sufficient experience,

it was decided to design and develop first a proto-type

windmill in the Netherlands which should be compl.etely

constructed with the materials, expertise and implements

which are locally available in Ghazipur. The total material-

costs should amount to Dfl. IGOOm-- (Rs 3000,-- up to Rs 4000.--


Materials like gas-pipe, angle-irons, flat- and sheet-irons,


bolts & nuts etc. would be available in plenty

there according to the information obtained.

The local blacksmiths in India are able to construct simple

implements from these, like wheat-threshers and many other

agricultural equipment.Moreover the electrification

of seve-

rat. villages sparksoff more or less sufficiently equipped black-

smiths and small industries. So welding and simple machining,

like drilling- and lathe-work, are rather current.

In comparison with steel (!977: angle iron RS 2.20/I@) wood is very zostly and later on it proved not tc be available in required size and quality. For example wood for the construction material for building a tower-structure would not be attractive as far as the lifetime is concerned. Moreover the Indian Government stimulates application of other construction-material rather than wood in order to prevent the severe erosion of the country due to deforestation.

Such considerations anl restrictions led to new ideas and alternative designs of some essential parts of the windmill. In September 1977 the construction of the first proto-type windmill was carried out at the testing field of the Working

1.1 The construction of the first proto-type I
This first proto-type windmill was fitted with a 4m dia-

meter I6 bladed rotor which was left after an education

program of the WOT in which it was constructed. It seemed

to be a fruitful

opportunity to integrate it in the design

which had to be developed at that time, but afterwards it

turned out that this rotor was too weak, too complicated

and too expensive. Moreover according to the estimation a

4 a. diameter-windmill

is rather small to serve the purpose.

So after some trouble-shooting

during the months of September-

November, the experience with the very first proto-type

caused a complete re-designing. This resulted in a set of

drawings of the tower construction, head construction, moving

parts and the windwheel of a new 5 a. dia. 12-bladed proto-

type. Besides also a small building manual of the new proto-

ty-pe has been edited,but actually the total set was not yet

completed. Still the design and the drawings of the most im-

portant and essential part of this wind-powered water-lifting

device, the pump, failed. In the next chapters extra atten-

tion will be paid to the pump designs which followed.

-Basic designs

The expected rotor efficiency of the newly designed windmill

amounts 0.38 theoretically

if the windmill is loaded at a tip-

speed ratic( XS~~ : tipspeed divided by the windvelocity) of

two. Via a simple crank .-connecting rod-crosshead mechanism the

rotation of the millshaft is converted into an up- and downmovement -* ,:.,;

of the pumprod which drives the single-acting-reciprocating




The tail of the headconstruction carries a windvane which turns ,"' ::A

the headconstruction armand the towerpipe in order to keep the windwheel perpendicular to the wind.

' ,":i.:..t_I2)

In case cf severe.storm a simple security-mechanism will automat ically the headconstruction from the tailstructure

dislo-ck after-,

-:''%, .":,'-'I `- "I',),...'I.


T -I--

- -.

Ir-- ---- - .-..- _ - --- -7
--._ - - _.

I *\ .
"\, :., `\`..

._-- -


- uw


which the head (and so the windwheel) turns 90' degrees out of the wind. In drawing Nz 1 the windmill has been drawn in that

secured position. The rotor-, head- and towerconstruction

have been composed

of parts of angle-iron of 1.5" (except-the four towerlegs: 2")

and flat-iron,

1.25" which are jointed by means of electric

weldings. The application of bolt and nut connections is limited

just to the mounting of sheet-iron parts and some connections

like hinges, pumprod-connection etc.

The millshaft (50 mm.) rotates in dustproof self-adjusting

bearings which are considerably oversized according to the


crankpin- and crosspin-bearings are

preferably furnished with nylon bearing-bushes,but

bronze bearings

can also serve the purpose if lubrication is applied from time to

time. Also the crosshead which is composed of two specially shaped

sheets needs the lubrication (see next chapters).

Characteristic of the headconstruction is the surrounding of the

tower-pipe by its framework of angle-iron which acts as a bearing.

In radial sense some tolerance is permitted,and

is even necessary

due to the unroundness of the towerpipe.

Positive experience has been achieved already by applying this

remarkable joint of the head and tower-construction

in. a

wooden sail-wing windmill which has been designed and constructed

in May 1976. Axial forces can be carried-over directly to the

top-edge of the towerpipe,so the radius of coulomb's friction

acts very near to the centre.

!Che total structure has been designed for maximum windspeeds of

at least 20 m/set and even the blade tips (free length of appr.

60 cm.) can in theory stand the forces which should act on it if the windwheel runs free of load under static conditions at

a windspeed of 65 m/set.

It is expected that the rotor is not the constraint factor of

strength of the total structure, However dynamic forces (e.g. due to rotor unbalance) probably cause vibration in the natural frequency of several parts which may even lead to fatigue cracks!

For example the natural frequency of the headconstruction situated

around the towerpipe comes to 4.35 rad/sec (theoretically) which means that during operation of the windmill at a R.P.M.

1 .I0

number of approx. 45, resonance may occur. Also the slender parts like the long ang7e-irons of the tower- and tail-construction are 1-rC?i? to vibrate within a scale of R.P.M.-numbers of the windmill if unbalances exist in the rotor construction. The single-acting piston pump applied in combination with the first prototype windmill consisted of a 6" p.v.c. pipe in which a wooden footvalve and piston with leather washers and sealing were acting. This kind of pump, which also was applied in combination with the former mentioned wooden sailwing windmill, :eemed to be very reliable as far as resistance against wearing and tearing of the Fealings and washers were concerned.

1.2. The first experiments in the Netherlands.

Due to shortage of time the prototype windmill was no-t yet: kept up

2,.I date according to the new designs, so the first experiments

have been undertaken with a rotordesign which was not represen-

tative i?f -the prototype which would be introduced in India.

Moreover the elevation head in Tndla was expected to be considerably

more than the 4.0 meter head which appears in the Netherlands.

Ir! unite of that the first experience turned out already that

a prcper fixing of the pump in the well zs well as jointingthe

sF:codenpumprods are rather difficult,

especially if highR.P.M.-

numbers occurred. The entire pump-construction jumped in its

suspension and the pump-rod joints got loose caused by the

redp-ocating forces after some time during operation. The cross-

head, constructed of two shaped sheets,started wearing out and

makizig noise. Neither did the connection of the crank to the mill

sl-:a.fT;turn out to be reliable. Although the security mechanism

functioned rather well, its critical adjustment did not remain

and demanded re-tighteningof

the adjusting bolt from time to


Especially <ts required and accurate hinges cause -that this

sensitive mechanism may never be successful, so other possi-

bilities had to be considered and developed in order to obtain

a simple and reliable security device.

A rather remarkable aspect has been discovered in those days. Depending or! the R.P.E. number the _a_m_.ou.-nt of water,delivered.


per stroke of,the piston pump exceeded the stroke-volume in a

percentage of more than 30 $. Later onthis effect was recognised

as an inertia-effect:

the kinetic energy of the accelerated water

column is convertedinto an extra delivery. hren after introduc-

tion of a small but permitted leakage in the piston sealing,

which is done in order to decrease the starting torque of the

pistonpump, this inertia effect still persisted. Also some experi-

ments with sealingless pistons has been carried out but

unfortunately the required time for a proper investigation

failed and the TOOL- windmill experts travelled to India at the

end of November 1977.

2.1. The first experiences with the 5 m. dia. windmills.

As mentioned above the experiences with the first prototype

windmill in the Netherlands were rather poor and still a new

and unknown security-mechanism and a complete new and untested

rotor was to be applied. In spite of that the realization of

the first windmill at the Technical School in Ghazipur progressed

smoothly. Also a quick prefabrication

of the next two prototypes

in the R.T.1. workshop, lateron,has been achieved by using

welding- drilling- and shaping-moulds. Small changes in the

original designs were caused by slight differences in quality

and size of the materials which are available locally.

Transport of the entire windmills to the site proved to be

rather uncomplicated. A truck or tractor with lorry can easily

contain all the separate parts (tower,head, windwheel) of a

complete windmill. Local people are always prepared to carry the

unloaded parts, lying on the roadside, to the site where the

windmill is going to be erected.

In general the windmill is built above an existing well, from

which water was drawn previously for irrigation purposes by

using bullocks as an energy-source. Since the maximum pumping capacity of the windmill amounts to appr.30 m3/h the water-

supply to these wells is often insufficient.

In order to increase this capacity a tubewell has to be applied

at the bottom of the well. The length of these tubewells

varies between 12 to 15 meters depending on the depth of the


The erection of the tower is achieved simply by pulling and pushing

it upwards by means of offered manpower of the local people.

After the tower has been positioned(exactly

in the centre above '

the tubewell)the towerlegs are poured in with concrete in the 40 cm deep holes whichhavebeenmade in advance.


After a five-days period the concrete is sufficiently



out to stand the extra load acting on it during lifting and

positioning of the headconstruction arid windwheel.Next page


&Technical developments of the windmi1'1s

In the beginning of December 1977 the windmill-project


in Ghazipur. Although a workshop was not at the disposal of the

counterpart organization: The Organization of the Rural Poor

(O.R.P.), the facilities

seemed to be sufficient in Ghazipur-


Several blacksmiths and institutes proved to be interested in

the windmill-technology

(there was no windmill-history

in the

ratherremote Ghazipur-area) and it was decided to construct the

first windmill in cooperation with theteachers and students of

the local Technical School.

On Friday the 20th of December, the first windmill (KK I) was

erected near the Kusumih Kalan village beside, the national-

highway Ghazipur-Benares and it was baptised by local villagers

and authorities. During the month of January two other windmills

have been prefabricated in the workshop of the Rural Trainings

Institute (R.T.I.) where agricultural

implements like plows

and threshers are manufactured by trainees and Staff members,

In the course of 1979 a newly built workshop, situated at

the ORP-campus, came at the disposal of the windmill project.

A complete set up of windmill production,maintenance

and trai-

ning facilities has taken place in this workshop. At the same

time all research- and development activities were taken up

at the Allahabad Polytechnic Allahabad (APA) where already a

modest windmill production was sparking off as well. During

the period of November 1979 upto March 1981 an intensive pro-

gram of pumpdesign, pumptesting, redesigning of windmillparts

and training was carried out over there.

shows how these can be carried out quickly a.nd safely by means of a simple lifting device.
The needed storage tan..k is ~~on:;tr*ucted from mud-made walls of about 1 mett'e high, which are partly bricklined and plastered in places where erosion or seapace is expected. The content of a storage-tank amounts 50 upto 150 m3, depending on the expected need and the available area which cansuitably be occupied by the tank. An overflow pipe is masoned into the wall in order to prevent that overflowing water will damage the mudwalls (if the tank is full). Since the local people often utilize the fresh watercdelivered by the windmill)for domestic purposes, like drinking, bathing and doing the wash,the mudwalls in the environment of the delivery pipe definitely will be damaged. Therefore a simple washing ac:Gi!lodafion is provided by means of some masonry.

me watercan bereleasedfromthe

tank by the farmer(s), using

a 2.5" hose, which acts as a siphon. Since this device is easy

to handle the water can be tapped at any place from the tank

and can be led to the plot(s), via one of the several irrigation

canals. The farmers have to be well-informed in handling the

windmills. During storms the windmills stop automatically,


afterwards.the farmers have to climb on their windmills in order

to reset the windmill into the wind. At the same time it is an

opportunity to inspect and lubricate the moving parts.

However,soon after the introduction of the first windmills

several breakdowns and long windless periods proved to be a

draw back for this particular technology. In spite of this the

farmers involved remained rather enthousiastic and local agen-

ties and Institutions

developed interest as well.

2.2 The prototypes under development.
During field-tests some parts of the windmill components needed redesigning in order to achieve a higher reliabiliiy cn the one hand and a costprice reduction on the other.

Although this seems to be more or less contradictory it is proved that both demands mentioned can partly be conceded by surging for th e most simple solutions and alternatives which can be realized with the local means and materials. Therefore it is a must to know the local market and the means of production to which the designs have to be adapted.

In practise still unknown factors like the extent of wear,

corrosion , vibration etcetera will remain and just a tho-

rough fieldtesting

will provide the answers in the long,run

and broaden the experience. This disadvantage of such re-

search is that failures can cause the distrust of the

local people in the windmill technology. To avoid this it

was decided to carry out the research and development as

much as possible under "laboratory" conditions. For this

reason a testing machine has been designed and built (see chapter 3) in order to speeden up the developement of the windmill.
These experiments as well as the field experiences led to the stepwise evolution of several windmill components which is described in the following paragraphs.

2.2.1. The towerconstruction.

The fabrication of the tower is very simple. The two pre-

fabricated halves are composed: the towerpipe is pushed in,

positioned and welded. In three hours the tower is completed with

the help of three persons.(kcluding

the cutting and

straightening of the materials).

The first and second platform are rather safe and comfortable to

stand on but actually the first platform is superfluous:

lubrication and repairs will always becarriedout standing on

the top platform.

Climbing on the tower is not done by means of a ladder: by

coincidence the diagonals of the lower tower section just

provide an adult the needed facilities

to reach the first

platform. For esthetic reasons the base-measurement of the

tower has been reduced from 2.50 to 2.00 meters and still

its foundation can surround the opening of the well which is

seldom more than 1.8 meters in diameter.

Previously the total material costs of the tower amounted to 2

l/3 of the total material costs of the windmill, so a reduction

in tower weight has been undertaken. Applying 1.5" instead of

2" angle-iron for the four towerlegs decreases the weight

considerably but the legs became too slender which means that

extra compartments in the structure must be applied in order

to achieve sufficient strenght in the entire construction.

Also some minor modifications, like guiding the towerpipe by

flats inst,,d of angles and cancelling the first platform de-

creased the weight and costs. It was also proposed to reduce the

length of the towerpipe,but extra angles and flats will be

necessary to fix +t properly and to provide sufficient rigidity

to the tower-top structure.

However the original simplicity of the construction will turn

over in an extra complicated one appearing during fabrication

and defenitely resulting in loss of' time due to mistakes.

Due to these above mentioned modifications the towerweight has

been reduced from appr. 225 to 125 kg.Also another advantage is

that the tower is easier to handle during transport and instal-

lation while the strength is almost the same. Climbing the

tower is facilitated by applying some welded flats in the





l U
2.2.2. The headconstruction.
Some minor parts in the framework of the headconstruction have been changed. The most important and valuable change
was reducticr of the total length of the tail. This has been done in order to obtain a better balanced structure; the length amounts to 0.8 D and a better*steering resulted. Since the bearinghouses, which are available in India,are rather wide some extra space had to be created for the support of the rear-bearing. The flange of this support (angle-iron 2" x 2") is pointing to the front of the headconstruction in order to achieve that. Worthmentioning is that using a simple fixture for welding the head saves time and moreover accurate work results.

The securitynechanism

In the very first desias the required accukate hinges of the

security-mechanism caused lots of problems concerning its

adjustment which did not remain constant.Initially

a new

design has been introduced in all the Indian prototypes which

proved to be very simple and reliable. A helpvane,situated behind

the rotor and besides the headconstruction,is

pushed backwards

by the wind Lf the adjusted preload of a spring is exceeded.

Due to that the achieved displacement unlocks the headconstruc-

tion from the tailstructure

by opening a simple lockmechanism.

The moment, caused by the windpressure acting on the helpvane,

initiates a small rotation of the head (and so the windwheel)

with regard to the tail and winddirection.

'1%~ windpressure acting on the windwheel provides a further :-otation of the headconstruction until the windwheel is turned godegrees out of the wind and a lock-system finally fixates the head to the tail.


!This lo-ck--

has to be appxd

in order

to prevent

that during

storms ,accompanied by sudden changes of wind direction, the

tail and head will move separately which can result in very

high impacts in case of concussion of both parts. During this

securing motion a toothed handle prevents such separate

movements and it will not allow the headconstruction to turn

back with regard to the tailconstruction.

Since there is almost no damping in the entire security system

the kinetic energy, stored by the wind into the rotating

* head, tail-con .struction properly balanced

old desi

gyroscopic moment


headconstruction, has to be dissipated. The impact taken by the final lock, proved to be considerable at the end of such movement of securing. Although there is the Coulomb's friction, its dissipation capacity is too less and moreover its friction is eliminated more or less by the gyroscopis moment which appears during this security movement. So even this damping can be neglected. In order to use this gyroscopic effect an extra generated Coulomb's friction could be added in positive sense by changing e.g. the direction of rotation of the windwheel. Since the windwheel runs clockwise in connection with the bolted crank pin, it is preferable to maintain this,but inverting the direction of turning cf the headconstruction with regard to the tail will also lead to the same achievement. I?owadays the headconstruction turns anti-clockwise out of the wind (topview!), in order to obtain the extra needed friction for damping caused by the gyroscopic moment. The impact is reduced considerably and hardly audib$e.

.. "

.( ..,.;.:



. .



. ..-

Although the w-indwheel is kept out of the wincl by the ntcering-

function of the tail, it will still be hit by the gusts during

storms. This is caused by the inertia of the total secured

structure: during stormy weathers it was observed that the rotor

changes its direction of rotation continuously :clockwise-

anticlockwise, etc, which means that it is also attacked by

these gusts from the back. This resulted several times in a

forward bending of the blade-tips because its rigidity is low

in forward direction.

In order to prevent such damage it was proposed to increase the

rigidity by hammering a profile in the blades_ in the environment

of the outside-ring blade-support. However, in this way the

airfoil will be spoiled and so the efficiency will decrease.

Another possibility is keeping the windwheelslightlyinto


wind in such a way that it will Just not be attacked from the back.

This is achieved if the windwheel is turned out of the wind less

than 90 degrees: the rotor maintains clockwise runningtoo.

Nowadays the final lock of the security mechanism is situated in

such way that the -windwheel turns appr. 75 degrees out of the

wind and forward bendings of the blade tips do not occur any


Though this automatic security device never failed the required reset procedure proved not to be practicalin places where wind speeds fluctuate too much. Therefore efforts have been undertaken to develop systems in orderto tackle this reset problem.

For instance a rope connected with the reset lever and

led via the tail end to ground levelfacilitates

a reset pos-

sibility which excludes the need of cl<mbing the tower. How-

ever this system still demands the presence of an operator

so it is not a great improvement.

As a matter of fact several fully automatic security devices

are common in commercial windmills which principle is based

on the momentum equilibrium of head and inclined hinged tail.

However such mechanism demands a frictionless

rotation of

the head over the towertop which does not apply for the

simple windmill design described here.

'kerefore it was tried to derive a fully automatic security device from the existing mechanism which finally led to the following designs: a. ratchet system b. stepless system
Both new systems have in common that the helpvane releases the fixation between head and tail in order to allow the head to turn backwards out of the wind. Y5i.s process continues till the helpvane does not detect any exceeding windpressure an,ymore with the result that fixation between head and tail is restored in that particular position. This fixation is achieved by means of a ratchet- and a stepless friction mechanism respectively in these systems.

Minor changes in wind direction cause minor oscillations of the tailvane which are most essential in order to return the head and rotor more or less facing the wind again if the storm decreases. This return motion is possible due to the "one way" character of the ratchet and stepless friction me-

- 2.13


chanism as long as the helpvane does not detect any overpressure in its newly found position.

So both the systems allow the head and rotor to turn out of

the wind to a certain extent depending on the actual wind-

forces. Theoscillatingtail

vane will see for the automatic

However for the simple ratchet system a minimum oscil-

lation of the tail vane is required during the automatic

reset procedure in order to reach for the next tooth of

the ratchet. In case of rapidly fluctuating wind directions

impacts on the teeth will occur which may lead to wear of

the ratchet and its counterpart in the long run. Also an

impact occurs at the end of the securing motion if suddenly

the helpvane puts the ratchet in action again.Observations

made during field tests just partly confirmed these disad-

vantages and presently the ratchet devices are still obser-

The "stepless" system, as the word suggests, responds to

very minor oscillations in a less hesitating manner than

the VatchetVg does. The pivoted catch induces sufficient

high friction between the disk (jointed to tailbeam) and

contact surface (jointed to head) resulting in a tight grip, In this system also the sudden impact exists at the end of


the security motion. !&is disadvantge has been eliminated


by allowing slip in between the disk and tail beam in order


to dissipate severe shocks-f



-.2 :

,4.",A;7$3Ij$-jti24,: i<$


."" ,, _._,I&,...\:3"5:a:a".i!_,4(:";~<<I':a.,$$s",I~*j<

2.2.3 The moving parts: Crank-mechanism, bearings, etc.
Although the principle of converting the millshaft rotation into the up-and down-movement of the pumprod was still maintained, the bearings as well as the simple crank-mechanism have been changed a lot. Many thousands of revolutions of the windmill were necessary to indicate and to eliminate the weak components which finally resulted in a complete redesigning of these essential parts which are still simple in manufacturing and assembling.
The mainshaftbearings
Two kinds of loads actuate on the two main-bearings. The front bearing carries a radial load of appr. 1000 N (rotorweight) and a thrust-load upto '7000 N, while the rear-bearing takes the alternating forces carried out on the crank-mechanism. If these bearings are nicely adjusted the above mentioned assumption will be actual but unfortunately there are constraints in practice. Due to considerable unroundness of the shaft and poor quality of the bearing-bushes (sleeve-type) a proper fixing of the bearings to the shaft often led. to complications. After some time the bearing-bushes got loose and the shaft started rolling and .clidj >lg in them. This rolling effect will increase while the shaft s=licipc backwards by the windpressure acting on the rotor


and the sharp edges of the :~ngle i r-on hut) start scraping n.t the

f`yoni; of the benr i nghoclse.0-l C:OIAt-:- c;few thous`ands of revolutions

are sufficient to destroy th? bearing-house completely &nd the

ball-bearing (due to the scraping-dust),

This problem is solved by

locking the bearing-bush to the shaft by a short spotweld so

that the shaft cannot roll and slide, Proper position of the

bearings to the shaft with regard to the axial clearance of the

bearings between the inside edges of the bearing-housings will


is going to carry whichload.

Besides,there were the problems with the bearings them-

selves. Since the bearinghousings are bolted to a welded frame-

work self-aligning

bearings (double-roller

type) have to be


double-roller selfadjusting - .


Although these bearings are rather oversized, according to the


the lifetime proved to be too short

(appr. 1 + 5 millions of revolutions).

This means that these


are not representative

of the bearings available

in the locai markets. Many bearings of several trademarks are

offered but specifications are unknown and the quality is not

guaranteed. Often these bearings are @Yepaired" old ones: the

rollersurfaees have been lightly ground and oversi.ze balls are

applied. The result is surprising; only insiders can detect

the minor differences from the new ones .Since these bearings

are not able to stand the loads and S.K.F. is not easily

The crank
In the previous designs an "open" .crar& has been preferred to a "closed" o!-L~.' The advantage of the application of an open crank is that the radius of the crankpin can be adjusted easily. Moreover the assembly of the connectingrod is very simple because there is no need of a partable construction of itsbearing-housing and bearing.

An adjustable crankradius always creates the opportunity to

"match" the windmill to the pump, depending on the actual

windspeed and the elevationhead.Therefore

the first windmills

have been proVid4 with a slotted crank in which the crankpin

can be positioned at any radius between 50 and 125 mm (2" and 5").

This slotted crank, composed of angleiron (1.5"), was clamped,

just like the hubconstruction of the windwheel at the end of

the mainshaft but this unkeyed connection proved not to be


led to fretting and the end of the shaft

as well as the crankhub were spoiled.Several sffortshavebeen

undertaken (central bolts, keys),but a lOO%rreliableclamping-

device has never been achieved.

Welding of the crank to the shaft-end solved the problem as long

as replacing of the mainbearings of the millshaft was not

required. In general welding to a shaft is not advisable,because

straigthening is difficult to ma,intain., but also the metal-structure

~511 be affected or even spoiled due to the enormous heat,

Since this welding takes place at the end of the shaft which

un-s traightning of it will occur in the rear section of the

shaft which can be eliminated in the crank itself. Also the affected


is situated behind the rear bearing, so if a

fatique-crack will appear in this place the rotor remains and

a further damage is avoided.

Although the slotted crank served its purpose very well,it

needs a very tight fixing of the central crankpin-bolt in order to

prevent that it gets loose and starts sliding in the slot.

This was a result of experiments with 8" dia pumps (normal pump-

dia 6"). Moreover in practice only two of the infinite adjusting

facilities were used: maximum and minimum position! This led

to the idea to introduce a new crank with three or even two

crank radia only. *

The crank is composed of several pieces of flat iron which
are welded to the shaft end. Later on it proved to be neces sary to create reinforcement by applying an extra flat
situated near the shaft in order to obtain sufficient
strength. The E-shaped crank provides a sliding fixation of


the crank pin nut which facilates assembling. For aesthetical and safety reasons the sharply edged tip of the crank is closed by welding a small piece of f1a.t to it.

A crank made of a solid steel block, directly screwed to

the shaft, showed also satisfactory results especially

regarding the revision to the rear bearing. However,

generally, the less sophisticated construction of the

composed crank, as described sofar, is preferred.

The bearing surface of the bolt-type crank pin loading the

crank plane proved to FIe too small for yield of the material


in a loosening crank pin- One can imagine

that this leads to disastrous damage of all the components

concerned (including the connecting rod and itsbearings).

Improvement has been achieved by applying a collar bush combined with a central crank bolt which latter ensured a sufficient strong connection of the bush to the crank plane. No indication of yield orloosecrank pins have been discovered

The connecting rod
To the upper-and lower-end of the connecting rod (flat iron) cylindrical steel-bushes are welded which eilclose the flanged nylon bearing-bushes. In the beginning these steel-bushes have

been maii~ of 1.5" gaspipe sockets after the screwthreads have been removed by turning. Tne remaining thickness in combination with the poor metal properties lead to unroundness and even cracks of these bearing houses, in spite of the f act that reinforcement-. flats have been applied. Therefore it was decided to apply stronger bushes machi!led of scrap which remains from the millshaft if it is sawed on the right length. Nowadays the thickness of the bushes amounts to 2 6 mm. During composing and welding the connecting rod, a simple devicpis applied in order to obtain a correct parallelism of both bushes. The unroundness of the bushes,

causea by the welding-procedure,

does not affect a smooth

function of the nylon bearings which fit nicely over the

crankpin with 0.2+ 0.3 mm. clearance in radial sense. An axial

clearance of appr. l+ 2 mm. is usual. Up till now no nylon-

bearingbushes wore out ,and were replaced accordingly.

The crosshead

The crosshead is situated and guided in the towerpipe. As mentioned before a rather high rate of wear of the very first crosshead has been noted, so another one had to be designed. Since the towerpipe ( gas pipe ) always has a seam inside a proper guidance of the crosshead is impossible unless the entire seam is removed. This can only be done by grinding, but the

5. :
"A y ~l."-"p` II /,


grinding equipment is- rather advanced, not always available

ad moreover the resulting SUrfaCf? remains rough.

F:f'forts under-1,n.ken in obtaining an inside surface of an

acceptable roughness were not succesful during constructing the

first prototype (KK I) in India.

The idea was born to install a P.V.C. insert inside the towerpipe.

This insert is made of a 4" P.V.C.-pipe of sufficient length

from which a strip is cut in such a way that the circumference

measure isa littleless

than the inside-one of the towerpipe.

The opening just provides enough space for the seam which is

completely counter-sunk then. At the bottom a small bolt,

screwed through the towerpipe,locks the insert in axial sense.

Since P.V.C. is rather soft the crosshead must have a sufficient

wide contact surface in order to prevent high specific loads.

Also sharp and hard edges must be avoided (scraping!). It was

decided to compose the crosshead of hardwood because this

material was expected to be a better counterpart for the

P.V.C.lining than e.g. steel.


11111 1 3

Lubrication is applied by grease from time to time, but also

soaking in expirc?d motoroil before assembling assures a smooth

motion,,a,nd oil in Lhe square hole pr&ovides a continuous lubri-



2.2.4. The rotor

The twelve-bladed rotor is composed by the rotorframe,

bladesupports and blades which are completely prefabricated

in the workshop. During installation

of the windmill the

rotorframe is lifted and installed first and the blades

are fixed afterwards.

The rotorframe.

Using thee shaft, bearings and headconstruction as basement

during composing and welding of the rotorframe the latter

is composed very accurately, However the disadvantage was

felt during transportation:

the diameter of the rotorframe

amounts to : 3.5 m, so its transport can be complicated.

Nowadays the rotorframe is composed of two halves: which

can be obtained easily due to the fact that the hub is


in the beginning there were some doubts concerning the

strength of the inside ring which is supposed to carry the

centrifugal forces acting on the blades. Although no defor-

mation has been indicated the rotorframe of the nex-t four

prototypes have been reinforced by extra flats which provide

extra support to the inner-ring.

Later on this precaution

proved to be superfluous bacause an overhaul of the KKI-

windmill pointed out that the inner-ring still was perfectly

circular . Since the rotorframe proved its reliability


combination with the broad blades of the old design no

doubts remain concerning its strength in combination with

the smaller blades. Nowadays the rotors are constructed

without these extra reinforcement flats.

The hub.

During the introduction of the b'igger pumps (6" and 8" dia)


and allowing the windmill to run in windspeeds higher than

12 m/s the hub of the rotor started slipping and screwed

backwards over the shaft.

Obviously the clamping of the hub was not sufficient

in order to prevent the rotor from slipping which res~l-ts

in wrong position with regard to the shaft (Rotorframe

is fabricated on its shaft to eliminate unstraightness)

Defenitely a "shaking'* windwheel will result,which leads to

vibrations and wearing and tearing wilI! result of several

parts like the security-lock,




so avoiding this is a must.

Practice proved that just a bolt through the shaft acting

as a key is not sufficient:

if slipping occur6 the two

hub-halves (angle-iron) will be bent open and the entire

hub- and rotor-construction

is spoiled.

Finally an accep'- ..dtjle solution for this problem was found

by applying two extra clamping angle-irons clamped firmly

by two I/2" bolts.

The blade-supports are manufactured from sheet-iron and if one is able to prefabricate two sets of Welve equal bladesupports an accurate running windwheel will result. This is achieved by making the supports in sets of twelve,while the

first of each set is used temporarily as a simple shape during '

cutting and drilling of the other eleven supports. After

marking and hammering nice blade supports result in satis-

factory shape and sufficient strength. However,in the first

prototype windmills the inside blade supports proved to be

not fatigue-resistant

as cracks developed at the bolt-hole

edges, near the inside ring. this problem was solved by apply-

ing 2mmsheet iron for the fabrication of the inside blade sup-

ports. The outside blade supports served their purpose best

and are still made of Imm sheet.

Some efforts have been undertaken to compose the blade supports of pre-drilled and curved flats which are welded to the rings. Unfortunately this experiment turned into a failure. Obviously due to the rigidity of these supports the forces setting of the blades in them was often accompanied by initial stresses resulting in increase of fatigue sensitivity of the blades

in places near the bolt holes. Such complications do not occur if the sheet iron blade supports are applied for these show g*elasticfv properties which are stress-relieving.
The blades.
The first three prototypes (KKI,KKII and RTI) have been equipped with sheet-iron blades which were curved 10%.



. .._._ - . - ----- ----

- -- -_--

-- _ __---


- ------

I - +----


_ --- --

-- _--_-















uneconomical (two blades per sheet). It was decided to narrow the blades by applying the properties of the 10% airfoils which resulted in a new blade-shape: three blades can be cut from one standardsheet. Unfortunately the starting- torque Scoefficient of the rotor decreased from 0.19 to 0.14 but the adva,ntage of costpricereduction made this quite acceptable. It was also proposed to cut the blade tips in a circular shape which might increase the efficiency by appr. 2.5 $. Although wind tunnel tests on this topic did not show any significant improvement. Some of the 12 PU 500 windmills have been fitted with blades of circularly shaped blade tips and are accepted (more or less) for aesthetic reasons.
2.2.5. The pistonpump.

From the beginning the major part of the time and energy

spent in developing the windmill-installation

has been

required for the evolution of the pumpsection.

Many problems had to be solved concerning the fixation of

the pumps and delivery-pipes into the wells and also the

design of the pistonpump changed from time to time.

Since the windmill drives the pistonpump with R.P.BS-

numbers of 15 up to120 the application of air-chambers

proved to be necessary in order to decrease the acceleration-

forces acting in the pumprod, pump suspension and trans-

mission. These forces were always underrated and several

times cracks of the pumprod=connectionSand loose pUmp


In some cases even the entire tubewell

was pulled -out and led to damage of the pump and pumprod

as well. .^.^_.

'lhe pump rod and pump rod,~connections
A.s mentioned before the first windmills have been fitted with woodenpumprods which proved not to be successfUle

The wooden rods, if available, do not exceed the length of appr. 2 m and they are always unstraigth. Since the length of a total pumprod is at least 10 meters many connections have to be applied.

These connections are realized by using simple Clamping

devices of angle iron. These seemed to be very strong but

sooner or later,due to the reverse of pumprod forces, these

connection got loose. Actually these failures resulted in

applying steel-pipe as a pumprod. In order to minimise the total

weight the first steel-pumprod has been composed Of 1" pipe with

1 mm. thickness only, but still the problem of simple joints


A simple rod connection,made by two belts through it,did

not serve the purpose. Such joints also got loose,worn

out and finally cracked. Flanges made offlatswhich


been welded to the pump rod ends offered a strong and reli-

able joint after bolting them.

The disadvantages of such connections were that the pump

rod sections were difficult

to align and moreover the

assembling of the pump-rod guides became complicated. me

latter had to be partable. Moreover the thin walled pump

rods needed many guides in order to prevent collapse during

downward motions. Therefore normal 3/4" gaspipe has been

applied for all the pumprods and the introduction of a new

and simple joint solved all the pump-rod troubles.

Besides the developing of the pumprod and its connections the piston became the bottle-neckwhen the stronger pumprods were introduced. Since the piston is made out of wood its strength was improved by increasing the thickness from 1.5Wup to 3)). The leather piston sealing proved to be very suitable as far as wearing and tigthening were concerned, but due to the high friction coefficient of the leather/P.V.C,-combination

&nd the high stati-c pressure the leather sealing set up and

caused a sticking slide of the piston in the cylinder resulting

in a very high startingtorque.

First a rather high windgust

(up-to 6 or 7 m/see) was required to start the windmill while

the minimum runningspeed amounted to 3 m/set. !Che idea was

born to apply a small but permitted leakage in the piston in

such a way that the windmill can slowly move the piston in the top-position and is able to start running unloaded during the first half revolution. Probably the stored kinetic energy in the rotor ~~i&-t maintain the rotation so the starting-windspeed can be r~d~;r~d .
DuI*i.ng these experiments the inducted friction due to the set up of the leather sealing still persisted and just a small improvement has been noted.

A new experiment with a sealingless piston showed a considerable

improvement due to the absence of the Coulomb's friction

acting on the P.V.C!.-leather-surface.

In spite of the fact that

the volumetric efficiency is badly affected at very low

R.P.M.-numbers this efficiency seemed not to be influenced

if higher R.P.M.-numbers are actual. Although the expected

lifetime was quite uncertain concerning the increasing of

leakage (decreasing of volumetric efficiency) sooner or later

this sealing-less piston has been fitted in the existing

prototypes. The experience with the sealingless pistons in

practice is rather positive: depending on the sliding-fit


the beginning, the purity of the lifted water and the actu-






al accuracy of the pump-rod guidance (!) the lifetime

varies considerably.

A worn-out piston/cylinder-combination

still operates,but the

efficiency becomes unacceptably low. Therefore maintenance

and repair must be carried out,and if possible in a quick

and uncomplicated way.

Since the leather valves needinspection and repair too the

above mentioned demands had to be taken into account with

regard to the pump designs. Finally the required airchambers

and the materials available determinated this pump design.

Right from the beginning the development of the piston pump

reached many stages resulting in type (A upto J) and still

the way of developing by trial- and error detected the

weaker sections in these designs. Time, wind, patience,

optimism and invention in particular proved to be the most

required tools in order to develop the piston-pump step by

step. Briefly types A up to J will be described.

A Band D

Basic type; very simple and reliable (handpumps) as

far as no high frequencies actuate!

C '?he need of airchambers were felt. Forces acting on

the internal chambers were underrated. Complete

chamber sections got loose and caused damage to

foot-valve and piston. Especially the airtightness

of the pressure chambers was difficult

to achieve.

When the P.V.C.-sockets, elbows and T-connections

were available, external air chambers have been

composed Gnd applied. But action-reaction


in the external pipe&Q caused vibrations while the

presence of air in the pressure section remained

doubtful after some time of operation. (Air solves

into the water under high pressure conditions.)

Suitable for moderate elevation head (no pressure

chamber). In this stage 7/8" pump rod and 7/8"

connection have been introduced.


Cheaper and simpler construction; by applying an

internal suction-air chamber several reducing sockets,

T-connection and elbow were superfluous. Total section

is clamped between two flanges.


Similar type but suitable for higher elevation head.

Air in pressure chamber is maintained due to small

but permitted leakage in suction section (by bicycle

tube-valve). Due to the location of the suction-air

chamber a foot valve with central supply has been

applied. The latter did not operate satisfactorily


combination with the sole-leather valve disk which

deformed in such a way that leakage resulted. The

pump is simple to open and to inspect.


The "combination pump" combines the advantages of

pumpty-pes F and G resulting in the following

properties of the pump:

- double air-chambers (reduces pump-rod forces)

- reliable valves (self-priming)

- proper pumprod guidance (life-time piston)

- "automatic air injector"(pressure

chamber remains


- total pump sections can be replaced easily (quicker

repair, maintenance)

- pump is opened quickly (repair, maintenance is

normally carried out in the workshop).


This pump is completely composed of steel components

which are jointed by weldings. The pump is'easy to

open in order to carry out maintenance and repairs. me

cylindc; is made of gaspipe from which the inner seam

is removed by means of a simple cutting device and final

finishing is done by honing. Though its construction is

more advanced than those of the otherpumps its reliabili-

ty proved to be better. (See chapter 4).




J. Same type but the footvalve is made of sheet irona This

allows a shorter cylinder which facilitates

removing the

seam. Moreover no doubts remain whether leakage occurs

between valve-body and cylinder as sometimes seemed in

the foregoing pumptype. Also this pump has been tested

thoroughly as described in chapter 4.

The air-injection

system in both the pumptypes I and J has

been abandoned. as pumptests showed (by means of waterlevel

gauge) that the air in the pressure vessel still remained.

Results of thorough tests of the pumps latter mentioned

are describer! in chapter 4.

2.2.6 Pump suspension
Practice showed that fixation of the windmill pump can be rather complicated. Especially during stormy weather considerable forces are acting on all moving parts and the pump structure in particular. One can imagine what happens if the suspension shows some play. In some cases the entire tube-well got loose and was simply pulled up till the piston started hammering on the foot valve!

Besides the need of an adequate fixation in vertical sense

supports are also necessary to keep the long pipe lines

centered. Previously this was achieved by means of wooden

crosses which have been wedged in the well by means of keys.

Though such arrangements served the purpose some danger

exists as people tend to stand on the crosses during climbing


into the well. A simple tripod-Tambrella*f, combined with the

pipe-line socket, proved to be very useful. . Centring is done

simply by adjusting the "pods" and locking them in the brick

wall by means'of a few hammerblows.

Proper fixation of the entire pump sections by applying the following constmnctions: - Anchors masoned in the well - Expanding clamping-devices - Tower-pipe connection - "Bridge"

have been achieved

Anchors masoned in the well

~..n case of an excentrically placed tower, which is often done

in order to faciiitate waterlifting

the well-structure

itself provides

section. Anchors are masoned into

pipe-line can be fitted.

by means of other devices,
attachments for the pump the wall so that the delivery

Expanding clamping device
Provided that the well casing is of proper construction the entire pump arrangement can be fixed suitably by means of an expanding clamping device. The latter is composed of I"gas pipes which are welded at right angle either to the bottom section of the pump or to the suction pipe. Mutual braces (angle iron) ensure a rigid construction. The clamping arrangement is obtained by fitting longthreaded bolts or screw studs into the extruding pipe ends. Tightening their nuts causes the expansion and fixation of the frame.

In order to prevent damage to the bricks of the well casing it is necessary to enlarge the bearing surface by means of pieces of angle iron.
Such clamping device proved its worth in combination with the "bridge" construction.
Tower piye connection
In cases of weak or brittle well casings the pump rod forces are hardly carried by masoned anchors and other devices to such well-casings. Therefore it is preferable to carry over the forces directly vi.? a tower-pipe extension, which is composed of two long angle irons and some clamps. As a matter of fact the entire pump section hangs from the tower pipe and all acting forces are led through this suspension instead of through tower structure, fundation, well casing and the pump suspension respectively.
Sideward vibration hardly occur and if so,retaining of the pump section is simply achieved by means of applying the "tripod umbrella" (as mentioned earlier), masoned anchors or the "bridgewg.

If a strong and proper well casing is available the simple bridge construction provides a rigid suspension of the pump section in horizontal and vertical sense both.

The delivery pipe is clamped to

the bridge by means of two

clamps. By opening these clamps,

and the pumprod-connection,the

entire topsection of the pump

(including delivery line) can

be lifted in otder to facili-

tate repair and maintenance


the moving parts of the pump.

2.p 3

2.3. The 4-bladed 2.50 m. dia. windmill for low-water-lifting.

In winterseason surface water, which remained from the rain-

season, is used in some cases for the irrigation of the winter-

crops. Several small farmers lift this surface water by

manpower to their fields; the idea was born to develop a small

and cheap windmill which can do this job.

Although ihis low-waterlifting

wasnot planned in the project-

schedule and its need was not actual,some time has been spent

by the Dutch windmillengineers in developing this windmillP

Therefore some attention is paid to this subject.

Actually it was the size of the scrap of steelsheet, which

was left from the production of the big windmills which

determinedthe dimensions of this small windmill. The first

prototype has been fitted

with a 4" pistonpump combined

with a 3" towerpipe. The total structure stands on its pump-

section and is held in position by guys. The latter are

simply made of 3/8" steel rods and anchored by pegs to the

ground. 5e conversion of the millshaft-rotation

into the

up-and downward -motion of the pumprod is obtained by means

of a crank-pumprod-mechanism. 5e elasticity of the rod allows

its outward bending besides the up-and down-movement so a

connecting-rod and crosshead are superfluous. Two wooden

guides, situated in the tower, provide a further support to the 6 m.-long pumprod.The piston is nzade sealingless in

order to enable the windmill to start almost unloaded. This is

necessary since the starting-torque

of the 4-bladed windmill

is very low. The measure of leakage is determined by trial

and error and a clearance of the piston in the cylinder of

aw= 112 % of the piston-diameter proved toleadto satisfactory results. In spite of this leakage the volumetric

cfficienckes are reasonable and even the value of 100% is ex-

ceeded due to the inertia-effects

in the watercolumn if high

R.P.M.-numbers occur.

In 1978 the total material-costs of the small windmill amounted

'o less than Rs. 500,- while the required manpower for

building it is estimated at appr. 50 manhours.

Lateran 3 other small windmills have been built and installed.

These were slightly different in size and design (3.2 m. dia.,

tip-speed ratio 5) and consecl_uentlymore costly.

In the longer run some technical problems (e.g. fatigue-cracks

in the blades) in combination with the low windspeeds made it

clear that the use of these small windmillSwas not quite

succesful in the Ghazipur district, but on the other had its

simplicity and costprice caused that it was adopted by the

WOT (TOOL-member in The Netherlands). Developing and testing

there , proved that this small windmill can still be attractive

for low-waterlifting

purposes while the basic design is main-

tained and applied in other WOTwindmill types.

In the framework of these windmill-types the small windmill is

coded: 4 PU 250 (4-bladed pumping unit of 250 cm. dia.)

Applying some small modifications The windmill can be fitted

with a transmission to groundlevel by means of a rotating shaft

for multi-purposes (4 MP 250). Also generating electPicity

is possible by applying this basic design and a 4 EG 250 will


_ .------ - _

. 1



L@,$,/,J,' / '


_ ----






- --

-I?3 x60 500 -.







TlGtlT FIT .IN 21





93 lO"`O~ ! T'ON






ROD 0 12-l=

x 3%

B&T- tiUi - WASHER

Ml?=20 lWyI'.P;I


Ml25125 (W'b-. 53


M2L IW+`l


H1 _ I.WWl


rla.l@ l.W%-.Cl


M 2L$ci c_wv . 6.1

xm i?@iimm


0 33 s 30

STEEL &lSH----



-0----13 I 33 fEik(-W0~0-0


SOLELEATtjER 9 290 .-6


-- 0 150 m 6


ROD .oe WI -~

..Em . LO


ulrul~ll~lbillbh I







\ ./

L t,,r/mw"-I



3. Test-Rig for 12 PU 500 Windmill

In the past the development of windmills took place on the

site itself where they have been installed which created a

lot of inconvenience with regard tobreakdowns and repairs.

It will be evident that such aprocedureof technical develop-

ment takes a lot of time and moreover the image and populari-

ty of the technology might be degenerated due to failures.

Therefore it is advisable to simulate the practical conditi-

ons in such way that the development-process

is done in a

shorter time and under heavy conditions without being de-

pendent on the fluctuating and unreliable wind. So the idea

was born to get, a TestlRig developed in which the essential

parts of the windmill can be tested thoroughly. This acti-

vity was carried out in the Research and Development section

at the Polytechnic in Allahabad.

3.1 Demanded features of the Test-Rig

Since the 12 PU 500 Windmill (12 blades Pumping Unit of 500 cm Rotordia) proved to be the most applied windmill SO far the urgency of developing its essential components is favourable. Cn the other hand results extracted from these tests can be extrapolated usefully for other windmill types like 12 PU 250, 12 PU 350 and 12 PU 700 Windmills which are under development too. Therefore it has been decided to get the Test-Rig develoflzd for testing the parts of the 12 PU 500 Windmill. Considering the structure of the essent2al moving parts and Pump of this particular windmill, testing of the following parts should be carried out with regard to wear and tear:

- Main shaft bearings - Crank Pin - Cross Pin - Cross head / P.V.C. - insert tower pipe - Wooden pump-rod guide situated in the lower
towerpipe. - Pump-rod guide (Bronze and Nylon). - Leather sealing / pump rod - Piston / Cylinder

end of the

- Leather valves (Piston-Foot-Valve). - Joint between tower pipe and rotating head construction.

Actually there is no need of testing the rotor with regard

to wear and tear, though fatigue cracks might occur. Field

experiences showed that fatigue cracks had been developed

right in the inner ring bladesupports which were made accor-

ding to the first designs. The improved bladesupport proved

to be sufficiently

resistant with regard to fatigue and no

failures appeared anymore. In this context there is no-need

to have the rotor tested.

Actually a normal 12 PU 500 Windmill installation the entire structure for a Test-Rig. Instead of flywheel, which is driven by an electric motor, installed in order to obtain a regular drive to ating loaded mainshaft.

provides the rotor a should be the fluctu-

3.1.1 Drive of the Test-Rig
To estimate the required R.P.M.-number for the Test-Rig we consider the 12 PU 500 Windmill under extreme conditions. In practice the maximum R.P.Mmnumberdoes not exceed the value of 120 sincethe windmill is locked in secured position automatically by means of its automatic security device. This R-P.M.-number represents a reasonable value for testing pumps and the moving parts of the windmill.

Applying an electric motor for the drive a reduction of 1:12 is needed in the transmission. !&is is simply achieved by applying a V-belt transmission (directly) from motor pulley to the circumference of the flywheel. In practice the nominal diametres of pulley and flywheel tuned out in 70 and 850 mmrespectively while three V-belts (B-type) completed the transmission. The required power of the electric motor has been estimated at 4 K W.

3.1.2 Selfmoving head construction

In normal condition the head just rotates if the wind-

direction changes which means that in the long run wear

and tear might be expected in the joint of the tower-pipe

and head-construction.

This rotation is made by piston and

cross-head as well.

However,the tests should be carried out under laboratory conditions independently from appearing winds and wind directions. This implies that the rotation of the head construction should be simulated by forcing the head to carry out this movement regularly.

In practice it is very rare that the head rotates more than 180 degrees while in most of the cases swinging movements within a certain range of some degrees occur for a certain wind direction. Actually this is the worst condition from point of view of wear and tear since still the same part of the bearing surface (here: tower-pipe) is carrying the load.

So an attractive simulation seems to be achieved by forcing the head to swing to and fro around the tower pipe within a certain range`of degrees. This cycle should be carried out not too quickly and not too slowly but in such a way that the ltibrication properties of the bearing are the same as under normal field conditions. However,it is quite difficult to quantify and to extract the required movement from the latter mentioned: wind patterns are different from place to place and from time to time.

As a compromise it has been decided to swing the head construction to and fro within the range of plus and minus 45 degrees (total swing 90 degrees). In order to obtain this movement a small crank is driven by a worm-wormwheel-transmission (1:8O). !l!he latter is fed by the main shaft (120 R.P.M.) on which the worm is fitted. The crank is linked to a joint of the tower by means of a con-

3.5 netting rod. This mechanism forces the head construction to swing over the tower pipe. Onecycle is completed in 40 seconds while this motion appears to be quite"natural",
c r

3.1.3 Simulation of the elevation head

In practice the windmills lift water over several elevation

heads upto 20 metres or even more. Then the dimension of

the pump applied depends on the actual head. Pump dimensi-

ons in the 12 PU 500 serie are: 3", 4", 5'1, 6" and 8"

(piston diameters.)It

is expected that, as far as pump-

testing is concerned, this range of elevation heads should

be applied during these tests.

In laboratory conditions it will be possible to simulate the elevation -heads by applying long pipe-lines leading to an overhead tank which should be situated on sexeral levels.

This method is expensi.ve and it creates inconvenience with regard to installation will not be practical.

quite a lot of and of course it

Simulation can be carried out in a simpler way by pumping the water from a sump (bottom basin) and getting it delivered into an air-tight vessel where a static pressure persists. This pressure simulates the elevation head and it is built up by compressing the existing air which remained above the water table in this vessel. The latter is achieved by restricting the exhaust opening at its bottom-side by means of an adjustable valve over there. Its resistance causes the required pressure to get the actual water flow forced through it,hack into the sump.


The main advantage of this system is that it is compact to be built. However its disadvantage is that the inertia-effects and pressure losses, which normally appear in long and narrow pipelines,are not simulated since they are not there. Instead the colossal vessel with compressed air is there which provides a huge airreceiver which acts as an capacitor

equalizing the irregular waterdelivery from the single acting piston pump. Therefore it is advisable to consider this air pressure not to be the static elevation head (geodetic head) but the actual hydraulic head! The latter is known as static head plus the pressure losses.
The pressure vessel is situated in center of the Test Rigtower in order to integrate it with the entire structure. The tower provides suitable attachments for the suspension to the several lattice joints.However a practical problem has to be solved with regard to the leading of the pump rod through the centre of the tank since its airtightness is most essential. This implies the application of a pump rod sealing which should not permit any leakage of air. The latter is achieved by applying a pipe in centre of the tank which bottom ends under water surface. On its topportions the pump rod sealing is situated. In this way air leakage is avoided in case the sealing proves not to be perfect
since in that case only water-leakage will appear without any effect to the functioning of the airchamber.
3.1 .A Instruments
In order to run the Test Rig in a proper and safe way and to gather information with regard to lifetime and efficiency of the pumps to be tested, several instruments have to be applied. These instruments are:
Switch and relais box Amps meter and KWH-meter Water meter Revolution counter - Pressure gauge Water level gauge
Switch and relais box
Besides the application of a main switch box it is preferable to have a relais box installed. Purposes of this device are:

- Switching-off the remainingphasesif

one of thephasesis

lacking due to power cuts. (If this is not done in time the

electric motor will burn definitely.)

- Switching-off the electric motor in case one (or more; of thephase-currents exceeds an unacceptable value. (By means of thermo-elements.)

- To enable the operator to start and stop the Test Rig by a "simple push of the button". The actual switch is made by the relais.

The relais box is attached to the head construction of the Test Rig in order to obtain a short multi wired connection to the electric motor. This also means extra safety since the Test Rig can only be started by the operator which is at the machine itself; so the machine can not be started from ground level while an operator is carrying out some job at the top of the Test Rig. In case of emergency the machine can always be stopped at ground level by using the main switch.

Amps meter / KWH-meter It is advisable to have an Amps meter installed in order to detect any overload. Moreover the Amps-meter provides information to what extent the machine is loaded. !lhe same applies for the KWH-meter (+phase)which counts the amount of energy which has been consumed to drive the test pumps under different conditions. Till a certain extent this power consumption can be derived for the specific pump under the specific conditions.

Water-meter This instrument measures the total amount of water which has been delivered during the pump tests. This reading is most essential in order to calculate the volumetric efficiency of the test pump. The Water-meter is situated in between the pressure tank and the adjustable valve through which the water is


flown back into the sump.

Revolution counter
The revolution counter registrates each revolution of the main shaft. Its counting is essential for calculating volumetric efficiency and judging lifetime features of the test pump and moving parts as well.

Pressure gauge
As mentioned before the pressure in the tank represents the total hydraulic head, which is indicated by the pressure gauge.

Water-level gauge
It is advisable to indicate the actual waterlevel in the pressure vessel. This in order to detect any leakage of air which will lead to a decrease of the capacitor effect of the airchamber resulting in a **hammering" run of the machine. Under static conditions the waterlevel gauge also provides a (rough) indication of the actual pressure in the tank.

3.1.5 Way of testing

Running the Test-Rig is done in day-time (working hours) in

ordertohave a close look at the experiments. Day by day the

following readings aretaken and noted:

- Water passed through water-meter.


- Wumber of revolutions indicated by counting device.

These readings provide the information for calculating the

volumetric efficiency. (See next chapter.)

Pumps to be testedc4" - 5" - 6") are run at 15, 10 and 7

metres head respectively simulating the hydraulic heads.

Adjustment of this head is achieved by tuning the adjustable

valve situated in the return-flow. One should be aware that

this adjustment is "tricky" for the following reasons:

- Its initial adjustment is quite a time'-consuming procedure

which is achieved by regularly made check- ups and re-

adjustments of the valve.

- After some time the features of the test pump change resulting in higher out-put (setting of valves) or lower output (wear) with the effect that the equilibrium of the system is going to be disturbed. Hence the regular checkup' are necessary too in order to give re-adjustments if I;eeded.

The pumps are tested for at least one million revolutions while recording wear and tear and calculating volumetric efficiencies is done at certain intervals. The out-come of these will provide the information to base conclusions on.

3.2 Pumps to be tested
Since the completion of the Test Rig (March 1980) four types of pumps have been tested each for one million revolutions. The test pumps were:
---- 4" s 500
---- 5" ss 500
---- 6" P.V.C. 500 ---- 6" S FjOO
Each pump has been tested turn by turn during cycles of 250,000 revolutions loaded at hydraulic heads of 15, 10 and 7 metres respectively. At the end of each experiment of 250,000 revolutions the pumps have been chscked up (and revised, if necessary).

In theseeheck-ups the qualification

of wear proved to be

rather complicated. This in the sense that e.g. play of


and pumprod/pumprod guide are difficult

to measure and to define due to unroundness of the several

parts. Although initial unroundness exists in the cylinder

(black pipe, P.V.C. pipe) and pump rod (gas pipe) which are

acceptable (from the point of view of Appropriate Technology)

wear and tear seem to cause an extra unroundness to these


particular parts. No doubt that the latter develops due

to improper alignment of the structure and slight bends in

the pumprod sections, (which is always the case in practice).

In order to judge the extent of wear of the parts in such

cases the non-measurable qualifications

have to be expres-

sed in degrees.

3.2.1 The 4" S 500 (Type I, see chapter 2)
This type of pump has been designed and introduced in the beginning of 1980 as a deepwell-pump. (Applied for the first time at Gohri.)

It consists of two main sections as follows:

- Topsection:

composed of 2 mm steel casing, delivery pipe, top flange and pumprod guide. 'Ihe latter contains a brass guide bush. The space between the delivery pipe and the casing provides the air pressure chamber.

- Eottom section: Composed of 2 mm steel casing, steel cylinder, top- and bottom-,fLange, wooden foot valve body and leather foot valve. The wooden sealingless piston has a proper fit in the cylinder which has been honed by means of a simple honing device.

Though the nominal cylinder diameter diameter amounts 106.6 mm. (Actually for 5" and 6" pumps.)

is given in 4" the exact such deviations also'exist

The eight holes in piston and footvalve body are situated at a pitch'circle of 0.65 D and the diameters are 0.2 D. This has been standardized for all the pumps.


3.2.2 The 5" SS 500
Basically this pump is of the same construction as the former though the construction of the footvalve differs and the piston has been fitted with a P.V.C. piston ring. These special arrangements provided the "special" label to this test pump hence ah extra S has been taken up in its code number.

Its foot valve arrangement consists of a wooden footvalve body and a 2 mm steel disk which has been lined with leather. This disk contains a steel hub which slides over the central bolt of the valve arrangement and is pressed to the valve body by a spring.

Since the rigidity of this valve disk does not provide the


if opened like the elastic leather valves

have, extra '*flow round" space had to be created. This has been

achieved by constructing a wider bottom part (of 6,') to the 5"

cylinder. Hence the outside dia of the wooden foot valve body

has been made at the nominal dia of 6" but of course the hole-

arrangement (dia, and pitch) refers to the actual 5" cylinder


As an experiment the piston has been fitted with a PVC-ring which should act as a sealing. This 8 mm wide ring has been cut from a 6"PVC-pipe and is situated in a groove in the wooden piston. This experiment could provide lifetime information of such a sealing.

3.2.3 The 6" ?.V.C. 500 (Type H, see chapter 2)
This pump, known as the "combination pump",has been developed in Ghazipur (1979) and proved to be of a quite satisfactory design. It consists of a Top - and a bottom - section.

- Top sestion

: Composed of a 6" PVC-pipe (1.20 m long), top flang and delivery pipe. The latter carries the pumprod guide (nylon). !Iere


also the space between delivery pipe and

PVC cylindercreates

the airchamber. The

wooden sealingless piston runs in the PVC-

cylinder with a proper fit.

- Bottom section:

Composed of a 6" PVC-pipe (0.75 long), wooden footvalve body and leather valve, PVC (4") airchamber and bottom flange.

Both sections are jointed by means of 4 anchor rods.

.2.4 The 6" S 500. Type J, see chapter 2
As a matter of fact this pump is basically the same as the other steelpumps. Only the foot valve differs: which is made of a sheet-iron disk in stead of wood. Therefore this pump is shorter in total length resulting in a more economical use of sheet materials during its fabrication.

3.3 Test results

The schemes given on the next pages show results of the check-

ups made after each interval of 250,000 revolutions.

As rrentioned earlier quantifying the extent of wear proves to

be quite difficult

with the result that judgements are given

in degrees of wear.

The first 250.000 revolutions
During this test interval it is found that the brass pumprod guide showed a better performance than the nylon one. The latter already wore. out inanoval shape which was already visable. The leather valves functioned without any breakdown and developed a good setting to their wooden counter-parts. 31 contradiction with the expectation the disk-valve of the 5" SS 500 developed a good setting too, though its "steel to steel" slide of its disk-hub over the central bolt showed some wear. The fit of the wooden pistons in the steel cylinder remained proper in the sense that some jerking fit in

some positions persisted. In the PVC-pump the fit already got so loose that some play between cylinder and piston
could be seen.

tie PVC piston ring,applied in the 5" SS 500, wore rather

seriously.The tips of the ring became sharp like a razor

blade. Moreover the Wear Was IlOt UIlifOrIn; the thickest pOr-

t-ion of the ring (4.0 mm)hardly showed any wear (approx.

0.2 mm) while the thinnestportion

measured 1.4 mm. It has

been decided to stop the experiments with the PVC ring for

the time being till the cylinder surface has been polished

thoroughly by the Plain piston after somenewlyundertaken

experiments. Probably a longer life of the ring would be

achieved then.

-After ihe first 500.000 reVOlUtiOns Still the brass pump-rod guides show a good sliding fit as well as the piston / steel cylinder combinations. All these moving parts showed nicely polished surfaces. The fit of the piston in the PVC-pump showed an increased play. Though difficult to measure, its play is estimated at 2m while a clear edge in the PVC is detected at the end and beginning oi' the track of the piston. Most problably due to the initial unroundness of this PVC-cylinder the piston seemed also to be worn off which is proved by a loose fit of it in an unused part of the cylinder. me nylon pumprod guide showed increased wear. Estimated play of pumprod 73-2 mm= The valves continued functioning WeI.:. Though once the central nut of the leather foot-valve of the PVC-pump ra loose (due to pcor fixation)= Obviously the hub of the steel disk-valve as well as I.ts central bolts showed no increase in wear. Probably its combination developed a good **run-in**.

After the first 750,000 revolutions

The brass pumprod guide still showsan acceptable fit while

the nylon guide wore out in an oval shape causing an esti-

mated play of 2 mm.


All valves are in good condition except the leather foot-

valve of the PVC pump which developed cracks due to fatigueso

it had to be replaced.

The piston / steel cylinder combinations are about to show the first wear, though not simple to quantify (say 0.5 mm play). It is assumed that wear in the steel cylinders (if any) is to be neglected: the inside surfaces just show nice polishing.

!Ihe experiment with the PVC piston ring failed again : at approx. 650,000 revolutions (so this ring just run approx. 150,000 revolutions) the ring wore offthat much that it slipped in between the piston and cylinder. This caused that the wedged piston has been pulled up to its upward position with such a force that a cleavage fracture occurredin the cross-head and the connection rod had to collaps during the downward stroke. (Fortunately the TestRig could be stopped immediately.)

The experiments have been continued again after removing the badly deformed ring.

After the first l,OOO,OOOrevolutions
!I!he brass bushes show detectable wear which is still acceptable, but the clearance between pumprod and nylon guide was increased upto 3 mm.

All the piston valves as well as the foot-valves were still in gcod condition. However,in the 5" SS 500 the spring, applied at the top of the disk-foot valve, broke (after approx. 860,000 revolutions) and had to be replaced.

Cle.arance in the piston / steel cylinder combinations was detected now and it had been estimated at approx. 1 mm. On the other hand the clearance ia the 6" PVC 500 increased to more than 2 mm.
3.4 Volumetric efficieacies
All the pumps &owed a decreasing volume?ric efficiency with in~:reasing numberof revolutions. Obviously this is due to wear cf piston and/or cylinder. Although in two cases (5" E5 500 and 6" PVC 500) considerable wear has been detected still the volumetric efficiencies exceed 75% for the 4" S 500, 80% for the 5" SS 500 and 80% for the 6" PVC 500 and 6" S 500 respectively. In all cases the decreasing behaviour of the volumetric efficiency does not seem to be due to malfunctioning of valves, since their condition did not change during the lmillion revolution tests.
It is hard to make a comparison between the pum&s based upon the registered volumetric efficiencies since relative leakage will depend upon size and pressure. Moreover taking this in consideration, it seems that the steel cylinders show better performance than the PVC cylinder as far as volumetric efficiency is concerned. Next pages show the testresults, which are calculated from the readings during 50,000 revolutions in intervals of 100,000 revolutions.
3.5 Conclusions with regard to life time- features of the test pumps
It will be clear that pumptesting to one million of revolutions just provide modest information with regard to lifetime features of the pumps. Though further testing should be applied, yet some conclusions can be drawn as far as the test results have provided obvious information.

- The combination of a wooden piston and PVC-cylinder can not compete with the wooden piston/steel cylinder comhina tion.

- Nylon Pumprod guides show a much quicker wear than those made of brass.

- A PVC-ring hardly covers a life of 250,OOC revolutions iii"ni 1 P it creates risks for the moving parts of the windmill (here: Test Rig) if it slips in between the piston and cylinder.

- me *tpplication of a spring on top of the foot-valve

(h ere: 5" SS 500) might fail due to fatigue (crack of the

spring), causing damage to piston/cylinder


- LeathLnAr _- valves, as applied so far, proved to be quite

reliable though one should apply a good quality of pro-

cessed leather (sole-leather).

The main disadv?-ntage of

leather is that after it has been dried its elasticity is

gone znd fatiguesensitivity


After testing four pumps (each during one million revolutions) the Test Rig completed another one and a half mil-

liorrs revolutions which should provide some significant in-

formation. As a matter of fact it drove pistons up and down

covering a total distance of almost 3000 kilometres. The head

construction was forced to swing to and fro more than 65,000

times in order to create the possibility

to study the wear

betweer tower pipe and head.

In thissee the following conclusions are drawn with regard

to life time features of the moving components of the

12 PU 500 v:indmill.

- NO visible wear has been noticed in radial sense in the joint

of tower-pipe and head. In axial sense the head gust sank approx. 0.5 mm during the run-in period. Obviously after

setting of the contact surfaces no additional wear appeared.

. s-




P% 90.




w .




' 3.49

. - ~-... __-_ -__
It is concluded that the joint will serve the purpose per-


- The bearings of the mainshaft (N.B.C.-double roller, self-

aligning) did not show any defects or play, while no additi-

onal lubrication was applied.

- Though once, the crosshead had to be replaced (cleavage due

to jerking piston) its bearing properties in combination with

the P.V.C.-towerpipe-lining

seemed to be excellent. However,

still the tests should be continued in order to find out to

what extent wear will develop in the long run.

- The nylon bearing bushes in the connecting rod did not show

any wear of relevance. Neither did the cross- and crank pin.

Though these results appear to be promising it is understood

that in a suitable windy area the 12 PU 500 windmill runs in

the range of 500,000 upto l,OOO,OOO revolutions per month.

Hence the experiences, gathered so far, just represent such a

windmill during the first three up to six months of its life-

time. In this regard these achievements should not be seen as

a milestone,but rather a beginning.


4. properties and characteristics

of the 12 PU 500.

As mentioned in the first chapter the windspeeds in Gha-

zipur are rather low and actucally the windmill sh!.ould be


adapted to these circumstances in order to

utilize the little availablewindpower in the most effi-

cient way. The size of the piston pump (piston dia. and

stroke) as well as the elevation head are the most impor-

tant parameters in the windmill installation

which define

the starting windspeed (cut-in windspeed) and the design

windspeed. The latter is the windspeed for which the maxi-

mum overall efficiency is obtained.

4.1. Input-Output relations, volumetric- and overall-efficiency.

Considering the total windmill installation

as a converting-

system the relationship of input and output is interesting

with regard to its efficiency, its capacity and the economics.

The applied measurement equipment at the R.T.I.-windmill


not suitable for the definition of the input-output relation-

ship since the windspesd is monitored interval-wise while the

R.P.M.-number and output of the windmill was fluctuating

continuously due to changings. in the.kriniispeed.

Accurate readings can just be taken if the circumstances are

stationary which means that the windspeed as well as the

R.P.M.-number are constant while theflowmeasurement is taken.

These data will provide a relation between input and output and

the required instruments are:

- anemometer

- R.P.M. counter

- flowmeter

Although these instruments were available the above mentioned

measuring could not be carried out satisfactorily.In


theflowmeasurements proved to be rather inaccurate due to the

irregularity of the waterdelivery of the single-acting piston



Therefore some efforts have been undertaken in developing a

simple flowmeter and finally two flowmeters resulted. The

first one consisted of a simple l-gallon oilcan. The flowing

water enters the open bottomside and is released through the


The static waterlevel in the can will repre-

sent the actual flow which is indicated by a "gauge-glass"

simply made from a plastic hose. These oilcans are standar-

dized so a similar flowmeter can be made in any place of the

world, but its use is restricted to a maximum flow of approx.

7 m3/hr. Moreover an irregular flow causesconsiderable errcrs

since this instrument reacts rather slow and airbulbs spoil

the readings depending on the way of entrance of the water.

The second device of flowmeasuring consisted of a shallow basin in which the flow enters the back-side and leaves this basin on the other side via a V-notch. The static waterlevel in the basin also representsthe actual flow according to severaltheo=tically defined tables, Unfortunately this equipment did not serve the purpose very well. The phenomenon of slowness, due to the large volume, caused that this device acts too late in order to determine' the actual output,


So both measuringdevicesfailed

because the slowness in indi-

cating made these V'instruments w unsuitable for direct flow


Due to this another way of flowmeasuring has been developed:

in a period of constant windspeed Bnd constant R.P.M.-number

(audible) a 20 1. drum is shifted under the output opening

of the pistonpump during the downward -stroke (no output!).

At the same time a stopwatch is switched on and from that

moment on the number of strokes is counted (e.g. four or five).

!l!hen during the downwards-stroke again, the drum is removed quickly and its conteilts will amount to the delivery of four

(or five) strokes. The timespan, registrated by the stopwatch,

the windspeed and the amount of water delivered will provide

sufficient information in order to calcuiate respectively the

input-output relation, the volumetric efficiency and the overall


The above mentioned method proved to be the most accurate and

reproductive way of measuring.

In the following table the results (screened) are printed of

the first measurement which have been carried out in May 1978.

Dia. pump: Stroke: Elevation head: Pumptype:

132 mm. 254 mm. (10")
6.5 m. A (see pistonpumps)

n (m,Iec .) (R.P.K)

I (dm3/sec.)































%f -
0.74 0.83 0.90 0.93 0.96 0.98 1 .oo 1.02 1.01 1.00

rl toot -
0.21 0.84 0.20 0.16 0.14 0.12 0.11
0.08 0.07

j "ww~+~&."'

;$,I "y77r>,,"


-*I~:`~, ~~;I&;**I~:li

i >J,S*<*se `> " &"&~

,$;, rq&pyp~






4 :+



~: /.' :`.,A

i" _.


i-8 -w:,


.-, ,'


The maximum overall-efficiency

amounts to 0.24 for a windspeed

of 3 m/see. For higher values of the windspeed the overall-

efficiency is decreasing up%o a few per cent only.

Improvements have been obtained later when the blades of the

rotor were replaced by the narrow ones and the pumpsize was

increased from 132 mm. to 150 mm. New measurements were carried

out in May 1979. The condition of the piston was rather bad;

wear of the piston caused a considerably affected output due

to internal leakage,but in spite of that higher overall-

efficiences have been extracted from these measurements:

Dia. pump: Stroke: Elevation head: Pumptype:

150 mm. (clearance
240 mm. 6.5 m.
E (see pistonpumps)


3.5 mm.)




(m/Iec.) [R.~.lL) (dm3/sec.)



































































Although these results are not quite representative (low volumetric efficiency) one can conclude that overallefficiencies of acceptable value can be obtained even if the pistonpump is not in proper condition. After replacing the piston new measurements have been carried out with respectively 0.5 mm. and 1.0 mm clearance.between piston and cylinder. !l!he difference in these measuring.data proved to be insignificant SO both data have been combined:


Dia. pump:

150 mm. (0.5+1 mm. clearance)


240 mm.

Elevation head: 6.5 m.


E (see pistonpumps)

't(W,-int' windspeed 2.5 m/set.





(in/set.) (RJ.Y.) (dm3/sec.)







I .8








































Decreasing the stroke from 240 to 160 mm. reduces the output and wcut-in*1 windspeed.

Dia. pump:

150 mm. (0.5+1 mm. clearance)


160 mm.

Elevation head: 6.5 m.


E (see pistonpumps)

Cut-in windspeed 2.2 m/set.




(m/set.) (R-LP.M.) (dm3/sec=)




















































liz the next paragraphs extra attention will be paid to the

:3 j :l

windmill-pump-combination. .- ,?_, . ,\

,",w'I,i',._, -,'. /,+..,: t


.'. " :`. ,:'.,,..'1,/,',I\

3.2. The powercharacteristics

of the windmill.

By means of the former described way of measuring several

properties of the windmill-pump combination can be defined

but still the powercharacteristics

of the windmill as well as

the pump are completely unknown. In order to investigate this

also torque-measurements must be carried out. In practice

many complications will be met in order to realize a reliable

torque-measuring because the actual torque will fluctuate

considerably due to the irregular drive of the pistonpump.

Knowing the characteristics

of the rotor the conversion of

windpower into useful mechanical power can be calculated

then for any windspeed and B.P,M.-numbers. This charac$eristic

is known as the C, -A curve;Cp is here the power coefficient (the rate of power conversion )while h stands for the tip-speed

ratio. Both dimensionless ratios are defined as follows:

Besides a dimensionless torque-coefficient follows:

is defined as .

The symbols are representing:


= useful converted power


= useful converted torgue

P = density of air


= windspeed


= radius of the rotor


= rotor angular velocity

IN4 Il.2 Kg./m3] [m/set .]
bl [rad/sec]

Both the CT - X curve and the C'J- h curve will represent the main properties of the rotor in a certain range of Reynolds numbers (Re). Since 10% arched steel plate airfoils are applied the curves will be valid if Reynolds-numbers appear which exceed the critical Re-number of lo4 since the applied airfoil properties, whichdetermined the blade design, are just valid under this conditionfR&>fO%By the help of some SWD-members(SWD=Steering Committee for Windenergy in Developing Countries, 1 -the C,-~and&P-Xcurvee of a scale-model (1.5 m. dia.) resulted from v&ndtunuoi-tests. In these tests the influence of the presence of several disturbing parts in the rotorframe has been investigated. Also the presence of the helpvane of the safety-device proved to affect the
G - curve resulting in lower efficiencies. The assumption that this G-Acurve will represent the 5 mm dia. windmill provides a lot of information with regard to the actual gerrerated power and torque. Also the powercharacteristic(s) of the pistonpump can be derived then. It is

4.8 &I!2PU 500




3 4

5 6

? 8

9 w fr 72

&? frod/s) .-

evident that the mechanical losses of power must be taken into account. An acceptable value of a mechanical efficiency will be 0.90. Probably a part of mechanical loss is related to the angular velocity of the system,but is supposed to be constant in a certain range of R.P.M.-numbers ,'J!hese assumptionsledto the approximation of the power-curves of the 5 m. dia. windmill. The cubic influence of the windspeed is clearly visible while the most attractive load-curve should cover the cube-curve for x=2-
4.3. Analysis of the windmill-pistonpump-combination.
Since the R.P.M.-number in relation with the windspeed is known, the load characteristics of the pump(s) can be drawn into the former derived powercurves. Considering this it can be concluded that the transmitted power to the pump depends on the pump-performance and the crank radius. !J!heconsiderable












reduction of the overall-efficiency

can be explained as

follows: for increasing windvelocities the combination ope-

rates at increasing values of the tip speed ratio so the rate

of the windpowerconversion drops. The mechanical efficiency .

and the efficiency of the pistonpump affects the overall-

efficiency as follows:

Since the mechanical efficiency is fixed at 0.9 and the

Cp-value does not reach values below 0.15f0.20

the low total


0.05 for V- &n/sec.)must be caused by a low

efficiency of the pump. This assumption can be proved simply

by considering the power-curve of the pistonpump.

In case of higher angular velocities increase considerably due to friction hydraulic losses.

the losses in power and all kinds of

Thedeterminatedpowercurves the following expression:

can suitably be approached by


P = Power


D = Diameter of piston


R = Crank radius R = Augular velocity

[I [rzd/sec]

9 = Density of water g = Gravity-acceleration H= Elevation head

[U.b31 [m/set*]

. The symbol Co stands for the coefficient of loss. As far as

just the hydraulic losses are concerned the hydraulic is defined by:
#-. BH @ = ~r/~~+rr',~~~z, = f + 'r. @!




_ I,,;,*

4, ,_ -`-.

7 z

.,\_: -'-,,:I`.,,,,f&':eT.,.,;! ; ;':j;
/:.y*.&',2),":$d,''I""&d3Jg;gi,,gg :';;$$ '_.-<,'/,,-:''_;,>-y,:j.-^:;,i,,35I"#,,_>:;,.'5:,,7";;<GA'$<s<;*`;2%>I@1<e$,"a4%*tf&,>J;:

while the total efficiency of the pistonpump is found by:

cp = volumetric efficiency)

In general the total windmill-pistonpump-installation

represented schematically in case of fixed parameters

elevation head,pistondiameter

and crank radius.

C~II be like

From this scheme it can be concluded that the total efficiency

is badly affected by the influence of hydraulic efficiency

since the measured tipspeed ratio (X ) in practice


exceeds the value of 3.0 while CP.Lmaintains reasonable values. From the two last mentioned measuring-data the value for Co

can be derived so an approach 08 powercharacteristics

in case

of different elevationheads and crank.radia will result. For

Co= 200,rather Vessimistic"but

acceptable powercurves can be

drzan through the powercurves of the windmill in order to de-

fine the meetingpoints and the actual angular velocity of the

system. Assuming the volumetric efficiency be 0.8 the input-

output relation is found by the calculated output for all

combinations of elevationhead and crank radia.

Although this determination is not quite valid the results will

provide a rather good impression concerning the output which

can be expected. One can conclude that in case of higher ele-

vation the output is hardly affected,but the starting wind

speed is higher.


, ? 2

5 6 ir 8 9 W V 72

4 3
2 /
3# 5 67
4 3 2


5. Defining optimal starting windspeed, windmill- and Eump dimensions

In the former chapter the "mis-match" of the windmill and single-acting piston pump has been fully described for the 12 PU 500 windmill. As a matter of fact the same applies for the other members of 12 P&family (250, 350 and 700) for their Cp-A curves are the same.

Apart from the problem which windmill type should be applied also the starting windspeed to be chosen is most essential with regard to the way the windmill is going to perform. To illustrate this we consider the following extreme situations which often occur in practice:

Too low starting windspeed:

Too small a pump applied. Windmill is almost always running

(psychologica?_ly attractive).

Poor output: too low quantity

in the long run.

Too high starting windspeed:

Too big a pump applied. Windmills is almost never running


unattractive.) Impressive output, but too

low a quantity in the long run.

Somewhere in between an optimum exists for which the most

optimal starting windspeed occurs for which the best

windmill-pump combination can be composed. To derive this

optimal combination there is a need for a simple and

systematic procedure by which the following can be


- starting windspeed - rotor diameter (250-350-500-700) - pump dimension ( 31*-4vp-5nt-6~M*~)
Whether such optimizing contradicts for instance social acceptance or required storage capacity cannot be foreseen as yet. However, as all components are available

in a finite number of sizes a certain deviation occurs. The latter provides a margin to create a compromise if necessary.

5.1 Deriving the optimal starting windspeed:

The essential point in power conversion by the windmill, com-

bined with the single acting piston pump, is that the highest

efficiency (7) occurs for the optimal (design) windspeedVo .

!The starting wixdspeed % is found slightly less. Further the

useful1 output of the system is almost proportional to the

windspeed and as a matter of fact the total output curve can

be suitably approached as follows:




Of course some deviations occur particularly val Vr-Vo and in case of higher windspeeds
supposed to be neglected.

aboutthe interbut these are



A = rotor area @ = flow (water delivery) V = actual windspeed Vo = optimal (design) windspeed 3f = elevation heaa s = peak-efficiency P = power fi = specific density of air r = specific density of water 9 = gravity acceleration

[ mL+ I m bP Ik/ g RI'1 [kg/m31
[mI s31

P (windmill)

= ~w%PO%~A

p bump)

= rsH*&

After equalization:

for V.VL :

$kMr,u*A = (ipH*&


[d/q _

for V>U:

zh this expression the influence of elevation head, rotor area and windspeeds are quite evident. We are interested in a maximum total water delivery during the time period considered r,,C : (e.q. January or a critical season):

It is noted thaz as far as optimizing is concerned the

expression ~~Vtfh fV SW

should reach a maximum.

However, the $tual windspeed is not presented in con-

tinuous functions although Histograms provide time dura-

tions of windspeeds belonging to certain ranges. There-

fore it is advisable to rewrite the expression in order

to facilitate an interval-wise calculation as follows:

Windspeedvi represents the mean windspeed of a certain interval (e.g. 5-6; V; = 5.5) and t; is the time of which Cndspeeds, belonging to the interval concerned, actually occur during the time span: Note that the expression indicates the extent of useful energy conversion. For convenience it is written in short
as E.

As a matter of fact the optimizing shown by means of this picture:

problem can be suitably d

E =


%ny bori-ng muitiplisations

carried out again and again for

an other starting windspeed x (as a parameter) are needed


to figure out for which value ofI& a maximum is found for the expression:

Such a processcan suitably be computerized,but a systematical calculation is simply carried out in a few minutes only without using any advanced calculation machines. This is done table-wise as follows:





0 f 11

2 I

7' 8

Here the whole numbers of the potential starting windspeeds V, are filled out, while these correspond with each start
value of the intervals given by the histogram. The required length of this column depends on relevant duration of intervals of high windspeeds and whether the windmill still functions as the security mechanism might have switched off the windmill. Generally no relevant contribution to the total addition is found for windspeeds which exceed values of 10 upto 12 m/set.

M--column As mentioned before w represents the mean value of the windspeeds which are represented in the interval concerned.
tL- -column
Time duration of windspeeds represented in the interval are preferably filled out in hours.
Vi*tt -column Values of the twofcregoing columns are multiplied and printed here.
SW*&* -column Here starts a tricky procedure! This column is filled out starting from the bottom adding the value of the foregoing COlUIlIIl.
Print values of the first cclumn in square.

Wltiply the figures of the last two columns to find a maximum arid so the optimal starting windspeed V, as a round figure.

Of course a real optimum is found for a value of Va which

is slightly more or less but as mentioned before the compo-

nents of the windmill installation

are bound to certain

series of fixed dimensions.

Therefore better continue the calculations with the "round

figure" found sofar but knowing where the real optimum exists.

(This saves time and calculations.)


/ 0 / /
1" 0 f

real optimum
s \ \ \ 0 \ \


VO,starting windspeed

The expected total delivery is calculated as follows:

which is only valid if time (6) has been expressed in seconds. In practice winddata are given in hours so far the practical case the foregoing formule turns into:

5.2 Deriving required rotor dia;
Zepending on water requirement and actual elevationhead the minimum rotor area is simply derived by applying the formula:

But in practice only a few windmill designs are at our disposal which means that each choice to be made will be a compromise:

12 PU 250 12 PU 350 12 PU 500 12 PU 700

- rotor area 4.9 9.6

P 2P

One should realize that a generous rotor area always provides some margins which, of course,, is more attractive than a very criticalLy chosen windmill which will hardly cover the need.


5.3 Pump dimensions
So far we are able to extract the optimal starting windspeed and windmill type to be applied under certain circumstances (winds, waterdepth, water requirement). The final match of pump and wi2dmill should prove the demanded properties of starting behaviour and output. To find out the optimal pump dimension the equilibrium is to be considered for

Windmill: Ro= !$

(5 6)


RO= angular velocity for V=V0

= crank radius

= pump dia 7 = volumetric efficiency

PI (0,85) I--

A0= tipspeed ratio @I


R = rotor radius


(5.6) substituted into (5.7):

In paragraph 5.1 was found forv=VO : after equalization of : (5.8) and (5.1)

It will be useful to create a graphical representation of the forementioned equation which is achieved as follows:

Kis proportional to qwhi1eA.d'

and 3 (which are fixed

figures) can be applied as parameters. As well known the

12 PU- windmills have three crank radia(r) so the ratio

leads to:

The graphical representation will be:

More pump types can be involved,so the KG-plane will

contain several overlapping fan-shaped figures representing

the pump size and crank-radius concerned.

Though in principlethe

choice is made for the biggest radius

5 an extra possibility

is available to give an additional

change in the starting windspeed. This is done byre-adjust-

ment of the crank which might be necessary during critical


Next pages show the graphs which are valid for the 12 PU 250-

350 and 500windmillsrespectively.

By means of these graphs

the most optimal pump is chosen and of course, most proba-

bly, a compromise has to be made. However, in many cases an

overcapacity of the windmill will be applied which allows us

to deviate from the round figure of V, , which was found SO far.

Preferably the "match point? is chosen which lies just on

left side of the V, which should be obtained. The reasons for

this are:

- A lower starting windspeed proved to be more attractive

with regard to social acceptance of the windmill.

- As the windmill is running during a higher percentage of

time its delivery will be more regular, which requires a

smaller storage tank in case of regular consumption.

. -5-_.1_0 _ :

: : , .- i _.. ,-


1 .I- i.. .

! .-.-..i& " !

! ,

: .

..--!If [ j

i ..4*j


; :

.--.--3..i._-.-- ,i

_-9_t'. ;*:..i,8;I.:.


-. ~; _

.* .: - .--..-

1 .:- ; ! L.-i. I

.I:..-+i?y.m. ._.ti'__
I_: .-;:.. : A_ :_,

j ! r




. -j.. -. __-._- -


(. I

; .;' /



:: -- - I ---7




- The forces transmitted acting on the moving parts are less

in case of lower starting windspeeds. TO a certain extent

a less quicker wear of such components and a lower probabi-

lity of breakdowns occur.

This might seem contradictional

for the windmill has a

higher degree of running, but such indications have been

noticed in practice.

5.4 Numerical example
In a certain project area the following winds appear during the most critical period of water requirement.

i !OO

Elevation head = 10 m

Water requirement = 1,500 m3

0 12

i. Wvhat is tile optimal starting windspeed (round figure)?

2. Which windmill is applied to cover the water requirement?

3a. Which pump dia should be applied to approach the maximum

output pos'sible?


3b. What will be the starting windspeed?

3c. What is the maximum output to be expected?

34. How many running- hours and -percentage are expected?

4a. Which minimum pump dia is possible to obtain at least

the required output and what will be the starting wind-

speed then?

4b. How many running- hours and -percentage are expected?

* . :
.,-7 . `_7
3. 5 A. :' 5.`; * .G.fl 7I . c_cI Pi.' . `-; `1. i; 10.'. 11 ."i

ti 200
60 20 IO
3 1

2 25
300 350 360 330 130 75 25.5





I 28~ 9

930 16

570 25

240 36



35 64 3.5 81





The optl-ma.1 starting windspeed amo.unts to 4(round figure)

(pump dia = 200) &* =21* - g= 4.58 %

x A-tually


but the graph

VElue of H.

figure reads

i s not represented at the Vo2-scale,
half the value of Vo'for half the

"bough the second match (pump dia explained before it is considered conseo*uences of such a choice.

= 200) is not desirable as

here to illustrate


jc. 5'ecalclJlation

of E by means of proportional



11520 + (14880~11520)(3.42-3)

= 12931.2

q2oG) = 14250 i- (14880-14250)(5-4.58) Obviously the second match is closest

= 14514.6 to the optimum.

%,O)~ "1(20,),

7.2 w 10m2 H USE


7.2 w 10B2 K lg.6 E 14514.6 10

= 1825 m3 = 2048 m3

3d. Total hours of functioning:

(from table)

T1150) = 1+3+10+20+60+80+(4-3.42)100 T(,,L,C. >* = 1+3+!0+20+60+(5-4.38)80

= 232 bfs.

= 127.6


Percentages: P(,50) = gg _?t 100% = 31% P(200) = I;;;" x 100% = 17%

4a. Q = 7.2 7t 1O-2 w + % E = 7.2 A 1O-2 ,J+jb

E =I500

E = 10629 e V,'S 9 (see table)

pump dia = 125 mm (see graphs)

with x2= 8.2 and. x- 2.86 m,'s _


for E:(from table)

'(125) 3 6320 + (1?520-632oj(2.86-2)

= 10792

This is slightly more than was demanded:

QO259) 7.2 x 10 -2 x w

x 10792 = 1523 m3

A 125 mm pump will serve the purpose so the starting

mindspeed of 2.86 m/s is confirmed.

4b. .Running hours: (from table)

T(125) fiunnins

= 1+3+10+20+60+80+100+120~(3-2.86) percentage:

P(125) = *

x 100% = 39%

= 288.4 hrs

Acrobat PDFWriter 3.03 for Windows

Search Any Device: