Field testing of water pumping windmills by CWD

Field testing of water pumping windmills by CWD

Citation for published version (APA):

van Meel, J. J. E. A., Smulders, P. T., Oldenkamp, H., Nat, van der, A. L. J. M., & Lysen, E. H. (1986). Field

testing of water pumping windmills by CWD. (TU Eindhoven. Vakgr. Transportfysica : rapport; Vol. R-816-D).

Technische Hogeschool Eindhoven.

Document status and date:

Published: 01/01/1986

Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be

important differences between the submitted version and the official published version of record. People

interested in the research are advised to contact the author for the final version of the publication, or visit the

DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page

numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:

openaccess@tue.nl

providing details and we will investigate your claim.

I

IJ~iU

EINDHOVEN

FIELD TESTING OF WATER PUMPING WINDMILLS

BY CWO

* Meel, J. v.,*Smulders, P.,*Oldenkamp, H.

*Nat, A. v.d., **Lysen, E.

October 1986

R 816 D

Paper of the 6th European Wind Energy Conference, October 1986, Rome

* TECHNICAL UNIVERSITY EINDHOVEN

Faculty of Physics

Laboratory of Fluid Dynamics and Heat Transfer

WIND ENERGY GROUP

P.O. Box 513, 5600 MB Eindhoven, Netherlands

CONSULTANCY SERVICES

WIND ENERGY

DEVELOPING COUNTRIES

P.O. BOX 85

3800 A8 AMERSFOORT

THE NETHERLANDS

**DHV Consulting Engineers, P.O. Box 85,

3800 AB Amersfoort, Netherlands

I •

'J. J.E. A. 'i.MEEL

*

,Po T _ SMULDERS· ,H.OLDEt'!!'IW!P·, A. L. J. M. v. <:!. NA1· ,E. H. LYSENH

cwo.,

Con.sulting Services Wlnd Energv DeveloPlng L'ountrles,

Amersfoort. lhe Netherlands

(~BSTRACI

rIP-let te-stjng o·f water' DumPlroG

windmills is an important part of CWD's

activities. Measurements are per/or-mea bv

means of Apple I I computers and

thereafter elabQrated by means ot the same computers, using software developed by the Wind Energy Group. lhlS artlcle discusses some aspects related to the hysteresis behaviour of windml 1 I,;:, driving pi ston pumps.

The behaVIour of a wlndmlll in the hysteresis region depe~s on the history of the wind speed. rherefore the average curves measured according to the

generally accepted lEA recommendations depend on the freQuency distributIon ana time history of the wind speed during the measurements. fhe IEI~ pract\ces were

concei veri to pr"ovi de a m2::\ns of mea sur 1 ng output CJJr"'ves \1) a repr Od14Clbie way. The

underlying as.umption is that an output

curve is a cha!'acterislJc of a wind

machine, valid at any slte, ln any wlnd regime. As indicated above, this is not the case for windmills driVIng piston pumps, or for that matter for any WInd machine, having hysteresis behaviour. Two examples are presented of output curves, measured for one machine at one site under different wind cond1tions. Con-siderable differences were found. Annual water outputs calculated on the basis of the two curves can differ as much as a factor of two!

Procedures for output predIction of water pumping windmills will have to involve three steps instead of two. The procedure must start by determining an output curve, including hysteresis effect, which only depends on character-istics of the wind machine. Subsequently the curve must be conver-ted into a simple but si te speci f i c curve by means of . probabilities derived from the site's wind speed frequency distribution. The third step is the conventlonal multi-plication and integration of output curve

*

J.J.E.A.v.Meel~ P.T.Smulders~

H.Oldenkamp. A.L.J.M.v.d.Nat

TechnicaJ UnIversity EJndhDv~n.

The Netherla"ds .

It" E.lLl YSE'll

DHV, Consult-lng EnnlneE'rs,

Amero sf oor' t. "I he Net her 1 ands

and frequency distribution. A slmple theoretical model has been developed including these three steps. Procedures for fIeld measurements based on thls concept will need to be developed. List of symbols

A rotor area

C.: energy production coefficient Cp power coefficient

C~ torque coefficlent D diameter

E ene..-gy

~ Weibull shape factor

N P p p q T v v number power probability probability of running in hysteresis region

idem. including wind speed variations within 10 minutes pumping rate

time wind speed

relative wind speed in hysteresis region ( - ) 1m) ( J ) ( -) i - ) (W) (-) (-) (lIs) (s) (m/s) ( - ) (V-V.tQP)/(Vs~.rt-VstQP) 1) efficiency

~ tip speed ratio

p air density Subscripts

( - ) (--)

0.2 19/m3)

d design, i.e. the point for which C_p1) reaches its ma~imum

max maximum

1. INTRODUCTION

CWD (Consultancy Services Wind Energy Developing Countries) is an organization initialized and funded by the Netherlands' Ministry of Development Cooperation. It aims to help governments, institutes and private parties in the Third World in their efforts to use wlnd energy and in general to promote the interest for wind energy In developing count.ries. The emphasis of the activlt.ies of CWO is Oil w"ter pumping windmills, coupled to single "ct.ing plston pumps. PartIcipants of CWD are DliV Consulting f=ngineer's (Ame"sfaor t), Eindhoven

Un; vers; t,' "t lechnol cogy. lwente Universitv nt lechnolGgy and ILRI,

Instit.ute of i_alld ReclamatIon and

Improvement (Wagernngell) ..

CWD d.,slgned ",,,,ler pumpers are In

Op~,.atl0r. 1'; Sr 1 Lant..a, Paklstar.,

lanzal",i 3, Mo.:dmblque, Sudan, Mauretanla~

: =" ,-"':.' !·t:...:, . ::\, r J. ' . ..., t -,?: . :;;'. ~ .1.;1:1..:" ~ I •.i -:.' : '.' nL .l 1., '.16

:if--',.' a.:\ -,,,lfi ... ,l!:

"

t

f,r O' :'\\-i:"l""~ J ~ [i.r.> l r· ! I ' :-1-....• hie 1'1'..::~ ·3!· G '.=. ':-,'r' t i. f::0 7 j lESt.; It :.) ', i t,.;~ tlf:" ~\ •• ; ,· l i~ ~ .... l ",~lr... ;i-;,. \ r vH; 1', ;1 ';; t'~sr':' ~p:.::c.. ," ~ i r :H lover" I (/ r' i e .':" El i ,ve e r, I.

t.

:

::.'=2 t= .', lv·\.·· ~ \... II.l"F:'t S~~ \...") ;:tp~ Aimet'"·~ ( I n e l i -! I-:'I f !" t-.. :.? ;'(Jl de,-s j ("l tl"l e- TGrnIPr' l')'~_'''' /' ''':) I ·.··er:: i-:WTJ t:E~ f' j. ; RJ r~-= !~,,;, ~ , ' : ; .' ':' 1- r: .'.t­ nur-poses~

ThE' A!O!£'i-P t e ' i t t i e ' d i',';!> l C !, r,:,­

71_,r.I, t-.- I.. .•• tf: ~-, tl3'(' ol ltr".t 1:. E'r 11.lr·tI.,..;r,'p {II , .... 1 f"'€' ltlr-i l '_ ",:,

..;::I-r.rj endurance' tests (t.o (' n o w f",hc

rna i ntenance and r epa.l r r :::> '11.11 I ~'r"'lr' ;; :-:' ) ;

nnly fully de~€laped Wlf,dmi ll s ~rp I.p~ti

there. See flgL~re 1.

Thr-~p. cammerci a l l v a V51 j e:.,t-. l '? Wl noml 1 ~ ~

have b.-en t2ste<i; GASJ" ( 2 m . ; 11 «mlic <".

manufactured In France ) t FlASflI t.~ _n••

r~rqentina), SOI.lthern r:ro~s ~5 .:~ m. .

''lustral i 2.) . Annther serl F.S o f tes t s 0:-'

tbr-ee commerc: i a I wi "dm:;, 1 1',:-; s't cP-teu

recently : BOWJON ('2 . :·6 tTl" (JrT '/ tnq ~rl ai.r

t:" ___ ,mpreossor , ar"l1 d l r 1 j f-t [""o LlnrP . IJSf'lJ, i UDING (1 . eJ I) m. ) W.. l:iEr"nli-illvJ ~ C:\q rj

r ~t.!.rtPS'j ER (q. . :~. H I . . u~,:., ~ . .'.~ i .-.;-. :. L \.III, _" "

( .o:.~ fT!. t "'. ·35 inst.dlled Fe)1' L"JJ',rjl lf ~,-: .=L" ~-.I~(',

\ h e V!:, .i1?~£DY~gjl t E'S t ~ i i-J ! r1 .i. 'S rl,':.. ,,- , -+UI1r::t: 10"a ] t es t ':;, at 'r OI It-r-.:.1j- pCtI' ·'nr'rn.::lnce mpaSi,r·pmF.-"ts o n L), ot c.-' t ,T ;"~. ,.:l ev , o pec t.:"'y'

1-r·l e WIJ,dmll l GrOLiC'", of ;, ~I. Ine C WO SUI)I)

{t"he r.tunoE" " ,"'e+G-rs T.;:' r~? :' c.t ::.- d12.rTI(?'t r r

In mm) h?s '-I een t ec;;;ced t l-,ere ;I ~':;' V er S IGI:

wi th 2 deepwE: l.i pt'mU i:oi' O O'IE.' W', tt· a t 0W

l i f t / hJgt"; '/ol'J me pum;-. f=-t I,,\ew; '/ d es::. q rie" :..,

m .. c11amc.. t et oro tot vpE? I : " belng ~ E-5~1_'(j.

The Ei fldhove,Q tt;::st-flelf~ ((:.J9ur-e '~~ ) .c::::

meant f 0, :

- fUnL: t . l o n a j tE'stll ' g C'it protG::: '-.' C"!?=,

- performance t J?st.: j ; ';~! Df G r ;,,·;; t, :-)1 ', 1=-1"0:'5

- special mE.'aSUr p.mRli t '::. ( 1 11,' ,,:,·. I ,L·"~ r;:, ' ,1.

measurements. liI F''::''=d lrer.H?r. t ~· r;f '~tt·eS·';I ,

and forcC?s)

- f l l::l d resr1 r:li nf Irll li',vatI Vp ('Of"\( I': I ,r~

- dev elopmp.nl.:. ~"rJ t p . s t l f l g r,)F ml11111 r lr lr: ' 1 systems. ~nemomet:el' t e s t s , . , tI, 07'l'' f." ~ 1:.-'''::0 I '" 1' 1'. • ( I "rr, : r t t t, ; ) ;, I ;, t :; 1 1 ~-, :-., __ I l l , al . ~ ' .' .JG t. I\(;or I , " , _, t, ....;, /. ~'J (WD 5QOOLW 11:!EU

i

1-3

.

.

..

FJQ. ~ lest sl te at the Technical

L1nlver s l t v EIndhoven

',., ~II)~ Lir·t l c l e special atter"'tlon ;0;-, D··.:;l

t 1 a~:1 t1 1 .lt j:')erf-nrmance measurement s +-'/1­

:ne-c:--'C\r"'l.-: :i1 \<'Ja t er- pLHT,p Jn q Wlnc1ml l l s , 1-1

~~i) ,="f I f i t' prnt.l @m 1,.0,11)1 he f1resented, ~..r~r,l ', t ,'f:: I'r',=,,:,pr'il E! nt .3 1a rQe 11Y ':.:Jt:prE:-s:~.

I",;n J I I t"h~ o;1".dt t/o::.l o p reg Ion of tr.e

Pllrrqlt {I ;( ve. '1t'lt:- C"Dr.Sf?QuerICf?'S For ell.; !..I-' ­

r-~Pr1 1 t.!,lrJns w11~ tH? showr:.

~. S "'''h !lNC, i,ND SlOF'F' Ir,G; HYS1ERESl ; e.u'!W lllUf, OF \·H NDM ILl S DR I V 1NG PI , 'H 'IJ f·I .I~IF::;

,..,r.orhanl r. C'~ WIndmIlls dr'i vi nq Ol =tnn

l)l.dll['1S hc-:;ve a .athe:or peculiar start ~ Cl

b~h~v ini. lrt I-o,jhlCh 15 qLllte d I f f e r e n t " r';·1.

wJ nd r:;.lec trir: ~ler lerators.

Ihe c:.. l-'E?C IAt behav I DLtr' 15 due tc- 1' 11(:­ c t-tar a cL er l S l l L : u f the l oad.

A P l .... ton pLlmD requires a t o rque \lJh .. c.t, I '='

1" -f ,l r c::::t apprC'l;~ lrTlallOn Indep endent c·.. i-IIE

-::. t::' f:-'d r ,1 rJDerat 1 Ol) ~ 1he torque is rJ t' e::-r ml ned hy t, lle strC1ke 1 ength anrt / .. ...

the w a t p.r ['1r e -l3 '""·l\r p. Qrl the pi stan . at"':~

arol 'v' "~L' ~E-~c<5E'~ sl l a h t !y at hJ Qh SOE'eo~

ri l l£:-' to t:'lr e.-:c:;.Ir£' losses Jfl the valvE':' '-~;-Id

"I

Des.

M( ,r(""Iclver . tilE? t-urqllE? IS CVCllC: durlllC1

tI ,E' upwar d ~h"ol.e the !J .i ston ho:\s t o J~ft

tIlE' ....;at. e r , r P'U(ll r inq a l arge tGrouE',

dt'r j "0 r),£l rl LH'Jll,,~ar'd stroke the t orQL1e j S

or cH: :-lt+d l l ~ Z(?r t;J. Ille olaximum torQl le

durir'l) In e CYL 1E' IS n times the avp r ;:(oEi

tnr 0~ tr~

~\t ,-'rt'Jlln +rorrl str?lnd s t I l l l r l ~I_.l~h r. s r t ".·\llC·'"\n !5 llL· ll.:.e d l .ff)cult -for' a \".11; ,0

11Ic~C"1 1 ne:

rl' r.,.-d fJr- t o st :l.r·t the load, ~ hlqh t(~I-t.ll~P

i,'"". r p(1U l f.-d. 1ft order to i l t t the P1S on

f-pr the f i r s t t i m!? , thE:' max imLlfTI torqLi. € I S

11~Eded [IT tlm~5 the average).

rhe wi , Mnll11 b: ade~ a re In s t a l l , and th~

t or' Ot'€? av~) lab) e t rom the "nndmIl 1 l ":'~ f"" C?l atj ·.Ie J ..; 10"'.1.

fhi s mtll es t ll,':It ti,E" Wll\U sneed at if.-Ih l '~

thE' YJitldndll star-ts f r OIT! s tand ~llll IS

rel~tJvF'lv hlg~l: V~t~~t.

[Inc r Uflr,) r.o . the ~lt L; :?t l l)P e-c c.,r.e -:;

mll(" h r'flUI' E: f .,YOIII' aU 1 e.

f I le l 'Ir~lc c~'a... r ter 0 4 the !,)aa 11.:\5 "·0

1' ," I l.IC?rt -e an\-' d,or c riltP t o th e iar:-;e ti,e:' t l 3 C'Jt t b ,=:- ""' lfldnil J: rot o r . lli.l " r-; ,f?

:3 '.ti2r.£u tol" qtlE.' .: S iJeer1eU .

I! .f.. I;J l II M rlll 1 j l aOE>:5 ,.,.t o r L, 1 n l ~;el r ~ l "'r 11;2 !

tt~,e ava;jable torqUf;: 1£:.. re::>-ioatlv'C:·:.-.. ' t>}qr-'~ fbi '7 mat_e~ tha.t -!~ Wlf,dqq 1: ~ :.:.-w.c..e r~yr-tt1f1'::i

stocs at 3 r~[at~veJ\' !nw wlrlM s~ep~=

V ~ 'I( ~\ p.

Due tC'· tt'l;~ st:lrt'~·t";:1r b~-t;~V·'C.I .• r'" onp

w:;.11 fIl"u:1 i:\ hvstPTE'%iS :t1C!!:.l In thE~ outotit

c~rve as lndl~ated In flnures 3 ~no 4.

v

p/sl •

Fig. 3 Hysteresis loop due to start! stop behaviour in output curve

Fig. 4 Hysteresis loop In Cr~ curvp Figure 3 presents an output p';;::formance curve as pumping rate versus wind speed, which is, of course, equivalent to a power curve, i.e. hydraulic power versus wind speed. Figure 4 presents the same information in a format which is more convenient for comparison of different machines: Cp~-V curve, the overall power coefficient as a function of wind speed.

A detailed analysis of start and stop wind speeds for different types of

windmills was presented in May 1986 at a conference in London, U.K., see reference 3. Measured and calculated values for

Vstop and VSt4~t proved to be in

reasonable agreement. Here some more emphasis will be put on the curves as a whole, and hence on output.

As a referencE!' for measurements afld for general output predIctions, i t is useful to have iii general expression for

the output curve, or - which is equivalent - for the Cp1! c .... rve. The

over.d 1 shape of the curves can be der i ved rather easi I y, when mal, i ng some assumptions (see also ref. 1).

A central role In the derIvation is pl.ayed by the value of VarIOUS Ql,antit,es at the deSIgn wind ~peed V_{d ,} I.e. the

WInd speed for WhiCh ttl", over.1l J power

coeffiCient Cp~ r~aches its maWlmum, see

also figure 4. Jhev are Indlcat~d bv an

1 rtr:iE'o.{ d,.

It ,">".:.';.~\AIi?r!! t,,=.t l;";,; Wll'C,ll,· . . .

;:1.:£<,rt~ct.I?" l::er;-• ..-.. ". ~ :lflp.ar tt,rn;'~""'':'t\J'.j.:,-~ rE?i,-,tJC)I''!'.~li' I t 2· once ru,'.r, ... \g. t t',e r- eq 10'1 Q t st;\ t 1 ) :

The pump i!S assltmed to demano a C or.sta,""".t

torque for .. 11 speeds, H'cll!ding the desu'In wind speed. therefore the "' .... 'd"1111

Will dell ver a constant torqlle at: a.;

pOInts 8+ operatIon:

L q 2

Old • d

From these two equatloos one flnos " as a function of V. Substituting the result into the .f" st eQuation, and uSIng the fundamental relatIonship Cp .. >- C ... one finds an expressIon for Cp illS a function

af V: C"

_"'<,I

~ ) ' . C K Vd 2. )..11 { 1 - - - ( 1 - - - ) } (3) C f,J2 >-11 V2 ~.«}( p • • "

Assuming a constant efficiency for the pump. ",xpression 3 also represents the shape of the Cp~ curVe (see fig. 4). Df course, i t can easily be converted Into a P(V)curve. For a windmill-piston pump system the point at which Cp~ is maXImum is sharply defined, since the locus of

maximum power pOlnts in the torque-speed

characteristic of ill windmllI IS a secona order curve, whereas the torque

characteristic of a piston pump is basically a horizontal line. The pOint of intersection corresponds to the design wind speed _{V d •}

ExpressiDn 3 d.scribes the upper br"anch of the hystereSiS loop (see fIgures 3 and 4). It is si Ifti 1 ar f or a wide range of mechanical water pumping windmills. Only the values Df start and stop wind speeds are different for different types of windmills.

3. FIELD MEASUREMENTS OF OUTPUT CURVES

WITH HYSTERESIS LOOP

In this section some problems of field . . asurements are presented which are caused by the hysterasis effect. This is done on the basi s 0+ _ _ _ nts performed at the Almer. test facility for the Argentina manufactured FIASA wind-mill.

The FIASA is a typical exaMple of a classical "American" windmill. The rotor is 3.0b m. in diameter (about 10') and

has 18 blade~. It is 'back geared

(1:3.29). During the test it was driving

a piston pump (diameter .101 m. and stroke .243 m. This configuration was specified by the manufacturer. This corresponds to a des1gn wind speed Va of

2.5 m/s.

A complete report of the tests and measurements is avallable, see reference 4. Some of the results were already D'Jb 1 1 shed at a <: on fer' enee in London.

rt-::'ff?renc:e 3 ..

Complete and relIable measurements were perf ar'med in the per j od October. November

1985. The measurements "'ere performed

ar<:or~<ng t.o the reco~endations of the

·H.'.",IJrefflE':-,to:;::- W.3S .30r:,r.ll~1'1 ~::'cept tC)r"

SF,1 f.?oct, .Jr. C)~· ... nd d 1 t pr L; 0""0;;;: ~S t-~­ ~ommende;j hy [EA.

mRa~ur~m~nts according to thp IEH recommendatlons, the results Jr. thE:<

hysteresiS r~gion will be some ,;,veraQf" c"

the upper and low2r branch of the

hysterests loop (see figure 4). Sometlmp.s the windmill is running (upper branch).

and someti mes i t is st andi ng stJ 11 (ll1wer

branch). For this region one expects to

find average ",,,lues below the theoretlud

curve, as well as a considerable spread

in the measurements: a large standard

deviation. ThiS was indeed the case as

can be seen in figure 5.

-

_I

_l'

-Fig. Sa Wind speed frequency distrIbutIon

Average wind speed 5.0 mls

..

,.

•

Fig. Sb V ~Isl • Cp1) curve Crosses: measurements

Drawn curve: theory (3)

Dotted curve. correction (5)

Fig. S Results of measurements of FIASA

windmill October-November 1985,

4143 measurements (ref. 4)

Figure Sa shows the frequency

distribution of the wind speed durIng the measuring period, which will prove to be an important reference.

Figure 5b shows the measured Cp1/ curve as

crosses. The centre of a cross represents

an average value, the length of the crOSSes corresponds to the standard deviation. The drawn curve corresponds to

the theoretic'",l Cp1! relationship derived

above (equation 3). The dotted Curve

corresponds to a correction which w,11 be presented in section 4 of this artIcle

(eQuation 5). In the hYsteresIs reglon

between V.~~ and Vsta~t 12 and 4 OI/S

respectively~

one sees trlat the measured

values are far lower than predicted bv

the SImple theoretical farmula (3). As

e}(pectE?d. a 1 an]e standard devI at HJn 1 S

",ouod In this regIon .. Below 2 m/s. one

finds some low. but non zero values for

C.,1!. This is ctl"E? to wind sp.,ed varlat,ons

within 10 minutes. Even in a 1.5 ""5

'

..

: wtnd =·~·.:;·f,>f_' ~~, .. ~;q

Bpt Wi:.~En ·4 =,1"'(1

LOlnC}\1~ \"E'f'" , .,...:edj ¥out;, the tr,~rlr.r> ~~~ blqt'. ~Jr~;;j f>o("fefis thE> 'ftt?~sLire,.d Y·';"!lH?""" =tri-lower dUE' tt.:l the s~ff-'tv '5-"l~'fpm W~,\(- .. : limlts the n\,tf'tttt~

Ctlt,m..t.,t.

Sl.!' "y_~? ~ a W1 ndftll j ! ~·C'1, fir·

svstem mav operate elther 111. t h " IIPOE?"'.

or in thE' lower- branch of the Cp~ cu..-v",

r+'QUre 4). What e)<actly happens dI?D"'''OS

on the h,story, Once the wlndmill lOS

runrH no i t wi 1 1 cont i nue doi n9 sf' wh.,." , t

enters-the hysteresis reg10n. Once

standlno s t i l l . i t will remain standlnq

s t i l l whE?n entering the hystereslS

replon. Therefore, the probabilIty of

ejther situation depends on the .. ind

speed dlstribution. In a period of stronQ

wlnds the windmlll will b" rI.lnnlng mest

of the timE? and the system will follow

more often the upper branch of the hvsteresis loop than the lower one. In '" period of weak winds thE? opPO!!Hte is expected.

In order to verifv thiS, two different

data series were chosen from the total

data base on which figure 5 was based.

One series was chosen so as to have mainly high wind speeds. Another one was chosen so as to have mostly low wlnd speeds. For these two series, graphs were made of frequency distribution, and power

coeffiCient, see figures 0 and 7. It 15

to be notE?d that the full data series wera used. no selection of data was

practised, except for selection of w,no

direction as recommended by lEA.

Fig. I:; Results of measurements of FIASA

wtndmill November 1 until November 4. 1985. 239 meas;ure·· ments. fh9h ."nd speeds prev.ul,

averace wind spe~d 4.6 m/s

Drawl" curve, tt'>eoretlcal C r " curve ~ rormLt~.a ~ .~.)

Uotten ~Llr~2~ correctIon

ltnTh

40 N 30

20~

.

iJ .

Vlm/SI:. 1 0 -

--C=t=

°0 1 2 3 4 5 (, 30 Cp~ ('Yol 20

1

10 V~ 0₀ 2 3 4 6

Fig. 7 Resu.lts of meaSGrernenT:S of FI{-\S;.\

wlnamili [,lctober 17 untIl October' 21. 1985, 194 me-asur-",···

ments. Low wind speeds prevail

averagp wind speed 2.4 m/s

Drawn CllrVe: theoret,,:al Cpll

curve (3)

Dotted curve: correct1on according to (0)

Indeed, the two curves of f'gure 6

and 7 are quite different. For the fIrst series, having mostlv high wind speeds the measured values approximate closelv

th~ simple theoretical relationship (3).

It i s concluded that, follOWIng the

recommendations of lEA, completely dif-ferent Cp1) curves can be obtalned for

one and the same wind jf.achine. This procedurE!' of measuring output curves does not yield reproducible results If the output curve has a hysteresis loop!

In order to judge how seriously thIS affects output predictions, yearly

average outputs were calculated, both on the basis of the measurements of figure 6

and of fIgure 7. Especially for low average wind speeds, in which water pumping windmills are often applied, very large differences were found: 501. to

1001., see reference 3.

3.3 Distribution of observations within !!!...!lin

Because of the large standard

deviations of the CI'1) measurements In the hystereSis region, i t seems of interest to study in more detail which values have

occurred, and with what frequency. Figure B shows two distributions of observations in tvpical bins: one WithIn the hysterE!'SlS regIon (2.5 to :.<. mil,.). "n" one outSIde thli' hvstE'reSlS reglon ('5.5 to

6 mts). The fIgure was derIved from th~

same series of mea5urements a.s figure 5. The difference between the two bins IS quite clear. Inside the hysteresi!> region one finds a wlde variety of values, ranging from ~ to 401.. OutSIde the hysterE?sis reqion one finds a

distribution, which is nicelv centered arolmd one wei 1 deH ned val ue.

FigGre 9 shows th~ l"fluenc~ of thE? Wind

regime or; the DIstributIon of C~1l v<"leles

iI-,S] de the hvstereS1S regIon ~ag:alrl f:::Jr

the 2_5 to 3 nils bin). FIgure 9a r~fers

to ~ per·lod of hlqrl wind spe::eds (sdtne·

period as fIqure 6., figure 9b to a

per10d Df 1.::<w !o'JlnrJs {~am£'l as flgure I J

Bm 25< V< 301M 30

IN

Fig. 8a Bin (Jf 2.5 to :.>.0 m/s 10 _{IIi!! ~,~V<tR.I.I} 60

I"

!il 40 30 lit

lli::1.~

10

I

0 0 2 3 4 5 6 7 8 9 10 11 12 13 Fig. 8b Bin of 5.5 to 6.0 mts

Fig. S Distrlbutlon of measured

e,n

value_ w,thin one bin

Measurements of FIASA windmi; J

October-November 19B5

o

Fig. 96 High Wind speeds, same perIod

as -fig. b 6 Bin; 2,5 < V < 10 ",,,

!

1N

~[

(p~ 1%1

_..

n

°0

10 20 3D 100

Fig. 9b Low wind speeds, same period

as fig. 7

Fl.g. 9 Distribution of measured Cp~

values within a bin in the

hys~eresis region (2.5 to

3.0 II\/s)

!il

50

Basic: .. Jly, the two figures show again the same range of values, but the frequencies have shifted. In the period of low Wind speeds (figure 9a) one finds only Cp~

values close to ~ero. For high wind speed_, one fInds higher values.

These results agaln confirm to the

remar~~ made earlier concerning the

probability of pumping in the hvsteresls region. It is to be noted that one does not only find Cr~ values corresponding to the upper and lower branch of the

hysteresiS loop (see figure 4). but one fjnds Instead a whole range of

intermediate values. This I S due to variat,ons of the wind within the 10

minutes averaolng perIod used for the measur ements: rhe WI ndmi Il rna" run for a

tew mi nutes {l~DOer t'r-aflL.:.!-,} done:' ;:.tar.o st,Ll~ dl\rlr,q the r·~tn~if'dEr Dr t:-.E.<> lit' mlnutes perlod .. resuJtlnq )0 SCJITH?

i~termediate 10 mlnu~e~ avera08E

From figLires Band 9 It :;'8V t"le

concluded that Jt does not maL"e ~urh

sense to take ~ simcle &rlthmatlc meAn D~

all observations In a bI,) wIthlf' the

hysteresis region. DOlng this, would not result ill a unique Cpt)-V relatl0nship. If high wind sp6'eds prevail, the frequency distribution of observations is distorted in favour of high Cpt) values, and a hlgh av ... ag.e value would be found. If low wintl speeds prevail one would find a low average value.

4. CONSEQUENCES FOR OUTPUT PREDICTION The total energy OLttput of a wi nd machine over a longer perIod of time depends both on characterlstics of the wind machine, and the site where it is installed. It is usual to separate the wind machine's characteristics and the site characteristics in the following way:

- Output curve of the wind machine, the relationship between power output of the machine and wind speed. T~is is assumed to be a unique characterlstlC of a given wind machine (with 10aO). independent of site characteristlcs, universally applicable.

Wind speed frequency distribution, summarizing information on the wind regime of a certain site.

The lEA recommendations for output performance testing <reference 2) are based on this concept. They describe how to measUre an output curve during a relatively short period of time. Total output at a certain site is to be calculated by "multiplying" the output curve, and the frequency distributlon. i.e. multiplying corresponding points and integrating the result.

For long i t has been attempted to conceive of output prediction models for water pumping windmills along the same lines. However, problems occurred with respect to the calculation of an

unambiguous probability of pumping in the hysteresis region, which was needed to determine a unique output curve in thIS region.

On the basis of the experience with output measurements presented above, the solution to the problem has become quife obvious: one must leave the concept of a unique output curve, which would be generally applicable for any site. Instead, a three step procedure will be required for the calculation of total output of a water pumping windmill:

An output curve is determined including the two branches of the hysteresis loop. A theoretical model to do this can be rather simple (see section 1 of this paper). A measuring procedure will be more complicated and is being developed.

- Using the actual wlnd speed fr~Quency distrlbution of a certain s i t . one calculates the probabIlity

0'

DUIDOlng in the hysteresis region and corrects the output to fInd a curve WhICh IS valid for this partlcular site onl ....

;:J rcb::lD 1 I l !:.~ (.) ~ .:.;. t: !'~.o , II 9 ... f"I t : I t ..

la(nor,. :_!:~ce 1:, thE' hvFt:~"·ec:.. 'I' ~"r::l ::,'C

tnl_,~t g.:::< baci. l~-. tlme ut"ltll f i i d l .... 1 speed cl;t~d\le the hYstereSIS ""'I?'YJO,';. J~

ttn s "no rid speed happens to be ~ll c.t"',e,.. t ... :: r, V::'t.j;!"'t~ thewindml111s-("un(l'll"\g~1~ tt",::=

tIme; 1f i t happens to be less that

" s t o p ' theN1ndmili is standll'9 sU;,.

Writln9~tt-,e probab.lity of a wind ;;10'''''0

greater than VstGrt as P(V>Vs~art'~ an~

wrItIng the probabilitv ot a WInd spe~d Ol.tsIde of the hysteresis reglnn as P(V>V •• o~t) • PCV<Vstopl, one may e"prps!'> the orobabi 11 tv of the wlndollil rur,nl "0

as follnwsL

P(V>V~t""t)

p '" (4 )

The probabilities P can be slmplv calculated from measured wind speed frequency distributions or estimatE"d from assumed Weibull distributions.

The probabiltty expressed by (4) 1S constant throughout the hystereSiS region. Applying this factor as a correction, would result in a rathE"r strange output curve, with a steep step

at V~tart' Also, a constant probabilIty

is not verv realistic.

The probabilities defined above refer to

10 minutes average wind speeds, whereas a windmill will react on a much shcrter time scale: i t may start or stop due to a gust or a lull of a few seconds onlv. For a 10 minutes average wind speed just below Vstart the probability of running must be practically unity; within the 10

minutes one will soon observe a gust above V$t~r~t the windmill starts and keeps on running. For a wind speed just above V_stop t~e probability of r~nnlng

will be practIcally zero fer a 51m1lar reasen.

Analyzing this process would be very complicated. For the time being, as a first guess, siMple linear relationships are assumed. At V!,~op a p,:"ob&bility of zero is assumed. In the .lddle between

V$top and Vst«rt a probability eq~al to p (see above). at V.i«rt a probabil1ty of

t. In between the probability is assumed to vary linearly with Mind speed,

summarizing. y<0 1

0<v<-:2 1

v=-"

--<

v<

1

"

_v>l p'=0 with v

=

p''''2vp p'=p (51 p'=l

This corrp-rtion has been Indicated with dotted.l,nes in figures 5,6, and 7. The probabilIties were derived from the measured distrlbution. It corresponds reasonably well to the measurements.

~h<F!' r:~_

.. ,

Cllt-\'P.~: "'.t w; t-~. t.!,~~ ~,-;,rr C-::-t l.l)r

3'cCGTChnQ tC. ~c:;! rm.:i'E.t be ltlt€2C.'~:~,tE·O (\ to?'r

mu!tlpl·i~at.irJn t~lt~ ~ Welbl.d ~ .

dlstrlhutlO .... The results C~tl t.p.

presented In oS. gef1era.~ .format bv aE-f~Plfl9

an energy orOdllctloro coefflClent. wr.1Ch

I S the ratio of enprgv oreduced and ~

refe,..,,,,nce energy:

E

This coefficient Will be a functIon of

the ratio of the desIgn wind speed of the windmill and the average wind speed at the 51 te of install ati on: V d IV.

Figure 10 shows some results.

2J

Fig. 10 Energy production coefficl'ent as

a function of design wind speed average wind speed

Drawn curves: theoretical model presented here

Dots: results oi measurements, see table below

Of course, important~oaraiiieter$ 'are start and stop wind speeds. Figure 10 shows as an upper limit the integration of the complete C~~ curve, i.e. assuming that the windmill is always running in the hysteresis region. Three more curves are indicated which aria' belleved to be

typical for three classes of windmills: c1.assical "American" windmills with and without balancing of the pump rod weight, and windmills having a starting nozzle in the pump and a balanced pump rod (like the CWO designs). For the start and stop wind speeds, valula's were assumed as indicatla'd in figure 10. These values were derived from earlier work, see rla'ference

3.

Some more assumptions, of minor

importance werla' nla'cessarv for calculating the graphs. The shape factor of the Wei bull distribution was taken to be 2, a usual value. The parameter A.~x/~d' see expression<:::;;) ~,as taken to be 1.8, a usual value (except for _{V<V d ,} where a

value of 2 was taken, since otherwise the maximum of CI''I? would not occur at V:J;). It

was assumed that the windmills safety system limits the output to a CDn5tant value inr wind speeds above three times the design wind speeo (} .e. V,,~~"d = :;;:

V~). and that it shuts down the windmill

c~mpletely above SIX times the design

WInd speed IV •• _t

=

6 Vd ' . yielding 3

coo-rection of not more than 10,. 0+ Cr:'

For the .. undnH.l with start.n" nozzle tohe

C .... 1) relatIonship was corrected fDr lo£sp.s

t~rough the nczzle~ according to

reSL,:t~· ['"I} tllc-dS',{rC'oTi£'J'" i.;>. 1~~ ... <7.' • :$Ld''j·;ti'r .. .?ec;· thesE' mt?C\:~t~lrements ?r.r:

indIcates the r.fer~r.ce5.

Tbe dots represent tMeo aVE'rage ... · .. d _.£?~ dllrlng the .. "hole mEle-surlnt;; oerloC', r:.f' tests of Southe,..~ Cross. Ftas-£lI:. ar.ct :!;3':;':' £;

were peo-formeo at the Almer .. test ~"'c.

with ve~v hiah average Wind soeeds. Therefor£> the corre5Pondlng pOInts "n:;

found In the left part of the grap;' .. 1he

te~ts of th£> WEU 1/3 and the CWO 200~

(both WIndmills WIth a starting n04~!e "'''0 a rei ati vel v hi gh V ~) ... ere per. orlt.",d

at the t,elilt 51 te in Ei ndhoven WI th rather

low average wind speeds, yielding po,nt~

more to the rIght In the graph. The table below summarizes some

infoo-mation of the measurements, and the

assumptions made.

Table 1. Measurements indicated as aots in figure 10. v ( - ) (-I (m!!!!) (1II/s) (-) Fiasa 0.35 8.B0 2.'5 5.0 0.37 4 So Cross 0.35 0. Bill 2.B 5.5 21.42 :; Oasis 111.35 111.60 1.9 4.1 0.24 6 WEU 113 0.35 0.60 4.1 3.7 0.86 l CWO 2000 0.35 0.35 4.6 3.3 0.B2 7

*

For the mechanical efficiency a value

of IIl.BIIl was taken fer high head pumps,

and Ill. 60 forI ow head pUIllpIl.

For the

cwn

2000 a value of 0.35 was taken because of friction in this flrst prototype, to Which

th . . . .

surements refer. Later i t was illlproved

considerabl y.

5.

CONCLUSIONS

In a general manner i t may be concluded that systematiC field measure-ments contribute significantly tD the understanding of the performance of wind machim.s.

More especifically field measurements of output p ... formance have produced a much bett . . understanding of the

importance of the hysteresis behaviour in the start/stop region on the total output of water pumping windmills.

It was found that field measurements performed according to the lEA re-commendations do not result in reproducible output curves for water pumping windmills having a pronounced hvsteresis behaviour. The differences ln m';"asured output curves may lead to differences In calculated total output as large as a factor two, especially for low

ave~aQe wind speeds, in which water

purnpi~g windmill~ are often applied.

On the basis of the experience with fleld measurements, a new procedure for output predictions i s proposed, involving three stpps:

DetermInation of the output cUrve Including hvsteo-esis 1 oC'f> , depending only upon characterIstics of the wl~d

machine.

- Conversion of this hysteresis output curve into a slmple (51te specific) output curve bv means of probabllit1e ..

derlVed fr-om t~e- .. and SO>?E?;j 'rt?]' IH" . 01 $,tri but 1 (lfJ ::O,t tr\E:2 "5) te' c.t

1 r"d;.t. a 1 ~ ~t.l 0[1.

Ml~jt.lpjlc.atlo;1 ::inC1 lr,taQl"ati ..

,r ...

H t t l t : 51 te :spec} f 1 C Dutpt.t C .. ,I" ',,'e !'\';Jj f f'ft. ..

sIte's wInd scee~ frRat~encv

~istrlblJtl0n.

A slmple theoretIcal modC! W'3?~

de"veloped inc' 1 ud} ng these thr E'l-' :-.tecs. j t seems to be in reasonable l'greerr,ent \;OH It'.

the measurements avaIlable so far. However. it WIll need further ·,,,,11 oat I on and refinement.

Procedures for field measurement", b",se" on this "three step" approach stIll neeo to be developed. This mav eventuallv lead to an extension of the lEA

re-commendations.

REFERENCES

( 1 ) E.Lysen "Introduction to WInd I::nerov"

.- CWD, P.O. Box 85, Amersfoort, -Netherlands, CWO 82-1 - 2nd eOitlor,-1983

(2) S.Frandsen, A.R.Trenka. B.Marlbo Pedersen "Recommended pract.ces for •• ind turbine testing. Power

performance testing" JEtl.

International Energy Agency. Roskilde, Denmark 1982

[3] J.v.Meel, P.Smuiders "Some results of CWO's test fields. Are the lEA recommendations sufficient for win~­

mi 1 I s dri vi ng pi ston pumps"" - BWE(~

conference on »Small ~ind Energy Conversion Systems" London - May 7 - 1986

[4] E.Lysen, M.Dlelemans, P.Lely "FIASA Water Pumping Windmill. Output performam::e test resul ts" - CWD test field Almere CWD DHV Consulting Engineers - Amersfoort - February 1986

[53 E.Lysen. M.Dieleman, P.Lelv

"Southern Cross water Pumping iIIlno-mill. Output performance test re-sults" - CWD testfi .. ld Almer ... CWD - DHV - Consulting Engin .. ers Amersfoort - February 1986

[61 E.Lysen, M. Dieleman, P. Lely "Oa",is Water Pumping Windmill. Output performance test results" CWD test field Almere CWD - DHV ConsultIng Engineers - Amersfoort - February

1986

[7] H. Oldenkamp "Field measurements on the CWD 2000 and the WEU 1/3

performed in the period 84.11.27 until 85.01. 14" - CWD -- Wi nd Energy Group - Laboratory of Fluid DynamiCS and Heat Transfer - Department of Physics - Eindhoven UniverSity of Technology R 709 D Februarv 1985