hJ/\c(\r-ix RB No.

E4I27

NATIONAL ADVISORy COMMITTEE FOR AERONAUTICS

WARTIME REPORT ORIGINALLY ISSUED

September 1944 as
Restricted Bulletin E4I27

END -ZONE WATER INJECTION AS A MEANS OF SUPPRESSING KNOCK IN A

SPARK-IGNITION ENGINE

By Rinaldo J. Brun, H. Lowell Olsen, and Cearcy D. Miller

Aircraft Engine Research Laboratory

Cleveland, Ohio

NACA
WASHINGTON

NACA WARTIME REPORTS are reprints of papers originally issued to provide rapid distribution of
advance research results to an authorized group requiring them for the war effort. They were pre-
viously held under a security status but are now unclassified. Some of these reports were not tech-
nically edited. All have been reproduced without change in order to expedite general distribution.

E-72 DOCUMEMTS DEPARTMENT

HACA R3 No. EUI27

NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS

RESTRICTED BULLETIN

END-ZONE WATER INJECTION ..S A MEAI-IS OF SUPPRESSING KNOCK IN A

SPiVRX-IGNITION ENGEIE

By Rinaldo J. Brun, H. Lowell Olsen, and Cearcy D. Killer

SUlfMARY

An investigation has baen made of the effnctiveness of vmter

injection into the combustion end 7,one of a spark-ignition engine
cylinder for the suppression of knock. Pressure-time records
obtained show that injection of vfater at 60"^ B.T.C. on the com-
pression stroke at a water- fuel ratio^of Q.;) rendered M-^j fuel as
£ood as S-J fuel from an antik/iock consideration. The optii.iura
crank angle for injection of v,':i,--er into the en'i zone was found to
be critical. As tlie injection ".ngle v;a3 increased beyond the
optimum, the quantity of v.atci required to auopress knock increased
to '^,6 Avatcr-fuel ratio a.^ 1^'-"'' P.T.C. The water quantity could
not be increased beyond 3.6 water-fuel ratio because of injection-
pump limitations; however, a furtiier increase in the injection
angle up to the earliest angle obtainable, wlii.-;h ^vas 20"'* A.T.C. on
the intake stroke, continuously increLised the jraock intojnsity.

The engine operating, conditions of the tests did -lot simulate

those encountered in flight, especially v/ith regard to the oper-
ating speed of 570 rpm. For this reason the results should only
be regarded as of theoretical importance untiJ further investiga-
tion has been made.

I?n.'R0DUC.TI0N

The investigation herein described was suggested by observation

of high-speed photographs of coiribastion i.n a spark-ignition engine
cylinder taken with the i'lACA high-speed camera at the rate of
ij.U,000 frames per second (reference 1;. These j-jhotographs con-
firmed the prevailing opinion that the combustion end zone involved
in the knock reaction includes otily a small fraction of the itiass
of the total cylinder charge. T]ie photographs also indicated that
this small fraction of the mass of the total charge is ?.o;.iprossed
into an extreriely siiiall and well-defined volume before knock occurs.
2 NAGA R3 No. £1^2?

A study of the photographs sucgestoJ that a great savinjj in anti-

knock additive might bo effected by injecting the additive exclu-
sively into the ond zone, pr-ssumably the only region in which it
1.3 needed.

Introd'jction oi water into the corbustion char.ber of the

internal- coir^ustion engine for the purpose oi" s lopressing knock
or cooling of the engine parts is v„ry old. In 1913' Hopkinson
(reference 2) reported successful results cf extensive tests v.ith
large gas engines with which the cooling was entirely by water spray
on tlie inner ourfaces of the combustion chamber. In 1921 Clerk
(reference 3) indicated tnat he had used vrat ^r to suppress extremely
violent knock in the year 1830*

More recent investigations of the effect of water and other

liquids introluceci into t!ie fuel-rdr mixture before a^taission into
the cylinder on Icnook-liiLited and tf-niperature-limited power output
are described in refcronccG U and 5» The water or oth-jr liquids
used in this manner are tei'med "i;.'t'>rnal coolant."

The object of the tests repor'"ed herein was to determine the

value of injection of vrater into the end ^one as an antikriock agent.
The engine used in these tests could not bo superchcurged and the
inlet-ail' temperature could not b-.; raised above room feniperaturej
therefore, the effjctiveness of the water was determined by lowering
the aritiknock value of the fuel. The engine operating speed was
far belov.' the speeds used in flight. The results should bo consid-
ered only on a theoretical basis jntil further investigations have
been made more closely appr achin, the operating conditions of
fligiit. Tnis vfarning is particularly important because difficjlty
may be expei-ieiiced in applying the- inetiiod to high-speed engine
operation.

The tests were perfonned at th"^ Aircraft Engine Research Labo-

ratory of the National Advisory Committee for Aeronautics d^aring
late 19'-t5 <and early 19l}i.

APPAiUTUS AND PROCEDURE

The engine used in the present tests was the NAGA combustion
apparatus described in reference 6. An injection valve v;as installed
in one of the cylinder openings to spray water into the last part of
the charge to burn. Photographs were, taken with the NAGA high-speed
cairiera to ooserve the exact end-sone position and trie shape and the
extent of the region permeated by the water spray.
MACA R3 No. SUI27

The snrine v.'c.s fired for om.- cycle only or each test run.
J

An c-lectric motor kept the engine speed constant until the firing
cycle occurred, a single-cycle clutch wat> engaged at the beginning
of the firing cycle betvjean the engine crankshaft and an accessory
sliaft to inject a single charp;e of fuel into the cylinder on the
intake stroke, ignite the charge at tnc-; proper tin^.e, and inject
water at t,lie predetermined injection anfle. The cylinder tenipera-
tiue was maintained by circulating heated glycerin through the
engine jackets. Three spark plW;S were used to brin^ the end zone
inbo the desii'ed position in front of the water-injection nozzle.
The ignition timing was so set that t!ie coiTipletion of burning mth
S-3 fuel occurred near top center. The water used for injection
into the cylinder v;as deaerated in order to avoid the fo.miatJ.on of
gas bubbles in the injection valve. The fuel-air ratio was set
slightly richer than that required i'o\- inaxim-iim Icnock with S-J fuel.
T^ffo valves vrere used, each serving for both intake and ex.haust.

Press'.aro-tinie records were obtained by paotographing an oscillo-

scope screen. The input to the vertical nlates of the oscilloscope
was r'i'oduced by a piczoelectrj.c prer.sii;-e dck?.ip installed in the
engine.

Engine conditions held con3tar:t were:

Engine speed, rom , 570

Fuel injection (intake stroxke), degrees A.T.G 20
Fuel-air ratio (approxiiiate) 0.U72
Cylinder tenioeratiare, '^F , 2,!|2±2
Go.-ripi'ession ratio 7.1
Intake pressure atrnosoheric
Exnaust pressure at.aosoheric
Intake teinj ;rat ire, °F 65-70 (room temperature)
Spark tiirdng, degrees B.T.C.:
Earliest plug 5011
Later plugs 26+1

RES-'.JLT:; AMD iHSCUSSIQAl

A photograph of about one-i'.alT the coniDustion chamber at the

time all the water spray had entered int/o the chamber is shoT;n in
figiir'T 1. The dashed cj.rcle shoves the conbusti on-chamber outline
as viewed froi.i the top. The letter G in the figure indicates the
position of the earliost-fired spark plug, E and F the positions
of the later-fired spark plug's; J indicates the position of the
vrater-injoction no'^-lc, H the position of the fnel-injection
nozzle, and I the oosition of the piezoelectric pressure oickup.
 MACA P3 No. Em27

A photOi^raph of the end zone a fev: e:agif.e crank-angle degrees

before the end of combustion is shown in figure 2. Tiie end zone as
seen in figure I is somewhat larger than the knocking end zone under
the most severe condition^ in'-Dosed in the tests. The photographs
in figures 1 and 2 are single frames from two series of pictures
taken at UO^OUO irar'ies per second vri. th the NACA high-speed camera
and srjow the develo-onent of the v^ater spray from the aozr.le and the
burning ox the charr^e in the ctiojiiber, respectively. In figure 2,
the Tfater-cpray outline obta:Lned ji-om f ij:ure 1 has b'3en di'avn in
to shovf its location -with reopect bo the end zone.

Pressure-time records of firing cycles are shown in fi^'ures 3

to ^. Tho lovfest trace in each li.-jure is a motoring trace taken
j'ost before the firin^^ cycles.

The traces of tv.'o coiisecutivs firing cycles superiimosed on

t!ie sane fi]m are jiicwn in fij-'ure J. The trace with the violent
knock, designated "unquenched ll-^i" in the fi£;urs, vfas taken A-ith
IJ-3 fuel vjithout water injection. The trace vlth incipient knock
and larger area under the trace, designated "quenched li-jj" in the
figure, Vfas taken vdth M-^ fuel v.c.th injection of xvater at
^9^ B.T.G. on the cornpressioa stroke. The weight of vfat^r injected
was three-tenths of the fuel weig'.it. All of tne operating condi-
tions of the engine and osciiloscoe were th',. saine foi' the two runs
of figure 3 except the vv-ater injecion. The amplitude of the vibra-
tions registered on tho qucnciiei trace was considered to be indica-
tive of incipient knock because a lower axiplitade of vibrations was
not readily detect3a visually on the oscilloscope. The '.'.'ater-fuel
ratio of 0,3 with injectioji at t/ie optiir.uiri crank angle consistantly
gavo incipient knock or no knock. The 0.;? water-fuel ratio v/as the
lowest obtainable .vith tho water-injection systeir.; therefore, the
siTiailest water quantity necessary to quench the knock v;as not deter-
mined iii this investigation.

High-speed Motion t;icturcs of the quenched cor^buction of

M-3 fuel indicated that the knock did not, occur in that part of the
end zone permeated by the water, -.ut that the incipient knock such
as registered on tii.. quenched trace in figui-e 3 came from a v^ry
snail fraction of the charge located on either side of the ^^'^ter
spray near th.e cylhider wa].l. If bhi small pockets of gas near the
water- injection nozzle alor^gside the cylir.der v^all could have been
sprayed with vfater, possibly all traces oi knock '©uld have been
removed.

The pov/er loss resulting from severe k:iock is shown in f ir-

ure 3^ The tn"0 traces are nearly identical aporoximately to the
point -;;here knock occurred in the unquenched trace. Th2 loss of
energy witli ..heavy knock cannot be acco^ontcd for as being involved
•)'•':
4 CA RB No. EUI27

in the energy ci vibratioA of the gases, because even after the

vibrations have been nearly damped out the unquenched pressure-
time tr'ice remains about the same distance below the quenched
trace, borne of the energy may be lost in the highei- heat transfer
through the cylinder walls with hea\^ Vrnock because of the greater
scrubbing action of the gases on the chamber Avails, as suggested
by iDitlirow and Rassweiler (reference 7). The greatest portion of
the energy loss might be accountable in the form of unreleased
energy in unburnsd carbon. During every violent knocking run made
in these tests, the engine released a large quantity of black smoke
on the exV'iaust stroke, Vilrien the knock was reduced to the incipient
level, eitiier by quenching or by increasing the antikiiock value of
the fuel, no traces of smoke v»-ere noted, KacCoull (reference 3)
has presented nieasurements of power loss with heavy knock.

The traces of ti/vo firi.ng cycles were super impose.i in figure h

to siiow tfiat quenched M-J fuel pi'oducea as much power as unquenched
S-3 fuel. The octane ratings of I'-';;, and S-3 fuels, as obtained
from the A.S.T.M. (Motor) Method, are about 20 and 100, respectively.
The quenched trace in figure has a shar'p rise ap'oroximately at the
Ij.

point where fciock v;oulJ occur in an- unquenched cycle. The sudden
rise in pressure at that jjoint is charactei'istic of all quenched
M-3 i-iins. Kotiou pictures of thci-si; runs, ho-^'revor, do not show the
cliaracteristic vibrations in vhe b'.irned gases I'.'hich accompany kiiodc.
The trace for unquenched S-3 fuel in figure h siiows about the same
amplitude of knocking vibrations a3 does the trace iu the same
figure for quenched M-3 fuel.

The traces shown in figures 3 ^nd J4 are representative of more

than 50 traces taken -onder the same test condJtions. The reproduc-
ibility of the test data was very goodj furtiiermore, the order of
taking the tvro traces on the carJs vv'as reversed many times and
identical resaits were obtained.

The ootimuiii angle for start of water injection with resoect to

knock was found to be critical; under the conditions oi" the tests,
the optimum angle was fx'om $0° to 60° B.T.C. on the compression
stroke. The total time require.! lor injection 01 i/vater, as indicated
by the high-speed photographs, was S'-* to 10° of crank angle. Start
of injection even as late as 10^ B.T.C. did not give good results
regardless of the quantity of water injected up to the capacity of
tlie water-injection t;ystem., which was 3,6 tiip.es -i-he fuel quantity,
or 12 times the water used at the optimum angle. The large time
lapse between knock an..l injection (50^'' irdnimum angle) is not encour-
aging from Uie consideration nf applying the listhod to the conditions
of flighit. If ttie mj.nimuifi time between water injection and com-
pletion of burning remains constcu'it for different engine speeds, the
miniinuia anfle of injection foi' good resists v.-ill inci-'ease from 50'^ at
 KACA R3 No. EUI27

570 rpm to 2h$° at 2800 rpm. The Droblerr' of injectinr: water when
the r)istcn is near bottom center in such :.!anner that it will ba
concentrated in tiie end zone when the piston is at top center will
be diificult.

With injection of water betvr en 6vO° and 13:?'^ B.T.C., the quan-
tity of vrater req'iired to prevent t.ie M-3 fuel fi'or:i exoeoding tho
incipient feock limit increased rapidly fi'oti. O.3 to y.b water-fuel
ratio. As the injeoi-ioii angle was advanced beyond 13^'^ 3.T.C. on
the cojT.pression stroke tc 20° A.T.G. on thf; intalte str-^ke, the
knock ijiteasity conbinaouol.y increased at u constant water-fuel
ratio of ^,6; also, the H.-aiie speed deci-eased and Icnock occurred
later in the cycle. The slower rate of b-jrninc lowered the cycle
efficiency considerably. Fr--"thei.T;i.i"a, so.rie cycles vrere drovvned
out with the 3.6 watci'-iuel ratio injected at 20° A.T.G. on the
intake stn ke

Injection earlier than 20° A.T.G. on the intake stroke was

not possible b&ca ise of mechanical lindtatioas. Injection at this
angle was as nearlj'' oompprable viith introducing the coolant into
the manifold near the intake valve as possible with the apparatus
used. In other engines, depending on the operating conJitioas,
introducing the coolant into thn manifold near the intake valve
will require different wat-.3r quantities for the same effective
knock reduction. The 3.6 water-fuel ratio used on this apparatus,
therefore, sho'old now b'j used in making connarisons witli other
engines.

In order to assiu'e as thorough a raixinp of the water and the

fu'3l-air ciiarge as possible, the plain valves used in the runs of
figures 3 =^nd were reolaced by shrouded valves for the run of
'4.

figure 5. The shrouded valves i;icreased the turbulence, -.)artic-

ularly durLug t^£ air intake. Figure 5> is =1 pressure-time 'trace
taken v.ith M-3 fuel, J). 6 v/ater-fu^l ratio, water injection at
20 A.T.G. on the intake stroke, u^d all oth^r conditions the same
as for the other runs. The knock intensity was not reduced appre-
ciably from that produced by unquenched 11-3 fuel, as shown in fig-
ure 3- The rate of pressure rise durin<.; combustion was appreciably
reduced as coirspared with the pressure rise at the optimum injection
angle (figs. 3 and U), particularly during th'.^ earlier stages of
combustion, uith shrouded valves, the rate oi' nrrssure rise v;as
considerably greater than vath pl.n.n valvos at tho earliest injec-
tion angle and the largest quantity 01 -'.'atei-.
NACA RB No. EUl27

SmMAEY CF RESULTS

The follovdng res'iLts wore obtained from limited water-injection

tests ccndactid on the NAGA combustion apparatus at an Oinginc oper-
ating speed of 570 rpra:

1. Injectinf^ vrater into the end zone with a v/ater-fusl ratio

of 0.5 reduced the knock intensit.';- of L-3 fuel to that of S-J fuel
without water.

2. The considerable power loss associated with violent knock

was prevented in the cycles in which, the knock v.^as quenched by
.end-zone viater injection.

3. The optimum uxir'/ie of water injection was critical.

Aircraft Engine Research Laboratory,

National Advisory Gvomndttee i'or Aeronautics,
Cleveland, Ohio.

RZlFEPJiMCES

1. Miller, Gearcy D. A Study by Hifh-Speed Photography of Combus-

:

tion and Knock in a Spar k-Ifnit ion Engine, NACA Rep. Ur, 727,
19l;2.

2. Hopkinson, Bertrar.: A New Met.iod of Coolirip Gas-Engines. Inst,

Mech. Eng, (Proc.j, PLs. 5-li., July 1913, pp. o79-71!j.

3. Clerk, Dufjald: Cylinder Actions in Gas and Gasoline Engines.

SAE J.jur., vol. VIII, no. 6, June 19^1, pp. 52>-539.

.'4.. Rothrock, Addison M. , Krsek, Alois, Jr., and Jones, Anthony YJ.:
Summary Report on the Induction of Water to tlie Inlet Air as
a Means of Intei'nal Cool-Lng Aircraft Engine Cylinders.
:i 1

NACA MiTi, AiiG. 1042.

5. KiJiring, M. S.: V/ater and Wat-;r-Alcohol Injection in a Super—

cliarged Jaguar ''urcrait EngLne. Canadian Jour. Res., sec. A,
vol. 16, Aug. iry], op. lli9-i7o.

6. Rothrock, A. !vi., anl Spencer, R. C.; A Photographic Study of

Goribustion and Knock j.n a Spar'/-Ir.nitioa Engine. NAGA
Rep. No. 622, I93S.
 VikCk RB No. Shl27

7. Withrowj Lloyd, and Rassvrei.ler, Ger-ild M,: Slor Ijiotion Shows

Knocking and Non-Knocking Sxolosions. 3A£ Joii'., vol. y9,
no. 2, Auj-. l?3.j, pp. 297-303, 312.

3. MacGouil, Neil: Power Loss Accompanjanj^ Detonation. 3AE Joiir.,

vol. Ui, no. Ij., A.rril 1?39, -o.' l^^-lbo.
NACA RB No. E4I27 Fig. I

Early
spark plug

Piezoelectric Fuel-i nject ion

pressure nozzi e-
pi ckup-

-Outline of
combust ion
chamber

C>639l

Uoter-inject Ion nozzle

Figure 1. - Photograph of uoter spray

in combust ion chamber
after uater injection uas complete. Combust ion
has not token place. Exposure time,
25 microseconds-
NACA RB No. E4I27 Fig. e

Early
spark plug

Plezoelecfrf c Fuel-Inject Ion

pressure noz? le-
pi ckup

NACA
C* 6 39 2

Flame
Flame-' ,
^ End zone
Vater-lnject Ion nozzle

Figure 2. - Location and shape of end zone a feu engine

crank-angle degrees before comp let Ion of burning.
Exposure time, 25 microseconds; relative position
of uoter spray, shoun by dotted lines, uas
obtained from figure 1.
NACA RB No. E4I27 Fig. 3

CO
o
»-

>-<
r) ^
cz
oo
to
SI
"•^ -^
.

mac w. 1

v.^
to
c; LU to

^< -too
<iy
OC
a o ••« .
3 <o 1^
-iP to to
6 •* c <w
o <>^ o o
K -^ a
*
1

1 <0
lATIO TTE

to w "^
«^.,
^i *.
to
a t en
S , Oh- o I

o o:t
o ««
o •.
>a.
.
to c to-.
H- a— to
•^ W ties
to tH*
ts CJ
«e to
fc « t: ..
•- o o c
a c •- a
<ow a o
to ». i.

o.to"t)
•--.to
<o t:^
to »o •—
O^
o O
c
V. ^
1

k. o to a
•». c t^ o*
•* o t
5 ass
c c
-. o ••
•w en .
C « 5j
1

to
k.-.
oo K. .

a . « .
(0 O O 4)
40 to-.
to'-t'^*
^ t oo»
«».-^ W.lfS

«n

5
NACA RB No. E4I27 Fig. 4

V3
O
>- = o
cc-< k
o z -c
1
»>
c/jO I. «o
_ oc •.
.^ t
>UJ •^
Q-< ago
><
OC ^ a— «o i^
.

—'O ^ 'i. 10 Q>

^"^ a <n ^ -fc.
•*- o1 k^
oi^ «-* **-5
UJ <n ^
»-»- 1 .o o
«<»-
^— *:— o c
tov_
z: t5 a •«
z Q) -^ Q)<Y>
o
o C
O
-^ 1
XJotO
c •• c
* fc CS «
a a —
Cr k. t <i)
O :>
tkot) -^"^
C % «.
•- ^ O
k. O «) •.
O C '^C
ck. to c a
e a "^ k.
OO
O k."0
1

to to
«o . >•* .c
to— o o
o to a c
e> a to
w.H^ .,
^ Or\
a
o-
f»> . R
to
«Co
o^ 1

•^
»- ti o .
1 to--^
Jto-c - .
k. o e»—
a c c .
«o to
«o a —
«ft

to e- to«
k. fc avo
«*. aH~«o

2J
NACA RB No. E4I27 Fig. 5

o
«^«o
C) •

ozo<
«> cvx
q:
>s *
(n w o
Q>
> UJ a «>.
•
t)^
<< CJ k. •

a: . •«b

<o
•
c Qa-^
a
Q) •k-O
O lij
-^
o V.OSI
1 o
>\^
o 4^ ^
o—
Q«a «>
t -*
•« o
•Mlk

o k^cn k.
o •". 1
'«>.
H» * CO
C3
^* «*«
«0 Q> o
t
—
:3 «.
«. U. fc
a •^.
O ^
<: •
<o Q> Q>
•^ *^.

% «o«o
o C C
o o o
i.
«.. c c
o •^
o
Q> •«.
^ «> '•l.
•«. o o
'*> q> %
1
•-^ •-%
% c •^ k:
k.
a 1 1

V) i. V.
«0 Q> 09
«> '*.. «_
k. ocs
•i. 3 a
 UNIVERSITY OF FLORIDA

3 1262 08104 951 1

IhSlTY OF FLORIDA
1
-f^^S QEPARTMEMT
Srosv^ SCIENCE U6RARY
.^.j.'-ja iiroii
GAINBSVIkLE. FL 32611-7011 USA