EPA-AA-TEB 66-1
THE EFFECTS OF WATER INJECTION
ON
NITROGEN OXIDES IN AUTOMOTIVE EXHAUST GAS
1966
PROJECT B-l-7-2
(In-House Report.)
D. M. Hollabaugh
-------
INTRODUCTION
Use of water injection in engines dates back as far as the
year 1880. During World War II, its use in aircraft engines
was quite extensive. In all these past situations, its use
has been exclusively to prevent knock. It had a secondary
advantage on wartime aircraft and that was its cooling effect
on the cylinder valves and the piston, which prolonged their
service life at high performance levels. The present investi-
gation of its use was undertaken to confirm its effect on
reducing the NOX emissions in automotive exhaust gases.
In a paper (1) given before the 53rd Annual Meeting of the APCA
in 1960, the theory was discussed on the way in which injection
can limit NOX. Briefly, since NOX formation is basically a
function of the peak combustion temperature and fuel-air mixture
ratio (availability of oxygen), any means which affect either
of these will have a corresponding effect on the NOX emissions.
Reduction of NOX by use of rich fuel mixtures are not desirable
since this results in marked increase in hydrocarbons and carbon
monoxide. With water, .combustion temperatures are lowered
by the heat used to vaporize, superheat, and finally dissociate
the water, which limits the NOX without significantly affecting
the other emissions.
(1)''Exhaust Gas Recirculation as a Method of Nitrogen Oxides
Control in an Internal Combustion Engine" by R. D. Kopa
and Kiroshi (Univ. of Calif.), APCA Paper 69-72, May, 1960.
-------
This earlier work showed that at water to fuel ratios, on a
weight basis, approaching unity, an 80 percent decrease in NOX
could be effected with only modest fuel consumption increases
and power losses, usually in the order of 10 to 12 percent.
Part throttle power losses, such as occurs at roadload conditions,
are always recoverable with either a throttle adjustment and/or
an ignition timing adjustment. In either of these.cases,
however, theoretically, changes will occur in the other compo-
nents of the exhaust gas emissions, as previously shown by other
investigators. For example, increasing either the fuel mixture
ratio or advancing timing should cause an increase in the hydro- .
carbon concentration, with the latter having the more pronounced
effect.
The objective in this project was to confirm previous findings
for roadload conditions and to test an intermediate load value
somewhat representative of a modest acceleration. Also, it
was considered advisable to examine the effect of an alcohol-
water mixture, at two different ratios, to observe the effects
on emission, since alcohol would be used in cold months to pre-
vent the water from freezing.
Conclusions
1. For part throttle operation with the engine tested, water
injection ratios of 0.9'lbs. of water per Ib. of fuel gave
NOX reductions of 75 to 80 percent without appreciable power
losses or effects on hydrocarbon or carbon monoxide emissions.
-------
2. Air-fuel mixture ratios are essentially constant over the
range of injection ratios investigated.
3. Exhaust gas temperatures increase with increases in the
injection ratio for part throttle operations and conversely,
under wide open throttle conditions, they decrease. Although
there is no precise explanation currently available, it does
appear to be an enleanment phenomena for the former.
4. Power loss1 was a minimum with water-alcohol injection.
5. As part throttle operation approaches higher loads, the
range of injection ratios that can be used becomes less
due to their effect on power and hydrocarbon emissions.
TEST EQUIPMENT
Testing was conducted with a 1963 Chevrolet 283 cubic inch
displacement engine and an Eaton Model 1519 DC dynamometer.
Hydrocarbon emissions, CO, and C02 were monitored with Beckman
NDIR equipment, Model type ISA. The NOX concentrations were
measured by the modified Saltzman technique; and oxygen by
polarographic means with a Beckman #777 analyzer.
Fuel weight consumed was measured with a Cox weighing system,
Model 402A, and the air used with a Meriam 6" laminar flow element.
Engine ignition timing was determined with a Sun distributor
advance meter, Model 214, and a timing light.
-------
An explanation of the ivater injection equipment can best be
accomplished by referring to the photograph in the appendix
section. Basically, this is a commercial system sold for detona-
tion control under the trade name "Octa-gane" and is made by
the Engine Accessories Manufacturing Company of Los Angeles,
California. Water is introduced through a plate sandwiched
between the carburetor and intake manifold. Through incorporation
of several toggle valves, the system could be made to operate
whenever it was desired, as contrasted to its former use only
at higher power conditions at wide open throttle. Air was
introduced with the water at the large control valve to help atomize
the water on discharge into the engine.
Test Procedure
Tests were conducted under the following conditions:
1. Roadloads equivalent to vehicle speeds from 30 to 70 mph,
in 10 mph increments.
2. Intermediate loads (22" Hg. manifold pressure) for the
same speed range as in step 1.
3. Wide open throttle at 30 and 50 mph.
For each of these operating conditions, the injection ratios tested
ranged from 0.3 to 1.1 Ibs. of water per Ib. of fuel. Water-
alcohol injection, was tested •only at 5y mph roadload on two
different mixture ratios (80/20 and 60/40 percent, respectively).
-------
Changes in engine torque at constant speed settings were compen-
sated for by throttle adjustments. Initially, spark timing
changes were tried to correct or hold engine performance, but
this was abandoned early because of its more adverse effect on
hydrocarbon emissions.
Results
Graphical results on the effects of water or water-alcohol
injection on emissions and engine performance are shown by the
curves in the appendix section. In all cases, the greatest
reduction in NOX was obtained at the highest injection ratio
(1.1). Through 60 mph, the NOX reduction at this ratio averaged
84.5 percent. Above 60 mph, the gains in NOx reduction became
less because of the limitation on injection ratios that can be
used with effecting power or the other emissions. At 70 mph,
roadload, the maximum ratio was 0.5. For wide open throttle
operation, the only ratio which gave an NOX reduction was 0.3
and this was restricted to .30 mph. At 50 mph, wide open throttle,
no reduction was made in either NOX or hydrocarbon emissions,
in fact both increased by 22 and 32 percent, respectively.
With respect to alcohol-i^ater injection, the 20/80 percent
solution decreased NOX concentration by 80 percent with only a
marginal increase of 4 percent in hydrocarbons. The 40/60
percent solution reduced NOX by 84 percent but hydrocarbon
concentration increased to 52$. Throttle adjustments to hold
power with the alcohol mixtures were very slight as contrasted
-------
with straight water injection. With the latter, increases in
manifold pressure become evident at about a 0.5 injection ratio.
At the maximum ratio, the manifold pressure has usually increased
between 1.5 to 2.0" Hg. above the baseline setting at any test
condition.
Recommendations
That if interest exists, the following be given consideration:
1. Test an injection system on the road and/or under cycling
conditions on the chassis dynamometer for emission eval-
uation and driveability.
2. Evaluate the continuous effects of water injection on
engine durability.
3. Investigate the design and economic aspects of incorpor-
ating water on a passenger car to help establish its
feasibility.
4. Determine what effect water injection has on specific
groups of hydrocarbon under similar test conditions.
Discussion
Under throttled or roadload conditions, the effects of water
injection on combustion are evidenced both by changes in manifold
pressure and exhaust emissions. In regard to the latter, both
NOX and hydrocarbons are reduced by this injection, except
at the highest ratios. At the higher ratios (0.9 to 1.1)
hydrocarbons begin to increase rapidly indicating a progressive
-------
deterioration in combustion. The same effect also occurs at
maximum power with any degree of injection. The increase in
exhaust gas temperatures in the manifold accompanying injection
at roadload conditions appears to be due to a slower burning
rate it imparts, simulating a lean fuel mixture, and causes
late burning in the exhaust manifold. This lower energy level
of combustion reduces nitrogen fixation but also allows further
oxidation of the unburned hydrocarbons before leaving the engine.
At higher load conditions above roadload, water injection has
quite pronounced adverse effects on engine power and hydrocarbon
emissions. Usually where water injection is used for maximum
power conditions, such as in racing engines, its use is predicated
on using leaner fuel mixtures than would be possible without
it. Under these conditions, water serves not only for detonation
control but also to cool the various combustion chamber parts.
The highest injection ratio which did not seriously affect
performance up to 60 mph, roadload (2500 engine rpm), was 0.9.
At this ratio, usually 75 to 80 percent reduction in NOX was
effected. Above it, hydrocarbon emissions and power loss increase
too rapidly. In most cases, something approaching 2" Hg.
increase in manifold pressure was required to maintain specified
torque. With water-alcohol injection, this effect in power all
but vanished and is probably explained by the energy contribution
of the alcohol. From the results,"it appears that 80/20 percent
-------
solution would be the more desirable of the two water-alcohol
mixtures tested, because it gave approximately the same NOX
reduction but with very little increase in hydrocarbons. The
80/20 mixture, however, would probably not be sufficient for
anti-freeze protection.
At part throttle operation, the air-fuel ratio remains essen-
tially constant, although slight changes do occur in brake
specific fuel consumption. However, from an over-all stand-
point, fuel economy should not be too drastically affected
with this control technique, unless the transient response
of the engine has been somehow affected by it. Since an auto-
matic arrangement of this system for part throttle operation
does not currently exist, it was not possible to test this
condition.
-------
10
OTHER REFERENCES
1. "Detonation and Internal Coolants" by E. F. Obert,
Northwestern University, SAE Quarterly Trans., Jan-
uary, 1948, Vol. 2-1.
2. "Antidetonation Injection" by C. H. Van Hartesveldt,
Thompson Vitameter Corp., SAE Quarterly Trans., April,
1949, Vol. 3-2.
3. "Aviation Fuels and Their Effects on Engine Perfor-
mance" (Manual of general information for U.S.A.F.
and Navy), by the Ethyl Corp., 1951.
------- |