WHEC 16 / 13-16 June 2006 – Lyon France

A Development of a High H2 with High Response by Using Common-Rail Injection System for Direct Injection H2 Fuelled Engines

Kimitaka Yamanea, Masakuni Oikawaa, Tomonori Kitauraa, Kouta Mawataria, Takashi Kondoa, Yasuo Takagia, Yoshio Satob, Yuichi Gotob

aHydrogen Energy Research Center, Musashi Institute of Technology 1-28-1 Tamazutsumi, Setagaya-ku, Tokyo, 158-8557, Japan [email protected] b National Traffic Safety and Environment Laboratory (NTSEL) 7-42-27 Jindaijihigashimachi, Fuchu-city, Tokyo 182-0012, Japan [email protected]

ABSTRACT:

The authors have developed a new generation hydrogen gas injector which may overcome the problems brought about by the previous activated hydraulically. The hydrogen gas injector is activated electro-hydraulically by working fluid at 100 MPa fed by the common rail system developed for diesel engines to open the needle of the injector. And the hydrogen gas at high pressure of 20 MPa in the injector is directly injected into the combustion chamber. An experiment was carried out concerning the injection flow rates affected by the injection crankangle at two engine speeds such as 1000 and 3000 rpm, the and pressure of the working fluid to obtain the characteristics of the injector.

KEYWORDS : Hydrogen Gas Injector, Electro-hydraulic System, Common-Rail

1. Introduction The development of hydrogen gas injectors has been being carried out for direct injection hydrogen fueled engines in Musashi Institute of Technology for many years[1], [2], [3], [4], [5]. Because it has been found theoretically that the direct injection of hydrogen fuel generates about 1.2 times as much output power as that of a gasoline engine and that the direct injection also brings about greater thermal efficiency than that of a hydrogen engine with external mixture formation owing to greater cooling loss than that by hydrogen fueled engines with direct injection. The previous hydrogen gas injectors were activated mechanically or hydraulically by working fluid at 20 ~ 30 MPa fed by a diesel plunger pump to open the poppet or needle of the injectors. And the hydrogen gas at high pressure such as 6 ~ 10 MPa in the injectors was directly injected into the combustion chambers. It was found experimentally that the previous injectors needed greater injection duration in crankangle as the engine was operated at higher engine speed. As a result, the thermal efficiency got worse as the engine speed became greater. It is necessary to have the injector activate in as small crankangle as possible to increase the thermal efficiency even at greater engine speed. The required injection crankangle is smaller than 30 deg. during which the of the engine seldom moves around the top-dead-center. And the direct injection hydrogen fuelled engine is subject to NOx emission even in totally lean mixture op- eration, namely low load, because of the heterogeneous combustion. It has been found recently that the de-NOx catalyst developed for fossil-fuel lean mixture engines was effec- tive also in hydrogen-fuelled engines when hydrogen gas was injected into the exhaust manifold upstream of the de-NOx catalyst [6]. It is very important to develop an injector with quick response activation by using the common-rail system developed for diesel engines where the pressure of the working fluid is large enough and the system works electro-hydraulically so that a swift and multi-injection control might be available.

1/7 WHEC 16 / 13-16 June 2006 – Lyon France

2. A New Generation Common-Rail Type H2 High Pressure Injector In order to operate the injector electro-hydraulically coupled with engine control units, for the needle valve of φ29 the injector to move up and down as swiftly as possible Receptacle and for practical use for truck engines with 4 valves for Flow Control Electric Magnet each , a common-rail type hydrogen gas injector Upper Plate Servo-valve was designed to know whether the common-rail system Flow Control Working Fluid was applicable to the high pressure gas direct injection for Lower Plate a hydrogen fueled internal combustion engine of a (Pressure:100 Map) medium-duty truck with the loading capacity of 4 tons by making use of the parts of a common-rail injector (Part No. Hydrogen Supply 23670-39015, Denso) developed for diesel engines. Drain Port (Pressure: 20MPa) Figure 1 shows the cross-section view. And Figure 2

shows the overall photograph of the injector developed. mm The diameter of the valve seat was 4 mm. The tip angle 53 of the needle valve was 90 degrees. The electric magnet 2 servo-valve, the flow control upper and lower plates and the fixtures were removed from the original common-rail Pushing Rod diesel injector to assemble the parts into this new Lift Sensor common-rail type hydrogen gas injector. The others were (Max. Lift:0.3 mm) newly designed for hydrogen gas injection at the design pressure of 30 MPa at the maximum and the design pressure of the working fluid was 200 MPa at the Needle Valve maximum. A rubber O-ring and a doughnut-shape resin sheet made of Teflon were used to seal between the φ4 hydrogen gas and the working fluid. A lift sensor (A φ12 Special Model; Applied Electronics Corp.) was embedded in the injector as shown in Fig. 1 to measure the Fig. 1 Cross-Section View of Common-Rail movement of the needle valve. Type High Pressure Injector The common-rail type high pressure hydrogen gas injector works as follows. Receptacles The command signal has the injector driver generate a large driving electric current. The electric magnet servo-valve moves WF Supply Inlet WF Drain up to open the outlet of the working fluid pressurized in the flow (100Map) control upper and lower plates. The pressure falls down swiftly and the pushing rod moves up together with the needle valve which is forced to move upward by the hydrogen injection pressure. As a result, the needle valve opens and the hydrogen at the pressure of 20 MPa is injected out. Once the command signal and the electric current are off, the electric magnet servo- 約φ64 mm H2 Supply Inlet valve moves down to close the outlet of the working fluid 3 5

(20Map) 2 resulting in building up the pressure in the flow control upper and lower plates. Consequently the pushing rod moves down while pushing the needle valve to the valve seat. In the end, the hydrogen injection stops. Injector Body

2. Experimental Apparatus and Method Figure 3 shows the schematic diagram of the experimental apparatus. The common-rail driving system developed for a 4- Injection Nozzle cylinder, -cooled diesel engine was utilized for the injector test rig in this experiment. The pump was driven by a AC motor and diesel fuel as the working fluid was fed to the common-rail, Fig. 2 Photograph of the injector then to the injector. The reason why diesel fuel was used as the working fluid was that all the parts to control the flow of the working fluid were originally made for diesel fuel. The pressure of the working fluid of 100 MPa was used in principle except for the experiments for the influence of the working fluid pressure on the injection flow rate. The pressure was adjusted by the relief valve in the line. The working fluid then went back to the tank. The temperature of the working fluid was measured at the working fluid tank. The temperature was, in principle, kept from 50 to 53 ℃. except for the experiments for the influence of the working fluid temperature on the injection flow rate. On the other hand, hydrogen gas stored at 35 MPa in the hydrogen high pressure tank was reduced by the pressure regulator to 20 MPa upstream of the mass flowmeter. The flow rate of the hydrogen injected out by

2/7 WHEC 16 / 13-16 June 2006 – Lyon France

Oscilloscope

Relief Valve Readout Outlet Cooling for Pressure Unit Water Control Pressure Current Probe Sensor Pressure Guage Injector WF Micro- WF Filter Driver Tank Processor

Constant Current Common-rail DC 12V AC 3-Phase Supply Amp. 200V Readout Injector Unit Motor Pump Silencer Pressure Sensor Mass Pressure Temp. Sensor Data Logging PC Regulator Flowmeter

WF: Working Fluid AC: Alternate Current Surging Tank Surging Tank DC: Direct Current PC: Personal Computer H HP Tank 2 HP: High Pressure Fig. 3 schematic diagram of the experimental apparatus the common-rail type hydrogen gas injector was measured by a mass flowmeter (Model 1.25 No. F-1238-A-12-11N, Oval Corp.) 1.20 calibrated in advance by injecting the hydrogen gas at 20 MPa by the tested 1.15

0

injector. The temperature and the pressure V) ( 1.10 Lift Rubber O-Ring of the hydrogen gas were measured by a E

t x H2 pressure sensor and thermocouple u Leading p 1.05 t Wire upstream of the injector respectively. u O The lift sensor employed in this experiment 1.00 H was an inductance type one. The 2 0.95 Pushing Rod H Valve Seat calibration was carried out identically in the Electric 2 Needle Coil Ball Bearing Joint Valve same manner as the injector works in the 0.90 experiment. Figure 4 was the calibration 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 curve obtained. It is found from Fig. 4 that Lift x (㎜) there is the maximum output at 0.4 mm Fig. 4 Calibration Curve : Needle lift vs Output from the zero lift. This is because the push- ing rod has a ball bearing joint where the area of the cross-section becomes smaller. So, in this experiment, the maximum lift of the needle valve was set to be 0.3 mm for fear that the lift might pass across the maximum output observed in the calibration curve shown in Fig. 4. Therefore, once the characteristics of the output of the lift sensor are improved by longitudinally unifying the area of the cross- section of the pushing rod, the maximum lift of the needle valve can be further extended. As a result, large amount of hydrogen will be obtained. The experimental measurements were carried out to know the characteristics of the flow quantity per injection concerning actual injection duration in crankangle at two engine speeds such as 1000 rpm (8.3 Hz) and 3000 rpm (25 Hz) and the pressure and temperature of the working fluid. And the characteristics of the movement of the needle valve were also studied by comparing the command signal and the movement of the needle valve measured.

3/7 WHEC 16 / 13-16 June 2006 – Lyon France

Command 1000rpm(8.3Hz) 3000rpm(25Hz) Signal Duration Pinj=20[MPa], PWF=100[MPa], Liftset=0.3[㎜], TempWF=50~53[℃] 5° Command Signal Duration 5° Command Signal 5V Electric Current 10A 10.7° 13.0° 5° 1.54° Lift Actual 4.50° Injection 0.3㎜ Delay Duration Lift Time

Start End

10° 10°

16.4° 24.2° ° 10 1.53° 4.47°

15° 15°

21.4° 32.1° 4.77° 15° 1.52°

20° 20°

26.1° 36.2° 1.53° 4.60° 20°

Fig. 5 Display of Comand Signal, Electric Current and Needle Valve Lift

4/7 WHEC 16 / 13-16 June 2006 – Lyon France

3. Results and Discussions

) 45.0 g. 3.1 Characteristics of the Needle Valve e 40.0 d

Movement (

nj 35.0 To study the characteristics of the needle i 3000rpm 30.0 valve movement, the lifts of the needle Δθ

valve were measured by the lift sensor on 25.0 i t a capable of the maximum lift of the needle r 20.0 u valve 0.3 mm set when the command D 1000rpm

n 15.0 o

signals were 5, 10, 15 and 20 deg. i t

c 10.0

crankangle (CA for short from now on) at e j n I the injection frequencies of 8.3 and 25 Hz, 5.0 l a

equivalent to a 4- engine speeds of u t 0.0 c 0 5 10 15 20 25 30

1000 rpm and 3000 rpm respectively. A

Figure 5 shows the results recorded on an Command Signal Duration Δθcs (deg.) oscilloscope screen. In Fig. 5, the results Fig. 6 Relationship between Comand Signal Duration and is displayed in two rows. The left row is the Actual Injection Duration Measured results in case of 1000 rpm. And the other one is the results in case of 3000 rpm. The results displayed vertically are 1200 l) Tip Angle of Needle Valve:90° different in the command signals. In each m Diameter of Valve Seat :4mm

N Maximum Needle Valve Lift:0.3mm ( 1000 one result, the uppermost is the command Working Fluid Pressure:100MPa nj i Working Fluid temperature:50~53℃

signal. The second one is the electric Q n 800 o current generated by the injector driver to i t

c 1000rpm

the electric magnet which controls the e j

n 600 servo-valve. And the last one is the lift of I r 320Nml (Vs=1liter,ηv=0.8

the needle valve. It is found that the delay pe

Stoihiometric Mixture) y

t 400 time crankangles were almost same even i t while changing the command signal uan 200 duration crankangle in the same injection Q

2 3000rpm

frequency such as 1000 rpm or 3000 rpm. H However, every actual injection duration in 0 crankangle is larger than that of the 0.0 10.0 20.0 30.0 40.0 50.0 Actual Injection Duration θ (deg. C.A) command signal. ing Figure 6 shows the relationship between Fig. 7 Hydrogen Quantity per injection Qinj vs the command signal duration in crankangle Actual Injection Duration and the actual injection duration measured with the output of the lift sensor. It is found that the actual injection durations measured are longer than the command signal ones in both the cases of 1000 rpm and 3000 rpm. Moreover, the actual durations in case of 3000 rmp are longer than those of 1000 rpm. The reason can be understood from the difference in the time history of the lift. In case of 1000 rpm, almost results shows that the lift reached the uppermost value of 0.23 mm. On the other hand, in case of 3000 rpm, the real time of the command signal duration was very short, a third of that in case of 1000 rpm, so that the large portion of the time was used almost for the time in lifting and pushing down the needle valve. From the mechanical character of the common-rail, even if the real time was the same as in the case of 1000 rpm, the injection duration in crankangle was longer because an engine in case of 3000 rpm makes three times as many revolutions as in case of 1000 rpm. It means that it is necessary to measure the lift of the needle valve in order to determine the opening and closing timings of the needle valve.

3.2 Influence of Injection Duration The influence of the injection duration on the hydrogen quantity by one injection was studied by measuring the flow rate injected by the tested injector while changing the command signal of the injection duration such as 5, 10, 15, 20 and 25 deg. CA at 1000 rpm (8.3 Hz) and 3000 rpm (25 Hz). To obtain the actual injection duration in crankangle, the output of the lift sensor were measured simultaneously as shown as in Fig. 5. Figure 7 shows the hydrogen quantity per injection Qinj against the actual injection duration in CA in both the cases of 1000 rpm and 3000 rpm. The injector studied here was developed for medium-duty trucks with the stroke volume of 1 liter. Then, the amount of the hydrogen gas necessary for the engine with direct injection to run in the stoichiometric mixture with the volumetric efficiency of 0.8 can be easily calculated to be 320 Nml. It is found in Fig. 7 that the amount of the hydrogen gas of 320 Nml can be obtained at the actual injection duration of 12.8 deg. and 37.5 deg. for 1000 rpm and 3000 rpm repectively. The actual injection duration obtained in case of 1000 rpm is small enough. However, the actual injection duration obtained in case of 3000 rpm is a little bit larger than the required crankangle of 30 deg. It is found that the maximum lift

5/7 WHEC 16 / 13-16 June 2006 – Lyon France set to be 0.3 mm should be increased much further in Liftmax=0.3mm,Temper=50~53℃ 3000rpm (25.0Hz) order to inject the hydrogen quantity of 320 Nml within the 350 actual injection duration of 30 deg. CA. ty

i 300

t on l) i n t m c 3.3 Infuluence of Working Fluid Pressure a e u N j

( 250 n Q I

In practice, the pressure of the working fluid is generated nj

i

r Q by the high pressure pump whose driving power is e H2 p 200 provided by the engine power through the engine driving shaft. It is very clear that the thermal efficiency decreases 150 as the pressure gets greater. Therefore, a low pressure of 5 the working fluid is preferable. An experimental study was as 4 G

) carried out to understand the influence of the pressure of 2 3 H %

d 2 the working fluid on the hydrogen quantity per injection Qinj, L( e

k 1 the static leakage flow rate through the needle valve Ls a e 0 and the actual injection duration Δθ in case of 3000 L inj 40 rpm (25 Hz) with the maximum lift set to be 0.3 mm, the .) 35 on injection pressure of 20 MPa and the working fluid i t deg c n (

e

o 30 j j temperature of 50 – 53 ℃ adjusted. i t n n i I a

r Figure 8 shows the influence of the working fluid pressure l 25 u ua D t on the hydrogen quantity per injection Q , the static Δθ inj c 20 A leakage flow rate through the needle valve Ls and the 70 80 90 100 110 120 130 Pressure of Working Fluid P (MPa) actual injection duration Δθinj in case of 3000 rpm (25 WF Hz). The static leakage quantity L in the Fig. 8 was Fig. 8 Influence of Working Fluid Pressure determined by multiplying the static leakage flow rate Ls by the time needed for 800 the engine to turn round by 720 degrees ) ml crankangle. But, in the figure, the static N 700 ( j n leakage quantity L is expressed in a i 600 Q (Command Signal Duration:15(deg.)) percentage of the hydrogen quantity per n 1000rpm o i

t 500 injection Qinj. c e j

It is found that the actual injection duration n 400 I became greater as the pressure of the r

pe 300

y

working fluid decreased. The long actual t i injection duration was attributed to the t 200 decrease in the pressure of the working uan 3000rpm (Command Signal Duration:15(deg.)) Q 100

fluid resulting in the working fluid’s taking a 2 long time to flow out of the space in the H 0 20 30 40 50 60 70 80 90 flow control upper and lower plates. In the Working Fluid Temperate (℃) same manner, when the needle valve was closing, the working fluid also took a long Fig. 9 Influence of Working Fluid Temperature time to fill the space in the flow control upper and lower plates. As a result, the needle valve was kept open, thus the actual injection duration became longer. It is also found in Fig. 8 that the hydrogen quantity per injection Qinj became greater as the pressure of the working fluid decreased. Concerning the leakage, the contact pressure with the needle valve and the valve seat got lower as the pressure of the working fluid decreased. As a result, to the contrary, the leakage increased. However, it is found that the amount was negligibly small as compared with the hydrogen quantity per injection.

3.4 Infuluence of Working Fluid Temperature It is conceivable that the working fluid temperature affects the hydrogen quantity per injection because the kinematic viscosity is subject to change with the temperature. The influence of the working fluid temperature was studied. The temperature was measured at the working fluid tank as a practical means. Figure 9 shows the results obtained by changing the from 30 to 80 ℃ with the command signal duration of 15 deg. CA, the working fluid pressure of 100 MPa in both the cases of 1000 rpm (8.3 Hz) and 3000 rpm (25 Hz). It is found that the influence was smaller than the expectation and the influence in the low temperature operation was larger than that in the high temperature one judging from larger difference in the hydrogen quantity per injection in the low temperature operation.

4. Summary The followings are the summary of this study though the injector was not installed in engines in actual firing operations.

6/7 WHEC 16 / 13-16 June 2006 – Lyon France

(1) The common-rail system developed for diesel engines is functionally applicable to hydrogen gas injectors. (2) It is found that the actual injection needs longer duration time than the command signal duration one as the injection frequency increases, namely the engine speed increases. To shorten the duration time at high injection frequency, much further effort is required to study the flow control in the upper and lower plates, weight-savings in the moving parts, the diameter of the pushing rod and the lubrication. (3) The characteristics of the lift sensor should be improved by simplifying the configurations of the crown of the needle valve. (4) The static leakage was small enough so that the amount might not disturb the combustion brought about by the direct injection. (5) It is found in this experimental study that the maximum lift of 0.3 mm was not enough to obtain the amount of the stoichiometric hydrogen flow of 320 Nml/injection at the injection frequency of 25 Hz, namely 3000 rpm. Further study should be done concerning the influence of the maximum lift. (6) The influence of the working fluid pressure on the hydrogen quantity per injection, the static leakage and the injection duration time should be studied coupled with engine control systems. (7) The influence of the working fluid temperature was small unexpectedly.

5. Acknowledgements The authors would like to express their appreciation to Mr. Takao Fukuma, Toyota Motor Corp., Mr. Hironori Iwamoto, Applied Electronics Corp., Mr. Takeshi Tanaka, Japan Air Ltd., Mr. Masayoshi Iwamoto, Iwamoto Ltd., Mr. Masayoshi Usui, Mr. Teruhisa Takahashi, Isao Yamaguchi of Usui Kokusai Sangyo Kaisha, Ltd. and Mr. Taizo Shimomura of Shimomura Seisakujo Ltd. for their considerate cooperation in this . In addition, the authors would like to declare that the hydrogen gas used in this experiment was filled at Ka- wasaki Hydrogen Station operated by Japan Air Gases Ltd. under the management by Engineering Ad- vancement Association of Japan in Hydrogen & Fuel Cell Demonstration Project (JHFC) organized by the Ministry of Economy, Trade and Industry.

References: 1. Shoichi Furuhama and Kimitaka Yamane, ”Combustion Characteristics of Hydrogen Fueled Spark Igni- tion Engine”, Bulletin of JSAE No. 6, p. 1, 1974 2. Shoichi Furuhama, Yoshiteru Enomoto and Yoshiyuki Kobayashi, ” Hydrogen Car with Two-Stroke Fuel Injection Spark Ignition Engine”, The 14th International Congress , A3-22, p.1, Sept. 1979 3. Shoichi Furuhama and Takao Fukuma, “Liquid Hydrogen Fueled Diesel Automobile with Liquid Hydro- gen Pump”, Advance in Cryogenic Engineering Vol. 31, p.1047, 1986 4. Masaaki Takiguchi, Shoichi Furuhama, Takayuki Suzuki and Makoto Tujita, “Combustion Improvement of Hydrogen Fueled Engine for Medium-Duty Trucks”, SAE Paper 870535, 1987 5. Katsuyoshi Koyanagi, Masaru Hiruma, Hiromasa Hashimoto, Kimitaka Yamane and Shoichi Furuhama, “Low NOx Emission Automobile Liquid Hydrogen Engine by Means of Dual Mixture Formation”, SAE Pa- per 930757, 1993 6. Tomohiro Fujita, Satoru Ozawa, Kimitaka Yamane, Yasuo Takagi, Yuichi Goto and Matsuo characteris- tics Odaka, “Performance of NOx Absorption 3-Way Catalysis Applied to a Hydrogen Fueled Engine €35, The International Association of Hydrogen Energy (IAHE), Hydrogen Energy Progress 10, The Proceed- ings of the 15th World Hydrogen Energy Conference, Yokohama, Japan, 27 June - 2 July, 2004, CD- ROM Issued by HESS, p.1-9, 2004

7/7