Sie sind auf Seite 1von 8

CONSEIL INTERNATIONAL DES MACHINES A COMBUSTION

INTERNATIONAL COUNCIL ON COMBUSTION ENGINES

PAPER NO.: 22 Utilisation of cylinder air injection as a low load and load acceptance improver on a medium-speed diesel engine
Christer Wik, Wartsila Finland Oy, FINLAND Soren Hostman, Wartsila Finland Oy, FINLAND

Abstract: Development of engine concepts for lower NOx emissions e.g. by means of Miller valve timing (early inlet valve closure) makes loading capability worse, especially at low loads. Continuous increase of cylinder output makes the situation even worse; larger absolute load steps, as kW or bar BMEP, and larger turbochargers mean longer rotor acceleration and slower pressure increase. Furthermore, Miller timings demand higher charge air pressure, i.e. the pressure ratio capacity of the turbocharger must be greater. This causes the optimum efciency of turbocharger to move towards higher pressure and decreased efciency at low load which results in poor load response at low load. Future engine concepts will probably also include a shorter valve overlap (scavenge period), which also deteriorates low load performance. Poor load response is directly linked to high smoke and particle emissions. All this sum up in the fact that low load operation of state-of-art medium speed diesel engines is known to result in fairly high smoke emissions and thermal loads. This is a problem in transient operation and especially for auxiliary engines that need to be fast reacting generating sets. There are different means available to compensate for the transient problems, of which, air injection in different ways before the combustion starts is one. Air could be injected directly on the turbocharger compressor; so called air jet assist or into the air re-

ceiver. Both these methods, however, always give a certain time delay in load response situation, and the air receiver injection may also force the turbocharger to stall. There is one additional method that has potential in bringing large benets compared to the available methods mentioned above and this is injection of pressurized air directly into the cylinders. In this paper, focus will be put on air injection into the air receiver or into the cylinders. Preliminary transient and stationary tests aimed for proving the potential on a medium speed diesel engine have been performed utilising the existing starting air valves. These tests resulted in considerable reduction of smoke opacity during engine start-up as well as ability to run 2-step load application fullling classication criteria. Final outcome of the tests will be presented in the paper. Design of a production system for an auxiliary engine, with its challenges, will be presented together with rig test results for system optimisation, verication, and validation. Ultimate engine test results, proving the concept, will nally be reported upon availability. This project has been a part of the Tekes National Technology Agency of Finland, nanced LOSPAC project and performed in cooperation with VTT Technical Research Centre of Finland, Yrkeshoegskolan Novia, and CITEC Engineering.

c CIMAC Congress 2010, Bergen

INTRODUCTION
Development of engine concepts for lower NOx emissions e.g. by means of Miller valve timing (early inlet valve closure) makes loading capability worse, especially at low loads, unless a variable inlet valve closure (VIC) system is utilized. Continuous increase of cylinder output makes the situation even worse; larger absolute load steps, as kW power or bar BMEP, and larger turbochargers give longer rotor acceleration and slower charging pressure increase. Furthermore, Miller timings demand higher charge air pressure, i.e. the pressure ratio capacity of the turbocharger must be greater. This causes the optimum efficiency of turbocharger to move towards higher pressure and decreased efficiency at low load which results in poorer load response at low load. Future engine concepts will probably also include a shorter valve overlap (scavenge period), which also deteriorates low load performance. Poor load response is directly linked to high smoke and particle emissions. All this sum up in the fact that low load operation of state-of-art medium speed diesel engines is known to result in fairly high smoke emissions and thermal loads. This is a problem in transient operation and especially for auxiliary engines that need to be fast reacting generating sets with demanding transient requirements and often operating mostly close to harbours where visual environmental aspects like smoke emissions are most critical. There are different means available to improve the transient operation. Transient smoke is formed when there is not enough oxygen in the combustion chamber in order to oxygenise all the supplied fuel, or if the oxygen is not available where the fuel is sprayed. To improve this, you could for example decrease the amount of fuel injected by utilisation of advanced fuel governors or governor settings or simply by applying a slower load increase rate. You could also improve combustion by better ignition quality of the fuel at low loads with an increased air inlet temperature or improve the combustion chamber design to make sure that most of the oxygen molecules find a carbon atom. But the most efficient way is to increase the amount of air into the combustion chamber. Increasing the amount of air could be done by upboosting the turbocharger (TC) and thus increasing the receiver pressure but this will have very little effect at start-up, since the TC does not provide any significant pressure ratio at loads <10%. Another option would be to apply an auxiliary blower in front of the TC compressor for air supply before the TC CIMAC Congress 2010, Bergen

starts to deliver enough pressure ratio, but it would have a relatively long time delay. Application of a VIC system is known to improve both steady-state and transient operation with strong Miller cycle engines since when applied it keeps the inlet valve open a longer time during part load operation for increasing the trapped air amount and thus decreasing thermal load and smoke emissions [1]. Pressurised air could also be injected directly on the turbocharger compressor; so called air jet assist or alternatively into the air receiver. Both these methods, however, always give a certain time delay in the load response situation, and the air receiver injection may also force the turbocharger to stall. There is one additional air injection method, though, that has potential in bringing large benefits compared to the available methods mentioned above. This is injection of pressurized air directly into the cylinders. In this paper, focus will thus be put mainly on cylinder air injection (CAI). There are a lot of patent publications existing regarding the application of air injection means for steady state and transient performance improvements. The first ones are originating from 1979 and 1984 and could be found in references [2] and [3]. But activities in this field are ongoing also in present time with one of the latest publications relating to improved torque on automotive turbocharged diesel engine applications [4]. Regarding load acceptance, most classification societies approve the rules by IACS (International Association of Classification Societies) [5]. This states that application of an intermittent load in more than two steps (0 x y 100%) can be permitted, provided that conditions within the ships mains permit the use of more than two steps. This should also be allowed for already in the designing stage of the electrical systems onboard the ship. In case these requirements are not fulfilled, the demand is load application in two equal steps, 0 50 100%. Split of the load steps depends on the nominal BMEP of the engine, and can be calculated according to the following equation, where 100% corresponds to the nominal output of the engine.
0% [800/BMEP]% [800/BMEP+ (100 - 800/BMEP)]% 100%

(1)

The general demands for loading capability are briefly defined as following (see also figure 1): Momentary speed variation maximum 10% of the rated speed when the load is applied Recovery time within 5 s, meaning that the engine speed must return to within 1% of the nominal speed in this time 2

Paper No. 22

At sudden unloading from full load to 0 load, the speed recovery time (< 1% of nominal speed) is also 5 s, but the allowed momentary speed increase is 5%

Applying a 2% speed droop helps the situation somewhat.


Instant load application, requirements on speed variation and recovery time.
110 Max. allowed recovery time = 5 s

depending on air pressures used. This was due to the rather simple control setup including a solenoid valve, mounted as close to the starting air valve as possible and operated with a control air pressure to open the starting air valve fed with compressed air. The test setup used is seen in figure 3.

Actual recovery time (4.1 s) 105 "Speed droop" 100 Max. allowed speed reduction = initial speed - 10% Actual speed reduction (8%)

Recovered speed = final steady-state speed 1%

95

Maximum pressure of solenoid valves 10 bar

90 Instant loading x -> y% 85 2 3 4 5 6 7 Time (s) 8 9 10 11 12

Figure 1 General demands for loading capability

Starting air supply 20 bar

PRELIMINARY TESTS
Figure 3 Air injection test setup on WV4R32LN Target with the preliminary tests was to show the potential of injecting air directly into the cylinders. The test engine utilised was a Wrtsil Vasa 4R32LN engine operating on 750 rpm constant speed with BMEP 23.3 bar at maximum power output 410 kW / cylinder. The engine is located at VTT Technical Research Centre of Finland in Espoo. A standard starting air valve was utilised for air injection before and during the compression stroke. An example of timing for the air injection is viewed in figure 2. .
Exhaust TEVO 41 TEVC 27 ######### 112 Inlet TIVO 29 TIVC 5 ######### 146 ppnar stnger lngd stngd 11.4095 EVO IVO EVC IVC 139.36 330.77 387.24 544.69 ATDC ATDC ATDC ATDC

Injection timing variation was done within the allowable limits regarding in-cylinder pressure buildup vs. pressure of the injected air (see figure 4). Rising cylinder pressure closed the starting air valve, regardless of control air pressure, around 50 CA bef. TDC. As an outcome of the injection timing variation tests it was concluded that air injection timing is not critical.
WV32LN (CR = 13.8, TIVC = 3 deg a.BDC), p comp 25.0 CAI pressure (abs) 20.0

Pressure - bara

ppnar stnger lngd stngd

15.0

10.0

######### Scavenging period

41 - 29

- 27

- 5

5.0

EX valve timing Cylinder air injection timing

IN valve timing

0.0 -180 -160 -140 -120 -100 -80 -60 Crank angle degree

-40

-20

Figure 4 WV4R32LN compression pressure curve Figure 2 Typical air injection timing used Different signal opening timings (200330CA bef. TDC) as well as durations (135315CA) were tested in order to screen the importance of injection time. The control system used had a delay of 100130 CA from signal start to valve open start, CIMAC Congress 2010, Bergen In the tests different control air pressures were tried and injection activation time was always 5 seconds with injection start being simultaneously with load application. Charge air receiver air injection was also tested for comparison.

Paper No. 22

Both control air and starting air pressures affected how the starting air valve reacted and thus the injected air mass and smoke opacity was considerably influenced as can be seen from fig. 5. Solenoid valves used were restricted to a maximum pressure of 10 bar and with this control air pressure the effect was clearly best as it allowed most air mass into the cylinder. Effect on smoke opacity reached its maximum already with about 9 bar air pressure and the air mass 20 g/cycle.
(rpm) Engine speed (%)

Supply / control air pressure = 20 / 10 bar Opening 100 bef. BDC Closing 40 bef. TDC duration 240CA

Total air inj. time 5 s, pulse length 50 ms


VTT 4R32. Instant loading 0 - 50% with & without air injection. Speed drop & smoke opacity.

800

Engine speed

240

700 600 Ref., no air inj. Receiver Air Injection Cylinder Air Injection

200 160 Smoke opacity (%)

g/cycle 30 air mass (g) 25 20 15 10 5 0 0.0 2.0 4.0

500 Smoke opacity 400

120 80

300

40

200 0 2 4 6 8 Time (s) 10 12 14

6.0

8.0

10.0

12.0

control air pres (bar)

max opacity 120 100 80 % 60 40 20 0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 control pressure bar

Figure 7 Load acceptance and smoke opacity with cylinder / receiver air injection vs. reference Influence on the load steps 0-50% and 40-100% is summarised in table 1 and the huge improvement in performance fulfilling classification requirements for 2-step loading (0-50-100%) can be clearly seen. Table 1 Load acceptance results
Reference Receiver air inj. Cylinder air inj. Classification requirements

Figure 5 Control air pressure influence Based on the sensitivity checks, the starting air pressure with 10 bar control air could also be defined to ~22 bar as higher did not increase the flow anymore restrictions over the valve (see figure 6). optimum pressure pressure due to

0-50% Speed drop [%] Recovery time [s] Opacity [%] 40-100% Speed drop [%] Recovery time [s] Opacity [%]

16 11.9 100

9.3 8.6 100

6.0 4.2 5.5

Max. 10 Max. 5 N/A

15.1 10 88

N/A N/A N/A

6.5 3.9 7.0

Max. 10 Max. 5 N/A

Figure 6 Starting air pressure influence A comparison of load acceptance results for the load step 0-50% can be viewed in figure 7. There is a huge influence especially on smoke opacity with cylinder air injection vs. air injection into the receiver or without any air injection at all. The air injection setup used in the tests was the following: CIMAC Congress 2010, Bergen

At the load step 40-100%, cylinder air injection was set to cut off when the receiver pressure reached 2 bar, not to exceed the cylinder maximum pressures allowed. The huge reduction in smoke opacity becomes clear when looking at the influence on the fuel rack position in figure 8. The clear decrease from full fuel rack position of 53.5 mm down 24.2 mm clearly reduces excessive fuel entering into the cylinder and avoiding black smoke formation. 4

Paper No. 22

VTT 4R32. Instant loading 0 - 50% with & without air injection. Speed drop & fuel admission.

PM mass from ELPI (mg/m @ 15% [-] PM mass from ELPI @15% O2 O2)

140 1.4

800 700

Engine speed

120 100 Fuel rack pos. (mm) Ref., no air inj. Receiver Air Injection Cylinder Air Injection

1.2 120 1.0 100 0.8 80 0.6 60 0.4 40 0.2 20 0 0


n n n 1 2 op er at io op er at io 3 ri nj ri nj ri nj op er at io
op er at io n

(rpm) Engine speed (%)

600 500

80 60

Fuel rack position 400 300 40 20

200 0 2 4 6 8 Time (s) 10 12 14

op er at io

ec ti o n

No rm al

No rm al

No rm al

Figure 8 - Load acceptance and fuel rack position with cylinder / receiver air injection vs. reference Maximum possible instant load steps achieved, fulfilling classification rules regarding speed drop and recovery time vs. the reference engine without air injection was the following: Without CAI: 0-40% and 60-100%, i.e. 2step loading not possible With CAI: 0-60% and 30-100%, i.e. loading in 2 steps 0-50-100% possible
FSN

Figure 9 Particulate mass concentrations with CAI vs. references

0.7 0.6 0.5 0.4 0.3 0.2

Some steady-state results with CAI were also done in order to screen its potential even though the usage of such a system for steady-state operation would demand a huge compressed air capacity when considering a typical air consumption of around 200 g/s. The points investigated were all run at 10% load with the following air injection amounts: Air injection 1 = Injection duration 225 CA, starting @ 45 after TDC; 236g/s CA Air injection 2 = Injection duration 180 CA, starting @ 45 after TDC; 190g/s CA Air injection 3 = Injection duration 135 CA, starting @ 45 after TDC; 183g/s CA

0.1 0
n n 1 2 n op er at io op er at io op er at io ec ti o n ec ti o n ec ti o n 3

ri nj

No rm al

No rm al

ri nj

No rm al

ri nj

Figure 10 Smoke number with CAI vs. references A drawback with cylinder air injection at steadystate operation was seen to be a big increase in NOx emissions of 6075% as seen in figure 11.
25 Nox ISO 8178 corr. g/kWh

20

Looking at the particulate mass concentration results obtained with SMPS and ELPI it more than halved when additional air was injected into the cylinder but actually the smallest air amount yielded the lowest particulate mass emissions (see fig.9). Looking at the filter smoke numbers in figure 10, they reduced with up to 60% and here the increased air amount yielded even lower smoke values even though the influence was not big. So one could conclude that as short air injection duration as possible would be enough since this would also considerably reduce the air consumption.

15

10

0
Normal Air Normal Air Normal Air Normal operation injection operation injection operation injection operation 1 2 3

Figure 11 NOx emissions with CAI vs. references To conclude the preliminary test results, a clear advantage of applying CAI was found both for

CIMAC Congress 2010, Bergen

Paper No. 22

No rm al

Ai

Ai

Ai

No rm al

Ai

Ai

Ai

ec ti o n

ec ti o n

transient as well as for steady-state low load operation. Based on preliminary test results, a specification for the Wrtsil 20 engine needs was done as follows: Relative to cylinder volume of WV32LN minimum channel diameter in W20 should be 13.5 mm To get enough air into the cylinder with 20 bar injection pressure and 150CA injection time, minimum channel diameter is 11 mm Opening period (closed-closed) of the valve should be 2030 ms (=120180 CA) Total air demand would be about 3.2 kg per load acceptance for a 6-cylinder engine Figure 13 CAI system principal setup FE-calculations were done for checking of housing and valve strengths and ensuring enough bolt forces. Worst case in this system is occurring when valve impact happens at the seat. Valve closing, speed is less than 1 m/s, which is possible because of the airbrake included in the system. Moving masses are 0.182 kg which results in a force for impact calculations of 13.5 kN. This gives stress amplitudes of 80 MPa for the housing and 181 MPa for the valve of which only the one for the valve is critical. There is also a very small area where contact pressure is less than zero but this should not be any problem. Valve stress calculation results are shown in figure 14.

PRODUCTION SYSTEM DESIGN


Based on the preliminary test results it was decided to go forward with design of a production version for the smallest Wrtsil auxiliary engine; Wrtsil 20. Design and FE-calculations were done by CITEC Engineering and the final valve design in its intended location on the W20 engine could be seen in figure 12. The cylinder air injection valves have been integrated with the cylinder safety valves and the compressed air distribution arranged with a rail. A principal sketch of the pilot/control and supply air setup could be seen in figure 13. The valve block is mounted on the cylinder head and connected to the combustion space via a 9 mm diameter hole in the cylinder head. The valve stem is supported with a non-lubricated guiding area and accurate closing speed is ensured with a spring and an air brake. A flame arrester (point 10 in figure 13) was included for safety into the system in case combustion would move upwards in the cylinder head bore when the distribution valves are open.

Maximum Principal stress is 362 MPa

Solenoid valve Control air Supply air

Figure 14 CAI valve stress calculation results The outcome of FE calculations was that the valve is the most critical part with a failure probability of <0.00125 at a temperature of 200 which was C assumed in the fatigue calculations. Simulations were also done with a 1D code to see the potential air flows and influence on engine performances with CAI having the specification defined after the preliminary tests. The conclusion was that CAI could be used up to about 50%

Figure 12 CAI system arrangement on the W20 CIMAC Congress 2010, Bergen

Paper No. 22

steady-state load after which the maximum firing pressure will be exceeded. The average injected air flow was expected to be around 39 g/s with a supply air pressure of 20 bar.

RIG TEST RESULTS


In order to verify the design and validate the air injection valve before engine tests, rig tests were performed at Yrkeshgskolan Novia as well as at Wrtsil Finland Oy. In the rig tests, one target was to verify the valve movement at different control and distribution air pressures. The second target was to measure the actual forces in the valve stem by applying a strain gauge measurement on the same (see figure 15).

First rig tests at Yrkeshgskolan Novia were done with a control air pressure of 7 bar due to restrictions in the compressed air pressure levels. The outcome of the first tests were that the valve opening was very slow but applying fast control valves and large control air distribution pipes, the valve could be opened and closed within 30 ms, i.e. as targeted. Operation was very unstable, though. Force levels measured on the valve stem during closing was very low, only 1050 N at a closing speed of 1.1 m/s, and as such the reliability of the valve was not expected to be a problem. In total the valve run 150 hours in constant open-closing mode. In the second rig test series at Wrtsil Finland Oy, the target was to verify valve movement when applying >20 bar supply air pressure together with 6, 8, or 10 bar control air. The test setup of this could be seen in figure 17.

Figure 15 CAI valve equipped with strain gauges To verify that the valve actually opens, a laser sensor (Wrenglor YP11MVG) was adapted onto the valve disk in the first rig tests (see figure 16) whilst it was done with strain gauges on the valve spring in the second rig testing phase. A pulse generator was added for giving input to open / close the valve. Figure 17 CAI valve measurement setup 2 Applying 10 bar control air pressure and a 20 ms control signal as well as 20 bar air delivery pressure on the valve disc, we get an opening time of the valve (closed closed) of 2030 ms, i.e. within the specification (see figure 18).
20

Control air press. [bar]


18

Valve movement [mm]

Working air press. [bar]

16

14 Movement [mm], Pressure [bar]

12

10

0 2,2025

2,3025

2,4025

2,5025 Time [s]

2,6025

2,7025

2,8025

Figure 16 CAI valve measurement setup 1

Figure 18 Valve movement in rig measurements Paper No. 22 7

CIMAC Congress 2010, Bergen

CONCLUSIONS
Air injection in different forms, and especially Cylinder air injection, to aid transient behaviour of diesel engines have been investigated. CAI was found to be the by far most promising technology for transient behaviour improvement, enabling 2step loading fulfilling the classification society requirements. Load steps of 0-60% and 30-100% were possible on the test engine having a full load BMEP of 23.3. bar. With a CAI system, load application can also be done without any visible smoke and opacity measurements dropping from 100 to 5.5%. Considering air injection technologies also for steady-state operation, the competing technology would be variable inlet valve closure (VIC). Comparison of these different technologies is seen in table 2. Steady-state performance at low load includes improvements in smoke and particulate mass emissions. Table 2 Comparison of different technologies for load acceptance improvements
System VIC system Steady-state perf. Transient / low load loading NOx emissions Comment
DPP: No impact on guaranteed NOx (given for load range 70..100%. Marine: Impact on IMO cycle value only at 25% load point, provided switch to late mode < 50% load. Basically only for temporary use at instant loading as usage at steady-state would demand very big air amounts Basically only for temporary use at instant loading as usage at steady-state would demand very big air amounts

ACKNOWLEDGEMENTS
This project has been a part of the Tekes National Technology Agency of Finland, financed LOSPAC project and performed in cooperation with VTT Technical Research Centre of Finland, Yrkeshgskolan Novia, and CITEC Engineering.

REFERENCES
[1] C. Wik, H. Salminen, K. Hoyer, C. Mathey, S. Vgeli, P. Kyrtatos, 2-stage turbo charging on medium speed engines future supercharging on the new LERF-test facility, 14th Supercharging conference, Dresden, September 2009 [2] Patent publication nr. FR2396869, Procd et dispositif pour l'amlioration du fonctionnement des moteurs thermiques, Alain KERVAGORET, 197902-02, Socit Alsacienne de constructions mcaniques de Mulhouse. [3] Patent publication nr. GB2127095A, Turbocharged IC engine with additional charge air supply Sadao ARAKAWA, Takeaki NOZAKI, Yuji MATSUBARA, 1984-04-04, Yanmar Diesel Engine CO Ltd. [4] Manuel MARX, Huba NMETH, Eduard GERUM, Verbesserung des Drehmomentverhaltens aufgeladener Dieselmotoren durch Drucklufteinblasung, Knorr Bremse Systeme, MTZ 06/2009, pp. 472-479. [5] http://www.iacs.org.uk

++

Receiver air injection Cylinder air injection

(+)

(-)

(+++)

+++

(- -)

As a conclusion CAI is definitely the best system for improving transient behaviour of a medium-speed engine and due to this, a production version has been designed for the Wrtsil 20 auxiliary engine. Considering steady-state operation, VIC is still considered to be the primary technology.

BIBLIOGRAPHY
Christer Wik started as Research Engineer at Wrtsil in Vasa, Finland in 1997 and has after this been working in various expert and managerial positions at Wrtsil in France, Finland, and Italy. He is currently working as General Manager at Wrtsil Finland in Vasa, heading Wrtsils 4stroke performance expert organisation. Christer has a M.Sc. degree in Mechanical Engineering from Helsinki University of Technology as well as an MBA degree from Helsinki School of Economics. Sren Hstman started as Calculation & Simulation Expert at Wrtsil in Vasa, Finland in 1993 and has since then had various expert and managerial positions at the same location. Sren is now manager of the 4-stroke Cylinder Head & Valve Mechanism expert group at Wrtsil Finland in Vasa. Sren has a M.Sc. degree in Mechanical Engineering from Helsinki University of Technology.

NOMENCLATURE
BDC Bottom dead centre BMEP Break mean effective pressure CA Crank angle CAI Cylinder air injection FE Finite element TC Turbocharger TDC Top dead centre VIC Variable inlet valve closure

CIMAC Congress 2010, Bergen

Paper No. 22

Das könnte Ihnen auch gefallen