An experimental investigation of fuel injection timing on

ISME2005-15-2

論文

An experimental investigation of fuel injection timing on NO X emission of marine slow speed engines

Borkowski Tadeusz*, Listewnik Jerzy**

Japan Institution of Marine Engineering

At present, air pollution Annex to Marpol 73/78 convention is being compiled to regulate NOx and SOx emission

limits. The paper describes the application of a CFD combustion model - KIVAII3v for a large bore, slow speed

marine diesel engine and explores the feasibility of using fuel timing injection variables to estimate NOx exhaust

emissions levels, at steady running conditions. The correlation analysis between the measured and calculated

gaseous emission caused by fuel variable injection timing (VIT) was performed. The analysis is based on

experimental data from Wð rtsila RTA engines test bed trials. Extracted and averaged measured history of NOx,

and other gas components emission concentration were used as a comparison basis for calculation results. An

examination of the measured and calculated results, particularly NOx concentrations, generally reveals a

difference of emission levels associated with engine load and VIT setting.

1. Introduction

Air pollution by exhaust emission is still a serious problem, and international concern has been raised for its control and reduction. High peak temperatures can be seen in typical uncontrolled and controlled diesel engine designs. Controlling both NO X and PM emissions requires different, sometimes opposing strategies. To control both NO X and PM, manufacturers need to combine their approaches to achieve optimum

performance. Several parameters in the combustion chamber of a diesel engine affect its efficiency and emission. Retarded timing reduces N0X because the premixed burning phase is shortened and cylinder temperature and pressure are lowered. Unfortunately, timing retardation increases HC, CO, PM, and fuel consumption [1]. Nowadays, both shipyards and engine manufacturers are under increasing pressure to reduce fuel consumption while meeting tightening standards of NO emission. This paper investigates the results of an experiment targeting lowering the NO X emission and developing a better understanding of relationship, between the air boost variables, fuel injection timing and output performances including the fuel efficiency and emissions. As a next step, the effect of fuel oil injection timing has been investigated. The experiment included a rough numerical simulation performed using a CFD code - the KIVA II 3V. Finally, calculated results of the manifold air pressure and injection timing variables affecting engine performance and emission were compared to these investigated experimentally.

2. Experimental details

2.1 General description

The main objective was to decrease emission, with NO as the primary target, of the Wartsila 6RTA 58T-B engine. It is known that NO rate emission can be reduced when retarding fuel injection timing - causing the deterioration in combustion. However, assumptions were it would not be sufficient to meet current IMO limit. As N0X emission depends also on other factors including compression pressure or charge air pressure. Presumably, a higher charge air pressure would be needed as a solution and re-matching the turbocharger for 6RTA58T-B engine. This engine has been used as the test engine for examination of new configuration to lower NO emission. The first step of the experiment was to establish performance parameters with greatest

possible accuracy. Further, fuel timing advance approach test was performed, which was focused on obtaining attractive SFOC and affordable N0X emission factor, close to IMO limit. In case the NO X emission factor value proved to fail IMO limits, the second expected step was a turbocharger re-matching and further fuel timing adjustments. At that phase, comparing the results of CFD calculation against the experimental based results (third step of testing) was important to determine the benefits of CFD use. The second case of engine VIT tuning concerned another slow speed marine diesel - Wartsila 7RTA 62U where the assumption was to get affordable fuel consumption with regard to NO X limit. The test data obtained from this trial have been used to compare measured results* Maritime University of Szczecin

**Maritime University of Szczecin

Journal of the JIME Vol. 41, Special Issue (2006) 24 “ú–{ƒŠƒ“ƒGƒ“ƒWƒjƒAƒŠƒ“ƒOŠw‰ïŽ• ‘æ41™É ú•Š§•†(2006)

An experimental investigation of fuel injection timing on NOx emission of marine slow speed engines

and predicted NO,, concentration.Basic performance measurement procedure of marine

engines on test beds provides engine data to assess the respective engine operating conditions. Equally

essential are the exhaust gas emission, particularly nitrogen oxides and other gaseous components that have

to be registered using gas analyzers in accordance to Annex VI of Marpol 73/78. Measurements were carried

out on a test-bed for main ship propulsion engines -specification given in table 1, over a range of power

settings.

Table 1- Test engines data

The mode of engine operation was determined using the ship propulsion engine test cycle. The load of the engines was set at four levels - Wartsila 6RTA 58T-B and at six levels - Wartsila 7RTA 62U. For each load and speed setting the engine under test was allowed to stabilize, prior to recording of the emissions level and cylinder combustion pressure. The stable running conditions were established basing on data logging

profile. For each of the test levels, all engine performance data was recorded in as much detail as industrial test bed equipment would allow. During the test engines were running on distillate fuel ISO-F-DMA. Samples of the fuel being bum were taken at the time of the trial and analyzed in accordance with standard industry procedure. For this project, a marine diesel engine electronic indicator (Premet-Lemag 1 ) was chosen. The pressure transducer was mounted on the indicator cock for measurement and then moved from one cylinder to another in order to complete the measurement on all engine cylinders. Manufacturer fuel injection timing characteristic is treated as the baseline at certain stage of adjustment procedure (marked by 0.0 deg). During the experiment given CA degree of advance (marked by "+"deg) or retard (marked by "-"deg) has been established for the analysis of cylinder

combustion effects.

2.2 Calculation method

The computer code used in the study was KIVAII-3V

[2], which is an updated version. In computation used a block structured hexahedron grid with 370,000 cells. In the discussed application, the initial pressures and temperatures in the cylinder and ports were experimentally established. Since the spray atomization

plays a decisive role in the development of subsequent

processes, its accurate prediction is a key issue. The

existing Taylor Analogy Breakup (TAB) atomization

model in KIVA-3V code was used for atomization

(droplet breakup). At the nozzle exit initial drop

diameter are prescribed and equal to nozzle hole

diameter (manufacturer specification). Breakup (TAB),

require some tuning of constants based on the injector

geometry and operating conditions. Given this

information, along with the equivalence ratio for fuel-air

mixtures allowed the subroutine to calculate the initial

species densities as a check of input data. In order to

enable an emissions formation in terms of NOR, the

model has been coupled with the extended Zeldovich

model [3].

N2+0•¨NO+N (1)

k1f=7.6.1013 exp[-38000/T]cm3 /(mol•Es)

N+02•¨k2fNO+0

k2f=6.41O9 exp[-3150 /T]cm3 /mol s

(2)

N+OH•¨k3fNO+H

k3f=4.1•E1O13cm3 /(mol. s)

(3)

The formation rate of NO can be written as

d[NO]/dt=kl,f[N2][o]+ k2f[N][o2]+ k3,,[N][oH]

-kl,r [No][N] - k2,r [NO][0] - k3.r [NO][H]

(4)

After simplification

d[NO] = 2k1l,f[N2][O]-2kl,r [NO][N] (5)dt

All data is averaged over the cylinder volume and is

intended for combustion calculation check. For

combustion runs, the data include energy balance and emissions data. The emissions data contains the crank angle and the amounts of HC, 02, C02, CO, and NO in

the cylinder expressed in parts per million. Computation was started at the closing inlet port position and ended

at the exhaust valve opening. Initial thermodynamic and turbulence quantities were specified to be uniform in the

ports and the cylinder. The exhaust gas mass flow and combustion air consumption are based on exhaust gas

concentration and fuel consumption measurement. Universal method, known as carbon-oxygen balance,

which is applicable for fuels containing H, C, S, 0, N in known composition is used.

3. Results and discussion

Injection retardation is the simplest way of reducing NO emission in internal combustion engines. The resulting late combustion leads to lower mean temperatures and lower NO products, but reduced effective pressure results in increased specific fuell Lehmann & Michels GmbH & Co

. KG, Germany

Journal of the JIME Vol. 41, Special Issue (2006) 25 日本マリンエンジニアリング学会誌第41巻増刊号(2006)


consumption. The main advantages of injection retardation, especially regarding marine applications, is that it does not require any significant changes to the engine and is easy to adjust. Furthermore, in view of evolving engine management systems, the retardation could offer the flexibility of changing the emission characteristics during ship operation. Adverse side effect is that it elevates the exhaust temperature by 10-30 K, what is tolerable. Preliminary assumption concerning NO X emission level on 6RTA 58T-B engine was to achieve nominal performance and attractive SFOC. It was a base line for further steps in order to fulfil NO X emission limit. First stage of test cycles of 6RTA 58T-B engine show NO X emission characteristics presented in fig, 1.

Engine load [%]

Fig. 1 Comparison of measured NO X concentrations and SFOC of RTA engine with two fuel injection settings

In that phase of engine adjustment it has been decided to transfer fuel injection characteristic by increasing advance of injection timing with maximum value 1.5° CA at 75% effective load. As a result, weighted NO X specific emission factor for fuel injection settings showed difference - 0.33 g/kWh placed above IMO limit - presented in fig. 2 and 5.0 g/kWh SFOC saving. The second stage of engine performance modification involved a turbocharger re-matching. Turbocharging affects exhaust emission through its strong influence on the combustion process.

Fig.2 Weighted NO X emission factor for 6RTA58T-B

engine with two preliminary fuel injection settings

Since NO formation is governed by local temperature, oxygen availability and residence time, and to a lesser extent pressure, the primary controlling mechanism relays on changes in the pressure and temperature of the air trapped. An additional factor is important when NO X emission is assessed on a specific power basis. Considering the effect of air-fuel ratio, it has two effects: oxygen amount increase and combustion temperature decrease due to the energy absorbed by the extra air. Then, the effect of air/fuel ratio on NO formation through reducing combustion temperature exceeds the opposing effect of increasing oxygen availability. Thus, NO formation decreases as air /fuel ratio is increased (weakened), at a constant temperature

[3].Consequently, the re-matching of turbocharger on the 6RTA58 T-B engine was performed. The main engine turbocharger performance after characteristic correction is shown in table 2 [4]. Turbocharging correction has changed not only the pressures in inlet and outlet receivers but also significantly decreased specific fuel consumption in optimization range (EOP).

Table 2 - Basic turbocharging parameters of 6RTA 58T-B engine after turbocharger correction

The more important factor, NO X emission concentration, was improved and fell below the limit. Thus, it was

possible once again to check NO formation process sensitivity on fuel injection timing change. For that reason VIT characteristic was transferred in wide partial load range of the engine (25%, 50%, 75%) and excluded nominal rating by l.0° CA. Results obtained from test cycle are presented in fig. 3 and 4. Because of the engine adjustment in second phase of the experiment, weighted NO X emission factor improved by 1.22 g/kWh against the preliminary measurement and effectively engine was approved for further IMO certification. Specific fuel oil consumptions versus engine load characteristic become more differentiated, and the optimization range is visibly narrower.



Engine load [%]

Fig.3 Comparison of NOX concentrations and SFOC of

6RTA 58T -B engine, after turbocharger correction

Then a possibility to achieve better SFOC came into view by VIT adjustment. Fuel savings has been expressed in diagram - fig.3. It gives a place for the optimization field of expected engine service load. This manipulation gives higher NO X emission factor by 0.5

g/kWh (fig.4) and flat SFOC versus engine load characteristic, with fuel savings ranged from 7.2 g/kWh at partial load to 0.3 g/kWh.

Engine speed [rpm]

Fig.4 Weighted NO X emission factor for 6RTA58T-B engine after turbocharger correction with two fuel

injection settings

The last stage of the experiment, covered comparison of calculated and measured results. As a preliminary confirmation of boundary conditions set involved in computations the cylinder pressure history was used. Generally, throughout the entire engine load range covered by test cycle procedure the combustion pressures were predicted well. These are shown in fig. 5, as an example. Cylinder combustion pressure can be used to assess the condition of marine slow speed engines. Cylinder pressure data helps improve power balance by individual fuel injection timing adjustment. These aspects give the technical basis for condition-based maintenance program on which engine is scheduled and in the future - exhaust emission levels, particularly NO,, would be included.

Crank angle [deg]

Fig.5 Measured and predicted cylinder pressure history at 100% effective load - 6RTA 58T -B engine

Cycle calculations were performed for two VIT settings 0.0° and +0.5° CA (advanced) for full E3 test cycle effective load of the considered engine. One case of average cylinder pressures history is plotted in fig. 6. The predicted cylinder combustion pressures for all cases (two VIT positions) show the same trend. The ignition delay data show strong dependence on charge air temperatures below 1000 K at the time of injection. Throughout this temperature range there is an effect of

pressure at the time of injection on delay - the higher pressure the shorter delay.

Crank angle [deg]

Fig.6 Predicted cylinder pressure at 100% effective load - 6RTA 58T -B engine

The computed average maximum combustion pressure rise exhibits difference ranging from 0.22 to 0.60 MPa resulting from alteration of injection timing. The

predicted and measured engine average combustion pressures for all cases (two VIT positions and all cylinders) show the same trend, while all predicted maximum pressures exhibit overestimation. It supposedly was caused by different fuel oil characteristics. Thus, experienced ignition delay varied in tested engine more than expected in calculation boundaries. An example of comparison for cylinder

pressure calculation and measured data is shown in figure 7.



Fig.7 Comparison of calculated and predicted maximum

pressure difference-6RTA 58T-B engine

Averaged measured history of NOR, CO and THC emission concentration were used as a comparison basis for computed emission factors. To test the used calculation model for its capability of prediction, the operating conditions of the engine during experiments were recorded. These included engine load - torque and speed, and fuel consumption. Hence, for all comparisons between predicted results and experimental data, recorded data of engine condition parameters, speed and fuel consumption served as data inputs for computation. Figures 8 and 9 show comparison of the measured and predicted NO concentrations versus engine load ranging between 25-100%, with two VIT settings; 0.0° and +0.5° CA (advanced). An examination of the measured and predicted data of NO concentrations - reveals generally similar trends and difference of emission levels associated with engine load and VIT setting. Measured results show gradual decrease of NO concentration with load increase, while VIT correction (by +0.5°) equally increases the emission.

Engine load [%]

Fig.8 Measured and predicted NOR, concentrations over

full spectrum of engine load with base VIT setting

Engine load [%]

Fig.9 Measured and predicted NOR, concentrations over full spectrum of engine load with +0.5° CA VIT setting

Early combustion process causes higher NO formation rate, as combustion proceeds. After peak pressure occurrence (angle after TDC), gas temperature decreases as the piston moves down what freezes NO chemistry. Nearly final NO concentration is formed within the 20 CA following the start of combustion. Hence, combined effect of the equivalence ratio (defined by engine load) and injection timing determines the whole formation processThe predictive capability of the calculation model is

quite acceptable excluding 25-50% range of engine load. The value of NO concentrations discrepancies in measured and calculated results were similar for two VIT settings - at partial engine load NO calculated concentrations were underestimated, while, in contrast,

predicted values at high load - 100% Pe, were overestimated. All discrepancies in the results of described comparison procedure were probably due to insufficient precision of engine condition data. Particularly, this may apply to range of interpolated

parameters if the engine operating at steady condition was not properly defined and stated. There were also a number of variables that were not considered - which might have a large effect on them. Exemplary factors are fuel oil quality and related inlet parameters or lacks of accurate scavenge air data. Most significantly, varying air-fuel ratio caused the results to be very poor during periods of low boost pressure - low engine load. This demonstrates the necessity of collecting accurate data of particular engine for computation. The VIT setting manipulation on the engine results also in fuel

penalty. The comparison of most important weighted NO emission factors considering cycle calculation and measurement results are presented in fig. 10. Currently achieved calculated results represented by NO emission factor tend to damp the effects of VIT settings. This allows them to be used with limited confidence.



Engine speed [rpm]

Fig. 10 Predicted and measured NO emission IMO factor comparison with two fuel injection settings

Considering another case, engine E3 mode load operation, related raw profile of NO emission is shown in fig. 11 [5]. Effects such as higher level of NO concentration due to fuel injection advance, initially set to +3.0°CA, were similar in RTA58T B engine - fig. 12. As shown in fig. 12 NO emission differences expressed by weighted factor are strongly controlled by fuel injection timing.

Power effective [kW]

Fig. 11 Measured NOR, concentrations and SFOC with +3.0° CA VIT setting change (only 75-85% engine load)

Engine speed [rpm]

Fig. 12 Weighted NO emission factor comparison with two VII settings

4. CONCLUSIONS

The general IMO procedure was adopted with the aim to achieve compliant engine performance - with current environmental legislation. The CFD KIVA-3V combustion model, incorporating the Zeldovich mechanism of nitric oxide kinetics, has been tested for

predicting NO concentrations of marine slow speed diesel engines, while fuel injection timing was altered. The predictive capabilities of the model have been tested by means of experimental data over a full spectrum of engine load. These included few cases of fuel injection timing changes and additional minor turbocharger re-matching. The exhaust emissions were measured in accordance to IMO standards. The

predictive ability of the model has been validated satisfactorily only at 75% engine load. Partial engine loads emission predictions suffered discrepancies even though reasonable preliminary agreement between calculated and measured cylinder pressure histories was achieved. Most likely, employed CFD program

predicted cylinder pressure with shorter ignition delay than that obtained by measurement, which caused higher maximum combustion pressure. The other discrepancy for the predicted engine performance was NO emission concentration. The used CFD combustion model code was restricted to calculate a single combustion cycle. The procedure of validating the CFD results was based on using the engine load and in-cylinder parameters of the test engine to tune the solution and fix all parameters to

predict the exhaust emission. Future efforts need to be done to improve the accuracy of the model inputs. Conclusively, experimental results show the possibility of VIT alteration to achieve best engine configuration in terms of SFOC value and acceptable level of NO emission. The recommended range of VIT settings expressed in terms of final NO emission for whole engine load test as weighted factor could be as high as 1.0 glkWh. If necessity to reach lower NO emission factor will arise, another solution than VIT change will have to be found.



References

[1] •gInternal combustion engine fundamentals•h, HEYWOOD J. B.,

McGraw-Hill, Inc. 1988

[2] •gA KIVA Program for Engines with Vertical or Canted Valves•h,

AMSDEN, Los Alamos National Lab., LA-13313-MS, 1997

[3] •gModeling Engine Spray and Combustion processes•h,

STIESCH G, Springer Verlag 2003

[4] •gExhaust gas emission measurement-Wartsila NSD

6RTA58T-B, acceptance test•h, BORKOWSKI T., ZIMNICKI B.

R&D Centre H. Cegielski Poznan, pp. 38, 2000

[5] •gExhaust gas emission measurement - Wð rtsilð NSD 7RTA62U,

acceptance test•h, BORKOWSKI T., ZIMNICKI B., R&D Centre H.

Cegielski Poznan, pp. 32, 2001

Authors

Borkowski Tadeusz

Born in 1955

Maritime University of Szczecin


Marine Engineering, Marine diesel

engines

Listewnik Jerry

Born in 1933


Technical University of Gdansk

Marine Engineering, Marine diesel

engines

Journal of the JIME Vol. 41, Special Issue (2006; 30 日本マリンエンジニアリング学会誌第41巻増刊号(2006)

Documents

An experimental investigation of fuel injection timing on