Factors influencing NO_x emission of a stationary diesel engine fuelled with crude rice bran oil methyl ester blend – Taguchi approach

Abstract

The main objective of this work is to control the NO_x emission of a stationary diesel engine fuelled with crude rice bran oil methyl ester blend with less sacrifice on smoke density and brake thermal efficiency (BTE) and also to investigate the factors influencing the objective. Fuel injection timing, percentage of exhaust gas recirculation and fuel injection pressure are chosen as the promising factors for the objective and NO_x emission, smoke density and BTE are considered as response variables. Tests were conducted as per Taguchi’s L₉ orthogonal array and the most influencing factor for each response variable and also the significance of each factor on the same was found out through analysis of variance (ANOVA). Response graph was drawn for each response variable and from the results of response graph and ANOVA the optimum combination of the factor levels in achieving the objective was obtained and the same was confirmed experimentally.

Keywords:

• Diesel engine
• NO_x
• smoke density
• ANOVA
• Taguchi’s orthogonal array

1. Introduction

Biodiesel, a promising renewable source of energy which is derived from vegetable oils has been paid more attention for the past two decades to meet the growing demand of the petroleum diesel. Apart from various advantages, its NO_x emission of the engine when fuelled with biodiesel was comparatively higher than that of diesel (Forson, Oduro, and Hammond-Donkoh 2004; Machacon et al. 2001; Masjuki et al. 2001; Narayana Reddy and Ramesh 2006; Nwafor and Rice 1996; Pryor et al. 1983; Rajasekar et al. 2010; Ramadhas, Jayaraj, and Muraleedharan 2005; Sharma et al. 2005) which limits its commercialization in market. NO_x reduction by combustion modification through retarded injection timing, exhaust gas recirculation (EGR) and water injection (Henein 1976) as applied for diesel were also attempted for biodiesel fuelled engine (Abd-Alla 2002; Bari, Yu, and Lim 2004; Mani and Nagarajan 2009; Pradeep and Sharma 2007; Saleh 2009; Sayin and Canakci 2009; Tsolakisa et al. 2007). NO_x reduction through modification of combustion process is the most economical method (Sarofim and Flagan 1976) when compared with treatment of exhaust gases with the help of different catalysts (Gilot, Guyon, and Stanmore 1997; Kummer 1979; Morimune, Yamaguchi, and Yasukawa 1998; Rosenberg et al. 1980; Takami et al. 1997) and the later method can be considered when the emission standards cannot be met by the combustion process modification alone. Researches conducted for the NO_x reduction of diesel and biodiesel through retardation of fuel injection and EGR reported that the reduction in NO_x emission is accompanied by increase in smoke density and decrease in brake thermal efficiency (BTE).

Rajasekar et al. (2010) reported that for a retardation of three crank angle degree (CAD) NO_x emission of biodiesel reduced significantly with drastic increase in smoke emission. For the same retardation angle Tsolakisa et al. (2007) achieved 16% reduction in NO_x with 20% increase in smoke emission for rapeseed methyl ester. Cooled EGR has been proved to be a very effective NO_x reduction technique (Sarofim and Flagan 1976) which reduces the peak flame temperature and oxygen partial pressure in the initial part of the flame and reduces the NO_x formation (Hayhurst and Vince 1980). Research work conducted on biodiesel with 15% EGR achieved a NO_x reduction of 74% with 20% increase in smoke (Rajasekar et al. 2010). It was also reported that increasing the EGR by more than 15% will result in increase in smoke emission and fuel consumption (Pradeep and Sharma 2007). Hence there is a need for an optimization of fuel injection timing and percentage of EGR to reduce the NO_x emission without increasing smoke emission and fuel consumption (Henein 1976). It was also inferred that among the fuel injection timing and EGR, the most influencing factor in the NO_x control has to be studied. In the present work, fuel injection timing and percentage EGR is simultaneously varied to investigate the effect on NO_x control of a stationary diesel engine fuelled with biodiesel blend. Since fuel injection pressure also plays an important role in IC engine combustion, it also varied in combination with injection timing and percentage EGR.

Every country concerned about utilization of vegetable oil for biodiesel production without affecting food industry. In the present scenario, attention has to be focused on non-edible vegetable oils extracted from the seeds cultivated from the land not suitable for agriculture (Saravanan et al. 2010). Biodiesel used in this investigation is crude rice bran oil methyl ester (CRBME) derived from high free fatty acid (FFA) crude rice bran oil (CRBO) which is a non-edible vegetable oil derived from rice bran. CRBO with high FFA content is a non-edible vegetable oil which can be utilized as a feed stock for biodiesel production and the obtained CRBME can be utilized in CI engine as an alternate to diesel fuel (Saravanan et al. 2010). Earlier research works on biodiesel indicated that B20 (20% of biodiesel mixed with 80% of diesel in volume basis) will be an optimum fuel blend for CI engine rather than neat biodiesel (Kalligeros et al. 2003; Pradeep and Sharma 2005; Win Lee, Herage, and Young 2004). Blending biodiesel with diesel minimizes the property differences between diesel and biodiesel. B20 is popular because it represents a good balance of cost, emissions, cold weather performance, materials compatibility and solvency. B20 is also the minimum blend level that can be used for the Energy Policy Act 1992 compliance for covered fleets (US department of Energy 2006). Hence, the present investigation is carried out using CRBME as a CI engine fuel in blended form (B20). CRBME blend has comparable properties as that of diesel and its combustion characteristics are similar with that of diesel (Saravanan et al. 2010). Properties of CRBME blend compared with diesel are given in Table 1.

Table 1. Properties of CRBME blend and diesel.

Fuel property	Testing method	Diesel	CRBME blend
Viscosity at 40 °C (mm^2/s)	Redwood viscometer	3.522	3.62
Flash point (°C)	ASTM D 92	70	80
Calorific value (kJ/kg)	ASTM D 240-02	43,356	42,455
Distillation temperature T90 (°C)	ASTM D86	335	337
Specific gravity	–	0.8	0.845
Calculated cetane index	ASTM D 4737	47	48

CRBME blend was tested successfully in stationary and automotive diesel engine (Saravanan et al. 2010). From the earlier research works on CRBME blend (Saravanan et al. 2010), it was found that CRBME blend has a potential to replace diesel oil in its neat and blended form and its NO_x emission was higher than that of diesel. Earlier research work by the authors on optimization process (Saravanan et al. 2011) used diesel as a fuel. In this work, an attempt was made to determine the optimum combination level of factors for CRBME blend.

The main objective of this work is:

(1)	To investigate the combined effect of fuel injection timing, percentage EGR and fuel injection pressure in reducing the NOx emission of diesel engine fuelled with biodiesel blend.
(2)	To investigate the most influencing factor in reducing the NOx emission, smoke density and BTE of biodiesel blend.
(3)	To find an optimum combination of injection timing, percentage EGR and fuel injection pressure in reducing the NOx emission with minimum sacrifice of smoke density and BTE.

2. Experimental programme

2.1. Taguchi design

Design of experiments method was employed to design the experiments and fuel injection timing, percentage EGR (by volume) and fuel injection pressure are considered as the factors influencing the objective. Three levels were chosen in each factor to study the significant effects of these factors on the set objective. Retarded and advanced fuel injection angle was taken as 2.5 CAD since further increase in angle will increase the smoke density and NO_x emission, respectively (Bari, Yu, and Lim 2004). Hence in addition to standard injection timing, retarded and advanced angle of 2.5 CAD was taken as the three levels of the injection timing. The upper level for percentage EGR was fixed as 15 and within that 0 and 10 have been chosen as the other two levels. For stationary diesel engine, it was suggested that the fuel injection pressure has to be maintained within 250 bar for smooth operation of the engine (Jindal et al. 2010). Hence by fixing 250 bar as upper limit, two more levels were chosen within that including standard pressure. The three levels of the chosen factors are given in Table 2.

Table 2. Factors influencing the objective with chosen levels.

Factor No.	Factors influencing the objective	Level of factors
		1	2	3
1	Injection timing	Standard timing	Advanced timing (by 2.5 CAD)	Retarded timing (by 2.5 CAD)
2	Percentage of EGR	0	10	15
3	Injection pressure	Normal pressure (200–210) bar	(220–230) bar	(240–250) bar

2.1.1. Taguchi orthogonal array

In full factorial experiment for three factors with three levels, the number of experiments will be 3³ = 27. To reduce the number of experiments to be conducted, experiments were designed using Taguchi orthogonal array (OA) technique. For more than two numbers of three level factors the recommended OA is L₉ (Rose 1988), which is given in Table 3. In Table 3, column 1 indicates the levels of factor 1 (fuel injection timing), column 2 indicates the levels of factor 2 (percentage EGR) and column 3 indicates the levels of factor 3 (fuel injection pressure).

Table 3. L₉ orthogonal array.

Trial No.	Column 1	Column 2	Column 3
1	1	1	1
2	1	2	2
3	1	3	3
4	2	1	2
5	2	2	3
6	2	3	1
7	3	1	3
8	3	2	1
9	3	3	2

2.2. Experimental set-up

Schematic of the experimental set-up is shown in Figure 1. The technical specifications of the engine used in this investigation are given in Table 4. A swinging field electrical dynamometer was used to apply the load on the engine. This electrical dynamometer consisted of a 5-kVA AC alternator (220 V and 1500 rpm) mounted on bearings and on a rigid frame for the swinging field-type loading. The output power was obtained by accurately measuring the reaction torque by a strain gauge-type load cell. A water rheostat with an adjustable depth of immersion electrode was provided to dissipate the power generated.

Figure 1. Experimental set-up.

Table 4. Specifications of engine.

Make	Kirloskar
Model	TAF 1
Type	Direct injection, air cooled
Bore × Stroke (mm)	87.5 × 110
Compression ratio	17.5:1
Cubic capacity	0.661 l
Rated power	4.4 kW
Rated speed	1500 rpm
Start of injection	23.4 CAD bTDC
Injector operating pressure	200–205 bar

2.2.1. EGR system

Schematic arrangement of EGR system is shown in the Figure 1. It has a piping arrangement of length 8 m to tap the exhaust gases from the exhaust pipe and to connect it into the inlet air flow passage. The exhaust gases were tapped from a point in the exhaust pipe which is 10 m away from the engine. This reduced the temperature of the exhaust gases admitted into the inlet manifold approximately equal to that of the ambient air which eliminates any additional cooling requirements. A control was provided in the pipeline to control the flow rate of exhaust gases and the mixture of exhaust gases and fresh air were admitted into the inlet manifold. Pressure difference in the ‘U’ tube manometer was used to obtain the volume of air replaced by exhaust gases from which the percentage EGR was calculated in volume basis (Agarwal, Sinha, and Agarwal 2006; Hayhurst and Vince 1980).

Percentage EGR was calculated using Equation (1):(1)

2.2.2. Injection timing and injection pressure

Injection timing was changed by changing the thickness of advance shim. The spring tension of the injector needle with setting screw was varied to get the different fuel injection pressure.

2.3. Testing procedure

Tests were conducted on the engine, with the selected factors at different levels to determine the effect of the factors on the objective. The engine was operated at nine times with the combinations of the different levels of the influencing factors as given in Table 2. Two replicates were conducted for each trial and the order of the trial was selected randomly. At each trial, the engine was tested at different loads and at each load the responses (NO_x emission in ppm, smoke density in mg/m³, time taken for fuel consumption in sec) were measured. NO_x emission was measured with MRU 1600 exhaust gas analyzer and the smoke density was measured with AVL smoke meter.

2.4. Error analysis

The errors associated with various measurements and in calculations of performance parameters are computed in this section. The maximum possible errors in BTE were estimated using the method proposed by Moffat (1985). Errors were estimated from the minimum values of output and the accuracy of the instrument. This method is based on specification of the uncertainties in various experimental measurements.

If an estimated quantity, S depends on independent variables like (x₁, x₂, x₃ … x_n) then the error in the value of “S” is given by:

where , , etc. are the errors in the independent variables; ∂x₁ = Accuracy of the measuring instrument; x₁ = Minimum value of the output measured.

2.4.1. Errors in BTE and exhaust gas emissions

Brake specific fuel consumption was calculated from the fuel consumption and BTE. The maximum possible error in the calculation of BTE:

The exhaust gas emissions are measured using exhaust gas analyzers and smoke meter. As per the specifications of the analyzer the maximum possible error in the measurement of smoke density and NO_x emission is ±5%.

3. Analysis of data

Obtained responses in each trial were analysed through analysis of variance (ANOVA) and results were tabulated to determine the significance and contribution of the selected factors in achieving the objective. Response graph was drawn for each response variable to determine the combination of the factors in achieving the objective and based on the response graph and ANOVA table the optimum level of combination was arrived.

3.1. Analysis of variance

ANOVA is a statistical method used to interpret experimental data and make the necessary decisions. The total variability of the NO_x emission, smoke density and BTE is measured by the sums of squares of those values using Equation (2) (Rose 1988).(2)

In the above formula, N is the total number of experiments, T is the sum of all experiments response variable and y_i is the ith response variable. The total sum of squares includes the sum of squares due to each factor (SS_f) and the sum of squares of errors (SS_e). The ratio of SS_f to SS_T is the percentage contribution (P) by the factor. F test has to be performed to identify the significant effect of each factor on the response variables. F test is the ratio of mean squared factor (MSF) to the mean squared error (SS_em), where MSF is equal to the SS_f divided by the number of degree of freedom (DF) associated with the factors. The larger the F value, the greater the effect on the response variable due to the change in the factor.

3.2. Verification

Confirmation experiment was conducted to verify the optimum combination obtained through ANOVA and response graph. Response variable of the confirmation experiment was verified by comparing it with the variables of normal operating conditions.

4. Results and discussion

4.1. ANOVA table

This chapter discusses the ANOVA table for the experiments conducted.

Tables 5–7 show the ANOVA table for the response variables for the tested OA. It can be seen that the calculated F value of injection timing and percentage EGR is much higher than the tabulated F value for all the response variables which shows their significance in controlling the chosen response variables of CRBME blend. It is well known that modification in the fuel injection timing and recycling a portion of exhaust gases alter the maximum gas temperature attained in the cylinder. Since the rate of NO_x formation and smoke density are function of combustion temperature, fuel injection timing and percentage EGR play significant role in controlling the NO_x and smoke density of CRBME blend as obtained through testing and ANOVA. Reduction in peak combustion temperature also reduced the availability of heat for conversion into useful work which has an effect on BTE. It can also be seen that fuel injection timing is the most influencing factor for NO_x emission and smoke density while percentage EGR shows its most influence for thermal efficiency. Atomization and vaporization of fuel have major influence on NO_x and smoke formation. Modification of fuel injection timing will affect the atomization and vaporization and hence the formation of NO_x and smoke. From the Tables 5–7, it was observed that fuel injection pressure is less significant for the chosen response variables of CRBME blend since its F_cal value is marginally lower than the F_tab.

Table 5. ANOVA table for NO_x emission.

Factor	SS	DF	MSF	Fcal	Ftab	P
Injection timing (SSf)	191,792	2	95,896.02	279.1	3.89	0.69
Percentage EGR (SSf)	78,569.8	2	39,284.9	114.3	3.89	0.28
Injection pressure (SSf)	2139.08	2	1069.542	3.1	3.89	0.007
Error (SSe)	3779.5	11	343.5871			0.014
Total (SST)	276,280.4	17

Table 6. ANOVA table for smoke density.

Factor	SS	DF	MSF	Fcal	Ftab	P
Injection timing (SSf)	3547.3	2	1773.6	9.3	3.89	0.42
Percentage EGR (SSf)	2103.8	2	1051.9	5.5	3.89	0.25
Injection pressure (SSf)	651.8	2	325.9	1.7	3.89	0.078
Error (SSe)	2096.2	11	190.6			0.25
Total (SST)	8399.2	17

Table 7. ANOVA table for BTE.

Factor	SS	DF	MSF	Fcal	Ftab	P
Injection timing (SSf)	0.34	2	0.17	3.9	3.89	0.10
Percentage EGR (SSf)	2.25	2	1.1	25.85	3.89	0.68
Injection pressure (SSf)	0.21	2	0.11	2.474	3.89	0.065
Error (SSe)	0.48	11	0.043			0.14
Total (SST)	3.28	17

4.2. Response graph

Response graphs for NO_x emission, smoke density and thermal efficiency at various levels of the chosen factors are shown in Figures 2–4.

Figure 2. Response graph for NO_x emission.

Figure 3. Response graph for smoke density.

Figure 4. Response graph for BTE.

It can be seen that combination 3-3-1 (third level in the injection timing, third level in the percentage EGR and first level in the injection pressure) and combination 2-1-2 give the lesser NO_x emission and smoke density, respectively. Optimum level for each factor has to be chosen based on the most influencing factor for the each response variable as indicated in the ANOVA Tables 5–7. It was observed from Table 5 that injection timing is the most influencing factor for NO_x emission and from the Figure 2 third level injection timing at which the NO_x emission is minimum is the optimum level for the same.

As reported by many investigators (Bari, Yu, and Lim 2004; Mani and Nagarajan 2009; Rajasekar et al. 2010) in their experimental work, modification of fuel injection timing resulted in considerable change in the NO_x emission of the engine fuelled with CRBME blend and the same was confirmed through ANOVA that fuel injection timing is the most influencing factor for controlling NO_x emission of the engine.

Modification of injection timing also modifies the start of combustion. This altered the distance travelled by the piston after the initiation of combustion during the end of compression stroke. This modifies the combustion temperature and hence NO_x and smoke emission and BTE. These parameters are also affected with percentage EGR and fuel injection pressure as the availability of oxygen is reduced with increase in percentage of EGR and fuel atomization is affected with fuel particle size. From the Figures 2 and 3, it is clear that at advanced injection timing and lower percentage of EGR, NO_x emission is higher with reduced smoke and at retarded injection timing and higher percentage of EGR it is reverse. At higher injection pressure fuel particle size will be less which resulted in an improved combustion and higher NO_x emission which is confirmed from Figure 2. An increase in smoke emission with injection pressure also observed in Figure 3 and this may due to the counter effect of the associated EGR. An increase in BTE is observed with EGR in Figure 4 is due to the reburning of unburnt hydrocarbons present in EGR (Saleh 2009).

It can be seen from Tables 5–7 that percentage EGR influences significantly for NO_x and smoke density and also most influencing factor for thermal efficiency. As EGR reduced the availability of heat for conversion into useful work, thermal efficiency of the engine restricts the maximum percentage of EGR circulated into the combustion chamber (Abd-Alla 2002; Henein 1976; Rajasekar et al. 2010). Hence, from the Figure 4, the second level of percentage EGR which has higher thermal efficiency is chosen has the optimum level for percentage EGR. It can also be seen from the Tables 5–7 that the injection pressure is less significant for all the response variables and hence the first level of injection pressure is chosen as its optimum level at which NO_x emission is lower, which is the objective of the present investigation. From the results of ANOVA and response graph, the combination 3-2-1 which is the eighth trial of the Table 3 is chosen as the optimum combination to reduce the NO_x with minimum increase in smoke density.

4.3. Response variables at optimized condition

Table 8 shows the comparison of response variables at optimum combination with those of the normal operating conditions.

Table 8. Effect of optimization on response variables.

Normal operating condition			Optimized condition				% Change
NOx (ppm)	Smoke (mg/m^3)	BTE (%)	Combination	NOx (ppm)	Smoke (mg/m^3)	BTE (%)	NOx (Decrease)	Smoke (Increase)	BTE (Increase)
756.6	54.6	16.51	3-2-1	497.6	76.4	17.49	34.23	39.9	5.9584

It can be seen that the combination 3-2-1 shows reduction in NO_x emission with increase in smoke density and BTE. As the injection timing retards, start of combustion and hence the combustion process also retard which retards the occurrence of peak pressure. This decreases the peak combustion temperature attained in the cylinder which resulted in lower NO_x emission and higher smoke density. As a result of the combined effect BTE of the engine was not decreased and shows a 5.9% increase in thermal efficiency. This was due to the reburning of unburnt hydrocarbons present in the EGR (Saleh 2009). Hence, the retarded injection timing, with 10% EGR at standard injection pressure is the optimum combination for lower NO_x emission with lower smoke density and higher BTE for CRBME blend.

5. Conclusion

In the present work, the most influencing factor in controlling the NO_x emission, smoke density and BTE of CRBME blend as a stationary CI engine fuel was obtained by employing ANOVA. With the help ANOVA and response graph, optimum combination of injection timing, percentage EGR and fuel injection pressure in reducing the NO_x emission with less sacrifice on smoke density and BTE was arrived. From the experiment results and ANOVA following conclusions are drawn:

(1)	NOx emission can be reduced with lower trade-off with smoke and BTE through combination of injection timing, percentage EGR and injection pressure.
(2)	Reduction in NOx emission is independent of a single factor and also depends upon the levels of other factors.
(3)	Fuel injection timing is the most influencing factor in reducing the NOx emission and smoke density of CRBME blend.
(4)	Percentage EGR is the most influencing factor in controlling the BTE of the engine fuelled with CRBME blend.
(5)	Retarded injection timing with 10% EGR at standard injection pressure is the optimum combination for controlling the NOx with less sacrifice on smoke density and BTE.

This work can be extended by including operating parameters of engine and optimized combination can be suggested.

Disclosure statement

No potential conflict of interest was reported by the authors.