Introduction

The requirement of fossil fuel is increasing day by day, which is used for both domestic and commercial purposes. The increase in the use of fossil fuels results in the depletion of its existence. This induced the interest among researchers to search for an alternate fuel. Many such alternate solutions were achieved in the form of bio-oils. Using transesterification process, these bio-oils were converted into biodiesels. These biodiesels were used as straight fuels or blended with conventional diesel fuel. The biodiesels are limited as an alternative fuel to use in diesel engine because of its high density and viscosity, which resulted in low fuel atomization and the formation of unburnt fuels in the combustion chamber (Kocak et al. 2007; Shameer and Ramesh 2017a). Calophyllum inophyllum (CI) biodiesel has been proposed as a superior source due to its higher oil yield and heating value than other biodiesel sources like karanja, neem, and jatropha (Shameer and Ramesh 2017b).

On the other hand, alcohols exhibit lower exhaust emissions in engines while using as additives with diesel fuel due mainly by the oxygen in their composition (Lapuerta et al. 2008). The low sulfur content and higher oxygen content in the alcohols attribute the lower emissions of alcohol added diesel fuel. Still, the lower alcohols show some disadvantages such as phase separation on blending with diesel and reduced lower heating value (Eugene et al. 1984). The lower alcohols also show poor soluble and miscible properties on blending with diesel fuel. These problems were overcome by the use of higher alcohols as additives with diesel fuel, which shows better solubility (Yilmaz et al. 2017). Among such higher alcohols, n-pentanol possesses superior qualities like higher cetane number and latent heat of evaporation compared to other higher alcohols such as butanol and lower alcohols (Yilmaz and Atmanli 2017 and Lapuerta et al. 2010).

Many types of research were carried out using alcohol/biodiesel/diesel blends in various proportions and were tested in diesel engines with varying parameters such as injection timing, load, and compression ratio. However, it was difficult to find out the proportion of blend at which it gives better results. So the optimization techniques which give better performance and emission characteristics were utilized by the researchers efficiently and effectively.

Kesgin 2004 used Genetic algorithm and neural network in the prediction of operational parameters, load, and speed, in the engine efficiency and NOx emissions with the help of a natural gas engine. Increase in efficiency of the engine was achieved and the NOx emissions were maintained under 250 mg/Nm3. The effects of speed, load, and injection timing on the performance and pollution of the engine were studied by Win et al. 2005. The input parameters were varied as 4 × 4 × 3 full factorial design array. Performance and emission results were captured and taken as output responses. Using response surface methodology (RSM) the input and output parameters were related and the objective functions were arrived (Shameer and Ramesh 2017c). Artificial neural network (ANN) modeling has been handled by Sayin et al. 2007 to predict the performance, emission, and exhaust gas temperature of a gasoline engine at different engine speeds. The predicted ANN results were compared with the experimental results and the correlation coefficients were in the range of 0.983–0.996 with a low root mean square errors. It was shown that the ANN approach accurately predicts the performance and emissions of IC engines. Xu et al. 2017 experimented on a diesel engine fueled with diesel-Jatropha curcas blended with diesel fuel. The experiments were conducted in the order designed with the help of design of experiments. The design was formulated on the basis of the fractional factorial design of RSM (response surface methodology). The study results show that for optimal input parameters, the performance and emission values were achieved at a high desirability. Choudhary et al. 2015 investigated the combustion and performance of bioethanol blended with diesel in a diesel engine using RSM and compared the results with the experimental results. At higher compression ratio, the fuel injection pressure and engine loads, higher combustion pressure were produced by bioethanol diesel blends.

The evaluation of multiple variables with multiple responses can also be carried out easily in an efficient manner (Ileri et al. 2013). Less consumption time in RSM (Pandian et al. 2011) and its studies on the optimization of diesel engine parameters (Namvar et al. 2008) and also the determination of improved optimized values of BTE, BSFC, NO, and HC were widely studied and depict a clear picture of the role of RSM in IC engines.

In this study, optimization was carried out using design of experiments (DOE) and the optimized values related to performance and emission were determined in a DI diesel engine fuelled with pentanol/Calophyllum inophyllum/diesel blends by varying the compression ratio (CR), load (in bmep), and fuel blends.

The pentanol-biodiesel-diesel blends showed lower BSFC than CI20 (Calophyllum inophyllum 20%+diesel 80%) fuel blend and lower NOx and CO emissions than pure diesel fuel and CI20 biofuel blend with the addition of pentanol (Ramakrishnan et al. 2018). The optimization of the percentage of pentanol to be mixed with the biodiesel-diesel blend is required to improve the engine performance, emission, and combustion of the diesel engine. However many optimization works have been carried out in the performance, combustion, and emission characteristics of DI diesel engine using various fuel blends with diesel fuel, so far the use of pentanol/Calophyllum inophyllum/diesel blend fuel has not been carried out in RSM which enhances the novelty in the paper. Also, this optimization work is the extension of the previous work Ramakrishnan et al. 2018.

Experimental set-up

The experimental work was carried out using a Kirloskar single cylinder, water-cooled, constant speed, four-stroke, direct injection, computerized VCR research test engine available in the Thermal Engineering Laboratory, Department of Mechanical Engineering, Government College of Technology. The layout of the test engine is shown in Fig. 1. The specifications of the engine are given in Table 1. The test fuels used for the preparation of the blends were neat diesel fuel, CI biodiesel, and pentanol. The diesel fuel was purchased from the commercial market. CI oil was prepared from the collected CI seeds and the biodiesel was prepared from the oil using transesterification process as carried out by Shameer and Ramesh 2017d. Pentanol was purchased from The Precision Scientific Co., Coimbatore, Tamilnadu, India.

Fig. 1
figure 1

Layout of the experimental test bed of the variable compression ratio single cylinder computerized diesel engine

Full size image
Table 1 Engine specifications
Full size table

The fuel blends used in the test were: (i) P10C20 (Pentanol 10%+CI 20%+diesel 70%), (ii) P15C20 (Pentanol 15%+CI 20%+diesel 65%), and (iii) P20C20 (Pentanol 20%+CI 20%+diesel 60%). The fuel properties of the fuel blends were given in Table 2.

Table 2 Physico-chemical properties of the fuel blends
Full size table

The engine emissions CO, CO2, HC, and NOx were measured using AVL MDS 250 gas analyzer and the smoke emission was measured using AVL smoke meter. The specifications of the gas analyzer and smoke meter were given in Tables 3 and 4. The measurement accuracy and the maximum uncertainties of the instruments were given in Table 5.

Table 3 Technical specifications of AVL MDS 250 gas analyzer
Full size table
Table 4 Technical specifications of smoke meter AVL 437C
Full size table
Table 5 Measurement accuracy and maximum uncertainties of the instruments
Full size table

Test method

The engine used was a constant speed engine which runs at 1500 rpm. Three blends were used namely P10C20, P15C20, and P20C20 at three compression ratios 16:1, 17:1, and 18:1. The engine was operated by fuelling with diesel fuel for 5 min in order to stabilize the engine before using each blend. The performance responses were measured using the IC Engine Soft that is connected to the engine by means of a data acquisition card. The emission responses were measured using AVL Modular Diagnostic System 350. Each blend was poured in the engine tank and test runs were conducted at compression ratios of 16:1, 17:1, and 18:1 for three varying loads (1.07, 2.15, and 4.16 bmep). The engine test run was made three times with each blend, at each compression ratio and load for statistical analysis.

Response surface methodology

The input factors considered in this work were fuel blends, compression ratio, and load. These factors affect performance factors such as brake thermal efficiency (BTE) and brake specific fuel consumption (BSFC), and emission factors such as carbon monoxide (CO), carbon dioxide (CO2), hydrocarbon (HC), oxides of nitrogen (NOx), and smoke, which were taken as output responses. The RSM analysis was carried out using “design expert” trial version 11. The experimentation was carried out with each factor varied in levels of 3 × 3 × 3 (27 runs) using full factorial design. As per the run order that arrived from the design matrix, the experimentation works were carried out. Derringer’s desirability approach is the optimization technique handled in this work, where the highest desirable solution is taken as the optimum result (Hirkude and Padalkar 2014; Pandian et al. 2011). The corresponding combination of the input factors was considered as the optimized input parameters to operate the engine. The obtained solutions of input parameters were validated by performing confirmatory experiments.

Results and discussions

Analysis of the model

Analysis of variance (ANOVA) was used in RSM to develop and validate the models for the output responses BSFC, BTE, CO, CO2, HC, NOx, and smoke. For each output response, the predicted ‘p value’ is less than 0.05, which shows that the factors were having a significant effect of confidence level at 95%. The maximum difference between the ‘Pred-R2’ and ‘Adj-R2’ among all the output responses does not exceed 0.2, which gives the reasonable agreement between the predicted and adjusted values (Pamnani et al. 2015). The signal to noise ratio is measured as ‘Adeq precision’. This value is considered desirable if it is greater than 4. The ‘Adeq Precision’ of all the output responses was greater than 4 and thus indicates that the model prediction is adequately accurate (Ganapathy et al. 2011). The ANOVA model evaluation of the output responses was given in Table 6.

Table 6 ANOVA for output parameters
Full size table

Optimization and validation experiment

The optimization criteria for the input and output parameters were represented in Table 7. Here for the responses BTE and CO2, the goal was set as maximize and for remaining responses the goal was to minimize. The importance of each response was set as 3. The solution with the highest desirability was preferred as the optimized solution. The maximum desirability of 0.88 was obtained for the input engine parameters 17.5 CR, P20C20 blend, and 2.5 bmep load.

Table 7 Optimization criteria
Full size table

In order to validate the obtained optimized result from desirability approach, a validation experiment was carried out in the engine using the obtained optimized input parameters. Three experiments were carried out to evaluate the predicted output responses. The obtained output values from the validation experiments prove that the optimized model is correct as the experimental and predicted values agreed with each other which are given in Table 8.

Table 8 Validation experiments
Full size table

Brake specific fuel consumption (BSFC)

The interaction effects of the input response compression ratio and load on the BSFC for different blends were shown in Fig. 2a–c. It is evident from the plot, as the pentanol percentage increased in the blend, the BSFC was decreased. This is because of the oxygen content in the pentanol. Though the heating or calorific value of pentanol is lower than that of diesel fuel, the calorific value of the pentanol/Calophyllum inophyllum/diesel blends is almost equal to the diesel fuel (Table 2). So, rather the heating value of the pentanol, the oxygen content in it influences the BSFC. As the pentanol level increases in the blend, the oxygen content also increased which attributes reduced BSFC (Imtenan et al. 2014). With the increase in compression ratio, the BSFC was reduced. As the compression ratio increases, it may enhance the evaporation rate and so the combustion process (Sharma and Murugan 2015). The higher BSFC at lower compression ratio is because of the incomplete combustion. With respect to increasing load, the BSFC was decreased due to the improved combustion efficiency (Bharadwaz et al. 2016). The highest value of BSFC is 0.65 kg/kWh at 16:1 CR, P10C20 blend, and 1.07 bmep. The lowest value of BSFC is 0.36 kg/kWh at 16:1 CR, P20C20 blend, and 4.16 bmep. From this, it can be inferred that both the CR and pentanol level in the blend influence the specific fuel consumption.

Fig. 2
figure 2

Variation of brake specific fuel consumption against CR and load (a) at pentanol 10%, (b) at pentanol 15%, and (c) at pentanol 20%

Full size image

The regression equation for the BSFC is generated as follows,

$$ {\displaystyle \begin{array}{l}\mathrm{BSFC}\ \left(\mathrm{kg}/\mathrm{kWh}\right)=3.99-0.0083\ast A-0.33\ast B-0.39\ast C+1.67e-4\ast AB+3.12e-05\\ {}\ast AC+0.013\ast BC+6.67e-05\ast {A}^2+0.0083\ast {B}^2+0.019\ast {C}^2\end{array}} $$

Brake thermal efficiency (BTE)

BTE plotted against CR and load for three blends was shown in Fig. 3a–c. It can be seen from the plot that the BTE increases with increase in the pentanol percentage in the blend. The increased oxygen content in the pentanol fuel blend resulted in the wider flame formation which attributes the increase in BTE (Yilmaz and Sanchez 2012 and Anand et al. 2011). Thermal efficiency increases with the compression ratio can be explained by the increase in temperature and pressure inside the combustion chamber. Higher cylinder temperature and pressure result in good combustion and increase the efficiency of the engine (Dhingra et al. 2014). As the load increases the BTE of all the blend increases, this is because of the reduction in BSFC as the load increases. The BTE increases with increase in engine load due to the occurrence of efficient combustion at higher engine load (Zhang and Balasubramanian 2016). The highest value of BTE (23.8%) is achieved at 18:1 CR, P20C20 blend, and 4.16 bmep. The lowest value of BTE is 10.88% which is achieved at 16:1 CR, P10C20 blend, and 1.07 bmep. Thus, the BTE value increases with increase in all input responses. The regression equation for BTE is given as follows,

$$ {\displaystyle \begin{array}{l}\mathrm{BTE}\ \left(5\%\right)=0.71+0.09\ast A+0.061\ast B+5.78\ast C-0.011\ast AB+0.04\ast AC+\\ {}0.19\ast BC+0.003\ast {A}^2+0.004\ast {B}^2-1.14\ast {C}^2\end{array}} $$
Fig. 3
figure 3

Variation of brake thermal efficiency against CR and load (a) at pentanol 10%, (b) at pentanol 15%, and (c) at pentanol 20%

Full size image

Emission parameters

Nitric oxide (NOx) emissions

The NOx emissions from the engine cylinder rely on air-fuel ratio, amount of oxygen present in the intake air and also the in-cylinder temperature. The NOx plotted against CR and load for three blends was shown in Fig. 4a–c. It is noted from the plot that the level of NOx emissions was inversely proportional to all the three input parameters. The NOx emission increases with increase in pentanol percentage in the fuel blend because of the presence of the non-reactive nitrogen gas in the intake air-fuel mixture (Shameer and Ramesh 2017e), and the unburnt oxygen available in the combustion chamber reacted as per the Zeldovich mechanism forms NOx emissions (Sakthivel and Ramesh 2018). The lower NOx emission was observed as 288 ppm at 16:1 CR, P10C20 blend, and 1.07 bmep and higher NOx emission as 1325 ppm at 18:1 CR, P20C20 blend, and 4.16 bmep. The NOx formation is attributed by means of high temperature and pressure inside the engine cylinder combustion chamber which is obviously occurred at high load and CR (Sharma and Murugan 2015).

Fig. 4
figure 4

Variation of NOx against CR and load (a) at pentanol 10%, (b) at pentanol 15% (c), and at pentanol 20%

Full size image

The regression equation for the NOx emission generated is as follows,

$$ {\displaystyle \begin{array}{l}{\mathrm{NO}}_{\mathrm{x}}\left(\mathrm{ppm}\right)=7392.12-96.47\ast A-907.18\ast B+586.69\ast C+4.72\ast AB+5.05\ast AC+5.39\ast \\ {} BC+0.56\ast {A}^2+26.94\ast {B}^2-102.01\ast {C}^2\end{array}} $$

Carbon monoxide (CO) emissions

The variation of CO plotted against the CR and load for varying pentanol blends were depicted in Fig. 5a–c. The CO emission was decreased as the pentanol ratio was increased in the fuel blend. This is because the pentanol evaporates easily in the cylinder which resulted in reduced spray atomization length. Thus with higher pentanol ratio in the blend, the CO emission is reduced (Yao et al. 2010). The presence of more number of oxygen molecules in the modified blends results in the complete combustion. This attributes the reduction in CO emission as the pentanol level increases in the blend (Fattah et al. 2014; Qi et al. 2014). At peak load, as the CR increases, a drastic reduction in CO emission can be observed. The reason is that the air temperature inside the cylinder gets increased with the increase in CR which leads to the complete combustion of the fuel and reduced CO (Sakthivel et al. 2018; Kassaby and Nemitallah 2013). The higher CO emission is observed at 16:1 CR, P10C20 blend, and 4.16 bmep loads whereas the lower CO emission is observed at 18:1 CR, P20C20 blend, and 2.15 bmep loads. The regression equation of the CO emission is as follows,

$$ {\displaystyle \begin{array}{l}\mathrm{CO}\ \left(\%\right)=-25.54-0.091\ast A+3.16\ast B+0.63\ast C+0.006\ast AB-0.008\ast AC\\ {}0.059\ast BC\hbox{--} 2.44e-4\ast {A}^2-0.095\ast {B}^2+0.12\ast {C}^2\end{array}} $$
Fig. 5
figure 5

Variation of CO against CR and load (a) at pentanol 10%, (b) at pentanol 15%, and (c) at pentanol 20%

Full size image

Carbon dioxide (CO2) emissions

The CO2 emission is always inversely proportional to the CO emission. This is because the O2 molecules available in the mixture oxidize the CO molecules and become CO2 (http://nptel.ac.in/courses/112104033/pdf_lecture/lecture7.pdf). The plot in the Fig. 6a–c depicts the variation in CO2 emission against CR and load for three blends. As the compression ratio increases the CO2 emission is also increased. The complete combustion of the fuel at higher CR attributes the occurrence of CO2 emission. The reduced clearance volume in the engine cylinder at higher CR increases the temperature and pressure in the combustion chamber which increases the CO2 emission (Bora et al. 2014; Bora and Saha 2016). The higher CO2 occurs at 18:1 CR, P20C20 blend, and 4.16 bmep loads whereas lower CO2 at 16:1 CR, P10C20 blend, and 1.07 bmep. The regression equation of the CO2 emission is as follows,

$$ {\displaystyle \begin{array}{l}{\mathrm{CO}}_2\left(\%\right)=-25.72+0.47\ast A+2.01\ast B+5.26\ast C-0.025\ast AB+0.0052\ast AC-0.002\\ {}\ast BC\hbox{--} 6.31e-04\ast {A}^2-0.041\ast {B}^2-0.55\ast {C}^2\end{array}} $$
Fig. 6
figure 6

Variation of CO2 against CR and load (a) at pentanol 10%, (b) at pentanol 15%, and (c) at pentanol 20%

Full size image

Hydrocarbon (HC) emissions

The HC emission plot against CR and load for varying pentanol blends was shown in Fig. 7a–c. It is evident from the graph that the HC emission increases with increase in load for all percentage of blends and is decreased at moderate loads. The similar trends have been already observed in previous studies (Sahin and Aksu 2015; Chen et al. 2012b; Giakoumis et al. 2013; Rakopoulos et al. 2010). The increase in the pentanol percentage in the fuel blend also reduces the HC emission. The pentanol in the blend reduces the viscosity which improvises the spray formation and atomization of the fuel. This reduces the size of fuel droplets and leads to complete combustion of the fuel and reduced HC emissions (Chen et al. 2012a). The higher HC emission achieved was 304 ppm at 17:1 CR, P10C20 blend, and 1.07 bmep and the lower HC emission was 126 ppm at 17:1 CR, P20C20 blend, and 2.15 bmep loads.

Fig. 7
figure 7

Variation of HC against CR and load (a) at pentanol 10%, (b) at pentanol 15%, and (c) at pentanol 20%

Full size image

The regression equation for HC emission is,

$$ {\displaystyle \begin{array}{l}\mathrm{HC}\ \left(\mathrm{ppm}\right)=-2458.47-6.42\ast A+356.51\ast B-122.49\ast C+0.45\ast AB+0.96\ast AC+\\ {}1.97\ast BC-0.31\ast {A}^2-11.33\ast {B}^2+17.93\ast {C}^2\end{array}} $$

Smoke emissions

The smoke emissions plotted against CR and load for three blends was shown in Fig. 8a–c. It can be seen from the plot that smoke emissions follow the same trend as CO and HC emissions. The smoke emission is occurred due to the soot concentration and unburned HCs available in the exhaust gases (Teoh et al. 2013). At all pentanol blends, the smoke emission is reduced with increasing CR. At lower CR the combustion will be incomplete which leads to higher smoke emissions and with an increase in CR, the in-cylinder temperature and pressure will be raised results in complete combustion and lower smoke emissions (Shameer et al. 2016). Because of the availability of the oxygen in the pentanol, with the increase in the pentanol level in the fuel blend the smoke emission is reduced (Ramakrishnan et al. 2018). The reduced smoke emission as the pentanol percentage is increased is also attributed because of the lower density and lower viscosity values of the pentanol (Qi et al. 2014). The higher smoke emission is 34.4% observed at 16:1 CR, P10C20 blend, and 4.16 bmep and the lower smoke emission are 20.6% at 18:1 CR, P20C20 blend, and 2.15 bmep loads.

Fig. 8
figure 8

Variation of smoke against CR and load (a) at pentanol 10%, (b) at pentanol 15%, and (c) at pentanol 20%

Full size image

The regression equation for the smoke emission is given as,

$$ {\displaystyle \begin{array}{l}\mathrm{Smoke}\ \left(\%\right)=-19.74-0.32\ast A+9.11\ast B-6.22\ast C+0.023\ast AB+0.021\ast AC+\\ {}0.0089\ast BC-0.023\ast {A}^2-0.34\ast {B}^2+1.24\ast {C}^2\end{array}} $$

Conclusions

This study dealt with the optimization of the input parameters of a DI diesel engine with the use of pentanol/Calophyllum inophyllum/diesel blends as fuels to find the optimal input parameters of the engine using RSM. The optimized input parameters of the engine to get good performance and lowest emissions using pentanol/biodiesel/diesel fuel is found to be 17.5 CR, P20C20 blend, and 2.5 bmep load. The conclusions that are noted from the analysis are as follows:

  1. 1.

    The output responses BSFC, BTE, CO, CO2, HC, NOx, and smoke for this optimized input parameters are found to be 0.43 kg/kWh, 20.02%, 0.1%, 7.34%, 156.82 ppm, 1047.55 ppm, and 22.17%, respectively.

  2. 2.

    It is noted that the BSFC value is decreasing with increase in pentanol level in the blend whereas BTE value is increasing.

  3. 3.

    It is observed that the emissions CO, HC, and smoke are reducing with an increase in pentanol percentage in the blend whereas CO2 and NOx are increasing.

The RSM is demonstrated using design expert software to find the optimized input parameters to achieve the desired output parameters which give enhanced performance and emissions. The desired optimized input parameters were found and were used in carrying out validation experiment in the engine, which gives favorable results. Thus the RSM is found to be effective in finding optimized results in a multi-objective optimization of a diesel engine fuelled with pentanol/Calophyllum inophyllum/diesel blends.