Performance and Emission Optimization of Jatropha-Pongamia Biodiesel Using CeO₂-TiO₂ Nano-additive: A Taguchi L16 Study

2.1 Feedstock preparation and biodiesel blending

Jatropha curcas and Pongamia pinnata oils were obtained from certified local suppliers and underwent the standard pretreatment involving filtration and drying before transesterification. Biodiesel was produced using base-catalyzed transesterification with methanol and potassium hydroxide; preliminary trials optimized the methanol-to-oil molar ratio, catalyst concentration, temperature, and reaction time to obtain methyl ester yields greater than 98%. After phase separation and washing, the biodiesel fractions were dried and used as a blend with commercial diesel to obtain B10, B20, B40, and B100 blends. Feedstock composition was varied by the preparation of Jatropha:Pongamia ratios (J:P) of 80:20, 60:40, 40:60, and 20:80 by volume. All blends were homogenized using magnetic stirring for 30 min and conditioned at ambient laboratory temperature for 24 h before testing. The physicochemical properties, including density, kinematic viscosity, calorific value, flash point, and cetane number, were measured according to the relevant American Society for Testing and Materials (ASTM) and International Organization for Standardization (ISO) standards; the results are shown in Table 1.

Table 1. Physicochemical properties for Taguchi L16 experimental runs

The B100 properties reported in Table 1 correspond to the mixed J:P methyl ester batches used in this study and should be interpreted as batch-specific measurements for the tested fuels rather than handbook values for neat single-feedstock biodiesel.

2.2 CeO₂-TiO₂ nano-additive preparation and dispersion

A CeO₂-TiO₂ nano-additive powder was prepared by a controlled co-precipitation route and calcined at 450 ℃ prior to blending with the test fuels. The same synthesized CeO₂-TiO₂ nano-additive powder was used for scanning electron microscopy (SEM) characterization and for all CeO₂-TiO₂ nano-additive dosage levels in the engine experiments. Fuel suspensions were prepared at nominal nano-additive concentrations of 25, 50, 75, and 100 ppm in the selected biodiesel-diesel blends. Dispersion was carried out by ultrasonication at 40 kHz and 200 W for 20 min, with intermittent cooling to avoid thermal degradation; a non-ionic surfactant (0.05 wt%) was used when needed to improve suspension stability. Before each engine test, the prepared blends were re-sonicated for 5 min. In the present study, the CeO₂-TiO₂ nano-additive-related experimental factor was dosage; the intrinsic powder characteristics were held constant across all Taguchi runs.

2.3 Scanning electron microscopy particle-size distribution

The morphology and size distribution of the CeO₂-TiO₂ nano-additive prepared by the controlled co-precipitation method were examined by SEM. The SEM images shown in Figure 1 correspond to the same CeO₂-TiO₂ nano-additive powder used for fuel dispersion and engine testing; the earlier description of separate pristine TiO₂, deposition-precipitation, and hydrothermal samples has been removed to avoid ambiguity.

Figure 1. Representative scanning electron microscopy (SEM) micrographs of the CeO₂-TiO₂ nano-additive synthesized by the controlled co-precipitation method: (a–c) different SEM fields showing quasi-spherical particle morphology and agglomerated nano-additive clusters

The original SEM images were calibrated in ImageJ using the 5 µm reference scale provided in the SEM micrographs. After calibration, 2 µm scale bars were inserted into all three panels. Particle-size analysis was performed in ImageJ by measuring at least 200 clearly distinguishable particles from each SEM field. Particles cut by the image boundary and severely overlapped agglomerates with unclear boundaries were excluded. The Feret diameter/equivalent circular diameter was used to represent the SEM-derived particle/agglomerate size.

The SEM-derived particle/agglomerate size statistics were as follows: Figure 1(a), n = 290, mean ± SD = 363.5 ± 227.8 nm, D10/D50/D90 = 111.2/309.0/668.7 nm; Figure 1(b), n = 336, mean ± SD = 323.7 ± 212.8 nm, D10/D50/D90 = 112.0/273.8/622.3 nm; and Figure 1(c), n = 321, mean ± SD = 318.4 ± 216.0 nm, D10/D50/D90 = 110.2/257.0/624.2 nm. Across all measured particles/agglomerates (n = 947), the overall Feret diameter was 334.1 ± 219.2 nm, with D10 = 111.2 nm, D50 = 281.5 nm, and D90 = 651.3 nm. The relatively broad distribution is attributed to partial agglomeration during drying and SEM sample preparation, which is typical for metal-oxide nano-additive particles with high surface energy.

2.4 Engine test rig and instrumentation

Experiments were conducted on a water-cooled, single-cylinder, four-stroke compression-ignition engine coupled to an eddy-current dynamometer for load control and brake-power measurement. Fuel consumption was measured gravimetrically using a calibrated burette and electronic balance, while intake air flow was recorded with a hot-wire anemometer. Exhaust gases were measured using a calibrated multi-gas analyzer: NOₓ, CO, and CO₂ by non-dispersive infrared (NDIR) methods, and HC by flame ionization detection. Exhaust temperature and in-cylinder pressure were monitored using K-type thermocouples and a piezoelectric pressure transducer, respectively. The annotated test rig and data-acquisition console are shown in Figure 2, and the instrument specifications are listed in Table 2.

Figure 2. Annotated compression-ignition (CI) engine test rig and data-acquisition system: (1) single-cylinder engine, (2) dynamometer, (3) exhaust line, (4) computer display, (5) control panel, and (6) data-acquisition unit

Table 2. Specifications and measurement accuracy of experimental instruments

2.5 Measurement uncertainty and calibration

All instruments were calibrated against traceable standards before experimentation. Uncertainty budgets were prepared for the principal performance metrics following the Guide to the Expression of Uncertainty in Measurement (GUM). Reported uncertainties were fuel consumption ±1.0%, brake power ±0.5%, NOₓ ±2 ppm or ±3% of reading, CO/CO₂ ±0.01 vol% or ±3% of reading, and HC ±2 ppm. Combined standard uncertainties for derived quantities were estimated by propagation of the component uncertainties and are summarized in Table 3.

Table 3. Measurement uncertainties and calibration summary

2.6 Experimental design

A Taguchi L16 orthogonal array was adopted to study four control factors at four levels each: blend percentage (A), J:P ratio (B), engine load (C), and CeO₂-TiO₂ nano-additive concentration (D). The L16 design reduced the number of experimental runs while preserving orthogonality for the estimation of main effects. Signal-to-noise (S/N) ratios were computed using the smaller-is-better criterion for BSFC, CO, HC, NOₓ, and CO₂, and the larger-is-better criterion for BTE. Analysis of Variance (ANOVA) was then used to quantify the percentage contribution of each factor and to report F- and p-values for statistical significance (Table 4).

Table 4. Factors and levels for the experimental matrix

Because the L16 array was selected to estimate main effects efficiently, interaction terms were not resolved independently. Potential couplings such as blend composition × CeO₂-TiO₂ nano-additive dosage, especially through viscosity and dispersion behavior, should therefore be interpreted as part of the residual experimental variation rather than as separately quantified effects.

This study used a Taguchi L16 orthogonal array to examine the combined effects of blend fraction, J:P ratio, engine load, and CeO₂-TiO₂ nano-additive dosage on a single-cylinder compression-ignition (CI) engine. Across the tested matrix, the models for BSFC, BTE, NOₓ, CO, HC, and CO₂ showed good fit, with R² values between 92.9% and 96.5%.

Blend fraction and engine load were the dominant factors governing the performance-emission response. Load had the largest S/N influence on BTE, whereas blend fraction contributed most to BSFC. For the emission metrics, blend fraction and load consistently accounted for the largest shares of the explained variance.

The CeO₂-TiO₂ nano-additive dosage and J:P ratio produced measurable but secondary effects within the explored design space. Their contributions were smaller than those of the blend fraction and load for every response, indicating that additive dosage should be viewed as a tuning parameter rather than the primary driver of engine behavior.

Within the tested L16 array, the B20 blend at 50% load with 25 ppm CeO₂-TiO₂ nano-additive and a 20:80 J:P ratio delivered the highest observed BTE and the lowest observed BSFC. This setting is reported as the best observed combination within the completed experimental campaign.

A separate confirmation run outside the L16 array was not available in the present dataset; future work should verify the optimum experimentally and examine interaction effects more explicitly.