Comparison of Used Locomotive Fuel Filters after Using B20 and B0 Fuels

The automotive and non-automotive sectors have followed the Indonesian government's policy regarding applying high concentration biodiesel use. Some issues arise from the effects of using a high biodiesel concentration on engine components, such as lter blocking or degradation. Therefore, various parties have challenged to improve both in terms of fuel quality and engine components. Meanwhile, testing of high concentration biodiesel fuel (blending ratio above 10 %) was still rarely published in the rail transport sector. Therefore, the rail test aimed to evaluate the effect of B20 fuel on locomotive engine components. The test was conducted during two periodic maintenance (6 months) using two trains with B20 and B0 fuels as a comparison. This study discusses the effect of B20 on fuel lters by performing tests on used lters for one periodic maintenance (3 months). The morphological analysis of deposits was conducted using a digital microscope, while TGA, GC-MS, FTIR, and elemental analysis were used to determine its components. The results showed that biodiesel ltration in the main lter was higher, and the subsequent ltration showed fewer deposits than that of pure diesel. Various types of fuel lter deposits have been identied, such as hydrocarbons and fatty acid methyl esters.


Introduction
The Indonesian government's policy on applying high concentration biodiesel has been followed by developing engine technology in the automotive sector, such as improving engine components, handling, and storage systems. Studies on the effects of using biodiesel fuel are easy to obtain from various publications. Nevertheless, biodiesel's use and test in the rail transport sector were lagging, especially for locomotive with high ratio fuel and a blending volume above 10 %. It was reported that using high biodiesel concentration causes problems in the intake system and engine components [1], such as lter clogging. Several case studies have reported that biodiesel was bene cial in minimizing exhaust emissions (except NOx) without signi cantly reducing engine performance but at risk of clogging the lter and reducing fuel economy [2].
The use of biodiesel with high concentrations affects deposit formation that caused accelerating the replacement time of the vechicle lter [3][4][5][6]. Apart from the contaminant factor, the excess deposit formation is also caused by ester compounds in the fuel. Testing of deposit characteristics was carried out by Csontos et al. with analysis using GC-MS, XRF-EDX, FTIR, and TGA. They identi ed carboxylic acid (CA) as fuel degradation and unreacted CA from biodiesel production, oxidized polymer compounds, glycerol, sterol, other impurities, and contaminants [7,8]. Meanwhile, Barker et al. identi ed the deposit or injector fouling as carbon in nature with C 16 -C 18 acid/ester [9]. However, several publications reported no signi cant difference in lters with pure diesel oil from petroleum [10,11]. Lammert et al. stated that low environmental temperatures such as during winter, the replacement of fuel with biodiesels disrupted the cold ow properties [12].
Our previous studies related to locomotive engines have identi ed that lter deposits are in the form of materials from fuel degradation, impurities from unreacted biodiesel production, as well as wear and contaminants. The results showed that the use of 20 % of palm oil during one maintenance period still guarantees a good ltration performance [13]. The effect of 20 % biodiesel (B20) fuel with strict quality control to the locomotive engine after being used for one maintenance period (3 months) from the rail test for 6 months, the used lters were investigated. This study reported the test results that compared them with petrodiesel fuel (B0).

Materials
B20 (20 % biodiesel and 80 % petrodiesel by volume) and B0 (100 % petrodiesel by volume) fuels were used for the locomotive engine. The biodiesel was derived from palm oil, which met the national quality standard (SNI 7182:2015), contained 0.8 % mass of Monoglyceride (MG), and less than 500 ppm moisture content. Furthermore, the petrodiesel met the national standard Decree of the Director-General of Oil and Gas Number 28.K/10/DJM.T/ 2016 and contained less than 2500 ppm sulfur with 48 cetane number.

Testing Methods and Analysis
The lter test was carried out on a coal-carrying train locomotive engine. One train uses B20 fuel, and the other uses B0 for comparison. Furthermore, the trains were run daily for 3 months, which is equivalent to 1 periodic maintenance. After 3 months, the lter was removed and replaced with a new one.
Trapped materials or chemicals were analyzed by a gas chromatography-mass spectrometry (GC-MS), Fourier-transform infrared spectroscopy (FTIR), an elemental analyzer, thermogravimetric analysis (TGA), and differential thermogravimetric (DTG). Before GC-MS analysis, the lter was cut around 1 cm 2 and then soaked in 10 mL of each acetone, chloroform, dichloromethane (DCM), and n-hexane as organic solvents. Meanwhile, solvents containing precipitate matter were ltered using a lter paper (Whatman, Glass micro ber lters GF/F diameter 47 mm, pore diameter 0.7 µm, Cat No. 1825-047) to prevent tiny solid particles from entering the GC-MS column. The use of these solvents was intended to determine the most effective in dissolving precipitate.
The GCMS model, 6890 N (Agilent Technologies Inc., Santa Clara, California, United States), was equipped with a mass spectrometer detector, 5975B and HP-5MS UI column (Agilent Technologies, 5 % Phenyl Methyl Siloxane, 30 m × 250 µm × 0.25 µm). Furthermore, the injector temperature was 250°C, and the carrier gas was helium at a constant ow rate of 1 mL min − 1 . The initial oven temperature of the column was 40°C, which was maintained for 1 min, raised to 300°C at 10°C min − 1 , and then maintained for 4 minutes at 300°C. The MS Source and MS Quad temperatures were 230°C and 150°C, respectively. Also, ionization energy was 70 eV, and the total ow was 104 mL min − 1 , column ow was 1 mL min − 1 , and the average velocity was 36.262 cm s − 1 . Chemicals identi cation in the samples was determined by comparing spectra and retention time of the individual compounds with the authentic references stored in the NIST14 mass spectral data library.
The FTIR (IR Prestige 21, Shimadzu instrument) technique was used in the wavenumber range of 4000-400 cm − 1 to identify functional groups of chemical contents in the lter engine containing precipitate.
Meanwhile, the used lter using B20 fuel was analyzed by thermogravimetric analysis (TGA) to ascertain its thermal characteristics. Also, a new lter was analyzed as a comparison. TGA is a technique to monitor changes in the mass of a material to temperature and time [14]. The small size of the used and a new lter as comparison were put into a thermogravimetric analyzer (STA PT 1600, Linseis), and then the sample was heated in N 2 atmosphere from room temperature (about 30°C) to 600°C at a rate of 10°C/min to obtain the weight loss pro le. Moreover, an elemental analyzer (CHN 628 series, Leco Corporation), according to the ASTM D 5373 procedure, was used to analyze elemental content in the used and new lters as a comparison.

Locomotive Engine for Testing
The locomotive engine speci cations used in this test are shown in Table 1. This engine has three levels of fuel ltration. The rst level was a strainer for ltering contaminants with sizes above 10 microns and water. The second level was the main lter for particles with a size of 5-10 microns, while the third one was the twin lter to ltrate smaller particles that stick to the engine. The schematic of the test engine ltration system is shown in Fig. 1. The main and twin lters were virtually analyzed using a 3D digital microscope, while the main lter was analyzed using GCMS, TGA, FTIR, and elemental analyzer.

Results And Discussion
This study determines using B20 on the main and twin lters after 3 months of use. Furthermore, lter analysis with B0 fuel was conducted on these two types of lters as references, and strainers for B20 and B0 were not analyzed because they only captured large-size material (such as rocks or gravel). Therefore, deposits of B20 fuel were not captured.

Photo of Fuel Filter
The visual inspection result of the used main lters with B20 and B0 fuel after 3 months using a 3D digital microscope at a magni cation of 500 times is shown in Figure 2. While the deposit's thickness was assessed at 100 times magni cation, as shown in Figure 3.
In the used main lter for B20, it can be seen that the bers have been tightly covered with a material deposit, while for the B0, the ber and porosity were still clearly visible. The result of deposit thickness measurement showed that B20 was higher with an average of 822 µm compared to B0 with an average of 699 µm. Meanwhile, the type of deposit on the main lter with B20 fuel was soft, and hence, even though it appeared to cover the surface, fuel still owed with less resistance, and the delta pressure lter test did not change signi cantly. Likewise, the performance test results showed that the decrease of power was not signi cant after using B20 for 3 months [15]. Figures 4 and 5 show the deposit's morphology on the twin lter for B20 and B0 with a 200 and 500 times magni cation, respectively. It can be seen that after 3 months of usage, the bers were still visible, both for the used lters of B0 and B20. However, the granule deposit which covers the lter surface had a different size. Petrodiesel (B0) tent formed a smaller size of granule deposit compared with B20.
The cross-section of the deposit thickness show that B20 has an average of 601 µm and B0 has 621 µm as shown in Figure 6. This can be considered because the contaminants for B20 have been ltered on the main lter, hence, the ltration process at twin lter is lighter.

Identi cation of trapped materials or chemicals on the lter
Analysis with GCMS was conducted to determine which deposit components were ltered on the main and twin lters with B20 or B0 fuels. Furthermore, the identi cation of trapped materials or chemicals on the used lter was analyzed by FTIR, elemental analyzer, and TGA. The used lters were soaked in four organic solvents before GC-MS analysis, namely acetone, chloroform, dichloromethane (DCM), and nhexane. The identi ed chemicals on this lter using B0 fuel for each solvent are shown in Table 2.
According to Table 2, among the four organic solvents used to dissolve chemicals in the used lter, chloroform appears to be better than acetone, dichloromethane, and n-hexane due to many chemicals identi ed. There were two compounds not identi ed with the chloroform solvent but identi ed with the nhexane solvent. The results of the identi ed compounds were a combination of compounds dissolved in chloroform and n-hexane. Therefore, the chemical compounds contained in the lters were obtained. Based on the combination identi ed compounds with chloroform and n-hexane (Table 2), the chemicals trapped on the used lter were hydrocarbons with the number of atoms of C 12 -C 27 , and the dominant one was pentadecane, 2,6,10,14-tetramethyl-(C 19 H 40 ). Alkane hydrocarbon compounds over C 16 are solid at 20 °C. Therefore, C 17 and above were naturally trapped and caused a blockage on the lter. C 12 -C 16 compounds were identi ed because there was no pretreatment to the lter before soaking in the organic solvent, meaning that it was still left on the lter, although the hydrocarbons C 12 -C 16 at 20 °C in the liquid phase.
Furthermore, naphthalene compounds caused blockage because of a high melting point and solid at room temperature. As the main diesel component, the alkane group dominated the precursor deposits. The deposit contained tetradecane, pentadecane, and hexadecane.
The identi ed chemicals on the used lter with B20 fuels for each solvent can be seen in Table 3. Based on Table 3, chloroform was a better solvent than others in dissolving chemicals trapped on the used lter with B20 fuel. However, some compounds could not be dissolved by chloroform but dissolve in acetone, DCM, and n-hexane. The chemicals identi ed in the four organic solvents were combined as shown in The hydrocarbons were almost the same as Table 2, which was derived from petrodiesel. Meanwhile, fatty acid methyl ester compounds derived from palm biodiesel. Chemical compounds such as Methyl tetradecanoate and Tridecanoic acid 12-methyl-, methyl ester were derived from Fatty acid C 14 (myristic acid) that solid phase at room temperature. Likewise, Hexadecanoic acid, methyl ester, and Methyl stearate derived from C 16:0 (palmitic acid) and C 18:0 (stearic acid) fatty acids, respectively, can cause lter blockage. Next, 8,11-Octadecadienoic acid, methyl ester, and 9-Octadecenoic acid, methyl ester, (E)-were derived from C18:1 fatty acid (oleic acid) that liquid phase at room temperature. Therefore, it can be concluded that the large peak areas (corresponded to high concentrations) of Hexadecanoic acid, methyl ester (C 15 H 30 O 2 ) and 9-Octadecenoic acid, methyl ester, (E) -(C 19 H 36 O 2 ) were caused by a high concentration of palmitic and oleic acid in the palm oil and no pretreatment on the used lter before soaking. Nevertheless, methyl esters derived from oleic acid do not cause lter blocking. Therefore, the precursor deposit component for the B20 lter was dominated by fatty acid methyl esters, although based on the measurement results, the deposit was also caused by alkane hydrocarbons.
Moreover, the used lter with B20 fuel was analyzed by FTIR to support chemical identi cation of the GC-MS results and compared with the new one. The FTIR analysis results and prediction of functional groups for each peak are shown in Figure 7 and Table 6, respectively.
Based on FTIR analysis, the peaks that appeared on a new lter generally also appeared on the used lter.
However, several new peaks appeared on the used lter, such as wave numbers 1259 and 1232 cm -1 which were predicted to be the C-O-C group from ester, which was probably derived from biodiesel.
Furthermore, the predicted wavenumbers 626 and 468 cm -1 were the naphthalene groups. The peak at wave number 609 cm -1 was predicted as -SO2-group derived from petrodiesel fuel.
The used lter is also analyzed with an elemental analyzer to determine the percentage of carbon, hydrogen, and oxygen in the lter compared to the new one, as seen in Table 7 (Table  6) because the elemental analyzer could only detect C, H, and N.
Moreover, the used lter was analyzed by Thermogravimetric (TG) and Differential Thermogravimetric (DTG) analysis to understand its thermal properties compared with the new lter. Figure 8 shows the TG and DTG analysis results of the used and the new lters at N 2 conditions with a heating rate of 10 C min -1 from room temperature to 600 C. Comparing the TG and DTG curves between the used (Figure 8a), and the new lters ( Figure 8b) showed thermal degradation of the used lter consisted of three decomposition areas, namely 35 -150 °C; 150 -320 °C; and 320 -400 °C, while the new one only consisted of two areas of 35 -150 °C and 250 -400 °C. In 35 -150 °C, it was predicted as water evaporation and volatile compounds decomposed from the lter material. Furthermore, in the new lter, the next decomposition occurred at 250 -400 °C, which was predicted the larger molecular weight material, such as polymer or composite, be decomposed. Next, in the used lter at 150 -320 °C decomposition or evaporation of fuel with the number of carbon atoms [C] ranging from C 13 -C 18 such as Tridecane, Tetradecane, Pentadecane, Hexadecane, Tridecane, and Octadecane (Table 2 and 3). Meanwhile, in 320 -400 °C, there was decomposition or evaporation of a larger number of carbon atoms [C] ranging from C 19 -C 27 , such as Nonadecane, Eicosane, Heneicosane, Docosane, Tricosane, Heptacosane, Tetracosane, Pentacosane, and Eicosane.

Conclusion
The amount of trapped deposit material on the B20 lter showed that the rst (main lter) was more than B0, and the material passed to the second lter (twin lter) was less. During one periodic maintenance, the main pores of the B20 lter have been closed by substrate, while on the twin lters, the use of B20 and B0 fuels still showed the porosity. Moreover, the used lter of rail locomotive fueled by B20 was identi ed various deposit compounds which contained hydrocarbon chains of C 13 -C 27 and fatty acid methyl esters C 15 -C 19 . Furthermore, deposit formation on both B0 and B20 was at tolerance level so that it did not require faster changing. The fuel ltration system of the test locomotive engine Figure 2 Page 11/13 The surface morphology of the locomotive main lter sample after 3 months of use Figure 3 The thickness of the deposit on the locomotive's main lter after 3 months of use  Photo of the used twin lter sample surface after 3 months of use Figure 6 Side view of the used twin lter sample after 3 months of use Overlay of FTIR analysis results from a used lter with B20 fuel (black graph) and a new lter (red graph)

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Tables.docx