The Investigation into the Tribological Impact of Alternative Fuels on Engines Based on Acoustic Emission

: The wide use of different alternative fuels (AL) has led to challenges to the internal combustion (IC) engine tribology. To avoid any unpredicted damages to lubrication joints by using AL fuels, this study aims to accurately evaluate the inﬂuences of alternative fuels on the tribological behavior of IC engines. Recent achievements of the acoustic emission (AE) mechanism in sliding friction provide an opportunity to explain the tribological AE responses on engines. The asperity– asperity–collision (AAC) and ﬂuid–asperity–shearing (FAS) mechanisms were applied to explain the AE responses from the piston ring and cylinder liner system. A new adaptive threshold–wavelet packets transform (WPT) method was developed to extract tribological AE features. Experimental tests were conducted by fueling three fuels: pure diesel (PD), biodiesel (BD), and Fischer–Tropsch (F–T) diesel. The FAS–AE indicators of biodiesel and F–T diesel show a tiny difference compared to the baseline diesel using two types of lubricants. Biodiesel produces more AAC impacts with higher AAC–AE responses than F–T diesel, which occurs at high speeds due to high temperatures and more particles after combustion than diesel. This new algorithm demonstrated the high performance of using AE signals in monitoring the tribological impacts of alternative fuels on engines.


Introduction
Many investigations of alternative fuels (AL) focus on the emissions and performance because of the environmental vulnerabilities and resource shortage of fossil fuels. The physical and chemical properties of alternative fuels are different from those of ordinary diesel, which inevitably lead to physical and chemical changes in the combustion chamber. The efforts to reduce fuel emissions may lead to further tribological problems such as a loss of lubricity and surface damage [1]. Moreover, the influence on the friction and lubrication condition of the engines using AL fuels is still unknown. Therefore, it is necessary to monitor the alternatives' tribological behaviors on engines under the working condition.
Biodiesel (BD) and Fischer-Tropsch (F-T) diesel have become a research focus in China recently because these two fuels show good potential to put into use on diesel engines with good performance and lower emissions. Furthermore, F-T diesel synthesized from coal can promote the clean utilization of coal in China. Biodiesel is a potential substitute fuel blended mainly with bio-oil and diesel. Numerous research achievements prove that biodiesel and diesel blends have progressive effects on emissions and power performance [2][3][4][5][6]. F-T diesel is also a promising diesel substitute fuel that contains a high cetane number, near-zero sulfur content, and low aromatic level [7][8][9]. However, the studies of the tribological impacts of F-T diesel and BD on engines are still not in-depth. This paper uses an adaptive threshold method based on the wavelet packet transform (WPT) to characterize the AE responses on the external cylinder surface to decipher the impacts of the alternative fuels on the engine. AE signals were acquired by fueling two substitute fuels, biodiesel and F-T diesel, and the baseline diesel with two different lubricants recommended from the engine manufacturer. An autocorrelated threshold-WPT noise suppression approach was proposed for enhancing the tribological features in this study. The denoised AE indicator based on the new method can quantify the tribological impacts by two AL fuels in different lubrication regimes of the piston ring and cylinder liner system.

Engine Test Rig and Test Conditions
The experimental studies were conducted on a single-cylinder direct injection diesel engine (Anhui Quanchai Engine Co., Ltd., Quanjiao, China) test rig. The schematic diagram of the diesel engine test system is shown in Figure 1. The engine was coupled by the eddy current dynamometer (Chengbang, China, Model: DW 160). The key specification of the test engine is given in Table 1. The material of the piston ring and the cylinder liner are chrome-plated alloy steel and alloy cast iron. The roughness parameter Ra of the ring and liner were 0.8 and 0.2 µm according to the surface finish grade numbers (ISO 1302(ISO :1992. Energies 2021, 14, x FOR PEER REVIEW 3 of 20 FAS model to decipher the FAS-AE frequency characteristics under the HL regime. Hence, in the piston ring and cylinder liner system, AE features under different regimes can be explained mainly with AAC and FAS models. This paper uses an adaptive threshold method based on the wavelet packet transform (WPT) to characterize the AE responses on the external cylinder surface to decipher the impacts of the alternative fuels on the engine. AE signals were acquired by fueling two substitute fuels, biodiesel and F-T diesel, and the baseline diesel with two different lubricants recommended from the engine manufacturer. An autocorrelated threshold-WPT noise suppression approach was proposed for enhancing the tribological features in this study. The denoised AE indicator based on the new method can quantify the tribological impacts by two AL fuels in different lubrication regimes of the piston ring and cylinder liner system.

Engine Test Rig and Test Conditions
The experimental studies were conducted on a single-cylinder direct injection diesel engine (Anhui Quanchai Engine Co., Ltd., China) test rig. The schematic diagram of the diesel engine test system is shown in Figure 1. The engine was coupled by the eddy current dynamometer (Chengbang, China, Model: DW 160). The key specification of the test engine is given in Table 1. The material of the piston ring and the cylinder liner are chrome-plated alloy steel and alloy cast iron. The roughness parameter Ra of the ring and liner were 0.8 and 0.2 μm according to the surface finish grade numbers (ISO 1302:1992).    The AE sensor was located closer to the contact part between the outer cylinder wall and the liner. The crankshaft position sensor is an electronic device located on the flywheel housing. A crank angle signal was measured by crankshaft position sensor 39180 (Hyundai Mobis, Seoul, Korea) at the same time to record the TDC (top dead center) of each revolution. The AE measurement system is composed of the AE sensor, pre-amplifier, AE detector, and PC. The wideband piezoelectric AE sensor (model SR800, Qing Cheng AE Institute, Guangzhou, China) was selected to acquire AE signals because this sensor has a good frequency response in the band from 20 and 400 kHz according to the calibration chart as shown in Figure 2. A number of trial tests also showed that AE signals of interest were more significant below 250 kHz on engine surfaces. Therefore, the frequency band in this study was focused around 20 to 400 kHz. and the liner. The crankshaft position sensor is an electronic device located on the flywheel housing. A crank angle signal was measured by crankshaft position sensor 39180 (Hyundai Mobis, Seoul, Korea) at the same time to record the TDC (top dead center) of each revolution. The AE measurement system is composed of the AE sensor, pre-amplifier, AE detector, and PC. The wideband piezoelectric AE sensor (model SR800, Qing Cheng AE Institute, Guangzhou, China) was selected to acquire AE signals because this sensor has a good frequency response in the band from 20 and 400 kHz according to the calibration chart as shown in Figure 2. A number of trial tests also showed that AE signals of interest were more significant below 250 kHz on engine surfaces. Therefore, the frequency band in this study was focused around 20 to 400 kHz.
The AE sensor was verified by the manufacturer before tests where the AE response of pencil-lead breaks on the engine housing was observed. Because the output voltage signal of the AE sensor is sometimes as low as a few microvolts. The signals are processed by a pre-amplifier (the magnification is 40 dB). The acoustic emission detector is the SEAU2S-1016-08 acoustic emission detector of Beijing Shenghua Industrial Technology Co., Ltd. The sampling length is up to 128 k sampling points under 16-bit precision.

Test Procedure
Three types of fuels (pure Diesel, F-T diesel and Biodiesel), the testing conditions, and fuel properties are presented in Table 2. Two new oils (CD-10W30 and CD-15W40, Sinopec Lubricant Company, Beijing, China) are taken as baselines to verify the effectiveness and sensitivity of extracted features fueling with diesel. The viscosity-temperature characteristics of the lubricating oils (CD-15W40 and CD-10W30) were measured by the sinusoidal viscometer (SV-10, A&D Company, Limited, Tokyo, Japan) as given in Figure  3. The length of the AE data record is 1,144,000 points for each working cycle. The sampling rate is 800 kHz, which covers over 100 engine working cycles for sufficient average and thus obtaining reliable results. The AE sensor was verified by the manufacturer before tests where the AE response of pencil-lead breaks on the engine housing was observed. Because the output voltage signal of the AE sensor is sometimes as low as a few microvolts. The signals are processed by a pre-amplifier (the magnification is 40 dB). The acoustic emission detector is the SEAU2S-1016-08 acoustic emission detector of Beijing Shenghua Industrial Technology Co., Ltd. (Beijing, China). The sampling length is up to 128 k sampling points under 16-bit precision.

Test Procedure
Three types of fuels (pure Diesel, F-T diesel and Biodiesel), the testing conditions, and fuel properties are presented in Table 2. Two new oils (CD-10W30 and CD-15W40, Sinopec Lubricant Company, Beijing, China) are taken as baselines to verify the effectiveness and sensitivity of extracted features fueling with diesel. The viscosity-temperature characteristics of the lubricating oils (CD-15W40 and CD-10W30) were measured by the sinusoidal viscometer (SV-10, A&D Company, Limited, Tokyo, Japan) as given in Figure 3. The length of the AE data record is 1,144,000 points for each working cycle. The sampling rate is 800 kHz, which covers over 100 engine working cycles for sufficient average and thus obtaining reliable results.     Figure 4 presents typical AE signals and the varying piston speed profile in the angular domain for an engine working cycle. The piston speed reaches its maximum in each middle stroke and decreases to zero at the top dead center (TDC) and bottom dead center (BDC). The large AE events to be attributed sequentially to the excitations of exhaust valve closing (EVC), inlet valve closing (IVC), fuel injection, combustion shocks, exhaust valve open (EVO), and inlet valve opening (IVO). Each of them exhibits strong AE bursts and reflects the short impulses of the sources. Between the strong bursts, there is nothing moving components except for piston assembly in the single-engine. It is worth noting that there are local stationary waves in the middle of each stroke. The amplitude of these stationary waves varied with the piston speed, displaying a close association with the FAS effects suggested in [42] and [44]. Within the same local range, some locally nonstationary signals are generated stochastically. The locally stochastical AE peaks varied with piston speed are related to the asperity-asperity collision based on the AAC model of the BL lubrication regimes.

AE Signals
Therefore, a more effective signal processing technique is required to emerge the weak pseudo-continuous AE for the evolution of tribological behavior between the ring and the liner.
Dynamic Viscosity ( Pa.s ) Figure 3. The viscosity-temperature curve of two tested lubricating oils. Figure 4 presents typical AE signals and the varying piston speed profile in the angular domain for an engine working cycle. The piston speed reaches its maximum in each middle stroke and decreases to zero at the top dead center (TDC) and bottom dead center (BDC). The large AE events to be attributed sequentially to the excitations of exhaust valve closing (EVC), inlet valve closing (IVC), fuel injection, combustion shocks, exhaust valve open (EVO), and inlet valve opening (IVO). Each of them exhibits strong AE bursts and reflects the short impulses of the sources. Between the strong bursts, there is nothing moving components except for piston assembly in the single-engine. It is worth noting that there are local stationary waves in the middle of each stroke. The amplitude of these stationary waves varied with the piston speed, displaying a close association with the FAS effects suggested in [42,44]. Within the same local range, some locally nonstationary signals are generated stochastically. The locally stochastical AE peaks varied with piston speed are related to the asperity-asperity collision based on the AAC model of the BL lubrication regimes.

AE Signals
Therefore, a more effective signal processing technique is required to emerge the weak pseudo-continuous AE for the evolution of tribological behavior between the ring and the liner.

Wavelet Packet Transform
As shown in previous studies, the measured AE signal will inevitably be affected by various noises. The nonstationary weak AE needs more effective approaches to extract the tribological features accurately. Compared with wavelet transform under the same bandwidth, wavelet packet transform (WPT) can process nonstationary signals with better frequency resolution. Moreover, WPT only adds a limited amount of decomposition calculations, which is lower than continuous wavelet transform and short-time Fourier transform. Hence, this study employed an adaptive threshold WPT method to investigate the AE denoising and feature extracting. WPT is widely used in condition monitoring and fault diagnosis in rotary mechanical systems such as gears [46,47] and bearings [48], diesel engines [49]. Remarkably, reference [50] analyzed AE signals to investigate the failure of tribological systems based on wavelet packet decomposition.

Wavelet Packet Transform
As shown in previous studies, the measured AE signal will inevitably be affected by various noises. The nonstationary weak AE needs more effective approaches to extract the tribological features accurately. Compared with wavelet transform under the same bandwidth, wavelet packet transform (WPT) can process nonstationary signals with better frequency resolution. Moreover, WPT only adds a limited amount of decomposition calculations, which is lower than continuous wavelet transform and short-time Fourier transform. Hence, this study employed an adaptive threshold WPT method to investigate the AE denoising and feature extracting. WPT is widely used in condition monitoring and fault diagnosis in rotary mechanical systems such as gears [46,47] and bearings [48], diesel engines [49]. Remarkably, reference [50] analyzed AE signals to investigate the failure of tribological systems based on wavelet packet decomposition.
The wavelet packet decomposition, shorten as wavelet packets (WP) or sub-band tree, is extended by discrete wavelet decomposition (DWT) passed through more filters than the DWT. Based on the DWT, the calculation function of wavelet packets can be defined as [51]: where n = 1, 2, 3· · · ; k = 1, 2, 3· · · 2N-1; W 0 0 (t w ) is the scaling function φ(t w ) and W 0 1 (t w ) the wavelet function ϕ(t w ). The superscript j presents the jth level of wavelet packets basis.
The algorithm of wavelet packet spectrum which contains the absolute values of the coefficients from the frequency-ordered terminal nodes of the input binary wavelet packet tree was first introduced by Wickerhauser [52]. orthogonal wavelet decomposition procedure separates the coefficients into two parts using the high pass filter H( f ) and the low pass filter G( f ).
The algorithm of wavelet packet spectrum which contains the absolute values of the coefficients from the frequency-ordered terminal nodes of the input binary wavelet packet tree was first introduced by Wickerhauser [52]. Figure 5 presents a full wavelet packet tree down to level 3. The terminal nodes approximate bandpass filters of the form at the j level of the wavelet packet transform is

Adaptive Threshold-AE based on FAS
The main idea of the traditional threshold denoising method is to retain useful information and denoise in different frequency bands according to appropriate threshold criteria. As shown in Figure 4, besides the large AE bursts aroused by the landing of valves and combustion process, it also has some small AE bursts that occurred randomly in the middle of the stroke. The amplitude of AE caused by FAS is very weak and its envelope is remarkably similar to the modified piston velocity curve. To accurately extract FAS-AE, a threshold needs to be determined under various working conditions. Hence, this paper designs a new adaptive threshold denoising function Yik to exclude the noise from valves, combustion and injection: where ik x is the raw AE data; k is the number of working cycles; i λ is the adaptive threshold designed to quantitatively check the similarity between a modified velocity curve and the envelope of the AE signals.

Adaptive Threshold-AE Based on FAS
The main idea of the traditional threshold denoising method is to retain useful information and denoise in different frequency bands according to appropriate threshold criteria. As shown in Figure 4, besides the large AE bursts aroused by the landing of valves and combustion process, it also has some small AE bursts that occurred randomly in the middle of the stroke. The amplitude of AE caused by FAS is very weak and its envelope is remarkably similar to the modified piston velocity curve. To accurately extract FAS-AE, a threshold needs to be determined under various working conditions. Hence, this paper designs a new adaptive threshold denoising function Y ik to exclude the noise from valves, combustion and injection: where x ik is the raw AE data; k is the number of working cycles; λ i is the adaptive threshold designed to quantitatively check the similarity between a modified velocity curve and the envelope of the AE signals.
in which c i is the iteration coefficient, c i is iteratively reduced until the distance d(x k , v pi ) turned to the minimum: in which x i is the mean value of 20 working cycles at the ith working condition. It can find that the data around some specific crank angles are too noisy and have less AE content of the FAS effect. The 'sign' function is applied to convert the piston speed around these crank angles into zero means. Therefore, the x i is closer to the velocity profile and to better reflect the FAS effect. Besides, the interval of the Y-axis is scaled at each speed by a factor of the quintuple standard deviation 5σ j calculated by: where n is the time index, running up to M of the sample number for the AE signal in an engine cycle; and the K = 20 is the number of engine cycles.

WPT Spectrum of AE Signals and Optimal Wavelet Basis
A Daubechies wavelet with order 35 is selected to identify the low amplitudes AE. The selection of wavelet basis is an important step to determine the effect of wavelet denoising. Daubechies wavelet has good symmetry and biorthogonality. The gradual attenuation profile is conducive to highlight the weak AE events with asymmetric characteristics and low amplitudes. Hence, Daubechies (db) wavelet is used.
To verify the effectiveness of AE thresholds to extract FAS-AE, WPT spectra of raw AE and threshold-AE are calculated for all different tested cases using db5 wavelet. The decomposition level for WPT is level 8 for a trial and error test. Figures 6 and 7 show the typical WPT spectra for the baseline pure diesel with the lube-oil 10W30 running at different speeds at 10 Nm. In both above two figures, it can find some semi-continuous distinguished peaks in four discrete frequency bands: 40-60 kHz, 70-90 kHz, 140-160 kHz, and 170-190 kHz. These AE events that emerged around each middle stroke in narrow frequency bands are correlating to FAS-AE according to the analysis of Section 3.1.
It should be noted that some little high irregularly AE peaks are accompanied by the FAS-AE in Figure 6 compared with Figure 7. These small bursts may be aroused by the AAC effects owing to the unevenness of the oil film thickness between the piston ring and cylinder liner surface at high sliding speeds. Hence, the threshold-WPT spectra analysis in Figure 7 provides a new way to differ FAS-AE from the other AE sources. Besides, the AE spectra around the low-frequency range (<40 kHz) were possibly less connected with the tribological-AE between the ring and liner with too irregularly high AE spikes.
Based on the critical understanding of WPT spectrum analysis, the optimal wavelet base selection principle which is according to the energy of signals is not suitable in this situation. Hence, a new criterion for the optimal order in Daubechies wavelet family is developed to submerge the wake signals from a strong background noise. For doing so, an average amplitude criterion is established in the time-frequency domain of a WPT spectrum as: where n is the frequency index which covers a frequency range from 40 to 200 kHz within which the spectral amplitudes at the four frequencies are more significant, and k is the index for engine cycles. The correlation coefficient between and modified piston speed curve was calculated as one of the criteria to choose the suitable order of Daubechies wavelet. In Figure 8, the average correlation amplitudes for different wavelet orders decline with the order increase and the correlation coefficients with the piston speed show very slight variations with the order increase. Therefore 'db4' is the maximal amplitudes with a shorter time for all the cases explored. Considering a slightly better smooth effect at the high order, the wavelet decomposition order is selected as 'db5' for further WP analysis in this study.     In Figure 8, the average correlation amplitudes for different wavelet ord with the order increase and the correlation coefficients with the piston speed slight variations with the order increase. Therefore 'db4' is the maximal amp a shorter time for all the cases explored. Considering a slightly better smooth high order, the wavelet decomposition order is selected as 'db5' for further W in this study. To select an optimal order for extracting weak AE signals, the average amplitude in the middle of the power stroke, correlation coefficients of the modified piston speed and the WPT spectrum, and the CPU computing time were chosen as an evaluation parameter. The baseline test case is the pure diesel with 10W30 oil. Figure 8a illustrates the average amplitudes under each testing speed calculated by different wavelet orders which are from 'db4' to 'db40'. The average correlation amplitudes for different wavelet orders decline with the order increase. Figure 8b shows the correlation coefficients of the WPT spectrum and the modified piston speed show very slight variations with the order increase. Moreover, the computation time of CPU is increasing linearly with wavelet orders as shown in Figure 8c therefore 'db4' is the maximal amplitudes with shorter time for all the cases explored. Considering a slightly better smooth effect at the high order, the wavelet decomposition order is selected as 'db5' for further WP analysis in this study.

Optimal Threshold-WP Based on the Auto-Correlation Analysis of the Piston Velocity
Autocorrelation analysis has great performance in noise cancellation by retaining periodic components to increase the signal-to-noise ratio (SNR) [53]. It is difficult to extract the steady and periodic AE components correlated to engine tribological behaviors because of the unknown characteristics in a non-stationary system. The AE modeling studies show that the FAS-AE are closely related to the piston speed. This FAS-AE feature provides an opportunity to increase the SNR of the denoising WPT approach for the tribological AE analysis.
According to the results of wavelet packet decomposition, four frequency bands are selected to denoise the semi-continuous signal. The Daubechies wavelet basis function 'db4' is selected. To obtain a finer angle-frequency domain analysis result, the autocorrelation coefficients between the WPT spectrum of different frequency sequence and the modified piston curve were calculated as follows: where V i is the modified piston speed V i = sign(v pi ); cov(W, V i ) is covariance of W and V i ; σ W and σ V i are the standard deviation of W and V i . Figure 9 shows the correlation coefficients of four selected bands under different working conditions with the baseline diesel-10W30. The correlation between the two variables is low when the correlation coefficient is less than 0.3. The WPT spectrum in the band 170-190 kHz has the lowest correlation to the piston speed which is less than 0.3 under most working conditions. That indicates the AE spectrum is less correlated to the FAS effect. Therefore, the frequency bands: 40-60 kHz, 70-90 kHz, 140-160 kHz were chosen as the target bands. where N is the frequency index which covers a frequency range from 40-200 kHz. The average residual wavelet packets coefficient can represent better the locally non-stationary AE bursts reflecting more AAC effects. To evaluate tribological impact using AE, AE signals are de-noised with an adaptive threshold-WPT approach, which is summarized as following key steps: 1. Apply the threshold given by Equation (2) to suppress the non-stationary AE bursts in the middle of the strokes; calculate the d value obtained Equation (4), and judge whether di-di-1 ≥ 0, otherwise reduce the iteration coefficient ci, and repeat step 1; 2. Apply WPT to threshold-AE signals (K = 20 for the limited memory in the PC used) with analysis parameters: J=8 and 'db5'.
3. Calculate the correlation coefficients between the envelope of WPT spectrums W and modified piston speed pi V , remove the frequency band with a low correlation is less than 0.3; 4. Calculate the residual WP coefficient RW as given in Equation (8) from  to remove the noise of other sources. 5. Perform inverses WPT to reconstruct the AE signals in the selected frequency bands; calculate the average envelope of 20 reconstructed signals of selected frequency bands, to enhance the similarity to the velocity profile sum the envelope signal matrix; 6. Select the local sequence in the middle of each stroke, calculate the mean standard deviation for 20 working cycles as the FAS-AE indicator and AAC-AE indicator for four strokes. Figures 10 and 11 show the differences of FAS-AE indicators between two baselines: 15W40 and 10W30. AE signals for 15W40 oil with higher viscosity exhibit greater amplitudes than 10W30 with lower viscosity. It shows that AE signals are sufficiently sensitive With applying this adaptive threshold method to AE signals, anther diagnostic parameters should not be neglect to reflect AAC effects corresponding to ML and HL regimes. The AAC-AE was observed in the WPT spectrum analysis as given in Figures 6 and 7. These AE bursts with tiny higher amplitudes than FAS-AE can be obtained the wavelet coefficients by subtracting between two sets of x ik and Y ik . Specifically, it is calculated by:

Diagnosis of Alternatives and the Baselines with FAS Effects
where N is the frequency index which covers a frequency range from 40-200 kHz. The average residual wavelet packets coefficient can represent better the locally non-stationary AE bursts reflecting more AAC effects.
To evaluate tribological impact using AE, AE signals are de-noised with an adaptive threshold-WPT approach, which is summarized as following key steps:

1.
Apply the threshold given by Equation (2) to suppress the non-stationary AE bursts in the middle of the strokes; calculate the d value obtained Equation (4), and judge whether d i − d i−1 ≥ 0, otherwise reduce the iteration coefficient c i , and repeat step 1;

2.
Apply WPT to threshold-AE signals (K = 20 for the limited memory in the PC used) with analysis parameters: J = 8 and 'db5'.

3.
Calculate the correlation coefficients between the envelope of WPT spectrums W and modified piston speed V pi , remove the frequency band with a low correlation which ρ(W, V pi ) is less than 0.3; 4.
Calculate the residual WP coefficient RW as given in Equation (8)  Perform inverses WPT to reconstruct the AE signals in the selected frequency bands; calculate the average envelope of 20 reconstructed signals of selected frequency bands, to enhance the similarity to the velocity profile sum the envelope signal matrix; 6.
Select the local sequence in the middle of each stroke, calculate the mean standard deviation for 20 working cycles as the FAS-AE indicator and AAC-AE indicator for four strokes. Figures 10 and 11 show the differences of FAS-AE indicators between two baselines: 15W40 and 10W30. AE signals for 15W40 oil with higher viscosity exhibit greater amplitudes than 10W30 with lower viscosity. It shows that AE signals are sufficiently sensitive to small changes in the properties of lubricants. Therefore, they can be based to detect and diagnose the FAS-AE effects of alternative fuels on engines. indicates that these two alternative fuels produce little negative influences on FAS effects for most operating conditions.  indicates that these two alternative fuels produce little negative influences on FAS effec for most operating conditions.   The FAS-AE exhibits the impacts on the lubricity of the oil film fueling th tives. Too high FAS-AE indicates high power consumption to overcome viscou using AL fuel, and too low FAS-AE shows the viscosity. of oil film decreasing wh AL fuels. The FAS-AE exhibits the impacts on the lubricity of the oil film fueling the alternatives. Too high FAS-AE indicates high power consumption to overcome viscous friction using AL fuel, and too low FAS-AE shows the viscosity. of oil film decreasing when using AL fuels. Figures 14 and 15 show the AAC-AE indicators of different alternative fuels under different loads. The AAC-AE trends are unstable at speeds. Especially, the AAC-AE of biodiesel has high values in the power stroke and exhaust stroke. This can indicate that the combustions of these alternative fuels are less perfect. Reference [47] suggested that the particles fuelling biodiesel were higher than fuelling diesel. The particles generate immediately after combustion in which the instantaneous temperatures are relatively high. Hence, more AAC effects are caused by the particles adhered to the lubrication surfaces. Therefore, between the two alternative fuels, biodiesel produces more impact because it has higher AAC-AE responses, which occur at high speeds due to high temperatures.

Diagnosis with AAC Effects
The diagnosis of AAC-AE indicator shows a slight abnormality accompanied by FAS reaction. This demonstrates the potential of using AE to conduct a comprehensive analysis of the tribological effects of alternative fuels. different loads. The AAC-AE trends are unstable at speeds. Especially, the AAC-AE of biodiesel has high values in the power stroke and exhaust stroke. This can indicate that the combustions of these alternative fuels are less perfect. Reference [47] suggested that the particles fuelling biodiesel were higher than fuelling diesel. The particles generate immediately after combustion in which the instantaneous temperatures are relatively high. Hence, more AAC effects are caused by the particles adhered to the lubrication surfaces. Therefore, between the two alternative fuels, biodiesel produces more impact because it has higher AAC-AE responses, which occur at high speeds due to high temperatures. The diagnosis of AAC-AE indicator shows a slight abnormality accompanied by FAS reaction. This demonstrates the potential of using AE to conduct a comprehensive analysis of the tribological effects of alternative fuels.

Conclusions
The AE signals can reflect the dynamic information of tribology behavior of engines by AAC-AE and FAS-AE. Around the middle of each stroke, significantly higher sliding

Conclusions
The AE signals can reflect the dynamic information of tribology behavior of engines by AAC-AE and FAS-AE. Around the middle of each stroke, significantly higher sliding speed leads to more hydrodynamic lubrication in which little AAC can occur, but high AE activities arise mainly from the FAS effects. A new adaptive threshold-wavelet packet analysis was proposed as an effective tool to extract the tribological AE from the ringliner contact surface with more details. The FAS-AE indicator was acquired by the mean envelops of the reconstructed WP coefficients in several frequency bands: 40-60 kHz, 70-90 kHz, and 140-160 kHz, because of a higher correlation of these bands towards the piston speed than 170-190 kHz. The AAC-AE indicator was acquired by the difference between raw WP coefficients and threshold WP coefficients from 40 kHz to 200 kHz. Therefore, the two indicators based on two classical AE generation mechanism on the sliding surface can give a comprehensive evaluation of the tribological impacts of AL fuel as follows.

1.
The FAS-AE indicators are increased with speed and viscosity increasing. The AAC-AE is less significant using diesel than using biodiesel.

2.
The developed FAS-AE indicators from AE signals for biodiesel show tiny higher than the baseline diesel with the same lubricant 10W30 and similar to the baseline using oil 15W40. The developed FAS-AE of F-T diesel is close to the baseline diesel using 10W30.

3.
The FAS-AE exhibits the impacts on the lubricity of the oil film fueling the alternatives. Too high FAS-AE indicates high power consumption to overcome viscous friction using AL fuel, and too low FAS-AE shows the decreasing the lubricity of oil film using AL fuels.

4.
Biodiesel produces more AAC impacts with higher AAC-AE responses than F-T diesel, which occurs at high speeds due to high temperatures and more particles after combustion than diesel. 5.
The AL fuel diagnosis of AAC-AE indicator shows a slight abnormality accompanied by FAS. That demonstrates the potential of AE to conduct a comprehensive analysis of the tribological effects of alternative fuels.
Author Contributions: N.W. and F.G. were involved in the full process of producing this paper including modelling, test designing, data processing and preparing the manuscript. Z.C. and Y.X. provided valuable comments for improving the model and test design, A.B. reviewed the model and improved data process and edited the manuscript at all stages. All authors have read and agreed to the published version of the manuscript. Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The data are not publicly available due to privacy.