Abstract
This study applies stepped loading and stepped velocity approaches to simulate the actual working conditions of gasoline engines. Accelerated wear tests were conducted for piston rings and cylinder liners under different lubricating conditions by using a self-made pin-on-disc wear machine equipped with an on-line visual ferrograph (OLVF) for wear monitoring. The wear coefficients for oil monitoring were extracted to distinguish between constant conditions and stepped changing conditions. A similarity model for oil monitoring was constructed and the monitoring data sets of similar working conditions were grouped together. Results show that the OLVF monitoring system can be used to obtain the real-time variation in debris concentration. The index of particle coverage area (IPCA) of OLVF increases abruptly after the load or speed changes. The similarity model can evaluate the similarity of the variation trend of IPCA under different operating conditions. The relationship between IPCA and working conditions was examined in this study and provides an essential support to wear monitoring and life prediction of engines.
Nomenclature
FN | = | Applied contact force |
g | = | Sampling volume of OLVF within one sampling period (mL) |
H | = | Hardness of the material |
H1 | = | Hardness of the ring segment |
H2 | = | Hardness of the gray cast iron disc |
K | = | Relative wear coefficient for oil monitoring |
KOC | = | Wear coefficient for oil monitoring |
k1 | = | Wear coefficients of the ring segment |
k2 | = | Wear coefficients of the gray cast iron disc |
L | = | Number of segments |
pi | = | Probability of the ith interval |
q | = | Speed of oil supply within one sampling period is denoted by (mL/s) |
S | = | Cumulated sliding distance |
s | = | Relative sliding velocity |
T | = | Length of the sampling period (s) |
tn | = | Experimental time of one segmentation |
V | = | Total wear volume |
V0 | = | Volume of wear debris in the oil remnants of the last sampling period |
v(i) | = | Volume of wear debris in the oil pool during the current sampling period |
Δv | = | Total wear rate of the ring segment and cast iron disc |
Δv1 | = | Wear loss of the ring segment during time dt |
Δv2 | = | Wear loss of the gray cast iron disc during time dt |
β(i) | = | Wear debris concentration of oil within the ith sampling period |
δ1, δ2 | = | Weight values |
APPENDIX: CALCULATION OF RELATIVE WEAR COEFFICIENT FOR OIL MONITORING
Given that the variation in working conditions affects the wear coefficient, the relative wear coefficient for oil monitoring is derived to calculate the wear loss of testing materials by using the wear model of mixed lubrication: [A1]
Let the relative sliding velocity be then Eq. [A1] can be expressed by [A2]
The total wear rate Δv of the ring segment and cast iron disc can be expressed by [A3]
Let KM denote the comprehensive wear coefficients of one data segmentation: [A4]
Under the same working conditions, coefficients KM are the same. Thus, when FN and Δs increase, the abrasive generation speed of friction pairs also increases.
For this experiment, remnants of the last sampling period are present because the peristaltic pump had supplied oil before the wear test began. Let the volume of residual oil W0 (mL) be equal to a constant value, then the oil volume in the oil pool within each sampling period W(mL) can be expressed as [A5]
The volume of wear debris in the oil pool during the current sampling period can be expressed as [A6]
Considering that V0 is very small, this variable can be replaced by a random variable γ with mean-zero Gaussian distribution and γ is on the positive half-axis of Gaussian distribution. Thus, v(i) can also be expressed as [A7]
Here, Z sampling periods exist in a single segmentation. The total volume of wear debris in the oil within one segmentation can be expressed as [A8]
For the same monitoring data segment, the wear state belongs to the same wear stage, and the load, speed, and lubricants are the same. Given that the random variable γ(i) in the last item in Eq. [A8] is very small or close to zero, then Vsum can also be written as [A9]
Regardless of the cumulative effect of the wear debris,the wear debris concentration of oil within the ith sampling period can be expressed as [A10]
Ignoring the last item in Eq. [A10], the wear rate Δv and β(i) within a sampling period show a linear relationship.
The total wear volume of Z sampling period (Vsum) can be calculated using β(i) through Eq. [A11]. The monitoring data of the OLVF during each sampling period are IPCA(t). The sum of the IPCA(t) after time tn under same working conditions is IPCAsum and can be expressed as Eq. [A12]. [A11] [A12]
Previous experimental studies showed that IPCA(t) is influenced by β(i) within a sampling period; that is, IPCA(t) has a positive correlation with β(i). Equation [A10] denotes that β(i) is primarily determined by load, relative sliding velocity, and wear coefficient. Thus, IPCA(t) is also associated with load and relative sliding velocities.
On the basis of the previous study that IPCA(t) is positively correlated with β(i) and by combining Eqs. [A11] and [A12], it can be inferred that a positive correlation exists between the sum of the oil monitoring data (IPCAsum) and total wear volume (Vsum) within a single segment. Thus, similar to the KM in Eq. [A4], the wear coefficient for oil monitoring(KOC) was derived from the IPCA. Vsum in Eq. [A11] is replaced by IPCAsum, and the new coefficient K is determined by Eq. [A13]. Given that Z and T are definite values and KOC = k*Z*T, KOC, is determined using Eq. [A14], which indicates the relationships of IPCAsum and FN and Δs. [A13] [A14]