Motor oil condition evaluation based on on-board diagnostic system

The condition of the motor oil in civilian cars is difficult to monitor; hence, we propose a method to evaluate the degree of degradation of motor oil using an on-board diagnostic (OBD) system. Three civilian cars and four motor oils (containing mineral oils and synthetic oils) were subjected to five groups of road tests under urban traffic and high-way conditions. The operation information, oil service time, mileage, engine operation time, idle time of the engine, and number of start-ups of the engine were obtained using the proposed OBD system. Physiochemical properties and changes in the components of motor oils during road tests were analyzed in laboratory. The theoretical model of the comprehensive indicators of driving parameters and oil properties were established. The proposed method was successfully applied to different cars, motor oils, and operating conditions in road tests. All the theoretical models had high accuracy and precision. Herein, we provide a method to monitor the oil condition with real-time driving parameters and provide a reference for end users to change their motor oil reasonably.


Introduction
Motor oil is an essential part of fuel-based vehicles. It provides wear protection, thermal management, and corrosion inhibition functions that are crucial for operation of the vehicle [1−3]. Regardless of the type of oil used in vehicles, degradation and/or contamination under complicated working conditions cannot be avoided [4−6]. Therefore, the motor oil must be changed to meet the normal working requirement. Although certain new energy vehicles have been developed, fuel-based vehicles continue to dominate the vehicle market. Fuel-based vehicles cannot be completely replaced in a short period of time. Excessively lengthy oil drain intervals increase wear in the engine and the likelihood of engine damage. If the intervals are too short, unnecessary preventive maintenance costs, energy wasting, and environment pollution are caused [7−10]. Study also shows that draining the motor oil too frequently may lead to a high concentration of additives in the oil. This can cause a reaction with the lubricant-surface and result in excessive wear [11]. Hence, a reasonable oil change interval is necessary for energy conservation, environment protection, and maintain cost saving.
Generally, there are two methods to determine the oil change interval. One method is sending oil samples to a laboratory to analyze the properties of the oil to determine whether the oil still meets certain criteria. The aforementioned oil analysis method can accurately determine the properties of oil; however, the long testing time and high cost limits its application [12−15]. Another more widely used method to change motor oil is based on the mileage or service time recommended by the original equipment manufacturers (OEM). As end users can easily monitor the miles that the vehicle has driven between oil change intervals, the recommended oil change interval has become widely accepted [16]. The recommended mileage that the OEM suggests for oil change intervals is based on various levels of severity of operation, which are rarely seen by consumers. It is impossible for the OEM to anticipate all operations of a user and list different oil drain intervals for each of them. In particular, most vehicles are used for more than one kind of operation. Hence, it is not easy to determine an optimum mileage for accurately changing motor oil [17−19]. Some scholars and OEMs attempt to use sensor technology to determine the oil life. Wang et al. [20] proposed a real-time sensor system that measures engine parameters and applies a special algorithm to indicate the oil drain interval. Jun et al. [21] applied the principal component analysis method to estimate the quality of vehicle engine oil based on oil viscosity indicators and certain engine operation parameters. General Motors has implemented an oil life monitoring system by monitoring the oil temperature and contaminations, and a penalty factor and engine speed are combined to simulate different operation speeds [22−26]. The application of such sensors and algorithms are limited due to cost, complexity, and limited utility of sensors and the errors caused by the algorithms. Jan Kral et al. [27] studied the features and qualities of 13 oil samples recommended for replacement by the onboard computer. The properties of kinematic viscosity, total base number, the amount of soot, oxidizing and sulphating products, water, fuel and glycol contamination, and high antioxidant presence were measured. The results did not correspond with the conclusions recommended by onboard computers.
Oil degradation is closely related to the operating conditions. The working load in the operation state and the idle state is different, which affects the working pressure of the engine oil. Frequently stopping and starting the engine results in continual oil temperature changes. Driving for short trips may cause unburned fuel and/or water to come into motor oil, which can reduce the viscosity and cause excessive wear of engine. Thus, it is necessary to establish the relation model between operation parameters and oil properties for scientifically determining motor oil change interval.
In this paper, a method to establish the theoretical model of operation parameters and oil properties based on road tests in urban traffic and high-way conditions was proposed. The theoretical model can directly reflect the change characteristics of motor oil properties with the operation parameters. This can be used to predict oil degradation in real-time based on the operation parameters of cars. This method can reduce the testing time and increase the accuracy for evaluating the oil change interval compared to the traditional laboratory oil analysis and stipulated operation mileage or service time, which helps change motor oil more economically and effectively.

Experimental details
The experimental cars and engine oils were tabulated in Table 1. The experimental cars include a 10 years old car (Experimental car No. 1), a 5 years old car (Experimental car No. 2), and a new car (Experimental car No. 3). All the experimental engines were port fuel injection-based and naturally aspirated. All the experimental cars were equipped with an on-board diagnostic (OBD) system. The OBD system was originally designed for monitoring emissions and fault diagnosis using a large number of sensors. The system can provide real-time operation information and trouble codes [28]. The OBD system has a connector for end users to access the diagnostic data. In this study, WiFi adapters are plugged into the OBD connector of cars, and the data can be manipulated using a cell phone application. Real-time engine operation time (EOT), mileage (MIL), service time (ST), engine idle time (EIT), and number of start-ups (NBS) were acquired using the OBD system and the cell phone application.
Oil samples were collected from the crankcase approximately every 30 days. Oil should be collected after the experimental cars stopped about half an hour. Collection via a vacuum tube inserted into the dipstick opening (sample a centimeter or two above the bottom of the oil pan). Sampling from the mid portion of the oil is preferable since the top and bottom portions are more likely to be contaminated, and the mid portion is more likely to represent what is flowing through the lubrication system. The component changes of oxidation, nitration, sulfation, zinc dialkyldithiophosphate (ZDTP) of oils were tested using the Integra software of infrared spectrometer (NICOLET iS10, Thermo Fisher Scientific, US). Total acid number (TAN) of oil samples were tested with reference to the ASTM D974-2014 standard [29]. The oxidation onset temperature (OOT) was determined using differential scanning calorimeter (NETZSCH HP 204, Germany) with reference to the ASTM E 2009-02 (the heating rate was 10 °C /min, the oxidation pressure was 3.5 MPa, the flow rate of oxidation was 100 mL/min) [30].

Results and discussion
Experiment No. 1-1 was taken as an example to demonstrate the details of the analysis and modeling process. The driving parameters and oil properties of Experiment No. 1-1 are given in Table 2. The driving parameters, ST, MIL, EOT, ITE, and NBS, obtained using the OBD system can completely represent the operation state of the cars in the experiment. The physicochemical properties (TAN and OOT) and component changes of experimental oil (oxidation, nitration, sulfation, and ZDTP relative change values) directly reflect the motor oil degradation degree.
As shown in Table 2, the motor oil in Experiment No. 1-1was effective to 410 days and the experimental was driven for 5723 km. The engine worked for 255.90 h and the idle time was 54.73 h. The engine was started 545 times. The TAN of the new oil was 2.17 mgKOH/g, which increased to 3.95 mgKOH/g after the experiment was concluded. The OOT value of the oil decreased from 243.7 to 198.9 °C during The detailed description of every step of the analysis method is provided below.
Step 1: Data pre-processing. The units and dimensions of the factors were different, and therefore, the original data obtained by the OBD system and the laboratory need processing before analysis. The average and initial values divided by original data are commonly used to pre-process data. In order to establish a theoretical model for oil degradation prediction, the initial value method was considered more suitable for this study. Some initial values of the factors (driving parameters, oxidation, nitration, sulfation, and ZDTP value) were 0, which cannot be considered as a dividend. Thus, the driving parameters and oil properties of the road test in 47 days were considered as the initial value. The result of pre-processing the data of driving parameters and oil properties is illustrated in Table 3.
Step 2: Comprehensive indicator calculation. This study attempted to establish a theoretical model between the comprehensive variation of driving parameters and the oil properties under different periods. The initial sequence was defined as the reference sequence X 0 . The data in the subsequent experiment were defined as the comparability sequence X i . The absolute variation represented the change degree of the two groups of data. The change degree of the driving parameters and oil properties can be calculated with Eq. (1). The comprehensive indicator γ i was calculated with Eq. (2) for modeling data.
where x 0 (k) is the element of the reference sequence, x i (k) is the element of the comparability sequence, Δ i (k) is the absolute variation of x i (k) and x 0 (k), and n is the number of the elements. The driving parameters are considered as an example to present the details of the calculation process. The X 0 and X 1 (in Table 3) values were presented as follows.  With the same method, the γ i of the driving parameters and oil properties in different periods can be calculated (in Table 4).
Step 3: Establishing the theoretical model. After the comprehensive indicators of driving parameters and oil properties were calculated, the theoretical model of driving parameters and oil properties was established. As shown in Fig. 1   The R square of the theoretical model was 0.999, which suggested that the theoretical model has high precision. The theoretical models for both Experiment Nos. 1-1 and 1-2 were high accuracy linear models.
The development trend of the oil properties of Experiment No. 1-2 (Fig. 4) can be determined with a similar method as that used for the Experiment No. 1-1. The OOT, oxidation, nitration, sulfation, and ZDTP values were 211.4 °C , 0.13 A/0.1 mm, 0.24 A/ 0.1 mm, 0.14 A/0.1 mm, and −0.11 A/0.1 mm, respectively, when the TAN reached the criterion for oil change (2.14 mgKOH/g). The limiting comprehensive indicator   of the oil properties and driving parameters were 2.4670 and 0.2115, which indicated that the motor oil should be drained when the comprehensive indicator decreased to 0.2115. The established theoretical model has high precision even though the road test exceeded the limiting comprehensive value; this also demonstrated that the proposed method has good reliability. Experiment No. 2-1 was carried out using Havoline mineral motor oil. The driving parameters and oil properties are presented in Table 5. Experiment No. 2-1 took 156 days, and the experimental cars operated for a total of 6317 km. The engine operated for 222.67 h, of which 45.56 h were in the idle state. The engine started 462 times during the experiment. The TAN of the oil used in Experiment No. 2-1 increased from 1.63 to 2.45 mgKOH/g. The OOT value of the used oil was 205.4 °C , which decreased by 27.6 °C , compared to that for the new oil. The Integra results of the used oil show that the oxidation, nitration, sulfation, and ZDTP relative change values were 0.09 A/0.1 mm, 0.20 A/0.1 mm, 0.13 A/0.1 mm, and −0.10 A/0.1 mm, respectively. The driving parameters and oil properties of 19 days were considered as the reference sequence. The theoretical model (Fig. 3) for Experiment No. 2-1 was y = 7.2328x + 1.6209, where x is the comprehensive indicator of driving parameters, and y is the comprehensive indicator of oil properties. The R square of the theoretical model was 0.960, which indicated that the linear model has high accuracy for representing the data.
The development trend of the oil properties for Experiment No. 2-1 (as shown in Fig. 5 (1) and (2). The motor oil needs to be changed when the comprehensive indicator decreases to 0.1197.   As per the operation information and oil properties presented in Tables 2, 5, and 6, the OOT values of synthetic oils were larger than 220 °C after the oils serviced more than 4938 km and 6471 km, which was higher than that for mineral oils (approximately 200 °C ). It is suggested that the oxidation stabilities of the experimental synthetic oils were better than those of mineral oils. The increments of the TAN of synthetic oils were smaller than that of the mineral oils, which indicated the advantages of the synthetic oil in reducing the production of acid products. The experimental synthetic oils have better comprehensive performance than the experimental mineral oils.
The average speed and idle ratios of the experiments are shown in Fig. 9. Since the Experiment Nos. 1-1, 1-2, 2-1, and 3-1 were conducted under urban traffic conditions, the average speeds for these experiments were 22.36 km/h, 23.43 km/h, 28.37 km/h, and 24.45 km/h, respectively. Experiment Nos. 1-1, 1-2, 2-1, and 3-1 have the characteristics of high idle ratios and low operation speeds, which are typical urban traffic conditions. The average speed for Experiment Nos. 1-1, 1-2, 2-1, and 3-1 were consistent with the average speeds of civilian cars in China's major cities (approximately 20−27 km/h). The Experiment No. 3-2 was carried out under urban and highway traffic conditions; the average speed for this experiment was 37.74 km/h, and the idle ratio was 18.2%. The average speed was higher and the idle ratio was smaller compared with those suitable for urban traffic conditions. All R-square values of the established theoretical models were larger than 0.96, which suggested that the proposed method used to establish the theoretical model between the driving parameters and oil properties had high accuracy and precision in both experimental mineral oils and synthetic oils under urban traffic and highway conditions.

Conclusions
This study was based on 41 oil samples in three experimental cars during the 575 days of road tests. The conclusions were as follow: 1) A method was proposed to establish the theoretical models of the comprehensive change characteristics of the driving parameters and oil properties. The proposed method has high accuracy and precision for both mineral and synthetic oils under urban traffic and highway conditions. The proposed method can help realize real-time oil condition monitoring with operation parameters obtained by the OBD system.
2) The results of the road tests in this study verified that the synthetic oils have better ability to restrain the increase of acid products and decrease oxidation stability compared to that of mineral oils. The oil change interval can also be appropriately extended by using the synthetic oil.