Optimization of Molten Salt Cleaning Process for Surface Roughness of Remanufactured 27SiMn Hydraulic Support Column

，，*，，，，，

1.Key Laboratory of High Efficiency and Clean Mechanical Manufacture（Ministry of Education），School of Mechanical Engineering，Shandong University，Jinan 250061，P.R.China；

2.National Demonstration Center for Experimental Mechanical Engineering Education，School of Mechanical Engineering，Shandong University，Jinan 250061，P.R.China

（Received 4 September 2019；revised 28 October 2019；accepted 20 April 2020）

Abstract: During molten salt cleaning of remanufactured 27SiMn hydraulic support column，oxidation occurs on the surface of metal substrate. This results in a change of the surface roughness of metal substrate after cleaning，which affects subsequent remanufacturing process. To decrease the effect is very important. This paper analyzed the oxidation mechanism of molten salt cleaning，explored the oxidation reaction that occurred during cleaning，and determined the key process parameters of cleaning that affecting oxidation reaction. By using central composite experimental design method and taking surface roughness variation of 27SiMn steel samples before and after molten salt cleaning as response variable to optimize the key process parameters，the optimal parameters of molten salt for cleaning remanufactured 27SiMn hydraulic support column could be obtained. The results show that the oxidation reaction of cleaning paint dirt can protect metal substrate from oxidation to a certain extent，and cleaning temperature and placement depth of metal substrate have a direct impact on the degree of oxidation reaction. When the cleaning temperature is 300 ℃and the distance between paint dirt and free surface of molten salt is 0.5 times the height of the parts，the surface roughness variation is minimal. Therefore，the cleaning quality will be improved under such parameters.

Key words：surface roughness；molten salt cleaning；hydraulic support column；central composite experimental design；remanufacturing

0 Introduction

As an extension of equipment manufacturing in?dustry chain，remanufacturing is one of the most ef?fective ways to make resource recycled［1］. Hydraulic support is the key system to make sure safety in un?derground coal mining process［2］. Hydraulic support column，as shown in Fig.1，is the main bearing part of hydraulic support，which can bear and transmit the main load from the roof of fully mechanized min?ing face. Due to long-term use in harsh environment of acid and alkali corrosion，hydraulic support col?umn is subject to corrosion，wear，impact and other comprehensive effects，resulting in the peeling of surface paint，and damage of substrate［3］. This will cause equipment failure and affect the whole coal mining process. Therefore，remanufacturing of hy?draulic support column is an inevitable trend for the sustainable use of mining machinery［4］. It is the first step in remanufacturing process to do a good job of cleaning paint dirt on hydraulic support column.Cleaning quality directly affects subsequent remanu?facturing. Molten salt cleaning is an efficient indus?trial cleaning method which uses molten salt as the cleaning medium. It combines physical and chemical action to clean carbon，paint，oil and other typical dirt［5-7］. Because of low energy consumption and lit?tle pollution，this method has been gradually ap?plied in cleaning of mining machinery.

Fig.1 Hydraulic support column

Under the premise of high efficiency and envi?ronmental protection，the effect of molten salt clean?ing on material properties of remanufactured hydrau?lic support column determines whether it meets re?quirements of subsequent remanufacturing. In re?cent years，domestic and foreign scholars have ex?plored the effects of molten salt cleaning on various properties of iron-based alloy materials commonly used in mining machinery［8-10］，including surface roughness，hardness and tensile strength，etc. The results show that molten salt cleaning has little ef?fect on hardness and tensile strength of iron-based al?loy materials，but has a significant effect on surface roughness. The reason is that oxidation reaction oc?curs continuously in the process of molten salt clean?ing，which leads to surface oxidation of metal sub?strate［11-13］. The resulting oxide layer has a great ef?fect on the surface roughness. Surface roughness is closely related to wear resistance，fatigue strength，contact stiffness，vibration and noise of mechanical parts，which has an important impact on service life and reliability of mechanical products［14］. Therefore，optimizing the molten salt cleaning process to re?duce the influence on surface roughness is of great significance to ensure its remanufacturing quality and improve its service life.

In this paper，27SiMn steel used for manufac?turing hydraulic support column was selected for the optimization experiment of molten salt cleaning pro?cess. By adopting central composite experimental design method，and using surface roughness varia?tion of metal before and after molten salt cleaning as response value，the process parameters of molten salt cleaning were optimized. In the end，the opti?mal parameters，which realized efficient and highquality cleaning of paint dirt on 27SiMn hydraulic support column，were obtained.

1 Experimental Design of Molten Salt Cleaning

1.1 Oxidation analysis of molten salt cleaning

Oxidation is the main action for molten salt re?moving paint［15-16］. The formula of molten salt used in this study is composed by KNO3-NaNO3-NaOH.Ion detection of the used molten salt showed that it contained a large amount of CO2-3，but the content of OH-was greatly reduced. Analysis of gas prod?ucts revealed that no acidic nitrogen oxide gas was found，which was due to the absorption of NaOH in cleaning process.

The specific reaction equations of molten salt in cleaning paint dirt were

In molten salt cleaning process，the substrate of iron-based alloy materials was also oxidized. On one hand，Fe element reacted with molten NaOH to form Fe3O4；on the other hand，high temperature cleaning conditions catalyzed the oxidation of Fe ele?ment in aerobic environment. With the removal pro?cess of paint dirt，metal substrate was inevitably oxi?dized in molten salt medium，resulting in the change of surface roughness after cleaning. Studies have shown that both the oxidation of paint dirt removal process and the oxidation of metal substrate con?sume a large amount of NaOH，while the content of NaOH in the mixed molten salt system is limited.So，the oxidation reaction of paint dirt removal pro?cess can protect metal substrate from oxidation to a certain extent. Speeding up the reaction rate of paint dirt removal can shorten the cleaning time，which will reduce the contact time between metal substrate and molten salt，thus reducing the oxidation of sub?strate.

As the key condition of oxidation，cleaning temperature directly affects the degree of each oxida?tion reaction in molten salt cleaning process. And the amount of oxygen dissolved in molten salt is dif?ferent at different depths，which also affects the re?action. Selecting proper cleaning temperature and placement depth can reduce the oxidation of metal substrate and the change of surface roughness while completely removing the paint dirt. This can effec?tively reduce the effect of molten salt cleaning on the properties of iron-based alloy materials and im?prove the cleaning quality.

1.2 Experimental materials and equipment

27SiMn steel，as an iron-based alloy material，is commonly used for engineering machinery. It is often used in manufacturing of hydraulic support col?umn［17］. Its element composition and content are shown in Table 1.

Table 1 Elemental composition and content of 27SiMn steel

In this paper，27SiMn steel was selected as the research object，and the material was processed into 13 cylindrical samples with a diameter of 20 mm and a height of 10 mm. The samples were treated to make the surface condition similar to that of the hy?draulic support column，as shown in Fig.2.

Fig.2 Experimental samples before painting

Surface roughness of the 13 samples was test.Each sample was tested three times，and the aver?age value was taken as initial value. After surface roughness test，the samples were painted. The main components of the paint were the same as those used on hydraulic support column. The thickness of paint sprayed on the samples was 0.05（±0.005）mm.After painting，the samples were dried in a ventilat?ed environment to meet experimental standard. The experimental samples after painting were shown in Fig.3.

Fig.3 Experimental samples after painting

In this experiment，a molten salt cleaning equipment which can control temperature was used to clean the samples. The main body of the equip?ment consists of a cleaning tank，a heating and hold?ing furnace and a temperature control box to achieve precise temperature control. The surface roughness test used Wyko NT9300 white light interferometer with high measurement accuracy.

1.3 Experimental program

This experiment intended to optimize the clean?ing temperature and the placement depth of painted samples. Surface roughness variation of the samples before and after molten salt cleaning was taken as re?sponse variable. Two-level full-factor experiment was carried out by using central composite experi?mental design commonly used in response surface methodology. The total number of experiments was 13. The range of the cleaning temperature was 270—360 ℃and the distance was 5—65 mm（0.5—6.5 times the height of the sample）. Level setting of experimental factors was shown in Table 2.

Table 2 Level setting of experimental factors

The experimental steps for molten salt cleaning were as follows：

（1）A sufficient amount of the mixed salt that consisting of KNO3-NaNO3-NaOH was poured into the cleaning tank to fully melt.

（2）The experimental samples after painting were subjected to molten salt cleaning under 13 sets of specific temperature and distance.

（3）When the paint dirt was completely re?moved，the samples were taken out. Then the resid?ual molten salt on the samples was removed by us?ing an ultrasonic cleaner. After that，the samples were dried.

（4）The surface roughness of each cleaned sam?ple was tested and compared with the initial value tested before. By subtracting the initial value from the final value，the surface roughness variation was then calculated and denoted by ΔRa. The central composite experimental design and corresponding results were shown in Table 3.

Table 3 Central composite experimental design and cor?responding results

2 Response Surface Optimization of Molten Salt Cleaning Process

2.1 Fitting of regression equation

（1）Regression analysis

Based on the central composite experimental design，the surface roughness variation of 27SiMn steel samples under various temperature and dis?tance was investigated. A corresponding quadratic regression model was established by coding and ana?lyzing the two factors. The coded estimated regres?sion coefficients in the quadratic regression model were listed in Table 4，where SE coef. is the stan?dard error for coefficients，and Temp is the abbrevi?ation of temperature.

Table 4 Estimated regression coefficients for ΔRa

As demonstrated in Table 4，the correspond?ingP-values of temperature and distance in linear terms were both 0.000，which was much smaller than 0.05. This indicated that they were significant.The correspondingP-value of distance*distance in square terms was 0.030，also smaller than 0.05.However，the correspondingP-value of tempera?ture*temperature in square terms was 0.865 and the correspondingP-value of temperature*distance in in?teraction term was 0.757. Both of them were much greater than 0.05，which indicated that these insig?nificant terms should be deleted from the quadratic regression model. The estimated regression coeffi?cients after correction were listed in Table 5. From Table 5，we could see theP-values for linear and square regression terms were all less than 0.05，which meant they were all statistically significant.

Table 5 Estimated regression coefficients for ΔRa after correction

（2）Variance analysis

Variance analysis for the surface roughness variation was carried out，and the specific data were shown in Table 6，where DF is the degree of free?dom，Adj.SS is the adjusted sum of squares of devi?ation from mean，and Adj.MS is the adjusted mean square. The adjusted R-Sq value，which was rea?sonably identical to the R-Sq value，indicated that extra variables were not included in the model. Fur?thermore，theP-value for lack of fit was 0.478，which was much larger than 0.05. This indicated that the data model does not have any lack of fit.

Table 6 Variance analysis for ΔRa(R?Sq=96.83%, R?Sq（adj）=95.77%)

The coefficients of regression equation for the surface roughness variation of 27SiMn steel samples were shown in Table 7.

（3）Residual analysis

Based on the residual condition，fitting effect of the model could be diagnosed by analyzing the re?liability，periodicity and interference of the data.The residual plots of the surface roughness variation were shown in Fig.4.

Fig.4 Residual plots of ΔRa

As can be seen from the analysis of Fig.4 above，the residual values were normally distribut?ed，and the scatter points were randomly distributed above and below horizontal axis. These indicated that the residual was not abnormal.

（4）Establishment of regression equation

By regression analysis，variance analysis and residual analysis，the regression equation for the sur?face roughness variation was obtained as

where ΔRastands for the surface roughness varia?tion of 27SiMn steel samples before and after mol?ten salt cleaning，Tfor the cleaning temperature，Dfor the distance between paint dirt on the samples and free surface of the molten salt，andξfor the re?gression error.

2.2 Response surface analysis and target opti?mization

（1）Contour plot and response surface plot analysis

Contour plot and response surface plot based on the fitting model were shown in Fig.5 and Fig.6.And the trend of ΔRavs. temperature and distance could be observed in conjunction with these two fig?ures.

Fig.5 Contour plot of ΔRa vs.temperature and distance

（2）Target optimization of ΔRa

The response optimizer was used to optimize target value. The optimization target was to make ΔRaequal to 0，which meant that the cleaning pro?cess had minimal effect on the surface roughness.According to the response optimization results in Fig.7，when the temperature was about 300 ℃and the distance was 5 mm，i.e. 0.5 times the height of the sample，the response value was optimal.

Fig.6 Surface plot of ΔRa vs.temperature and distance

Fig.7 Response optimization plot of ΔRa

3 Conclusions

（1）Molten salt can effectively remove the paint dirt on hydraulic support column by oxidation. The iron-based substrate is also easy to be oxidized to Fe3O4，which will make the surface roughness of cleaned substrate changed. However，the oxidation reaction for paint removing will restrict the oxida?tion to substrate. In order to minimize the substrate oxidation，an optimal experiment was designed to explore the effect of cleaning parameters. Research shows that the cleaning temperature and the place?ment depth of metal substrate are the main factors for the degree of oxidation reaction.

（2）In this paper，the cleaning temperature and the distance between paint dirt on the samples and free surface of molten salt were optimized by using experimental design method of central composite and surface roughness variation of 27SiMn steel samples as response variable. With ΔRatending to zero as the optimization target，the optimal process parameters of molten salt for cleaning remanufac?tured 27SiMn hydraulic support column were finally obtained. Results show that when the cleaning tem?perature is 300 ℃and the distance between paint dirt and free surface of molten salt is 0.5 times the height of the parts， the optimization target is achieved.