Research on the Jet Milling Process of Oxidizer for Solid Propellant

The particle size of the oxidizer used by the solid rocket propellant has a great impact on the combustion performance of the grain. In order to study the influence law of the oxidizer used for the jet milling, the jet milling process test system is constructed. The influence of the main process parameters on the average particle size was studied by single-factor test.The orthogonal test was designed, and the prediction model of the relationship between particle size and milling pressure, milling chamber material mass and classification wheel speed was established using multiple linear regression method, and the significance analysis and the prediction test were conducted.The results show that the particle size prediction model is significant, and the classification wheel speed affects the average particle size the most, followed by milling pressure and milling chamber material mass.The conclusion is that the deviation of the milling average particle size under the prediction model acquisition process parameters is between 1.79%∼5.88%, and the prediction model acquisition process parameters can be obtained to mill the AP with the average particle size requirements of different specifications.


Introduction
As an important part of solid propellant, oxidizer provides a guarantee for reaction, and the wide variety of oxidizer materials, commonly used oxidizer materials are ammonium perchlorate (AP), lithium perchlorate, ammonium nitrate (AN), ammonium dinitroamide (ADN), nitroform hydrazine (HNF), etc [1].The current widely used composite solid propellant (CSP) is an HTPB solid propellant, which mainly includes oxidizer(AP)(60%~80%), terminal hydroxy polybutadiene(HTPB)(10%~15%), aluminum powder and other metal fuels(15%~20%) [2].Many researchers have studied the positive role of ultrafine ammonium perchlorate in propellant, including Buckmaster J and K.V. Suresh Babu et al. [3][4][5], studied the effect of ultrafine ammonium perchlorate in CSP on propellant viscosity, and proved the necessity of ultrafine ammonium perchlorate, and obtained the optimal ratio of AP particles in each size in order to get better propellant performance and the combustion rate .The study of V. A. Strunin et al. [6] showed that the AP particle size will affect the propellant combustion rate and the combustion limit, and that the contradiction between particle volume and particle surface due to different particle size will cause AP endothermal processes and combustion failure.Sunil Jain et al. [7] studied the effect of AP particle size and shape on the propellant performance, and as the AP particle size decreases, the propellant viscosity increases, the combustion rate accelerates, and the shape of the particles will become increasingly irregular, which is reflected in the reduction of the shape factor.However, for the research and description of oxidizer milling process is not much, in the field of particle milling many methods can achieve the requirements of AP milling, such as jet milling, mechanical milling, chemical milling, supercritical fluid method and other methods, but from the perspective of environmental protection, milling energy consumption and milling quality jet milling is the most efficient, clean, low cost and high quality method.Fluidized bed jet milling technology is widely used in the chemical industry, metallurgy, mining, military industry and other fields of material milling, its milling accuracy can reach 0.1 μm, and plays an important role in the production of ultrafine powder and other fields [8].
In this paper,we will mainly study the jet milling process of AP.Based on oxidizer milling equipment, study the influence law of milling process parameters on average particle size, conduct theoretical analysis through prediction regression mathematical model and verify through test.

Milling process test and analysis
Because the particle size distribution of oxidizer in solid propellant has a great impact on the combustion performance of the grain, it is necessary to prepare a variety of oxidizer particles with particle size distribution to obtain different combustion properties.The propellant oxidizer used in the process test is ammonium perchlorate (AP).The AP selected in the test is provided by Dalian Gaojia Chemical Co., Ltd., and the particle size is not more than 630 μm before milling.Figure 1 is the equipment of the jet milling AP.It is an automatic milling system for QLD350 oxidizer produced by Shanghai Xichuang Powder Equipment Co., Ltd.The milling pressure can be adjusted within 0.6~1.2MPa, the processing capacity is 20~250kg/h, and the maximum speed of the classification wheel can reach 8000 r/min.The milling principle is that after the material into the milling chamber in the supersonic nozzle formed a negative pressure area, under the action of

Establishment of the mill prediction model
The factors affecting the milling particle size distribution and the milling efficiency include the physical parameters and the process parameters.This paper mainly studies the influence of process parameters on the average particle size after milling.

3.1
Single-factor test Perform single-factor tests based on the process parameters involved in the actual milling process.In the single-factor test, single-factor tests are conducted on the milling pressure, milling chamber material mass, classification wheel speed, and the average particle size after milling is the target value.The results are shown in Fig. 2   According to Fig. 2, when the pressure is greater than 0.7MPa, the average particle size change is no longer significant.The analysis of the reason is that the increase of the milling pressure causes the agglomeration and bonding of the ultrafine particles, so that the ultrafine particles cannot leave the milling chamber at the classification wheel, and thus enter the cycle of the milling chamber again [9].According to Figure 3, the average particle size is gradually declining with the increase of the milling chamber material mass, but the particle size has slowed down significantly after 18kg.The reason is that excessive milling chamber material mass increases the retention of the milling chamber.Although the production of ultrafine particles is increased, the distance among the particles is also reduced, making the movement speed of the particles decline, and the frequency of the collision and the collision probability is also limited [10].
In Figure 4, the influence of the classification wheel speed on the average particle size of the oxidizer is basically consistent with the trend of the single-factor test curve of milling pressure and milling chamber material mass, but the inflection point in the average particle size at 14Hz~16Hz.This phenomenon stems from when the speed of the classification wheel is too high, resulting in the formation of clusters near the blade, and the clustered particles cannot enter the re-cycle through the classification wheel, so that the amount of ultrafine particles is decreased, and less particle with a smaller mean particle size were also detected [11].

3.2
Orthogonal test design and analysis Orthogonal test design is a multi-factor and multi-level test design method based on the probability theory and mathematical statistics.According to Galois theory, it uses standardized orthogonal tables to select an appropriate number of representative points from all the horizontal combination data of factors for testing, effectively reducing the number of trials and representing the analysis of comprehensive trial situation.According to the results of single-factor test data, the orthogonal test was designed at three group levels corresponding to milling pressure P, milling chamber material mass m and classification wheel speed f were selected respectively.The data and results are shown in Tab.1.The range analysis of the data is conducted according to the results of the orthogonal test (Tab.2).Since the range is the changing range of the test index when the level of the factor is changed, the primary and secondary factors can be judged by the range R value of any factor.Therefore, the range R of milling pressure, milling chamber material mass and classification wheel speed is calculated.According to the data, the primary and secondary order affected is: classification wheel speed> milling pressure> milling chamber material mass.

3.3.1
Establishment of the regression prediction model.The nonlinear function model was constructed based on the data of the process test in Tab.3 and was used to establish the prediction model of the AP jet milling process parameters and the average particle size [12].Suppose that the AP average particle size dp satisfies: (1) Where dp is the average particle size of AP, P is the milling pressure, m is the milling chamber material mass, f is the classification wheel speed, and k, α, β and γ are to set the coefficient to be determined.Through digital processing: Where lndp lnP, lnm, lnf are respectively y, x1, x2, x3, and lnk, α, β, γ are a0, a1, a2, a3, the corresponding linear equations is: ), the linear regression equation is established from the n (n=16) group data: , can be expressed in matrix form: Calculated by the least-squares method can obtain a vector: The parameters and measurement results of the orthogonal test are substituted into the equation 6, and the calculation can be calculated .The prediction model of the particle size of AP jet milling is: The final predictive regression mathematical model is obtained by bringing the coefficients into equation 7:

Significance test and validation test.
In order to verify the accuracy and reliability of the prediction model of particle size mean in equation 8, SPSS software is used, and the input variables are the milling pressure P, the milling chamber material mass m, and the classification wheel speed f for the significance test.The verification results are shown in Tab.4 to Tab.6.
According to Tab.4, the influence value of three main process parameters on the change of average particle size after milling is 95.7%> 70%, that is, three process parameters of milling pressure, milling chamber material mass and classification wheel speed can obviously reflect the change of milling particle size average and the reliability of process test parameters.The Durbin-Watson (D-W) is 1.753, so the mathematical model of the mean prediction regression of the milled particle size consistent with independence.As can be seen from Tab.6, the significance P of each process parameter is less than 0.05, so the prediction regression mathematical model of milling average particle size is highly significant.The process parameters on the impact of particle size average in order, the classification     According to the predicted value of the regression model of Tab.7, the combination of process parameters with the smallest average particle size is 6th: the milling pressure is 0.7MPa, the milling chamber material mass is 20kg, the classification wheel speed is 20Hz, and 8 milling tests (Fig. 6) are conducted, so as to verify the consistency of milling and the stability of the average particle size.

Conclusion
In this paper, through the collection and analysis of the process parameters of oxidizer AP milling, the influence of each factor on the average particle size is analyzed based on the single-factor test, and the primary and secondary order of each factor on the average particle size is determined by the orthogonal test range analysis.According to the parameters obtained by the process test, the regression prediction model of the milling average particle size is constructed, and its effectiveness and significance is verified.The regression coefficient test results verify the accuracy of the range analysis on the primary and secondary order of the influencing factors.The milling system is stable and reliable with small prediction error, and has certain reference value for the research and parameter control of the jet milling process of AP.The future work should focus on the internal mechanism of milling, and simulate the milling process through discrete elements method, so as to realize the purpose of numerical simulation instead of milling process test, reduce the test cost to a certain extent, and provide a new idea for the optimization of oxidizer milling process parameters.

Figure 2 .
Figure 2. Single-factor test diagram of milling pressure.

Figure 3 .
Figure 3. Single-factor test diagram of milling chamber material mass.

Figure 4 .
Figure 4. Single-factor test diagram of classification wheel speed.

Fig. 5
show the equipment for the milling process test.In order to further verify the accuracy of the mathematical model, further test verification was done, and the average particle size obtained from multiple trials obtained the mean of three trials, and test parameters results are shown in Tab.7.The error between predicted value and test value was 1.79%~5.88%,indicating the effectiveness of the mathematical model.

Figure 5 .
Figure 5. Equipment for milling process test.

Figure 6 .
Figure 6.Average particle size change in milling process tests.

Table 1 .
Single-factor test parameters table.

Table 2 .
Results of range analysis and influence order.

Table 3 .
Milling test scheme and test results.

Table 4 .
Model analysis result.