Experimental Study of Parameters of Grain Milling Product Separation in Pneumatic Screw Classifier

Present study was aimed to improve efficiency of sieve and screen separators by reducing specific load on a sieve (screen) through preliminary separation of the initial mixture of crushed material on air delivery stage in pneumatic screw classifier. Parameters of grain milling product separation in pneumatic screw classifier using planning of multifactorial experiment and processing of statistical data are presented. We have built pneumatic screw classifier prototype and performed its production check basing on these results.

Sorting of a material mixture into several fractions differing in size, density, and aerodynamic properties is one of the most important operations in many industrial branches, where grinding of raw material with subsequent classification of obtained milling products or simply sorting of the original product into several fractions is required.One of the promising ways to intensify mentioned technological processes is to use pneumocentrifugal devices.
Efficiency of a grain processing facility depends on reliability of technological lines and a number of losses of raw materials at all process flow stages, while product quality becomes top priority.Sorting processes on such facilities are based on screen and sieve separators 11 .
Many studies of Russian and foreign scientists are devoted to improvement of their efficiency.Obtained results suggest that efficiency of sieving depends on following factors: specific load on the sieve; homogeneity of granulometric composition, shape and state of processed material surface, methods of cleaning and aligning abilities of sieves 8,9,10 .
Fractional technology is one of the promising ways to intensify bulk material sorting: fractionation of source material according to aerodynamic properties 13,14,15 .
In this regard, it is reasonable to divide initial mixture into several fractions similar by particle size and shape and send them to the appropriate systems to enhance sieving efficiency.
This problem may be technically solved by using pneumo-centrifugal separators for fractionation of milling products during their transportation.They are more energy-efficient compared to separators with linear air currents.
Studies aimed to improve efficiency of separation based on the patterns of particle movement in centrifugal force field allowed to elaborate pneumatic screw classifier for separation of grain milling products 7 .

METHODS
Experimental studies were conducted to confirm basic theoretical regulations, as well as to determine the rational parameters of grain grinding products separation in pneumatic screw classifier.Experimental unit was built for laboratory studies.The unit consisted of pneumatic screw classifier Figure 1, two cyclone-separators each with own ventilator, bunker with dozer, and material lines.
Research single-factor experiments were conducted to determine the significance and optimal interval of factors for subsequent studies.During these experiments we evaluated parameters influencing grain milling product separation in pneumatic screw classifier 2 .
It was found that the best results are obtained in selected intervals of technological and design parameters, Table 1.
Total extraction factor h (%) was selected as optimization criterion; it is determined by formula (1):

...(1)
Where h 1 , h 2 , h 3 are extraction factors of large fraction from the first outlet, medium fraction from the second outlet, fine fraction from the third outlet, respectively Extraction factor h 1 , h 2 , h 3 (%) is characterized by the ratio of extracted particles P i to their quantity in the initial mixture P 0 .It is calculated according to the formula: ...( 2) Basing on a priori information, it is assumed that the response function is described by a second-order polynomial: .... (3)   We selected composite symmetric threelevel design B 4 considering the number of significant factors and recommendations for choosing experiment designs [5].This design was selected on the basis of recommendations for choosing designs with the best joint characteristics.Design matrix is shown in Table 2.
Pneumatic screw classifier was built and tested on flour-grinding mill MVS-01 with performance of 1000 kg/h.
The screening and factorial experiments justified values of geometrical and technological parameters of pneumo-centrifugal air separation system.Pneumatic screw classifier was designed and built according to the technical task.
Test program of experimental unit was based on IS 101.3-2001.Test program included: examination of unit construction, assessment of working conditions, determination of performance indexes for optimal performance.Technical characteristics of the unit are based on the results of laboratory tests, designed and manufactured experimental sample of pneumatic screw classifier.Technical characteristics should contain indicators resulting from cyclone type 6 .
Adjustment experiments were conducted in order to determine optimal adjustment operation mode.One should focus on extraction purity of required particle fraction while setting the air flow rate.Three experiments in each mode with each grinding system were conducted.
Material was fed into pneumatic screw classifier through designed receiver with adjustable flow splitter to regulate material load on the classifier.
The experiment began in steady-state mode.Material flow was blocked by the flap, and all outputs were directed to containers; after each repeated experiment material flow was stopped on signal by closing the flap, and time of sampling was placed.Extracts were weighted and sampled for analysis.All output samples had labels.Unit performance was determined according to the air flow rate per hour.

Analysis of samples taken during unit test consisted of -
Collection of sample weights for determination of quality of the source material and extracts from the unit; -Determination of fractional composition of source material and extracts.Obtained results were processed using mathematical statistics, dependency graphs were plotted, the the total extraction factor h (%) with assessment of unit competence in the process flow design.
In order to calculate economic efficiency, we drew milling balance before and after the introduction Table 4 and 5.This document fully reflects all features of technological process in this facility.
Milling balance is a tabular record of distribution of all products on technological systems, as well as extracts of products from all systems.Milling balance reflects not only technological process according to its design, but also process management, therefore it provides a complete analysis of the process flow in the facility.
Calculation of milling balance requires setting the load on I break system as 100%, i.e. grain mass changes in the preparatory division of a mill are not taken into account due to removing of impurities and moisturization of grain.Therefore, the amount of obtained flour and bran, as well as semolina (if present), should be 100%.Mass of all products is expressed in percentage to I break system.
These balances are recorded as tables: for each system separately or for total milling -as so-called cross tables.
Pneumatic screw classifier is a part of the experimental unit.The main body contains pneumatic crew channel with radial flow into axial pipe windows.
Pneumatic Pneumatic screw classifier was mounted directly behind the grinding mill of the first break system in production line.Air flow rate in outlet pipes was regulated by TRIAC transducers and grinded grain load -by a splitter set into the drift from grinding mill on the plansifter.

RESULTS
Determination of aerodynamic properties of grain milling showed that particle suspension velocity ranges from 0.5 to 5.5 m/s, estimated equivalent diameter ranges from 122 to 1040 µm, respectively.
Research data on the influence of mean rate ratio in the axial and tangential pipes indicate that extraction factor maxima in pipe samplers are obtained under the ratio of the rate in pneumatic duct to the rate in axial pipe V pd /V ax H" 0.8.
Study of influence of the conical part of the unit on grain milling product separation in pneumatic screw classifier allowed to determine rational angle of conical part disclosure as 60 degrees.
It was also found that grain milling products divided according to suspension velocities after I break system have different internal friction coefficients.
Experiments with design matrix were followed by data processing and building of mathematical model.Quadratic model coefficients were determined according to formulas 5 .
Processing of experimental results were conducted using recommendations 1,3,4 .
In accordance with the hypothesis adopted in this study, we attempted to create a comprehensive study of the pneumatic screw classifier that would ensure the grain milling product separation into three fractions with different suspension velocities.Table 2 presents the results of experimental research.
We obtained regression equation in form of (x 1, x 2, x 3, x 4) based on experimental results and their statistical We obtained regression equation η (x 1 , x 2 , x 3 , x 4 ) based on experimental results and their statistical processing using standard programs.. Mathematical model Check of the model adequacy was carried out using Fisher criterion.Regression equation in decoded form: ...( 5) Equation ( 4) was cited to canonical form for the analysis and systematisation:  pneumatic duct pipe U 2 = 6.7752 m/s, material load on the classifier, q = 0.0285 kg/s).Cross-section of X 1 and X 2 were obtained through substitution of x 3 =0 and x 4 =0 in equation ( 4): We differentiated equation (7) for each variable, equated the derivatives to zero and got a system of linear equations consisting of two equations.Solution of linear system of equations is new response surface center coordinates: x 1 = 0.2547; x 2 = 0.1368.
After substitution of found values x 1 and x 2 in equation ( 7) we obtained the value of the total extraction factor in the surface center Y S = 91.69.
Canonical transformation of equation ( 7) was conducted; it is expressed by the following equation: ... (8)   Response surface is a paraboloid (Figure 2).Both factors B 11 and B 22 have the same signs.Centers of ellipses represent minima, as factors are positive and ellipses are extended along the axis x 1 : ... (9)   In this case factor value x 1 of the adopted factor variation interval moves a 0.2547 variation step away from the plan center and makes 13° inkind; factor value x 2 moves a 0.1368 step or 10.3 m/ s, while Ys = 91.69%and angle of coordinate axis rotation from the initial state is á = 11°.
Analysis of the response surface Figure 2 indicates that the rate changes in axial pipe to the right and to the left from the response surface center leads to a higher extraction factor (92.8-96.4%,which makes 3.7%) than change of pitch angle (95.29-96.40%-1.1%).Therefore, the rate in axial pipe (X 2 ) has a greater impact on the total extraction factor than the pitch angle (X 1 ).
Study of influence of X 1 and X 3 on optimization criterion was conducted in a similar way.Regression equation (10) and response surface (Figure 3) were obtained, center of factor variation intervals was shifted, in coded form:  the response surface -hyperbolic paraboloid.Analysis of the response surface Figure 3 indicates that the rate changes in axial pipe to the right and to the left from the response surface center leads to a lower extraction factor (90.7-93%, which makes 2.5%) than change of pitch angle (90.7-89.5% -1.3%).Therefore, the rate in pneumatic screw channel pipe (X 3 ) has a greater impact on the total extraction factor than the pitch angle (X 1 ).
Study of influence of X 1 and X 4 on optimization criterion was conducted in a similar way.Regression equation (11) and response surface (Figure 4) were obtained, center of factor variation ... (11)   Analysis of the response surface Figure ).Therefore, the material load on the classifier (X 4 ) has a greater impact on the total extraction factor than the pitch angle (X 1 ).
Study of influence of X 2 and X 3 on optimization criterion was conducted in a similar way.Regression equation ( 12) and response surface (Figure 5) were obtained, center of factor variation intervals was shifted, in coded form:   right and to the left from the response surface center leads to a higher extraction factor (89.5-93%, which makes 3.7%) than change of rate in pneumatic duct pipe (90.7-89.5% -1.3%).Therefore, the rate in axial pipe (X 2 ) has a greater impact on the total extraction factor than the rate in pneumatic duct pipe (X 3 ).
Study of influence of X 2 and X 4 on optimization criterion was conducted in a similar way.Regression equation ( 13) and response surface (Figure 6) were obtained, center of factor variation intervals was shifted, in coded form: ... (13)   Analysis of the response surface Figure 6 indicates that the rate changes in axial pipe to the right and to the left from the response surface center leads to a higher extraction factor (90.4-93.9%,which makes 3.8%) than change of material load (93.9-95.7%-1.9%).Therefore, the rate in axial pipe (X 2 ) has a greater impact on the total extraction factor than material load on the classifier (X 3 ).
Study of influence of X 3 and X 4 on optimization criterion was conducted in a similar way.Regression equation ( 14) and response surface (Figure 7) were obtained, center of factor variation intervals was shifted, in coded form: x 3 = -0.0926,x 4 = -0.4669;Y S = 92.17,axis rotation angle á = 30.27°.Regression coefficients B 33 = -2.4781;B 44 = -1.5285: can be seen from equation(6), regression coefficients of canonical equation have different signs, therefore, minimax-type response surface[1] with coordinates of the figure center x 1 = 0.4017; x 2 = 0.1262; x 3 = 0.1124; x 4 =-0.364(factors are, respectively: pitch angle, ° = 13.6068°;air flow rate in axial pipe U 1 = 10, 2524 m/s, air flow rate in x 1 = 0.1988; x 3 = 0.0191; Y S = 91.75,axis rotation angle á = -2.41°,regression coefficients B 11 = 1.20665;B 33 = -2.24292:....(10) In this case, coefficients B 11 and B 33 have different signs.Hyperbola are stretched along the B 11 axis with a lower absolute value of coefficient in canonical equation.In this case, the response value increases from the center of figure along this axis and decreases along the B 33 factor axis. Center of the response surface is called saddle or minimax,

Table 1 .
levels of experimental factors

Table 2 .
Experimental research results

Table 3 .
Results of production check of pneumatic screw classifier

Table 4 .
Basic quantitative milling balance

Table 5 .
Designed quantitative milling balance