List of Figures

Figure 1.1:

Control variables of an electrostatic separation process: high-voltage level U; roll-speed n; angular α1 and radial d1 position of the corona electrode; angular α2 and radial d2 position of the electrostatic electrode; angular positions γ1 and γ2 of the splitters.

27

Figure 1.2:

Schematic representation of a powder tribocharging installation and corresponding variables: fluidised bed pressure, injection pressure, dilution pressure, vortex pressure, charging device configuration and coating, powder granularity.

30

Figure 1.3:

Two electrodes configuration for a free fall separation chamber.

31

Figure 1.4:

Charge measurement setup.

33

Figure 1.5:

High voltage measurement setup.

34

Figure 1.6:

Electrostatic voltmeter: surface potential measurement principle [96].

35

Figure 1.7:

Circuit diagram of an electrostatic voltmeter [95].

37

Figure 1.8:

Instrument control system for voltage measurements [93].

38

Figure 1.9:

Front panel of the virtual instrument for single particle charge measurements.

40

Figure 2.1:

Charge insulating particles “pinned” to the surface of the grounded electrode of a roll-type corona-electrostatic separator; 1: grounded rotating roll; 2: wire type corona electrode; 3: charged particles.

45

Figure 2.2:

Insulated particles charge measurements setup: schematic representation (a) and physical embodiment (b).

47

Figure 2.3:

Design architecture of the charge measurement virtual instrument.

50

Figure 2.4:

Schematic representation of type I, II, and III particles

55

Figure 2.5:

View of the corona charging experimental set-up; 1: wire type corona electrode; 2: charged insulated particles; 3: grounded roll electrode

56

Figure 2.6:

Charge Q of type I particles as function of spacing d, for an applied voltage U = 20 kV and a distance s = 50 mm between the electrodes

57

Figure 2.7:

Charge Q of type II and type III particles as function of discharging time tdisch

58

Figure 2.8:

Custom Faraday cage for powder charge measurements

60

Figure 2.9:

Schematic representation of the experimental setup for powder tribocharging

61

Figure 2.10:

Physical embodiment of the charge measurement setup for powders

62

Figure 2.11:

Command structure for virtual instrument for charge vs. time measurements.

63

Figure 2.12:

Typical charge to time graph obtained with the virtual instrument for the powder tribocharge measurements.

65

Figure 2.13:

Graphs of the charge to time characteristic from the powder tribocharging measurement system obtained with the virtual instrument

66

Figure 2.14:

X-bar chart showing results from experiments 1, 2 and 3.

75

Figure 3.1:

Experimental set-up for the study of corona charging and discharging of granular insulating layers at the surface of a grounded electrode. In the photography left to right: electrostatic voltmeter, potential probe and corona electrode, high voltage supply.

81

Figure 3.2:

Insulated granular layer at the surface of a grounded electrode; spring actuated potential probe and dual corona electrode (detail from Figure 3.1).

82

Figure 3.3:

Virtual instrument for surface potential decay measurement - front panel.

83

Figure 3.4:

Two discharge curves obtained with the virtual instrument for the nonconductive fraction of a genuine chopped wire waste; Y-axis in kV, X-axis in measurement points corresponding to a 5 Hz sampling rate

101

Figure 3.5:

Forces acting on a charged insulating particle at the surface of a rotating roll electrode connected to the ground.

88

Figure 3.6:

Surface potential decay of fine (sample S1.1) and coarse (sample S1.2) PVC particles. Each curve is the average of at least 5 measurements.

91

Figure 3.7:

Surface potential decay of PE (sample S2) versus silicon rubber (sample S3) particles. Each curve is the average of at least 5 measurements.

92

Figure 3.8:

Surface potential decay of fine PVC particles before and after drying for 30 min. in hot airflow (60° C). Each curve is the average of at least 5 measurements.

93

Figure 3.9:

Surface potential decay of HDPE for RH=20% and RH=40%

97

Figure 3.10:

Surface potential decay of PVC for RH = 20 % and RH = 40 %.

98

Figure 3.11:

Surface potential variance due to RH dVHDPE and dVPVC

98

Figure 3.12:

Surface potential decay for PVC and HDPE at RH=40%

100

Figure 3.13:

Surface potential decay for PVC and HDPE at RH=20%

100

Figure 3.14:

Potential difference between HDPE and PVC: dV20=VHDPE-VPVC at 20% RH

101

Figure 3.15:

PE versus rubber; Surface potential ratio VS.2 / VS.3

102

Figure 4.1:

Roll-type corona-electrostatic laboratory separator (CARPCO Inc, Jacksonville, Fl) with high-voltage probe; 1: feeder; 2: corona electrode; 3: electrostatic electrode; 4: high-voltage connector; 5: high-voltage probe; 6: grounded roll electrode (carrier); 7 collector.

107

Figure 4.2:

Experimental set-up for HV monitoring, using a digital electrometer and a virtual instrument developed in a LabVIEW environment.

108

Figure 4.3:

Voltage restoring curves for the two HV supplies. The graph resulted from the superposition of the curves recorded for two series of spark discharges, one for each of the HV supplies.

110

Figure 4.4:

Voltage restoring curve for the: HV supply #1 with 20 ms/div (a) and HV supply #2 with 500 ms/div (b)

111

Figure 4.5:

Electrostatic field at the surface of the grounded electrode in a corona electrostatic separator.

113

Figure 4.6:

HV variation before and during granular material processing. Voltage level 30 kV; sampling frequency 20 Hz.

114

Figure 4.7:

Diagram of the Virtual Instrument employed for HV monitoring, using a HV probe and a digital electrometer. Developed in LabVIEW 6.0

115

Figure 4.8:

Plots of the predicted response σHV and of the respective 95% confidence intervals as function of metal concentration c (%), for n = 80 rev/min and U = 30 kV (a); as function of roll-speed n (rev/min), for c = 25% and U = 30 kV (b).

119

Figure 4.9:

Plots of the predicted response σHV and of the respective 95% confidence intervals: as function of metal concentration c (%), for n = 80 RPM and U = 28 kV (a); as function of applied voltage U [kV], for c = 25% and n = 80 RPM (b).

121

Figure 4.10:

Low pass, 3rd order filter used for signal enhancement

124

Figure 4.11:

Input signal and filtered signal

125

Figure 4.12:

High voltage variation for a 3 mm particle

126

Figure 4.13:

High voltage variation for a 10 mm particle

126

Figure 4.14:

Average feed rate data obtained for 10-second intervals.

128

Figure 4.15:

Measured data on an empty run (a) and on a regular run (b).

130

Figure 4.16:

Normalised coefficients of the quadratic model

131

Figure 4.17:

The linear correlation of noise and metal content. Confidence 99%.

132

Figure 5.1:

Variables of an electrostatic separation process.

138

Figure 5.2:

Graph representation of the factors under study (a) and the Taguchi’s linear graph associated to the L16 orthogonal array (b).

141

Figure 5.3:

Linear graph representation of the factors under study (a) and the Taguchi linear graph associated to the L8 orthogonal array (b).

143

Figure 5.4:

The “design factors”, voltage and roll-speed, and the “noise factors”, granule size and copper content, represented as respectively inner and outer arrays of an experimental design.

152

Figure 5.5:

Plots of the predicted output/noise ratio Y as function of U and n. The computations were carried out with MODDE, for a confidence level p = 0.95.

158

Figure 5.6:

Contour plots of the simplified response function (percentage of middling) computed with MODDE, for n = 80 min-1 (a); n = 90 min-1 (b); n = 100 min-1 (c).

160

Figure 5.7:

The arrangement of the electrostatic separation “design” and “noise” factors as respectively inner and outer arrays.

163

Figure 5.8:

Plots of the predicted output/noise ratio Y [-] as function of U [kV], n [RPM], and γ[°]. The computations were carried out with MODDE 5.0, for a confidence level p = 0.95 computed with MODDE 5.0.

168

Figure 5.9:

Contour plot of the output/noise ratio Y [-].

168

Figure 5.10:

Plots of the predicted amount of middling M [g] as function of RH [%], and c [%] , for U = 28 kV, n = 90 RPM, and γ = -6°. The computations were carried out with MODDE 5.0, for a confidence level p = 0.95.

171

Figure 5.11:

Plots of the predicted global performance index G [%] as function of RH [%], for U = 28 kV, n = 90 RPM, and γ = -6°. The computations were carried out with MODDE 5.0, for a confidence level p = 0.95.

172

Figure 5.12:

Contour plots of the global performance index G [-], computed with MODDE 5.0, as function of n [RPM] and γ [°], for U = 28 kV, c = 25 %, and RH = 40, 50, and 60%.

173

Comments