Search

Micro metal droplet is the basic building block of three-dimensional metal parts fabricated by micro droplet deposition manufacturing (MDDM) technique. In this paper, the effect of wetting behavior between liquid metal and spray nozzle on the generation of micro aluminum droplets produced by pneumatic drop-on-demand (DOD) technique was investigated by simulation and experiment. A finite element model of liquid-gas flow was established based on the improved level set method (LSM). Then the generation of micro liquid aluminum droplets under different wetting conditions was simulated. A series of spraying experiments were also performed on micro droplet deposition experiment platform. The results show that the generation and flight of micro aluminum droplets are influenced by wetting condition between liquid metal and the nozzle surface significantly. Additionally, the effect of wetting behavior on the droplet size was analyzed to achieve the smallest building block. It was found that the droplet radius decreased with the increase of contact angle exponentially, which agreed with the numerical calculation and experiment results. On this basis, a wettability criterion was proposed for selecting nozzle materials. These works would be helpful for the processing optimization and equipment improvement of MDDM technique. [ABSTRACT FROM AUTHOR]

The mutual evolving processes of signals' waveforms and their spectra were numerically analyzed in time and frequency domains. The purpose was to research the essential relation between the signals' waveforms and their spectra. Then, the mutual transform principle was applied to analyze moiré pattern spectra, acquiring phase distribution information of the pattern. The rectangular window function was used to simulate the mutual transform between the impulse signal and direct-current waveform. Many rectangular window signals with deferent widths were obtained by changing the window width The unit impulse signal was obtained by changing the width down to zero, and the direct-current waveform obtained by changing the width up to +infinity. For smart, quick, and easy implementation of discrete Fourier transforms to rectangular pulses and obtain signals' spectra, a simple FFT system was worked out. With its calculating, the mutual evolving processes of signals' waveforms and their spectra were tracked deeply. All signals here were transformed with it. As the result, first, the spectra of rectangular window signals were in the form of sampling function [Sa(x) = sin(x)/x]. Second, with the change in the window's width, the waveform of Sa(x) changed. Third, when the width decreased, the waveform of Sa(x) extended, and vibrated more slowly. It changed into direct-current waveform when the width decreased to zero. Last, when the width increased, the waveform of Sa(x) shranked, and vibrated faster. It changed into impulse waveform when the width increased to +infinity. Signals' waveforms were in mutual transforms between the time and frequency domain. The transforming essence was considered as that the frequency component principle in Fourier series theory is reflected in the frequency domain. According to the principle of mutual transforms between signals' waveforms and their spectra, the first order spectrum of the moiré pattern was extracted out and normalized to a constant one when the moiré patterns were analyzed for acquiring their phase information. By the normalization, the moiré pattern should take on the sampling function model, which showed high contrast level. This new pattern was convenient for acquiring the phase information.

A new deflection tomography was suggested and verified with a simulated flow field. The deflection programs for projection and inverse-projection were produced based on optical principle of refraction and mathematical, physical significance of tomography. With the authors' home-made simple self-correlative algebraic reconstruction technique (SSART), the program system for the new deflection tomography was worked out and named Nonlinear Iteration Tomography Based on Deflection Spectra. A section of a complex flow field was simulated with Gauss function. The deflection spectra of a beam of parallel rays, as they passed through the field, were figured out according to the deflection program for projection, and the relative projections were acquired from the spectra. The section was reconstructed with the projections by SSART. The reconstructed result was compared with the model in order to test the deflection programs for validity. The reconstruction effect was scaled with mean-square error (MSE) and peak error (PE). As a result, with SSART, the deflection programs could be used to reconstruct the simulated field accurately. The MSE of reconstruction in this case was about 0.000 09 to 0.000 11 after 503 iterations, and PE was about 0.007 to 0.013. So, the program system, Nonlinear Iteration Tomography Based on Deflection Spectra, can accurately reconstruct complex flow fields based on deflection information.

Laser Moire Interference spectra of rocket exhausted plumes were analyzed. Projections for reconstruction were figured out, and air density distribution was reconstructed. Moire deflectograms of the plumes were produced and captured with a home-made, wide-range, high-sensitivity Moire deflectometer. Moire interference spectra were processed by phase unwrapping technique with Fourier transform, so the space phase distribution of Moire deflectograms was extracted. The background was worked out based on the margin of the Moire deflectograms, and so was the space phase distribution of the background. The phase shift distribution of distorted Moire interference spectra could easily be obtained by calculating the difference between the two space phase distributions. The relative projections could be extracted from the phase shift distribution. The air density distribution of the plumes was calculated by simple self-correlative algebraic reconstruction technique (SSART) based on deflected angles for projections. Eight projections with equal direction interval of 20 degree were used for reconstructing the air density distribution of a plume section. As a result, with phase unwrapping technique based on Fourier transform, any cross section projection could conveniently be worked out from the Moire deflectogram of rocket exhausted plumes. The air density distribution of the cross section could be reconstructed by SSART. So, SSART with deflected angle for projection is a kind of super nonlinear deflection tomography.

The present article focuses on improving algebraic reconstruction technique (ART) for optical computerized tomography (OCT). With spectroscopy, a series of traditional ARTs were analyzed in detail, then a new improved ART was developed. The new ART was named simple self-correlative algebraic reconstruction technique (SSART). With numerical simulation technique, SSART reconstructing efficiency was demonstrated by reconstructing a simulated field with an opaque object. The simulation result and reconstructing accuracy were analyzed. Many current representative ARTs were studied in order to compare the reconstructing efficiency of the new ART with that of others. These ARTs included basic algebraic reconstruction technique: ART, simultaneous algebraic reconstruction technique: SART, and modified SART: MSART. Each of those was used for reconstructing the same simulated field as that SSART was used to reconstructing. The reconstruction results under the same condition were compared with each other. Reconstruction accuracy of each ART was weighed with three error indexes, mean-square error(MSE), absolute value error (AVE), and peak error (PE). As the results, each error index of the reconstruction result with SSART declined a lot. The MSE of SSART was decreased about 26.6% compared to that of ART at the level of 10(-4) magnitude, and PE 77.6% at the level of 10(-2) magnitude. As far as the field with an opaque object is concerned, SSART improves the reconstruction result much, and increases the reconstruction accuracy a lot. SSART improves ARTs greatly.

With spectroscopy, the principle and process of sampling and reconstructing a continuous-time signal are discussed. A symmetrical frequency-finite spectrum function F (omega) is constructed with three modified rise-cosine pulses. Its corresponding time-domain signal f(t) is worked out theoretically. f(t) is sampled with a comb function deltaT (t). By modifying the value of T, Shannon sampling signal is obtained. With Fast Fourier Transform (FFT), the frequency spectrum F(d) (omega) of the sampling signal is figured out. The calculated F(d) (omega) is compared with the constructed F(omega). The processes to reconstruct f(t) with the sampling signal and its digital frequency spectrum F(d) (omega) are discussed. As the result, there is little difference between the calculated F(d) (omega) and the constructed F (omega). The original signal can accurately be reconstructed with the sampling data in time domain, so can with the frequency spectrum F(d) (omega) by FFT. As soon as signal storage is concerned, we can store the sampling data or its digital frequency spectrum.

A new improvement was brought forward in algebraic reconstruction technique (ART), named simple self-correlative algebraic reconstruction technique (SSCART), and its reconstructive result was discussed in detail. With numerical simulation technique, both ART and SSCART were numerically simulated with computer, and discussed in terms of simulated result and reconstructive accuracy with error indexes such as mean-square error (MSE), absolute value error (AVE) and peak error (PE). As a result, the accuracy of SSCART was markedly improved. Its MSE was decreased sixty-five percent compared to that of ART (the difference between MSE of ART and that of SSCART divided by the MSE of ART) at the level of 10(-4) magnitude, and PE decreased ninety-nine point nine percent at the level of 10 magnitude. The iterating program of SSCART was not a new program to ART, but only a simple factor was added to. SSCART was considered a superior reconstruction technique with many good virtues such as simple iterating program, persistent convergent iterating process, and high reconstructive accuracy. It is the most superior iterating reconstruction technique in current ARTs to our knowledge.

The field to be reconstructed was simulated with numerical simulation technique. The effect of a three-dimensional field's characteristics, especially, its frequency components, on the reconstructive accuracy was discussed with spectroscopy. A double-peaked field model was built with Gaussian function, and its frequency components were analyzed by Fourier transform. With algebraic reconstruction technique, the model field was reconstructed. The reconstructed field was analyzed with many error indexes. As a result, the reconstructed field was similar to the model one in respect of the position, pointing and shape of the peaks, but the heights of the peaks were shorter than those of corresponding ones in the model. The border of the reconstructive field showed obvious fluctuation. It was considered that the main causation of the reconstructive result was the filtering inserted in iterating process. To confirm this idea, the convergence factor, represented with A, of Gaussian function was changed from six to thirty in order to change the frequency components of the model field, and the same numerical simulation as the one for the first model field with A=18 was carried out for these new model ones consisting of different frequency. As a result, the idea proved to be right. At the same time, it was found that the characteristics of the field at the edge of the reconstructed zone had a deeper effect on the reconstructive accuracy. When the precondition that the field edge approximated to zero could no longer existed, the reconstructive accuracy declined, and furthermore, it would make iterating process divergent. From all the above, the characteristics of the field at the edge should draw much attention when we reconstruct a field with Algebraic reconstruction technique.

The effects of filters and filtering methods on the field reconstructed by algebraic reconstruction technique (ART) is discussed with spectroscopy. The field to be constructed was simulated with numerical simulation technique. The effects of many kinds of filters and filtering methods on the field reconstructed by ART were analyzed with spectroscopy. The filters used included quarter, fifth, eighth averaging filter, fifth center filter, and third row, column filter. The filtering methods included filter being inserted in the iterating process or set at its end. The averaging filter mentioned above meant that the value was acquired from the average of the four data, named quarter averaging filter, and so on. The fifth center filter meant that the value was acquired from the center, the third one in ascendant or descendant sequence of the five data. As a result, the authors found that a larger difference existed in the effects of different filters and filtering methods on the reconstructed field. First, the field reconstructed by ART contained much big noise and needed to be smoothed. Secondly, the reconstructive accuracy declined much when an averaging filter, such as quarter, fifth or eighth filter, was only set at the end of iterating process. Thirdly, the accuracy could be improved a little with a quarter or eighth averaging filter inserted in the iterating process and activated according to iterating times. Forthly, a better accuracy could be obtained with a fifth center filter inserted in the iterating process than with a fifth averaging one inserted. Fifthly, the best result in this work was obtained with a fifth averaging filter and a fifth center one inserted in the iterating process and activated alternately according to iterating times.

Aluminum (Al) plasma was induced with a pulsed Nd: YAG laser beam ablating Al target in Ar. Time-resolved information of the plasma radiation was taken with time-resolved technique, and the spectra of the radiation were recorded with an optical multi-path analyzer (OMA III ), whereupon, time-resolved spectra of the plasma radiation induced by pulsed laser were acquired. Based on the experiment data, Al resonant double lines, Al I 396.15 nm, Al I 394.40 nm, were respectively fitted with Lorentz, Gauss and their linear integrated function (abbr. Integrated function), whereupon, Lorentz and Gauss elements were separated from the experiment data profile curve. By contrasting Lorentz with Gauss curve separated, it was found that the experiment curve mainly consisted of Lorentz element, a with little Gauss. By contrasting Lorentz with Integrated fitting curve for experiment data, a visual picture of the characteristic lines broadened by Doppler effect was exhibited. According to the visual picture, the increase of full half-high width of the characteristic line broadened by Doppler effect was estimated. It was about 2 x 10(-)3 -8 x 10(-3) nm, approximating the theoretical value 6.7 x 10(-)3 nm. As a result, Doppler effect on the width of characteristic lines in the plasma could be reasonably explained by curve fitting analysis and theoretical calculation.

Aluminum plasma was obtained with a Nd:YAG pulsed laser beam ablating target aluminum. The radiation information of the plasma was recorded on a time and space resolution basis. So the time-resolved spectra of Al plasma were obtained with a time delay of 10 to 10,000 ns. From those spectra, we acquired the time-resolved spectra of Al I 396.15 and Al I 394.40 nm, the characteristic lines of Al plasma radiation. The shape of the two characteristic lines was studied from 800 to 10,000 ns delay. Further more, the two characteristic lines were respectively fitted in Lorentz profile and Gauss profile with Lorentz and Gauss function. As a result, all the characteristic lines after 1000 ns delay well fitted Lorentz profile, and before that time, they also fitted each other, but slightly. In the whole range of time delay, the experimental lines departed by far from the fitted Gauss profile. So we could conclude that the shape of Al I 396.15 and that of Al I 394.40 nm were all Lorentz lines, not Gauss ones. The half-high-full widths of the two characteristic lines were determined, aided with the Lorentz lines fitting. Compared with the natural widths of the two lines, the experimental results were much wider than the theoretical values.

The plasma was acquired with a pulsed Nd:YAG laser beam ablating aluminum. The pulse energy was set up to 145 mJ per pulse. Ar was used as surrounding gas with the pressure of 100 kPa. The radiation of the plasma was recorded with time- and space-resolved technique, so we obtained time-resolved, in 10(-9) s, spectra of the plasma radiation in front of the surface at 0.5 mm, 1.5 mm, 2.5 mm and 3.5 mm, respectively. Based on the spectra, we briefly discussed the characteristics of the plasma radiation in each zone, and studied the behavior of the continuum and aluminum radiation in detail, so the spatial profile of the plasma was obtained. As we found, at the beginning of the plasma coming into being, the radiation of the plasma was dominated by continuum radiation, but aluminum radiation became dominant later. Characteristic and ion radiation was much weaker than continuum radiation. The aluminum radiation lasted much longer than continuum and ion radiation. As a result, it was suggested that the absorption of continuum radiation of plasma by aluminum atoms should be the main mechanism for the formation of the aluminum double lines and the broad absorbtion spectra.

In order to obtain the aluminum resonant double lines, Al I 396.15 nm and Al I 394.40 nm, we have recorded the time-resolved spectra of the plasma induced by Nd:YAG laser ablating metal aluminum in Ar, under the Ar ambient pressure 10 kPa, at 0.1 mm in front of the target, with laser pulse energy 145 mJ/pulse. Relative intensities of the double lines were acquired by integrating the area under the double lines, respectively. According to the ratio of the intensities of the double lines, the ratio of transition probability of the double lines was deduced, and contrasted with the theoretical value. As the result, we found they tally with each other exactly. Limited to our experimental condition, we could only estimate approximately the ratio of transition probability of the double lines. Further experiments were needed for precise measurements of respective transition probability of the double lines.