粉末处理粉尘排放料斗出口对粉尘羽流的影响【中文6204字】
收藏
资源目录
压缩包内文档预览：
编号：15025201
类型：共享资源
大小：3.51MB
格式：ZIP
上传时间：2019-02-26
上传人：闰***
认证信息
个人认证
冯**（实名认证）
河南
IP属地：河南
20
积分
- 关 键 词：
-
中文6204字
粉末处理粉尘排放料斗出口对粉尘羽流的影响【中文6204字】
- 资源描述：
-
粉末处理粉尘排放料斗出口对粉尘羽流的影响【中文6204字】,中文6204字,粉末处理粉尘排放料斗出口对粉尘羽流的影响【中文6204字】
- 内容简介：
-
粉末处理粉尘排放:料斗出口对粉尘羽流的影响Renaud Ansart , Jean-Jacques Letourneau, Alain de Ryck, John A. Dodds法国F-81013阿尔比，拉普索迪阿尔贝矿.图卢兹大学贾拉德校区法国F-81013阿尔比拉普索迪中心，贾拉德校区，CNRS文章信息文章历史:2011年1月5日2011年6月10日以修订后的形式接收，2011年6月23日接收2011年7月2日上线关键词：粉尘排放.羽.图像分析.PIV固体废物处理过程中产生的尘埃，包括散装物料的自由落体和对物料的影响，会在工业界引起许多问题，并对营办商的健康构成极大威胁。本文描述了一种描述粉体从料斗自由落体形成的尘羽的实验装置，并研究了各种出口几何形状对尘羽的影响。为此，开发了一种图像分析技术来量化尘埃羽流的特征。1. 介绍Cooper和Arnold1已经区分了自由落体柱行为的两个相反和极端的情况。第一种情况是由大量粒子组成的大块物质从料斗中落下。在这里，流中的所有粒子都以等于重力常数的速度加速。这种情况下，由于在加速过程中产生了很高的动能，碰撞时产生了很高的粉尘排放。第二种情况与研究自由落体过程中的粉尘排放更为相关，即一股非常细的粉末从料斗出口流出。这种情况在自由落体过程中产生重要的粉尘排放(图1)。研究了不同料斗出口几何形状对细颗粒物自由落体过程中粉尘排放的影响。利用PIV系统，我们可以确定下落颗粒的速度场，并获得与Liu2,3定义的夹带常数相关的羽流扩散角测量值。此外，垂直粒子速度向上通过高斯曲线拟合3，表示为：其中为中心线处的最大粒子速度，为半最大值处高斯分布的全宽度。然而，速度PIV系统的一个限制是当颗粒流过于集中时(例如靠近出口)或质量流量过大时，无法进行相关的测量。在这些羽流高度集中的区域，互相关算法无法计算出每个粒子的运动，这意味着粒子的速度会被低估。因此，我们不能使用PIV技术来估计羽流半径。为此，我们开发了一种补充PIV分析的图像分析方法，以确定即使存在高度集中的颗粒流时的羽流宽度。本文首先介绍了实验装置和不同料斗出口几何形状。然后，阐述了羽流半径估计的图像分析方法。然后将该方法应用于筒仓出口的两个镶件上，确定其对粉尘排放的影响。最后，我们解释了如何用这种图像分析方法来估计羽流中的颗粒浓度。2. 实验装置实验装置分为两部分(图2)，第一部分为将固体物料送入筒仓的气动输送机(图中未示)。筒仓可配置不同的出料口结构，可生产不同质量流量的粉体。整个装置安装在三个质量流量传感器上。固体材料从料斗中流出，然后自由落入作为测试外壳的第二部分。该箱体通过孔板水平分成两部分，该孔板分离由自由落下产生的灰尘和由于对储存堆的冲击产生的灰尘。盒子顶部有两个通风口，可以让羽流像静止的环境空气一样自由膨胀。这些通风口必须足够大以避免由诱导空气产生的压力降低，这可能导致粉末柱下降而收缩。为此，使用压力传感器测量测试箱顶部和周围环境之间的空气压力。我们使用霍尼韦尔DC系列，精度为满量程的0.25（2 500 Pa）。另外两个通风口已经放在箱子的底部，以避免颗粒撞击储料堆产生过压。这些通风口的尺寸通过压力传感器以相同的方式确定。这个盒子安装在一个可以升高或降低的框架上，以改变掉落高度。图1.落下的物料流产生的粉尘。图2.实验装置示意图。在试验过程中，观察到料斗出口质量流量的波动。这是由于料斗内置换的间隙空气流动阻力造成的。众所周知，粉末流道上的空气压力梯度会引起喷嘴外的振荡流(滴答声)4,5。这是因为空气在多孔介质6中流动受阻。这等系统尤其重要,在这里处理粒度小于100m。放电时的质量流量是恒定的宽容0.5%。图3为流动剩余波动情况。我们还可以注意到，射流从料斗出口出来后立即收缩，然后在相当长的一段时间内保持一个近似恒定的半径，然后膨胀。在本研究中，我们想评估不同的出口设备在自由落体过程中对粉尘排放的影响。在试验箱中间设置6mm的释放管。这是垂直插入位置附近的插座料斗(图4)。此外,10m多孔管直径25毫米的长度(图4)是最终释放管。通过一个连接到这个释放管的泵，我们能够在出口下方创建一个径向吸气，并测量这个吸气对羽流宽度的影响。然后，我们延长了释放管的长度，以改变多孔管在羽流中的位置。第一个结果引导我们单独研究释放管的影响。PIV系统位于设备顶部，用于拍摄烟羽。激光Nd: YAG(波长:532 nm，能量/脉冲:30 mJ，脉冲持续时间:4 ns，频率:15 Hz)产生脉冲激光片，照亮流场中的平面(图5)，记录颗粒位置。CCD相机可以获得高达30hz脉冲重复率的图像。然而，由于上面讨论的小波动，不可能确定粉末的稳定流动。因此，平均粒子图像是通过添加500个图像序列来创建的。最后,它已经表明,超过500的图像校正的平均值小于0.5%。图3.从料斗出口流出的粉末流的前30厘米照片。图4.漏斗与管。在这项研究中使用的硅胶是40到60m的筛分数。粒子密度,。相同的粉末被重复使用，在第一次和最后一次实验之间没有发现测量性能的特殊变化。如果发生摩擦，它对这里研究的空气动力特性没有可观察到的后果。 由于料斗由PMMA制成，释放管采用不锈钢，因此有必要研究这种粉末是否易于静电充电。 首先，测量粉末的静电弛豫时间：将硅胶铺在接地的钢板上，然后通过电晕充电，并测量电位下降到其初始值的36.8的时间。 硅胶的弛豫时间小于30毫秒，它只能达到初始电位的10。 因此，如果该粉末获得任何电势，则其放电将是准瞬时的。 此外，摩擦电试验证明硅胶对不锈钢和PMMA不敏感。 图5.PIV系统。3.图像分析本节提出了一种基于图像分析算法的替代方法，该算法用于颗粒流速过高时的羽流测量。算法开发与Matlab及其图像处理工具箱,目的是确定柱宽度的值用获得的图像用于PIV实验设置。每组图像对应505张频率为15Hz的图像，其中下落的粒子被激光片照射。图6给出了500张图片中的一张。相应的实验设置是:质量流率= 0.9 g s1,落差= 22厘米,管=是的,管长度= 17厘米。在这幅图中可以看到粒子流动的管道。图7为灰度图像6的像素直方图，显示了像素强度值在0-255范围内的分布。大部分的像素的图片(900000年在10001016像素)低强度和对应的背景图像。在直方图中，它们被分组为“区域I”。II区对应的像素不直接被激光束照亮，但接收到足够的衍射光和反射光，可以在图像上检测到。从区域I和区域II的像素必须被减去，并且只有那些在激光表中被考虑。我们使用一个全局阈值将图像集的每个图像转换为一系列二进制图像。强度级别首先使用Otsu的方法7来计算，但是由于100-254范围内的强度像素的总个数与强度255的像素不相关，所以我们简化了算法，使用恒定的阈值级别127。由图6生成的二值化图像如图8所示。在这张图片中,每个白像素(110110m2)对应材料在激光表的存在。通过添加相同系列的所有阈值图像构建的平均图像的示例在图9中示出。该图像已经获得用于图6中所示的管的实验。可以注意到所得到的图像（图 9）是非对称的。 这是由于在管表面和颗粒上存在激光片的反射。 即使没有管，这种非对称性也与激光器在同一侧发生，但只有高流速才能看到它，并且当增加羽流中的颗粒密度时它会增大。图6.带管的实验图。图7.图片6的直方图分布。在平均图像中，归一化到范围0,1的像素的灰度强度给出白色像素的数量与系列的图像的数量。平均图像已被用作计算表征羽流的宏观特性的基础。在x，y，强度空间中该图像的强度表面中存在高频噪声。可以使用DFT（离散傅立叶变换）或样条滤波来部分地抑制该噪声。图10示出了利用图像9的中心部分产生的该表面的示例。在该图中，可以看到管底部的凹陷。随着颗粒流变窄，该凹陷从管的下游缓慢减小。对于给定的高度，强度是水平位置的函数。我们使用的Levenberg Marquardt算法到FI吨与3参数高斯模型这个曲线示于图11.对于实验没有一个出口管，该曲线连接TS如图3所示几乎是完美的。12.对于这两个网络连接gures，所选择的高度是相应平均图像的中间高度。已经对平均图像进行了几项试验，以便在视觉上评估颗粒流羽流的宽度。最好的相应定义似乎是图11中所示的ABCD梯形的基极（CD）的宽度。梯形的点A和B是高斯曲线和线（AC）的每一侧的反射点。（BD）与曲线相切。 已经注意到，诸如图9的平均图像的非对称性对所得基底（CD）的对称性没有影响。每个实验的梯形基底的值已经用于比较具有和不具有管的不同质量流率的实验。图8.来自图6的阈值图像。 图9.阈值序列的平均图像。4. 结果与讨论通过图像分析确定了羽流的宽度。图13为无管自由下落过程中粒子高斯拟合强度分布图的演变过程。可以看出，当下降高度增大时，高斯曲线的最大值变小，梯形基底的宽度增大(图14)。因此，我们可以定义一个与扩散角接近4%的羽流斜率。这种扩散角，与羽流的膨胀行为相对应，强烈地依赖于粉末的特性，尤其是其粒径分布。对于从料斗中落下的大量颗粒，流体呈柱状。另一方面，对于从料斗出口落下的非常细的粉末，其流动应该发展成混相烟羽。还可以注意到的是，曲线下的积分整体上随高度的降低而减小。如上所述，这个表面与激光片内的粒子数量有关。所以这种下降很明显与粒子羽流的膨胀有关，它在各个方向上都是下降高度的函数。4.1.吸气的影响我们试图通过连接到泵的多孔管吸入引入羽流的环境空气来减少羽流宽度，该管设置在出口下方（图4中L = 0 cm）。第一次测量显示，由于粉末粘在管的多孔表面上，因此在该区域中产生径向抽吸是非常困难的。而且，如上所述，射流在从出口离开后立即自行收缩。由出口结合到料斗斜面之后的颗粒加速产生的较低压力在出口后几厘米处产生羽流部分的减少。我们还测试了羽流膨胀区中出口（L = 15 cm，图4）下方的多孔管组的影响。对于Qm = 3.3 kg.s-1和Qm = 0.9 kg.s-1的两种质量流量，在羽流上测试了三种不同体积流量的吸入率。羽流的宽度估计距出口22 cm处。对于不同强度的径向抽吸，测得宽度完全相同（表1）。这个结果可以通过对图像的研究来解释。平均图像9显示管周围的颗粒流具有管的精确形状。因此，我们必须研究单独的不锈钢管对羽流的影响。图10.图像9的样条曲面。4.2.管的影响为了确定释放管的影响，我们测量了释放管下方7厘米处的羽流宽度。该管设置在离射流出口15厘米处，没有多孔管。对Qm=0.9 kg的低质量流量进行的测量。s1,表明羽没有管与管厚比约11%。对于45厘米的管，图像是在距离出口50厘米处拍摄的，并且获得的强度分布在图15中示出，其显示了长管对强度分布的影响。羽流宽度由梯形的基部长度估算。管子的羽流比没有管子的羽毛窄。对于质量流量等于Qm = 0.9kg.s-1，宽度减小约为14。此外，具有管的羽流的高斯分布的强度的最大值高于没有管的羽流的强度的最大值。这可能与较高的颗粒体积分数相关。实际上，强度对应于一定量的粉末。该管本身引起空气/颗粒混合物的较低孔隙率。值得注意的是，这种减少取决于管长度。对于45cm的长管，羽流半径的减小约为14，对于15cm的管，羽流半径的减少为11。如上所述，该粉末不易于静电荷，因此该宽度减小可归因于粉末与管之间的摩擦。因此，可以认为该效果是由于粉末和管之间的摩擦。但是，必须进行进一步的实验以确认这一结果。然而，这种关于摩擦的假设似乎是由图16所证实的，图16给出了有管和不管的垂直粒子速度分布。我们注意到，该文件是高斯曲线，其中心位于最大值。管的最大速度比没有管低约12，高斯速度随管变窄。此外，我们可能会注意到管的高斯速度并不完全平滑。这个事实可以分配给在管底部后面重新形成流的方式。流中的管的存在使羽流收缩，保持高的颗粒体积分数并防止颗粒从核心脱离并变成空气传播。在自由下落期间颗粒与管的接触产生剪切，这是由于颗粒通过管上的摩擦而减慢时颗粒速度降低的原因。图11.管的强度分布。图12.无管强度分布图。图13.强度分布与下落高度的演变，Q m = 0.18 gs -1图14.曲线的羽流宽度（图13）。表1工艺参数。图16.粒子速度，Q m = 0.9 g s -1，下落高度= 50 cm。4.3 粒子衍射的光强度与粒子体积分数的相关性用高斯曲线拟合了粒子衍射后的光强分布。该强度与颗粒浓度成正比，可以写成:其中imax为高斯曲线峰值处的最大光强，lg为拐点处定义的高斯宽度。作为第一步，我们假设该光强度与羽流中的颗粒浓度成比例：i =a（p），其中a是常数参数，p是颗粒密度，是颗粒体积分数。图15.尘埃管中的管子的影响，Q m = 0.9 g s -1，下落高度= 50 cm。 在右侧图像上，显示左侧图像的两个梯形。图17.参数的变化。表2用于计算参数的过程参数。通过羽流的水平截面S的质量流率可写为：其中up是由式（1）表示的垂直粒子速度。这些方程导致重新排列质量流率：在自由落体过程中，对于不同的下落高度，图像系列采用相同的质量流率。假设参数a是常数，参数由下式定义：也应该是不变的。为了进行该计算，从粒子衍射的光的强度分布获得imax和lg。我们无法测量所有流中的粒子速度，特别是在非常高的浓度区域，所以峰值upmax的最大粒子速度是从我们所拥有的Liu 2的数值两相流模型估算出来的。以前在其他实验中验证过3。对应于表2的操作条件的图17示出了对于相同的自由下落过程，参数几乎恒定，除非下降高度太高。实际上，当流量太稀时，的值会减小。在这种情况下，羽流宽度不容易测量，因为相应的配置非常明显。对方程的各种参数进行灵敏度分析表明，该曲线的宽度lg是对表达影响最大的参数。值得注意的是，这里开发的图像分析方法用于浓缩颗粒流，但由于羽流宽度lg的变化太大，因此不易应用于稀释流动。最后，值得注意的是，常数参数随着研究过程而变化。将粒子浓度与粒子衍射的光强度连接起来的参数a的值是用于每个图像序列的光学设置的函数。因此，根据PIV获得的粒子速度，对于恒定质量流率的相机系统的校准使我们能够确定粒子浓度。我们假设参数a是恒定的，其将粒子的浓度与由一系列图像上的粒子衍射的光强度的平均值相关联。参数a是用于每个图像系列的光学设置的函数。对于略微密集且对称的粒子流，我们能够使用特定校准来获得参数。因此，我们可以获得几个高度的羽流半径以及流的每个点处的颗粒浓度。然而，对于集中流动或非对称，我们可以确定羽流的边界并估计羽流的扩散角度。该扩散角是预测羽流行为所需的唯一参数5. 结论本文提出了一种图像分析的方法，以确定下降羽状粉末的特征，即使是在颗粒流动过于集中的情况下。在这种方法中，我们为给定的实验生成了一系列阈值图像的平均值，这允许我们定义粉末羽流的宽度，并量化各种出口几何形状的影响。图像分析方法表明，连接泵的多孔管所产生的径向吸力对羽流的膨胀行为没有影响。作者认为羽流的径向愿望可能是一个有效的想法控制粉羽扩张,但使用一个版本管连接水泵和多孔管不是一个合适的方法在飞机释放管的存在扰乱了颗粒流。然而，这个释放管通过降低粒子在空气中传播的可能性，减缓了粒子的速度，并通过摩擦力使羽流集中，从而产生了另一种现象。通过使用这个管子，我们能够将羽流的宽度减少14%，自由落体50厘米。最后，该图像分析方法使得对称流能够接近尘羽稀释区颗粒浓度。在羽流密集带，可以通过对羽流半径的测量来确定羽流的演化过程。致谢作者要感谢国际自然资源研究所的Jean-Raymond Fontaine、Fabien Gerardin和Michel Pourquet对这个问题的讨论。作为ACI项目的一部分，这项研究由INRS、CNRS和FNS资助。参考文献1林志强，张志强，等。固体颗粒流对空气的影响，国立台湾师范大学学报(自然科学版)。2刘泽琴，自由落体材料中的空气夹带研究，卧龙岗大学博士，2001。王志军，王志军，王志军，等。粉末处理对粉尘排放的影响j .北京:清华大学学报(自然科学版)。4X.-l。吴克杰。马洛伊。A.汉森。M.阿米。D.毕度。眼镜滴滴答的原因。物理评论通讯71(1993)1363-1366。张志强，张志强，等。沙漏中颗粒流动的动力学。北京:清华大学学报(自然科学版)。马迪亚斯莫比乌斯，自由下落颗粒射流的聚类不稳定性，物理学报e74(5)(2006)。一种基于灰度直方图的阈值选择方法，IEEE系统事务，Man和控制论9(1)(1979)62-66。Contents lists available at ScienceDirectPowder Technologyjournal homepage: www.el sevier. com/l ocate/ powtec Powder Technology 212 (2011) 418424Dust emission by powder handling: Inuence of the hopper outlet on the dust plumeRenaud Ansart , Jean-Jacques Letourneau, Alain de Ryck, John A. DoddsUniversit de Toulouse, Mines Albi, RAPSODEE, Campus Jarlard, F-81013 Albi, France CNRS, Centre RAPSODEE, Campus Jarlard, F-81013 Albi, Francea r t i c l ei n f o Article history:Received 5 January 2011Received in revised form 10 June 2011 Accepted 23 June 2011Available online 2 July 2011Keywords: Dust emission PlumeImage analysis PIVa b s t r a c t Dust generation in solids handling involving free fall of bulk materials and impacts on a stockpile can cause many problems in industry and be a great danger for operators health. This paper describes an experimental set up to characterize the dust plume formed in free fall of powders from a hopper and investigates the inuence of various outlet geometries on the dust plume. For this purpose an image analysis technique was developed to quantify the characteristics of the dust plume. 2011 Elsevier B.V. All rights reserved.1. IntroductionCooper and Arnold 1 have distinguished two opposite and extreme situations of the behavior of free falling particle columns. The rst corresponds to when a bulk material composed of massive particles falls from a hopper.Here, all particles in the stream accelerate at a rate equal to the gravitational constant. This situation generates high dust emission at impact because of the high kinetic energy developed during acceleration. The second case, more relevant for the study of dust emission during free fall, is when a stream of a very ne powder falls from the outlet of a hopper. This situation produces an important dust emission during the free fall (Fig. 1).The aim of this present work is to study the inuence of different kinds of hopper outlet geometries on dust emission during the free fall of ne particles. By using a PIV system, we were able to determine the velocity eld of the falling particles and obtain a measurement of the spread angle of the plume which is correlated to the entrainment constant dened by Liu 2,3.In addition, the vertical particle velocity up was tted by a Gaussian curve 3, expressed as:Here, upmax is the maximum particle velocity at the center line and bg is the full width of the Gaussian distribution dened at half maximum.However, a limitation of velocity PIV systems is that pertinent measurements cannot be made when the particle ow is too concentrated, for example close to the outlet, or for high mass ow rates. In these high concentrated zones of the plume, the algorithm of cross correlation cannot calculate the motion of each particle, meaning that the particle velocities will be underestimated. Hence, we cannot use the PIV technique to estimate the plume radius. For this reason, we have developed an image analysis method, complemen- tary to the PIV analysis, to determine the plume width even when there is a highly concentrated ow of particle.First of all, this paper describes the experimental set up and the different hopper outlet geometries used. Then, the image analysis method for the estimation of the plume radius is explained. This method is then applied to two inserts tted at the outlet of the silo and the consequence on the dust emission is determined. Finally, we explain how this image analysis method can be used to estimate the particle concentration in the plume.2. Experimental setup up = upmax:exp ln2:* Corresponding author. 2r!2!bg;1The experimental set up is divided into two parts (Fig. 2). The rst comprises a pneumatic conveyor (not shown in the gure) to transfer the solid material into a silo. The silo can be tted with various outlet congurations to produce different mass ow rates of powder. The entire arrangement is mounted on three mass ow sensors.The solid material ows out of the hopper and then undergoes free fall into the second section which is the test enclosure. This box isE-mail address: ansartimft.fr (R. Ansart).divided horizontally into two parts by an orice plate which separates0032-5910/$ see front matter 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.powtec.2011.06.022R. Ansart et al. / Powder Technology 212 (2011) 418424419 HopperEntrained air Dilating material by the induced air, which could cause the falling column of powder to contract. To achieve this, a pressure sensor is used to measure the air pressure between the top of the test box and the ambient environ- ment. We use Honeywell DC series with an accuracy of 0.25% of full scale (2 500 Pa). Two other vents have been put on to the bottom part of the box to avoid an over-pressure generated by the particles impact on the stockpile. The size of these vents was determined in the same way by means of a pressure sensor. This box is mounted on a frame that can be raised or lowered to change drop height.Fugitive dust Core of particle plumeBoundary layerDuring the test, uctuations were observed in the mass ow rate from the outlet of the hopper. This is caused by the ow resistance of the displaced interstitial air in the hopper. It is known that air pressure gradients across a owing bed of powder can cause oscillatory ow (“ticking”) out of the nozzle 4,5. It occurs because the movement of air is impeded inside a porous medium 6. This is especially important for systems such as that dealt with here where the particle size is less than 100 m. The mass ow is constant during the dis- charge to a tolerance of 0.5%. Fig. 3 shows the residual uctuations in the ow. We may also note that the jet contracts immediately after stockpileexiting from the hopper outlet and thereafter maintains an approx- imately constant radius for a considerable height before expanding afterwards.For this study, we wanted to estimate the inuence of differentFig. 1. Dust generation for falling stream of material.the dust produced by free fall and that due to the impact on the stockpile. Two air vents are located on the top of the box to allow the plume to expand freely as in quiescent ambient air. It is necessary that these vents be large enough to avoid the lowering of pressure createdoutlet devices on dust emission during the free falling process. A release tube of 6 mm was set in the middle of the test bin. It was inserted vertically into position near the outlet of the hopper (Fig. 4). In addition, a porous tube of 10 m diameter with a length of 25 mm (Fig. 4) was set at the end of the release tube. By means of a pump connected to this release tube, we were able to create a radial aspiration just under the outlet and to measure the inuence of this aspiration on the width of the plume. Then, we lengthened the release tube to modify the position of the porous tube in the plume. The rst results lead us to study the inuence of the release tube alone.A PIV system is located in the top part of the equipment to take pictures of the plume. A pulsed laser sheet, generated by a laser Nd: YAG (wavelength: 532 nm, energy/pulse: 30 mJ, pulse duration: 4 ns and frequency: 15 Hz), illuminates a plane in the ow (Fig. 5) and the position of the particles are recorded. The CCD camera is able to acquire images up to 30 Hz pulse repetition rate.However, due to the small uctuations discussed above, it is impossible to determine a stationary ow of powder. Thus, a mean particle image is created by adding a sequence of 500 images. Finally, it has been shown that over 500 images the correction on the mean value is less than 0.5%.Fig. 2. Sketch of the experimental set up.Fig. 3. Photo of the rst 30 cm of a powder stream from the hopper outlet.420R. Ansart et al. / Powder Technology 212 (2011) 418424instantaneous. Moreover, triboelectrication tests have demonstrated that the silica gel is insensitive to stainless steel and PMMA.3. Image analysisFig. 4. Hopper with the tube.The silica gel used in this study is a sieve fraction of 40 to 60 m. The particle density is p = 1000 kg.m 3 and loose poured bulk density , b = 550 kg.m 3. The same powder has been reused and no particular variation in the measured properties has been noticedbetween the rst and last experiments. If attrition occurs, it has no observable consequences on the aerodynamical properties studied here.Since the hopper is made of PMMA and the release tube is in stainless steel it was necessary to study if this powder is prone to electrostatic charging. First of all, the electrostatic relaxation time of the powder was measured: the silica gel is laid on an earthed steel sheet, then charged by a corona and the time for the potential to fall to 36.8% of it initial value is measured. The relaxation time of the silica gel is less than 30 ms and it is just able to reach only 10% of the initial potential applied. Hence, if this powder acquires any electric potential, its discharge will be quasi-This section presents the alternative method based on an image analysis algorithm which is used for plume measurements when the ow rate of particles is too high for the standard PIV technique. The algorithm has been developed with Matlab and its Image Processing Toolbox, and the aim is to determine the value of the plume width using the images obtained with the experimental setup used for the PIV.Each set of images corresponds to 505 pictures taken with a frequency of 15Hz, in which the falling particles are lit by the laser sheet. Fig. 6 gives an example of one of the 500 pictures. Thecorresponding experimental set up is: mass ow rate= 0.9 g s 1,drop height= 22 cm, tube= yes, tube length= 17 cm. The tube around which the particles ow is visible in this picture.Fig. 7 is the pixel histogram of the grayscale image 6, which shows the distribution of the pixel intensity values in the range 0255. Most of the pixels of this image (900,000 among a total of 1000 1016 pixels) have a low intensity and correspond to the background of this image. In the histogram, they are group together as the Zone I. Zone II corresponds to the pixels that are not directly lit up by the laser beam but receive enough diffracted and reected light to be detected on a picture. The pixels from the zones I and II have to be subtracted, and only those which are in the laser sheet are to be taken into account. We use a global threshold to convert each picture of a image set to a series of binary images. The intensity level is rst calculated by Otsus method 7, but since the overall number of pixels of intensities in the range 100254 is not relevant compared to those of intensity 255, we simplify the algorithm and use a constant threshold level of 127. The thresholded binary image generated from picture 6 is shown in Fig. 8. In this image, each white pixel (110 110 m2) corresponds to the presence of material in the laser sheet.An example of a mean image constructed by adding all the thresholded images of a same series is presented in Fig. 9. This image has been obtained for the experiment with tube shown in Fig. 6. It can be noticed that the resulting image (Fig. 9) is non-symmetric. This is due to the presence of reects of the laser sheet on the surface of the tube and on the particles. Even without tube this non-symmetryFig. 5. PIV system.Fig. 6. Picture of an experiment with tube.Zone IZone II 2 104 pixelswith intensity=2552 x 1041.81.6Number of pixelsR. Ansart et al. / Powder Technology 212 (2011) 4184244210127255IntensityFig. 7. Histogram distribution of picture 6.Fig. 9. Mean image of the thresholded series.occurs on the same side as the laser one, but it becomes visible only for high ow rates and it enlarges when increasing the particle density in the plume.In the mean image the gray intensity of a pixel normalized to range 0,1, gives the number of white pixels versus the number of images of the series. The mean image has been used as the base to calculate macroscopic properties characterizing the plume. A high frequency noise is present in the intensity surface of this image in the x, y, intensity space. A DFT (Discrete Fourier Transform) or a spline ltering could be used to partially suppress this noise. Fig. 10 shows an example of this surface generated with the center part of the image9. In this gure, the concavity at the bottom of the tube can be seen. This concavity slowly decreases downstream from the tube as the stream of particles become narrower.For a given height, the intensity is function of horizontal posi- tion. We use a Levenberg Marquardt algorithm to t this curve with a 3-parameter Gaussian model as shown in Fig. 11. For experiments without an outlet tube, the curve ts almost perfectly as shown in Fig. 12. For these two gures, the chosen height is the middle height of the corresponding mean images. Several trials have been done on mean images to evaluate at sight what could be the width of theFig. 8. Thresholded image from picture 6.particle ow plume. The best corresponding denition seems to be the width of the base (CD) of the ABCD trapezium shown in Fig. 11. The points A and B of the trapezium are the inection points of each side of the Gaussian curve and the lines (AC) (BD) are tangent to the curve. It has been noticed that the non-symmetry of the mean image such as Fig. 9, has no inuence on the symmetry of the resulting base (CD).The values of the trapezium base for each experiment have been used to compare experiments with different mass ow rates with and without a tube.4. Results and discussionThe width of the plume has been determined by image analysis. Fig. 13 illustrates the evolution of the Gaussian tted intensity prole of the particles during free fall without tube. It may be noted that when the drop height increases the maximum of the Gaussian curve becomes less, while the width of the trapezium base increases (Fig. 14). Hence, we may dene a slope of the plume corresponding to the spread angle which is close to 4%.This spread angle, corresponding to the dilation behavior of the plume, is strongly dependent on the characteristics of the powder, in particular its particle size distribution. For a bulk material of massive particles falling from the hopper, the stream takes the shape of a column. On the other hand, for very ne powder falling from the outlet of a hopper, the ow should develop as a miscible plume.What can also be noticed is the overall decrease of the integral under the curve with the drop height. As mentioned above, this surface is linked to the quantity of particles inside the laser sheet. So this decrease is obviously correlated to the expansion of the plume of particles, which occurs in all directions as a function of the drop height.4.1. Inuence of the aspirationWe have tried to reduce the plume width by sucking in the ambient air induced into the plume by means of a porous tube connected to a pump, set just below the outlet (L =0 cm Fig. 4). The rst measurements show that it is very difcult to create a radial aspiration in this zone since the powder sticks on the porous surface of the tube. Moreover as mentioned above, the jet contracts by itself immediately after exiting from the outlet. The lower pressure created by the particle acceleration just after the outlet combine to the slope422R. Ansart et al. / Powder Technology 212 (2011) 418424Fig. 10. Splined surface of image 9.of the hopper converging generate a reduction of the plume section just on few centimeters after the outlet.We have also tested the inuence of the porous tube set further down from the outlet (L =15 cm Fig. 4) in the expansion zone of the plume. Three different volume ow rates of aspiration have been tested on the plume for two mass ow rates about Qm = 3.3 kg.s 1and Qm = 0.9 kg.s 1. The width of the plume is estimated at 22 cmfrom the outlet.The width is measured to be exactly the same for various intensities of the radial aspiration (Table 1). This result can be explained by the study of the images. The mean image 9 shows that the ow of particles around the tube takes the exact shape of the tube. Thus, we have to study the inuence of the stainless steal tube alone on the plume.4.2. Inuence of the tubeTo determine the inuence of the release tube alone, we have measured the width of the plume at 7 cm below the tube. This tube is set at 15 cm from the outlet in the jet, without the porous tube. The measurement done for a low mass ow rate equal to Qm = 0.9 kg.s 1, shows that the plume without the tube is about 11% thicker than with the tube.For a tube of 45 cm, the images have been taken at 50 cm from the outlet and the intensity prole obtained is presented in Fig. 15 showing the inuence of the long tube on the intensity prole. The plume width is estimated by the base length of the trapezium. The plume with the tube is narrower than that without the tube. The width reduction is about 14% for a mass ow rate equal toQm = 0.9 kg.s 1. Moreover, the maximum of the intensity of the Gaussian prole of the plume with the tube is higher than that without the tube. This may be correlated to a higher particle volume fraction. In fact, the intensity corresponds to a quantity of powder. The tube induces a lower porosity of the air/particle mixture behind itself.It is noteworthy that this reduction depends on the tube length. The reduction of the plume radius is about 14% for a length tube of 45 cm and 11% for a tube of 15 cm. As discussed above, this powder is not prone to electrostatic charge so this width reduction may be attributed to the friction between the powder and the tube. For thisFig. 12. Intensity prole without the tube.80intensity400-30-20-100x mm102030Fig. 11. Intensity prole with the tube.Fig. 13. Evolution of the intensity prole with the drop height, Q m = 0.18 g s 1.R. Ansart et al. / Powder Technology 212 (2011) 418424423 tube no tube320Particle velocity m/s30plume width mm280.526124222018162636465666drop height cmFig. 14. Plume width of curves (Fig. 13).-1.540302010010203040X mmFig. 16. Particle velocity, Q m = 0.9 g s 1, drop height= 50 cm.the free fall generates shear which is responsible of the reduction of the particle velocity as the particles are slowed down by the friction on the tube.Table 1Process parameters. 4.3. Correlation of the intensity of light diffracted by the particle and theQ a = 0 ml.s 1Q a = 2 ml.s 1Q a = 4 ml.s 1Q m = 0.9 g.s 1W = 27.09 mmW = 27.53 mmW= 26.82 mm 1particle volume fractionQ m = 3.2 g.sW = 32.54 mmW = 33 mmW= 32.34 mmThe intensity prole of light diffracted by the particles has beentted by a Gaussian curve for a given drop height. This intensity isreason the effect may be taken to be due to the friction between the powder and the tube. However, further experiments must be done toconrm this result. Nevertheless, this hypothesis made on the frictionproportional to the particle concentration and may be written as: !2!seems to be conrmed by Fig. 16 which presents the vertical particle velocity prole with and without the tube. We note that the prole is a Gaussian curve with a maximum at the center. This maximum of the velocity is about 12% lower with the tube than without, and thei = imaxrlexp ;g2Gaussian velocity is narrower with the tube. Moreover, we may note that the Gaussian velocity with the tube is not completely smooth. This fact may be assigned to the way the stream is re-formed behind the bottom of the tube. The presence of the tube in the stream contracts the plume, keeps a high particle volume fraction and prevents the particles from breaking away from the core and becoming airborne. The contact of the particles with the tube duringwhere imax is the maximum of the light intensity at the peak of the Gaussian curve and lg the width of the Gaussian dened at the inection points.We assumed, as a rst step, that this light intensity is proportional to the particle concentration in the plume: i = a (p), where a is a constant parameter, p is the particle density and the particle volume fraction.Fig. 15. Inuence of the tube in the dust plume, Q m = 0.9 g s 1, drop height= 50 cm. On the right image, the two trapezoids of the left images are displayed.424R. Ansart et al. / Powder Technology 212 (2011) 418424 Case 1Case 2Case 3Case 4 0.0120.01Constant b0.0080.0060.0040.002002004006008001000Drop heightFig. 17. Variations of the parameter .Table 2Process parameter for the calculation of the parameter .connects the particle concentration with the light intensity diffracted by the particle, is a function of the optical set up used for the each image series.Thus, a calibration of the camera system, for a constant mass ow rate, according to the particle velocity obtained by the PIV, enables us to determine the particle concentration. We assume that parameter a is constant which correlates the concentration of particles with the mean value of light intensity diffracted by particles on a series of images. The parameter a is a function of the optical set up used for each image series.For a ow of particles slightly dense and symmetric, we are able to use a specic calibration to obtain the parameter. Hence, we can access to the radius of the plume for several heights and the concentration of particles at each point of the stream. However, for a concentrated ow or non symmetric, we can determine the border of the plume and estimate the spread angle of the plume. This spread angle is the only parameter needed to predict the plume behavior.5. ConclusionThis paper presents an image analysis method for determining theCase 1: 60 m Qm = 0.2 g.s 1Case 2: 130 m Qm = 0.5 g.s 1characteristics of a falling plume of powder even when the ow ofCase 3: 130 m Q m = 1.9 g.s 1Case 4: mix. 60/130 m Q m = 2.2 g.s 1The mass ow rate through an horizontal section S of the plume may be written as:SQm = pupndS;3where up is the vertical particle velocity expressed by the Eq. (1).These equations lead to rearranging the mass ow rate as:particles is too concentrated. In this method, we generate a meanimage of the thresholded series of images for a given experiment, which allows us to dene a width of the powder plume and to quantify the inuence of various outlet geometries.The image analysis method has shown that the radial aspiration created by a porous tube connected to a pump has no inuence on the dilatation behavior of a plume. The authors think that a radial aspira- tion in the plume is probably a valid idea for controlling powder plume expansion, but using a release tube to connect the pump and the porous tube is not a suitable way of doing this as the presence of the release tube in the jet disturbs the partic
- 温馨提示:
1: 本站所有资源如无特殊说明，都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
2: 本站的文档不包含任何第三方提供的附件图纸等，如果需要附件，请联系上传者。文件的所有权益归上传用户所有。
3.本站RAR压缩包中若带图纸，网页内容里面会有图纸预览，若没有图纸预览就没有图纸。
4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
5. 人人文库网仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对用户上传分享的文档内容本身不做任何修改或编辑，并不能对任何下载内容负责。
6. 下载文件中如有侵权或不适当内容，请与我们联系，我们立即纠正。
7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
人人文库网所有资源均是用户自行上传分享，仅供网友学习交流，未经上传用户书面授权，请勿作他用。
川公网安备: 51019002004831号