Particle-shape monitoring and control in crystallization processes
Particle-shape monitoring and control in crystallization processes
复制标题
DOI：
10.1002/aic.690470922
复制
发表时间：
2001-09-01
期刊：
影响因子：
3.7
通讯作者：
Rawlings, JB
中科院分区：
文献类型：
--
作者：
Patience, DB;Rawlings, JB
用于固相性质在线测量的传感器的局限性限制了结晶器和其它颗粒过程的改进设计、监测和控制方法的开发和实施。图像分析技术是近年来发展起来的一种监控结晶过程中晶体尺寸和形状的流行技术。Plummer和Kausch（1995年）使用图像分析来测量聚甲醛在显微镜下的静态台上结晶时的实时晶体尺寸密度（CSD）。Mon-nier等人（1997）使用图像分析来测量己二酸在水结晶中的最终CSD。Puel等人（1997）使用图像分析来测量作为时间函数的对苯二酚晶体的两个特征长度。然而，这项技术还没有完全自动化，晶体浆料是手动取样并置于显微镜下的。此外，成像需要操作员偶尔中断自动尺寸测量并在获取更多样本之前丢弃测量结果。Braatz和Hasebe（2001）在他们的综述中讨论了最近可用的原位Lasentec粒子和视觉测量（PVM）系统。该传感器提供随机取向晶体的二维图像，然而，作者指出，图像的质量限制了成像软件自动识别单个颗粒的能力，但适用于定性故障排除。图像分析尚未被证明能够真实的实时自动监测悬浮结晶器中晶体的尺寸和形状特征。然而，图像分析在前述应用中的强大优势在于，它几乎不需要关于颗粒尺寸或形状的假设，并且不需要信号的反转和所得颗粒尺寸或形状密度的干扰。虽然图像分析是一种直接观察技术，但它是一种二维测量，如果要推断三维形状，必须小心。有限的信息，形状可以从二维二值化的图像中提取的对象给予足够的采样。Podczeck（2000）回顾了用于从二维图像评估颗粒形状的一系列方法的历史。Podczeck（1997）提出了定义新的形状因子的需要，这些形状因子易于计算和计算，
Limitations of sensors for the on-line measurement of solid~ phase properties have restricted the development and implementation of improved design, monitoring, and control methods for crystallizers and other particulate processes. Image analysis has recently become a popular technique to monitor crystal size and shape in crystallization processes. Plummer and Kausch (1995) used image analysis to measure the real-time crystal size density (CSD) of polyoxymethylene as it crystallizes on a static stage under a microscope. Mon-nier et al.(1997) use image analysis to measure the final CSDs of adipic acid in water crystallization. Puel et al.(1997) use image analysis to measure two characteristic lengths of by-droquinone crystals as a function of time. The technique is not yet fully automated, however, and the crystal slurry is manually sampled and placed under the microscope. In addition, imaging requires an operator to occasionally interrupt automatic sizing and discard measurements before taking more samples. In their review, Braatz and Hasebe (2001) dis-cuss the recently available in situ Lasentec particle and vision measurement (PVM) system. This sensor provides two-di-mensional images of crystals in random orientation, however, the authors state that the quality of the images limits the ability of imaging software to automatically identify individual particles but is suitable for qualitative troubleshooting. Image analysis has not yet been demonstrated to automati-cally monitor in real time the size and shape characteristics of crystals in a suspension crystallizer. The strong advantage of image analysis in the preceding applications, however, is that it requires almost no assumptions about particle size or shape and does not require inversion of a signal and interence of the resultant particle size or shape density. Although image analysis is a direct observation technique, it is a twodimensional measurement and care must be taken if one is to infer a three-dimensional shape. Limited information regarding shape can be extracted from two-dimensional binarized images of objects given adequate sampling. Podczeck (2000) reviews the history of the range of methods used to assess the shape of particles from a two-dimensional image. Podczeck (1997) motivates the need to de-fine new shape factors that are simple to calculate and incor