Chowhan Z T, Chi L H
J Pharm Sci. 1981 Mar;70(3):247-51. doi: 10.1002/jps.2600700305.
The multicomponent mixing for cohesive powders was evaluated by multivariate statistical methods. Tests were carried out for the sampling technique, completely random state and completely segregated state. Hotelling's statistics were not helpful in testing the practical sampling technique. Comparisons of the mixing indexes based on univariate and multivariate statistics indicated excellent consistency in optimizing mixing time. Neither mixing index approached unity because cohesive powders do not reach a completely random state. The multivariate mixing index was smaller than the univariate indexes largely due to interparticular forces among small cohesive particles.
采用多元统计方法对粘性粉末的多组分混合进行了评估。对采样技术、完全随机状态和完全分离状态进行了测试。霍特林统计量在检验实际采样技术方面并无帮助。基于单变量和多变量统计的混合指数比较表明，在优化混合时间方面具有极好的一致性。由于粘性粉末不会达到完全随机状态，因此两个混合指数均未接近1。多变量混合指数小于单变量指数，这主要是由于小粘性颗粒之间的颗粒间作用力所致。
