Application of Low-field Nuclear Magnetic Resonance Combined Partial Least Square Regression in Simultaneous Prediction of Water and Oil in Oily Sludge
  
View Full Text    Download reader
DOI:
KeyWord:oily sludge  waste treatment  nuclear magnetic resonance(NMR)  water content  oil content  principal component analysis
  
AuthorInstitution
ZHENG Xiao-yuan,WANG Bo,YING Zhi,CHI Yong,JIN Yu-qi 1.上海理工大学能源与动力工程学院,上海市动力工程多相流动与传热重点实验室;2.浙江大学能源清洁利用国家重点实验室
Hits: 2218
Download times: 1059
Abstract:
      In the process of oily sludge treatment,it is important to simultaneously determine the water and oil contents rapidly and accurately,which is helpful to assess the oil recovery efficiency from the oily sludge and to optimize the process parameters on site.In this study,the calibration models of oily sludge samples were developed by the low-field nuclear magnetic resonance(LF NMR) together with the partial least square regression(PLSR) while the water and oil contents attained by Dean Stark apparatus as reference values.Both echo decay curves and T2 distribution curves were employed for the model construction to investigate the performance of the calibration model.In the end,the general calibration model with three sample sets was constructed.The results indicated that the model with echo decay curves is superior to that with T2 distribution curves.The general calibration model with echo decay curves was satisfactory with correlation coefficients(R2)of 0.965 7 and 0.978 5 for the water and oil contents,respectively.The corresponding RMSECV were 2.73% and 2.22%.The 3 batches of samples were used to verity the general model.The R2 of 0.914 1 and 0.924 7 was obtained for the water and oil contents,respcetively.The corresponding RMSEP of 1.85% and 2.04% was achieved,which was close to the RMSECV of 2.73% and 2.22%.The above results showed that the general model was steady and could meet the standards of accurate and rapid determination with the advantages of time-saving(less than 5 min),solvent-free,and non-invasion.
Close