Research On Quality Control Of Trimethylolpropane Production Process Based On Near Infrared Technolog

Trimethylolpropane is a ternary alcohol.Near infrared spectroscopy is an easy,fast and efficient method to analyze samples without causing damage.It can determine the content of chemical components in the sample and give its physical parameters at the same time,the analysis results are different from HPLC and GC.For online testing or field analysis,using statistical regression techniques,near-infrared spectroscopy can provide real-time chemical information for controlling chemical production processes,solvent recovery processes,mixing and extraction processes,etc.At present,there is no study on the application of near infrared spectroscopy to the quality control of trimethylolpropane synthesis.This study is eager to build the near-infrared spectrum model on the production of trimethylolpropane,so that it can better detect the production process of trimethylolpropane,and hope that this model can be applied to the actual production of the factory,so as to control the quality of trimethylolpropane.The following work has been carried out in this regard:（1）According to the detection of GC-MS in the previous stage,in addition to the formation of trimethylolpropane,there will also be the formation of 2-ethylacrolein,2-ethyl-2-hexanal,2-ethyl-1,3-hexanediol.In order to know the concentration of trimethylolpropane and its byproducts at different time,a GC method was developed.Through the investigation of repeatability,linearity and detection limit,stability,precision and recovery rate of addition,it is found that:The RSDS of trihydroxymethylpropane and its by-products were 2.42%,2.08%,2.39%and 2.40%,respectively,which were all less than 3%,indicating good reproducibility.The correlation coefficients of each component are all above 0.999,indicating that the method has good linearity.Precision and stability,RSDS of each component in the recovery rate were all less than 3%,indicating that the detection method met the requirements of methodological investigation.（2）The scanning mode of near infrared spectrum was determined as transmission analysis module,which was scanned 64 times with a resolution of 8cm-1 within the spectral range of 10000～4000cm^-1.By comparing different modeling methods,RMSEC=0.354 and Rc=0.9980 of partial least square method were finally in the optimal state.In addition,the precision,repeatability and stability of the near infrared modeling method were investigated,and the correlation coefficients were all greater than 0.99,which could meet the requirements of subsequent experiments.（3）Based on the parameters of trimethylolpropane produced by Chifeng Ruiyang Chemical Co.LTD.,the effects of reaction time,reaction temperature,n-butanal feeding rate,molar ratio of raw material and aldehyde condensation temperature on the product concentration of trimethylolpropane were re-investigated.With the mass concentration of trimethylolpropane product as the investigation standard,the response surface experiment was conducted using Design Expert 12.0 by selecting the molar ratio of raw material,the temperature of aldol condensation and the reaction time.Its regression equation is Y=-26215.67420+6.65952X₁+16846.78375X₂-9.19735X₃-1.86625X₁X₂+0.008075X₁X₃+3.36750X₂X₃-0.003706X₁²-2714.93750X₂²-0.034219X₃²,R2=0.9884.The model has high reliability.The optimal experimental parameters were obtained:reaction time:169min,molar ratio of formaldehyde and n-butanal:3.061,condensation temperature of hydroxyl aldehyde at 40℃,and the optimal TMP concentration of 12.744mg/m L.In the experiment of synthesis of trimethylolpropane,samples of different time were taken,and GC and near infrared spectrum were collected after treatment,and the experimental results were corresponding one by one.By comparing the mean square error and correlation coefficient changes of different spectral treatments and models with outliers removed,the near infrared models of trimethylolpropane and its by-products were finally established,and their R values were 0.9759,0.9412,0.9346,0.9221,respectively.RMSEC values were 0.862,0.143,0.042 and 0.0212,indicating that the four quantitative models were accurate and reliable.It can be applied to the quality control of trimethylolpropane production.（4）The best experimental parameters of the response surface method were verified by the four models established,and the predicted values of the near-infrared model were compared with the measured values of GC detection,and the deviation rate was calculated.At the same time,this model is used to predict the sample concentration of different parts of actual production,and compared with the measured value.The results show that the model established in the laboratory can well predict the concentration change of the optimal parameter experiment,and the error rate of the four substances is less than 3%.However,when comparing samples collected from different processes in the factory,except for the predicted error rate of reaction condensation liquid stage of less than 3%,the other stages are not ideal.It is speculated that the reason is that the original data used in the establishment of the model are from the condensation stage,and no data collection for the extraction,evaporation,distillation and other steps of the reaction.To achieve quality control for each stage,data need to be taken separately for modeling.