Methodology Study On The Paper Structure Simulation And Property Prediction Based On Paper Fiber Property Modeling

In the pulp and papermaking field,due to the complexity of raw material(fiber)proper-ties and the correlations between different properties,the study of relationship between fiber properties and paper properties is in its infancy.Nowadays,aiming at prediction and qualita-tive analysis of paper properties,the main research method is to establish mathematical model(such as Page model)by experimental data and human experience.However,this method is kind of subjective.Although several studies have employed the computer simulation method for modeling the paper sheet structure and predicting the paper property,it is difficult to bal-ance the simulation speed and simulation accuracy.To solve the above problems,this paper concerned the tensile strength of paper sheet,which is the most important property and be noticed widely,as an example,and deeply explored the application method of machine learning,multivariate statistical analysis and Mass-Spring Model(MSM)on the pulp and papermaking field.Due to the study is a exploratory method research with highly interdisciplinary and com-plex.To simplify the research process,based on the laboratory conditions and the feasible lit-erature experiences,the following assumptions and simplifications are proposed:(1)Only the handmade papermaking process(including refining,filtering,pressing and drying)are consid-ered in the papermaking experiments,and the technological processes of filling,flow,sizing and so on are ignored temporarily.(2)Assumptions about the fiber model: the initial form of single paper fiber is a hollow circular tube? the width and wall-thickness of any cross-sections in one fiber is uniform? the material and mechanical properties of fiber are homogeneous.According to the above assumptions and simplifications,This paper has carried out the following research work:5 types of paper fiber materials with different properties are selected to measure the paper fiber properties,to make paper and to measure the tensile strength of paper sheet.Finally,317 groups of experimental data were obtained.In this process,the 5 selected fiber types include eucalyptus fiber,coniferous fiber,cotton fiber,bagasse fiber and bamboo fiber? the consid-ered fiber properties include weight-average length,width,coarseness,wall thickness,softness,elasticity modulus and fiber strength index? the beating degree is considered as papermaking process parameter.After obtaining the 317 groups of experimental data,the back propagation neural network(BPNN)is employed to build the statistical model of relationship between the paper fiber properties and the paper tensile strength,and then,3500 groups of trend data of the relationships are generated and analyzed by the BPNN model.The study result shows that: the tensile strength of paper sheet increases at first and then decreases with the increase of most fiber properties,which is basically consistent with the existing reports.Further,analyzing the 3500 groups of trend data by grey correlation analysis,the sort of influence degrees(impact factors)of different paper fiber properties on the paper tensile strength are as follows: fiber strength index(0.924)> fiber width(0.880)> fiber wall thickness(0.839)> fiber weight-average length(0.823)? fiber coarseness(0.590)? fiber flexibility(0.590).The above results about the trend and quantitative analysis results would provide important research basis and verification data for the subsequent computer simulation research.According to the experimental data,the focus of this study is to model and simulate the paper fiber,paper structure and related properties.In this research,an Improved Mass-Spring Model(IMSM)is proposed by adding the center spring to build the paper fiber model,and the IMSM can effectively avoid the collapse of fiber model cross section by using the traditional Mass-Spring Model(MSM).The parameters of IMSM are calculated by the input fiber length,width,coarseness,wall thickness,elastic modulus and material mechanics formula.Using two groups of measured data with different pulping methods(chemical pulping and mechanical pulp-ing)as inputs,simulate the bending of paper fiber model under transverse load by IMSM and FEM as comparison.The simulation results show that the average relative error(MRE)of IMSM and FEM are 6.55% and 4.57% respectively,both less than 10%.Although the simula-tion accuracy of IMSM is about 2% lower than FEM,the simulation speed of IMSM in the two groups of simulations are 116.75 times and 28.44 times faster than FEM.Thus,compared with FEM,the IMSM has higher efficiency and practicability under acceptable simulation accuracy.Based on the paper fiber modeled by IMSM,the paper structure model is built for predic-tion of paper properties.To simulate the binding force between fibers in the paper structure,the segmented oriented bounding box(OBB)collision detection method is used to calculate the con-tact area between different fibers,and the binding springs are employed to simulate the binding force(hydrogen bonding interaction)and the binding distance between different fibers,while the number of binding springs can indirectly represent the relative binding area(RBA).Then,simulate and predict the tensile strength of paper structure model by constant rate of elongation method in reference to national standard.The simulation results are as following: compared with the experiment results,the relative error of all simulation results are less than 10%? com-pared with the similar simulation process by FEM in literature,the simulation speed of IMSM is about 9.4 times faster than FEM,and the memory consumption is about 2.4 times less than FEM model.Furthermore,using the different values of target fiber property and the same values of other fiber properties to build different paper structure models,the relationship trend between the target fiber property and paper tensile strength can be simulated.Select the weight-average length of eucalyptus fiber as an example,the trend of paper tensile strength is increased at first and then decreased with the increase of fiber weight-average length,which is basically consis-tent with the relationship between the fiber weight-average length and the paper tensile strength obtained by BPNN model.At the end of this research,based on the Qt5 and OpenGL,this study develops a com-puter simulation platform by the design methods of ”three-layer structure”(presentation layer,business logic layer and data access layer)and interface oriented.This software platform con-veniently realizes the simulation function of paper fiber,paper structure and tensile strength of paper,and it is easy to use by related technologists and researchers,and easy to optimize and extend in the subsequent research.