3D打印地聚合物配合比设计及响应面优化模型

3D printing geopolymer mix design and response surface optimization model

摘要:

为了最大限度提高固废资源化,以建筑再生微粉和矿粉为前驱体,开发一种基于数字化的3D打印地聚合物,研究了液固比(L/S)、胶砂比(C/G)及水胶比(W/C)对3D打印地聚合物砂浆(3DPGM)流变-力学性能的影响机制,并基于响应面法(RSM)构建多目标优化模型。结果表明：L/S、C/G、W/C能显著影响3DPGM的可打印性和流变特性,动态屈服应力与打印条带相对宽度呈线性负相关、打印墙体相对高度呈非线性负相关；抗压、抗折及粘结强度随变量递增呈先减后增的变化规律,基准组的Y向抗压、抗折及层间粘结强度分别达到峰值41.9 MPa、5.4 MPa和2.71 MPa,Y向受载试件具有最大强度,引入各向异性系数(I_a)定量表征力学各向异性,抗折I_a>抗压I_a>粘结I_a.最后,基于方差分析建立的动态屈服应力-抗压-粘结强度均值的回归模型显著,通过满意度函数多目标优化,确定最优配比：L/S=0.482、C/G=0.675、W/C=0.344,试验与预测响应值相对误差<5%.

Abstract:

In order to maximize the recycling of solid waste, a 3D printed geopolymer based on digitization was developed using building recycled fine powder and mineral powder as precursors, The effects of liquid-solid ratio (L/S), cement-sand ratio (C/G) and water-binder ratio (W/C) on the rheological-mechanical properties of 3D printed geopolymer mortar (3DPGM) were studied, and a multi-objective optimization model was constructed based on response surface method (RSM). The results show that L/S, C/G and W/C can greatly affect the printability and rheological properties of 3DPGM. The dynamic yield stress has a linear negative correlation with the relative width of the printed strip and a nonlinear negative correlation with the relative height of the printed wall. The compressive strength, flexural strength and bond strength decrease first and then increase with the increase of variables. The Y-direction compressive strength, flexural strength and interlaminar bond strength of the reference group reach the peak values of 41.9 MPa, 5.4 MPa and 2.71 MPa, respectively. The Y-direction loaded specimen has the maximum strength. The anisotropy coefficient (I_a) is introduced to quantitatively characterize the mechanical anisotropy. The flexural I_a > compressive I_a > bond I_a. Finally, the regression model of dynamic yield stress-compression-bond strength mean based on variance analysis is significant. Through the multi-objective optimization of satisfaction function, the optimal ratio is determined: L/S=0.482, C/G=0.675, W/C=0.344, and the relative error between the test and the predicted response value is less than 5%.