Design And Field Application Of Magnesium Phosphate Cement Based On Computational Optimization

Magnesium phosphate cement(MPC)is a novel type of cementitious material that forms a chemical bond between phosphate and dead burned magnesium oxide through acid-base reaction.It has the performance characteristics of fast hardening,high early strength,high bonding strength,good volume stability,and excellent durability.As a result,MPC has broad application prospects in many fields,such as repairing material in highways and airport runways,emergency repairs and construction of military engineering,solidification of nuclear waste,and 3D printing.However,existing studies have not yet proposed a unified and accurate mixture design criterion due to the complex hydration process of MPC and many influencing factors.According to the previous literature,most of them focus on the effect of a few influencing factors and single performance enhancement.However,there is a lack of research on the interaction mechanism of multiple influencing factors and the joint enhancement of multiple performances of MPC,which fundamentally limits the improvement of its overall performance.In addition,complex and diverse repair engineering put forward higher requirements for MPC.In order to solve the above issues,this study introduces a computational optimization method to establish a mathematical prediction model that scientifically and accurately,under the goal of synergistic enhancement of workability and mechanical properties.Moreover,the effect of the three factors,including magnesium-phosphorus molar ratio(M/P),water-binder mass ratio(w/b),and boraxmagnesium mass ratio(B/M),on early performance is explored,which could effectively guide the design of MPC.In addition,based on the application requirements of the repair engineering,the MPC with the excellent performance is designed by prediction model and has been applied in the facade repair of subway prefabricated segments,providing the technical guidance for the application of MPC in facade repair.Furthermore,in order to improve the prediction accuracy of the early performance of MPC,this paper explores the feasibility of adopting a back propagation(BP)neural network to establish early compressive strength and five factors prediction model.It is expected to provide a theoretical basis for the intelligent development of MPC preparation technology.The specific research content is as follows:(1)The mix design of MPC is optimized via the response surface method(RSM)and a mathematical prediction model between 3 factors(M/P,w/b,B/M)and early performance(setting time and 3h compressive strength)is established.Correlation tests are performed on the model through analysis of variance and regression coefficient calculation.The results show that the prediction model has a good degree of fit and high prediction accuracy.In addition,the w/b is the key factor that has the greatest impact on the setting time and the development of early compressive strength,which means that the w/b is the most critical factor affecting the early performance of MPC.Moreover,the M/P and w/b have a mutual coupling effect.The optimal value of M/P depends on the value of w/b.The smaller the w/b,the greater the optimal M/P.(2)The prediction model of the early performance of MPC established by RSM is used to solve the optimal mixes,according to the requirements of the rapid repairing engineering of the facade.In addition,the studies of comprehensive performance and micro-performance are carried out on the optimized mixture ratio.The results show that the optimized mix ratio has excellent mechanical properties,bonding properties,water resistance,and the compact microstructure.Besides,it initially reveals that the water resistance of MPC is closely related to the mixture ratio.(3)The optimized MPC is successfully applied to the facade repair engineering of subway prefabricated segments,and corresponding on-site repair construction measures are proposed.The repair result shows that the optimized MPC has a better bonding effect,and the standardized construction process can complete the repair more quickly and high-quality.(4)To further explore the optimization model with more influencing factors,higher prediction accuracy and high intelligence,this study establishes a nonlinear prediction model between early compressive strength and 5 factors based on BP neural network(BPNN).The model regression analysis results show that the BPNN can quickly build an implicit prediction model,and exhibit the optimal prediction result on the testing data,that is,the model has a good generalization ability.Therefore,the above results show the feasibility of the adoption of an artificial neural network(ANN)in the multi-factor mix design of MPC.