Researchers use machine learning technology to develop stronger 3D printed buildings

China3D printingNet, June 24, researchers from Swinburne University of Technology and director of French construction company Bouygues Travaux Publics have used machine learning technology to better understand3D printingCompressive strength of building materials.
In order to develop a3D printingThe process of classifying geopolymer samples of the research team targeted specific variables and used machine learning methods to optimize3D printingThe composition of the material.This research can not only produce building composite materials with higher compressive strength, but also provide other materials used in the construction industry.3D printingRoadmap for classification of compound stability.

The team explained: “The purpose is to introduce a feasible method to classify geopolymer samples made by additive manufacturing technology. This research uses popular recursive partitioning functions, including rpart and ctree, to establish individual Classification model. Based on the survey results, these features are shown as3D printingThe powerful ability of geopolymers to create models. “



One of the construction sites of Bouygues Travaux Publics, France. The director of the company is a research project. Picture from Bouygues Travaux Publics.

Evaluation of additive applications in buildings

In recent years, many structures have been used in the construction process3D printing, The methodology behind these projects has developed rapidly. From the invention of the Construction Digital Fabrication (CDF) system in 1998 to the powdered “D-Shape” by Italian engineer Enrico Dini in 2007 3D printingThe technology is developing exponentially.Using cement mortar, the structural components in these enterprises were separately carried out3D printing, And summarized in each construction site in an effective way.

However, these companies also use a large amount of cement, which produces high spontaneous shrinkage, heat of hydration and costs related to building materials.As we all know, cement manufacturing will lead to higher greenhouse gas emissions, which will lead to higher energy consumption and reduce3D printingThe overall sustainability performance of the concrete structure.

On the other hand, polymers provide a quick set-up, cost-effective and environmentally friendly alternative. Compared with conventional cement composite materials, this material also has enhanced fire resistance and durability. Despite these benefits, the use of silicate compounds can be disadvantageous, not only because silicate compounds are known to cause environmental problems, but also because they are corrosive. As a result, researchers have made many efforts to substitute other elements for the silicon and aluminum atoms in the geopolymer matrix that are known to cause such harmful effects.

The research team set out to use the large amount of data generated in civil and construction engineering to learn3D printingPatterns and classification of materials, and find ways to overcome these shortcomings. Due to the complexity of the information, the team used a modern calculation method, including conditional inference trees (ctree) and recursive partitioning (rpart) methods to draw conclusions.For example, when performing geopolymer binders3D printingWhen, the number of effective factors of its strength will increase due to the printing parameters used. Given the range of independent variables, attempts to predict the compressive strength of printed geopolymer samples without the use of machine learning can produce high errors. Therefore, researchers used learning algorithms to evaluate printing variables and studied the factors that have the greatest impact on the compressive strength of the material.




A plot matrix showing the variables analyzed by the research team. Picture from Material Advances.

Using machine learning methods to classify geopolymers

During the test, a customized small3D printingMachine to produce geopolymers. The extruder with piston operation extrudes fresh geopolymer from a rectangular nozzle measuring 30 mm x 15 mm. When filling the fresh mixture, external vibration is temporarily applied to the extruder to ensure that the internal mixture is sufficiently compacted. Then, for each sample, 3 horizontal geopolymer filaments were printed in two rows, each with a size of 250 x 30 x 30 mm.

A total of 114 samples were measured and an average conversion factor of 1.95 was applied to the data set. The preliminary analysis of geopolymer formation using the ctree function confirmed the importance of slag in the design of geopolymer mixtures. The slag-based mixture design can produce higher compressive strength, while increasing the proportion of silicate to above 0.45, which can increase the strength of geopolymer materials. In addition, using the rpart function that did not use the sodium ion ratio to create a predictive model, the team was able to accurately predict the compressive strength category of 70% of the samples produced.Overall, the research team is able to3D printingTo create a classification model for geopolymers, use the ctree function to achieve a positive predictive value of up to 100%, and use the rpart function to achieve a positive predictive value of up to 81%.

Using these supervised machine learning algorithms, the research team can accurately classify and predict3D printingThe compressive strength of boron-based geopolymer concrete. Although the rpart function uses only two factors to generate these predictions, ctree uses four factors, which is reflected in the higher strength that the rpart function gives cement samples. In addition, the importance of slag percentage and boron ion ratio in the composition of geopolymers was successfully determined through the ctree and rpart functions, respectively.Therefore, the research group may become a3D printingThe important foundation of building materials has even become the starting point for the development of guidelines.

The research team concluded: “Supervised machine learning algorithms are used to3D printingThe boron-based geopolymers are classified.This study may be an excellent starting point for the development of guidelines or standards that can3D printingThe boron-based geopolymer samples are divided into several categories according to their compressive strength. “

The team used the DT flow chart of the ctree function to optimize the composition of its geopolymer cement. Picture from Material Advances.
Additive manufacturing in the construction industry

In recent years, researchers have used3D printingA series of cement mixtures have been developed, showing enhanced stability and durability, which can be used in the construction sector.For example, researchers at the University of Messina have formulated a lightweight foam concrete that can be used more effectively3D printingBuilding structure without any formwork. The new material (3DPC) has a high viscosity, so it can keep its shape in a “melted” or “fresh” state.

Armatron, a 3D architectural printing company based in Arizona, has obtained high-speed extrusion to produce reinforced concrete structures3D printingExtensive patent for the method.With this technology, the company aims to adopt sustainable full-size3D printingStructure to overcome the current limitations of conventional buildings.

A research team at Purdue University has developed a cement substitute that is resilient, torsion-resistant and crack-resistant.Through research3D printingResearchers are challenging the brittleness of the current weaknesses in the cement structure (such as uneven microstructure).

China3D printingNet compile article!

(Editor in charge: admin)