Cyber Physical 3d Printing

This research project aims to make 3d printing more adaptable with the help of sensors feed back, 3d scanning and other sensors. We have been 3d printing on small scale from last many years but now 3d printing technology can solve the unaffordable housing issue on architectural scale. Where there is an opportunity of 3d printing buildings there are challenges also and one of them is buckling and shrinkage of 3d printed structures in the 3d printing phase. On architectural scale repeating the whole process each time manually to fix the printing errors is cost prohibitive. So to minimize the 3d printing errors we can learn from the 3d printing errors themselves and let the system intelligent enough to avoid such circumstances which can cause them.

Context Evaluation – Non-standard Concrete Structures

More than half of the total cost for non-standard concrete structure goes in Formwork Labor & Material

Evolution – AM for Concrete

With every year the demand for 3d printing concrete is increasing and the technology is getting matured.

Research Scope 

Scope of the project is to correct the printing path and robot parameters according to the deviations in printed layer in z axis due to buckling and printed layer width due to over and under extrusion. There are lot of parameters which cause the structure to buckle and shrink during 3d printing phase i.e. temperature, humidity, path planning etc. From the past research in IaaC with LDM extrusion processes is that each time the 3d printing is impacted by the mix of the material. Assuming the material mix is not 100% uniform each time then the adaptable control of the 3d printing parameters with the help of sensors feed back and computer vision can minimize the errors as well.

Linear vs Proposed Workflow – Additive Manufacturing

  

Objectives

3d Printing Test Setup – ABB

Robot used for the test is IaaC’s ABB – 120. The Pneumatic control, extrusion motor speed and robot speed are automated by imbedding the  switch commands in the g-code. Temperature and Humidity sensors are used to take environmental data during 3d printing to make a dataset.

 

 ABB 3d Printing – [Case 1]

Target height of the 3d printed model is 150mm and diameter is 40mm. Extrusion speed 25mm/s.

<p>

</p>

 ABB 3d Printing – [Case 2]

Target height of the 3d printed model is 150mm and diameter is 40mm. Extrusion speed 25mm/s.

<p>

</p>

ABB 3d Printing – [Case 3]

Target height of the 3d printed model is 300mm and diameter is 40mm. Extrusion speed 25mm/s.

<p>

</p>

3d Printing Test Setup – Kuka

Robot used for the test is IaaC’s Kuka – 150. The Pneumatic control, extrusion motor speed and robot speed are automated by imbedding the  switch commands in the g-code. Wasp Extruder and Re-filling pump is used in test.

3d Printing – Test on KUKA

Target height of the 3d printed model is 600mm and diameter is 100mm. Extrusion speed 500mm/s.

<p>

</p>

1d Scan Setup 

2d Profile Scanning of 3d printed geometry with Bosch sensor mounted on the linear slider.

1d Laser Scan – Working

This 1d laser scan method of scanning is more economical and takes less computational power as compared to the 3d scanning methods.

3d Laser Scan – Setup

In this case the the 3d printed object to be analyzed is in 360 motion and the depth camera is stationary.

Data Extraction

Geometries within 300mm to 400mm range have better accurate results. 

3d Scanning – ABB Printed Geometry

Extreme buckling in the bottom layer due to weight, Under extrusion has resulted in elastic buckling, the max deviation is 30mm

<p>

</p>

3d Scanning – Kuka Printed Geometry

The deviation noted through 3d scanning after 30 min of printing is 10-15mm from 2 sides and the rest 2 have shown no deviation.

Real-time 3d Printing – UR Setup

The robot to be used is Universal Robot 10e which has built in force torque sensors to be safe for human-machine collaboration. In this setup universal robot UR10e is used with Machina frame work.

Strategy (Sensing & Response Conditions)

3d Depth Camera is used take the depth frames and RGB frames. From RGB position on the frame is selected by call back function and with depth image the depth is extracted for real time path planning. Depth value is determining the flow rate of 3d print.

Configuration (Depth Camera)

The camera is mounted on the linear slider at the distance range of 450 – 550mm from the 3d printed geometry. The purpose to use the linear slider is to avoid the distortions and for more surface area coverage.

Test 1 (Real Time 3d Printing)

The geometry in this experiment is square column 3d printed with  real time feedback of depth sensor.

<p>

</p>

Test 2 (Real Time 3d Printing)

The geometry in this experiment is 45 degree twisted column 3d printed with  real time feedback of depth sensor.

<p>

</p>

Test 3 (Real Time 3d Printing)

The geometry in this experiment is 60 degree twisted column 3d printed with  real time feedback of depth sensor.

<p>

</p>

Test 4 (Real Time 3d Printing)

The geometry in this experiment is 70 degree twisted column 3d printed with  real time feedback of depth sensor.

<p>

</p>

Post Evaluation –  (3d Scanning)

3d Printed geometry was in fresh state during 3d printing and the post evaluation was also done by 3d scanning to compare the designed geometry and 3d printed geometry deviation. The deviation observed through this method was 6% in volume.

The deviation observed in this case of 600 mm tall and 100mm diameter column was 5mm.

Catalogue of 3d Printed Models

Note: Research is continuing…

Cyber Physical 3d Printing  is a project of IaaC, Institute for Advanced Architecture of Catalonia developed at the MRAC-02 program in 2020/21 by:

Research Student: Mansoor Awais

Tutor: Alexandre Dubor, Aldo Sollazo