E-Archive


Vol. 21
November Issue
Year 2020
BACK

Articles


in Vol. 21 - November Issue - Year 2020
Digitalization of the Shot Peening Process



Figure 1 3D Topography Measurement from Confocal Microscope (left) Converted to Abbott-Firestone Curve (middle) and The Use of Model to Define Coverage Percentage (Right)


Figure 2 Image Captured by Camera and Conveyer System (left) and Processed for Contour Detection (Top right) Followed by Checking of Circularity Criteria using Algorithm Built for Media Shape Check (Bottom right)





Figure 3 Experimental Setup Involving Sensorized Nozzle (left) and Experimental Intensity Values Against Pressure Sensors Data (1,2)

Introduction

Shot peening is a well-established process that has been extensively used in manufacturing for over 70 years. The process has gone through several phases of innovation, such as the introduction of Programmable Logic Controllers (PLCs) that allow control and regulation of input process parameters, the introduction of different manipulators for air blast systems like the 6-axis robotic arm for motion flexibility during the peening process and even various nozzle designs and media types to suit different peening applications. With the ongoing trend of Industry 4.0 and digitalization, which has been accelerated due to the COVID-19 pandemic, how can the shot-peening machine or process be transformed to fit into the current Industry 4.0 ecosystem?
The Data-driven Surface Enhancement (DSE) team at the Advanced Remanufacturing and Technology Centre (ARTC), Singapore, has been developing the shot peening process over the past few years and has identified two technologies from Industry 4.0 that can be applied to the current shot peening process – Automation and Sensorisation. Each technology will be explained in the subsequent sections, and it will include some of the development work conducted by ARTC in each trend for the shot-peening process. 

Automation

Automation is the use of technology such as robotics and software to execute tasks with minimum human input. Automation can help increase overall quality, reliability and efficiency of repetitive tasks which is generally affected by subjective inputs from humans, performance of humans due to fatigue or being distracted, or safety issues. This often results in an increase of overall production output with reduced costs required due to waste generated. Taking a deeper look into the shot-peening process, there are existing process control inspections like coverage and media size inspection that are currently manual and largely rely on human judgement to make key decisions. 

Coverage

Coverage is a key process control parameter in shot peening and it is important for coverage to be at least 98% as indicated in SAE J2277 to prevent premature failure at un-peened areas of the component. However, there may be instances of insufficient coverage due to slight variations of pressure and mass flow rates in shot-peening machines. Therefore, coverage inspection should be conducted after each component has been peened. 
The current method for coverage inspection relies on experienced operators to visually inspect different areas of the component with the help of a handheld magnifier or microscope. However, this method has high variance due to the subjective quantification amongst operators, especially at high coverage percentages. 
With the rapid advancement in measurement systems, the ARTC team explored the possibilities of using 3D topography to automate coverage measurements without reliance on human visual inspection. It was found that using the Abbott-Firestone curve generated from a confocal microscope could be useful in determining the coverage percentage of a peened specimen. As shot peening involves media constantly indenting onto the surface of a component, the material is pushed inwards, leaving a difference in height between a peened and an un-peened area. Since the Abbott-Firestone curve is a plot that shows the cumulative distribution of profile heights over a measured area, the cumulative height of indented areas could be easily obtained by using the coverage percentage of the specimens as a reference. 
A correlation model was built to create a relationship between coverage percentage and the normalized heights obtained from the Abbott-Firestone curve using several flat Ti-6-4 samples with various coverage percentages. The model was then validated using a separate set of samples and was able to accurately determine coverage percentage within +/-10% for higher coverage percentages (>80%). 
With this model developed, it can be applied to any Abbott-Firestone curve generated from a confocal microscope or any measurement system, and the coverage percentage can be determined with high accuracy and repeatability without any form of subjective human inspection. With this proof of concept validated using a laboratory confocal microscope, the next step for this development is to further validate the performance of the model by testing on actual production components. At the same time, a suitable quick and low-cost portable measurement system that can generate an Abbott-Firestone curve has to be identified to meet the final goal of having an automated inline coverage measurement system for the shot peening process. 

Media inspection*

According to AMS2430, media inspection is required at a minimum of every 8 hours for cast steel shots to check the shape and size condition of the media that is currently used for the peening process. The current method of checking media is to first remove a small amount of shot from the machine and bring it to a sieve shaker for a size check. After which, a separate check is conducted by taking a small amount of media to visually check under the microscope for out-of-shape media. This entire process of media inspection is not only manual, but also, the shape check under the microscope creates subjective quantification amongst operators. 
As a result, the ARTC team explored an alternative solution to automate the media inspection for both size and shape check using 2D image analysis. This was done by creating an algorithm trained to detect non-compliance media shapes using circularity, and then applied to images captured by a camera system. In addition, the diameter of each individual shots could be easily calculated automatically using a simple formula based on the 2D image captured to check against the acceptable media size range. 
With this development, media inspection could be conducted at a higher accuracy and repeatability without any subjective human inspection. In addition, this could potentially lead to continuous monitoring of media condition during the peening process with suitable hardware installed to shot peening machines, which would not only reduce media inspection frequencies during production, but could also be used as a form of machine health monitoring for the machine’s sieves and spiral separator.

*This development work was done in conjunction with A*STAR’s Institute of High Performance Computing (IHPC)

Sensorisation

Like automation, sensorisation sets aside resources for more demanding and analytical tasks by eliminating manual monitoring. Smart sensors can be used as a part of a wireless remote monitoring system that sends a prompt through a mobile application when a trigger is detected. This is done by gathering input data from the physical environment and then translating to a desired output through onboard computation. When sensors with high sensitivity and resolution are used, data collected over time can be further used for optimization of manufacturing through analytics and simulation. This can enable other technologies such as machine learning to model after key features in the process, which then could be used for predictive and preventive maintenance. In addition, sensors installed in multiple machines could also be connected via machine-to-machine (M2M) communication, as a step towards i4.0 where machines work together by sharing data.
It is widely known that in shot peening, air pressure is the most crucial parameter out of all key input parameters (air pressure, media flow rate, impingement angle and standoff distance). An analog pressure gauge is used in most legacy machines, while a single low-tier sensor is used at the air inlet of modern machines. However, neither provides detailed monitoring and insights into downstream characteristics and relating to intensity. 
Therefore, the ARTC team conducted extensive experimental trials and development towards a system capable of monitoring air pressure at various points of the shot-peening machine such as the air intake as well as the nozzle. In addition, other sensors such as accelerometer and acoustic emission sensor were employed on the nozzle in hopes to monitor media flow rate indirectly. 
Through detailed analysis, it was found that air pressure has the highest correlation to intensity out of all the shot-peening parameters and the intensity correlation of the pressure sensor at the nozzle is higher than the pressure sensor at the air intake. This is because the downstream pressure sensor at the nozzle could represent the kinetic energy of the shot stream better due to its location, where it could detect the differences in setup (hose configuration and nozzle type and size), and also, the kinetic energy is mostly accelerated at the convergent section of the nozzle. 
Furthermore, it was also found that the air intake pressure remained rather constant while the pressure sensor at the nozzle could detect fluctuations during the peening process from turbulence caused by changing media flow rate. As such, all experimental and pressure sensors data were fed into machine learning which generated a model that could be used for live monitoring of intensity. (1)
With this development, it could be clearly seen that the pressure sensor is a good example of a sensor for the shot-peening process to monitor the peening intensity in real time. There could well be other potential sensors yet to be discovered that could be used for this process. Once sensors capable of directly or indirectly correlating to other peening parameters (e.g. media flow rate) are implemented (2), the new generation shot-peening machines can make use of the information to better monitor and regulate the peening process. This will not only improve consistency of the peening process, but also reduce the amount of rejects and waste from the production of parts, and more importantly, improve the performance of components during operation. Potential applications include remote monitoring through a mobile application, standalone smart sensors for shot peening and predictive intensity peening.

Summary

Industry 4.0 envisions intelligent machines or systems that not only communicate with one another, but also analyze and diagnose issues without a need for human intervention. Amidst the COVID-19 pandemic, Industry 4.0 has become more relevant especially with the use of technologies like automation to assist in social distancing, or even digital twins for a more remote and virtual work setting.  It is clear that such technologies are here to stay, as they bring forth higher efficiencies through increased productivity and reduction of costs. Therefore, one should start implementing such technologies through baby steps, as this could not only inspire workers within, but also improve overall competitiveness of an organization.

References

(1) Teo, A., Ahluwalia, K., & Aramcharoen, A. (2020). Experimental investigation of shot peening: correlation of pressure and shot velocity to Almen intensity. The International Journal of Advanced Manufacturing Technology, 106(11), 4859-4868.
(2) Teo, A., Yicheng, J., Ahluwalia, K., & Aramcharoen, A. (2020). Sensorization of Shot Peening for Process Monitoring: Media Flow Rate Control for Surface Quality. Procedia CIRP, 87, 397-402




For Information: 
Aldrich Chua, Augustine Teo, Marcus Ang
Advanced Remanufacturing and Technology Centre (ARTC)
3 Cleantech Loop, #01-01, CleanTech Two Singapore 637143
Tel. +65.6715-6972
E-mail: marcus_ang@artc.a-star.edu.sg www.a-star.edu.sg/artc