In Part 2 of this article, I looked at the pressure-sensitive devices (safety edges) themselves. This part explores the stopping performance requirements that engineers and technologists must consider when applying these devices.
Full disclosure: I use examples from Rockwell Automation and Pepperl + Fuchs in this article. Neither firm has any relationship with me, and no financial or other considerations were offered or solicited concerning this article or any other work on this blog.
Stopping Performance
Pressure-sensitive devices need physical deflection to detect the presence of an object. For instance, the P+F PSE4 needs 12.8 mm of deflection for the initial detection of an object.
Since we are dealing with the closing speed of the guard and not the speed of the human approaching the hazard, ISO 13855 and, by extension, CSA Z432 do not provide any guidance. Instead, we need to focus on the behaviour of the guard itself.
Below is the force-path diagram from the P+F PSE4 manual [14]. Using the force-path diagram, we can determine how much force is required to deflect the edge profile enough to trigger the sensor.
We need to understand the relationship between the closing speed of the guard, the deflection of the pressure-sensitive edge, and the force with which the guard is closing. If we assume that the guard is closing at the “safe speed” of 250 mm/s, we can assume that a person might be able to avoid an impact from the guard as it closes. Calculating the reaction time, we take the detection distance from the force-path diagram and the closing speed:
Fifty-one milliseconds is exceptionally fast. The safety relay used to monitor the pressure-sensitive edge requires 32 ms to react, without considering how much time is needed to stop and reverse the guard. Here is the reaction time budget for a sample application:
| Device | Reaction Time (ms) |
|---|---|
| Total available reaction time | 51 |
| PSE2-SC-02 Safety control unit | 32 |
| SMC SY3000 solenoid valve | 16 |
| Mechanical System (cylinder and guard) | 3 |
As you can see, there are only 3 ms remaining for the mechanical system. Three milliseconds is simply not enough time for a mechanical system like a sliding guard to stop and reverse direction. Most mechanisms take between 50 and 250 ms to stop, let alone reverse direction. The mass of the moving parts, the velocity and the efficacy of the braking method chosen make a big difference in the time required.
Although the edge can deflect past the detection point, called the “cushion zone,” there is significantly more force required as the deflection increases past the detection point. The table below shows the increasing force necessary in the “cushion zone.” Referring to the table, you can see that a force of 250 N (56 lbf.) is required to create 22 mm of deflection or about three times more force than is required to reach the detection point.
| Point | Deflection (mm) | Force (N) | |
|---|---|---|---|
| A | Detection Point | 12.8 | 80 |
| B | Resistance | 22.0 | 250 |
| C | Resistance | 23.0 | 400 |
| D | Resistance | 24.0 | 600 |
Considering these factors, it is clear that it is very unlikely for the system to be able to react quickly enough to prevent any injury from the guard. Slowing the closing speed of the guard down to the point where adequate time is available for the system to detect – and react – to the presence of a person or an object is needed.
The closing speed must be set so that the control system and the mechanical systems have adequate time to react under the worst-case condition – the point at which the guard closes against an obstruction and a fixed object like the frame of the machine or an opposing door.
The “safe speed” used in our first calculation comes from the robot safety standards. This is the maximum speed a robot can move at when a human is within reach of the robot. The assumption is that this speed is slow enough for a person to recognize that they are about to be hit by the robot and move out of the way. There is significant research [18], [19], [20], [21], [22], going back to 1991, showing that even 250 mm/s is too fast for most people and that speeds in the 140-170 mm/s range are more avoidable.
Reducing the closing speed of the guard makes it easier for people to avoid being hit by the guard as it closes without making the closing time excessively long. For example, if the guards on the CNC machine in the photo above have to close a 1.5 m (60″) wide opening, each door needs to move 750 mm. Using 140 mm/s, we get a closing time of
Rounding up, five and a half seconds is fairly slow, so some increase in speed is probably okay. A decrease to 3.5 s for closing yields a closing speed of about 215 mm/s. Using the reaction time calculation,
Again rounding up, 60 ms gives the following reaction time budget.
| Device | Reaction Time (ms) |
|---|---|
| Total available reaction time | 60 |
| PSE2-SC-02 Safety control unit | 32 |
| SMC SY3000 solenoid valve | 16 |
| Mechanical System (cylinder and guard) | 12 |
Twelve milliseconds is still not enough time for a mechanical system to react. Additional measures will be required.
Additional Measures
Realistically, 100-250 ms is likely required for the mechanical system to react. As we can’t slow the guard closing speed enough to provide 300 ms of reaction time, we need to address the closing force. If the closing forces are limited by restricting the air pressure so that no more than 80-90 N (18-20 lbf.) can be created by the door closing mechanism, then based on the force-path diagram, we can see that we have sufficient pressure to trigger the device without a high probability of injuring the person. Even if the person is briefly trapped, they are unlikely to suffer a significant injury.
As you can see, there is a need to balance the guard’s operating speed with the actuator’s closing force while ensuring minimal impact on the production time of the machine. If there is a risk of someone forcing the guard open once it is closed, you may need to consider guard locking or a dual pressure closing system where low pressure is used to close the guards, and high pressure is used to hold the guard closed through the production cycle. There are many ways to design a safe and functional system.
Type-C Standards
Numerous type-C (machine-specific) standards anticipate the use of pressure-sensitive devices and will provide some guidance concerning their application. One example, EN 12978 [23], applies to garage doors and gates, providing application and testing requirements for this type of machinery. Many other examples of machine-specific standards anticipate the use of pressure-sensitive devices to reduce people’s risk.
Credit
Thanks to one of our readers, Mr. Philip G Horton, for asking the questions that inspired this article and for being patient with me while I carved out the time to write it.
References
[1] “SafeInd Custom Machine Safety Guarding – cprsafe.com.au”, cprsafe.com.au, 2018. [Online]. Available: https://www.cprsafe.com.au/products/guards/custom/. [Accessed: 23- Apr- 2018].
[10] Guardmaster Safedge Pressure Sensitive Safety Edge System Installation and User Manual 440F, 3rd ed. Milwaukee, WI: Rockwell Automation, 2015.
[11] “Safety Edges”, Pepperl+Fuchs, 2018. [Online]. Available: https://www.pepperl-fuchs.com/global/en/classid_2794.htm. [Accessed: 27- May- 2018].
[12] “How to Read Pneumatic Schematic Symbols….”, www.valmet.com, 2018. [Online]. Available: https://www.valmet.com/media/articles/up-and-running/reliability/FRFluidDwgs1/. [Accessed: 24-Aug-2022].
[13] “Solenoid Valve – STC Valve”, Stcvalve.com, 2018. [Online]. Available: https://www.stcvalve.com/Solenoid_Valve.htm. [Accessed: 30- May- 2018].
[14] Safety edge PSE4-RUB-01. Mannheim, DE: Pepperl+Fuchs GmbH, 2017.
[15] Safety control unit PSE4-SC-01. Mannheim, DE: PPepperl+Fuchs GmbH, 2017.
[16] Safety edge PSE4-SL-01. Mannheim, DE: Pepperl+Fuchs Group, 2016.
[17] Sensors for Safety Applications Product Overview. Mannheim, DE: Pepperl + Fuchs GmbH, 2017.
[18] Y. Beauchamp, T. J. Stobbe, K. Ghosh, and D. Imbeau, “Determination of a Safe Slow Robot Motion Speed Based on the Effect of Environmental Factors,” Hum. Factors J. Hum. Factors Ergon. Soc., vol. 33, no. 4, pp. 419?427, 1991.
[19] W. Karwowski, T. Plank, M. Parsaei, and M. Rahimi, “Human Perception of the Maximum Safe Speed of Robot Motions,” in Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 1987, pp. 186-190.
[20] S. Haddadin, A. Albu-Schäffer, M. Frommberger, and G. Hirzinger, “The role of the robot mass and velocity in physical human-robot interaction – Part I: Non-constrained blunt impacts,” in Proceedings – IEEE International Conference on Robotics and Automation, 2008.
[22] S. Haddadin, A. Albu-Schäffer, and G. Hirzinger, “Requirements for Safe Robots: Measurements, Analysis and New Insights,” Int. J. Rob. Res., vol. 28, no. 11-12, pp. 1507-1527, 2009.
© 2018 – 2022, Compliance inSight Consulting Inc. 
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
