Pivoting to Ventilators: Q&A with Kash Behdinan, President of Pointfar Automation (Part 2)
This post is the second and final part of an interview we conducted with Kash Behdinan, president of Pointfar Automation, on his company's digital twin model for ventilator manufacturing. Pointfar is making this software available for free to companies interested in manufacturing ventilators to help fill the shortage induced by the COVID-19 pandemic. To read the first part of the interview, go here.
ISA: You've given me an overview of the digital twin platform for modeling ventilators, and we've talked about why this technology is so important. Can you share a few details about how it actually functions?
Kash Behdinan: So this is how it works. You have a real PLC, a real control system, that will run the digital model in this process. There are sub-processes that are discrete, like open and closing the valves, or turning the machine on or off. There are also some complex algorithms that give feed-forward PID control. There are core formulas for the response model, expressing the response of all the emulated elements. Finally, there's the ventilator visualization, where the 3D model is animated using high-level calculations of behavior.
Anything in nature that we face, we can model its vibration and stability and controls—a car moving, or a lung pushing air, or a pump that is pumping oil. We, as engineers, know these problems. When it comes to medical systems, we seek assistance from physicians and other engineers who know those specific models, but we can still help them out. Let's say we have a ventilator located somewhere in a Tesla factory. If it's damaged, nobody can reach out to it, but I'm an engineer and I can tunnel through my internet network to that system. Then I can start actually using control systems with a simulation to turn on and off the real valves on that machine.
Say you want to test only part of your system, or a specific device. You can't use a human, so you don't have a lung to test with. The device will be connected to your computer in your home. This is a real ventilator, built by someone, that needs to be tested. The signals that feed back and forth are able to transfer the air pressure and other electronic signals and convert them to digital. You receive that information, interpret it, and send it to the lung. You ask the lung, "Okay lung, this is how much air we're going to give you. Are you going to collapse?" This all happens on a digital platform.
So all the actuator functions can be allocated. You can test different frequencies and see the dynamic behavior. Functional testing and verification is useful for the designers as well as the physicians. Both sides can validate and verify the model, and again, it costs just about nothing to implement. It's just a simulation platform.
We are hosting this software on our servers, and we're ready to give it out to any companies that are trying to leverage our capability to do this kind of simulation. That's what we do. We're very glad to help those industries that are providing aid in this pandemic.
ISA: Have you had any manufacturers approach you about this yet?
KB: Yes, we are a preferred vendor with Tesla. When we were working on the Model 3 production line, we built almost all the robot junction boxes and control systems for their facility. Since the factory was shut down due to the recent crisis, they are now working on building ventilators. We haven't rolled out the simulation yet with them, but we are starting on the concept.
The simulation is absolutely essential for those working at GM, Ford, and other companies like this. It can even benefit medical companies, like Medtronic, in the advancement of the next versions of their ventilators. Like I said, it's a digital twin; it doesn't cost anything. Everything can be done remotely with the collaboration of engineers. We don't need to be actually on site. We don't even need real patients; we can model the human body with the help of physicians.
ISA: How does your model for a human lung work, and how does it interact with the ventilator digital twin?
KB: If you want to model a lung, you can start with principles of basic physics. Let's consider a lung as a cylinder and a piston with a force of air that's pushing it up. We'll adjust the resistance; we'll adjust damping functions. We'll include spring functions because the lung is also going to work as a spring.
By the way, I want to make sure we don't get full credit for this concept. Other researchers have modeled mechanical lung models. We're an engineering company, not a medical company. My wife, Nooshin Jahangiri, is a physician, and I referred to other research as well.
But this is the model we came up with—a piston, a spring, a damper, and adjustments for the offset of the force. Then we connect it to the ventilator to get the pressure for air, which can be adjusted with the flow and the amount of oxygen that the body needs. The control and the temperature of the inhale can be adjusted. Then we push the air into the lungs, and we do the inhale and then exhale. We have a tube and a valve so we can also control the exhale. All these things are going to be working together in this model.
If we get additional funding for this project, we can develop the model further with a group of physicians.
ISA: Thank you so much for taking the time to speak with me today, Kash. Is there anything we didn't discuss that you would like our audience to know?
KB: To summarize, as control systems engineers, we can leverage digital twins for production without needing to be highly skilled in model construction and simulation. With computer-aided engineering, simulation tools can be used to reach maximum performance, and then we use the isolation model to enhance our design. We also have an object-oriented information base that includes models of the production system, so you can scale these capabilities into different areas. All of these things can be done remotely, as well as integrated data management. Everything will be working exactly as it is on site, or in the operation system.
If you want to take a journey into the future and see where the automation industry lands, you'll see that all tests and all verification will be done in a virtual environment. I'd like to express just how important digitalization will be. We're on the edge of Industry 4.0, and that's where we've got to go.