The High-Stakes Race to Build More Ventilators


As he loaded the 3D printer into the back of his Jeep, Kevin Zagorski figured he wouldn’t be back in his office for a while. It was Friday night, March 20, and Virgin Orbit, the satellite launch company that spun out of Richard Branson’s Virgin Galactic a few years back, had limited work at its Long Beach, California, operation to limit the spread of the novel coronavirus. The engineer drove home, thinking he’d spend some time experimenting with 3D-printed attachments to let his colleagues open doors without touching the handles. But that first weekend at home, his manager called with a new assignment.

Before Monday rolled around, Zagorski was working with a group of doctors to design a new low-cost, scalable ventilator to keep air in the lungs of the sickest Americans suffering from Covid-19. By mid-week, he and about a dozen fellow engineers had produced a prototype of a mechanized bag valve mask, also called an Ambu bag, the handheld device EMTs use to pump air into a patient’s lungs. Now they’re on their third generation, which they think they can start mass producing in coming days, pending regulatory approval.

Read all of our coronavirus coverage here.

In the battle against coronavirus, which can inhibit a person’s ability to breathe, ventilators are a crucial tool in short supply. Estimates range, but the University of Washington’s Institute for Health Metrics and Evaluation projects the US could need roughly 55,000 ventilators in mid-April, assuming that social distancing measures stay in place through May. On Thursday, Governor Andrew Cuomo said that New York could run out of the devices in six days. That surging demand has overwhelmed an industry that makes a few thousand of the machines a year. And so companies whose normal businesses have been sidelined—Ford, General Motors, Tesla, Dyson, and others—have rallied to fill the gap.

While smaller outfits like Virgin Orbit, vacuum maker Dyson, a group of MIT researchers, and others are designing new kinds of ventilators, America’s major manufacturers are sticking with proven technology. Ford and General Electric have licensed a design from Airon, a small Florida outfit that typically builds two or three ventilators a day. The two giants say they’ll produce 50,000 in the next 100 days, and 30,000 a month thereafter. Much of that work will happen at Ford’s Rawsonville components facility in Detroit, with 500 United Auto Workers members covering three shifts. General Motors is planning to bring in 1,200 UAW workers to build 10,000 Ventec Life Systems ventilators a month at its plant in Kokomo, Indiana.

Virgin Orbit engineers have created a new kind of ventilator by mechanizing the traditional Ambu bag, using off-the-shelf parts and some 3D printing.

Courtesy of Virgin Orbit

Those workers and plants are much needed, but the automakers’ biggest contributions are their huge breadth of manufacturing and logistical savvy. Ford’s payroll includes supply chain maestros, product designers, industrial engineers, and facilities experts, all of them trained to work on products that need thousands of parts from all over the world, with minimal room for error. “We’re able to pull together a huge amount of talent very quickly,” says Adrian Price, Ford’s head of global core engineering.

To enable a 500-fold increase in production, a Ford team spent a weekend disassembling a ventilator that Airon had shipped overnight from Florida. They 3D-scanned more than 250 parts and tore down and rebuilt sub-assemblies. They filmed each step of production to ease worker training. They mapped out how many workstations they’d need to hit the faster pace, along with the required tooling and equipment. Then they reached out to suppliers to get it all in place.

GM is doing similar work, sending a team to Seattle to see how Ventec builds its ventilators, what components it needs, and which suppliers can meet the demand. Breaking bottlenecks requires digging through the proverbial parts bin to find automotive supplies that roughly match what they now need. GM won’t use a radiator hose in lieu of a ventilator hose, says manufacturing chief Gerald Johnson, but it gave one supplier the criteria and specifications for a ventilator hose, along with a picture of the part. That company then went to work getting the tooling in place to meet the new demand. GM is doing similar things with DC motors, circuit boards, wiring bundles, and more.

Meanwhile, the automaker is hustling to convert its Kokomo plant into a ventilator factory, tearing out existing equipment, dismantling walls, and installing new workstations and conveyor lines. The automakers’ partners will help ensure the products are built to their exacting standards and properly tested before they’re sent to hospitals.

The US Food and Drug Administration will have to approve Ford’s and GM’s new setups—the agency oversees not just products, but the facilities in which they’re made. But this kind of unconventional, high-speed ramp up raises questions about quality control. “Any time you start rushing to get something out, you’re bound to have problems,” says Alan Schwartz, the executive vice president of MDI Consultants, a former FDA official who has advised device makers since the 1970s. Making medical equipment requires tracking every part and product, collecting vast amounts of data, and documenting numerous details so that any problems can be sniffed out and squashed, stat. The necessary quality control, Schwartz says, is “an enormous task.”

The automakers have the advantage of working with established suppliers on certified products—GM is technically a contractor for Ventec, in an odd inversion of how it usually works. Virgin Orbit took the tougher tack of designing a new machine. Rather than the boxy devices traditionally used in hospitals, they turned the Ambu bag into a device that doesn’t need human hands to keep human lungs going. Instead, it uses a motor to rotate a potato-shaped cam, pressing a patch of metal onto the bag. “It’s a very simple machine,” says Zagorski, who leads Virgin Orbit’s advanced manufacturing work.

Along the way, Zagorski and his colleagues took direction from the Bridge Ventilator Consortium, an ad hoc group of doctors and engineers based at UC Irvine and the University of Texas Austin. The consortium coined the term “bridge ventilator,” for a simple device with fewer capabilities than the more complex machines typically used in hospitals, but good enough for many patients with relatively mild issues, including some Covid-19 patients. “Not everyone needs the Cadillac,” says Govind Rajan, an anesthesiologist at the UC Irvine Medical Center. Ventilator shortages were common when he worked in India, he says, and he’d sometimes teach family members to use an Ambu bag to keep their loved ones breathing for a day or even two. So he says it was natural to ask for a mechanized version of the device, to bridge the gap between ventilator supply and demand.

Image may contain Wood and Plywood

“It’s a very simple machine,” says Kevin Zagorski, of Virgin Orbit’s new ventilator.

Courtesy of Virgin Orbit

Focused on scale and speed, the Virgin engineers followed “an absolutely ruthless approach to the simplicity and manufacturability of the design,” Zagorski says. “We’d go through every single part in the assembly, every nut, every washer,” making sure it was both necessary and up to the task.

That meant minimizing the amount of programming needed. The goal was to control three key variables: breaths per minute; tidal volume, which is how much air comes out of the bag; and the inspiratory-expiratory profile, which the time spent breathing in versus breathing out. The first is a function of motor speed, adjusted with a dial. The others can be adjusted by swapping out parts. To change tidal volume, a user can hook up a smaller or larger version of the piece that pushes on the bag. The shape of the rotating cam (think of a lumpy potato versus a round one) defines how much of the time the bag is being compressed, and thus the inspiratory-expiratory rate.

To put the ventilators in the field, Virgin Orbit began seeking approval from the FDA. “We are essentially ready to move into production the moment that that happens and start delivering these to hospitals,” says head of special projects Will Pomerantz. Rajan says early conversations with the regulators have gone well, and he expects approval soon. The FDA didn’t respond to questions about its talks with the company.

The Virgin Orbit engineers are well suited to this work. Their regular jobs involve designing new sorts of rocket-based satellite launch systems, and they spend much of their time spinning up new mechanical and electrical designs. They have access to clean rooms, 3D printers, and other manufacturing devices. To make this new kind of ventilator, they needed to write new instructions for metal-fabricating machines and their operators, but not much more. “They’re just building things of a slightly different shape for a slightly different purpose,” Pomerantz says.

Some question the usefulness of such a system. Standard ventilators allow for adjusting many variables, including oxygen content. “There’s a reason why these things have so many knobs and buttons,” says Barry Belmont, a biomedical engineer at the University of Michigan. Further, it’s critical to consider how a new kind of ventilator would fit into the constrained space of a crowded hospital. It can’t interfere with other medical tools, physically or electronically. It must withstand cleaning with hospital grade disinfectants, or the extreme heat and pressure of an autoclave. And health care workers must be trained to use it, at a time when they’re already pushed to their limits.

Ventilators may be mechanically simple, but they typically cost around $10,000 because they have to work nearly perfectly in a specific and demanding environment. The coronavirus sweeping the country and the planet leaves less room for error than ever, but also the need for a bigger swing.

More From WIRED on Covid-19