Open Ventilator Design and Production

One of the very interesting things about our response to covid19 has been to watch who responded, who felt both the need and the capability to help. Now, I don’t want to discount the admirable efforts of the formal healthcare system and its many practitioners, but I want to focus here on people that changed what they were doing and attempted to expand the capability of our society to respond to this pandemic. Seeing Dave Franchino’s post on open hardware design this past week, I’d like to express a few observations, to enumerate some of the breakdowns I saw and preview how my company Mechanomy is working to fix these issues for the next supply chain emergency.

Mid-March, I gave some thought to the improvised/open ventilator efforts and covid, looking to see if I and Mechanomy could help in a way that wasn’t nakedly marketing. I was quickly dissuaded by the disparity between the predictions of impending tragedy in mass media, necessitating significant action, and the lack of any formal engagement from healthcare providers and manufacturers. Locally, I saw a job posting as GE Healthcare stood up 2nd and 3rd shifts, but they were not seeking engineers or really attempting to change their product or design around supply chain shortfalls. Likewise there were no anesthesiologists or other practitioners explaining to the broader world what a ventilator needed to do, what features they can do without, and how they would vet or come to trust non-name-brand solutions.

(Of course it’s possible that both device manufacturers and health practitioners lacked the tools or institutional wherewithal to make a competent outreach to local communities, even the wide-scale donation of masks and protective equipment has been met by inflexible protocols[1].  It’s also possible that they’ve done incredible, agile work in responding to this challenge and just haven’t had time to tell the wider world how they’ve changed their work to respond to this unique situation.)

There were clearly public health reasons for strong messaging, but given the lack of understanding of the virus, its spread, and its effects in patients I think many stories reached for a simplified narrative: “Covid patients require ventilators and we simply don’t have enough, so STAY INSIDE and help save your neighbors.” In this sense the number of ventilators and their supply is extraneous to the objective of encouraging people to take the precautions seriously and self-isolate. But incorrectly characterizing covid as a device design and manufacturing problem motivated the maker community to ‘fix covid.’

The question that many makers, inventors, and entrepreneurs asked is: do we need to wait for our global supply chains to reopen or retool to the present need, or can we do something to lessen the impact of this pandemic? In most cases these creators had no relevant experience in healthcare, artificial ventilation, or professional product development, but they do have respect for the institution of healthcare and take it at its word: if the media says ventilators are needed, and ventilators are electro-mechanical-pneumatic machines, and I can make machines, then the only outstanding task is to get from the medical community the precise settings that they need. Voila, the ventilator shortage can be solved via off-the shelf and 3D printed parts assembled in my garage[2].

These efforts, organized through shared documents, forums, and hasty websites provide a fascinating but uncompelling look into the state of bootstrapped collective workflows. The Ambu bag efforts proceeded directly, drawing almost entirely from the participants’ existing knowledge: knowledge of how to hack together simple machines and the awareness that manual respirators exist and are readily available. The primary ignorance was on the operation of the device, what air pressures at what rate need to be supplied to a covid patient. While these pressures are displayed on many ventilators, this clinical knowledge is not easily found outside hospitals or medical schools.  That is, the healthcare institution controls access to information, dispensing it only to certain people in formal medical education and practice contexts. This approach differs significantly from the open access to information assumed and expected by ventilator #makers. There is no ‘open medicine‘ movement to complement open hardware, or open software. This absence is dearly felt, responsible for the high cost of care in normal times and the recent failure to enlist local communities in responding to covid.

From a broader social/cultural lens, the ventilator shortage appeared to many open hardware makers to offer a chance at legitimacy, to show spouses/friends/communities/academia/media that clairvoyant makers and the open source ethos can help society. The high idealism of open source has sustained many projects but it hasn’t made the true believers wealthy or otherwise shown that open developments are superior at creating societal value. I am quite sympathetic to the argument and agree that an open ventilator would have been a good win for those suffering covid and for the movement, especially when compared to the recalcitrant approach of device manufacturers. But here too the open hardware approach has been stymied, quickly assembling devices that no practitioner is willing to use.

Medtronic’s release of drawings and bills of materials for their PB560 ventilator is insufficient, appearing mostly as an exercise in placatory PR. Drawings and bills of material tell you what to build and how to assemble it. Assuming you can purchase all of the subcomponents and custom machined parts, you can assemble and program your replica ventilator. But the limiting element of the ventilator supply chain is quite likely not in final assembly but upstream in the sourcing, production, and delivery of the 3rd party subcomponents.  Global production has been severely impacted by the covid cessations, and because few of these suppliers are obviously or exclusively supplying the ventilation industry, may not have the supply chain transparency to see that their parts are essential for the covid response. (This is frankly the most important thing to know, where is the traditional ventilator supply chain limited and how can we overcome this?) Competing with supply chain ignorance, any awareness of covid-utility may be enough to prevent parts from leaving the producing country, in case that country is short of that particular element. Either way, no parts are incoming for the foreseeable future, no mater how many organizations attempt to produce a Medtronic replica.

Out of fairness to Medtronic, they simply can’t distribute their ventilator design. A product design cannot be pointed at, it is not a thing. Product designs do entail physical and digital documents, but an apocryphal 63% of the design heads home each evening. What Medtronic and the other ventilator companies could have done, were ventilators an actual problem that they desired to help, future sales notwithstanding, is (re)design a ventilator to be hardware-store producible. That is to employ their employees to design a new product that is maximally producible while still meeting the ventilation requirements. (Whether this is possible is the point of the effort.) The distinction is that a ventilator design is a series of choices made under particular requirements and certain supplier or manufacturing constraints that are eventually distilled into design documents and physical assemblies. The documents Medtronic released are useless because only Medtronic is able to use them to produce a ventilator. The design documents do not contain the series of choices and supporting context that is required to modify the ventilator design to use widely available machine parts. Now, engineers can review these documents and produce a mental model of the system function but, fundamentally, they do not know why the system was designed as it was and cannot be certain that matching the subcomponent specifications will produce an acceptable ventilator.

Mechanomy is a solution to the general problem of reusable system design, applicable here and in many other product engineering domains. From my vantage as founder, the open hardware community needs to move up the value chain to modeling the functions of ventilator systems. Modeling what a pump must do is inherently more valuable than modeling a particular pump; the first allows many pumps to be considered while the second cannot explain why it was chosen, only that it works. Especially in open developments, where team members may come and go and the development is not tied to the commercial success of a product, it is essential to have development systems that enable contributors to make the largest contribution they can. Sadly, today’s tools are insufficient to this task, costing too much to install and use while locking work product away in proprietary formats and brittle workflows. We are working to upgrade systems design and will have quite a bit more to say as our tools mature; say hi and we’ll keep you in the loop.

There is much to be commended in the open hardware response to covid19, that thousands of people felt the need and ability to help respond to covid is a great cultural statement. It is also a watershed, showing how our ossified, fragile approaches to healthcare, engineering, manufacturing, and business need to improve to be able to robustly deal with the challenges and opportunities of tomorrow.



1] Hospitals haven’t shown much ability to adapt their protocols to meet the situation, e.g. retaining and cleaning disposable masks or inverting patients for gravity draining. There have been some stories but if they have adapted they’ve done a poor job explaining it to the wider world. Obviously they’ve been busy so I look forward to reading more.

2] This criticism extends to many manufacturers who also made Ambu bag ventilators. The point of being a product manufacturer is not that you manufacture random things, it is that you make what people need and want. In this case Dyson, Virgin, and others issued press releases and videos without studying their intended customer, producing only a flash of PR.


Leave a Reply

Your email address will not be published. Required fields are marked *