Having had success with not catching covid during the Gabriola Studio Tour this year thanks in part to our use of clean air technologies like the Far UVC box shown in the image above, Mrs Wonders and I have started thinking about making the whole house, not just the studio space, safer.
We don't live in a big place, but having worked on medical devices and environmental sensing problems for a lot of my career, as well as in more conventional medical physics, I am a strong believer in overkill when it comes to safety. Our living room is not a particularly large space, and it's adjacent to the studio and connected to it via an open archway, but I'd like it to have its own clean air system. We do relatively little entertaining, but it would be nice to do more, and one case of covid can ruin your whole life.
I also just wanted to design a "UV Corsi-Rosthenthal Box": a simple, relatively low-cost, unit that is within the scope of the average handy-person to build, requiring no more than basic wood-working, electrical wiring, and crafts, although attention to detail is vital.
The big difference between CR boxes and what I'm working on is that unlike a box fan and HEPA filters, if you screw up with UVC you can damage your eyes and skin, maybe permanently. I will in the fullness of time post open-source plans for this unit, assuming I can make it work, but I cannot caution people enough to be ultra-careful if they are doing anything with UV, and to not do it at all if they have any doubts about their own ability to do it safely.
The thing that I don't like about conventional CR boxes is that like all HEPA filters they require fairly routine maintenance--which in the case of a CR box is being dismantled and rebuilt, as the HEPA filters are the structure. While they are great stop-gap measures, they really should be replaced with more easily maintainable systems, eventually. I'd like something more permanent than that, but still home-built.
The advantage of UVC over HEPA is two-fold: one is that maintenance is much rarer, with bulb lifetimes in the ten thousand hour range, which running 8 hours a day gives a lifetime of over three years. My big SmartAir Blast recommends changing the filter every two years, and my little Levoit recommends every six months or so.
The other advantage is that it has zero pressure differential: there is no mechanical filtration, so you get the full flow rate of the fans up to the effects of head loss through the unit, which can be made quite small. In HEPA units the fans have two functions: drawing air through the unit, and forcing the air through the cleaning element. In UV units the fans just act as air collectors and distributors, with no added requirement that they be able to force the air through the filter. This means the fans can be quieter, as they are doing considerably less work.
To get a sense of this, HEPA filters often work to 50 mm of water equivalent pressure, which is the (tiny) pressure you'd experience under a depth of two inches of water. That's not a lot, but the Arctic P14 fans I favour for their ridiculously quiet operation (10.6 dB, barely more than a pin dropping!) generate just 2.4 mm of water equivalent pressure.
The SmartAir Blast delivers 950 cubic metres of clean air per hour at 43 dB, which is noticably loud if you're standing beside it. Four P14 fans should be able to deliver about 280 cubic metres of clean air per hour with a fraction of the noise, in part because this design should produce close to 100% kill for the air passing through it, and because it's ceiling-mounted that air will be taken from the upper part of the room and delivered downward to the centre of the space, acting to create displacement between the people seated around the periphery. At least that's the theory: fluid mechanical modelling will come later, I hope.
Regardless: not having to push air through a filter gains us a lot in terms of power and noise.
The only thing we lose is being inherently safe: I cannot emphasize enough that UVC can do real, lasting, harm if you're exposed to it, so ensuring that no one is ever exposed must be part of any upper-room UV system design of the kind I'm talking about here, and it is ultimately the sole responsibility of the builder and/or installer to ensure that no exposure can take place.
It turns out you can buy 36 W "Swordfish" UVC furnace fixtures from Canadian Tire for about $170, which are the core of the unit I've got in mind.
My usual design process is to think up a variety of ways of solving the problem under the known constraints, kick them around, see which one looks most promising, and pursue that one in more detail. If there is more than one apparently equally good approach, I give them all a look.
The constraints in this case are:
1) An upper-room fixture with low profile, as our ceilings are only a bit over eight feet, which is too low for any commercially available units I could find.
2) Not much more than four feet on a side, and no more than foot deep.
3) Zero sightlines between the UV tubes and anyone standing anywhere in the room, up to a few inches below the ceiling. The rule is: if something is possible, it will happen, so the design must make any kind of casual exposure impossible.
4) Mount four P14 fans to get at least 6 air changes per hour (ACH) in the space under consideration, which is about 33 cubic meters, depending on precisely how one measures it.
5) Minimum impediment to airflow, about which more below.
Mrs Wonders has a degree in architecture and worked in architecture and design for decades before starting to paint full time six or seven years ago. I'm an engineering physicist. Between us we knocked a bunch of ideas back and forth and eventually landed on pretty much what I'd first thought of: a base board of 1x6 pine that's three feet long and has a mounting bracket for the UV bulb, which lies along the length of the board in the middle. The light is only about 20 inches long overall, so this leaves some room outboard for the fans, which are mounted as follows.
The sides consist of pine 1x6 which has been cut up at an angle, as shown in this sketch, which was one of many I did in the course of a few days of thinking about it on and off:
I generally find thinking is best done in bursts: focus on the problem for a bit, play around with ideas, ideally reach a point where you think you've got a really good one, then let it sit for hours or a day and come back to it, at which point you'll often find a very obvious flaw. We have limited attention, and when we're focused on something new it's easy to exhaust it, which results in the "obvious" flaws becoming invisible. If it's not within the scope of our attention, it may was well not exist.
But having iterated through a bunch of ideas this one looked like it could fulfill the requirements. My big concerns were: could the light be seen through the fans, could the light be seen through the air intake opening, and would the power be high enough to actually kill covid, the flu, and other airborne nasties?
I'll take up the last question later, but this weekend I've spent five or six hours putting together a simple visualization of the design using VTK which is probably not the tool that a normal person would use for this--the learning curve is steep and the docs are just OK--but I've been using it for decades so it was the simplest way to go. All I wanted to do was throw together a quick visualization that would accurately represent the sightlines so I could figure out fan placement. By interacting with the visualization I could try different ideas quickly, once I'd got the basics working.
The Swordfish light is a twin-tube model and I wanted one tube on top of the other to maximize power in the horizontal direction. This puts the light in the opposite orientation of what it's designed for, but since the base dimension is square doesn't change the geometry.
Like most things, visualization development is iterative: starting with the dumbest example I managed to get something to show up on the screen using the python bindings for VTK. This required stupid linux tricks: the first time I ran the demo code it failed with a complaint about not being able to connect to the X server, which I solved by adding myself to the "video" group. Of such knowledge is expertise made.
The big thing that people who aren't involved in the creation of novel devices, ideas, art, or similar do not realize is that every step of the way is like walking through molasses. Less experienced people get frustrated by this, and may even be tempted to quit because everything, even the stupidest, most simple things, is apt to fail the first time you try it. I don't know why this is, precisely, although I have theories, mostly do to with our limited attention: in things we know how to do, learned, automatic behaviour is doing the bulk of the work. As soon as we step outside that charmed circle, we are required to pay attention to far more stuff than we can fit into our tiny little circle of mental light, so we miss things, we forget things, we trip over things, and everything breaks.
Acknowledging this, accepting it, and learning to take each failure as a guide to where we should focus our precious, limited, attention right now while never losing sight of the ultimate goal is the key to making progress. The ultimate goal in this case is to build a working upper-room UVC fixture that will be reasonably priced, not too hard, and unlikely to give me a sun tan all the way through.
It's important that we commit to the ultimate goal in a way that does not require us to pay attention to it, because attention takes effort that we can't afford if we're to make progress. The goal has to be there, all the time, in the background, as the context for what we're doing.
This requires that it be internalized in a way that it becomes part of our identity, which means if we fail--and we often will, when we're trying to build something new--we need to set things up so it doesn't do too much damage to our ego. Protecting ourselves requires both resilience and an even stronger identification with the idea of ourselves as explorers, as the kind of person who does dangerous, psychologically difficult things, and who knows that failure puts us in a very exclusive club: people who dare to step into the unknown.
It may seem silly to talk about building what is ultimately nothing but a glorified lamp-shade in these terms. It's not a moon shot, after all. And yet, if more people had the resilience to do dare to fail, even at small things, the world would be a better place.
After getting past the initial failure to launch, I incrementally added various bits and pieces: the base board, the lamps themselves, and a ring that represents the barrel of the fan, which is a bit under five inches in diameter and a bit over one inch high. I just used rough measurements until I got the basics working, because who cares if it's an accurate representation of the world when the geometry is all rong, and then measured everything up carefully and put in the correct data.
There's a few lines of code in the model that does all the heavy trigonometric lifting to set the position and orientation of the various components in the coordinate system I defined for the problem. This is typical for this kind of work. I once told a client who wanted a particular feature added to a surgical guidance system that it would only take one or two lines of code, but they would probably take me three weeks to write, which in fact they did.
With that done, it became immediately obvious that the simplest implementation of the idea would not work:
The grey slab is the 1x6 baseboard, the red tubes are the UVC light--red because it shows up well... I tried purple for verisimilitude and it was harder to see, which seemed like a bad choice for a visualization application--and the gray ring is the barrel of the fan. The view is from underneath and off to the side, and the light is clearly visible through the ring.
And sure, the fan has blades and the motor in the centre, so this is a worst case, but engineering is all about assuming the worst case. If a fan disintegrates, I still want the people brushing shards of fragmented plastic out of their hair to be safe from UVC.
My first thought was to raise the fan up, but Mrs Wonders pointed out that a cowling also work, and after a moment's reflection I saw she was right. I like to think of this as an instance of "mediocrity borrows, genius steals".
Fiddling about with cowlings sticking up above the barrel of the fan--which I initially modelled as just longer fan barrels, and which showed up a couple of bugs in the first implementation of the tricky trigonometry because nothing works the first time--it turns out a half-inch high cowling won't quite do the job:
But a one inch one will:
There then remained the question: would UVC escape the intake in a way that endangered people?
Unsurprisingly, I took a conservative approach: the only UV escaping the fixture had to be angled upwards.
To test this, I added two more elements to the visualization: a line representing a ray leaving the top of the upper tube (which resulted in my finding a typo in the tube placement that make them off by a millimeter or so because... you know the answer) and a "cube" that represented the panel that would enclose one half of the fixture. For reasons lost to history--but which probably involved a looming deadline--the "cube" source in VTK has the feature of "not all sides have to be the same length". In any other context I'm sure this would bug the hell out of me, but VTK and I go far enough back that it just seems like an understandable quirk of personality.
If the line intersected the panel--which I modeled as a 2x3 foot section of 1/8 inch ply--all would be well. If not, not.
It did not.
My initial design had assumed a three-inch opening for air intake, but that would let downward-going light out. Reducing the opening to two inches cuts off all the downward-going light and has a comfortable margin of error:
I also added a "ceiling" to the visualization to get a sense of what the gap would actually look like. The fact that the horizontal ray from the top of the lamp intersects the plywood panel demonstrates that no downward-going ray can escape.
Paint is typically not UV-reflective, although I have it in my future to buy a UVC meter (which are, sadly, not cheap) to ensure that the amount of spill from the as-built design is negligible.
At that point, it was Sunday evening--I'd started on Saturday afternoon--and all that was left to do was write this up, but during the writing--specifically the last item on the list of constraints, above--I realized that it was silly to have a cowling that ran all the way around the fan barrel when I only needed it on one side, to block the light. I went in and switched the cowling to a separate class rather than simply representing it by a longer fan barrel, and added a cutting plane to create a half-cylinder, which I then fiddled with the orientation of until it blocked the light from below but didn't block the fan intake in the "forward" direction very much:
Which eventually gave me one last idea: there's no need to make the cowling cylindrical, and two advantages to making it planar: it's easier to make by folding a bit of heavy black card around two edges of the fan and gluing it in place, and doing that will protect the fan's plastic housing from the UV light, which is always a good idea. Plastic and UV don't mix. The cowling doesn't need to be heavy or strong, it just needs to be opaque, and by wrapping it around the square fan housing it gets a lot of support from it. I'll likely make the real cowling a little higher than this one to protect the whole fan a bit better.
Although shown in the image as blue for good visualization, the cowling will be black for functional reasons: if you look up, you don't want to see anything inside the unit, and if it's black, you won't. Anything else might attract your attention, and your attention is your most limited resource. It would be rude to attract it for no good reason. Ergo: black. I also have some lying around.
The next step in this little adventure is probably to run a fluid mechanics model on the fixture using OpenFOAM, which unlike VTK I'm a complete newb at. But I want to model the airflow, but because I want to know what mixing in the room looks like, and what the typical piece of air does passing through the unit, which will allow me to make some estimate of the kill fraction.
So that's next week's exercise in things that won't work the first time, but I'll get there, eventually. I often do.
Getting there! 🙂