The other day, my girlfriend Katelyn pulled her Yeti mug out of the dishwasher with annoyance, hoping it hadn’t been ruined. I told her that, while a double walled plastic mug would concern me, I didn’t believe a stainless steel yeti designed to hold boiling water would have any issue. But why take my intuition as gospel when we can do entirely overkill kitchen science experiments and get real data?
(Spoiler alert: Yeti ramblers are, indeed, dishwasher safe)
One of the lab tools I keep in my back pocket is a Flir One iPhone thermal camera. There are plenty of things about it that annoy me, but having a thermal camera with decent resolution at the ready is EXTREMELY useful, and basically nobody can beat the $400 one of these runs. One of the best uses in an electronics lab is finding faults on new PCBs during board bring-up. Can’t figure out why the power rail is sagging and sinking a ton of power? Look for the hot spot(s). With 70 milikelvin sensitivity, you can even spot which microcontrollers are in sleep mode and which aren’t, all at a glance. But I digress.
If Katelyn’s Yeti had been damaged by the dishwasher, surely the external surface would get hotter, therefore transferring more heat to the surroundings, vs my un-damaged rambler of the same design. That’s something we can see directly with a thermal camera!
In this case, I used the app Vernier Thermal Analysis Plus. Unfortunately, Thermal Analysis has a tendency to crash all-the-damn-time, and strictly CANNOT load “experiments” that are larger than your phone’s available RAM. It’ll let you take a 15-minute video, but then crash on any attempt to open it afterward. It also won’t tell you what the max recording length is before hitting that limit, because of course.
But what it DOES do beats the Flir app for experiments like this: it lets you put crosshairs on multiple points in the thermal image, then plot them over time. EXACTLY what we want to do here! And luckily, it now has a time lapse mode to generate much less data, and give long experiments a prayer of exporting successfully.
Without further ado, the Yeti comparison:
The experiment here is to fill both ramblers with boiling water at ~the same time, and see what the exterior surface temperature of both looks like. A higher exterior surface temperature means higher convective heat loss to the ambient environment, which also means higher thermal transfer from the liquid contents to the external wall, since heat in=heat out looking at that external wall.
Of course, to look at the video and the trend lines, the mugs seem to be performing almost identically.
We do see something pretty interesting here though – if the vacuum flask is working right, there should be basically zero heat transfer from the interior wall to the exterior via either internal convection between the walls or conduction (since they’re not touching). The only possibility is radiation, and conduction through the joined lip at the top of the mug. If conduction is indeed the dominant transfer mode, we’d expect to see a steep thermal gradient from the hot top of the mug to a colder bottom, since towards the bottom it’ll have had a chance to cool off in the air. Indeed, we do see that, which seems to indicate the mugs are working just about as well as a double-walled vacuum flask could.
At this point, Katelyn mentioned that she thought her Hydro Flask coffee mug was also getting cold sooner than she thought it used to, since her coffee used to still be hot by the time she got to work.
Well, let’s run the same experiment on the Hydro Flasks!
I thought these results were ESPECIALLY interesting. For one, we see what we’d have expected in the first run if, indeed, the dishwasher Yeti had been damaged – higher surface temperature meaning higher thermal leakage. It sure looks like her mug, on the right/bottom and red line, is actually performing worse than my identical mug.
But we see something else, too – that sharp thermal gradient on the Yeti isn’t nearly as pronounced here. The top near the lip is definitely warmer (don’t mistake the cap, which is black plastic and leaks much more heat than the metal walls), but the wall heats up almost uniformly below that, as if the lip-conduction isn’t nearly as dominant over either internal convection (possibly true of Katelyn’s mug) or perhaps wall-to-wall radiation.
This interests me greatly because of an AvE video a while back where he dug right into a Yeti – https://www.youtube.com/watch?v=zVPLX6LY5HM. The most obvious takeaway there was that they copper plate at least one of the inner walls specifically as a barrier to radiative transfer. I wonder if maybe Hydro Flask doesn’t, and therefore suffers on performance in comparison.
Unfortunately for now, I like having coffee on my 1.5-hour way to work, so I’m going to hold off on cutting my mug in half. Sorry. Hope the rest of the post makes up for it 🙂
A quick note about experiment methodology: You might be thinking, “but what of emissivity!? All the mugs are different colors!” True. But they have the same surface finish, and they’re all painted, so I don’t suspect color plays a large role. Also, remember when I said long video in Thermal Analysis crash on re-load? This whole post was re-created after the ORIGINAL experiment, and just to be safe, I put some masking tape targets on the mugs to correct for possible variance in emissivity. Same results.
You might ALSO be thinking, couldn’t you have just put water in and measured the temperature with a thermocouple to see if one gets cold faster? Yes. But that’s much less fun:
|Time||Red Hydro Flask||Teal Hydro Flask|
|0||96.5 C||96.3 C|
|?||88.1 C||86.4 C|
|?+55m||75.6 C||72.5 C|
|?+1:52||66.7 C||62.5 C|
|?+2:48||58.7 C||54.1 C|
See? I told you. Way less fun. Also, red > teal.