Your jerry-rigged second tube may fail to convey outside air to the A/C effectively. It also may leak outside air into the room.
I expect conveyance issues to be handled well by just making the tube big. If we use a Poiseuille flow approximation, throughput scales like R^4. I was planning to use boxes with ~1ft square cross section, compared to the 6-inch diameter of the air conditioner’s hose, so my jerry-rigged second tube should have over 10X the capacity of the output hose even after some inefficiency.
(That might not work if we’re in a regime where boundary layer effects dominate, but in that regime I expect there will be plenty of airflow regardless. Also, the Darcy-Weisbach equation says R^4 is a good empirical approximation regardless.)
That still leaves turbulence in the corners as a potential issue, but I only plan to have two corners, and hopefully just making things big will also minimize corner issues.
I do plan to check for leaks by running my hand around to feel for airflow. Leaks small enough to not be caught by that method are probably small enough that they won’t impact the results much. (Note that the probable direction of a leak would not be to leak outside air into the room, but rather to leak inside air into the cardboard intake tube. It’s an intake tube, after all, so it should have slightly negative pressure compared to the room.)
It seems important to track the outside temperature to ensure that results are comparable across tests.
I do plan to measure outside temperature, though I expect it’s fine if there’s some outdoor temperature difference between the tests. The tests are looking at temperature delta relative to outdoor temperature, so I just need to have an outdoor temperature measurement taken at roughly the same time as the indoor measurements for each test condition.
I’d suggest confirming that your thermometer(s) are reliable before running the A/C experiment.
Any particular suggested calibration method? I do have at least two different thermometers to compare against each other.
My guess is calibration won’t matter much, so long as I use a single thermometer for all the measurements, so it doesn’t need to be perfect.
I’m also not sure if endpoint temperature equilibration is the right call.
Remember, the thing we really want to test is whether single-hose air conditioners yield bad outcomes for consumers. Temperature equilibrium is what I most cared about as a consumer.
I doubt the speed to equilibrate will be significantly different, but it makes sense to report it regardless.
I think it would be good to preregister either an objective method for stopping a measurement, or a maximum time limit on waiting for equilibration.
When the change over 15 minutes is less than the uncertainty in my measurements (e.g. thermometer precision or retest discrepancy, whichever is larger), I’ll call that an equilibrium.
I expect conveyance issues to be handled well by just making the tube big. If we use a Poiseuille flow approximation, throughput scales like R^4. I was planning to use boxes with ~1ft square cross section, compared to the 6-inch diameter of the air conditioner’s hose, so my jerry-rigged second tube should have over 10X the capacity of the output hose even after some inefficiency.
(That might not work if we’re in a regime where boundary layer effects dominate, but in that regime I expect there will be plenty of airflow regardless. Also, the Darcy-Weisbach equation says R^4 is a good empirical approximation regardless.)
That still leaves turbulence in the corners as a potential issue, but I only plan to have two corners, and hopefully just making things big will also minimize corner issues.
I do plan to check for leaks by running my hand around to feel for airflow. Leaks small enough to not be caught by that method are probably small enough that they won’t impact the results much. (Note that the probable direction of a leak would not be to leak outside air into the room, but rather to leak inside air into the cardboard intake tube. It’s an intake tube, after all, so it should have slightly negative pressure compared to the room.)
I do plan to measure outside temperature, though I expect it’s fine if there’s some outdoor temperature difference between the tests. The tests are looking at temperature delta relative to outdoor temperature, so I just need to have an outdoor temperature measurement taken at roughly the same time as the indoor measurements for each test condition.
Any particular suggested calibration method? I do have at least two different thermometers to compare against each other.
My guess is calibration won’t matter much, so long as I use a single thermometer for all the measurements, so it doesn’t need to be perfect.
Remember, the thing we really want to test is whether single-hose air conditioners yield bad outcomes for consumers. Temperature equilibrium is what I most cared about as a consumer.
I doubt the speed to equilibrate will be significantly different, but it makes sense to report it regardless.
When the change over 15 minutes is less than the uncertainty in my measurements (e.g. thermometer precision or retest discrepancy, whichever is larger), I’ll call that an equilibrium.