Whenever vibration data is being collected from an electric motor, you want to fit the sensors as close to the bearing as possible, a placement where there are as few barriers or frame component joins as possible. That way, the important vibration signals don't attenuate before they get to the sensor. You can see from the pictures that one is installed on the motor cooling fan cowl. Not an ideal placement at all. You could argue that this position moves more with flexing of the thin steel cowl. Then there is the placement on the terminal box. I've never seen this before , there cannot be any meaningful data gathered here, there are at least four fitment joins and then a nice rubber gasket just underneath the terminal box lid to attenuate any signals getting to the sensor. As for the sensor on the outer frame, maybe someone had a spare sensor and stuck it there for a bit of fun?

These sensors have 3-axis accelerometers with a 1kHz frequency response, that basically means it will measure overall vibration but not pick up bearing defect frequency. They are, in my opinion, an entry level sensor.

They have their place, but it does need someone to understand sensor placement, and the data that comes from them, to be useful to a plant's condition monitoring routine.

If I had this air handling unit and these sensors. What I'd do is to fit one of them, probably close to the centre of the motor or close to the fan end. I'd look for a baseline vibration and then only monitor that. As it's a fan, I'd be looking for any increase in vibration caused by imbalance in the impeller which could then cause bearing issues. But that's just my opinion.

Like I mentioned earlier, you can fit sensors wherever you've got the space or money to fit them, the choice is up to you.

It does pay though to have someone who understands sensor selection, placement and can interpret the data to really have an optimised monitoring system in place.