Air Gap Between the Thermowell and Temperature Sensor

4mm OD stem RTD fixed in 1/2"ID thermowell (Compressor area). RTDs getting frequent failure without any physical damage. Just i want to know is there any standard to maintain air gap b/w sensor and well.

For good measurement, one wants heat transfer to be as quick as possible. So you do not want an air gap.

The idea is to use thermally conductive compound/grease to fill the void between the sheath and the well so that there is no air gap to create thermal lag. In fact, the element should be spring-loaded to press it up against the bottom of the well for conductive contact.

But you've got a 4mm OD sheath banging around unsupported in a 13mm ID thermowell?

Have your machine shop make an adapter sleeve that slip-fits the sheath in the well. Then fit them together with some thermal grease.
The OD of the RTD must match the ID of the thermowell. Otherwise the RTD will vibrate and fail. The RTD should also be spring loaded so that it makes firm contact with the bottom of the thermowell for best thermal conductivity. Any temperature sensor vendor can sell you a RTD and head to match your thermowell.

good luck
The OD of the RTD sheath and ID of the thermowell should tightly fit. Air gap should be as minimum as possible. Compressor area, so there must be high vibration in the line. Few question will help to better understand the failure:

1. What kind of failure? RTD sheath is breaking? Wires are breaking or getting short?

2. What is the length of the RTD? RTD sheath end is touching the end of the thermowell or hanging?

3. What is the temperature range you are measuring?

4. RTD sheath is mineral insulated?

5. RTD conductor insulation - PTFE or Glass?
You've a great set of responses, but to address you question about the sensor not appearing to fail.

RTD's are highly sensitive to sustained vibration. The ASTM standards calls out a basic test of 3 G's for several hours, but thermowell suppliers report complete distruction when the sensors are "banged" about at 20+ G's. The welded connections to the extension wires experience work hardening resulting in sensor loss. Wire wound sensors also experience decalibration over time simply due to high vibration.

Start collecting sensor failure data. We can all use it.

1 G is equivalent to dropping the sensor 1 m to the floor.
In many cases where there is already or would-be minimal air gap. we use thermal paste/grease (aka heat sink compound) to reduce the air gap to zero and improve thermal transfer, thus performance.