from this link:
PSI-PCI - Filling Cylinders In Water - Time to Review
it overall doesn't matter (but dry is better) as long as the person doing the fills understands that it's best to slow fill: cylinders, when filled at the industry recommended fill rate of 300-600 psig/min, do not get hot
The article mentions several issues with wet filling which makes it harder to get a proper/correct fill.
Happy reading.
I read the article and there are some thing said that need to be properly qualified.
For example, if you are filling a tank directly off a compressor, the tank will indeed get hot at a 300 to 600 psi per minute fill rate. I have a 4 CFM compressor and it takes about 20 minutes to fill an AL 80 from empty to 3000 psi. That's just 150 psi per minute and the tank will get hot, because the freshly compressed air is itself hot.
As noted above the options are to use a water bath to keep the tank at close to room temperature, or to fill the tank to around 3200 psi so that it cools off to 3000 psi at room temperature.
If you crunch the numbers you'll find that a tank filled to 3200 psi at a temperature of 106 degrees F will cool to 2996 psi at 70 degree F. That means the tank is not in fact over filled. But some shops will insist that doing the initial fill to 3200 psi with the tank heating to 106 degrees F is "overfilling". Those same shops will slow fill a tank to 3000 psi at 70 degrees F and then insist it is not overfilled when it is sitting in the sun on the boat deck at 106 degrees F.
PSI suggests that a water bath is not effective unless it's an ice bath, and then they insist that this will result in over filling. If the tank really does chill all the way down to 32 degrees F, then yes a 3000 psi fill will be about 3230 psi at 70 degrees. But think about this for a minute. What drives cooling efficiency is the temperature gradient - the higher the difference in temperatures, the better the heat transfer. If you are filling a 106 degree tank in 70 degree water, PSI suggests that 36 degree temperature gradient is not sufficient. Yet they insist the 38 degree temperature gradient between 70 degrees and 32 degrees is enough to cause a over fill. It's basically the same gradient so their logic fails.
In my experience, if I use a water bath on a 70 degree day, I can reduce my initial fill pressure to about 100 psi over the target pressure and be pretty much on the money once the tank cools to 70 degrees. If I am filling tanks on a 95 degree afternoon, a fresh water bath with 70 degree water is a real help as it means I can fill to about 100 psi over the target pressure instead of around 300-400 psi over the target pressure to get the proper service pressure at 3000 psi. In that context, a water bath on a hot day eases the strain on both my compressor and my tanks.
Similarly, on a 95 degree day in N FL, if I want a 3600 psi fill once I'm actually in the 70 degree water in a cave system, I do in fact need to fill the tank to 3800 psi (3771 psi to be exact). That's assuming the tank is actually at the ambient temp of 95 degrees during the fill. That won't happen unless I'm both slow filling and using a water bath (where the water will be close to the ambient air temp) to improve the heat transfer. I not, I am once again needing a 3900-4000 psi initial fill pressure to get a 3600 psi fill once I'm in 70 degree water - and that is really pushing the test pressure of the tank. In that regard a water bath makes a great deal of sense and that's why you see almost every diver shop in N FL using a water bath.
Obviously I am a fan of cave fills. PSI is correct to the extent that DOT considers a tank to be over filled if it exceeds the service pressure when it is at 70 degrees F. Yet every dive shop in cave country has been doing 3600 psi cave fills in 2400 psi service pressure 3AA steel tanks for at least the last 2 decades and DOT has not shut any of them down. Why? Partly it's a jurisdictional issue as DOT's jurisdiction applies to tanks used in interstate commerce. DOT inspectors can try to insist that a scuba tank filled in N FL can potentially be taken across a state line, and thus be used in interstate commerce, but it's weak argument when divers are getting fills that morning for a dive in N FL that same day. Part of the reason DOT ignores the practice however is the fact that 2400 psi steel tanks that are cave filled on a regular basis to 3600 psi are not exploding, and in fact are not even failing requalification every 5 years.
PSI also cites concerns for water getting in a tank due to being filled in a water bath. However, provided the tank monkey cracks the tank valve to ensure there is no water lurking in the valve, and likewise ensures there is no water in the fill whip, no water will get in the tank. Water just does not seep in against greater tank pressure. The same threat exists on boats, where spray can get in a valve or in the fill whip just before the connection is made and the tank is filled, An even greater threat is posed by compressors that do not have adequate moisture separation and/or inadequate desiccant in the filter stack. If there is excessive moisture, it will condense in the tank once it cools to its dew point.