Small details can (will) contribute to the welders' quality of work
and his opinion of working with you in the future; good prep, clean well marked, cut and fit material and sequences of work all laid out will make a better relationship than dropping off a box of cut parts and expecting a welder to be able to understand what you planned.
Keyhole welds, pocket welds, some even say socket welds but that confuses with a type of welded pipe fitting of that name, are welds where a hole is put in one material and the two materials are stacked and a weld inside the hole on the top sheet; holds the two pieces together. Sort of a welded bolt or rivet - a single point of fusion of the two metal pieces. This is a completely legitimate weld joint type, but a picky one, and especially picky if you aren't a welder and therefore ignore what is about to be shown?
We've talked about the arc, not that tank builder is necessarily going to weld, but; we needed to explore the
Why of some joints for reasons we've shown above. This post explores another small detail that can come back as a problem in a tank build and cost you shop time that could be avoided - if this information is incorporated into your build.
The subject is the preparation the keyhole or pocket welds' joint prep. Specifically, we'll explore the
walls of the holes cut in the top material; these walls should be conic/flared out/wider at the top than bottom and hopefully we'll all agree why that is the case.

This sketch, above, is of two pieces of tungsten, the non-consumable electrode used in TIG welding to provide the arc and therefore the heat of fusion of the aluminum. NO gas cup is shown, no argon gas cone is shown, and no filler wire, this sketch introduces just the two arc cone's proportions.
There are two blue cones of different proportions coming off the sharpened points of the electrodes and their different proportions are from the angle of the ground tips. Notice one electrode tip is more like pencil, sharpened to a longer cone that the right side - blunted cone? The more blunt the cone the narrower the arc, the sharper the tip is ground the wider the cone of the arc. We don't care why that is true, but it is. What we do care about is how this can be a problem for our weld preparation?

All the holes in this sketch, above, have vertical sides so the top and bottom are the same width. The smaller hole to the left would have to have a pretty narrow arc cone to reach down inside and melt to the yellow angle extrusion's top surface in order to make this keyhole weld work.
The weld should start at the bottom edge of the top material where the two parent metals come together.
The wider hole may allow SOME arc cones down inside- in other words the wider hole may be the solution but you can't control how your welder grinds his tungsten, and even with a narrow arc width from a blunt ground tip... the arc may not reach the bottom of the hole.

IN this third sketch, our view point changed down close to the plane of the work, to see what the problems can be. The arc cone has 'swung' or wandered to the side of the top materials, the arc is finding the closest path to the welding arc Work connection to the welding power supply (sometimes called 'ground' but the is AC TIG so that term is not too useful). When the arc wanders to the top side edge; it will not direct heat onto the top of the back up material first. That arc wander happens because of the two features in this image.
First, the arc cone may not 'fit' down inside the hole for the weld and (Second) with the shoulders of the drilled holes closest to the electrode there is a natural tendency for the arc (even of narrow arc cones) to heat the sides of the keyhole first! NOT GOOD.

What is the solution? This image shows the smallest hole, left, that has be made into a cone instead of a cylinder and the longer weld to the right has also been widened at the top surface edges.
Now the arc cone, even a wider cone will reach the angle extrusion's surface before the top edge of the sheet is melted. This allows the welder, even if using a MIG welder (not preferred for this weld but 'doable') to melt/fuse/weld at the boundary between the two metals- first. Each weld will be initiated on the back-up material's upper edge edge and the conic shape of the hole's sides allows the lower edge of the two materials to be fused first, building up to the top of the weld last.

This last sketch in this series of 5, above, just closes in to show each arc cone can now reach the bottom material and the lower edge of the top material to initiate the welds; instead of being hard to control and 'running off' to the top edges of the top material.
The areas under the arc cones now, are the edge of the two materials, before there was no assurance this could be accomplished? Both MIG and TIG arcs will wander to the closest 'shoulder' of a drilled hole and therefore fusion of the bottom is in question.
I'm sure if you're not used to welding these types of welds, or have done them but simply made bigger holes in the top sheet to 'get down there' ?? you're probably wondering why this matters in a tank builders' set of methods of work?
The reason is that if the welds are not initiated down at the two materials closest edges, where the meet at the upper surface of the back up material- then the line of fusion in the entire keyhole weld is faulty and the weld may fail- pop loose- come undone! IF the top surface of the back up material is not seen (in the welder's hood while welding) to fully melt and fuse to the lower edge of the sheet overlay(?) then the chances of failure are too high too accept.
Where do these welds happen? They happen on tanks that have a top added last, where one sided welding on the perimeter seams is common, but there are sufficient structural elements, like baffles for example, that remain to be welded to the top. In order to fuse the 'field welds', which are not part of the top's outside seams, keyhole or pocket welds are used to weld the tank top to the structural elements inside. (also used to put decks in welded boats)

Here's a picture of a router bit that will do most of this work- the beveling work not the hole drilling work. This type bit can be run with a top table guide on the base to keep the router in a given hole, elongated or not, and also with a bearing guide from below, clamping a plate with an over sized hole to guide the bearing under the work piece.
Why not just make the holes big enough to get all the different types of welding done done inside a bigger opening- why bother to bevel these keyholes' edges? The reason is to keep the keyholes as small as your welder will accept them. These welds will 'pucker' the tank top (any panel can be distorted if these welds are not put in correctly), the larger weld can more easily have a crater crack in the center as the bigger puddle cools more slowly and the "arc core" may not be cooled to solid before the weld top is cooled (cracking) and that will leak in the pressure test- then leak fuel.
What happens when the tank is pressure tested? Some or another form of soap is spread on the tank and you look for bubbles. Then take the pressure off, and TIG the "leakers" . But its not that easy... the soap gets into the leak and now the amount of re-work gets bigger as the welder has to carbide burr out the crater in the keyhole and reapply the weld, what if it leaks from the soap contamination down in the weld metal... well the result is sailor like language by the welder about your Mother! Not that she has his profane rant coming, nope , just that the welder is frustrated by having to do this work again, and possibly again, again.... and you're paying for that time too!
So if the tank design is large enough that entire panels have to be put on where they need to attached to the interior structure, like baffles or supports, AND keyhole/pocket welds are used; consider taking the time to make each weld #1 the correct diameter for your welder's preferences, #2 bevel the sides to allow the welder to reach down in the weld to start without arc wander and #3 done once, correctly, each weld will probably test tight, and the work time to weld your tank will be reduced or conserved.
A hand held die grinder with a conic or pear shaped carbide burr will bevel the edges of these holes as well but are sometimes less controllable depending on the bit to hole size ratio.

This last photo in this post shows keyhole welds holding a piece of a skiff deck to the framing. The deck in this case is an
air tank, and was pressure tested to 3PSI to insure the integrity of all the welds. All the keyholes were bevel sided and none leaked because they could be welded in the correct sequence- bottom up to the top, cooling at the top and then stopping the arc on TOP OF cooled welds that were fused from the bottom first.
Cheers,
Kevin Morin
Kenai, AK