[paulgato]

I've decided not to use glass fibre, just in case I need to return the battery under warranty. But if your battery is out of warranty California98Civic, then the fibreglass would be the easier option.

I phoned a local aluminium fabricator shop and found out how cheap aluminium plate is. They can cut me four rectangular plates - 6mm thick for the battery ends and 3mm thick for the long sides. Gonna cost about £15 total. I have some 3mm thick angle aluminium pieces about 70mm wide left over from a previous project, which just happen to be cut 130mm long, so just perfect for this. I shall epoxy bond and/or bolt those angle bits onto the 3mm thick long-side plates, and then I can slide those three-sided long-side pieces over the thicker end plates, thus holding the battery together securely, end-to-end. A few machine screws will hold the long-side pieces to the end plates, ...but one could equally well use gaffer tape or cable ties, as it is mainly end-to-end expansion we need to worry about.

The tech guy said that although the casing can sometimes expand in the sideways direction, expansion is only usually an issue in the lengthways direction as that is when plates start to separate. He said that the base and the top of the battery don't need reinforcement. In fact he said that when they test batteries at high temperatures they use two metal plates at the ends, clamped together with four threaded rods and nuts, and put the battery in the 'oven' like that to test.

That home-made metal jacket will be fully removeable if I ever need to return the battery under its 3-year warranty, and it is slim enough that the battery will still fit in my existing battery tray/clamp and existing battery box with no modification.

[Edit: Then again, if I use four plates rather than two, then wrapping with glass fibre will not involve glueing anything to the battery itself, so it could all be removed if necessary by cutting through the fibreglass at two corners. Much less work than all that drilling and bolting! Also there would then be no exposed conductive surfaces that could cause shortcircuits.]