A plenum lid is merely a sealing device. The primary reason to use a plenum lid is to make sealing near perfect, which guarantees all cooling air mass flow passes between hot parts. Any cooling mass bypassing the hot parts through seal leaks does nothing but provide drag.
The secondary reason is to relieve the cowl of the significant "balloon load" created by conversion of dynamic pressure to increased static pressure in the upper plenum space. Don't underestimate the load. I would not design for less than 230 KTAS/1000 feet/85%Q. That's about 150 lbs per square foot. If your lid is three square feet, that's 450 lbs.
I have seen plenum lids which don't seal very well at their perimeter, a mechanical design issue. Recall how the rear edge of a standard cowl will pooch outwards in flight? That's internal pressure. The edges of a plenum will do the same thing and leak if the attachment is poor.
I see a lot of plenum lids with sketchy sealing at the inlets, which negates the whole concept. May as well stick with flap seals.
I see quite a few with crappy duct connections between the inlet and the plenum space. Bad duct shape reduces dynamic pressure capture. As a general rule, if the cowl inlet area is small as compared to exit area (a high Vi/Vo inlet), the ducting shape becomes more critical. If it is large (a low Vi/Vo inlet), the duct shape is not critical, so much so that really low ratio Vi/Vo requires no duct at all, just a hole. See Mooney Acclaim, Cessna TTx, or a certain well known Rocket displayed at OSH this year, all of which have no plenum lid, BTW. Their flap seals, even if well done, will not match the leak rate of a really good plenum lid and inlet seal combination, but pressure recovery is very good. My next one will be a hybrid; I'd like to try molded seals with a reinforced upper cowl half.
For now, here's a low Vi/Vo inlet, urethane rubber and glass cloth flexible ducting, and a conventional fiberglass lid. Screws directly to the cylinder head ears on #1 and #2, with clamp plates. the rubber duct experiment is now at 800 hours. The system has near zero leakage. Easy to tell; look for black dirt tracks on the inside of the white cowl.