Plan A - Lead Acid, high quality ? 2Sx12V
Shelf Life of Lead Acid batteries is not a common spec as they are prone to sulphation from lead and antimony which insulates the electrode and raises ESR by orders of magnitude.
Only battery supplier with test results to backup their claims can be replied upon as process contaminants influence the results.
I might consider major brands and ask for R&D test reports for shelf life. from my experience, some small pulse discharged boxes are effective in preventing the rate of sulphation under a float charge as well as breaking down sulphation and restoring the specific gravity.
Plan B - Li primary 8S4P
- may be cheaper since no charger, but cell costs and reliability vary greatly.
- $20 at Home depot 10yr shelf life
- 12x Primary Lithium cells 3V 1.5Ah x12 = 54Wh
- Each cell may have an ESR of 150 mΩ depending how measured
- if load is actually 200W/24V then equivalent , Req =V²/W = 24²/200 = 2.9 Ω
- then 8x cells would be 150m total of 1.2 Ω
Assuming a permitted load regulation error of 10% then ESR of 24V battery pack must be <10% of 2.9 Ω or <290 mΩ which means you need 1.2/0.29=4.1 parallel arrays of 8 in series or 8S4P = 32 cells
Fusing each string or using PTC polyfuse would be prudent.
Plan C - LiPo 1P7S
The same could be done with 1 string of 7S. V =26V LiPo cells if the Mfg guarantees shelf life at >50% SoC float voltage 3.6~3.7V charge since they have ESR < 10mΩ
The problem with LiPo cells is that shelf life reduces with > 2/3 SoC so they ship them with 50% to 66% SoC. Shelf life may be less than 10 yr and depends on cell temp.
Depending on need for fault detection reliability, a circuit to measure ESR and V drop with a 1ms pulse could be done every month with historical data and a failure threshold determined by design tests.
.. These ones have a 20y shelf life 7S?P
but may be more expensive but then reliable, very the ESR specs with shelf life report, so perhaps fewer Parallel strings.
This could be the best bet for reliability and simplicity. 7S?P