@houseofancients @Culchee I hear and acknowledge what you are saying regarding the 2p16s, and had done some research earlier on it - there are several pro/con threads on here, and I decided to take a managed risk approach with it - I'm already quite safety conscious even with 3x16s packs, in terms of detectors and extinguishers, SOC and voltage cut offs, maximum loadings and C charge/discharge rates. I plan to make the new box lined with 12mm fire board and have 4mm board between each cell.
In the fire example you posted, I'm not sure that not having parallel connections would have prevented the problem he had, even he says "The BMS will stop power, but self discharge continues" - presumably a faulty cell could have caused this runaway even in a 1p16s configuration? Even so I acknowledge the increased risk of a faulty cell going undetected.
My first reason to choose this way was because I currently have constant issues with multiple BMS SOC and the inverter resulting in yo-yo charging at high and low SOC points - I think this is becoming a common issue as more people have multiple packs where the inverter receives multiple SOC readings from the stringed BMS. Redundancy is not an issue for me since I am grid connected and also have a 2nd 3kw inverter with its own solar string (no battery).
I'd really like to retain a single battery to inverter connection but wouldn't be adverse to running 16s2p internally, if there was a way to use two internal BMS fronted by the one master BMS (preferably the seplos I already have). Is that even possible while retaining use of the active balancer from the seplos? I don't think so.
Edit.. I re-found the doc I had read before which goes into a little depth on many issues with parallel strings, but parallel cells appeared to be less problematic:
https://www.orionbms.com/manuals/pdf/parallel_strings.pdf
