It could have to do with the trip curves of the breakers, and/or the ambient temperature differences between panel locations.
Trip Curve
For example, let's say all the circuits on Phase A are drawing 3 times the rated current. The trip curve for these breakers, say they will trip between 10 and 30 seconds at 3 times current. So the total current on Phase A at the main panel is 270 amperes, or 2.7 times the rated current (100 A). The trip curve for the breaker in the main panel, says that it should trip between 12 and 35 seconds at 2.7 times current. If the breaker in the main panel is a bit more sensitive, you can see that it might actually trip before any of the secondary breakers.
Ambient Temperature
As well as a trip curve, circuit breakers will also have an ambient temperature curve. Breakers are designed and calibrated to operate at 40°C (104°F) ambient temperature. If the breaker is operated at an ambient temperature of -10°C to 24°C (14°F to 75°F), the breaker will be able to carry more current. However, if the breaker is operated at an ambient temperature between 41°C to 60°C (106°F to 140°F), it will carry less current and could lead to nuisance tripping.
For example, if the 100 ampere breaker is operating at 140°F ambient, you might find that it actually trips at only 85 amperes.
High Altitude
When breakers are installed at higher altitudes, the current has to be adjusted due to the reduced cooling effects of the thinner air. According to the documentation I could find, the adjustment multipliers are as follows:
- 0 - 6600 ft. (0 - 2011 m) -> 1
- 6600 - 8500 ft. (2011 - 2591 m) -> 0.99
- 8500 - 13000 ft. (2591 - 3962 m) -> 0.96
If you were installing a 100 ampere breaker at 10,000 ft., the breaker should only be expected to carry 96 amperes (100A * 0.96 = 96A).
Though in your situation, the breakers are likely all installed at a similar altitude. So this it likely not the problem, unless the air near the main panel is significantly thinner for some other reason.
That Federal Pacific panel has to go as they are dangerous. As far as concerns about your skill, you sound like that particular kind of newbie who is well capable of learning to do it all safely and well; however my hunch is you are still thinking too much, and need to read a little more. It's OK, we all start there.
Normally, just replacing a sub panel is a straightforward thing. Change panel, reattach wires, done. However, this is only part of a project with a much larger scope. You must contemplate (i.e. ask your permitting authority) whether you have crossed the legal threshold of a remodel. If you have, everything in-scope must be done to all current codes. Even ADA!
The 2-circuit requirement for kitchen outlets is not intended to mean "1 circuit for this wall, 1 circuit for the other wall". I'd encourage some more research but if it was me, I'd interleave the outlets, every other outlet on a different circuit. Also there's nothing wrong with more than 2 outlet circuits in a kitchen, the whole point is to prevent trips when the chef is madly at work, so the chef isn't hobbled with limitations like having to put the toaster here and the George Foreman over there, and avert ugly workarounds like extension cords draped across sinks or stoves, etc.
I see you plan to go 12 AWG wire for almost everything (that's what 20A breakers mean) - that's awesome. Feel free to kick the refrigerator and smoke detector up to 12AWG also - that way you don't have to buy any 14AWG wire. I don't own any! If you have some other reason to use a 15A breaker you are welcome to use that on a circuit wired in 12AWG.
You may want to run the water heater circuit in 10/2 or even 8/2. That will allow you to easily upgrade to an electric water heater in the future. Still use a 15A or 20A breaker because the outlet is still only good for 20A. There is a trick to fitting 8+ AWG wire on a 15-20A outlet, just ask.
I would go with a much larger panel. You have either -1 or 3 circuits left, and that's too little headroom for my comfort. Getting a larger panel is dirt cheap compared to the cost of replacing perfectly good breakers with duplex breakers (I call them double-stuff) merely to shoehorn everything in. Also, larger panels in combo-packs come with more breakers and that is far-and-away the cheapest way to buy breakers. Another reason to avoid duplex breakers is if you ever need AFCI, GFCI or whatever future thing comes out - those are much more expensive in duplex breakers because of the miniaturization required. Don't think you must use a 100A panel - you can use a larger panel (200A), you just can't use a smaller one (70A).
Remember each sub-panel must have its neutral bus bar separate and isolated from its ground bar. That means removing bond straps, magic green screws, neutral bar kits, whatever the panel requires to do that. You might consider a panel with a neutral and ground bar on each side of the panel. That's a convenient feature so wires don't have to cross over the panel.
Keep in mind how your house got a dangerous Federal Pacific panel. The last guy bought cheap. Feel free to research the good-better-best that each manufacturer offers, the price differential for "best" is quite small compared to the overall cost of a remodel. You may find better selection and better prices at a real electrical supply house.
There is nothing wrong with more ground rods. Go nuts. The key is that all the grounds are connected to each other by wire - and they are not connected to neutral anywhere except one place - the main service panel.
Best Answer
An AFCI is more a protection from "fire issues" than anything you'd typically associate with "having electrical issues."
NEC allows use of NM cable in residences. NM cable is sadly prone to being gnawed at by rodents, and rodents are sadly prone to be in houses, even nice ones. Some folks estimate that a human is rarely more than 15 feet from some form of rodent (mouse, rat, squirrel etc.) I've seen multiple-feet-long sections of NM cable gnawed back to copper in remodeling jobs, and those were live, operational, and not having any electrical issues - because the copper had not touched the next bit of copper in the cable yet. IF a solid connection was formed between them, you'd expect a breaker to trip. But if something not as robust made a connection, you could get a hot plasma carrying current between the conductors without necessarily drawing more current than the breaker was rated for - and that hot plasma can set anything combustible on fire.
An AFCI is looking for the "signature" of a plasma discharge, and some of the early ones were rather terrible at discerning the difference between a fault and a brushed motor, which has tiny arcs between the brushes and the rotor. I believe they are somewhat better now.
People who should STRONGLY consider AFCI retrofitting when not required by scope of other work requiring code updates would be those with 15 & 20 amp circuits on the problematic old Aluminum wiring (1960's era, vaguely) that is a particular fire hazard, and those with other "very elderly and suspect" wiring. Lacking those signature items, consider it, sure, but how critical it is will be a personal call, possibly influenced by your level of thinking your house is utterly rodent-free or otherwise.
Other things can cause arcs, such as poorly done work (loose wire-nuts, using backstab connections) or the classic 'nail into a cable' when it does not trip the breaker.