Here goes: If you are looking at a 100A supply fuse as overload protection then yes, it probably needs to be 35mm cable in duct. However, you could get away with 25mm for the full running current of both motors and for a volt-drop point of view. Personally I would go with 35mm as then it is met for open 70C cable and so the end accessories are not at risk of working with 90C or thereabouts cable.
Estimated 10A three phase for ancillary might be high, that is 6.9kW, but not really much to argue over.
Looking at the Hager guide for 45kW motor assisted start it has 80A fuse as protection, 22kW individually as 50A fuse, so your 100A fuse is fine even for simultaneous start of both motors.
100A fuse in Table B3 of the OSG has 0.34 ohms as measured (i.e. cold) Zs for meeting 5s and also adiabatic for 16mm CPC so no issues there.
If the supply is TN-C-S and reasonable, say better than 10% drop on full load, then your supply Ze is going to be around 0.23 ohms (23V = 10% at 100A, probably less, but to be measured or inquired) so max R1+R2 for sub-main limit is 0.34 - 0.23 = 0.11 ohms.
Assuming 35mm 5-core cable then R1+R2 as seen from volt drop tables as 1.25 mOhm/m so length limit is 0.11 / 1.25E-3 = 88m and as longer than your 50m planned length is met.
If you are
really paranoid about worst-case cable fault you would look at armour resistance and work out the worst place for a fault, etc. Looking at the Prysmian data sheet for 35mm 5-core the armour is 1.6 mOhm/m so the worst location works out at 34.9m from the supply end and R1 there is 0.022 ohm and R2 is 0.028 ohms (armour from source in parallel with remaining armour to end and the 5th CPC core back to source) so R1+R2 worst-case fault location is 0.05 ohms and less than the above limit of 0.11 ohm based on a guestimated Ze and fuse for 5s sub-main.
Alternatively you could say max Ze = 0.34 (max fuse) - 0.05 (worst case R1+R2) = 0.29 ohm.
At the pump end you are either looking at 63A D-curve MCBs or 50A BS88 fuses for the pumps, based on Hager data sheets (page 102 of commercial one here
Hager Brochures, Catalogues & Documentations - https://hager.com/uk/p/catalogues-brochures). From the OSG Table B6 final circuit values for 63A-D we don't have Zs limit nor does the OSG give 50A final circuit fuses (not really domestic or common commercial I guess?). So let us compute the 63A-D MCB from max "instant" trip of 10-20 * In so max = 20 * 63 = 1260A so we can compute Zs limit as:
- Zs = 230 * 0.95 * 0.8 / 1260 = 0.14 ohm
From our guesstimated Ze 0f 0.23 ohm we can
never meet < 0.4s for the 63A D-curve MCB, we would need to rely on and RCD to meet disconnection times if socketed. If hard-wired, as probably the case, we can allow 5s as for the sub-main and that is listed in OSG Table B6 at 0.28 ohms for the MCB so we might be able to meet it if the supply Ze is a little lower.
Further more looking at the Hager tables on fuse-MCB selectivity we have
none with a 100A fuse, so this is
not a good design as a hard phase-neutral or phase-phase fault could take out the supply fuse(s) with any RCD seeing nothing wrong with such a current path. That is surprising as the 63A C-curve MCB is selective to 4.4kA faults so probably fine at end of cables, but alas not for the D-curve recommended for such a motor!
50A BS88 fuse from table on Fig 3A3 has 0.4s at 380A so Zs limit is:
- Zs = 230 * 0.95 * 0.8 / 380 = 0.46 ohm
As we already have to meed 0.34 on the sub-main for the supply fuse we could allow a further R1+R2 of 0.12 ohm and meet 0.4s final disconnection. Furthermore a 50A BS88 fuse is totally selective with a similar 100A upstream fuse, so this is a good choice. Just not so convenient as a couple of breakers in a board.
As above, if hard-wired then 5s is permitted and the OSG table B3 has Zs max at 0.79 ohm and adiabatic met for 6mm or more at that time.