Electric Utility Network and subnetwork controllers

[LS811]

Hello!

We are setting up an electric Utility Network currently for our LV network only at this time. 

Is there a recommendation for what would make a better subnetwork controller: fuse or transformer. 

For context, our network has transformers both at pole mounted and ground mounted levels.

Typically a pole mounted circuit feeds a single house but there are many occasions where it feeds up to 90 in rural towns.

In ground mounted substations there is typically 5/6 fuses each feeding their own circuit, and ideally we want to keep each circuit with their own naming convention e.g. TEST-FU1, TEST-FU2 and so on so that we can identify the substation name and the fuse it’s fed from. 

This naming convention would also have to apply to circuits coming off a pole mounted transformer which is connected to fuses. 

Currently it’s configured as fuses as the subnetwork controller at ground mounted transformers only which is preventing us naming circuits like on site. We have been recommended to change to transformers at ground mounted level which again does not met the needs. 
It appears to be if we make all fuses at both PM and GM level a subnetwork controller we would be creating approx. 100,000 subnetworks. 

Is this too many subnetworks for the LV tier to handle? Will there be performance issues especially in respect to updating the subnetwork? 
Has anyone any experience with this volume of subnetworks and could give context on timings, server loads, performance etc. 

We need to represent digitally each feeder on site and unsure what is best practice to represent that without bringing the system to a halt with performance issues. 

Any help is greatly appreciated! 
Thanks! 

[RobertKrisher]

You'll need to weigh the business value of having more granular network with the time it takes to manage all the individual networks. Individual subnetworks will trace and update quickly, but the biggest problem with managing a dataset with many thousands of subnetworks is the amount of time it takes to update ALL your subnetworks. This means you need to account for this additional time when you first go live, and every time you need to disable/enable your network topology to make schema changes.

An alternative would be to only model subnetworks at the transformer level, and for the fuse-level transformers use something like the batch trace tool to calculate the individual fuse-level circuits.

Note: ArcGIS Pro 3.7 introduces an optimization that allows the first time you run update subnetwork to be significantly faster, that will help with your initial deployment. I recommend you test this.