The systems an electric cooperative cannot run without

Data-centre developers now approach rural cooperatives for hundreds of megawatts at a time, and they often send the same project to several utilities at once. I treat each enquiry as a record that has to earn its place in the load forecast. At Merrowind every request over 20 megawatts received by any member goes into one register, is matched by parcel, developer and nearest substation so that a project counts once, and moves through four gates: an enquiry, a study agreement with a deposit, a facilities agreement with financial security and a signed electric service agreement. Only the last two enter our base forecast; the study stage appears as a scenario. I work with our planning, transmission and member services teams, and the resource decision is not mine. My responsibility is that the numbers our board and lenders see describe commitments rather than correspondence, and that members' commercial information stays within the access rules we agreed with them.

Our control centre, transmission substations and generating plants fall under the NERC CIP standards, the mandatory cyber security requirements for the bulk power system, and our operations and compliance teams maintain that programme with my team's support. My particular concern is the boundary those standards draw. Market submissions, gas nominations, energy accounting and outage management sit outside it, yet losing any of them for three days would stop us scheduling, fuelling or settling power. I classify systems by consequence, asking what stops, for whom and for how long, rather than by which network they sit on. Fourteen applications now sit in an operations-essential enclave with separate identity, hardware-key authentication, offline backups and written manual procedures. I give the audit and risk committee the measured restore time from our latest test alongside the compliance position, and I design the board's annual tabletop exercise, which in 2025 opened with our trading desk unable to file.

A cooperative's billing system does more than send bills. It applies power cost adjustment riders, runs prepaid and levelised plans, holds the payment histories that collection decisions rely on and supplies the purchase records from which capital credits are allocated to members. I have replaced a billing system at an investor-owned utility, kept one running through a storm at a Texas retailer and now lead the shared billing conversion for our members. My method starts before the software. Each member's board reviews its rate schedules and retires what it no longer needs, because every surviving rate code must be configured, tested and explained at the counter. Conversions then run in waves, with two bill cycles in parallel and every bill that differs from the old system's by more than 2% held until a person has examined it. Member service representatives test the system before go-live, not only the vendor's analysts.

An outage management system predicts which device has operated by grouping calls and meter outage events against a model of the network: which transformer serves which meter, which fuse protects which tap, which phase each line carries. When the model is wrong, the prediction is wrong and crews drive to the wrong place in the dark. I treat connectivity data as operational equipment. After a February 2023 ice storm in which only 58% of first-device predictions were right, we spent two years verifying connectivity with members' field staff and with meter event data, cutting transformer association errors from about 11% to under 2%. I also plan capacity for the worst hours rather than the average day. The voice response system, the outage map and text alerts all draw on one outage engine, so I size it for the first three hours of our largest storm. In major storms my staff work in members' storm rooms beside their dispatchers.

Fourteen of our seventeen members use some part of Merrowind's shared technology service, which provides billing, outage management, mapping, network operations and security monitoring to cooperatives too small to run them alone. It is a cooperative within a cooperative and has to be governed as one. Each member pays for what it uses under a cost allocation that participating managers review every year, and decisions about shared configuration go to a user committee of member staff rather than to my team alone. I have also learnt to make leaving possible. When our largest participant decided in 2023 to buy its own billing system, we agreed an exit plan with data extracts, a fee schedule and a date, and my team supported its conversion. A shared service that members cannot leave stops answering to them. One that lets them go on fair terms has to keep earning their participation, and that pressure has improved ours.