Chief Technology Officer / Canada
Four decisions that traded visibility for evidence
- Customers served by Quillmere Hydro
- 287,000
- Homes and businesses in fourteen Ontario communities, 2026
- Meter replacement programme
- C$112m
- Feeder-sequenced, 2025–2029; 121,000 meters installed by June 2026
- Longest drill running the grid in isolation
- 72 hours
- October 2025; the first drill, in May 2022, lasted nine hours
Decisions & consequences
01Sequencing the meter replacement by feeder rather than by route
By 2023 the 287,000 meters installed across Quillmere's territory between 2008 and 2010 were reaching the end of their life. Under federal metrology rules, sampling had failed for three lots totalling 41,000 meters, which therefore had to be replaced by the end of 2026, and our supplier would stop supporting the collection system in 2027. Meanwhile our load was changing street by street: heat pumps and vehicle chargers were concentrated in newer subdivisions, and voltage complaints had risen to 312 a year. The supplier offered a three-year, route-based replacement at C$318 a meter installed, which our chief financial officer favoured.
I proposed a five-year programme sequenced by feeder: first the failed lots, then the feeders with the fastest load growth, the most voltage complaints or the poorest outage visibility. Every new meter would report fifteen-minute consumption, voltage and a last-gasp message from the first day. Because crews would move between areas, the unit cost rose to C$349 and the total to C$112 million, against C$104 million. Our chief financial officer objected to the premium and to two collection systems running in parallel for longer. Our regulatory affairs director worried that the regulator would question it. Aileen Strachan approved my plan on condition that by the 2027 rate application I could show deferred spending that exceeded the premium.
We ranked all 540 feeders and began installing in January 2025. By June 2026, 121,000 meters were in service on 214 feeders. Their data identified 2,140 customers outside the normal voltage range, and transformer tap changes corrected 1,870 of them. Measured loading deferred C$19 million of transformer replacements. For metered customers, the time to detect an outage without a telephone call fell from 38 minutes to six. The premium paid so far is C$3.8 million.
Transformer replacement in our capital plan now starts from measured loading rather than nameplate estimates. I would change two things. I assumed that 3% of meters would need an appointment for access, but in older town centres, with meters in basements, the figure was 9%, which cost us seven weeks. I also published our statement on the use of interval data three months after the first installations, following a question from a councillor. It should have come first.
Automated switching instead of burying two town centres' lines
In February 2022 a freezing-rain storm took 166,000 customers, 58% of the total, off supply at its peak, and full restoration took six days. The centres of our two largest communities were without power for up to three days, as ice-laden limbs from their mature street trees brought down overhead lines. The council of our largest shareholder municipality asked us to bury the lines in both centres over eight years, at an estimated C$138 million. The request was public, popular with downtown businesses and supported by our own vice-president of operations, who argued that underground cable would also be safer for crews.
I analysed the storm by feeder segment. The two town centres accounted for 9% of customer-minutes lost; long rural feeders with heavy ice loading accounted for 61%. I proposed a four-year, C$44 million programme: 212 automated reclosers and switches with fault location, isolation and service restoration on 96 feeders, 290 kilometres of covered conductor on the worst spans, and undergrounding of 2.4 kilometres in the centres, where tree canopy and access made overhead repair slowest. We gave up the visible answer. The mayor said in council that we had chosen spreadsheets over main streets, and some business owners remain unconvinced.
The board accepted the proposal after I presented the customer-minute analysis alongside the shareholder's letter. We completed the work in late 2024 at C$41.6 million. The late-March 2025 ice storm took out 47% of our customers at its peak. Automatic switching restored 37,000 of them within five minutes of their faults, and 90% of all customers were back within 52 hours, compared with four days in 2022. The most remote rural lines still took eight days, longer than I had hoped.
Storm investment is now ranked by customer-minutes avoided per dollar. What I would do differently is the communications design. Our reclosers relied on commercial cellular service, and 31 sites lost contact when towers ran out of backup power after about four hours. We have since spent C$2.3 million on our own radio links with battery backup. I should also have taken the customer-minute analysis to the council before the shareholder's request was made public, rather than after.
Closing the supplier doors into our control systems
When I became chief technology officer in April 2019, an inventory confirmed what a review had suggested a year earlier. Fourteen always-on supplier connections reached our control system, our relay engineering tools and the meter collection system, and data from the control room's historian reached the corporate network through a firewall rule nobody could explain. In 2021 a large US fuel pipeline shut down for several days after an attack on its business network, and our board asked whether we could keep running if the same happened to us. The honest answer was that we did not know.
I proposed three changes: no always-on supplier access, with every session brokered, recorded, time-limited and approved by two people; a one-way link for data leaving the control network; and a standing ability to run the grid for seven days with the corporate network disconnected. Our vice-president of operations objected that a fix would now take hours rather than minutes when a supplier was needed urgently. Two suppliers asked for higher support fees, adding C$640,000 a year, and one refused to work through brokered access, so we replaced its relay engineering tool at a cost of C$1.2 million. The chief executive accepted the cost and asked for a drill report to the audit and risk committee.
Our first isolation drill, in May 2022, broke down after nine hours. The outage management system depended on the corporate identity service, and field crews' dispatch tablets could not reach it. We gave the control room its own identity and dispatch arrangements, printed switching procedures for our 96 automated feeders and trained every operator to use them. The October 2025 drill ran for seventy-two hours with no loss of switching capability. No supplier has held always-on access since December 2021, and urgent supplier sessions are now approved in a median of 22 minutes.
Isolation drills are now an annual item on the committee's calendar. If I did this again I would start with the paper procedures and the people, not the network architecture, because the drill exposed dependencies that no diagram showed. I would also have included the meter collection system in the first drill. We left it out to keep the scope manageable, and it turned out to share a server with the corporate network.
Recommending a capture pilot instead of a capture plant
In 2012 Kelsvar Oil Sands Corporation's parent, a listed European energy company, announced an ambition to capture a million tonnes of carbon dioxide a year in Canada by 2020. Our in-situ site emitted about 2.3 million tonnes a year from steam generators and the cogeneration plant, none of it as a concentrated stream. A full-scale plant on the steam generators' flue gas was estimated at C$1.4 billion. I led the technology work to say whether and how it could be done.
Our studies put the cost at C$118 a tonne captured, but the capture plant needed about three gigajoules of steam per tonne, and raising that steam meant burning more gas. Per tonne avoided, the cost was about C$146. The provincial compliance price at the time was C$15. I recommended deferring the plant, running a 30-tonne-a-day pilot for C$31 million to test the solvent in our own flue gas, and putting money into reducing our steam-to-oil ratio, which cut emissions faster per dollar. The parent's climate strategy team and our vice-president of corporate affairs wanted the plant announced; a solvent supplier offered to share the cost of a larger demonstration. In accepting my recommendation, we gave up a headline and a possible partner.
The pilot ran from November 2013 to January 2015 and captured about 9,800 tonnes. Solvent losses were 2.3 times the supplier's estimate, because of oxygen and nitrogen dioxide in the flue gas, and the steam requirement was 3.4 gigajoules a tonne. The revised estimate for a full plant rose to C$171 a tonne avoided. Over the same period our efficiency work reduced the steam-to-oil ratio from 3.1 to 2.7 and avoided about 190,000 tonnes a year at a fraction of the cost.
The parent withdrew its 2020 ambition and began reporting capture claims in avoided rather than captured tonnes. I would do two things differently. My first paper quoted the C$118 captured figure on its opening page, which flattered the case. I should have led with avoided tonnes. I would also have explored a shared pipeline and storage arrangement with neighbouring operators instead of appraising our site alone, because transport and storage were a third of the cost.