Where the money actually goes at an off-grid tower site

Tower power is usually budgeted as a fuel line, because fuel is the part that is visible. The costs that move the total are the ones attached to sending a person to the site.

A diesel genset at an off-grid tower needs a person to visit it roughly once a month. Oil, filters, coolant checks, top-ups, and the refuelling run itself all require someone to physically arrive. Almost everything expensive about off-grid tower power follows from that single fact, and almost none of it appears in the budget line the site is measured against.

Operators plan tower power around fuel because fuel is metered, invoiced, and easy to attribute to a site. The other three lines are diffuse. They sit in field-operations budgets, in vehicle fleets, in shrinkage write-offs, and in service-credit accruals. Added together they frequently exceed the fuel bill, and they respond to different levers.

The four lines that matter

Fuel is the first line and the best understood. It scales with load and run hours, and it moves with the local delivered price, which at a remote site includes the cost of getting the fuel there. Two sites with identical loads can differ substantially on this line purely on the strength of their road access.

Service labour and travel is the second line, and it is the one most often mis-modelled. The cost driver is the journey. Parts are cheap: a filter set costs very little. A vehicle, a driver, a technician, a day of working time, and in many territories a security escort, are expensive. Because the driver is the journey, this line scales with how remote a site is and how hard it is to reach, and it does so almost independently of how large the site is. A small site four hours off the tarmac costs more to maintain than a large site on a main road.

Fuel loss is the third line, and it has two distinct components. Theft is the one that gets discussed: fuel siphoned from the tank between visits, sometimes opportunistically and sometimes systematically. Under-delivery is the quieter one: the tanker that invoices for a full load and delivers less than a full load. Both show up in the same place, as a gap between fuel purchased and energy produced, and both are hard to isolate without metering at the site. Exposure varies widely by territory and by site, so the honest answer on scale is that it is site-specific.

Downtime is the fourth line. When a site goes dark, the operator pays twice: once in SLA penalties or service credits to the tenants on the tower, and once in subscriber experience, which over time reads as churn. Operators rarely book either against the power budget, even though power is the cause. That accounting choice makes the power line look cheaper than it is and makes reliability improvements look less valuable than they are.

What changes when the visit count drops

A microturbine genset is specified for one service visit per year. Set against a diesel unit that wants monthly attention, that is twelve visits a year against one.

Twelve to one is a different operating model, and the second-order effects are larger than the arithmetic on the service line suggests. Security escorts are commissioned per journey, so eliminating eleven journeys eliminates eleven escort bookings. Spares logistics changes shape: a consumables pipeline sized to deliver filters and oil to every site every month becomes an annual replenishment exercise, which changes warehousing, forecasting, and the working capital tied up in stock. Field-engineer headcount is planned against the total number of visits an estate generates, so the number of engineers a given estate requires falls with it. Vehicle fleets, fuel for those vehicles, and the overtime attached to long drives all move in the same direction.

There is a risk dimension too. Every journey to a remote site carries road risk, and in some territories carries security risk. Fewer journeys means less of both. That is difficult to put a number against, but operations directors recognise it immediately.

What the field data shows

The figures below are the ones published on this site's own case study pages. Each links to the page it comes from.

Abbott Technologies in South Africa recorded an OPEX reduction of over 50% across off-grid telecom sites, moving to one service visit per year, with particulate emissions 97% lower than the diesel gensets replaced.

A West Africa operator ran a proof of concept that recorded a monthly maintenance saving of $276 at a single site, zero routine maintenance across the trial, and operation on three fuels: diesel, kerosene, and a blend of the two. At estate scale, a per-site monthly figure is the number that compounds.

In Papua New Guinea, a hybrid deployment across 70 sites covering 500,000 km² of territory reported over 90% diesel displacement, with one service visit per unit per year.

Beyond those figures, the numbers are site-specific. Delivered fuel price, road access, escort requirements, and local shrinkage rates vary enough that a single published average would mislead more than it helped.

How to model it for your own network

Model the whole estate. Take a representative site archetype (or three: roadside, off-road, and hard-access) and cost each of the four lines separately for each archetype, then weight by how many sites of each type the estate holds.

• Cost a service visit fully loaded: vehicle, driver, technician time, escort where required, and the day of working time it consumes. Parts are a rounding error against that total.
• Put a percentage on fuel loss per archetype and be willing to write down a wide range. Metering a handful of sites for a quarter is the cheapest way to replace the guess with a measurement.
• Book downtime against the power budget, using the SLA credit rate the tenant contracts specify.
• Run the comparison over a multi-year period, because capital differences amortise and operating differences compound.

The TCO calculator on this site does exactly that arithmetic, with editable inputs and regional presets, and will produce a written report you can circulate internally.

Fuel is the line everyone can see. The visit is the line that decides the total.

This article is part of the Bladon archive and reflects the position at the date of publication.