Why household energy surveys overestimate demand

Before building a mini-grid, a developer has to estimate how much electricity the community will buy. That number decides how big the system should be and what it will earn. The traditional way to estimate it is a household survey. The research says that method often goes badly wrong.

What the studies found

• Kenya: a study of eight rural mini-grids found survey predictions were off by an average of 426 Wh per customer per day, while customers actually used only about 113 Wh per day. The error was almost four times the real consumption.
• Tanzania: researchers found survey-based estimates typically overpredicted demand. Their best model still had a median error of 37% at the site level.
• East Africa: using two years of data from 31 sites, researchers found that each customer's survey answers did not predict their actual use.

Why surveys overshoot

• Hope: people describe the appliances they'd like to own, not the ones they can afford this year.
• Wanting the project: communities naturally want to look like a good site.
• Guessing: people who have never had metered power find it hard to estimate kWh.

The cost of getting it wrong is real. An oversized system costs more to build, earns less than planned and takes longer to pay back.

Better ways to estimate demand

• Count what people spend today: fuel, generator repairs, phone charging and batteries are real money already being spent.
• List the appliances people already own, rather than asking about plans.
• Find the anchor customers: mills, cold rooms, clinics and workshops drive most of the demand. Researchers suggest focusing surveys on them.
• Compare with similar sites: real data from mini-grids in similar villages predicted use far better than surveys in the Kenya study.
• Build to grow: start with a sensible size and leave room to add capacity.