CLEVEREST estimates how long an EV charge will take, when it'll be done, what it'll cost, and how far you'll drive afterwards.
Charge time
The energy needed is simple — your battery size × the percentage you're adding. Turning that into time is the hard part, because charging power isn't constant: it changes with the state of charge.
So rather than one division, the time is added up in 1% steps. At each step the power used is whichever is lower:
- the charger's output — the speed you pick, or
- what the car accepts at that level — its max rate × the curve below.
Each step's time is its energy ÷ that power. Add them up, plus ~10% for charging losses:
time ≈ Σ ( battery × 1% ÷ powersoc ) ÷ 0.9
The charging curve
Every model's real curve is different, and there's no free, reliable database of them — so CLEVEREST starts from one typical fast-charge shape, scaled to your car's max rate. But you can tune it per car: open a car on the Cars screen and drag the curve to match your model. Whatever a car's curve is — the default or your custom one — is exactly what its estimate uses.
This is the curve for your active car:
The shape is why the estimate mirrors real charging:
- AC or slower DC — the charger is the limit the whole way, so it's near-linear (the curve barely matters up to ~80%).
- Rapid DC near the car's peak — the curve takes over and the last stretch slows right down.
What it can't know up front
It uses a generic curve, not your car's exact one, and it can't see the battery temperature or whether you've preconditioned — both change real-world speed. It's most accurate on a fast charger in the 10–80% range. A solid ballpark, not a stopwatch — but it gets sharper the more you tell it (below).
Learning from your charges
On the Sessions screen you can log what actually happened after a charge — start and finish %, the real time and cost, optionally the energy delivered (kWh) and the temperature. The app compares each one against what it predicted and quietly corrects future estimates. Nothing leaves your device.
A few principles keep it honest:
- Recent charges count most. Each session's influence halves roughly every 30 days, so a tariff change or a new habit is followed, not drowned out by old data.
- It stays cautious when data is thin. One session nudges the estimate part-way, not all the way to its own result; it takes several consistent, recent charges to move it fully. With no data it does nothing.
- The pieces are kept separate. Time is corrected per car, split into AC and rapid DC (which behave very differently). Cost is corrected per charger — and if you logged the energy, it learns your real price per kWh rather than guessing at how much energy went in.
Reshaping the curve, not just scaling it. For rapid DC, the time correction isn't a single blanket multiplier. The charge is split into even 10% bands, and your sessions are unpicked (a small regularised fit) to work out which part of the curve was off — a charge that only partly covers a band counts toward it in proportion. So if your car dawdles from 80% up but is spot-on lower down, only the top of the estimate stretches. Bands you've charged through little fall back to the overall figure until there's enough data.
Temperature. Cold slows rapid charging; AC is barely affected. Once a few DC sessions carry a logged temperature, the app learns how much your car slows per degree. Put today's temperature in the optional field on the calculator (it appears for DC chargers) and the rapid estimate adjusts to match.
Cost and range
Cost = energy added × your price per kWh. Range = battery size × charge level × your car's efficiency (mi/kWh).
Privacy
Your cars and settings stay on your device — nothing is uploaded. The app counts anonymous, cookie-free page visits (via GoatCounter) so I can see it's being used; no personal data, and no tracking across other sites.