Suivi compteurs
English · Français
Android app for tracking water, gas, electricity and heating oil meters. It
gives you charts, a forecast for the year and a year-on-year comparison
adjusted for degree days, so that a harsh winter isn't mistaken for
consumption creeping up.
Free, with no account and no ads: all your data stays on your phone. The only
thing fetched from the internet is the weather, from
Open-Meteo.
Project page
· ⬇ Latest release
· ♥ Donate
Download
| Where | What |
|---|
| Google Play (closed test) | join the testers group, then accept the test and install the app from the Play Store with the same Google account. Automatic updates; your help is needed to open the public release |
| F-Droid | automatic updates, no proprietary components: you type the reading in by hand |
suivi-compteurs.apk | reads the meter from a photo (Google ML Kit, on the phone): choose this one if unsure |
suivi-compteurs-libre.apk | same app as on F-Droid, without photo reading: no proprietary components, you type the reading in by hand |
To install an APK, open it from your phone's file manager and allow the
installation when Android asks. Since version 1.6.2, the F-Droid build and these
APKs are signed with the same key, so you can switch between them. The Google Play
version is signed differently: switching to or from it means uninstalling, so
export your data to Excel first. The same applies once if you installed an F-Droid
version older than 1.6.2.
What the app does
- record a meter reading, with automatic recognition from a photo that you
always confirm;
- consistency checks: a meter doesn't run backwards unless it rolled over past its
maximum, and a date holds only one reading;
- after each reading, what it changes in the forecast;
- forecasts for the year, from one annual reading to the next or by calendar
year;
- detail for each energy: running total, monthly consumption, energy signature,
meter readings, years compared at normal weather, costs;
- comparison of years, weather, accuracy of past forecasts;
- several houses, meter replacement, tariffs, events (building work…);
- heat cost allocators: one line per radiator with its factor, annual resets
included;
- prepaid meters: read the balance, note your purchases;
- heat (district heating, heat meters), and meters you get consumption figures
for per period rather than a reading;
- heating oil tanks with a gauge: read the level, note your deliveries;
- points to check, including readings that were probably mistyped;
- Excel import and export, which doubles as a backup (Google Drive included);
- reading reminders;
- French, English, Dutch and German; your choice of currency.
How the numbers are calculated
The calculation engine is in
android/app/src/main/java/be/suivicompteurs/app/moteur/,
in pure Kotlin with no Android dependency.
From readings to consumption
A reading records the number on the meter, not a consumption. Two readings may
be seven days apart or three hundred. To compare years, each difference between
readings is therefore spread over the days it covers (Consommation.kt).
Spreading it evenly is fine for water. For gas it's wrong: 300 m³ used between
October and March doesn't fall evenly across those months, it follows the cold.
The thermal model
So the app fits a straight line, meter by meter:
consumption for the day = base + k × degree days for the day
base covers what doesn't depend on the weather: hot water, cooking,
standby power.k is the sensitivity to cold, in m³ (or kWh) per degree day. In essence,
it's the thermal cost of the building.
The fit uses least squares weighted by the length of each period: a 300-day
period counts for more than a 7-day one. It's kept only if R² ≥ 0.5; otherwise
the app falls back to an even spread and says so.
The model is fitted only for gas, heating oil and electricity. Water
consumption follows how the home is occupied, not the weather: an apparent
correlation over a few years would be a coincidence, and "correcting" for it
would distort everything.
The weather adjustment
Once k is known, each year is brought back to an average winter:
adjusted consumption = measured consumption + k × (normal DD − actual DD)
A colder-than-normal year therefore has its consumption revised downwards, and
vice versa. This adjusted figure is the one to compare from year to year.
Normal DD is the ten-year average for each calendar day.
Forecasts
What was actually measured is never rewritten: only the days after the latest
reading are estimated (Previsions.kt).
- Thermal model (when it's reliable): the remaining days are estimated from
the season's normal degree days, so a mild autumn isn't carried over into
the following January.
- Seasonal profile (otherwise): past complete years show what fraction of
the year's consumption has usually been used by this date, and the amount
measured so far is divided by that fraction.
With an annual reading, the year runs from one anniversary of that reading
to the next. The gap between the actual reading date and that boundary is
estimated, so years stay comparable even if the reading slips by a few days.
This is the beginning of the README. Read the rest at the original. Read the full README ↗
