← How ProTankr Works
How the temperature prediction works
Product temperature is one of the biggest hidden variables in a load, colder product is denser, so the same gallon count weighs more. Guess wrong and you're either leaving gallons on the table out of caution, or finding out you're overweight at the scale.
ProTankr predicts product temperature from real weather data and the physics of a large storage tank, then keeps correcting itself against what actually happened at each terminal, while always leaving the final call to you.
30 hours of weather historyWe pull hourly temperature, wind speed, and cloud cover from OpenWeather for the past 24 to 30 hours at this terminal's location. This gives the model a full picture of how the environment has been heating or cooling the storage tank.
Solar gain calculationUsing the terminal's exact latitude and longitude, we calculate the sun's elevation angle for every hour of the past day. Higher sun angle + clear skies = more radiant heat absorbed by the tank surface. Overcast or nighttime hours contribute zero solar gain.
Wind cooling adjustmentWind accelerates how quickly product temperature chases the ambient air. Higher wind speeds increase the effective cooling rate, pulling the prediction closer to ambient on breezy days and reducing the impact of solar heating.
Large tank model, intentionally conservativeWe model a large above-ground storage tank (~1 million gallons). Large tanks have enormous thermal mass, they heat and cool very slowly, lagging well behind ambient swings. This is intentional: we'd rather predict the product is colder and denser than it turns out to be, which keeps you safely under your weight limit.
Live ambient blendingThe current ambient temp is gently blended into the final result to account for the last few minutes of temperature change. This keeps the prediction current without overreacting to short-term spikes.
Self-training bias correctionEvery time you complete a load and enter the actual observed product temperature, ProTankr computes the difference between what it predicted and what you actually saw. This error is stored per terminal, per hour of day, and per month of year.
Over time, the model learns terminal-specific patterns, for example, that a particular terminal's tanks run 5 to 9°F colder than predicted at 3am in March, or warmer on sunny afternoons. The correction is applied automatically on the next prediction at that terminal.
The correction is weighted by confidence, it takes at least 3 observations before any correction is applied, and grows to full weight around 10+ observations. This prevents a single outlier from throwing off the model.
Over time, the model learns terminal-specific patterns, for example, that a particular terminal's tanks run 5 to 9°F colder than predicted at 3am in March, or warmer on sunny afternoons. The correction is applied automatically on the next prediction at that terminal.
The correction is weighted by confidence, it takes at least 3 observations before any correction is applied, and grows to full weight around 10+ observations. This prevents a single outlier from throwing off the model.
Confidence levelsHigh: Clear skies and calm winds over the past 24h. Solar gain was predictable and the model is well-constrained.
Medium: Partly cloudy. Cloud variability introduces some uncertainty in how much solar heat the tank absorbed.
Low: Heavy cloud cover or high winds. Use the number as a starting point but lean on what you know about this terminal.
Medium: Partly cloudy. Cloud variability introduces some uncertainty in how much solar heat the tank absorbed.
Low: Heavy cloud cover or high winds. Use the number as a starting point but lean on what you know about this terminal.
Use your judgement. Override the prediction freely, you know your terminal better than any model. It is strongly recommended not to set the planned product temp above ambient unless you have full confidence from a recent BOL. When in doubt, err cold. A colder planned temp predicts denser product, which protects you from overweight loads.
Equipment Setup