Three numbers, everything else derived
Window length, start count, total run time. From those: run fraction is run over window, cycles an hour is starts over hours, average on time is run over starts, average off time is the rest over starts. The defaults — 24 hours, 96 starts, 720 minutes of run — give 50 percent run time, four cycles an hour, 7.5 minutes on and 7.5 minutes off.
Everything on the page is one of those four figures rearranged. That is the useful thing about a run time log: it is only three measurements and it settles arguments that otherwise run on opinion.
What a wider differential actually does
It stretches both legs of the cycle by the same factor and leaves the run fraction alone. Go from 2 °F to 4 °F and the on time goes from 7.5 to 15 minutes, the off time does the same, cycles an hour halve from four to two, and the run fraction stays at 50 percent. Annual starts fall from 35,040 to 17,520.
The reason is that the box needs the same amount of heat removed either way. A wider band lets the temperature travel further before the control acts, so each leg takes proportionally longer, and the ratio between them — which is the run fraction — does not move. What you are buying with the halved start count is a temperature swing twice as wide, and whether that is acceptable is a question about the product, not about the compressor.
This holds while load and capacity are steady. Over a real day they are not — the morning delivery and the afternoon ambient both move them — so treat the projection as the shape of the trade rather than a forecast.
Run fraction is a capacity statement
If the capacity table says 12,000 BTU/h at the conditions on the day and the compressor ran half the time, the average load over the window was 6,000 BTU/h. That is the most useful thing a run time log gives you, because it is a load measurement that needed no instrumentation beyond a logger clipped to the contactor.
It also gives you a sizing statement in the other direction: the same load fitted into 75 percent run time needs 8,000 BTU/h of equipment. Refrigeration is normally sized on a run time well under 24 hours so there is room for defrost and for a bad day, and this is where that number comes from.
One day is not a sample
A Tuesday in October and a Saturday in July are different boxes. The annual start count on this page assumes the window repeats, which it does not; take a week if the logger will hold it, and take it in the season that matters.
Questions people ask
How many compressor starts an hour is too many?
That is set by the compressor data sheet for the model in question, which is why it is a field on this page rather than a number the page supplies. The page counts your starts, works out the rate, and restates it beside whatever figure you entered from the data sheet without passing judgement on it.
Does widening the differential save energy?
Not directly — the run fraction does not change, so the compressor runs the same total minutes and uses the same energy. What it changes is the start count. On the defaults, going from 2 °F to 4 °F halves 35,040 starts a year to 17,520, at the cost of a temperature swing twice as wide in the box.
How do I work out the load from a run time log?
Multiply the capacity from the manufacturer table at the day conditions by the run fraction. With 12,000 BTU/h of capacity and 50 percent run time, the average load over the window was 6,000 BTU/h. It is the cheapest load measurement there is.
What run time should a refrigeration system have?
Systems are normally selected so the daily load fits inside a run time well short of 24 hours, leaving room for defrost cycles and for a hot day with the door open. The page shows what capacity a given run time target implies from your measured load; what target to pick is an engineering choice about that installation.
Why does my average off time not match the anti-short-cycle delay?
If the average off time is longer than the delay, the delay is not what is setting the rate — the box is simply taking that long to warm to the cut-in. If it is shorter, the delay is holding the compressor out and the control is asking for it sooner than the timer allows. The page says which of the two applies to your log.