Why two planners give different runtimes

The short answer

Most technical planners, Dive Kit included, run the same published model: Bühlmann ZH-L16C with gradient factors. Give two of them the same dive and the same settings and their schedules land within a few minutes of each other. So when you see a gap of 20, 40, or 60 minutes between two apps, the model is almost never the reason. One of the two plans describes a different dive.

Before you distrust either engine, line up the inputs. In our experience every large gap reported so far has traced back to one field that differed between the apps.

The checklist: what has to match

Work down this list in both apps. The items at the top move a schedule the most.

Input	Why it moves the schedule
Gas composition, including helium	The single biggest lever. On open circuit that means every cylinder’s O₂ and He. On CCR it means the diluent: at a fixed setpoint the loop’s inert gas follows the diluent, so a 10/70 diluent and a zero-helium EAN10 diluent produce very different schedules at the same depth and the same setpoint.
Gradient factors	GF Low sets the first stop, GF High sets how fast the shallow stops clear. 40/90 and 50/80 are different plans.
Setpoints (CCR)	The bottom and deco setpoints set how much inert gas the loop carries in each phase.
Last stop depth	A 6 m last stop and a 3 m last stop redistribute the whole shallow schedule.
Bottom time convention	Some planners count the descent inside the bottom time, some start the clock on arrival. Dive Kit lets you pick (Include travel in level time). A 40 min level with a 5 min descent is a 35 min or a 40 min exposure depending on this switch.
Ascent and descent rates	The deep ascent rate decides where the first stop lands. Planners also differ on shallow-rate defaults.
Water type and altitude	Salt, fresh, and altitude change the depth-to-pressure conversion.
Stop rounding	Whole-minute stops versus exact stop lengths add up across a long ladder.

Numbers like CNS also depend on conventions. Dive Kit tracks CNS against ambient PPO₂, which reads higher than planners that track the inspired value. See How the engine works for that distinction.

A real case: 42 minutes from one gas field

A CCR diver planned 90 m for 40 min in Dive Kit and in another planning app. Dive Kit said runtime 296. The other app said 244. That looks like a broken engine until you read the other app’s own output: its bottom rows showed END/EAD 101/101 and its stop rows showed nitrox loop equivalents (EAN30, EAN75, EAN92). Those numbers only appear with a zero-helium diluent. The plan had been entered with EAN10 as diluent instead of trimix 10/70.

We re-ran both configurations through Dive Kit’s engine:

• The diver’s exact plan, 10/70 diluent: runtime 296.
• The same plan with only the diluent swapped to EAN10: runtime 257. One field, 39 minutes.
• Every input matched to the other app (EAN10, constant setpoint 1.2, same rates, 3 m last stop): runtime 250 against the other app’s 244, with the stop ladder matching stop for stop. First stop at 57 m in both. At 15 m, 14.3 min against 14. At 6 m, 34.3 against 34. At 3 m, 50.2 against 52.

So the two engines agree within about 2.5 percent once they plan the same dive. The 42-minute gap was helium: Bühlmann’s fast helium compartments load hard during a 40 min bottom on a 70 percent He loop, and that gas demands a longer ascent than a nitrogen-only loop at the same setpoint.

The EAN10 plan also fails a sanity check. At 90 m that loop gas would sit near 12.7 g/L density with a 101 m equivalent narcotic depth. No one dives that. If a shorter runtime comes with warnings like these, the shorter runtime is describing an impossible dive.

If the inputs match and the gap remains

Small differences survive a perfect input match: stop rounding, how travel time is booked, where each engine decides a stop has cleared. Expect a few minutes across a long schedule, usually spread over the shallow stops. We track these engine-level differences openly against MultiDeco across 26 scenarios; How Dive Kit compares to MultiDeco lists where they agree and the known differences, with reasons.

A gap bigger than a few percent after you have matched every row of the checklist is worth reporting. Share the plan link from both apps if you can. The share link embeds every input, which is exactly what made the case above solvable.