I know that the actual printing time is longer than the time estimated by Cura, due to acceleration and jerk factor.
However, I have printed a small pyramid for which Cura estimates 4.0 minutes, while the measured printing time is 2 minutes and 40 seconds. What reasons could produce this difference?
For clarity, I'm using a Delta Wasp 20 40 equipped with a clay extruder for which parameters such as E-step (for E axis) are unpredictable since I have no filament.
If you properly define your own machine with a delta_wasp.def.json file you can fill in the acceleration and jerk settings of your printer, so that Cura will use the correct values for print time estimation.
For example, take a look at how the Ultimaker 2 is defined.
Exposing these settings to the Custom FDM Printer wizard hasn't been implemented (yet).
As you note, the estimates are usually optimistic because they're calculated without knowing the printer's actual acceleration as well as communication overhead time losses, etc. However, many estimation algorithms put in a bit of a fudge factor to approximate a typical printer or to approximate startup time. I think in this case, you're just on the wrong side of the fudge factor.
Estimating time for any CNC based machines are measured in this formula:
The length of pulses that machine travels x The feedrate of the pulse itself
It gives you the time for whole movements. If you're familiar with NC codes, information of any movement is listed in a single line, having its own feedrate, that's what we call pulses.
As machines are not ideal, some delay occurs between pulses, approximately 1 or 2 ms according to machine type. That's what is not measured in the formula.
If you were able to measure that delay time and multiply it in number of pulses, by adding it in the estimated time found by machine, you could find a much more definite estimated time, for your process.
