Cura 3.0 Time Estimate Not Accurate

[CovertLemon 1]

I've been using cura with an Ultimaker 3 for nearly a year now and something I've noticed is that the time estimates aren't always very accurate. It may be that they never have been and I've only noticed them being wrong occasionally.

A recent print I set to go was predicted at 2hr 10mins, but ended up taking an extra 30mins, while another print was predicted 30mins and took 40mins instead.

I've been looking for some sort of answer on other forums but while some say they don't have the issue, others say they do, but most of these people are not using an Ultimaker 3D printer. I do use an edited version of the 'Fast' print profile which reduces print times mainly by decreasing the top and bottom layer heights, but by also increasing most of the speed settings. I don't know whether this affects the estimate, but I'd assume that these settings should be taken into account when calculating the print time.

Wondering if anyone else is having this issue, and if someone has managed to fix it?

Thanks in advance n_n

Edited November 9, 2017 by Guest

[CovertLemon 1]

*Edit*

I've been monitoring my prints to see how the print time changes with the percentage completion, and for the first time I've actually seen the print time go down... It's a longer print which contains 2 parts instead of just one, and it hasn't changed as much as the other prints have. Nothing else has been changed except for which nozzle I'm using.

[Marco_TvM 34]

Basically, you're comparing the estimate from Cura with the estimate from the UM3 printer? Am I correct?

The printer will adjust the estimated time based on 'real time' measurements.

So the mileage may vary.

Sometimes layers will go faster which will reduce the total printing time, then layers will take more time, adding to the total printing time.

That may explain the differences you've seen.

[CovertLemon 1]

I was actually comparing the 'printing time' on cura it self, it's been changing as the print goes along, but I'd assume they are effectively the same thing?

ah okay, that makes sense. So is there no way to increase the accuracy of the initial estimate? I'm keeping track of how long prints take and how much material they use as the printer is used for work based projects, so knowing that the data is likely not correct is rather annoying.

[kmanstudios 1,120]

I have always found my time estimates to be within an acceptable range and a marked improvement from what I had researched before getting my printer in January.

But, I may be a bit more lenient than others because I am used to dealing with rendering estimates that can really vary wildly based on what Marco_TvM said. A render could speed along and hit areas with a lot of reflections/refractions and just bog down and throw everything off.

Over the last year, I have developed a feel for the time approximate and allow for a bit of slop in either direction.

[Marco_TvM 34]

I was actually comparing the 'printing time' on Cura itself, it's been changing as the print goes along, but I'd assume they are effectively the same thing?

Unless @msuurmond corrects me, I think you are right. Cura should be reporting the remaining estimate time with the printer.

[Msuurmond 28]

We are investigating how accurate the estimates are and see if we can improve these. Please be aware that as they are 'estimates' this is not high on our priority list.

[CovertLemon 1]

We are investigating how accurate the estimates are and see if we can improve these. Please be aware that as they are 'estimates' this is not high on our priority list.

Yes I understand, I was just wondering if there was a way to increase the accuracy as I sometimes get error margins of 20-30%. So trying to set up multiple prints through the day would have been a bit easier

[kmanstudios 1,120]

We are investigating how accurate the estimates are and see if we can improve these. Please be aware that as they are 'estimates' this is not high on our priority list.

Yes I understand, I was just wondering if there was a way to increase the accuracy as I sometimes get error margins of 20-30%. So trying to set up multiple prints through the day would have been a bit easier

Wow! That is really off. I have never had anything that far off. Seriously, on a several day print, it may be off by 30 minutes. On an hours long print it may be off by only a few minutes.

I wonder why you would be getting such a wide spread of estimation:result......

[CovertLemon 1]

Wow! That is really off. I have never had anything that far off. Seriously, on a several day print, it may be off by 30 minutes. On an hours long print it may be off by only a few minutes.

I wonder why you would be getting such a wide spread of estimation:result......

That's what was confusing me, and I don't always get estimates that are that bad, others are within a few minutes when the print is around 2-3 hours. I was wondering if it is possibly down to what action the printer is currently taking. For example, it may be that when it's putting down support there's a big margin of error, so prints with more support vary more form the orginal estimate.

[kmanstudios 1,120]

Wow! That is really off. I have never had anything that far off. Seriously, on a several day print, it may be off by 30 minutes. On an hours long print it may be off by only a few minutes.

I wonder why you would be getting such a wide spread of estimation:result......

That's what was confusing me, and I don't always get estimates that are that bad, others are within a few minutes when the print is around 2-3 hours. I was wondering if it is possibly down to what action the printer is currently taking. For example, it may be that when it's putting down support there's a big margin of error, so prints with more support vary more form the orginal estimate.

I am not so sure that supports have anything to do with it as I use a boatload of PVA on my prints.....many have thin and/or spindly parts. And it runs consistently between those prints and non support prints.

Edited November 10, 2017 by Guest

[tomnagel 126]

For most of the prints, Cura estimate is within a few % of the true printing time. But there seem to be cases where the difference is really large. @CovertLemon maybe you can provide us with a real world example, or maybe even more?

An STL and your gcode would really help.

[DPI_1999 0]

Hi there,

most of our prints have been within 1-2% of the estimated time in Cura, except one. It was estimated at 12 hours and 51 minutes but is still printing currently and has been doing so for more than 22 hours... no issues with the output so far but just taking much longer than anticipated. I have gcode and STL if you would like me to upload it somewhere, in case it will help your investigation.

[gMaker 0]

I am also struggling with the estimated times. I would love to know how to improve them.

I used this model: https://www.tinkercad.com/things/g2wf4flqvjs-anycubic-i3-mega-dial-indicator-mount

Which resulted in this G-code: https://gist.github.com/brnl/af0fc099c0c81589a79fd732e9f4bf9d

My printer is an Anycubic i3 Mega, and I have added it as a "Prusa i3", Gcode flavor 'RepRap'.

Hope this helps, and I hope you will be able to help me :-)

[cgottlieb 0]

I am also having huge differences between the time shown in Cura and the actual print time on the Ultimaker 2+.  For example, I am printing something about the size of a luggage tag, and Cura shows it will take 3 1/2 hours.  Yesterday, it took over 19 hours to print.  I reset the Ultimaker 2+ to the factory default, and it is still projecting 11 hours, nearly 3x the estimated time.  My laptop was recently reimaged, and the latest version of Cura was installed.  I'm using the recommended settings in Cura.  Any advice?

[ahoeben 1,948]

Did it actually take 19 hours, or did the display just show 19 hours from the start of the print? The first layer is typically printing much slower than the rest of the print. The printer figures "if the first few commands of the file are taking this long, then the whole print is going to take a looong time to print". In reality, the following layers will print much faster and the estimate shown on the printer will come down. I think. But I never printed from SD on my UM2.

[cgottlieb 0]

I'm not sure it took 19 hours because it finished after I left for the day.  But the current print has been running for about 2 hours and now the estimated time is 22 hours left.  It does seem to be printing pretty slowly.

 

Update - the print did finish pretty close to the time shown in Cura.  The Ultimaker estimate is just way off - ranged from 11 to 22 hours as the print was running.  I guess I will just ignore the Ultimaker estimates.

Edited March 15, 2018 by cgottlieb

[kmanstudios 1,120]

With some of the weird stuff I am printing now, my time estimates have been a bit off. More than usual. But, I am really telling the printer to do some really wacky things, with strange areas that have great slowdowns and speed ups.

 

So, my times lately have been off in a net effect of Positive and negative. But, on a print that is taking 6-7 days to print, a time differential of 3-6 hours is not all that bad percentage wise.

 

Edit: This would be an example of radically different times over areas: At the bottom when it is mostly solid it moves a lot faster. But as it gets higher, and the spires and spirals take over, it does take longer than the bottom and it just gets more and more dispersed with individual spires and such as it goes up.

Edited March 15, 2018 by kmanstudios

[Brulti 177]

@kmanstudios I'm always fascinated with your fractal sculptures...

 

I've been doing quite a lot of small prints lately, just serviced the printer and running the test prints, and the estimate Cura is giving me have been spot on down to the minute.

Did a couple huge prints last week while I was away (dragon head my partner found somewhere online) and the estimate for top part was 6 days and some hours, 3 days for lower jaw, and, as far as I can tell, it was spot on as well.

[CovertLemon 1]

On 21/11/2017 at 9:03 PM, tomnagel said:

For most of the prints, Cura estimate is within a few % of the true printing time. But there seem to be cases where the difference is really large. @CovertLemon maybe you can provide us with a real world example, or maybe even more?

An STL and your gcode would really help.

 

Sorry I didn't add any files before, I completely forgot about the post due to other problems occuring!

 

For a while this issue stopped occuring, or at least was reduced to the point that I no longer noticed it, but I've just had quite a bad estimate from cura. The original estimate from cura was 35 mins, the estimate during the print sat pretty consistently around 1hr 20mins, and the actual print has taken around an hour...

 

During the print, the estimated time left would increase for a while before suddenly dropping by a few minutes.

 

I've attached the Gcode, STL file and the profile I used to see if anyone can spot the issue.

Prototype Print R.curaprofile

Pi Pack Lift Stopper v2.STL

UM3E_Tray stopper v2.gcode

[Tomhe 21]

The STL file and the gcode are not for the same model, but I do see that both models have high resolutions in the rounded corners.

The curaprofile file is not really clear, as you have made your own materials (material = empty_material). In dual extrusion we spend a lot of time making the heatup/cooldown of the idle nozzle happen while the other nozzle was printing. A small change (in standby temperature, heatup speed, cooldown speed, etc) could disrupt this. That would cause the printer to wait between layers, untill the nozzle is heated. This waiting time is (of course) not included in the Cura estimation, as it assumes that the values are tuned so it doesn't have to wait.

 

Back to the rounded corners: using the Cura 3.3 beta and the latest firmware, the estimation may be a lot better (less slowdown on high resolution parts).

 

[CovertLemon 1]

28 minutes ago, Tomhe said:

The STL file and the gcode are not for the same model, but I do see that both models have high resolutions in the rounded corners.

The curaprofile file is not really clear, as you have made your own materials (material = empty_material). In dual extrusion we spend a lot of time making the heatup/cooldown of the idle nozzle happen while the other nozzle was printing. A small change (in standby temperature, heatup speed, cooldown speed, etc) could disrupt this. That would cause the printer to wait between layers, untill the nozzle is heated. This waiting time is (of course) not included in the Cura estimation, as it assumes that the values are tuned so it doesn't have to wait.

 

Back to the rounded corners: using the Cura 3.3 beta and the latest firmware, the estimation may be a lot better (less slowdown on high resolution parts).

 

 

Ah my mistake, I'll add the correct file to this. Not sure why the material has changed as I use Ultimakers default material, all I've done is add a price to it to see how much each print costs.

 

As far as I can see the printer does not stop at all between layers, so if i've understood you correctly, what you have suggested shouldn't cause much, if any, change to the print time.

UM3_Pi Pack Lift Stopper v2.gcode

Edited April 12, 2018 by CovertLemon