Ultimaker for Architecture project

[a-max 0]

Hello guys,

I'm an architecture student and currently I'm thinking about buying an Ultimaker to help me create a physical model of my building concept, that's part of my graduation project due in September. However since it's quite pricey for a student, 5 months salary, i'm not sure if it's the right tool for the job.

That being said, I've installed Cura to get an estimate regarding the material per module as well as time needed to print each module (total 41modules- with a total size of 2.1m x 1m x 12cm).Initially it was like almost 3 weeks to print all modules (5hours for each module, 2 modules/day) and 8 roles of PLA.

This proved to be quite time consuming and costly, however I've played with the settings and would like to know if they are correct and that for an architecture print these settings would come up with decent results, I'm confortable of doing some post production using a dremel afterwards.

 

[illuminarti 18]

Hi, and welcome!

While I won't say it's impossible to get good results printing the sort of pieces that you show here with an Ultimaker, it's going to take you quite a lot of practice to get good results. The main challenge is going to be supporting both the curves and the straight line parts. The printer cannot lay down plastic over thin air, so there needs to be proper support under the curves. You also need to make sure that the piece stays in place while it is printed, so you need good contact with the print bed. For both those reasons, almost all the parts that you show oriented vertically in your screen shots would need to be laid down flat on the bed.

The speed estimates in Cura aren't particularly accurate at the moment - they assume constant speed moves (whereas the head in fact accelerates and decelerates on each move), and in some cases get it flat out wrong. 100mm/s is rather fast for good results, especially with a 0.2mm layer height - and you will need to configure the printer to allow adequate time for cooling between each layer.

I'm inclined to think that other print technologies such as stereo-lithography might be better suited to these sorts of complicated, detailed geometries - but those are going to be even more expensive, and harder to get started with.

It's certainly not going to be a simple click-and-print experience with the Ultimaker - it will take you a lot of trial and error to do this right, and if you're going to do it, I recommend you get started with the learning curve as soon as possible. Even if not buying a printer immediately, are there perhaps local hackerspaces or labs where you could get some time with a printer to experiment? Whereabouts are you located?

If you would care to send me a couple of the STL files, I'd be happy to look at them in more detail, and maybe try a test print to get a sense of what's possible.

 

[a-max 0]

Thank you Illuminarti for the ultra fast & detailed response!

I've been looking around the forums for quite a lot of time recently and I've been impressed with the quality the printer can achieve, that's the main thing why I wanted to make a model for my graduation project using an Ultimaker. I've also looked at stereo-lithography, yes it's true it might be better due to high resolution, but I've heard the materials are quite expensive, it involves a messy post production and they are extremely slow, like 15mm/hour and I'm also based in Romania, Europe and the cost + import duties are currently way out of my budget since I'm just finishing college

The reason why I thought it would be great for architecture is that I kept seeing either colorfabbs models or other yoda's, ultimaker robots, which seam quite more complex and have more details then architecture models, and they seamed to look good even at 0.2mm. Unfortunately there aren't any hackerspaces or labs where I'm at, only companies that sell this service and I think they use powder based 3d printers and the cost to print my architecture model was almost the same price as buying the Ultimaker, so it's quite expensive for a model that you will showcase once.

I've placed the models vertically since I've read somewhere that 3d printers have better resolution on the Z axis and thus providing a smoother finish then if layed flat + I think that it will require more material if layed flat due to it needing to build support and Cura seamed to show the need of little support when I placed them like this.

Also I`m currently a little behind schedule, so your help in looking over some STL files is extremely helpful since hopefully I will know how much material/time will be needed to print since the model neads to be done by the end of August the latest so don't know if that's enough time to get it done and also as you mentioned to get familiar with the workflow/tweaking/trial & errors + building it.

Ohh and one more thing... in my CAD software if I set my walls to be 0.6 mm width then in Cura if I set a infill rate of 20% does that mean that my walls will have a 0.12 fill then in the middle they will be "hollow/empty" and then another 0.12mm shell for the other part of the wall? so if i want actual 0.6mm walls the infill should be 100%?

You also mentioned getting speeds of 100mm would require allowing adequate time for layer cooling, does this involve installing a secondary fan or will the single one be enough?

STL FILES:

https://www.dropbox.com/sh/30v5mdxvhjubjg6/TPHDQPTcH4

Thank you again for the reply!

 

[a-max 0]

PS: The STL files need to be scaled to 1.5 in all directions in order to reflect the correct size since there's a bug in Revit ( CAD software ) that it exports STL files in the wrong units.

 

[illuminarti 18]

I'll take a look at the STL's and get back to you soon.

Regarding the wall thickness... the setting in Cura has recently been renamed 'Skin Thickness' as this gives a better idea of what it means. Basically, the skin thickness needs to be some multiple of the nozzle size (this isn't quite true, but its a good approximation to begin with).

So, if you're nozzle is 0.4mm, and you set a skin thickness = 0.4mm, then you will end up with a single perimeter pass around the outside of your object. If you set skin = 0.8mm then you will get two passes around the outside of each STL object. Any space left inside the object then gets filled with infill, at whatever percentage you specify. 100% fills it in totally, to make a solid object. 0% leaves it hollow. Values in between draw a mesh.

This determines the minimum size of object you can print. Cura currently will not do a single pass to print and fill an object (e.g., to do a wall only 0.4mm thick). It always tries to draw all the way around the outside of the shape. A wall that is 0.8mm will get a pass along each side; slightly thinner walls it will try to print, but the passes may overlap, and it will make a mess. As walls get thinner and thinner, Cura will give up, and not even try to print parts of STL's that are too thin.

It would be best to design your objects to not need something with a cross-section thinner than 0.8mm, but you can cheat a little bit by declaring your nozzle size to be thinner than it really is. Within reason this will work (although actually having a smaller nozzle would be best). So if you specify a 0.3mm nozzle, and 0.6mm skin thickness, then it should try to print your 0.6mm wall with two 'fake' 0.3mm passes (i.e., it will extrude as much plastic as if the nozzle opening was 0.3mm, and you have to hope it all ends up in the center to make a thinner bead).

 

[illuminarti 18]

To answer some of your other questions - layer cooling time is a parameter built into Cura. In general you don't want to be laying a new layer of plastic on top of the previous one until the first one has had 3 to 10 seconds to cool, depending on the thickness of the layers, and the geometry of the object. This is typically the biggest constraint on print speed when printing fairly small objects. Even modest linear speeds of 50 to 75mm/s can result in layer times that are too quick, so Cura slows them down. This extends the necessary print time (although there is a big in the current version of Cura that actually causes it to report faster, rather than slower, print times in this case). More/better fans might reduce the amount of time needed for cooling, but the built-in fan isn't too bad.

Regarding orientation, in practice the resolution in x/y/z is rarely a consideration. While the absolute resolution in z is higher, this is balanced by the fact that in X and Y you can extrude a continuous bead of plastic, whereas in Z you're limited to discrete layers, and the ability of the head to move in X and Y on those layers - and also the difficulties of extruding overhanging layers that aren't supported underneath. So mostly you can't really resolve fine detail at anything like the layer height. The benefit of thinner z-layers is mostly in representing curves more smoothly.

Against those considerations, more practical limits are that you cannot typically print unsupported horizontal beams of plastic that extrude more than a mm or less beyond the print on the lower level. You can add support to help with this, but support is typically hard to remove, and tends to damage the finished surface. So you need to orient your part to minimize the amount of support needed, and to make sure that where there is support, it avoids obvious finished surfaces if possible.

Finally, you need to avoid creating top-heavy parts that will vibrate loose, or fall over as they get taller, and the bead of freshly extruded plastic tugs on them as the head moves on the higher layers.

Like I said, there are a lot of subtle effects that need to be taken into account - there's really no substitute for lots of practice.

 

[gr5 2,172]

I'm not sure why you need to use so many spools of plastic. Can't you just print it another 1/2X scale? If you do it will use 1/4 as much filament and print 4X faster.

The cooling time is the least of your problems. It's only a problem on very very tiny layers - for example the tip of a point on the top of something and only slows you down for those layers. 5 seconds per layer is plenty slow enough.

Printing the roofs on their side like you showed will definitely improve quality but if the roof falls over due to not enough adhesion at the base will cause the print to fail completely.

Actually being able to see the layers in the objects is a fun, unique feature that reminds people that the parts were made with a 3D printer. Rather than thinking of these thin lines as ugly, think of them as wonderful.

Getting an ultimaker working nicely is a ton of work and may take you 20-40 hours even with help from this forum. You should factor this into your time calculations.

 

[a-max 0]

Thank you for the in depth info, you helped shed some light regarding 3D printing and indeed as you mentioned it sounds like it involves a lot of practice to get a hang of what the best settings are and now I see some of the challenges that my model may face when 3d printed.

For now I'll wait for you to check the STL files in order to see your opinions regarding if those can be 3D printed, as well as how fast and how much material will be used.Since these are kinda the key things that are keeping me from taking a decision and since each day matters, I would like to finally have my mind set on what I will be building my physical model of.

Also if you somehow manage to do a test print, that would be extremely helpful since I could see first hand how it will look like.

Thank you for the tremendous support!

 

[a-max 0]

Thanks for the info gr5, unfortunately I can't print it smaller since at the moment at 1:200 of it's actual scale the height at the ends are around 7.5cm and making it half that scale it will result in a model that's only around 3.75cm in height, which is way too small for an architecture model.

Personally I don't have a problem with seeing thin lines,as you said they are wonderful, well only if they aren't really really visible like 1-2 mm in thickness. Also the 3d printed lines can be hidden by sanding and laying a coat of primer + painting over with an airbrush, this way it would make it look smooth & glossy.

Actually this was an option I was actually thinking off, in case I'm short of time, like get the 3D parts printed as fast as possible while keeping the rough desired shape, then use a dremel to fix possible print errors + sand it + primer + paint... but I don't know the speed to quality ratio of the parts in order to avoid ending up with strings all over the place or even worse, the structure collapsing while printing, I assume these are found during test prints, end this technique would be somehow similar to the steps while finishing up a sculpture. Would like to hear some opinions regarding this method, since I've seen 3d prints get polished, fixed, painted, here on the forum.

Regarding the time investment, yes 40 hours seams quite a lot but I was actually planning on printing on August 1st, taking into consideration shipping time + installation + calibrating + other errors. But for now I'm interested more on some feedback from the community regarding how fast can I print them out to get a relative good quality print that may or may not require some heavy post production, more exactly rough estimates regarding what Cura settings you recommend to get a good balance between speed/quality/cost.

 

[ian 32]

what scale are you printing this ?

 

[a-max 0]

The scale of the building model is 1:200 (metric system),resulting in a building model that would be around 2.1m x 1m with a height that varies from 7.5cm to around 12cm. However the STL files as mentioned are smaller due to a bug in the cad software, that's why they need to be scaled to 1.5 to reflect the proper scale.

 

[ian 32]

split your model up into as many small clean components as possible, use cura software to place a few at a time on the print table. think clever and place the parts lieing flat and not stadning up.. gravity is your enemy when printing and dont give it a chance to screw up your openings...

then just keep tight belts on your ultimaker !!!!!!! and use sowing maschine oil on all the bars to keeps things slippy and nice....

and... print like a lunatic... it is possible, it will be simply stunning when finished, you will just probably kick two or three cats out the window with frustration before its all finished...

good luck my friend !! :-)

ill try and help you when i can.. i do this as a business...

Ian

 

[gr5 2,172]

Just to give you some ideas...

Since this is so large I would definitely print .2mm layers.

Looking at the first image you posted, I would print these all seperately. The large part closest to the "camera" can be printed at high temperature to allow the plastic to flow fast (like honey versus low temperature the plastic is like toothpaste). I would print it at 240C and 150mm/sec at .2mm layer. Cura can tell you how long that will take to print.

The curved roof I would lay flat with support. You might be able to print it as shown - you would need to add a brim. Or you could use cad to add some coin shaped pads in 3 or 5 places along the bottom edge and then remove them afterwards. Again you can print at 150mm/sec at 240C. I know you don't want to waste plastic on support but you can do sparse support. It's worth it. Support makes for a bit of an ugly surface but that surface will be mostly hidden to eyes on the final model. You could create the support with CAD and it will take very little extra PLA.

The open ladder shapped parts (remaining parts) will be very difficult. The cross bars are printed in the air and will droop and not look as good. You might be able to print them as is but you would have to print slower and cooler (50-75mm/sec, 190C maybe). Some types of PLA will get you lots of stringing on these parts. Not so hard to clean up though.

So the open ladder's I think also need to be printed flat on the bed so that nothing is printed "in the air". You shouldn't use support with these as it's too much support for too little gain and removing the support might break some of the cross pieces.

These "ladder's" are by far the hardest thing to print if you can't lay them flat. Too much air.

 

[a-max 0]

@Ian @gr5 Thank you guys for the tips, I really appreciate all the advice and effort put into replying. I didn't think that the hardest part would be the windows (aka ladder shaped parts), I actually thought the roof would be the thing that will give me the headaches.And the tips to actually break them into smaller parts, like the Y shaped collumns, then the windows, walls, is quite an excellent idea, haven't thought at breaking it into so many smaller objects, but I see the advantages of this option and I can now see clearly how it can be approached for 3D printing in order to not waste too much material on support and thus also resulting in smaller print times. Luckly I can break the model into as many smaller parts as needed as well as make the elements smaller/bigger when needed due to being the one who made them, thus having the flexibility to do so.

The print speed settings & temperatures are absolutely brilliant, since I'm not familiar with how those work, and at least I'll have an estimate where to start at. Also the PLA I was planning to buy is actually from Colorfabb, the bluish white, since it looks like the print quality is really high end and I've read that it kinda keeps stringing away from happening.

Also....I've checked the model in Netfabb and can't seam to fix the errors and I can't really fix them in my CAD spftware, although when viewed in CURA (sliced mode) I don't see anything "troublesome".

If you also look at the roof, in the 3D view (first image ) , you will notice a lot of squares, each of those is a 3D "box" and there's a small gap between them. When viewed in normal mode Cura doesn't show the gap, but when viewed in sliced mode they are clearly visible and don't know if they will stay welded or will they fall although there's a structure beneath them that should hold them together, but I was planning on removing that structure in some areas to save material.

All that being said I will post in the "please print for me" section of this forum and hopefully someone will be able to print a small portion of them too see how they look and hopefully by Friday I will be placing my Ultimaker order.

 

[gr5 2,172]

>Cura doesn't show the gap, but when viewed in sliced mode they are clearly visible

 

This is a problem and easily fixed. By default, everytime the slicer passes through a wall it switches form "solid" to "air" and back again. However there are settings in cura... In the "expert" settings under "fix horrible", play with those 4 checkboxes. Unfortunately there are 16 possible combinations but fortunately cura slices very fast. Usually I want the first, second or last checkbox only which will help me fill in "holes" in a model that I didn't want.

 

STL files have a "normal vector" for each polygon which tells the slicer which side is solid and which side is hollow. Cura uses this. Sometimes. I think. Not sure how it all works. I think that's one of the 4 checkboxes.

 

>it kinda keeps stringing away from happening

 

I haven't bought from colorfabb but they have a fantastic reputation - you don't want to get cheap PLA as if it is slightly too large for the tube it will get stuck and other issues which can force you to throw it all away. Since you say you don't mind fixing the strings it might not matter but for many people and many models it is an issue. It's very nice to have a PLA that doesn't string at high temperatures so you can print fast. For example my "white" pla from printbl.com prints with strings no matter what I do but the "grape" pla from printbl.com rarely prints strings even at higher temperatures (usually lower temperatures are less likely to string but you have to also print slower because the pressure in the nozzle is high). So even though I haven't used "colorfabb bluish white" it sounds like a good choice to me.

 

I wouldn't worry too much about stringing though. Not until you own a printer. That is a minor problem.

 

If you do buy a printer, show photos of your problems and you will get lots of help.