[G-Spot]

I've started to build a machine I plan to use to print aluminium structures. It will use a welder and feeder to print Aluminium. Then after the Aluminium is cold it will mill away any errors/excess the welder has made.

It's going to be used to print aluminium structures 2.2m x 2.2 x 15m

I think I am going to need a lot of help especially on selection of components.

Here is the design criteria;

1. Accuracy only needs to be about 1mm but over the full 15m.

2. It could take a year to print each structure. So weeks to setup is acceptable.

3. It will be OUTSIDE because the y axis is going up 15m vertically. The Y axis HAS to go up vertically because molten metal and gravity.

5. I will build the y axis as it prints. Like the top of a crane or building form work climbs the structure it is making.

6. I will set 4 aluminium posts into the ground with helical gear rails attached to the sides. This will be the y axis which the rest of the printer will climb, I will weld the structure that is being printed to the posts as the height increases. I am printing ridiculously rigid structures which will be far more rigid than the CNC printer itself, so the height will not be a problem as the structure being built is welded at intervals to the climbing posts. Even when the gantry is at full 15m height it will only be a meter from being physically attached to the structure being printed.

7. There will be no physical machining contact between the welding head and the object being printed. The only physical machining contact with the printer will be the spindle used for milling what the welder laid down.

8. The milling spindle is only there to follow the welder around and clean up any excess, shaving off a few millimeters. No deep cutting/milling required. Same grade of aluminium every time behind the welder.

9. Both the welding/printing head and milling spindle will be located close to each other. It will either be printing or milling never both together.

10. I plan to use AI and a camera to make measurements from a reference(not on the machine) and adjust the welding/milling as required.

11. If the machine distorts/bends/vibrates while in operation then I will cut it, straighten it then reweld it.


From my design requirements I believe the smallest motor spindle is all I require. Am I wrong? missing anything? The welder has to be water cooled so cooling water is going to be on the gantry may as well get a water cooled one and avoid the noise. If am only going to be removing a few mm from the just printed aluminium then a 400W motor should do? Yes/No? Is the quality of the aluminium finish going to be reliant on the quality of the motor? Will a good cutting tool and small cuts give me as good a finish as a high quality motor taking chunks off?

I also decided to use Helical gear rails throughout for linear movement because of the long lengths involved and the possibility to add extra rail when needed. Are there any better solutions, cheaper? I can just afford the helical gears solution so unless there is a better and cheaper option then I think I am just going to bite the bullet and go for helical gears.

I am trying to find the cheapest rail solution. I think linear bearings are too expensive(15m x 2,5m x 2,5m) and the accuracy way more than I actually need. I am thinking of using the aluminum beams that I am going to build my gantry on as the rails and use regular bearings as wheels. Will I have problems with the helical gears if the gantry is not on linear bearings? I see this is how most cheap 3D printers do it, but they use belts for motion and not helical gears. I never bothered considering a belt setup because my belt lengths would be 10m long. I could not find anyone that was using these type of belt lengths for CNC or 3D printing. And the 25mm belt cost $12 per meter. Can belts be used successfully at those sort of lengths? Saw a Youtube video where the stretch on a 500mm belt was visible.

Anyway I am going to visit the scrap yards and Aluminium suppliers this week and find 12.5m of the biggest aluminium box sections I can afford and weld them up into a 2.5m x 2.5m square for my frame and then another 2.5m beam that will traverse them. First build my x and z axis without servos, connect a cheap motor and measure how much vibration and deflection I get when taking off a few mm of aluminium weld, then strengthen/dampen as required. Weigh the beam I end up with and then get the servos/motors to shift that.

I would be very grateful for any information/opinions on rails and motion solutions that would suit my application.

[awerby]

Have you made a picture of this thing? I'm trying to visualize it - especially the part about it building its own Y axis as it goes - but not having much success.

I'd suggest using steel beams rather than aluminum ones. Steel is much stiffer, and it's also cheaper. I have seen 3D printers based on MIG welders, so that's not hard to picture, but I'm not seeing how this could possibly be rigid enough over that huge span to carve away at the weldment without chattering madly, especially since you're using wheels on your beams instead of real linear rails.

[joeavaerage]

Hi,
great sounding project.

I agree with Andrew, use steel in the construction, it is way cheaper to build a structure of given rigidity out of steel than aluminum.

To be honest I think a 400w spindle is a bit of a joke. This project will cost many thousands, what matter if you up the ante in the spindle power to 2.2kW, they are only a few hundred
dollars?.

I am rather dubious that the bearing arrangement you have described will be sufficiently accurate to ensure optimal mesh of the helical gear and pinion. I would have thought that the mesh
needs be +-0.02mm or there abouts....and that would probably exceed the bearing idea.

The only other idea to a rack and pinion is a rotating nut ballscrew. They are used on very large machines and I suspect prohibitively expensive for your project....could be wrong of course.
Even transporting a 15m long ballscrew will give you grief.

Craig

[peteeng]

Hi G-Spot - Some feedback. 30 years ago I used to program welding robots and train people to robotically weld. One of the exercises I designed was to get the welder to write their name on a plate then start stacking it up and see how high they could make it. So I was 3D printing metal a long time ago At about 50mm high the signatures would get wobbly. That's because of the weld shrinking and generating lots of stress that pulls the stack around. So you have a lot to figure out. There are large scale welding 3D printers around. I'd do a lot of research and make a small a scale version first. I'd look at laser welders this is the best technology to get what you want to happen, happen.

Your Q's
10) using AI sounds great but again its not easy. As the weld progresses the structure moves. There are several camera based weld following/steering systems around so there are precedents, do some more research
11) good luck with that approach, It won't work. This sort of machine has to be correct at the get go. Again a small scale version will provide a lot of failures and learnings

400W spindles have very small shafts and fatigue very fast. Mine rarely got over 300 hrs before they snapped and that was majority timber. When I used them on aluminium they bent and once straightened they then failed. I would use an er20 or an ISO20 spindle 1.5kW minimum prefer 2.2kW. Milling AL weldment is a tough thing. Welds are soft and gummy you will need lubricant and high pressure air to keep the tool clean and sharp. I used to manually dress AL welds and even burrs gummed if you were not paying attention. If the machine head is a bit wobbly you will choke the tool real quick. Single flute tools are the way to go to help on this issue

In regard to the finish this comes down to the dynamic stiffness of the total machine. I have been researching this off and on for a few years. What you propose will have many vibration modes that will be easy to excite. You need to look at 3D printer drivers that have input shaping, That will be the best defense against this type of vibration. You are proposing a light weight structure that is in no way what a mill of this size would look like. They are massive for a reason. If you look at 15m high structures such as bleachers, scaffolding, temp buildings etc you will see they wave around in the breeze. You will have global and local vibration issue to solve.

Helical or straight gears? Hmmm helical will be difficult to get the correct mesh over such a length in such a structure. I'd use straight with a tensioner like basic R&P cnc's use. The spring loaded drive takes care of the intolerances. Rack comes in 1.4m lengths and you will need to align all of those somehow. Longer is available but costs a lot and transport is always an issue. They can arrive bent. Its easier to align straight gears then helical. By align I mean get the pitch across the gap correct you use a short piece of rack across the gap. You can't do that with helical due to the wind in the gear.

It is possible to use belts but I'd use a 50mm wide for the long axis. But I think straight R&P will be better for you.

Linear bearings are actual cheap. By the time you get straight tube, several bearings and then make carriers accurate enough your well past the cost of the rails and cars. I have tried and won't go there again its too painful. Plus home made goes out of tolerance regularly.

I would not spend any $$$ until you have sorted this on paper or CAD. I can see a lot of pain in this project. I have been involved in all aspects of your project and in each bit there are high hurdles to overcome. These compound exponentially when placed into the same machine. I'm not saying don't do it, but go small scale first... Even large companies and universities collaborating in this area have taken years & Mega $$$ to get something together that works.

Good Luck on your adventure and keep us posted... Peter

[G-Spot]

Have you made a picture of this thing?

I only know how to use ACAD 3D-Studio and it's the Johnny Depp version which I need to rehack onto my PC. So that will take at least a week to get working again.
I am keen as this is probably going to be a commercial project, eventually, if I am lucky..lol.., to use something a lot simpler and cheaper. What do you use?

Imagine a 4 post car lift. Going to attach a square frame which prints the x and z axis. Standard 3D printer design but moving the top of the printer up and down and not the bed.

I live within 50 miles of the biggest Aluminium smelter and mill in the country so I am not getting much difference between the price of Aluminium and steel for the same dimensions, like only 15% more for the Aluminium.

From my experience in sailing I am not convinced that a steel structure is more rigid than aluminium. In sailing the more rigid boat wins. If your mast or boat flexes you spill the wind. But there isn't a single modern racing boat with a steel mast, the trick is they use steel stays to brace the Aluminium mast. They use a little bit of steel where it works best, in tension, to make the aluminium structure more rigid than steel would be. Of course Carbon fiber is what is used now but before that came along it was all aluminium masts.

I also happen to be working on my own DIY construction project where I am making a pre-stressed concrete beam 20m long x 1.5m x 0.2m. I spent the last 2 years sinking 15 6m deep reinforced concrete piles into the ground with about 50 tons of concrete on top. I now have the ability to prestress large objects up to 20m in length to 100s of tons. Had to design and make a machine capable of digging 6m deep holes only 30cm x 30cm through stone or soil that could be operated by 1 person. Been there done that got the t-shirt. Only cost me $100 for the steel and 3 months from start to first hole finished. In case you are wondering why I didn't just hire a pile digger, there was a massive flood here 18 months back and for 3 months you could not hire any. Anyway my machine can go through stone and concrete while the ones you hire are augers and cant. Also my machine could do undercut piles which only piling machines costs hundreds of thousands can do which cost thousands to hire.

Was thinking of using large Aluminium box sections only 5mm thick. Fill them with concrete with dozens of high tensile steel rebars running down the length then pull them between my end posts leave for 30 days then cut the rebar away from my end posts, thread the ends and then use them for my machine.

Not sure why so many DIY builds go for the post stressing option on their concrete interiors. The amazing thing about prestressed concrete is after the cement is dry you no longer have to carry around the machine that did all the stressing. The energy it exerted is now locked in your cement. So if you now encase the prestressed concrete in your machine you are adding the energy stored inside to resist flexing to your machine. If you use the post stressing method then your machines tensile strength is being used to stress the concrete. And because the rods running through your concrete are not attached to the cement they provide a pathway for vibrations to travel directly through the concrete bypassing any dampening properties your cement may have.

If you try to insulate a house exterior wall from vibrations if you have anything rigid between the 2 walls like bolts etc then it bypasses whatever insulating material you put in the middle. Same with submarines, no use filling a layer between the outer and inner hull with vibration absorbing material if you then attach the inner hull to the outer hull with hundreds of bolts.

[G-Spot]

Hi,
great sounding project....

Thanks Craig

Yes I usually look at the most expensive option then the cheapest and try to get something in the middle.
I also hate noise. I recently switched from electrically produced air to petrol because I have a residential electricity supply, now I am always looking for electric tools and not the air ones because I can't stand to work with the racket the petrol compressor makes. So I am definitely going to go water cooled.

I also hate gearbox problems. When they go wrong its hard to tell why when all you are left with is pieces. So I am going to take yours and Peter's advice and go with linear rails of some sort.

I already started this project before COVID hit but had to shelve it until now. But I did spend a couple of months researching balls screws and as you pointed out getting long ones delivered without damage is the biggest problem. Half the stuff I get delivered to me here in South Africa comes with physical damage. And shipping is criminally expensive because of how bad the Customs and Duty clearance is. Anything more than 700mm in dimension is usually $250 for shipping.

Paul

[joeavaerage]

Hi,
Steel Youngs Modulus =207GPa
Aluminum Youngs Modulus =70GPa

Steel is three times stiffer than aluminum, fact.

Filling hollow tubes, be they aluminum or steel is a waste of time. You need rigidity and concrete filled tubes are not it. The same tube but with double the wall thickness is better.

Craig

[peteeng]

Hi G-Spot - pre or post tensioning a structure does not make it stiffer. It does make concrete stronger by placing more material in compression. As concrete is weak in tension. The internal strains are in equilibrium and do not contribute to the structural stiffness.

Steel masts especially stainless steel have been used in the past. But weight high up in a sailing boat is important or the reducing of weight that is. By the time you make steel thin enough to achieve the required weight it is too thin and can buckle. So after wood which is a great material, then steel, then aluminium came composites... Aluminium is viable because it can be extruded which is a cheap way to form metal.... Peter

[G-Spot]

Hi G-Spot - Some feedback. 30 years ago I used to program welding robots and train people to robotically weld.

Awesome!!! My strength and formal qualifications are in electronic engineering and computer programming. I have been building my own welder for this project, that was the first part of the project, but I had so many unanswered questions on the welding side that stopped me from choosing the correct components and I already had a couple of thousand dollars spent on components and didn't want to spend more blindly so I decided to buy the best TIG machine that I can interface to my machine, get the answers I need then finish my own welder, and like you say Laser is the way to go, and the prices are dropping to within the reach of DIY now.

I went for TIG and not MIG because with TIG the arc and adding the filler material are separate operations. I want to use a camera with infrared to see the temp of the metal and then add the aluminium to the weld pool. I have a whole bunch of theories about how I can control the weld pool using software(AI) that I want to explore.

I have done a lot of research in the past using cheap cameras and Small Board Computers to measure objects accurately and plan to get the rails of my machine laser etched with registration marks then use the camera and computer to position the machine and not the position of the motor shafts or encoder. I plan to use a mechanical brake in conjunction with jogging the motors to obtain resolutions far exceeding the resolution of the motors alone. This is why I am going to go for the straight cut gear rails, as long as they can move the axis then it wont matter how worn or how much backlash there is as long as they don't jam I will get excellent resolution. Simple closed loop routine that jogs the motor about the point you want to be at then apply brake and check.etc etc. My resolution on the axis which I use this strategy will be dependent on how well etched my rails are and not the ballscrew or geared rail I use. And as long as it is not an axis that requires rapid speed and cutting precision, like the y axis on a 3D printer, it will be a winner. I predict that it will take up to 30 minutes to weld each layer and no cutting or welding is done when moving the y axis and it is the 15m axis.

I need to make my first attempt 2.5m by 2.5m because we talking about ships here...lol and anything on a ship is supersize compared to other engineering. I can earn good money immediately with the machine just working to 5mm accuracy only welding, the parts are traditionally sent elsewhere for milling. If worse comes to worse I can just stay up all night and do them by hand. Even if it only works as a giant set square it will save hours compared to working with cardboard cut outs and visual approximations and hours of grinding.

I think I am going to make this version good enough to make the money that will pay for the next version......as opposed to to something that will bankrupt me before I literally get it off the ground.

I am going out early tomorrow to visit the scrap yards that keep industrial beams, aluminium and steel. I will report back on whatever is available and would be grateful for your input before I make a choice.

Thanks for your excellent advice and information.

Paul

[joeavaerage]

Hi,

mild steel has a tensile strength of about 500MPa. Hardened and tempered 4340 ( the stuff my Grandpa always called '90 ton axle steel' and his eyes glazed over) has
a tensile strength of about 1000MPa (depending on the temper), so about double the strength of mild steel.

Take a 50mm x 50mm bar of each 1m long, supported at both ends, and then load it with a 20kg weight in the center. Which deflects the most? Mild you might say,
after all its weaker right?. But no, they both deflect the same because the Youngs modulus of mild steel is the same as 4340, both being 207GPa

If you used an aluminum bar of the same dimensions (50mm x 50mm x 1 m) and loaded it with 20kg it would deflect three times as much as steel, because its Youngs modulus is 70GPa.

It beggars the imagination, but strength is NOT the same as stiffness.

For instance if you used a 100mm x 100mm x 5 aluminum section, its highly probable that any load you put on it is a small fraction of what it can withstand, ie it won't break but it will
spring or deflect. If you use 100mm x 100mm x 10mm aluminum section neither will it break but it will spring or deflect less.

Craig

[G-Spot]

Hi,

mild steel has a tensile strength of about 500MPa. Hardened and tempered 4340 ( the stuff my Grandpa always called '90 ton axle steel' and his eyes glazed over) has
a tensile strength of about 1000MPa (depending on the temper), so about double the strength of mild steel.

Craig

Yes I understand the difference between strength and stiffness.

If you apply tension to something you make it stiffer.

If you apply more tension to a washing line then it deflects less for a given load. But you reduce it's strength.

And a thin walled steel tube filled with concrete is a lot stiffer than just the steel tube alone.

"The Effects of Filling the Rectangular Hollow Steel Tube Beam with Concrete: An Experimental Case Study"

https://www.researchgate.net/publica...tal_Case_Study

[joeavaerage]

Hi,
BS. Concrete is about 50GPa, compared to steel of 207GPa.

If you fill a thin walled steel tube with concrete you increase its stiffness by the stiffness of the concrete alone and given the modest Youngs modulus of concrete not by much.
Double the wall thickness of the tube as the stiffness increases markedly, way WAY more than filling with concrete.

If you apply tension to something you make it stiffer.

BS. All you do is shift the load point up the curve, but the slope of the curve , ie the Youngs modulus remains the same.

If you apply more tension to a washing line then it deflects less for a given load.

The washing lines ability to withstand a load is not due to the stiffness of the line, but due to tension.

Craig

[peteeng]

Hello Paul - Readers - Complex concepts trigger warning

Your project is interesting and tough on so many levels. Lets talk about geometric resolution or accuracy to begin with. The geometric control system that most CNCs use is called deterministic. What you are describing is deterministic control. This means that the structure determines where the tool centre point (TCP) is. The machine actually does not know where the TCP is we assume that it is at the correct place because we told it to go there. You mention 1mm (so I assume that means +/-0.5mm) over 15m. This means the control system and the structure have to be good to 0.1mm which means it has to be made to 0.01mm. This is typically done by machining lands to the required accuracy using bigger machines then the machine your building (called a mother machine) In your case this is not possible as its a multi piece lightweight structure and its very big. You cannot control a TCP from within a deterministic system unless the entire system is built to the required accuracy. In your case this is impossible. So lets talk about humans. Our bodies are indeterminate and very complex. We use several systems to be able to touch our nose or pick up a glass of water. Particularly we use a system called Proprioception (PP). This is an electromagnetic referential system that is external to our mechanical deterministic system. The PP gives feedback to our motion systems and steers (reactively and predictively) our limbs so they get to where we want them to get to. Plus it not only handles position it also handles velocity and forces. Amazing. You are going to need a machine perception system.

I have been involved in a similar system that repair welds large mining equipment particularly extremely large excavator buckets, much bigger than you are talking about lets say 20m by 30m by 10m high....The company has gone down that road and stalled and started a stalled a couple of times. But size does not matter. The issues are the same as if it's on a benchtop or on a mining site.

So since you can't solve this issue in the deterministic universe your machine (and others) will have to have an external laser, lidar, precision gps or ultrasonic system (the perception system. Or machine perception system MPS) that measures where the TCP is at all times absolutely and then compares that to the deterministic information, create an error differential then feed that back to the controller for correction. This is done on some machines via glass scales and on very large mills its done with lasers these days. So I imagine you will have to have a scaffold or flagpole above your machine (as it needs to be separate from the deterministic machine), a transmitter/receiver on the machine head and a lidar scanner looking down on your job doing the MPS work. Your cameras will do the local work within a local geometric reference frame at the machine head and this will be integrated into the global reference frame and corrected in real time.

Then there is the issue of being a large lightweight machine and it will vibrate. Plus thermal expansion/contraction will be very large and I'm not sure how you will compensate for that, perhaps the MPS can cope...

All of this is not trivial... as you may have already guessed. Peter

as a note a lunar lander has literally fallen over on the moon because they did not spend the M$$$ to check the laser inertial system, they assumed that it being a commercial system it would work. Plus they left the system initialisation as a manual switch that had to be human operated prior to launch and it was not on a checklist with a big RED TAG so of course it was overlooked, so the mission which costs say $80M has basically failed. Inertial control is tricky. Musk has had heaps of rockets blow up trying to land the critters but his over the hump on that now. Why do I say this? because Paul many of these issues you are about to get entangled with are all the same stuff just smaller scale... Control, Control, you must learn control... Yoda

re: tension in washing line. The apparent stiffness increase is due to the foundations of the line not the line itself. If your lines are attached to tree twigs there is no gain in stiffness. If they are embedded in tonnes of concrete, then it's the foundation stiffness that improves not the line stiffness. Yoda probably had something to say about that as well.

[G-Spot]

Here's research that confirms filling a steel tube with prestressed concrete improves it's strength,stiffness and ductility even more so than just filling it with normal concrete.

Both experimental and analytical results show that the prestressed strands could significantly enhance the confinement effect of the core concrete under bending, which, in turn, improves the prestressed CFST beam performance in strength, stiffness and ductility.

https://ui.adsabs.harvard.edu/abs/20....144Z/abstract

[peteeng]

Hi G-spot - You have to be careful when reading this sort of work. They are done to prove something in particular usually hand in hand with an industry partner who wants a particular outcome and all of those statements are relative. Strength improved relative to a thin hollow beam YES, stiffness improved against the hollow beam YES and ductility improved against the hollow YES again. But it could have been a thicker tube and all the comments are then the same. Construction industry loves concrete so to use a light steel tube then fill with concrete is sort of attractive... You have to broaden your vision sometimes to find the "best" structural solution. Peter

thick hollow tube is stronger than thin hollow tube - obvious, thick hollow tube is stiffer than thin hollow tube - again obvious and thick hollow tube is more ductile than thin hollow tube (can you guess)

As an after thought i suggest you consider timber for your structures vs metal. Being involved in boats I expect you are familiar with timber, epoxy and fibreglass. Mass timber and laminated timber is now a very popular building material even for multi story buildings. Its stable, light stiff damp... Easily screwed to , repaired etc etc. Most of my routers are made from formply and its proven to be stiff, damp stable for its purpose. Peter

[peteeng]

Hi G-spot - I read the article and the reason the stiffness of the prestressed beam improves is that in the non stressed beam the concrete cracks, this means the structural efficiency in tension is zero. When the concrete is prestressed the compression holds the cracks together and the load is transferred via the steel members. This is discussed in the paper its not my interpretation. So there is a mechanical reason why the apparent stiffness improves. The concrete people are always looking for ways to improve their product. The metal people don't as its pretty much a given. Portland concrete is low modulus, low strength and it cracks. Its main construction benefit is it can be cast in place... There are concretes that don't crack and there are concretes that self heal. The Romans used them 2000 years ago, that's why their huge domes are still standing... In this papers case the tubes could have been made thicker or the pre-stressing strands could have been used by themselves so it was a tensegrity structure (sort of) Many ways to skin the cat. Its the concrete people driving that research, they need solutions........ Peter

[G-Spot]

OK been looking for steel.

New price is $1:15 Kg

I managed to find a scrap yard with a great selection of beams at $0:70 Kg so I am going to take the 50% cheaper option and use it to get thicker bigger beams than I could afford new.

They even have railways tracks, so tempted to use them and make my own bearings, 1000Mpa and hardened by trains, 100 years of stress relief.

Now I must choose a size, I need 10m and my budget is for 500kg.

Any recommendations for web and flange thickness ?

Is it better to go for beam height or flange thickness? I know that flange thickness is better if you want it to be stronger but web thickness is important for twist and buckling.

I think for a machine base/frame then resisting twist is just as important as load bearing strength. They are not going to be used for load bearing in the direction of gravity.

I want to go for I beams because they are easy to model in FEA and easy to increase stiffness later.

[G-Spot]

The concrete people are always looking for ways to improve their product. The metal people don't as its pretty much a given. Portland concrete is low modulus, low strength and it cracks. Its main construction benefit is it can be cast in place... There are concretes that don't crack and there are concretes that self heal. The Romans used them 2000 years ago, that's why their huge domes are still standing... In this papers case the tubes could have been made thicker or the pre-stressing strands could have been used by themselves so it was a tensegrity structure (sort of) Many ways to skin the cat. Its the concrete people driving that research, they need solutions........ Peter

I lived in Rome as a teenager. When I arrived there I bought a guide book and visited every great monument, took me about 2 years to cross them all off. The Pantheon is amazing and I visited it on a rainy day when the rain falls through the hole in the roof. The fact that it doesn't have a capstone at the very top and instead has to support itself just on the strength of the concrete makes it awe inspiring. And that's when I decided to join the masons(the concrete people) and I have been pushing it ever since...lol

There are 2 different types of concrete. One mixed by builders and one mixed by scientists. If you mix it with measuring jugs and scales and make sure the temperatures while mixing and even more importantly when curing are kept constant then it is like freaking kryptonite. With steel or other materials you can pull out a blow torch and it melts away. Concrete can break a man who tries to defeat it.

People need to stop looking around them at builder's concrete when judging concrete and try making their own under lab conditions.

[peteeng]

Hi G - Design the machine before you buy anything. Its free to design something. If you have CAD and FE then make all those decisions first. Making decisions based on cost like you are doing this early always ends in grief. Define what the structure has to do and design it to work. And remember the CAD model is perfect - the real stuff is twisted, crooked and imperfect. Peter

[hanermo]

Yeah right.
About 4 trillion dollars per years is spent on concrete and buildings by builders, globally.

There is probably exactly nothing that high-end builders don´t know about concrete ..
In Finland for example concrete can be cast in sub-zero temps, with the right additives.

2000 years ago the romans already had concrete that could be cast underwater, and they used it to build viaducts -- some still in use today.

ANY industry that spends over a trillion dollars per year at anything will quickly achieve a mini-max of results/cost/ease-of-use within a very few years.
PV panels are an example.
So is concrete.
So is steel.
So is alu manufacturing.

No lab anywhere is going to do concrete better than builders spending hundreds of millions of dollars per year at it over decades.

Endless materials are better than concrete. Depends on what and how your measure and how much You can afford to spend.

E.g. The best laminated wood beams, are about 30% stronger than steel by mass, and stiffer, and can take any shape including complex compound curves.
The best carbon fiber beams are very light vs strong, but were abandonded by SpaceX after a few years in favour of stainless steel.

And they, SpaceX, are the most successful, the best, engineering company on earth.+/-
With pretty much limitless money.

Buckyballs and other exotics are theoretically stronger by about 10x, but You cannot buy them by the ton at Your local steel merchant or builders supply.
Spiderwebs are much stronger than steel, by mass, but there exists the problem of getting a 100 M spiders weaving a web for You, on time, to budget, to spec.

Real anecdote. Feel free to laugh. Or invest.

I researched this.
Really, I did.
Apparently If You feed spiders amphetamines You can increase web production a lot.

Potential investors were very impressed when I suggested a farm with 100 M spiders in the sahara desert, fed on amphetamines.
This was to build a space elevator, around 2002.

There are technical challenges, one of which is the potential energy in a space elevator 36.000 km tall, should it fail.
Firstly it will pretty much circle the globe when falling, and secondly it is likely to crack all the continents and vapourise a lot of seas.

So who exactly will insure it, permit it ?
The permits process is probably a 20-year process.
Financially, 2T$ should do it, and Elon Musk and or Jeff Bezos could carry it off, maybe Larry Ellison.
It c/would potentially give humans free spaceflight, and asteroid mining, and endless raw materials from the belt, all kinds of goodies.

There are 2 different types of concrete. One mixed by builders and one mixed by scientists. If you mix it with measuring jugs and scales and make sure the temperatures while mixing and even more importantly when curing are kept constant then it is like freaking kryptonite. With steel or other materials you can pull out a blow torch and it melts away. Concrete can break a man who tries to defeat it.

People need to stop looking around them at builder's concrete when judging concrete and try making their own under lab conditions.