[piotr07da]

Hello,

I am designing a CNC machine and plan to cast it using concrete (kind of UHPC or at least a mix with a low water/cement ratio). I would like to ask for your advice.

The travel distances of the machine in the X / Y / Z axes are:
- Metric: 400 mm / 300 mm / 380 mm
- Imperial: 15.75" / 11.81" / 14.96"

Initially, I assumed the heaviest machining scenario I am aiming for is:
- Material: D2 steel (HRB:95|HB:209)
- Tool diameter: 10 mm (0.39")
- Number of cutters: 4
- Cutting speed: 70 m/min (230 sfm)
- DOC: 12 mm (0.47")
- WOC: 0.3 mm (0.012")
- Feed per tooth: 0.1 mm (0.004")
which gives:
- 225 N (50.6 lb) of tangential cutting force
- 1.12 Nm (9.96 in-lb) torque at cutter
- 0.47 kW (0.4 hp) spindle motor
- 94 cm^3/min material removal rate
- 2228 RPM
- 891 mm/min (36 ipm) feed rate

To be sure, in my FEM simulation I used 300 N (67.4 lb) of force applied to the spindle nose instead of 225 N (50.6 lb).

The FEM results show a maximum deflection of:
- 0.032 mm (0.00126") in both the -X and +X direction
- 0.037 mm (0.00146") in both the -Y and +Y direction

My first question is whether those cutting parameters are to ambitious/aggressive for such a small machine.
I have very little experience in machining; all I have done previously was to machine aluminum with a DOC of 1 mm and a WOC of 3 mm in slotting.
Thus, I lack a reference point.
One thing that tells me that it is to ambitious is the material removal rate of 94 cubic centimeters per minute - I would probably be happy having it 10 times slower. However if such a performance could be achieved that would be nice.

The second question is about the consequences of the deflections mentioned above. Could they lead to tool breakage? I'm not worried about the dimensions of the parts since finishing passes will have much lower deflection. Or should I be worried about it?

I was planning to equip the machine with 1.5 kW (2 hp) spindle motor and 8.5 Nm (75 in-lb) steppers with encoders.
I did the calculations but for even heavier machining.
But now I wonder if that might be overkill, given the low rigidity of the machine's body.

I would be grateful for any advice.

[G-Spot]

I have been doing FEA analysis on this same design. Looking at thermal distortions from using steel on concrete.

I would recommend you do the same, simulations like the machine is warm from sunlight, or just ambient air temp...then you put coolant on the steel rails, or any steel part of the machine.
Or you have a heated workshop, then you go home and switch the heating off and outside temps are below zero.
When I did these simulations it showed stresses that would crack the concrete and cause the rails to warp.
Also the warp was worst when the steel was at lower temp than the concrete and insignificant when the concrete got warmer than the rails.
Which meant that if you made sure your coolant was always at a higher temp than the concrete you would not have problems. I was planning to pump my coolant through the concrete which would guarantee the coolant would never be at a lower temp and also the lowest temp for the machines current temp.

I was planning on bolting my rails and steel completely through the concrete because it seemed that if you tried to embed them in the concrete then they could break free if the temp of the steel fell 20 degrees below the concrete.

Only been doing the thermal analysis for the last week or two, don't take anything I say as gospel but you should definitely take a look using your FEA

[peteeng]

Hi Pio - What modulus are you using for the concrete? If your model is correct (so check all the connections and geometry) then the spindle nose static stiffness is 300N/37um ie 8N/um. This is not very stiff compared to many machines. Try to set up your FE so it mimics ASME B5 attached, I think you have done that. Concrete is a low modulus material say 30GPa to 50Gpa depending on the mix, so think about using steel which is the stiffest material generally available at 200GPa. So for the same geometry it will be 4x stiffer. I see your column is hollow, in concrete its generally solid for ease of casting and maxing the stiffness. There are other machine configurations that are stiffer by nature that can help as well. Any particular reason for a C frame machine in concrete?

Looking at your FE - I'd use a finer mesh, there are some very large elements in there close to small features. The table seems to be deflecting quite a bit so maybe look at that. Is the table steel? You can isolate the structure at different parts to identify where the main deflection is coming from. These are called structural loops. So in one freedom case you may restrain the column at its base, the next restrain at the table bearings etc. The loops sum deflection should add up to the global deflection. So subtract the outreach def from the next loop (column) etc and you will be able to identify the loop with the biggest deflection. That's the loop that needs attention. Then move to the next biggest loop etc...

With a C column machine each loop is in series so its straightforward. In a more complex structure the loadpath can sometimes be a bit complex to find... Peter

[G-Spot]

This video was posted by Jack on my thread about using concrete in machines.
What he says about keeping symmetry works, if you keep the same amount of steel both sides of the concrete then there is little distortion.
He also discusses the consequences of allowing coolant to run on your rails surface. Everything this guys said checks out in FEA.
https://www.youtube.com/watch?v=_XGF2FZUTms

[peteeng]

Hi Pio - A couple of things. What sort of machine are you aiming at? Maker/Hobby level? Commercial? Toolmaker grade? and what do you want to do with it? Your design is clearly above say a Taig but below a HAAS or a KERN. These machine steel "well" at their level. I expect your design will do well, but within its limits and that's the problem you have. Not having a comparison machine or experience with a similar machine makes it difficult to predict what it will do. But if you pick a commercial machine and use that as a benchmark or muse you maybe able to estimate how it will go. My experience at designing this size machine and trying to get to a commercial machine stiffness (say 150N/um) is that you won't get there in concrete. Commercial machines of this size are made from cast iron which has an E=100GPa. Well above concrete. Concrete and EG machines are usually much bigger and take advantage of the large geometry. Geometry wins over material. As machines get smaller material stiffness comes into play and becomes important. A 300mm machine is sort of in the transition zone. A steel machine will be stiffer I feel. Then there's the C frame config which is called an "open" frame. Closed frames like gantry machines can be stiffer because the structure is more efficient. Your on the right track, keep at it.. Peter

[routalot]

I know almost nothing about machining steel.I am curious about the spindle motor you are proposing to use and how it generates enough torque at the sort of speed that steel is best machined at.I have no up to date knowledge of high speed machining of steel and it is a topic I would like to know a bit more about.My lingering suspicion is that 0.47Kw is less than ideal.

[piotr07da]

G-Spot
Thank you. I gave it some thought, but I didn't expect it to be such a significant issue. I'll definitely look into that. Thank you for the video - I will check it for sure.

peteeng
Thank you.

I apologize for not including the basic assumptions and information about the project in my initial post. I am aiming for a hobby-level machine that can cut steel, soft metals, and wood. It's hard to specify the exact parts - it could be a steel gear, a wooden sculpture, a brass trophy, or pretty much anything... If I had to compare it to a benchmark machine, it might be something like the Tormach 770/1100. The Haas VF1 is fantastic, but I cannot imagine the required stiffness for utilizing its 22kW spindle at its full power.

I used 32 GPa for concrete and 210 GPa for steel.

I need to review the ASME B5 document you sent, but it will take some time. For now, I've skimmed through it and, if I understand correctly, I should measure the deflection at the center of the work zone. My current setup is quite different because I measure at the maximum extreme positions of the work zone. I've attached images showing my current setup of fixed and force constraints.

Why use concrete for the base and column? There are three reasons:

1) Ease and safety of manufacture - It might sound funny, but it's due to the fact that I currently live in a small apartment in a multifamily building on the 3rd floor, which, strange as it may sound, is a real constraint I must consider, at least for the near future. Using concrete avoids fire hazards, toxic fumes, omnipresent steel dust, and noise from welding or grinding. Concrete seemed to be an easier material to work with inside my apartment. Moreover, I have no experience in welding, while creating a mold for casting seems easier.
2) Cost - Concrete is cheap.
3) Vibration dampening as an added bonus.

Why a column design? It’s a compact design that fits within the limited space of my room.

Why is the column (and also the base) hollow? It's a compromise. I chose this because I wanted to reduce the weight, as I don’t want the machine to risk falling through to the floor below. I’ve run numerous simulations and found that I can reduce the total weight of the base and column from 624 kg to 405 kg (from 1376 lb to 893 lb) without losing much stiffness:
- In the X-axis, it drops from 9 N/µm to 8 N/µm.
- In the Y-axis, it drops from 11.5 N/µm to 9.5 N/µm.
This comparison is based on my current FEA setup. I expect the results might vary with a different setup, but generally, I anticipate that filling the voids yields diminishing returns.

Regardless of my current reasoning, I am open to a complete redesign or may consider postponing my project for an undefined time until I have access to a garage with more space to potentially weld the machine.

I will read the document thoroughly and repeat the FEA.

routalot
While choosing the spindle, I plugged my most ambitious cutting parameters mentioned above into this calculator: https://www.kennametal.com/us/en/res...and-power.html
, and it indicated:
- 1.12 Nm (9.96 in-lb) torque at cutter
- 0.47 kW (0.4 hp) spindle motor
So, to have some margin, I am thinking about a 1.5 kW motor.

[G-Spot]

If I had to machine accurate parts in an apartment I would go the EDM route, especially for gears.
You don't require any stiffness, basically a 3D printer in a fish tank.
You can get or make your own power supply for a lot cheaper than the $2000 or something he charges for his kit.

https://www.youtube.com/watch?v=2dsrLD52Mv0

[peteeng]

Hi Pio:
1) thermal issues for a machine in a flat is unlikely to be a problem when aiming at a Hobby level machine. The principles of thermal symmetry if they can be applied is worthwhile especially if you have more than one material in the machine
2) Concrete has perceived cost advantages but on a small machine I think its a tough call to go down that path and the main costs are in machining the lands and the stuff bolted to the machine frame. Steel is actually the cheapest material $/E/m3
3) Since you want to build in an apartment consider a bolted together machine in steel or aluminium. It will be stiffer then your 32GPa concrete, able to be disassembled and reassembled somewhere else. It will be adjustable to get to the tolerances you want.. There are issues with cast machines, then machining the inserts. There are a couple of threads here that share their pain through this process.
4) If the tormach and HAAS are your benchmarks you won't achieve their stiffness, sorry. They have considerably more metal in them then you can have on a 3rd floor....
5) I few years ago I got very excited with EG and concrete. Made several test coupons using various ceramics and the usual ingredients. Talked to a very good concrete chemist at length and had my samples tested. All of that was very disappointing. Even with all alox reo (which is E=300GPa) I only got to 20GPa in bending in EG . I think the usual compression test is biased. So I have gone off concrete and have gone back to metal. The best grout I can get is 50GPa which is qualified via compression not bending. Most machine parts are in bending.
6) so if you have access to basic metal working tools or contract machinists or a machinist friend I suggest you consider a bolted machine. There is a thread here by katran that makes a very good aluminium machine about your size.
7) I do a lot of analysis on bolted connections and very large mining equipment and I think bolted connections for hobby machines are misunderstood and under rated. Commercial machines are driven by production costing requirements so casting in iron is one of the best ways to get economic machines. Making a bespoke machine is another story.
8) Now basic FE includes separating connections and preloaded bolts its quite straightforward to analyse this sort of structure and compare to welded structures. Efficiencies in the order of 90% are achievable which when you compare to concrete at 32Gpa and steel at 210x0.9= 189GPa its got to be attractive
9) then theres the C frame config. Its a historical design based on the requirement to prevent stick slip in the ways. These days there are several other ways to config the machine that gets the screws and workings up out of the muck plus they are stiffer, eg gantry designs.
10) could go on but thats a lot for now... Cheers Peter

oh you mention damping - if the machine is stiff enough damping is a non issue at maker level. Commercial level trying to pump out stuff asap damping is a big issue. Even the big guys cannot fully design around damping for their machines. Look up tap testing of machines, you will see that even the big makers sort that out after they get into production. I have done dynamic analysis of a large 5 axis machine and its not simple... not enough data and too much guessing at various figures...

[piotr07da]

G-Spot

EDM is cool, but it is a different machine that has its limitations. But thanks for pointing that out as an option.

peteeng

I am glad that I posted my doubts here. You have given me a lot of information and a lot to think about. Speaking of EG, that was my first plan. I conducted tests on 10 different samples, experimenting with various ratios of epoxy and different aggregates: granite, sand, powdered quartz, and electrocorundum. It turned out to have poor stiffness and be very expensive. Then I switched to concrete, made at least 20 specimens, and the results did not differ much from the 32 GPa generic concrete from FreeCAD FEM, so I am using that in my simulations.

The quick conclusion is that I must create two all-steel designs - one for a column and one for a gantry for the same size work zone as in my original project, compare all three of them in FEA, and do some research about the cost of materials and required machining in my area.
I will also check katran's thread.

[bravernoe]

Comparing those designs in FEA should give you some solid insights. Good luck with your research and simulations!

[G-Spot]

I few years ago I got very excited with EG and concrete. Made several test coupons using various ceramics and the usual ingredients. Talked to a very good concrete chemist at length and had my samples tested. All of that was very disappointing. Even with all alox reo (which is E=300GPa) I only got to 20GPa in bending in EG . I think the usual compression test is biased. So I have gone off concrete and have gone back to metal. The best grout I can get is 50GPa which is qualified via compression not bending. Most machine parts are in bending..

I think if you did some research you will find that concrete is 20 > 25 GPa in bending and if you treat it like any other material and work with it's actual properties and not the properties you imagined then it is not a bad material. Just because some contractors made a mess of 3 of your floor slabs is not a scientific argument.

And one of the reasons we use FEA is so we can see whether materials are in bending or not and your statement about steel being the best stiffness for dollars is ludicrous when you just look at the world around us. $100 of cement without any steel at all can give you a 500mm x 500mm x 500mm work area at 100N/um stiffness.

It would be nice to see some real data or FEA simulations that back up you claims of which parts of a machine would be under tension/compression.

[peteeng]

Mr G-Spot - I won't get into various discussions with you in this thread or other threads. Unfortunately, what you say cannot be removed, it is unfounded. I have been designing & building machines and structures for over 40 years so there is a large trail of work behind me. Take it or leave it. As I said to you in your thread, I leave you to your journey, good luck. Peter

[peteeng]

Hello Pio - I am not against concrete or any material. I pick the material for the job. If I were designing a large machine with integrated electricals and cooling, engineering grout CSA would be my first pick. As the machine gets smaller the geometry diminishes and material stiffness becomes important to achieve the stiffest machine. There is a cross over point so a trade study needs to be done to figure out the path. Ultimately its up to the Maker which way they go... Peter

[RaderSidetrack]

I have removed two posts from this thread. Personal attacks will not be tolerated.

[piotr07da]

peteeng

I quickly modified my FEA setup according to the document you attached, and now it shows 11 µm at 300N, which gives 27 N/µm. It is better, but the model is simplified, so I expect the real-world performance might be about half of that.

I am also slightly surprised that this is the method of measurement because it does not represent the worst-case scenario. In my initial setup, I assumed that I wanted to know the deflection at min or max X, min Y, and max Z, not in the middle of the work zone as stated in ASME B5.54. However, if this is the standard for measurement that allows for comparing apples to apples, then I am okay with it.

What you are saying makes sense. For now, I am not abandoning the idea of making the machine out of concrete, but I am very curious about the results for an all-steel design. If it turns out to be better, I will go that route. I will start with a C design to have a more direct comparison between concrete and steel. What appeals to me is the possibility of an even lighter machine and the fact that any mistakes during the build can be easily fixed by replacing a single element if everything is screwed together.

However, it took me a few weeks to create this design, so it will also take some time to redesign it using only steel.

[G-Spot]

My experience at designing this size machine and trying to get to a commercial machine stiffness (say 150N/um) is that you won't get there in concrete.

That's interesting, would love to see one of your 150N/um designs, what sort of machines were they, C shaped? Gantry? How did the FEA predicted stiffness vary from the actual stiffness? how much did they cost to build? How long did they take to build?

[peteeng]

Hi Pio - For a trade study use simplified geometry, as simple as possible. The aim is to establish a direction. Your current model is nicely detailed nearly ready for manufacturing specifications... But time spent in the design phase is time well spent....

The mill design I like is the Mori M1. It has a fixed table and raised mechanics so they are out of the muck. Since the payload is not moved the mechanics can be optimised neatly as the machine parts are fully known. I expect it to be better then a gantry machine at your table size. One example of a concrete gantry machine is worthwhile but I can't find the image. I'll keep looking.... Peter

edit - found it - the designer has used the concrete for the walls which forms a perfect well for a wet machine. The high wall and rail has proved to be very, very stiff compared to the column designs I have built... Re modelling - I have always modelled at worst case. I have been searching for an actual ASME B5 report but have not found one yet. Some time ago I was chatting with a HAAS engineer and asked what was the target static stiffness of their machines and he said he did not know?? said he'd get back to me but didn't

My current design approach is to make block geometry which is the bounding box geometry or package geometry of the part and these are all solid. Run the FE to establish solid stiffness. Then shell each part out until the stiffness starts to be affected. Then look at those shapes and results and decide how to get to it in plate or casting or whatever... With Fusion you can run it in form finding, generative and shape finding to see what it thinks about it as well. Good Luck...

[G-Spot]

]
1) thermal issues for a machine in a flat is unlikely to be a problem when aiming at a Hobby level machine. The principles of thermal symmetry if they can be applied is worthwhile especially if you have more than one material in the machine

oh you mention damping - if the machine is stiff enough damping is a non issue at maker level.

I'm confused as to your statements on vibration and thermal issues with machine. What is it about a Hobby machine that makes these effects insignificant? Do you have any data or one of your thousands of FEA simulations that can show when these effects become insignificant.

[RaderSidetrack]

I'm confused as to your statements on vibration and thermal issues with machine. What is it about a Hobby machine that makes these effects insignificant? Do you have any data or one of your thousands of FEA simulations that can show when these effects become insignificant.

A "hobby machine in a flat" refers to a machine that is located inside an apartment or flat, by definition a residential building typically occupied by multiple non-related persons / family groups. Such a situation imposes significant limitations on a machine as to size, noise, weight, vibration, 'operating time/hours' and electrical power consumption.

A machine inside an apartment that draws the displeasure of other residents or the landlord is a recipe for getting the machine owner evicted from the apartment.

Lets get back to the main topic of the thread.