Casting design practices

[Arlin]

As I don't do much casting/mold design, I am a bit curious about how people do this sort of work all the time.

From another discussion here, it seems that many actually design the 'as cast' part first and then use a configuration or insert part or assembly feature(s) to cut away material for the 'as machined' part.

As a designer, this process seems backwards to me. In my mindset, I would want to specify the end result (as machined)in my primary documentation. Tolerances should be specified that dictate what needs to be machined and where a rough cast is acceptable. The manufacturing experts then use the model and add material to allow for machining operations.

I do a lot of sheet metal work where we use the above design/manufacturing process. My drawing only contains the final, bent configuration with proper tolerancing. Manufacturing is responsible for how the raw material gets there (flat pattern, cut/punched/machined holes, etc.). Manufacturing may create their own drawing detailing their process, but that drawing model is based of the design model, not the other way around.

Thanks, Just curious...

 

[TheTick]

Arlin:

This is a great topic! One more case of one model does not do it all.

I agree. From a design standpoint, the process is backward, and the as-cast model should be designed from the finished piece. Manufacturing flows in the opposite direction of design.

The point where it starts to unravel is when the post-casting process changes, such as including additional holes. Oftentimes, this happens months or years after a casting has been in production. Most documentation trails go in the direction of process flow.

I doubt if there is any single process that can capture all the interplay between castings and finished parts.

All this machinery making modern music can still be open-hearted.

 

[ctopher]

Arlin,
I have always wondered this my self.
I always design as if the part is finished, then go from there. Let the mold shop work out the in-betweens.
I have seen people go other routes, but seems backwards and they usually take more time creating designs.
You are on the right path.

 

[TateJ]

I guess it depends on who you work for:
the casting supplier
or the manufacturer who's buying the casting for machining.

We buy our castings - we have only a vague idea what the molds look like. the casting supplier designed the molds to produce the castings we asked for. They did this from our machining drawings.

This will look favorable on my next review.
tatejATusfilter.com[/u]​

 

[MadMango]

We don't do a lot of casted parts either, but when we do we always do the finish machining in house. This saves us a nice chunk of change at the end of the day.

So in our case, we model the finished part to look like the "as cast" part, since that is what we recieve and inspect. After that the part is sent on for finish machining. This is where the model has an extra configuration or two to represent this finish machining and the final part.

I hope I didn't muddy the waters too bad.

MadMango
"Probable impossibilities are to be preferred to improbable possibilities."
Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[Rocko]

We do a decent number of aluminum castings for mold purposes and we always machine our own patterns. We then have the foundry cast and polish to a particular surface finish. We will design the as finished part and add the scale feature for shrinkage. We then only need to suppress the scale factor and it will have all of the hole locations and other requirements. The key factor is the larger the casting the more the tolerance increases on the shrink factor matching what is predicted. We typically end up within .001-.012 of the mold dimensions required. In our industry those tolerance standards are acceptable.

 

[Boosie]

I am in a unique situation in that I design the finished part. Then also design the permanent mold tooling to create the cast part. I work hand in hand with our foundry to make sure that our product is castable and the tooling is as efficient as we can make it. We also build all of our molds in house. This allows us to keep a firm grip on design intent should there need to be some concessions when it comes time to cut the mold cavity.

I always design the finished part and add then configurations for the as cast state as well as any other versions that we may need along the way. This way every thing is tied to the parent design. The mold design is breeze with a simple insert cavity feature.

If you have any other specific questions feel free to ask. We have already produced 9 molds this year and have about the same number on the books.

Mold Design with SW
Boosie

 

[Andy330hp]

I am very curious about this! So, what you are saying is that take the model of the finished part and add material to it via extrusions, etc to get a mold shape?

It seems logical. It also seems like you could run into problems if you changed the machining that the casting accidentally get overlooked and not get modified to suit. Whereas, if the casting is the basis for the machining, it is immediately evident if the casting no longer has enough material!

It does not sound like this has been a problem for you. How do you avoid it?

 

[TSTUEBER]

I work for a large sand cast and permenent mold foundy. Most cases when we receive a model from a customer of a line ready part cast and machined I will add stock to the casting. If I am starting from scratch I will develope the casting first and add the machining to it later. Alot of the problems that come into play are all of the different cad systems that our customers use and that we have in house. A simple revision in a software and igesed or parasoilded to SolidWorks that is a machined casting either means updating the mold model file by comparision or starting over. There is no good way to keep associativity between other cad systems. One nice thing is we run SolidWorks, Catia, Pro-E.

 

[drawoh]

Arlin,

I am very aggressive about using assembly centric design features of SolidWorks, however, I try to make my models conform to the manufacturing procedure.

Someone else may be responsible for manufacturing, but you are responsible for holding manufacturing costs down. If, for example, you design a housing with a negative internal draft angle, you drastically increase the cost of the casting. Either an elaborate mould is required, or you need an expendable internal pattern. You need to consider jigging of the casting for the clean-up machining. You need to consider how easy it is to cast and/or machine and/or bend the tolerances you specified.

If you want to provide understanding of the design of your model, try renaming all the features as you add them into SolidWorks.

JHG

 

[macPT]

As many of you mentioned, the design process is not a one way process.

If you are closed in your office designing parts, you can design the greatest parts, but your company can't manufacture then because don't have the knolege, or the tools, or the money, or the clients simply will not buy them.

And to avoid that you have two ways (in an ideal company, you should have them together): 1 - you know very well the production capabilities and you can design for manufacturing; 2 - in some stages of your design you must call the production people to discuss and validate the design. In fact, you should call more people (marketing, quality, management,...). It's a surprise how we can simplify the design and how we can predict problems (both technical and costs) when all these people are analysing the design by their point of view!

So a casted or bended part, wich are designed, in the first place, as a final part (that is ignoring all the features that make them a cast part or a flat part), can have changes in it's design according to the production tips (even if the part will be subcontracted, we allways ask to the subcontractor what possible changes the part should have in order to have a better quality and lower cost).

In the end of the design, we have a "real" finished part. Normally we also design the "real" cast/flat part for reference. If you change subcontractor, maybe you will need to redesign something so you must put these designs to it's analysis.

Regards

 

[pdybeck]

I'll throw an extra variable into this discussion - PDM. We do not allow users here to define more than one part number in a file (via configurations) because of data managment issues. It keeps all our files simplified and easy to manage. It also makes the rev process easier, as now we don't have to rev a configuration and a file, just rev the file. There aren't very many PDM systems that can treat a configuration of a file as a seperate entity, and usually these are quite expensive programs. Because this is the way that we want to manage files, we insert the cast part into the machined part file. This creates a "made from" part (the machined part is made from the cast part). If we look at the machined part, it has a reference -the cast part. If we look at the cast part and look at where it is used, it shows any machined parts made from that casting. I don't see how configurations can capture that aspect of data management and in fact I think it complicates matters when you start to make revs. As far as designing both parts, I prefer to do both at the same time. I'll start with the casting, obviously, as that gets inserted into the machined part. But then I can go back and forth tweaking the casting and machining to get the results that I'm looking for. We actually design both parts here, so we have this luxury. A drawback occurs when we have a part designed that we originally did not need a casting for, and then decide that we will make it from a cast part. In this case, the parts have to be recreated to ensure the correct structure that was described above. It makes for some extra work sometimes, but it correctly captures the BOM strcuture of the design and allows us to manage our parts the way we desire, without complicating the PDM Environment.

 

[meintsi]

Here's another two cents worth....

We develop many parts/castings here for our hydraulic product lines.
Valve bodies, covers, manifold blocks, housings, brackets, flanges, etc.....

pdybeck's statement, "As far as designing both parts, I prefer to do both at the same time." comes very close to what we find as the most efficient way to work with castings, core blocks, and machined parts.

The design intent of the finished part is still the same if the part is to be machined complete from a piece of blank stock or whether it is to be machined from a casted blank. The method we use allows us the most flexibility for anyone who becomes involved in the design process at any time.

The casting should have it's own part number and therfore be a stand-alone part file. You should have some idea of what your finished part will look like and that should lead you to what the first base-extrude will look like. Proper selection here can save a few extra hours for those people who like a "tidy" feature manager.

Also, depending upon the complexity of the casting, we model the core of the casting (if there is one) either as a series of 'cut' operations within the part or sometimes as a seperate solid model which is then 'assembled' with the casting using SW cavity feature.

The finished "machined part" or "machined blank" is then its own seperate part number where the casting is inserted as a "base part".

SW then makes it very easy then to "tweak" any part of the design process as other ideas/problems arise.



Remember...
"If you don't use your head,
your going to have to use your feet."

 

[JNR]

We do a lot of castings which are then machined. None of the part manufacturing is done in-house. I can't say there is an ultimate way to to this.

After much experimentation, having two configurations - one as cast, one as machined, seems to be the way to go. It would appear at first glance that designing the finished (machined) configuration makes most sense for the designer. It is then perhaps possible to suppress the machined features, add any cast features plus machining stock in the cast configuration. Unfortunately this is not ideal. Often you get geometry where (for example) you end up with parts of fillets being removed in machining, but you fully filleted it in the machined config. So now you add a straight boss for allowance in the casting and it does not come out right. So infact we use a combination. In those cases we go into the cast config. and make sure the basic casting is correct, then remove the machining stock in the machined config. We do this on a feature level, rather than in one fell swoop. It takes some to and fro but seems worth it to get the part logic right for the manufacturing process. (Here's another example. Slightly raised pad to be machined. You don't want a stress raiser, so you chamfer or fillet it to the surface it grows out of. In pactice this would be done in the casting, so the resulting ACTUAL diameter of the machined pad depends on tolerances and how much is machined off. If you fillet/chamfer in the machined configuration, you either have a complex job to extend the shape for the machining allowance in the casting or you add a straight boss - but this could still result in a mchined part with the stress raiser!) You defniitely never end up with a part you can't make from the casting (particualrly if you later revise it). Since casting usually requires expensive hard tooling of some kind this seems to be a good idea.

BTW: We are a bit strange. A machined "part" usually includes other features like press fit bearings, dowel pins, etc. (anything "inseparable&quot. The vendor makes the whole thing and there is no assembly drawing for it (in fact they even buy the castings themselves to our drawings). So we have one SW PART with the casting and as-machined casting configurations, plus a SW ASSEMBLY (but a "part" to us) which has all the added part in it. tThen we have a casting part drawing from the cast SW PART config. and a machined part drawing from the SW ASSEMBLY. interesting, but works extremely well for us.

3/4 of all the Spam produced goes to Hawaii - shame that's not true of SPAM also.......