Fully constrained vs no constraints in an assembly

[dtharrett]

We are running 7.5 w/ TC integration but I think my question is related to all versions.

I work in the equipemnt design group of a large comapany that primarily designs consumer products. We just make the equipment that assembles the product.

Because of the nature of the workflow, I get to handle others models & assemblies and noticed the following. Our equipment assemblies contain parts that are fully or nearly fully constrained while the product assembles do not contain a single constraint. The latter of the two is very forein to me as I was brought up in another CAD package that required everything to be fully constrained.

I understand that NX does not require us to constrain parts within an assembly but how would one go about getting all of the parts in their proper location?

What are others here doing and what are the pros and cons of fully constrained vs no constraints?

 

[loki3000]

fix the parts (in their location). but to get them there you usually need some constraints, so...

 

[JHTH]

Personally I'd use constraints, but you can also use the Move Components function to position them and then (if desired) apply a Fix constraint so that they can't move. This does prevent poorly-constrained assemblies 'exploding' if a part is removed/replaced/redesigned but has two disadvantages. It preserves no design intent and it also requires frequent interference checking and diligent modification of the surrounding parts.

Another approach, used a lot in the auto industry, is to design in 'car space' - a reference point (e.g. the centre of the front axle) is defined as a global 0,0,0 and all components (seats, fuel tank, windscreen etc) are modelled relative to that in their own part files and then inserted into the parent assembly with absolute positioning so that the part origin is coincident with the global origin and everything then, in theory, fits together.

Hope that's of some help and I was at least vaguely clear what I meant.

Jon

JHTH
NX 7.0.1.7 + TC 2007

 

[JHTH]

Forgot to say that 'car space' also requires you to define beforehand the spatial envelopes in which the components and subassemblies must be modelled. If everyone sticks to these, you shouldn't need much interference checking, but it's still worth doing on occasion!

Jon

JHTH
NX 7.0.1.7 + TC 2007

 

[dtharrett]

Hmm, the "car space" technique sounds interesting. Probably not a good fit for us but the concept interests me. Does it offer some sort of advantage with file size as it does not have a huge constraint network? What other advantages/disadvantages does it have?

 

[ewh]

The same methodology is used in aerospace... unique components are usually designed in aircraft coordinates. Commmon components (hardware, etc) need constraints.

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[JohnRBaker]

An observation, for what it's worth: if you're worried about the impact of adding constraints will have on the SIZE of your assembly files, then you've got bigger problems than worrying about whether your components are constrained or not.

Another observation: while it is true that the automotive and aerospace industries have successfully used the concept of 'ship coordinates', where all the major parts (and when carried to extremes, even minor ones, which led one major auto manufacturer to discover that it had something like 27 DIFFERENT, yet currently in production, part numbers for a 'Gas Cap', most all of which were interchangeable, but for which each had been designed relative to the global origin for the particular vehicle that that gas cap was going to be used on) are designed and modeled relative to a common X,Y,Z absolute origin and orientation, this was first adopted long before 3D CAD was being employed and over time has been somewhat relaxed, particularly when the trend moved toward using common components (primarily in automotive, as the above 'gas cap' example showed) across multiple product lines.

Now your ID designates you as being 'Mechanical' and while that can cover a multitude of 'sins', you did use the term "equipemnt" so we'll assume that you mean machinery (an area I'm fairly familiar with). In that case there often is NO natural X,Y,Z origin (BTW, in aircraft design the 'origin' is generally NEVER ON the actual airframe itself, but often is at some arbitrary location AHEAD of and BELOW the lowest possible point on the aircraft with the only predictable characteristic being that it does lie along the centerline) and therefore the common approach is to FIX some major frame component of the machine and then constrain, based on both position and potential movement of the various Components. Let me emphasis this last aspect, the potential MOVEMENT of components.

When people mention design intent when talking about constraints in an Assembly, it's not only WHERE something is located relative to some other Components, but also how that Component MOVES relative those same Components. Often it's the DEGREES OF FREEDOM of a Component that's as important as the location since if properly defined and constrained where appropriate, the 'network' of constraints can also allow you to articulate the moveable parts of your machine in such as way to be simulate it's mechanical or kinematic behavior as well as it's spatial locations. So please keep that in mind as well, just because it may be easy to simply FIX something in absolute space, if that Component does more than just 'fly in formation' with the rest of the Assembly, it may be better to take the time and provide the constraints which represents the ACTUAL physical 'relationship' between individual parts rather than just their overall rigid-body or fixed spatial relationships.

Anyway, those are just a few observations, for what their worth.

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

UG/NX Museum:

http://www.plmworld.org/p/cm/ld/fid=209

To an Engineer, the glass is twice as big as it needs to be.

 

[dtharrett]

Thanks for the information. Yes I am an equipment design engineer, designing assembly machinery for consumer products. For now, I am going to continue with my current constaint practices.

I was just trying to wrap my head around how someone would design without constraints and if there was any value to pursuing this approach. Sounds like I would loose the ability "articulate the moveable parts of your machine in such as way to be simulate it's mechanical or kinematic behavior" which basically rules out this approach for me.

 

[edgray]

I have a particlar peeve about modeling in "car position". as mentioned above by John, if you model in car position it is unmanagable to reuse a part. One should assemble in "car position" not model in car position. I beleive this is a leftover from modeling in Catia V4 which had feeble assembly capabilities but huge involvement in the car industry. Old habits die hard! I still encounter people that want to model in car position in Catia V5, which has passible assembly capabilities.

 

[looslib]

When I worked for an air bag supplier to the auto industry, we modeled the components to their own coordinate system and made the assembly by mounting all components to the base plate. We left it up to our customer to assemble this module into car position. This has been a while as we were doing this in UG v7-v9 and had only begun evaluating V10 when I left that company. All wireframe, except some parts at the end done in V9 solids. Assemblies and components was nothing like it is today!




"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli

 

[ewh]

edgray,
This was not a feeble left over from modeling in Catia V4 (which I have no experience with). Why would a part modeled in "ship" coordinates become "unmanageable"? The only problem that I have personally come across with this method is getting the machinists to accept the coordinates, which no longer becomes a problem with proper education and training (same as with the modeling/design aspect). Designing in "ship" coordinates allows more people to easily work with less geometry relative to the space in the aircraft it occupies. If I want to design a bracket in the vertical stabilizer, all I need do is add the affected and adjacent components to my layout without having to load larger files, even taking into account load options which still require a larger scope in which to work with. I have only the information at hand to effectively continue the design.
John does point out some weaknesses, and I have seen many of these addressed by adding constraints as the design matures. The gas cap example should be in the "common components" catagory I mentioned earlier. A tip tank or wing leading edge would not, as they are unique and usually "'fly in formation' with the rest of the Assembly".
As an aside... my first drafting boss was named Ed Gray. ;-)

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[JohnRBaker]

Just so that everyone understands, I was not arguing that designing in 'ship coordinates' is a bad approach or one which should be avoided. After all, in automotive, aerospace, marine (shipbuilding), etc, there are some good reasons for designing the MAJOR components relative to some common pre-defined X,Y,X origin and orientation, but like many things in life, the key is moderation ;-)

The primary reason for my commenting (when I first saw this thread I hoped to let others do all the 'talking') but when it was clear that the original poster was designing machinery, an area in which I do have experience (14 years as a machine designer, albeit only 3 using CAD) and that he was 'debating' the pros and cons of working in 'ship coordinates', I felt I needed to put in my 2 cents worth since capital machinery can be very different than automotive/aerospace/marine industries when it comes to how major subsystems and component relate to each other and what the most practical approaches might be, particularly when there are mechanisms involved (they're NOT called 'machines' for nothing). Also, machine design generally involves a higher percentage of common or reusable parts than does the other industry segments mentioned above, which, by definition, makes designing relative to a common origin/orientation not only impractical, but potentially problematic when later someone decides to leverage the fact that many parts were indeed common between different product lines.

Anyway, please don't take my comments so much as coming from a representative of Siemens PLM but rather as coming from someone who has spent time in the world of machine design and who merely trying to make sure that there was some clarity in understanding that not all situations were the same when it came to getting the maximum advantage out of a capability such as assembly constraints and that how one approached the overall problem could have implications later on when more value was being asked of the models, assemblies or otherwise, which had already been created.

Sorry if I confused anyone, it was NOT intended.

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Industry Sector
Cypress, CA

http://www.siemens.com/plm

UG/NX Museum:

http://www.plmworld.org/p/cm/ld/fid=209

To an Engineer, the glass is twice as big as it needs to be.

 

[ewh]

No confusion here, John... I whole-heartedly agree that with other types of design, such as machine design, "ship" coordinate type modeling is unnecessary and may not add any value, and all things are better in moderation.
My response was more of a knee-jerk reaction to a statement which was based on the posters poor understanding of the methodology in question (which happened to be in the automotive industry).

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[ewh]

Just a bit more explanation of modeling in aircraft coordinates...
Many different disciplines are involved in the design of an aircraft (or automobile), and it is very important that they each have confidence that the geometry that they are working with truely represents what and where the object is in aircraft 3D space. There is no luxury of picking parts from a pile and assembling them to come up with a working aircraft. Each part and subsystem relys on the correct location of the many other components to be designed properly. When you have hundreds of designers involved using absolute coordinates as their work coordinates, that can be difficult.
Using aircraft (absolute) coordinates, whoever needs to design a new part knows that every other part he may need to work with is properly located in 3D space. Work coordinates are relative and are easily manipulated. Absolute coordinates do not change.
Aircraft components aren't designed one piece at a time, but as components of a very complex system. You can't just sit and decide to design a control bracket without knowing the environment (location and relationships) in which it will be operating.
Modeling the components in aircraft coordinates is an added insurance policy that whomever may open and use a component in their design, the likelyhood of that component being mislocated is lessened.
Think of the differences between painting a picture as opposed to putting together a puzzle. A puzzle is limited in that each piece must have a corresponding piece to which it can attach before you can continue (constraints). A painting can be approached from countless directions, filling in the gaps as you go to form a cohesive whole, and the location of what has already been painted won't change.
As noted before, different types of design require different approaches. While an automobile or an aircraft may be "machines", machine design CAD methodology does not necessarily apply. Again, common components (those used more than one or two places) do not need to, and should not, be modeled in aircraft (ship) coordinates.

(This is a "hard" forum, right?)




"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[edgray]

EWH
I have only worked with components far inside the machine. I think my point is that to me it makes sense to model the part where the base coordinate system is relevant to the part. with the capabilities of NX, where one can model either in part mode or assembly mode it seeems one can have the best of both worlds. I prefer to import the parts I mate to from the customer, in ship/car coodinates and use constraints to position my part in the vehicle space. I also think that my emphasis is on using constraints when in assembly space. I have received too many sub components with small bu significant mis positioning of sub parts. if the modeler of the sub component had used constraints I would never have had the problem.

I doubt it was me. I don't think I have ever been anyones boss. There are very large number of Ed Grays out there but I have never met another one.

 

[ewh]

if the modeler of the sub component had used constraints I would never have had the problem

You haven't had to work with many legacy assemblies yet, have you?

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[edgray]

I admit that proper modeling proceedures would also have avoided the problems, but so would assembly constraints and or mating conditions.

The models that I was having the problems with were from Catia, so the terminology is not right but the concepts are appicable.

Legacy data is always a challenge. In fact i have been the senior CAD designer here for over twenty years. dealing with several migrations from previous CAD systems and operating systems.

Back to the point. Proper CAd practices are for the purpose of making the process easier they should never tak the place of engineering proceedures. Then again never is a very long time.

 

[alj722]

If you are inserting a sub-assembly into an assembly, that sub-assembly acts as a single entity as long as it is not the work part. The sub-assembly doesn't have to be constrained within itself as long as you don't mess with it and you can still constrain it at the top level assembly. Once you make the sub-assembly the work part, all bets are off.

I've avoided assembly constraints because they sometimes have a mind of their own. I don't know how many times I have an assembly turn into a hairball for no apparent reason. I use them when I have to but often delete them when the design is settled just to avoid them behaving like a transformer and turning into a monster.