Helical Datum Curve? 1

[oneilldm]

How can I create a helical datum curve? Or how about a helical pipe?

Thanks

 

[Skiddy]

to create a helical datum curve you will need to create a datum from equations:

in cylindrical coordinates:

r=IR + t * (OR-IR)
theta=t*360*Rotations
z=A+(t*height)

ir= inner radius
or= outer radius
A=starting height off z plane
height= distance and direction can be +/-

egs cone:
r=3+t*7
theta=t*(3*360*5)
z=0+(t*30)

wavy circle
r=4
theta=t*360
z=1*sin(360*t*5)

or cartesian coor:
x=20/2*cos(t*360*4)
y=20/2*sin(t*360*4)
z=50*t+1*sin(360*t*100)

hope this helps???

 

[oneilldm]

Thank you so much Skiddy. This is great!

 

[im4cad]

You can also create a helical pipe as a helical protrusion (same as a spring).

im4cad
Pro Design Services, Inc.

http://www.cadproe.com/pds/home.asp

 

[HotSteam]

Skiddy,

Thanks for the tip on datum curves from equations. I found it great but Pro/E 2001 wont let me use a curve with a varing z as a swipe path. Whats with that?

Also conditionals(if else), floor and ceil function can not be used in path equations. I realize that these functions can create discontinuties. So what.

I am modling a small steam generator and wish a specific pattern in the coils. I wonted to define a (healical like) sweep where the z is constant for all but 2"(or what ever the transition distance) of the wrap. The transition from one wrap to the next is made in the last 2"s of the wrap. I bought the personel version of Pro/E which turns to be a student version. Is it crippled in this area?

Andy Patterson

 

[noppenbd]

If the path is continous and tangent, use a swept blend. You will have to sketch the cross section more than once, but the curve can be 3-dimensional (varying Z).

Since in real life there will likely be a smooth transition from the flat part of the coil to the changing Z part, you should try to model that in. Therefore all your path segments should be tangent.

 

[mjcole]

Andy,

YOu should be able to use a Swept Blend or Variable Section Sweep to use your curve as a path. You would only need to set up something to control section orientation for Variable sweep. I think the easiest thing for you to do would be a swept blend which has an auto orient option.

If you are using a closed curve for your trajecory you must split it using Datum Curve Split using a point on curve and keep both sides. The other thing you need is to sketch two sections the minimum allowed for swept blends. Just pick the start and endpoints for your section and link Section 2 to Section 1 dimensions if needed.

You should be able to create a surface to drive the z value of a curve with a relation that controls surface height. You can do this by choosing a flat xy curve as the orig trajectory and adding a relation for z.

I suggest trying to use the Swept Blend feature first because it's the easier of the two advanced sweep features.

Let me know if you have any questions.

Michael

 

[HotSteam]

Thanks. I will try the sweep blend again. Had truble geting it to work.

I modeled the curve in mathcad first and thranslated that to the Pro/E script. The steam generator is made of of concentric wraped coils of tubing. I wonted to do n layers but don't think that will work. I need floor and ceil to work to do that. I used a sin function to model the transition bend from one coil to the next. The sin was made to vary from -90 to +90 through the transition so there was a smothe transition. But even the most simplest z function doesn't work with a simple sweep.

r = 1.625
thata = 360 * t
z = 0.30 * t

But make:

z = 0

And it works.

OH:

I am trying to use a thin protusion sweep.

I am new to Pro/E. Learning it on my own. I am a software engineer with lots of math background.

Andy

 

[noppenbd]

If you can't get the swept blend to work with your cross section, try creating a construction surface using swept blend with a single line as your cross-section. Then you can do a Var Sec Sweep using your original path as the main trajectory and the edge of your cons. surface as the X-Trajectory.

 

[HotSteam]

Thank you all for the help. The swept blend works. But I can not get it parameterized and have it work right.

I would like to generate m coils. I tried using a pattern but the best I could get is about 5 coils. It's hard to explain but after sumnumber of generations the sweep goes haywire. It's a thin circle. I made two datum curves. First one is a circuler arc z constant z = 0. Then the transition arc which is:

z = spacing * (sin(180*t - 90)+1)/2

The two datums make a total of 360 degrees.

It is this second curve.

I had to put my profiles at an offset from the curve in order to have a dimemsion to vary in the pattern. At first I had a lot of problems with the section profile distorting(not being normal to my curve and seamed to be shifting out, r increassing although r was defined is constant in the curve equations).

I was able to generate an instance of my coil by making all the datum curves needed to generate the instance and doing one swept blend. But I really would like to be able to paramertize the number of coils and layers in the generator.

 

[noppenbd]

I am not sure I understand your design intent exactly. But one solution that works for difficult to generate patterns is to use surface transforms. Generate the first coil+transition as a surface sweep rather than a solid. If needed, merge multiple surfaces together that you wish to pattern. Then Surface/Transform/Copy the quilts. You should be able to pattern the transform easily. Then create solids by Use Quilt. You can pattern the solid created from the first Surface Transform as well.

 

[HotSteam]

Hi noppenbd

I don't think what I am doing is all that hard to pattern. It just doesn't work right. It may be in the way I created the origional part or it is a bug in Pro/E.

I created the origional as a sweep blend along two datum curves. The first curve has a constant z. the second datum curve makes a z transition of the spacing between the tube coles.

I did a swept blend along the two curves.

I made a pattern of the swept blend to generate more coils. The pattern was parameterized in the relations to generate n coils. But at some n 8 or 9 the pattern generates deformed coils. The coils are fine up to that point. I have tried several things grouping the datum curves and the swept blend and patterned the group. That did the same as just patterning the swept blend. The only thing that seams to work correctly is to make datum curves for the full length of the coil and then one swept blend along that path. But I can not pattern grouped data curves. Or at least I see no way to offset z in the pattern. And if I could how would one do a swept blend a long a patterned set(group)of datum curves.

If I could use conditionals, floor and ceil in my datum equations I could generate the whole path. But I get an error when I us an if statement or a floor or ceil function that I can not use these in a datum equation.

I spent a day working out equations in mathcad to generate a n(wraps) x m(layer) coil. I wont to be able to parameterize.

n number of wraps in a coil layer.
m is number of layers.
ID is in side diameter of the inermost coil layer.
tube_od is the outside tube diameter
wall is the tube wall theckness
hcc is the tube spacing center to center along the z axis
vcc is the radial tube spacing center to center (layer to layer).
trans is the radial length that the transition. The length of the two datum curves ard calculated from this parameter.

There is also a transition curve between layers that of course if different. And the normal transition curve in odd numbered layers is the negitive of the even layer transition curve.

Some of my initiual equations might show more of my intent.

rc = floor(t * m)*vcc + (tube_od + OD)/2

Is the radus of the the coil layer except during the layer to layer transition period. The transition occured during the last part of the last coile of each layer. The layers have a stagered pattern. odd layers have n - 1 wraps to a coil so the layer to layer transition point is a bit complicated. the calculation of z is also a bit complicated depending on odd or even layer. z is increasing in hcc steps on even layers and -hcc steps on off layers.

if (floor(t*m/2)==floor(t*m)/2)
/* even layer
wraps = n
else
/* odd layer
wraps = n - 1
endif

I will try and figure out your transform copy quilts method. I have only been working with Pro/E for a couple of months and that is not full time.

Andy

 

[HotSteam]

I got it to pattern by inserting a Coordinate System which I use as the rederance for the Datum Curves. Create a swept blend. Group the Coordinate System, datum curves and the swept blend and pattern that.

No distorting of any of the patterened objects. Now I need it to pattern the diameter of the curves.

Andy