# What is the correct sequence for getting curved roofs right?

## Recommended Posts

I am pulling my hair out trying to get a little test plan to give me some roofing numbers.

I've a 4.5:12 roof that gets a 30" overhang that flares with a 9' radius, point of tangency at the building line.  The angle of the 4.5 roof is 20.556045°.

Am unable to get the curved roof to rotate at its ridge line while keeping the required radius.  But the plan I want to apply my numbers to has curved flares everywhere so I was able to get it to work before.  What am I doing wrong right now?

We should be able to examine the main roof's specs to get its angle in degrees, copy that, go to the flare roof overhang, lock its ridge top height (the point of join, the point of tangency), set the radius (which I want to fix at 108" and Chief wants to call negative 108), and have the roof build.  It rotates about that specified and locked ridge height.  But it won't build and hold the radius and top angle.  Why not?

An interesting thing about these is that when you get it working right, the sequence of specs so your curved reachout is good, you then have to manually go in and lower the roof to have its top edge match the roof it is to join.

##### Share on other sites

Followup.  Shouldn't one be able to lock the ridge, specify a radius and top angle, and let the roof build to whatever bottom angle and fascia height gets determined by the program?  I cannot get it to work.

##### Share on other sites
14 minutes ago, GeneDavis said:

Followup.  Shouldn't one be able to lock the ridge, specify a radius and top angle, and let the roof build to whatever bottom angle and fascia height gets determined by the program?  I cannot get it to work.

It's really not difficult if you know exactly what you want to achieve. I'll use a section view and use cad lines as a reference to refine the data. Identify all the coordinates of your desired end result:

where the curved plane connects to another roof (x or y coordinate) and z coordinate

where the bottom of the curved roof will end (x or y coordinate) and z coordinate

desired radius of the curved roof plan and...(x or y coordinate) and z coordinate

Compare this info to what's being used in the roof plane dbx. Then, fairly simple to adjust as all you'll need to do is lock the pitch and enter the radius using a negative number.

##### Share on other sites
2 hours ago, GeneDavis said:

little test plan

Can you post it?

##### Share on other sites

For complex curved roofs, I usually do roughly as Robert described and use some CAD to get all the numbers I need.  Commonly I do so off to the side right there in plan view.  Once I have my roof shape drawn in CAD, I copy the relevant numbers one by one and paste into the appropriate places in the roof plane settings.

Also, it's super helpful to draw your baseline at a good and appropriate location.  I suggest you ignore the wall and just draw the baseline where it can be used to better control the roof.  For complex roofs, my baseline may be located at the ridge, it may be located in the middle of the room at some beam reference, etc.  For curved roofs in particular though, I commonly locate it at the very bottom of the roof plane (no overhang beyond baseline).

##### Share on other sites

Been there done that, but when I get the o'hang to curve to the 108 radius, the top angle does not match that of the abutting roof above, and I believe that for true tangency, it should.  That is why I want to be able to spec the overhang by ridge and radius and let the rest calculate.

Here is the plan.

Curved roof edge study.plan

##### Share on other sites

I've had to do a fair amount of these flared eave roofs now and while I can usually get what I want, I have to say that it's definitely not as intuitive as it should be... and I know with my current knowledge I wouldn't be able to match an existing condition with an autobuild using the walls 'roof' panel to set the lower pitch then curving that lower pitch.
Since you guys seem to have it under full control, a few questions:
My mindset has been don't add the flared eaves until late in the design because they just add a ton of time to any changes that get made.  So that means starting with roofs set to the Upper pitch so the look is mostly like what you'll end up with, sans the roof flares.  BUT.. I do wonder if there's an advantage to starting with what would be the Lower pitch...and then curving it and then manually building the upper pitch by snapping to the lower planes ridge?

Something...More to Gene's question that you guys haven't mentioned;

Let's say I've got my lower and upper roof pitches defined using the wall dbx roof panel, and I now select my lower roofs to set their curve.  I LOCK the RidgeTop Height... and I check the Curved Roof checkbox, then I enter the Angle at Ridge to match the angle of the Upper Roof pitch, and the Radius I want...  at THIS point, I should have defined everything necessary for this lower roof to form as I would like it to... but it seems that the the lower roof baseline/plate sticks whether you want it to or not.

From what I see... the reason for this is that the Lower Roof Pitch over-rides everything else that's happening in the 'curved roof' settings even if you do not have the pitch locked.

My solution has been that I always just let the curve radius be what it wants to be and I get the look of what I want by setting that lower roof pitch rather than a radius.

##### Share on other sites

Depending on the structure type, I use a couple of priorities to set a curved roof.

1. height above the exterior corner of the wall plate must be a whole, even number in inches

2. radius must be a whole, even number in inches

3. upper roof pitch must be a typical roof pitch; that is to say a whole number in 12.

4. curved roof pitch not be less than 3:12

5. A Radius over 10' tends to be too subtle IMO.

So, in a section view draw the upper roof pitch line and a 3:12 roof pitch line intersecting and experiment with a circle that fits nicely, sticking to rule #2. Then, plot them using rule # 1 and boom, set the roof planes. Easy peezy, lemon squeezy!

##### Share on other sites
14 minutes ago, robdyck said:

Depending on the structure type, I use a couple of priorities to set a curved roof.

But notice... although minor, your upper roof slope is 45.0° and your curved roof slope at ridge is 43.126°.
Also notice that you said pitch must be a whole number, but yours is 5-5/16:12 ?
If you change the curved roof slope at ridge to  45.0° then the radius you set changes.  How to set radius AND slope at ridge?
What is your understanding (if any) of the relationship between the roof slope and the curve radius?

##### Share on other sites
23 minutes ago, DzinEye said:

Also notice that you said pitch must be a whole number, but yours is 5-5/16:12 ?

The upper roof pitch, not the curved roof pitch. I believe Gene's original plan had an odd number for the upper roof pitch.

26 minutes ago, DzinEye said:

and your curved roof slope at ridge is 43.126°.

Who cares?! (not meant rudely)

I don't even look at that. All I'm looking at is the smoothness of the transition, and for simple number to provide to the carpenters. Radius, sub-fascia elevation, and the intersection are all they need.

##### Share on other sites
36 minutes ago, DzinEye said:

What is your understanding (if any) of the relationship between the roof slope and the curve radius?

You may be onto something by insinuating I have a complete lack of understanding...at least that's how I'm reading it. I thought you were a nice guy?

This is where typing breaks down and a voice conversation tool in the forum would be awesome. I'd like to explore this further. For that relationship, Are you referring to the upper roof slope, or the slope of the curved roof?

##### Share on other sites
9 minutes ago, robdyck said:

You may be onto something by insinuating I have a complete lack of understanding...at least that's how I'm reading it. I thought you were a nice guy?

You got an honest to goodness LOL out of me there.   Simply wondering if you are aware of the relationship, or if there is one.  It seems to me the slope setting over-rides other settings in the curved roof part of the dbx.

9 minutes ago, robdyck said:

Fo﻿r that relationship, Are you referring to the upper roof slope, or the slope of the curved roof?

The curved one

##### Share on other sites
17 minutes ago, robdyck said:

Who cares?! (not meant rudely)

Because you very carefully pointed out that those all had to be set at nice WHOLE numbers?!?!

##### Share on other sites
Just now, DzinEye said:

Because you very carefully pointed out that those all had to be set at nice WHOLE numbers?!?!

Ahh. I guess I should clarify. I'm referring more to the design criteria than the roof dialog box. In fact, I don't even use a roof plane until I know what it'll look like and then, I'm ignoring almost every option in the dialog box. I can then draw the roof plane over the wall, set its fascia height, set the ridge height, set the radius, then done.

##### Share on other sites

My preferred "whole numbers" are shown in the green clouds in the image below.

##### Share on other sites

It is a minor little thing in the whole scheme of building, but the curvature generally specified in a section view of eaves such as this, at least in the few plan-sets that I have encountered (likely not drawn in Chief but instead in Revit or AutoCad) dimension it out as a.) tangent to the upper slope, b.) clearly showing the point of tangency, and c.) the radius of the curve.  And that radius is always shown to the top of the rafter.

The curve follows and is the sawn top edge of that rafter.

But Chief does not do the curve that way.  The curve begins at top end at the top of the rafter, but it begins to deviate away from the rafter immediately as it moves downslope, and it ends precisely at the tip, the outboard top tip, of the fascia.

So for Chief, two other things come into play in developing the curve.  The thickness of the sheathing, and the thickness of the fascia.

So if I do a section of the eave for con docs, annotated for building, I show, as the Revit- or ACAD-user has done in the detail shown here, the overhang reach, the point of tangency at the building line, and the radius of the curve, with my callout arrow pointing clearly to the rafter top.  To make a pattern, the cutter will either loft it out on the floor with a beam compass and some straightedges, or he might use his BuildCalc phone app to get a half dozen chord points and then draw it using a thin rip flex stick, and that'll be it.

But it won't be the actual curve Chief develops for the 3D model.

Note in the attached pic, no fascia height is given.  Plate height, yes.  Heel height, yes, because the slope (4.5") and framing (2x12) is given.  But all that is there is radius and reach.  Not my drawing.

And I can tell this is a copied-in CAD detail from something the architect must have done previously.  See the leftover note about "existing second floor?"  This is a new build, and there is no second floor.

##### Share on other sites

@GeneDavis
If you're only given the information you mention, I think you'd have to draw it in 2D to figure out what the slope of that lower portion of the roof will be, from eave to tangent point.  Then use that slope for the lower roof slope setting... and set your radius to match the spec. and lock the ridge height.
In Chief, the slope setting for the lower roof is really the main driver of how things come out...at least it's the most controlling.  It stays fixed no matter what other things you change, and any radius changes take place between the points where it bisects the upper roof and the top of fascia.
Also note that the plate for the lower roof  as it's shown in your detail is totally irrelevant... so just base it on the ridge height matching the upper (main) roof tangent point.

##### Share on other sites

Just another quick note...We don’t actually have to use actual curved roofs at all.  For complex curved roofs, I will commonly use a series of standard flat roof planes with heights and pitches set appropriately.  It’s like manually positioning each facet instead of changing the facet angle.  Again though, I draw the whole profile in CAD and take all my numbers from that.

##### Share on other sites
6 hours ago, GeneDavis said:

The curve follows and is the sawn top edge of that rafter.

But Chief does not do the curve that way.  The curve begins at top end at the top of the rafter, but it begins to deviate﻿ away from the rafter immediately as it moves downslope, and it ends precisely at the tip, the outboard top tip, of the fascia.

I noticed that in your previously posted file on the other thread but I was able to correct it so that it did follow at the 108" R. which I then joined at the hip to get the 207 11/16" R. hip radius.  I found it was best to check it by creating a cad detail of the section and then snapping an arc to the top of the rafter.  It seemed to work fine.

##### Share on other sites

Just got back to the office and had a chance to open the provided example file.  Not sure I've ever made a video for you before Gene, so here's your first custom tutorial on-the-house...

• 2

##### Share on other sites

N﻿ot sure I've ever made a video for you before Gene, so here's your first custom tutorial on-the-house...

Thanks for making and sharing that Michael.  After I decided to really dig into understanding this yesterday I figured out, just as you suggest here that the slope (or chord angle) of the curved section of roof is the key to the whole thing.  Great idea to grab the chord angle from the 2D cad.