What is the centre-to-end dimension of a long radius elbow?

One and a half times the nominal size, in inches — a 6-inch long radius elbow is 9.00 in centre to end. That holds at 31 of the 33 sizes ASME B16.9 lists. It fails at the two smallest, NPS ½ and ¾, which are both held at the 1-inch length of 1.50 in rather than shrinking with the pipe.

What is the difference between a long radius and a short radius elbow?

The centreline radius. Long radius is 1.5 times the nominal size, short radius is 1.0 — so a short radius elbow turns in a tighter space and a long radius one is gentler on flow. The short radius table starts at NPS 1 and the identity holds on every size it lists, with no exceptions at all.

Is an EN 3D elbow the same as an ASME 3D elbow?

No — they are a factor of two apart, and the names are the trap. EN counts the bend in nominal diameters, ASME counts it in radii, so an EN "model 3D" is an ASME long radius while an ASME "3D radius" elbow is twice as long. At DN 25 that is 38 mm against 76 mm. Ordering one and receiving the other is a mistake made entirely on paper.

How long is a concentric reducer?

It depends only on the large end. A 12 × 10 and a 12 × 5 reducer are the same length, because B16.9 gives one end-to-end dimension per large end and the small end never moves it. The length also jumps sharply at one point in the range: from NPS 12 to NPS 14 it goes from 8.00 in to 13.00 in.

Do butt-weld fittings have a wall thickness in the standard?

No, and that surprises people. B16.9 does not specify the wall away from the ends — the fitting is made to match the pipe it welds to, and the design decides. So there is no fitting wall thickness table to look up, and any chart offering one has taken it from somewhere other than the standard.

Pipe Fitting Dimensions

Butt-weld fittings to ASME B16.9. A long radius elbow is one and a half times its nominal size, centre to end — except at the two smallest sizes, where it is not. And an EN “3D” elbow is twice an ASME “3D” elbow, because the two standards count the bend differently.

The Rule, and the Two Sizes That Break It

A 90-degree long radius elbow is defined by a centreline radius of one and a half nominal diameters, so its centre-to-end dimension ought to be 1.5 × NPS at every size. Across the 33 sizes B16.9 lists, it is — at 31 of them, exactly, including the fractional sizes.

It fails at NPS ½ and ¾. Both are held at the 1-inch elbow’s 1.50 in rather than shrinking with the pipe, so a ½-inch long radius elbow is very nearly twice as long as its own definition.

The centreline identities B16.9 was tested against, and where each one stops holding. Computed from the standard’s own tables.
Identity	Holds	Of	Fails at
A = 1.5 x NPS	31	33	½, ¾
B = 1.5 x NPS x tan 22.5deg	13	33	20 sizes
A = A of the long radius elbow at the LARGE end	54	54	no exceptions
O = 2 x A of the long radius elbow	20	21	1 ¼
A = 1.0 x NPS	19	19	no exceptions
O = 2 x NPS (twice the short radius elbow A)	19	19	no exceptions
A = 3.0 x NPS	32	32	no exceptions
B = 3.0 x NPS x tan 22.5deg	32	32	no exceptions
C = C of the straight tee at the RUN size	213	213	no exceptions
R = 1 x DN	3	31	28 sizes
R = 1.5 x DN	3	31	28 sizes
R = 2.5 x DN	0	31	31 sizes

The short radius and 3D-radius identities hold on every size they list. It is the long radius series, the commonest fitting in the catalogue, that has the exceptions.

An EN “3D” Is Not an ASME “3D”

This is the most expensive fact on the page, and it is entirely a naming problem. EN 10253 counts a bend in nominal diameters; ASME B16.9 counts it in radii. So an EN model 3D elbow has a radius of 1.5 diameters — which is an ASME long radius elbow — while an ASME 3D elbow has a radius of three diameters and is twice as long.

The same name on both standards, measured. Every row here is exactly a factor of two.
DN	EN model 3D (mm)	ASME 3D radius (mm)
25	38	76
40	57	114
50	76	152
65	95	190
200	305	610
250	381	762
300	457	914
450	686	1372
500	762	1524
550	838	1676

The two carry different type names in our data for that reason, and are never compared as though they were the same fitting. For the related trap in the pipe itself — where DN and NPS name different diameters — see what DN actually measures.

A Note in the Standard That Fails in the Standard

Tables 6.1-3 and 6.1-5 of B16.9 carry a note saying that dimension A is one half of dimension O. In the inch column that holds on 20 of the sizes listed. In the millimetre column printed immediately beside it, it fails on 9.

Nothing is wrong with the fitting. Each column is rounded independently from the same underlying quantity, and halving a rounded number is not the same as rounding a halved one. It is worth knowing because it is the clearest demonstration available that a standard’s two unit columns are two separate documents that happen to describe one object — and that reading across them will eventually produce a contradiction that is nobody’s error.

A Reducer’s Length Ignores Its Small End

B16.9 gives one end-to-end length per large end, and the small end never changes it. A 12 × 10 and a 12 × 5 reducer are both 8.00 in long. That is a rule rather than a coincidence, and the build refuses to publish this page if the data ever stops supporting it.

End-to-end length by large end, ASME B16.9 concentric and eccentric reducers. The small end appears nowhere in this table because it makes no difference.
Large end	End to end (in)
1	2.00
2	3.00
3	3.50
4	4.00
5	5.00
6	5.50
8	6.00
10	7.00
12	8.00
14	13.00
16	14.00
18	15.00
20	20.00
22	20.00
24	20.00
26	24.00
28	24.00
30	24.00
32	24.00
34	24.00
36	24.00
38	24.00
40	24.00
42	24.00
44	24.00
46	28.00
48	28.00
¾	1.50
1 ¼	2.00
1 ½	2.50
2 ½	3.50
3 ½	4.00

The series is not smooth. Between NPS 12 and NPS 14 the length jumps from 8.00 in to 13.00 in — a step of 5.00 in, larger than any other in the range.

There Is No Fitting Wall Thickness

B16.9 does not specify the wall of a fitting away from its ends, and it says as much: the bore between the welding ends is not a specified dimension. A butt-weld fitting is made to suit the pipe it joins, so its wall follows the pipe’s schedule rather than a table of its own.

This matters when reading charts elsewhere. A fitting chart that offers a wall thickness column has taken it from somewhere other than the standard — usually by assuming a schedule on the reader’s behalf. For the schedule that actually governs it, see how the pipe schedule system works and schedule 40 pipe dimensions.

What We Could Not Verify

ASME B16.9-2018 is held here, so its rows are judged against the standard itself — 672 of 1035 rows are verified. EN 10253 is a third-party repost with no chain of custody, so no EN row rises above single-sourcehowever cleanly it checks out. That distinction is kept in the data rather than blurred. How this data was verified.

EN 10253-2:2007 Table 14, reducing tees: The G column is vertically merged across two to four outlet rows and the span is not recoverable from the text layer. Guessing it is the silent parser bug this project has shipped four times. Not compiled at all.
EN 10253-4 (draft): prEN 10253-4:2003 is a DRAFT and may never anchor a published row.
EN 10253-3:2008: Carries tolerances only; its dimensional tables are those of part 2.
EN 10253-1:1999: Same dimensional data as part 2 — see clone_evidence. Read to prove the bloc, not published as a second source.
EN 10253-2:2007 Table 13, the four largest equal tees: DN 1050 to DN 1200 print F with a footnote marker rather than a number because the outlet branch is shorter than the run. The rows are not compiled; the footnote's own values are read into en_equal_tee_outlet_exceptions.
EN 10253-2 Table 16 conditional cap rows: Rows from DN 700 upward print two heights with a wall-thickness condition in prose. Only unconditional rows are compiled.
Wall thickness of any fitting: B16.9 does not tabulate it — clause 2.2.2 leaves it to design, and clause 6.2.2 says bore diameters away from the ends are not specified. There is no fitting wall thickness to publish.
Tolerances (B16.9 Table 11-1, EN Table 19): Not parsed in this pass.
MSS SP-75, ASME B16.28, B16.11: No primary document held.
the sixteen vendor pages in _sources/fittings/firecrawl: Secondary republications of this same standard. MANIFEST.md section 4 found five clone clusters among them in one pass. With the primary standard in hand they add nothing and are not read.

One more worth stating plainly: the standard itself contains a single-cell typo, found by arithmetic rather than by reading. One reducer prints its small end as 355.4 mm where every other appearance of that size in the same document — including a table three pages earlier — prints 355.6, which is what the inch value converts to. It is listed with everything else at errors found in published pipe data.

Every butt-weld end here is a real pipe outside diameter, checked against steel pipe sizes — a fitting that did not match a pipe would not weld to one. For the other component that bolts rather than welds, see pipe flange dimensions, and for the full diameter table, the full pipe size chart.

Frequently Asked Questions

What is the centre-to-end dimension of a long radius elbow?: One and a half times the nominal size, in inches — a 6-inch long radius elbow is 9.00 in centre to end. That holds at 31 of the 33 sizes ASME B16.9 lists. It fails at the two smallest, NPS ½ and ¾, which are both held at the 1-inch length of 1.50 in rather than shrinking with the pipe.
What is the difference between a long radius and a short radius elbow?: The centreline radius. Long radius is 1.5 times the nominal size, short radius is 1.0 — so a short radius elbow turns in a tighter space and a long radius one is gentler on flow. The short radius table starts at NPS 1 and the identity holds on every size it lists, with no exceptions at all.
Is an EN 3D elbow the same as an ASME 3D elbow?: No — they are a factor of two apart, and the names are the trap. EN counts the bend in nominal diameters, ASME counts it in radii, so an EN "model 3D" is an ASME long radius while an ASME "3D radius" elbow is twice as long. At DN 25 that is 38 mm against 76 mm. Ordering one and receiving the other is a mistake made entirely on paper.
How long is a concentric reducer?: It depends only on the large end. A 12 × 10 and a 12 × 5 reducer are the same length, because B16.9 gives one end-to-end dimension per large end and the small end never moves it. The length also jumps sharply at one point in the range: from NPS 12 to NPS 14 it goes from 8.00 in to 13.00 in.
Do butt-weld fittings have a wall thickness in the standard?: No, and that surprises people. B16.9 does not specify the wall away from the ends — the fitting is made to match the pipe it welds to, and the design decides. So there is no fitting wall thickness table to look up, and any chart offering one has taken it from somewhere other than the standard.