Warren Truss viaducts on a curve

[philip-griffiths]

Hello,

Nine years ago I started building a warren truss viaduct for the 1883 competition. I ran into trouble with numerous hurdles (house move, new job, etc) and except for some odd moments of peace making some track I haven't had a chance to wield soldering iron or saw in earnest since 2002. This year I've tried to reinstate modelling time.

The subject on the bench at present is the viaduct. This would be simple if I had decided to build it straight, but I was trying to build the curved section of the Crumlin Viaduct. The straight section is easy (relatively); I have detailed drawings from a Victorian book, and have been able to produce the required CAD drawings. The pillars are straight forward too. I started with the construction the warren truss girders, though these days maybe a 3d-printer may be a better solution instead of plastrut.

However when it comes to the curve I'm stumped. I've looked at photos but they haven't answered my question - were the truss girders on the outside of the curve longer than those on the inner? They should be. The pillars are not wedged shaped to accommodate and change, while the surviving viaduct abutments look like their are uniform. Also I have nothing to tell me whether the outer girder of the viaduct was to the standard length of was the inner one? The photographic evidence shows that the girders are angled to each other so there must be a difference in length, but the diagonal struts look like they are forming the 60 deg angle and there seem to be the same number of struts in the outside girders as there are in the inside ones.

If there is a Civil or structural engineer or someone with inside knowledge, I'd be very grateful if I could get some pointers as to what exactly I need to do to replicate the Kendon section of the Crumlin Viaduct.

(for those of you who do not know about the viaduct, it was, until demolished, the UK's highest viaduct, at 200ft from the river, spanning the Ebbw Vale at Crumlin, carrying the Taff Vale Extension of the Newport, Abergavenny and Hereford Railway, later to become part of the GWR - google the viaduct and you'll find a website full of pictures but none really resolve my problem).

with many thanks

Philip Griffiths

[Alan Turner]

Look here: http://www.crumlinviaduct.co.uk/photo.htm

and at photos 26, 39, 50 and 64.

You will see the trusses were not curved.

To achieve the curve one truss would be longer than the other on the curved span.

I would expect the piers to be orientated to allow for this change in length.

regards

Alan

[craig_whilding]

To achieve the curve one truss would be longer than the other on the curved span.

regards
Alan

I think the question was related to how much longer or shorter the one side was from a standard length.

I had a look on Access to Archives and the Welsh archives but neither seems to have anything. I wonder if there are records anywhere related to the listing of the structure and the later BR move to overturn that listing? Would CADW/English Heritage have ended up with some sort of record?

I've had a look at some shots I have from the Daffodil Express excursion over there but those don't reveal much. Must have been a good trip that though, City of Truro at Crumlin High Level and a Pullman observation car getting to Neath!

Could you name the book you got those detailed shots from, i'd like to have a look at the shots too
.

[newport_rod]

Philip
Is there anything in the footage of the film Arabesque that might be of help?
Cheers
Rod

[Alan Turner]

I think the question was related to how much longer or shorter the one side was from a standard length.
.

That's easy to calculate/determine.

Draw a circle of the required radius and lines from the centre to the circle circumference. If you now draw straight lines representing the trusses between these lines this will give you the lengths of the trusses to achieve the circle.

Regards

Alan

[craig_whilding]

I think the question was related to how much longer or shorter the one side was from a standard length.
.

That's easy to calculate/determine.

Draw a circle of the required radius and lines from the centre to the circle circumference. If you now draw straight line representing the trusses between these lines this will give you the lengths of the trusses to achieve the circle.

Regards
Alan

Curves.jpg

This could work but only if there is an accurate plan of the pier positions and abutments to work off.

[Armchair Modeller]

Look on old-maps.co.uk. I typed in Crumlin as the place to search on and several 1:2500 maps of various dates came up. They should be good enough to work from - though I must say the viaduct doesn't look very curved to me, so you may be worrying unduly about this potential problem.

[LesGros]

Look on old-maps.co.uk. I typed in Crumlin as the place to search on and several 1:2500 maps of various dates came up. They should be good enough to work from - though I must say the viaduct doesn't look very curved to me, so you may be worrying unduly about this potential problem.

I did the same, then moved the locus to co-ords 321248 198520 The main Section does indeed look straight but a straight edge on the screen reveals a gentle curve, However the track on the south west spans is clearly more curved, It should be possible to use Alan's method of plotting, or Templot, to get accurate dimensions.

good luck with the project

[Armchair Modeller]

Look on old-maps.co.uk. I typed in Crumlin as the place to search on and several 1:2500 maps of various dates came up. They should be good enough to work from - though I must say the viaduct doesn't look very curved to me, so you may be worrying unduly about this potential problem.

Incidentally, the viaduct has a superficial resemblance to Bennerley Viaduct

see http://en.wikipedia.org/wiki/Bennerley_Viaduct

though Bennerley does not have Warren Truss spans, the design of the piers and overall appearance is very similar. There are drawings of Bennerley viaduct in one of the "Great Northern in the East Midlands" series - probably volume 3. This viaduct still exists.

[Alan Turner]

The Warren trusses of Bennerley Viaduct are the conventional way around. Those of Crumlin are inverted.

The position of the piers are determined from the geometry and essentially "split the difference" of the angle between the two truss spans. It follows the line drawn from the centre of the circle.

I assumed that you would not be slavishly copying the Crumlin prototype so there is a need to calculate this in the model in any event.

Looking at the 1952 map (because it has Grid lines and allows scaling) the radius of the curved section comes out at 654m. The spans appear to be 90m so this gives a difference in length of the inner and outer trusses (measuring a deck width of 15m) of 2040mm. About 27mm in 4mm.

As a general proposition the difference in truss lengths is given by: truss span (average or mid line) x width between trusses/curve radius.

Regards

Alan

[philip-griffiths]

Thanks for the replies. The film doesn't help as the views under the bridge were on the straight part.

Yes my trusses are straight but support a curved deck. While measuring them is straightforward, there is a problem in that the pictures, poor quality as they are, do not indicate where the extra length is added in to the trusses on the outside of the curve. There are 17 triangles within the truss on both the back and the front. They seem to line up.

If I had designed the layout it straight it I would have had no problem, as the dimensions given by Liddell in his Victorian book are excellent, but it wouldn't have fitted the space.

So for the Civil's expert out there, would a gradual lengthening of the dimensions work?

regards

[frizby]

Looking at the 1952 map (because it has Grid lines and allows scaling) the radius of the curved section comes out at 654m. The spans appear to be 90m so this gives a difference in length of the inner and outer trusses (measuring a deck width of 15m) of 2040mm. About 27mm in 4mm.

As a general proposition the difference in truss lengths is given by: truss span (average or mid line) x width between trusses/curve radius.

Regards

Alan

Thanks for the replies. The film doesn't help as the views under the bridge were on the straight part.

Yes my trusses are straight but support a curved deck. While measuring them is straightforward, there is a problem in that the pictures, poor quality as they are, do not indicate where the extra length is added in to the trusses on the outside of the curve. There are 17 triangles within the truss on both the back and the front. They seem to line up.

If I had designed the layout it straight it I would have had no problem, as the dimensions given by Liddell in his Victorian book are excellent, but it wouldn't have fitted the space.

So for the Civil's expert out there, would a gradual lengthening of the dimensions work?

regards

Not an Civil but an engineer .The pragmatic Victorian Engineer would have increased each truss segment on the outside span by 120mm (4½"), or possibly lengthened the outer by 60mm and shortened the inner by 60mm. This would have negligible effect on the engineering structure and would be very hard to detect on the photographs.

[Alan Turner]

One truss would be 1m longer than the mid span and one would be 1m shorter. This means that each end would either be 0.5m longer or 0.5m shorter. in 4mm that would be 6mm.

This is easily accomadated in the last truss section. Unlikly that the difference would be spreadout along the truss because that would mean each truss would be unique (don't forget the inner trusses) and hence very expensive to fabricate.

Only the sections shown in red would have been altered.



Alan

regards

Truss 2.PNG (6.78 KiB) Viewed 1420 times

[Armchair Modeller]

Don't forget that if you sharpen the curve,compared to the real thing, you will possibly need to make the piers wider and the girders further apart to maintain sufficient clearance for the trains

[frizby]

Not sure if these help - especially as still standing.

http://www.semgonline.com/structures/struct_55.html
http://www.transportheritage.com/find-h ... sobi2Id=65

Andy