on the Stratford-on-Avon Canal, near Bearley, west of Stratford-upon-Avon
General
An impressive canal aqueduct, having a cast iron trough supported on brick piers. Completed 1816.
Engineer: William Whitmore [1]
There are two much shorter aqueducts nearby, of similar design, at Wootton Wawen and Yarlingstone
Edstone is the longest canal aqueduct in England, and the second longest in Great Britain. It crosses a minor road, the former Birmingham and North Warwickshire Railway and also the trackbed of the former Alcester Railway. There was once a pipe from the canal trough that enabled locomotives to draw water to fill the loco's tank.[2]
The length of the iron trough is 497 ft 6 in., made up of fourteen spans of approximately 34 ft. The cast iron panels are 14 ft 2 in. long[3]
Edstone (or Bearley) Aqueduct is the longest of three iron aqueducts on a 4 miles (6 km) length of the Stratford-upon-Avon Canal in Warwickshire. All are unusual in that the tow paths are at the level of the canal bottom. The others are at Wootton Wawen Aqueduct and Yarningale Aqueduct.
These aqueducts are of simple design compared with the longer, higher and wider Pontcysyllte Aqueduct. The trough is assembled from bolted flanged rectangular cast iron plates. Additional support provided by cast iron trusses seated on brick piers. In contrast, the Pontcysyllte side plates are 'voussoir' shaped, and the troughs have additional support from cast iron arches.
Cast Iron Panels
The panels measure approximately 14 ft by 5 ft.
The foundry responsible for the iron castings has not been identified. It may have been William Whitmore's own foundry.
There are obvious surface defects in the castings which serve to illustrate the difficulties faced by iron founders of the day - and indeed now, 200 years later.
Photo 5 shows the honeycomb appearance on the flat side of the castings, showing that, for simplicity, the iron was poured into an open mould. Gas bubbles would rise towards the surface, and some would be trapped as the iron solidified. So, the side we see in contact with the water represents the top surface of the casting in the foundry.
The flanged face of the casting, corresponding to the bottom of the mould, shows two types of defect. Photos 6 & 7 show rough protrusions, these being the result of weak areas of the sand mould becoming detached. Ideally the detached sand would float to the top of the molten iron rather than being trapped inside. The other type of defect appears as unevenness, valleys, or apparent fissures in the casting. These features reflect protrusions or ridges on the face of the mould, caused by expansion and contraction of the sand under intense heating from the molten iron. These irregularities are sometimes known as 'sand buckles'.
Making large castings in open moulds is far from ideal. Much better to have closed moulds where the metal is poured in through channels known as 'runners', while other channels known as 'risers' allow gas, loose material, and excess metal to escape upwards. Having some depth of molten metal in the runners and risers means that the casting is pressurised to a small degree by the static head of the molten metal. This reduces the risk of pores or cavities being present in the casting. Using closed moulds would introduce its own problems, especially when moulds were large and mechanical handling provision was primitive.
This is not to imply that casting in an open mould is simple. One problem is that the mould must be absolutely level. Any slope will obviously affect the thickness of the cast plate.
The cast panels would have been made in a limited number of standard sizes for the aqueducts, effectively modular construction. This does not imply that the castings could be mass-produced, as each mould required careful preparation.
A simple wooden pattern would have been made, in the form of four sides of a picture frame. Referring to Photo 6, the top flange can be readily compared with the 'moulding' of a picture frame. Things get much more complicated with the side and bottom flanges, which have bosses and holes for the bolts. It is the forming of the holes which causes the difficulty. An explanation is beyond the scope of this brief entry, but suffice to say that the raised bosses (Photo 7) on the patterns for the flanges would be made so as to leave recesses (pockets) in the sand when the pattern is removed from the sand mould. These pockets would be used to support small sand 'cores', whose presence results in bolt holes being formed in the finished casting.
It must not be thought that the wooden pattern was simply 'pressed into the sand'. In fact the pattern had to be put in place on a carefully-levelled bed, and sand thrown in and carefully rammed against the pattern. Inadequate or excessive ramming of the sand, or careless removal of the pattern, could cause detachment of lumps of sand, resulting in defects in the casting - as seen here.
The pattern was likened earlier to a picture frame. The space within the frame would be open, ready to be filled with sand. The top surface of this infill would then have to be scraped down to a level which determines the thickness of the cast plate. The probable way of achieving this is described below.
The four sides of the 'picture frame' would be screwed to a rectangular wooden frame, the thickness of which corresponds to the required iron plate thickness (i.e. the thickness of the main panel, about 1 inch). The wooden assembly would be placed face down on the previously-levelled sand bed, and the sand infilled and rammed. The top face would then be 'strickled'. The strickle is a straight wooden batten, with a notch at each end, the depth corresponding to the thickness of the screwed-on wooden frame, and the width between the notches corresponding to the gap between the frame's sides. The strickle would be drawn across to remove excess sand, and any finishing done with a trowel.
Returning to the starting point, the levelled bed: nothing simple about this. A pit would be dug in the foundry floor, and the bottom roughly levelled. A cast iron grid would be placed on the floor, then a layer of coke, then straw, then sand. The coke and straw provide a porous layer for gas and steam to escape from the sand. The bed of sand would then have been made truly level, using wooden battens as a guide for a strickle. The sand nearest the top would have been of a fine grade with the aim of giving a good surface finish. Too fine and dense, and this facing sand can encourage sand buckles…..
Thanks to Dan Mackay for his insight into the probable method of moulding these superficially simple castings.
Photo 8: A curious feature, revealed by a missing bolt, is that the cored hole for the bolt is square, rather than round. Reason not known. Also note the small size of the flange, and the less than ideal amount of thread engagement on many nuts.
Other Iron Aqueducts
Edstone and its neighbouring aqueducts were preceded by others, most notably Longdon-on-Tern Aqueduct and Pontcysyllte Aqueduct. Like Longdon, Edstone Aqueduct has the defect of its trough being of small cross section in relation to the size of the boats (difficult for displaced water to get past).
The Longdon and Pontcysyllte troughs were assembled from a variety of shapes of iron panels, whereas Edstone's are all rectangular. Comparing the surface finish of the castings, those at Longdon appear superior to Edstone on the outer face. On the canal-side (upper face in the mould), the surface at Longdon is rough, as would be expected with an open mould, but differs from that at Edstone. It is thought that sand would have been thrown onto the molten surface immediately after casting the Longdon panels.