The Refurbishment of Union Chain Bridge: Part 2

Alison Church
Good afternoon, everyone, and thank you for joining us today. We have a lot to cover in this webinar, so I’ll pass straight over to Ryan Convery and Simon Rudman from Northumberland County Council.

Ryan Convery
Hello everyone. Welcome back for Part Two of the Union Chain Bridge presentations. Today we’ll give an overview of the works undertaken during the refurbishment and provide insights into the challenges we encountered and how we resolved them. There’s a lot to get through, so we’ll get started.

Here’s a reminder of the bridge before the works, with the Scottish end toward the bottom of the image. As you may recall from the previous presentation, we also showed an archive photo illustrating flood risks at the site.

The original proposal for removing and reinstalling the bridge involved using a crane positioned on a pontoon in the River Tweed — a heavily used salmon fishing river and a protected watercourse. This approach severely restricted the working period due to environmental limitations associated with fish‑spawning seasons.

Instead, we installed an overhead cable highline system, using a Blondin‑type arrangement commonly seen in quarrying and forestry. This allowed for controlled movement of loads. It was used alongside an under‑deck gantry, man baskets and rope‑access techniques. The system was temporarily anchored to the rock face on the English side, routed over a steel temporary tower behind the Scottish tower, and connected to temporary pile foundations. This created a safe elevated route for removing and reinstalling the bridge, eliminating the need to work within the watercourse. As a result, environmental risks were greatly reduced and we had a reliable temporary‑works system that could operate year‑round.

Simon Rudman
To put that sketch into real‑life context, here you can see the under‑deck platform, which acted as a fall‑arrest system. Above the deck are two man baskets and the overhead cable crane.

The deck was dismantled from the English end, working toward the Scottish side, taking down what we called “one bay at a time”, a bay being three hangers’ worth of deck. All components from each bay were removed together.

We gradually worked toward the Scottish tower. You can see from the condition of the river in these photos that choosing an aerial ropeway was absolutely the right decision. We didn’t lose a single day of work to flooding.

Once the deck was fully removed, only the chains remained. These were removed by connecting each chain rod to a supporting catenary wire via a strop and pulley. Once all rods were connected, the catenary rope was tensioned to lift the chain rods slightly, allowing the load to be reduced and the connections released. The pulleys were linked with a haul rope, so once disconnected at the Scottish end, each rod could be hauled back in and removed in turn.
Workers did need to walk across the chains initially to install the strops and pulleys, but once the system was established, the process became straightforward. We repeated it twelve times until no superstructure remained. At that point, we could begin addressing the other major challenges.

Ryan Convery
These were the challenges we faced for the remainder of the project, which we’ll now go through. As mentioned previously, the wire rope was fully removed from the scheme, so you won’t see it reinstalled during reconstruction.

Anchorages
Ryan Convery
This schematic shows the new arrangement for the English anchorage. Rock anchors were installed through a new interceptor beam. After the beam was cast, the chains were connected to anchor plates fixed in place with Dywidag bars.

Here’s the English tower with the chains connected to the original anchorage. To install the new arrangement, the top section of the tower was removed, the original anchorage was cut, and the material behind it was excavated to provide access for the new design. The interceptor‑beam reinforcement cage was prefabricated in the compound, lifted into place, and then cast. The rock anchors were installed through it. The exposed Dywidag bars at the front were then used to connect the anchor plates. The chains were attached to the outstand plates through spade ends of Macalloy bars, which Simon will discuss shortly.

Here is the Scottish anchorage‑block design, which functions as a dead weight resisting the chain pull. A cofferdam was installed behind the Scottish tower, and the ground was excavated to install the reinforcement for the anchor block. Approximately 800 tonnes of concrete were placed in two pours. The temporary works were removed, the area was backfilled, and the front face of the block was exposed, revealing the Dywidag bars ready for the anchor‑plate installation, similar to the English side. Inspection‑chamber walls were installed, the anchor plates were fixed, the backstay chains were installed, and the concrete lids were cast around them.

Masonry
Ryan Convery
Here are the English and Scottish towers before refurbishment. We produced sample panels to finalise the mortar mix and confirm the proposed repair methodology, with Historic England involved in approving the approach.

Both towers were scaffolded to provide access. Repairs included replacement corners on the English tower and indents on the Scottish tower. Targeted removal of cementitious mortar was carried out to avoid unnecessary damage to historic stonework.

These photos show the towers after renewal. Over time, the replacement stone will weather and blend in with the existing masonry.

Handrail
Ryan Convery
The handrail was removed, inspected, tagged and sent to the conservation metalworker for repair. These photos show typical damage at connection points caused by bridge movement and hanger interaction.

Here are the removed rails awaiting repair, along with replacement rail ends to be welded where the originals were too damaged. Further examples include cracking within the rails and several posts requiring replacement feet.

After repairs, the handrail returned to the compound for painting, with tagging preserved for full traceability so each piece could be reinstalled in its original location.
These images show the reinstalled handrail and the new stainless‑steel backstays providing support. You can also see the transition between the original handrail and the new section introduced to provide a consistent parapet height along the central portion, which was necessary after moving the handrail inboard.

Deck Flap
Ryan Convery
This photo shows the topside of the previous deck flap, which provided articulation near the English end. Here it is again after all deck timbers were removed.

After inspecting the flap, we concluded the arrangement needed redesign. The new design consists of a sealed box‑section frame with steel angles supporting the deck — far more efficient to fabricate. It incorporates a rotational dampener using washers and disc springs, pre‑compressed with locknuts.

We carried out trial assemblies to confirm performance. After testing, the frame was positioned, the pivot‑stub joints were tightened, and the deck components were installed.

Simon Rudman — Chains
You may remember this diagram from the first presentation, showing all the components of a chain knuckle for each pair of chains. In 1974, a significant refurbishment replaced many pins and links. The numbers show that several washers were replaced, a large number of gib keys and pins, and just over half of the links. What we’ll do now is look at what happened during our refurbishment and compare it with what the original consent allowed.

Hangers
Simon Rudman
As we discussed previously, the only real solution to the fatigue issue was to replace the hangers with high‑yield steel versions. They are relatively simple round‑bar elements to replicate. However, we made a key change to the “fishtail” at the top. Originally, the hanger had a fishtail and a separate wedge hammered into the top of the hanger cap. We later discovered why that caused problems.

The replacement fishtail design included a captive locking wedge. The hanger could be lifted from below, the wedge placed into the dovetail, and the hanger lowered — no hammering required. The slight curvature on the outside of the fishtail allowed some angular adjustment inside the hanger‑cap pocket, making installation easier.

As expected, we replaced every hanger on the bridge. None of the originals were retained.

Washers
Simon Rudman
There were many washers on the structure, and most were heavily corroded. They were straightforward to replicate by water‑jet‑cutting steel plate. In the consented scope, we estimated replacing 584 washers, but we actually replaced 428. Still significant, but none of the originals could be kept.

Gib Keys
Simon Rudman
Gib keys prevent the whole knuckle assembly from falling apart. Many were fused into place by corrosion and had to be cut out; others were simply too corroded to reuse. They are easy to cut from flat plate. We ended up replacing 203, compared to the 584 we initially estimated.

Pins
Simon Rudman
There were 64 original pins remaining in the bridge. Their condition was remarkably poor. These were generally near the Scottish tower at higher positions, where access was more difficult during the 1974 works — likely why they were left in place.

We also found issues with the spheroidal graphite pins installed in 1974, including casting defects and cracks. The replacement pins were made from high‑yield steel, CNC‑machined from billets. An important detail is that the pins are not perfectly round; they are slightly oval to suit the link geometry.
We replaced slightly more pins than originally estimated. Again, none of the originals were reusable.

Links
Simon Rudman
The original wrought‑iron links were heavily corroded. We carried out significant finite‑element analysis to explore whether any could be retained, but the corrosion was too severe. We also discovered problems with the 1974 replacement links: many showed high porosity in the castings, and several exhibited cracking at the pin‑bearing surfaces. These also had to be replaced.

As with the pins and washers, the replacement links were straightforward to cut from plate. We ended up replacing a very high number of links, and no original or 1974 links could be retained.

Chain Rods
Simon Rudman
The chain rods were the biggest surprise of the entire project. After blasting, we discovered significant corrosion that had not been visible during inspections. In many cases, there was severe loss of section.

The heavy paint build‑up — combined with flexible fillers used historically — had concealed corrosion inside the joints. None of this was mentioned in the 1974 records, and the rods’ condition had simply been invisible without dismantling.

Because these rods were handmade, their original cross‑section varied by about 20–22%. This meant a smaller rod with minor corrosion might fail, while a larger rod with major corrosion could still pass. The variability made assessment complex.

We created a 3D‑printed nylon prototype of a typical chain‑rod end to establish a representative geometry for new replacements. Using this, a machine tool was produced to upset‑forge new rod ends from high‑yield steel blanks.

These were manufactured in bulk. The new ends were then welded onto straight bar — technically challenging due to the volume of weld required. Every weld was radiographed, and early in the process several needed rework until fabricators became familiar with the technique.

We ended up replacing 282 chain rods, but 110 originals were retained, all returned to their original positions.

During inspection, we also noticed hairline cracks at some rod ends — previously seen in our initial investigations. After further destructive examination, we found these were not structural defects but natural features from the original forging process. This allowed us to confidently reuse the original rods deemed structurally sound.

Hanger Caps
Simon Rudman
All hanger caps were original; none had been replaced in 1974. After blasting, every cap showed cracking — sometimes one crack, sometimes as many as four in the corners.
To understand the cause, we fabricated a test rig, fitted strain gauges and applied load using a hydraulic ram. We tested at various load factors.

Up to 2.5 times working load, the behaviour was essentially reversible and elastic. At 5 times load, behaviour became nonlinear, and permanent crack propagation occurred. This suggested that the cracks in the original caps were likely caused during original installation — specifically hammering the wedge into place — rather than service loading.
With confidence gained from testing, we concluded that almost all hanger caps could be retained safely. Only eight needed replacement, mostly due to casting quality issues. These replacements were CNC‑machined and, once painted, were practically indistinguishable from originals.

Reconstruction
Simon Rudman
As with many engineering projects, reassembly essentially followed the reverse order of dismantling. Chain rods were installed one by one onto strops and dollies and hauled across the river. They were delivered in manageable loads and stored on the bank before installation.

Twelve repetitions later, the chains were fully in place.

At the English anchorage, the first rod from the anchor was a Macalloy bar with adjustable couplers. This gave us about ±300 mm vertical adjustment at midspan, allowing precise positioning of the chains following survey checks.

With the chains correctly aligned, three hangers at a time were dropped into place, the bottom‑chord angle was installed, and the timber framework for the deck began to be built out.

Under‑deck photos show progress from below, including the original Bowering support for the central stringer.
Work continued back toward the English tower using the aerial ropeway and platforms, until the full deck framework was completed.

Deck boarding followed. Two ducts were installed — one for an existing BT service and one spare for future use. GRP deck panels were then installed first on the carriageway, then on the footways. These modern panels replicate the historic ACME panels in appearance while providing improved longevity.

Photos show the handrail reinstated. Despite being moved inboard and supported by stainless backstays, it remains visually unobtrusive relative to the parapet.

Further painting works were undertaken from the gantry. Finally, the entire structure was completed, having been fully dismantled, refurbished or replaced, and reassembled.

Alison Church
Thank you, Simon and Ryan, for that very interesting walkthrough of how you rebuilt the Union Chain Bridge. And apologies for not introducing myself earlier — I’m Alison Church, Senior Structural Engineer at Historic England.

A quick note: we will be publishing a technical case study on the bridge, so keep an eye out for that in January.

We also have Patrick Smith from Northumberland County Council with us to help answer questions.

We’ll begin with a couple of questions from the last session that we felt were better answered today.

Q&A
Question:
Was the listed building consent flexible enough to allow changes after inspecting all parts onsite, especially if more or less original fabric needed to be retained than anticipated? Was this true on both sides of the border?

Patrick Smith
It wasn’t so much the consent itself; it was the working relationship between us and the heritage authorities. There was a site presence, everyone was aware of what was happening, and as we discovered new conditions, everything was wrapped into a component‑retention report at the end. It didn’t present any major issues.

Alison Church
It’s all about communication.

Question:
With hindsight, is there anything you would change now that wasn’t necessarily considered previously?

Patrick Smith
We might have anticipated more hidden defects and looked a bit closer during early inspections. It wouldn’t have changed the overall approach — dismantling was always the right solution — but it would have given us earlier warning about the extent of chain‑rod replacement and allowed better planning.

Simon Rudman
We did make one major change during development: recognising that complete dismantling was the only appropriate solution. Beyond that, not much would change — just forewarning.

Question:
What method of procurement was used, and why?

Patrick Smith
It was an open tender. The first phase used a consultant-contractor arrangement for early contractor involvement. Then we moved to an NEC3 Option C (target cost with activity schedule). It suited the project well: a bespoke piece of work with well‑defined elements but still some risks. The strong quality weighting allowed us to filter for suitable contractors.

Question:
What paint systems were used, particularly regarding paint archaeology, surface preparation and final colour decisions?

Simon Rudman
A standard DTP Type 2 bridge‑painting system was used. Both old and new components were treated the same, including aluminium metal spray as a base coat. We had on‑site facilities for blasting and metal spraying, so everything could be treated consistently.

Non‑setting sealants were applied over joints to stop moisture ingress. In inaccessible areas — such as around the towers — some components were galvanised, with a wax‑oil coating applied over the top. This can be steam-cleaned and re‑applied when needed.

Question:
What corrosion resistance does the high‑yield steel have compared to wrought iron?

Ryan Convery
Slightly better, but the main focus was preventing corrosion from starting at all. Maintenance will still be required: minor painting after around twelve years, major after twenty‑five.

Simon Rudman
Future repainting should involve removing old paint properly — unlike past decades.

Question:
You mentioned that many replacement components in the 1970s were on the main span due to easier access. Where are the original rods now?

Ryan Convery
The original chain rods were returned to their exact original positions. Everything was tagged and traced. More original components tend to be on the English side, where the hanger cap didn’t trap moisture.

Question:
What species of timber was used for the deck, and how was it preserved?

Ryan Convery
Douglas fir, pressure‑treated to the highest class.

Question:
How was the use of modern materials, such as GRP deck panels, justified in heritage terms?

Ryan Convery
The ACME panels they replaced were not original and had been renewed several times. GRP provides improved durability while matching the appearance. It’s essentially a modern equivalent of a non-original historic element.

Question:
What is the design life of the new deck timbers? Did you consider other materials?

Simon Rudman
Around 25 years for the timber deck. We didn’t consider other materials, as a timber deck is part of the historic character of this Grade I / Category A listed structure.

Question:
How did you estimate replacement quantities for consent purposes?

Simon Rudman
Partly based on inspections, partly on experience with other steel bridges, and partly educated guesswork. It was important to present a realistic expectation from the outset.

Question:
How was movement of elements considered after standardising chain‑link sections and changing materials? And can you explain the finite element analysis?

Ryan Convery
The finite‑element models were non‑linear contact analyses created in LUSAS. They simulated pin‑to‑link behaviour under unit load, stretching the components to create stress‑strain curves. We repeated the analysis with various cross‑section reductions to determine minimum acceptable section sizes.

The chain acts as a catenary, meaning tension is constant throughout. Strength isn’t affected by position along the chain.

Simon Rudman
Movement doesn’t matter much because the holes and pins are oval, not round. Loads transfer at specific contact points. As long as minimum required areas are met, the behaviour remains unchanged.

Question:
Did you build a pause into the programme between removal and reinstatement to assess components?

Simon Rudman
Although the presentation seems linear, the programme was highly overlapping. While chain components were being assessed and repaired, rock anchors were being installed, masonry repairs were underway, and so on. The contractor’s programme included cleaning, assessing and refurbishing time. Only unexpected replacement volumes created deviations.

Question:
Was traditional hand‑forging considered for replacement chain links?

Simon Rudman
No — traditional forging wouldn’t meet modern bridge‑design standards.

Question:
Were any timbers painted? What does ongoing maintenance involve?

Ryan Convery
Timber wasn’t painted. Its protection comes from pressure treatment. Ongoing maintenance includes cleaning stainless and steel components, routine inspections, and periodic anchor testing. Paint cycles are roughly twelve years (minor) and twenty‑five years (major).

Question:
Were parts beyond reuse salvaged, recycled or upcycled?

Simon Rudman
Yes — samples went to Historic England and several universities. Remaining wrought iron was sent to a specialist blacksmithing firm, where it re-enters the conservation supply chain. High‑quality wrought iron is increasingly rare, so it was gratefully received.

Question:
How has material quality improved for the new parts?

Ryan Convery
The basic materials are similar — wrought iron historically, cast iron in the 1970s, timber throughout. Improvements primarily come from modern high‑yield steels and better coatings, allowing improved performance while retaining original form.

Question:
Did heritage requirements constrain the choice of paint system?

Simon Rudman
We used a robust system including aluminium metal spray, which is no longer standard but gives excellent base protection. The main challenge wasn’t the top coat — it was ensuring joints were properly sealed. That’s where historic corrosion occurred. In some inaccessible areas we used galvanising and wax‑oil coating.

Question:
Are any real‑time monitors installed, or are tests carried out periodically?

Ryan Convery
There is no real‑time monitoring; the bridge is performing well. Regular inspections continue, and rock‑anchor lift‑off tests are carried out periodically as required by standards.

Question:
What were your findings regarding lime‑mortar specifications?

Alison Church
Our records show an NHL 3.5 lime binder was used, based on specialist analysis identifying semi‑hydraulic lime in the original samples. This is entirely typical for structures of this type. Sample panels ensured visual consistency.

Simon Rudman
Yes — very standard for historic masonry. Nothing unusual.

Question:
In hindsight, would you have used a different contract form? Did the outturn cost differ significantly from budget?

Patrick Smith
Costs did rise significantly, particularly due to the much larger‑than‑expected number of chain‑rod replacements. But we wouldn’t change the contract form. NEC3 Option C worked well for a project with defined works but inherent uncertainties. It gave flexibility where needed.

Alison Church
Thank you to everyone for your questions and for joining us across both presentations. The recording will be available online later this month, and the technical case study will be published in January.