Fish Plate Welding: Rail Joint Bars, Standards, and Modern Welding Methods

Every time a train crosses a small rail joint with a distinct click, it’s a reminder that rails don’t always run as one continuous piece. For decades, joint bars (commonly called fish plates) held separate rail lengths together, but as speeds and loads grew, railroads turned to welding to create continuous welded rail. Understanding how fish plates fit into that transition is key to safe, reliable track.

Introduction to Rail Joints and Fish Plates

Every rail section on a main line is a chain of shorter pieces, and between them sits a deliberate, engineered gap. That gap is a rail joint, and the hardware that closed it for most of the last century was a fish plate, fishplate, or joint bar.

The name comes from shape, not function: long, flat steel plates that flare slightly on each side, so seen end-on they resemble a fish laid flat. Look at any joint and you see a pair of matching plates, one bolted to each rail end, drawn together by four or more bolts.

What a fish plate actually does

Two jobs. First, alignment: the plate holds two rail ends at the right height, gauge, and angle. Without it, rails bolted face to face would be held only by friction and bolt bearing, and would wander out of square in service. The plate gives the joint a positive geometric reference, so the running surface stays continuous.

Second, load transfer. As a wheel crosses the joint, load must pass from one rail end to the other, and the plate provides the shear and bearing path. Bolts carry longitudinal forces from thermal expansion, contraction, braking, and traction; the plate takes the vertical bending load the wheel applies.

It must also absorb thermal movement. A 39-foot (12 m) rail changes length noticeably across seasonal swings, and the joint bar gives the rail ends room to move as the bolts flex and the plate bears on the web. That is why gaps are checked and adjusted rather than rail ends crammed together.

Why bolted joints were the standard

The fish plate suited the railways of the late nineteenth and early twentieth centuries. Rails came in limited lengths, and bolting them together was fast, cheap, and needed nothing more than a wrench, a spiking hammer, and a rail drill. A crew could replace a broken joint bar in minutes with no heat, no electricity, and no welding equipment, and the joint tolerated misalignment.

Joint bar designs were refined to match each rail section, with hole pattern, bar height, and bolt size tailored to the profile. Today they are specified by rail section and weight per yard, with drilled or slotted holes depending on whether the bar allows movement or locks the joint solid.

Bolted joints are discontinuities, and railways care about discontinuities more than most people realize.

Each joint is a small step in the running surface, and the joint bar takes the shock as the wheel climbs onto the next rail end. Railroads spend a meaningful share of track maintenance budgets on joint upkeep: tightening bolts, replacing worn bars, correcting gaps, replacing fasteners. Joint noise is among the loudest complaints on urban and suburban corridors, and in some jurisdictions it has driven spending to remove joints altogether.

The shift to continuous welded rail

All of this set the stage for the change that dominates modern track practice: rail sections got longer, mills improved, and welding matured enough to join rail reliably in the field. The goal became a rail running hundreds or thousands of feet with no joint for long stretches.

That is continuous welded rail, or CWR: weld the rail ends together, fasten the rail to the sleepers or a ballastless track slab at selected points only, and let the rail expand and contract around a fixed, “anchored,” midpoint. Rail joint welding practices are built around thermal management, since forces in a locked rail can buckle it if anchoring is wrong.

The benefits follow directly from removing the discontinuity:

• Lower impact loads on rails, wheels, sleepers, and ballast, since the wheel no longer drops into a gap.
• Reduced maintenance: no bolts to check, no gaps to adjust, no joint bars to replace.
• Smoother ride and lower noise, which is why transit agencies and high-speed lines moved to CWR earliest.
• Longer rail life, since impact and fatigue cracking at joints is a leading cause of failure.
• Better electrical continuity for track circuits and signalling, and a simpler return current path.

The downside is real: CWR is unforgiving of bad work. A bolted joint slightly out of square or a little loose still functions, but a weld that is misaligned, under-filled, or full of porosity is a defect from day one and it tends to grow. That shift in consequence is why welding rail is governed by formal standards, procedures, and inspection regimes rather than the judgment of a track gang.

What this article covers

Fish plates have not disappeared. They remain on lighter traffic track, branch lines, industrial and yard tracks, turnouts in some applications, legacy jointed track in many countries, and anywhere a joint must be opened and closed without a welding plant on site. The comparison of fish plates and welded joints comes up constantly in track engineering, and the honest answer is that both belong on the network, in different places.

What follows covers joint bar design and materials, why railroads moved to welding, field and plant methods including flash-butt and thermite welding, the AWS D15 series, weld preparation and welder qualification, inspection and non-destructive testing, and the common mistakes behind failures. Our guide to all-position welding rod selection refreshes the filler metal basics.

Whether you specify joint bars, run thermite kits in a cold rail cut, or decide to bolt or weld a crossing on a spur, the deciding factor is the same: know what the joint is doing, match the method to the duty, and never leave it uninspected.

Fish Plate Design, Materials, and Fit-Up

Before a welder touches a rail, someone has to specify the hardware that held rails together for a century: the fish plate, or joint bar, a heavy steel plate that straddles the rail at a joint and carries bolts clamping two rail ends face to face.

One rule governs everything else here: a fish plate must match the rail profile it clamps. Each bar is rolled or forged for a specific rail section, with the fishing edges, base, and bolt holes machined to sit flat against that head and web geometry. That is why a bar that looks correct may not fit an existing joint. Comparisons of fish plates versus welded joints make the same point: a bolted rail joint is a precisely engineered connection, not a general-purpose clamp.

Common Types of Joint Bars

Joint bar type	How it works	Where you see it
Single fish plate	One plate, one side, bolted through both ends	Light and medium rail, older track
Double (twin) joint bar	Matched pair, one per side, sharing bolt load	Heavier rail, high axle loads
Keyed joint bar	Head and base shapes carry torque, not friction	Curved track, creep and rotation
Unkeyed / flat bar	Plain plate, strength from bolt clamping	Straight, stable track
Insulated joint bar	Bushed or non-conductive bar	Circuit boundaries, turnouts, interlockings
Thermit-welded fish plate	Bar fixed to the rail end by thermite	Field repair, lower-speed track

The last type bridges both halves of this article. A thermit-welded fishplate is a joint bar pre-fitted to a rail end and joined by aluminothermic reaction, so bar and rail act as one welded assembly, not a friction joint. How fishplate joints are handled alongside rail joint welding varies by railroad, so the owning agency’s specification wins.

Materials and Condition

Joint bars are predominantly rolled carbon steel, chosen so strength and ductility match the rail section they serve. The North American reference specification for steel plates in track joints is ASTM A30/A30M, with railroads adding requirements for bolt hole size, edge distances, and proof loading. Cast and forged bars exist, but rolled plate dominates for consistent section properties and accurate machining.

Condition matters just as much: cracks at bolt holes, wear and grooving along the fishing edge, bent or out-of-flat geometry, corrosion pitting, and stretched bolt holes. An elongated hole is not cosmetic, it shows the joint moving and the bolts no longer clamping. Replace a failed bar with a new bar of the same section, never a mixed one.

Hole Patterns and Bolt Compatibility

Bolts pass through matching holes drilled in the bar and the rail web. Most joints use four holes, two either side of the centre line, while heavier and higher-speed track often carry six for clamping. Hole diameter, position, and bolt size and grade work as a set. Oversizing a hole to reuse an old bolt is false economy, because the extra clearance cuts clamping force and lets the rail ends rock under load, which leads to cracking. Square-head track bolts of the specified grade go in sequence and must sit flush, clear of the passing gauge.

Insulated Joint Bars and Track Circuits

Where signalling relies on track circuits, a solid steel bar conducts current across the joint and the circuit stops detecting the train. Insulated joint bars provide the electrical separation the signalling system requires, as a composite bar with fiberglass or epoxy insulation or as a metal bar with insulating sleeves around the bolts. That insulation must stay intact and the bolt holes properly sealed, so installation is more delicate than a plain bolted joint. Running bonds and jumpers must carry current around the insulated length, or the circuit reads as a broken rail. Insulated joints are common at turnouts, crossings, and interlocking limits.

Fit-Up: Getting the Plate to Sit Flat

The joint bar must bear fully on the rail with no rocking gap. Fitters work from the gauge side and the field side in turn, using shims and packing to close the gap, then check the joint with a gauge and straightedge before tightening. Rail ends need the correct gap for the ambient temperature, because a joint set up too tight in summer has nowhere to move when the rail expands. Gauging and torque checks are repeated once the first traffic loads have passed over it.

Removing a damaged bar in the field usually means cutting and air arc gouging, as explained in the guide to air arc welding and carbon arc gouging. Shop-side repair of steel hardware also calls for sound filler metal selection, covered on alloy steel welding electrodes.

Fit-up is where most joint bar failures start. Misalignment, a proud bolt head, a gap under the fishing edge, or a mismatched section all show up later as cracked rails and loose fasteners. Getting the plate right is unglamorous work, but it makes every later welding decision on that track safer.

Why Weld Rails: From Joints to Continuous Welded Rail

Every bolted joint is a small engineering compromise. Two rail ends sit side by side, joint bars clamp them, and bolts hold the assembly together. It works, but it works by letting things move. As axle loads, traffic volume and line speed climb, that compromise becomes the limiting factor in the track, and railroads eventually replaced it with something simpler: weld the rail ends and make the running surface continuous.

The Impact Problem

A wheel crossing a joint does not travel smoothly. The rail flexes just before the joint, the wheelset drops slightly into the gap, then the far rail end rises abruptly to meet it. That sudden change in elevation produces a dynamic impact force, the classic source of the sharp click you hear and feel in a passenger car. The effect is modest on a lightly used branch line and severe on a heavy main line, where the dynamic force at the joint can be a multiple of the static wheel load.

Fish plate joints must absorb that cycle for decades. The bars carry repeated cyclic loading, bolts loosen as the joint works, and corrosion attacks the bar seats and contact surfaces. Joints also let the rail slide slightly, which is why jointed track handles temperature changes well and why rails in bolted track can be laid in shorter, lighter lengths. The trade is a recurring maintenance bill and a quality hit on every wheel passage.

Lower Maintenance and Fewer Components to Inspect

Continuous welded rail (CWR) removes most bolted joints, and with them a large share of the maintenance workload. Fewer joints mean fewer components to inspect and less risk of joint-related defects: no joint bars to check for fatigue cracking, no bolts to re-torque, no joint chairs or cribbing to repair after a rough patch. That matters on a network where a single crew must cover hundreds of miles. In practice, welded rail has meant lower routine track maintenance costs and fewer speed restrictions, a large part of why it became the default for new main line construction.

Ride Quality, Noise and Rail Life

A continuous running surface removes the joint dip, so cars ride more smoothly and passengers feel less of the track. Vibration and impact noise drop as well, which benefits riders, trackside neighbours, wheelsets and track fastenings. Fewer impacts also mean less cumulative damage to the rail head, so rail lasts longer between resurfacing or replacement. A useful side-by-side look at the two joint types is in this comparison of fish plates and welded joints.

The Trade-Off: Thermal Stress and Anchoring

Welding is not a free upgrade. CWR has to be managed as a stressed system rather than a loose collection of pieces. Steel changes length with temperature: a 100 m rail lengthens by roughly 23 mm for every 20 °C swing. A welded rail cannot slide to relieve that movement, so it must be anchored to the ties and allowed to expand into controlled stress. Rail is tensioned at installation so that its stress-free temperature sits above the coldest expected temperature, keeping cold rails in tension rather than compression. Get that wrong and the same physics produces two very different failures: sun-heated rail that buckles into a kink, and cold rail that snaps in tension.

What Changes in Practice

Factor	Fish-plated joint	Welded joint
Wheel impact	Repeated at every joint	Eliminated along the welded length
Ride quality	Dip and bump at each joint	Continuous running surface
Maintenance	Bars, bolts, chairs, re-torquing	Fewer components, more stress management
Thermal movement	Rail can slide and expand	Anchored; compressive stress must be managed
Repair	Fast, localized joint replacement	Cut out and re-weld the rail
Best suited to	Light branch, yard and industrial track	High-speed, heavy-traffic main line

That is the logic behind the shift: jointed track buys simplicity and tolerance of thermal movement at the price of impact, noise and constant upkeep, while welded rail buys a smoother, stronger, cheaper-to-maintain running surface at the price of tighter thermal control. Not every line makes the switch. Branch lines, sidings, yard tracks and much of a turnout area keep bolted joints because the traffic does not justify the capital cost. Modern practice also mixes methods, with plants welding rail by flash-butt and thermite processes, and crews bringing in-track flash-butt welding equipment to the line itself. Where rails must be cut or repaired, arc-based processes such as stick welding remain a practical option. All of this work falls under the AWS D15 Committee on Railroad Welding, which sets requirements for rails and rail vehicles alike.

Rail Welding Methods Used in the Field and Plant

Rail welding happens under three conditions. In a factory, rail ends are joined before the rail reaches the track. On an open line at night, the joint must be made with equipment that fits between the rails, in whatever weather shows up. In a plant, rail strings are built joint by joint before installation. Each setting has its own qualification and acceptance rules, which is why the AWS D15 Committee on Railroad Welding exists.

Thermite (Aluminothermic) Welding

Thermite is the traditional in-track method. It needs power, but surprisingly little of it, and it joins rail ends in place without moving the rail.

A mold around the prepared rail ends holds a charge of finely divided iron oxide and aluminum powder. Ignited, the aluminum reduces the iron oxide in a highly exothermic reaction, producing molten steel that fills the mold and solidifies. Excess metal is ground flush.

Because the reaction is self-contained, the crew needs no heavy welding power source, which matters on lightly used lines and in remote territory, and the method suits older rail sections and confined spaces. The joint is a cast structure, not a forge weld: less ductility, more sensitivity to shock loading, and porosity and inclusions that must be controlled. Every weld needs proper preheat, mold preparation, and heat treatment. Since the mold is a consumable, thermite-welded fishplate systems that capture molten metal in a fish plate are used to reinforce a joint where the base rail is older or lighter.

Flash-Butt Welding

Flash-butt is the high-strength method. It produces a forge weld: the joint is compressed while molten, so separated grains rejoin into continuous metal. The result is usually stronger than the surrounding rail itself, and the automated, repeatable process delivers excellent consistency. It is the preferred method for continuous welded rail in modern practice.

The process is resistance heating. Rail ends are brought together with a small gap, heavy current passes across the joint, and touching points heat rapidly. Thin films of molten metal flash and burn away, which is where the name comes from. The machine then forces the ends together under pressure, upsetting molten metal out of the joint, and grinding restores the head.

• In-plant flash-butt welding happens in a rail mill or plant, where clean, matching rail ends are joined under controlled conditions and cut to exact length. The highest quality is achieved here.
• In-track flash-butt welding is performed on the line with a self-propelled rail-bound machine that grips the rail and makes the weld in place. In-track flash-butt machines work on continuous welded rail, correcting shifted rail, replacing damaged rail, or joining rails at switches and crossings.

The practical limitation is capital cost and a stable power and hydraulic supply. The process is also less forgiving of badly prepared or mismatched rail ends than thermite.

Arc-Based Welding Methods

Arc welding has a narrower role: repair welds, attaching guard rails, rail end blocks and conductor rails, and joining where a rail cannot be removed. SMAW and gas-shielded arc welding both appear in railroad maintenance practice, and the enduring advantages of stick welding are what make it useful here: simplicity, portability, and tolerance of surface condition.

Arc welding is not the method of choice for new continuous welded rail joints. It needs more heat input control than a forge weld, and the result has lower fatigue properties. State agencies specify field arc welding practices in detail, and the NCDOT field welding procedures manual shows the detail such repairs require.

How the Methods Compare

Method	Joint	Setting	Strength	Limitation
Thermite	Cast exothermic	In-track field	Low power, portable	Lower ductility
Flash-butt (plant)	Forge weld	Mill or plant	Top repeatability	Matched rail, heavy gear
Flash-butt (in-track)	Forge weld	Trackside machine	CWR correction	Capital cost, access
Arc-based	Fused buildup	Repairs, attachments	Portable, versatile	Heat input, low fatigue

No single process wins everywhere, and some track still relies on fish plate welding with bolted fish plates instead of a welded joint. A side-by-side of flash-butt versus thermite welding helps weigh quality against cost and crew logistics.

How Method Selection Is Actually Made

1. Where is the work? In-plant is almost always flash-butt because conditions are controlled. In-track work may mean thermite for a joint on older, lighter rail, or a flash-butt machine for CWR correction.
2. What rail section? Light and older sections commonly use thermite, since flash-butt needs matching ends in good condition. Heavy sections benefit more from forge welding.
3. What logistics? Power, access windows, weather, and crew skills narrow the options. A thermite kit can go almost anywhere.
4. What quality does the service demand? Heavy-haul and high-speed mainlines call for flash-butt, while secondary and industrial track often runs on thermite joints, and some fish plate versus welded joint decisions turn on the same tradeoffs.

Chosen method still has to satisfy the AWS D15.2 requirements and the railroad’s own specifications for preheat, alignment, heat treatment, and inspection.

Standards and Codes: AWS D15 and Railroad Welding Requirements

Standards are where fish plate work and rail welding meet paperwork. A bolted joint bar needs correct length, correct holes, correct torque. But when a welder joins two rail ends, cuts out a failed weld, or repairs a cracked locomotive frame, the joint becomes safety critical. Somebody has to write down how the weld is made, and prove the welder can make it repeatably. That is the job of the AWS D15 Committee and the railroad specifications built on it.

What the AWS D15 Committee Actually Does

The AWS D15 Committee on Railroad Welding develops the AWS D15 series of welding standards. Its scope covers railroad welding applications across both manufacture and maintenance, which is the key point for anyone working around fish plates. The same committee that writes requirements for building railcar bodies also covers repair and maintenance, including welding of rails themselves.

That dual scope matters because the failure risks are not symmetrical. A fish plate is a mechanical fastener. A rail weld is a structural joint that a loaded wheel passes over thousands of times a day. A welded rail or rail vehicle carries dynamic, cyclic loading, and standards exist to keep those welds in service for decades.

D15.1 and D15.2: Two Documents, One Goal

The two most-referenced documents cover different scopes but share one philosophy: a weld is only as good as the procedure and the people behind it.

Standard	Scope	What it addresses
AWS D15.1 (D15.1/D15.1M)	Welding of railcars and locomotives, including components	Minimum requirements for welding, procedure qualification, welder qualification, and inspection of rail vehicle structures and components. The standard carries formal endorsement, reviewed through the D15.1 endorsement process.
AWS D15.2	Welding of rails and rail vehicles	Guidelines and minimum requirements for welding rails and rail vehicles, including thermite and flash-butt welding practices, alignment, heat treatment expectations, and acceptance criteria for finished welds.

Neither document tells a track crew how to strike an arc on a rail end and walk away. Both tell you what must be documented first: a qualified procedure, a tested welder, defined inspection, and a record trail that outlasts the crew.

What the Standards Actually Require

Every railroad welding program has to answer for four things:

• Qualified procedures. A welding procedure specification (WPS) must exist before production welding. It defines joint configuration, filler metal, preheat, cleaning, sequence, and heat input control. Welders work to the procedure, not to memory.
• Qualified personnel. Welders and welding operators must be tested against the applicable standard, with continuity maintained through documented training and testing intervals.
• Defined inspection and acceptance. Visual inspection standards, nondestructive testing requirements, and repair or cut-out criteria are set in advance, so a bad weld is caught by a rule rather than an argument.
• Traceability. Records tie the weld to the procedure, the welder, the inspection results, and the specific location in the track or on the car.

The same logic shows up in aerospace welding standards: AWS D17 in aerospace and AWS D15 in rail both treat procedure and personnel qualification as the foundation, not paperwork added later.

Where Railroad and Federal Requirements Layer In

AWS documents do not operate alone. Railroads work under federal track safety regulations, engineering department instructions, and industry manuals such as the AREMA Manual of Railway Engineering. Owner-published welding procedures are the working documents on the ground. A state specification such as the North Carolina DOT field welding procedures manual shows how much detail gets written down once a public agency takes over welding on its projects.

For materials and consumables, the referenced specifications dictate what is permitted. Choosing filler metal within a qualified procedure is covered in our guide to alloy steel welding electrodes. For field repair work using stick welding rather than thermite or flash-butt, the case for that process is laid out in why SMAW still wins for many repair applications.

Compliance is not paperwork theater. A rail weld that fails an inspection or a destructive test in the shop gets cut out in the field, where a poor joint is measured in delays, replacement rail, and passenger complaints. Treat the D15 series, the railroad’s specifications, and the qualified procedure as one package, and every fish plate, rail weld, and repair becomes a joint somebody has already proven can hold.

Weld Preparation, Procedure Qualification, and Welder Qualification

A rail joint that is not prepared correctly does not fail loudly. It quietly costs you. A mismatched joint, a contaminated surface, a skewed fit-up, or an unqualified weld can sit in track for months before it turns into a cracked weld, a broken rail, or a derailment.

The governing framework for U.S. railroad welding is the AWS D15 series, published by the American Welding Society’s D15 Committee on Railroad Welding. D15.1 covers railcars, locomotives, and their components, while D15.2 addresses rails and rail vehicles and was significantly consolidated in its 2022 revision. ANSI’s overview of the AWS D15.2:2022 update is a good place to start.

Joint Preparation: Getting the Ends Right

Rail ends are prepared by milling, grinding, or abrasive blasting to a defined geometry. For flash-butt welding the ends are dressed to a specific angle and must be clean and square. For thermite welding they are squared, with a gap set for thermal expansion of the molten metal. Fish plate and joint bar welds follow the same rules.

• Surface condition. Mill scale, rust, oil, and graphite prevent fusion. Surfaces must be ground back to bright metal and kept clean until the weld is made; degreaser alone is not enough on a rail sitting in a yard.
• Alignment and offset. Ends must be in line vertically and horizontally before welding. A small lateral offset creates a bent weld that concentrates stress, and an out-of-position joint leaves a bending moment that later shows as a vertical split in the head or web. This is fit-up, not welding.
• Gap setting. The gap must account for shrinkage and expansion. Too tight and metal cannot flow; too wide and you get incomplete penetration, slag inclusion, or porosity.

Preheat, Sequence, and Post-Weld Conditioning

Preheat is not optional. Specifications typically call for it when rail temperature falls below a minimum, commonly around 50 °F (10 °C), and the joint must be free of ice, frost, and moisture. A cold rail pulls heat from the weld too fast, so the pool freezes before it fuses properly, leaving a brittle weld that cracks later. Agency field manuals spell out thresholds, hold times, and torch positioning, as in the NCDOT field welding procedures manual.

The sequence follows the process: preheat the ends, set the mold, ignite the charge, let the aluminothermic reaction run, pull the plug and slag, let the weld cool, trim excess metal, and grind flush with the running surface. Flash-butt welding differs: the ends come together under pressure, current melts and upsets them, and no filler metal is added. In-track flash-butt welding runs on the track with the rail still in service in some configurations.

Excess metal is trimmed and ground back to the correct profile, the surface is checked for flatness, and where required the rail gets post-weld heat treatment to relieve residual stress. Grinding rework can introduce a defect, and removing a failed section means gouging it out, which is where air arc welding and carbon arc gouging come in for cutout work.

Procedure Qualification: Prove It Before You Weld for Production

A procedure qualification record, or PQR, shows that a welding procedure meets the governing standard. Under the AWS D15 series, including AWS D15.1 for railcar, locomotive, and component welding and D15.2 for rail and rail vehicle welding, the PQR is built around test welds made under the conditions you will use in production.

• Base metal specification, rail section, and heat identification
• Process, joint design, position, filler metal or thermite charge
• Preheat and interpass temperatures
• Electrical parameters or thermite reaction parameters
• Post-weld heat treatment details, where applicable
• Coupon results: tensile, bend, macro-section, hardness
• Acceptance criteria for each result

Qualification testing finds the limits of the process. You qualify a range of conditions, then weld only inside it. If rail temperature, joint gap, or alignment falls outside the qualified window, you need a new qualification, not a workaround.

Welder Performance Qualification and the Paper Trail

Procedure qualification asks whether the process works. Welder performance qualification asks whether the person can do it. A WPQ ties a welder to a qualified range of variables and confirms they can produce acceptable welds inside it, based on a test weld. The record lists exam results, qualified position and parameter range, the PQR behind the test, and a requalification interval set by the standard and agency.

Maintenance of qualification is not automatic. Welders requalify at the intervals the standard specifies, and anyone off the job or working outside their qualified range requalifies first.

That documentation gets audited. A traceability file links the finished weld back to:

• Rail heat number, section, and segment identifier
• Joint location in track, by stationing or milepost
• PQR and WPQ behind the weld, plus welder ID
• Date, time, ambient and rail temperature, weather, and NDT results
• Any rework, repair, or cutout performed later

Without that chain a failed weld becomes an archaeology problem. With it you can trend defects back to a crew, shift, temperature band, or heat of steel. The paperwork is the weld.

Inspection, Testing, and Acceptance of Rail Welds

A fish plate can mask a defect for years, right up until the bar cracks; a welded joint offers no such hiding place. A weld in a running rail carries a locomotive’s full weight over the same spot thousands of times a day, so a missed defect becomes a fatigue crack through the rail head and, at worst, a broken rail on a mainline. Wheel loads concentrate stress at the weld and its heat-affected zone, and any discontinuity becomes a crack initiation site, so inspection is a mandatory stage for every railroad that welds rail.

Visual inspection comes first

Visual inspection is the first and most frequent check, in the plant and in the field. A qualified inspector examines each new weld before release, against the governing standard and the owner’s welding manual. A welder’s own look at the bead never substitutes for independent inspection.

• Offset or misalignment of the two rail ends
• Undercut, overlap, or lack of sidewall fusion
• Cracks of any length, including hairline cracks and arc strikes
• Slag, spatter, porosity, or oxidation on the surface
• Excess reinforcement, poor grinding, or a narrowed rail head

Ultrasonic testing and supporting NDT

Ultrasonic testing is the workhorse for internal discontinuities: lack of fusion, slag or inclusion pockets, voids in the weld metal, and subsurface cracking. Field and plant welds are both subject to UT, with method, coverage, and rejection limits set by the specification rather than left to inspector judgment. The NCDOT field welding procedures manual sets out field weld inspection, UT, and acceptance criteria in detail.

Not every weld receives UT. Many railroads specify it on critical weld types or a defined sample, then rely on in-service detection. Wayside and onboard ultrasonic rail flaw detector cars scan running rails continuously and flag suspect locations, including welds.

Magnetic particle testing finds surface and near-surface cracks in ferrous rails, dye penetrant identifies fine surface-breaking flaws, and radiography is rare, reserved for special cases. Flash-butt welds carry their quality evidence in the process: the upset and flash action is the proof of fusion, and each weld is recorded with its upset force and measurements during in-track flash-butt welding. A weld that misses its upset parameters is rejected on the spot.

Mechanical and geometric acceptance

Beyond flaws, inspectors confirm the joint does the job the track needs:

1. Offset and alignment: within the specified tolerance, typically a fraction of an inch at the rail head, since a step creates wheel impact and bending moment.
2. Level and cant: correct elevation and cant, checked with track geometry equipment or on-foot measurement.
3. Hardness and samples: thermite welds are checked for rail head hardness and for a soft or hard heat-affected zone; many specifications require destructive testing of sample welds from each run.
4. Electrical continuity: the joint must conduct as part of the running rail circuit used for signals and track circuits.
5. Rail stress: rail temperature and neutral temperature are recorded and the anchor pattern adjusted to limit buckling risk, because welding adds heat to a stressed rail.

Acceptance criteria are never generic. They come from the applicable standard and the railroad specification: AWS D15.1 covers railcars, locomotives, and components, AWS D15.2 addresses welding of rail vehicles, with AREMA and owner requirements on top. A railroad welding specification such as AWS D15 sets the floor, and owners often add tighter limits.

Defect	Usual cause	Caught by	Typical disposition
Lack of fusion or penetration	Misalignment, dirty ends, low heat	UT	Cut out and replace
Slag, inclusions, porosity	Bad thermite reaction, contamination	UT, visual	Cut out and replace
Cracks, surface or internal	Fatigue, overheating, bad preheat	Visual, UT, detector cars	Cut out immediately
Offset, misalignment, poor cant	Bad end prep, track movement	Visual, geometry check	Grind within limits, else cut out
Undercut, excess reinforcement, narrow head	Wrong parameters, poor grinding	Visual	Grind to profile or cut out

When a weld must be cut out

If the defect is inside the rail, the rail comes out. Any internal crack, lack of fusion, or inclusion rejected by UT means the rail is removed and replaced, then re-welded. Grinding out a subsurface flaw leaves the fatigue crack starter in place, so it is not an acceptable repair. Surface and geometry defects are graded: minor reinforcement and profile issues get ground and re-inspected, while offset outside limits, cracks of any size, or repeat failures at a joint end mean a cut-out.

The cut-out is itself a welding job. Record the defect location, cut the rail out with a rail saw, drop in the replacement, and weld it in the plant or in track. Gouging the weld metal is where air arc and carbon arc gouging technique matters, since a flame-cut rail end must be ground back to clean, unfractured metal. Some crews tie the track together in the interim with a thermite-welded fishplate so service continues while the permanent weld is made.

Acceptance is not a rubber stamp at the end of welding. It is what keeps a welded joint in service for decades instead of failing in year three, and it only works when the criteria come from the governing standard and the railroad specification, the testing actually happens, and the paperwork survives long enough to be useful.

Fish Plates vs Welded Joints: Practical Comparison

Railroads pick between fish plates and welding based on traffic volume, track class, climate, terrain, signaling design, and maintenance capability. This fish plate and welded joint guide covers the tradeoffs.

Initial Cost

A bolted joint is cheap on paper: joint bars, four to six bolts, a torque wrench, no capital equipment, preheat, or procedure to qualify. Continuous welded rail (CWR) needs flash butt plant or thermite kits, welding crews, rail handling and destressing gear, temperature monitoring, and a program built around rail stress. Compare cost per mile, not per joint: 39-foot rail gives about 40 joints per rail, roughly 80 per mile of single track. Welded, that falls to a handful or zero, so capital rises while recurring repair spreads over fewer locations.

Maintenance Load

Fish-plated joints are small machines that loosen. Bolts relax, holes wear oval, joint bars corrode and crack, and bolts need retorquing every inspection round. Bolted track consumes labor continuously, and the load grows with traffic as impact works the fasteners loose. Fish plates make short rail sections easy to replace, their biggest practical advantage.

Welded joints cut the joint and fastener count but swap fastener upkeep for thermal and structural upkeep: ultrasonic testing of welded lengths, rail grinding, rail neutral position adjustments, and distress heating to relieve locked-in stress.

Ride Quality

The click of a bolted joint is a real defect signal. Each joint leaves a gap and a stiffness discontinuity that deflects under wheel load, causing impact, vibration, and wear on rail ends and ballast. CWR removes that discontinuity, improving ride quality, cutting wheel and track component wear, and extending rail life. Welded ride quality depends entirely on weld quality and rail surface condition: excess reinforcement, a dip, or a rough profile creates its own impact problem, and a failed weld behaves worse than any bolted joint.

Electrical Continuity and Signaling

Joint bars do electrical duty too. In many territories they are bonded to both rails so signal current can pass through and detect a broken rail. In block signal territory, insulated joints keep a rail break in one block from energizing the adjacent block and giving a false clear.

Even in predominantly welded track, insulated joints remain necessary in some signaling territories. Railroads use insulated joint bars with insulating sleeves, or insulated rail, and that insulated joint may still be a bolted bar joint even with continuously welded rail on both sides. State departments of transportation follow the same logic in their field welding procedures, which require insulated joints and joint bonding within welded lengths.

Expansion, Contraction, and Thermal Stress

Bolted joints handle rail length change by letting rail slip at the joint and fastening points. Welded rail locks that movement in, turning seasonal temperature swings into compressive stress in summer and tensile stress in winter. Unaddressed, it buckles in heat or breaks in cold. Managing that takes rail stress measurement, de-stressing, rail adjustment, and correct anchorage and fastening practice. Bolted track skips the thermal monitoring program but still needs slide joints, expansion provisions, and properly set fastener torque.

Repairability

A failed bolted joint is familiar work: remove the bars, cut out a short section, drop in a stick of rail, fish-plate it, and torque it. A failed weld is harder: the cracked weld is cut out and a replacement length welded in, calling for thermite equipment or an in-track flash butt welding machine plus a qualified crew. Cold weather complicates flash butt work because preheat may be required first. Removing damaged weld metal often starts with air arc and carbon arc gouging, so gouging equipment and consumables are standard rail repair kit items.

Thermite and flash butt are not interchangeable in the field, and the choice shapes repair logistics in ways often overlooked; this flash butt versus thermite welding comparison details the tradeoffs. Choosing CWR means owning or contracting that repair capability.

Where Each One Fits

Criterion	Fish-plated joint	Welded joint (CWR)
Initial cost	Low, no capital gear	High, plant and crews
Maintenance focus	Bolts, torque, joint bars	Stress, grinding, weld inspection
Ride quality	Click and impact at joints	Smoother with good welds
Electrical path	Bonded or insulated bar	Bonding detail; insulated joints remain
Thermal response	Rail slides, less stress	Stress must be managed
Best fit	Light rail, yards, turnouts, low volume lines	High-speed mainline, heavy freight, high traffic corridors

Neither method wins everywhere. Turnouts and crossing tracks stay bolted because the geometry and rail movement make welding more complex, and light rail or industrial spurs often cannot justify the capital. CWR pays for itself on high-traffic mainline, but only when the railroad commits to the thermal stress and inspection program that comes with it.

Common Mistakes, Challenges, and Practical Solutions

Rail joint failures are usually ordinary, repeatable mistakes, not exotic metallurgy: a joint bar torqued to the wrong value, a thermite mold set up on a dirty rail end, a weld checked with a flashlight and nothing else.

Loose Fish Plates, Worn Bolts, and Joint Batter

A bolted joint squeezes two rail ends with a bar bearing against the rail web. When bolts loosen, the ends drop, tip, or roll and the bar stops supporting the joint, so the ends take the full impact of every wheel crossing. Loose or worn plates show up as joint batter, rail end wear, a dip in the running surface, and cracked or flaked rail ends.

• Torque bolts to the railroad’s specified value with a calibrated wrench; there is no universal figure, since bar designs, bolt diameters, and clip styles differ widely. Boltless bars rely on clips or keys with a manufacturer-set tightening value.
• Check that the bar bears fully against the rail web and the joint is closed, not just that the bolts are snug; a joint can be “tight” by torque and still not be seated.
• Replace bars and bolts once bolt holes elongate or the bar shows cracking, bent flanges, or bearing wear; a reused or “reshimmed” bar is a defect waiting for a broken rail.
• Where the joint carries signal current, use an insulated fish plate or a bond; substituting a solid bar silently breaks track circuits. That trade-off between joint bars and welded joints recurs in this comparison of fish plates versus welded rail joints.

Poor Weld Preparation and Thermite Mold Setup

Thermite welding is a chemical reaction inside a disposable mold, so joint quality is decided before the reaction starts. Improper setup or contamination causes most failed welds.

• Misaligned or mismatched rail ends. Cut ends square, match for height, and align to the rail’s specified alignment (typically neutral, not kinked) before the mold closes. Gap and offset must stay within railroad tolerance.
• Contamination. Rust scale, grease, paint, water, or grinding dust cause porosity, inclusions, lack of fusion, and cold shuts. Grind and brush clean immediately before the mold goes on.
• Damp or cold molds. A wet mold turns to steam inside the cavity and ruins the pour. Molds are preheated, and the reaction temperature is what fuses the rail ends, so a cold setup weakens the joint.
• Wrong or expired consumables. Thermite kits have shelf limits and batch consistency requirements. Track the lot number of every kit used so a suspect batch can be traced back to specific welds.

Thermite is still a durable field process, though flash-butt and thermite welding differ sharply in how the metal is joined and where each fits best. For gouging cracked rails or cleaning arc-weld prep, air carbon arc gouging is the practical tool, covered in Air Arc Welding and Carbon Arc Gouging.

Inadequate Inspection and Weak Traceability

Most missed defects come from shortcuts in documentation. A weld with no traceable ticket, lot number, qualified procedure, or recorded inspection result cannot be defended when it fails months later. Standards such as the AWS D15 Committee on Railroad Welding emphasize qualified procedures and welders, and programs such as the NCDOT field welding procedures manual spell out inspection and record-keeping in detail.

• Photograph the joint geometry and finished weld before the crew leaves the track.
• Log weld location, rail section, procedure number, welder ID, consumable lot, and inspection result on the ticket attached to the work order.
• Use ultrasonic inspection, not visual checks alone, on heavy rail and on welds made in cold or wet conditions.
• Hold a short “quality huddle” each shift to review the previous shift’s tickets and rejected welds.

Thermal Issues in Continuous Welded Rail

Effective CWR maintenance requires monitoring of rail temperature and stress. A welded rail has no joints to absorb movement, so every degree of temperature change becomes force: in summer heat, compressive stress drives buckling, the most dangerous CWR failure; in winter cold, tensile stress drives brittle fractures, sometimes with no warning. Rail thermometers, rail stress detectors, and strain-gauge measurements establish each rail’s neutral temperature and stress state, while buckling detectors, temperature-based slow orders, and ultrasonic flaw detection show crews where the risk sits. Beyond compressive limits, destressing or rail cutting with crib blocks relieves the force before it moves the track.

Field Logistics and Weather

Rail work happens in traffic windows, often at night, in weather that does not cooperate, and thermite welding and many arc processes have minimum ambient temperature and moisture limits. State DOT manuals specify when welding must stop and the joint must be protected. Stage molds, consumables, and preheat equipment before the possession begins, and build weather-based “go or no-go” criteria into the plan rather than deciding at 2 a.m. in the rain. In-track flash-butt work needs its own planning, since the machine must occupy the track, carry its own power, and work around traffic, as described in this guide to in-track flash-butt welding.

Conclusion and Best Practices

Fish plate welding sits at a crossroads in rail infrastructure. The joint bar was invented for a railway that could not weld rail, and it did that job for over a century. Modern track leans on welding for ride quality, rail life, and reduced maintenance, yet the fish plate has not disappeared. It remains the right answer for temporary joints, emergency repairs, and insulated joints where signal circuits must be kept separate. Each method has a place, and that place is decided by a qualified procedure, documented workmanship, and honest inspection.

What to Remember

• Fish plates still earn their place. Joint bars remain practical for temporary work, break repairs, and insulated joints.
• Welding is the backbone of rail performance. Continuous welded rail cuts impact noise and vibration, extends rail life, and lowers maintenance cost.
• Inspection closes the loop. An uninspected weld is an assumption, and assumptions fail under load.

Best Practices for Track Engineers

Engineers set the conditions every welder and inspector works within, and a few upstream decisions prevent most downstream problems.

1. Specify the process, not just the result. Name the process (aluminothermic, flash-butt, or a qualified arc method), the applicable AWS D15 document, consumables, preheat where relevant, and the acceptance criteria the weld will be judged against.
2. Require a qualified procedure and qualified welders before the first cut. The qualification record and welder performance test should be on file, current, and matched to the joint configuration and welding position.
3. Account for the environment. Rain, temperature, and wind affect in-track welding, and state DOT field welding manuals address how weather changes preheat, shielding, and cure time.
4. Plan for continuous welded rail stress. Bringing CWR into a jointed track section needs a designed plan for rail neutral temperature, stress management, and thermal forces. Weld quality will not save badly managed track.

Best Practices for Maintenance Personnel

Day-to-day crews are where a good weld stays good, or a small defect becomes a big one.

• Match the method to the repair. Flash-butt welding in track gives repeatable, high-strength joints suited to in-track repair, aluminothermic welding is the classic field answer, and fish plates are fastest when rail must be made continuous, as during a thermit welded fishplate installation.
• Prepare the joint properly. Clean the rail ends, verify alignment and gap against the procedure, confirm the head profile, and remove contamination. Misalignment causes defects that no good wire can fix.
• Follow the documented sequence. Preheat if required, weld to the specified sequence, maintain interpass temperature, and control heat input. Heat control governs the joint’s microstructure, and microstructure governs fatigue life.
• Record what you did. Date, location, rail section, process, welder or operator, consumable lot, preheat readings, and inspection results. Traceability exposes systemic problems instead of repeated failures.
• Inspect on a schedule and after an event. Visual inspection catches surface problems cheaply, while ultrasonic and other nondestructive methods find internal flaws. Re-inspect after impacts, derailments, or heavy tamping near a joint.
• Cut out bad welds early. A weld with a detected defect gets scheduled for removal, not left in service hoping it holds. A scheduled cutout always costs less than a failure in service.

The Standard That Holds It All Together

The AWS D15 Committee on Railroad Welding exists because rail welding carries consequences far beyond the cost of a failed joint. Standards such as AWS D15.2 for welding of rails and rail vehicles define what a defensible weld looks like, and the AWS D15.1 document for railroad and locomotive welding components adds years of committee work on rail-specific problems. Citing the applicable D15 document in your specification, procedure, and inspection report is the most effective way to make the whole process defensible.

For perspective on how these joints are evaluated in practice, our comparison of fish plates and welded rail joints covers the underlying processes in detail.

Bottom Line

Fish plate welding is not a dying technique. It is a decision framework. Bolted joint bars remain correct where speed, reversibility, and signal isolation matter most; welded joints remain correct where performance, longevity, and low maintenance matter most. Either way, whoever makes the joint must know the standard that governs it, hold a qualified procedure, and inspect the finished work with a method suited to what was made. Get those three right, and the difference between a joint that lasts and one that gets cut out is largely decided before the first arc is struck.

Frequently Asked Questions

What is a fish plate in railway tracks?

A fish plate (joint bar) is a steel bar placed on each side of two rail ends and bolted together to connect them. It provides alignment and transfers loads across the joint while allowing for some flexibility.

Is fish plate welding the same as welding rails together?

Not exactly. A fish plate is a mechanical joint bar used with bolts; ‘fish plate welding’ is sometimes used informally, but true rail welding refers to fusing rail ends (thermite, flash-butt) to form continuous welded rail, not welding the fish plate itself as the primary rail connection in modern CWR.

What welding standards apply to railroad rail welding?

The AWS D15 Committee develops railroad welding standards such as AWS D15.1 (railcars, locomotives, components) and AWS D15.2 (welding of rails for rail vehicles), which establish minimum requirements for procedures, qualifications, and workmanship.

What are the main methods to weld railway rails?

The primary methods are aluminothermic (thermite) welding for field joints and electric flash-butt welding (often in-plant or in-track) for high-quality, consistent welds. Other arc methods may be used for repairs per applicable standards.

Why is continuous welded rail used instead of fish-plated joints?

Continuous welded rail eliminates most joints, reducing impact loads, noise, maintenance requirements, and improving ride quality. However, it requires careful thermal stress management and specialized repair techniques.

Are fish plates still used today?

Yes. Fish plates (joint bars) are still used for insulated joints, temporary track, certain repairs, and in locations where CWR is not practical or where electrical separation is needed for signaling.

How are rail welds inspected?

Rail welds undergo visual inspection and non-destructive testing, most commonly ultrasonic testing, to detect internal defects. Acceptance is based on the governing standard and railroad specifications.

What causes problems in bolted fish-plate joints?

Common issues include loose bolts, worn joint bars, rail end batter, and gap mismanagement. These lead to increased impact forces, noise, and accelerated track component wear if not properly maintained.

Can fish plates be welded to rails?

In some repair scenarios components may be welded, but fish plates are primarily mechanical fasteners. Welding of rail components must follow applicable railroad welding standards and is not a universal substitute for qualified rail-end welding procedures.

What is the difference between thermite and flash-butt rail welding?

Thermite welding uses a chemical exothermic reaction to produce molten steel poured into a mold around prepared rail ends. Flash-butt welding uses electric resistance heating and forging (upsetting) to fuse ends together, often producing very consistent welds with high automation potential.