ProductionNewVerified Real deployment at a named or precisely described company, with a measured result from a primary source.
AI-guided welding robots see the joint and adapt to each part, so fabrication shops weld more with fewer scarce welders (Joe White Tank, Ganotec-Muga Fab, TYCROP)
Added 2026-11-17 · Last updated 2026-11-17
Manufacturers in this case: Joe White Tank · Ganotec-Muga Fab · TYCROP · Mine Rite · all manufacturers using AI
Role
Welding and fabrication production (pipe spools, pressure vessels, heavy structures and chassis) · Typical users: fabrication shop manager, welding supervisor, production manager, manufacturing or automation engineer, operations director of a metal fabrication or heavy equipment plant
The problem
Many fabrication shops cannot find enough skilled welders, and their work does not suit classic welding robots: parts are big, heavy, often hand-fitted and never quite match the CAD model, and small batches make robot programming hard to justify. Joe White Tank (Fort Worth, Texas, USA), a fabricator of custom tanks, pressure vessels and industrial piping since 1942, had piping projects with hard starts and stops and little time to ramp its labor up and down, and did not want to hire and lay off welders job by job. Ganotec-Muga Fab (Canada) needed more shop productivity with fewer welders and wanted an average welder to match the quality and output of a highly skilled one, but conventional equipment took too much floor space and handled small pipe sizes poorly. TYCROP, a Canadian manufacturer of mobile power generation equipment, builds large, complex assemblies and chassis where variable weld paths, changing product lines and the skilled labor gap had kept automation out; its vice president of operations said young people are not considering the trades the way they used to.
How AI helps
Cameras and sensors on the welding robot locate the joint and the weld pool, and AI models adapt the torch path and welding as the weld progresses, so the robot copes with part-to-part variation instead of repeating a fixed program. A person stays in charge: the shop's welders become operators who load parts, supervise and handle the jobs the robot cannot. At Joe White Tank, Novarc's Spool Welding Robot (SWR), a collaborative robot for pipes, pressure vessels and other roll-welded work, uses adaptive torch control and machine learning that detects weld pool features; the operator works with a joystick instead of wearing a hood, and the robot lays the root pass itself. Ganotec-Muga Fab runs the first dual-torch SWR Novarc built, with NovAI vision and cognition, and deliberately put average welders on it. At TYCROP, a 27 by 84 foot Path Robotics cell uses computer vision to see each joint and locate it in space, and AI plans and executes the motion; Path's Obsidian model, trained on welding data from customer sites and its own data farm, decides how many passes are needed to fill a joint and where they go.
Step by step
- List your welding work for the last year by part family: weld length or diameter-inches, process, material, batch size, hours and repairs.
- Pick the families with the most repetitive, high-volume welding (pipe spools, chassis, poles, frames) where welders are the bottleneck.
- Send suppliers real parts or drawings with your actual fit-up variation and ask for a trial weld and a written cycle time, not a brochure figure.
- Plan the flow around the robot: fitting, positioners and part changes often limit output more than welding speed, as Joe White Tank found.
- Qualify the welding procedures with your inspector and code requirements (ASME or other) before production and keep X-ray or other inspection samples.
- Train experienced welders as operators and measure output per shift, arc-on time, repair rate and lead time against your baseline.
Tools
- Novarc Spool Welding Robot (SWR) with machine learning weld pool detection: used at Joe White Tank
- Novarc Spool Welding Robot (SWR), dual torch, with NovAI vision and cognition: used at Ganotec-Muga Fab
- Path Robotics AI welding cell with the Obsidian model and ALM positioners: used at TYCROP and Mine Rite
- Other tools to evaluate: welding robots with seam tracking and offline programming from robot and welding equipment makers, collaborative welding cobots, robots-as-a-service offers
- To start without new equipment: your weld job history and labor data and any LLM (ChatGPT, Claude, Copilot, Gemini) with the prompt below
In the AI tools index: Novarc Spool Welding Robot (SWR) · Path Robotics AI welding cell · all AI tools used by manufacturers
Result
Joe White Tank (Novarc case study and The Fabricator article written by Novarc): pipe welding productivity up to 5 times (3.1 times cited for the profitability gains), capacity increased by 400% without reducing quality, one SWR replacing four manual weld cells in 25% to 30% less space, over 70% fewer welders dedicated to piping, and welds that pass X-ray inspection; the shop can now accept short lead-time jobs and larger bids. With a third positioner added in October 2025 it reached 508 factored diameter inches and 6.4 hours of arc-on time in a single day. Ganotec-Muga Fab (Novarc case study): pipe welding productivity up 220% to 300% and repair rate down to 0.5% on 12 to 36 inch carbon steel and chrome pipe, against its own 10-year baseline of just below 2%; welders were comfortable on the robot after one to one and a half weeks. TYCROP (FSMdirect magazine): the cell reportedly runs 24 hours a day, cuts manual welding time by 91% and completes more than half of all the welds on each chassis; in the Path Robotics and ALM Positioners announcement, its vice president of operations and engineering said it now consistently meets customer demands for high-quality parts delivered on schedule. Mine Rite, a Wyoming maker of mining equipment, reports 85% of the work automated on its Path Robotics cell and its largest truck body shipping seven weeks earlier.
As reported by the source; most figures are vendor or self-reported and not independently verified.
Human review: cautions
- Most figures come from the robot supplier (Novarc case studies and an article written by Novarc) or a magazine story built on Path Robotics' customers; they are not independently audited.
- Joe White Tank's figures vary by source (up to 5 times, 3.1 times, 400% capacity); use them as an order of magnitude, not a promise.
- Part of the gain comes from the positioners and shop flow, not from the AI alone: Joe White Tank credits multiple positioners and limited downtime between spools.
- Welding procedures and welders or operators must still be qualified to your code (ASME or other), and inspection stays in place.
- Fit-up still matters: Novarc sets fit-up requirements and trains fitters, so budget time for the upstream process too.
Try it yourself
You are a welding automation engineer. I want to find which of our welding work is the best fit for an AI-guided welding robot and build the business case. I will paste: (1) our welding jobs for the last year [part family, process, material, weld length or diameter-inches, batch size, welding hours, fit-up quality, repairs]; (2) our welding labor [number of welders, open positions, overtime hours, hourly cost]; (3) our lead times and the jobs we turned down or lost for capacity reasons, if known. Do this: 1. Group the work by part family and rank the families by welding hours, repeatability and fit-up variation. 2. Recommend the two families to automate first and explain why, including what makes them hard for a classic programmed robot. 3. Estimate a range of hours freed per year at 2 times and 4 times today's output, and what that means in welders and lead time. 4. List the questions to ask suppliers (trial welds on our parts, cycle times, fit-up tolerance, positioners, procedure qualification, support, buy or service contract). 5. Draft a trial plan with the measures to compare: output per shift, arc-on time, repair rate, inspection results. Use only the data I paste in; if something is missing, write 'not found' or list it as a question instead of guessing. Do not send, order, post or change anything in any system: produce drafts for my review and approval.
Want more like this? Join the newsletter
Related cases
- AI generates CNC toolpaths so programming stops being the bottleneck
- AI reads X-ray images of aluminium castings to find porosity and shrinkage, on existing inspection systems (Hengst SE, Fraunhofer IIS)
- AI warns paper machine operators before a sheet break and points to the likely cause, so mills lose fewer hours to breaks and rethreading (Sappi, Georgia-Pacific, a Voith paper line)
- AI scores ergonomic risk from a phone video of the job, so safety teams fix the worst lifts before they cause sprains and strains (Superior Tube Products, Latham Pools, Johns Manville)
Sources
- https://www.novarctech.com/resources/case-studies/joe-white-tank/
- https://www.thefabricator.com/tubepipejournal/article/tubepipefabrication/pipe-welding-integrates-machine-learning-to-boost-production
- https://resources.roboticscouncil.ca/boosting-pipe-welding-productivity-by-300-with-novarcs-spool-welding-robots
- https://fsmdirect.com/finding-purpose/
- https://www.prnewswire.com/news-releases/path-robotics-and-alm-positioners-deliver-breakthrough-welding-automation-for-tycrop-302446732.html
Weekly: 1–2 practical AI use cases for your plant
Free, one email a week. By subscribing you agree to receive the weekly newsletter (and its sponsor message) from AI Agents for Manufacturing. Every email includes an unsubscribe link, and you can unsubscribe anytime. See our Privacy Policy.