Tunnel borer components can be large enough to require workstands to reach areas for plasma cutting.

A new tunnel boring machine, viewed from the front end, on the Robbins assembly line.

Another Robbins borer, with a material conveying system trailing behind. Note that different style of cutterhead.

Cleaning up a notch cut into a heavy cylindrical part.

At any given moment, human workers are creating more than one hundred tunnels to bring water to cities, channel waste to treatment sites, and create shorter routes for highways and rail lines. Up until the middle of the last century, the only way to tunnel was either through the use of picks and shovels or explosive charges to break up the rock. The hole then had to be braced, the fragments removed and the cycle started again. The work was costly and dangerous, and it took months and years to progress a few miles.

James Robbins, who was a consulting engineer in the coal industry, completed the first practical tunnel boring machine in 1953. To date, The Robbins Company (www.robbinstbm.com) has produced TBMs that have excavated tunnels from 1.6 meters to 12.87 meters in diameter, with machine weights ranging from 50 metric tons to more than 1,500 metric tons. The modern machines can bore 4,000 ft. per month, on average. Each machine is unique while similar in basic design to the original Robbins TBM, every machine must be designed to address a different boring environment.

Design and materials
Like other huge-scale construction equipment, the structure of a tunnel boring machine is primarily welded together for strength. Chewing through hard rock and earth demands high power — the entire machine, during operation,is subjected to a constant 4 gs to 6 gs of acceleration. Rotating the cutterhead into hard rock puts a million in./lbs of torque on the main gear. Holding the machine on each side are gripper shoes, which during tunneling operations must withstand 4 million lbs. to 6 million lbs. of thrust. Pushing the cutterhead into the rock face are four or six propel cylinders, each exerting 21/2 million lbs. to 4 million lbs. of thrust force. There’s not much room on a tunneling schedule for work stoppages to repair popped welds, nor is there a crane readily at hand to hold a 150 ton cutterhead with cracks, when the machine is 200 ft. below the surface of the earth.

Robbins’ TBMs are designed initially from 2-dimensional CAD drawings, which allow the engineers to fit together the job specific of each machine. Cutterhead design usually differs from machine to machine, with carbide components sometimes added at key points. The cutterhead may include a simple system, installed around the outer circumference to help break rock down to finer chunks, that resembles a large-opening grill constructed of heavy hardened steel bars (“grizzly bars”). The machine specifications may be for an open or a shielded design, the latter incorporating a steel outer cylinder that protects workers while the machine processes soft or unstable rock. Other component options include scrapers for removing crushed rock and muck to a conveyor system that traverses the center of the machine from the cutterhead to the aft.

3D modeling and finite element analysis programs, based on the 2D drawings, then help the designer examine potential stresses that would affect the machine’s structure. Gary Thomas, production and operations manager at the company’s Solon, Ohio, office, says that use of such software tools lets the company determine if thinner plate, more gussets or other structural strengthening enhancements can be applied to a given area.

Material cutting and welding
A typical Robbins TBM is constructed from a combination of A36 steel, with average yield strength of about 48 ksi (ultimate strength about 70 ksi); and A572 high-strength, low-alloy columbiumvanadium structural steel. Inserts are made from alloy materials. Most of the pieces that go into making a machine are from plate stock, burn-outs and pre-machined stock, with thicknesses from 2-in. to 8-in. Some Ibeam and pipe stock is used for structural integrity, and a few other shapes (W-formed steel, for example) may be used, depending on the machine. Outside job shops cut the plate on plasma-burning or water-jet tables, the latter allowing for the creation of complex shapes, such as ring gears, up to 6-in. thick without need for post-machining. Thomas cites how in past years the manufacture of a probe drill ring gear required many steps of heat treatment, machining and inspection. “An average probe drill ring gear would cost $20,000 to machine,” remarks Thomas. “Now we make the same gear for $4,000 on a plasma or water-jet table.”

Other cutting tasks are performed in-shop with oxy-acetylene-, propane-and natural gas-fed torches. The main skeleton of a cutterhead is made of A36, with the face assembled from semi-circular segments. A cylinder of A36 supports the installation of the gear drives, bearing and ring gear that turn the cutterhead.

Most welds on a TBM are full penetration, with a depth of 1 1/4-in. to 1 1/2-in. The material to be welded is pre-machined for a root pass. The first pass is laid down with rod or wire, followed typically by a magnetic-particle check to make sure the pass is clean and has no cracks. The remainder of the weld is built-up with stringers — weld weaving is not permitted on Robbins machines. The stringers are required to be 1/2-in. or less in thickness, because of the higher potential for cracks at larger sizes. Once the first side is fully welded, the back of the joint is air arced to ensure full penetration. Carbon that is introduced during air arcing into the surface is removed by grinding down 1/4-in. to 3/8-in. of thickness, bringing the weld down to the original root pass. Another magtest is performed to check again for any cracking, and ultrasonic-testing is performed routinely during manufacturing to make sure the weld does not have any porosity. Submerged arc welding is used on large round parts and rod stock.

Maintaining quality
If a weld does not pass final testing, the joint is hand ground, remachined and re-welded. “Hopefully, we catch all flaws before stress relieving,” says Thomas, “because stress relief adds other issues to the structure and joints. If we do have to go back in and repair a joint, we use TIG to ensure less heat and distortion.”

Inspection and quality control at Robbins goes deeper than the assembly line. The company closely monitors all machinery and wire lots coming from their suppliers.

Lots and codes on welding materials are number matched for tracking purposes, should there be a problem. Each production supervisor, at the beginning and end of shift, has to record amperage settings and serial number on each welding machine, as well as wire feed rates, where appropriate. Anyone performing stick welding must put the remaining stub of used stick in a box at each station. The boxes are collected daily, and the final stub lengths are recorded to show that the stick wasn’t used past the flux. Likewise, a record is made if a rod is dropped and some of the flux comes off.

Of course all of Robbins’ weld staff, including new hires, are certified. A Level 2 quality inspector is on staff to test applicants, who are rarely hired with less than 10 years of welding experience. Mandatory retraining with an outside training firm takes place every two years. Says Thomas, “We have a lot of welding applicants who are production welders, and have done a lot of spot and tack welding. This is different, with weld-work taking place inside a piece that’s already heated for post-heat treatment to 375 degrees to 500 degrees to prevent cracks. The work is hard.”

As a result, failure rate at Robbins is low. “In our business view,” Thomas comments, “any error from the office to the shop is multiplied ten-fold. If the error goes uncorrected from the shop to the field, that’s another tenfold. So, if an order comes from the office specifying a 3/4-in. weld, and it should have been a 1 1/4- in. weld, and the product cracks in the field, the correction amounts to 100 times the effort of doing things right from the start. The costs to correct an error are exorbitant.”

On the average, a new 20 ft. diameter TBM can take 8 months to one year to build. Because each machine is made for a specific job, Robbins buys back the machine and refits it for another application. As a result, every boring machine is made for a life expectancy of 20 years to 30 years. It’s important to Robbins, then, that their staff is dedicated to the work. As Thomas, who has 25 years spent designing powertrains for TBMs, observes, “Typically, when people start working on these machines, they find it gets in their blood. We have some welders who’ve been with us more than 15 years.” The machines are like UFOs, he adds — few people ever see one, but working on such a huge project is not easily forgotten.