Jaws That Grip a Cubic Foot of Ballast BNSF Replaces the Hand Dug Sample Pit With a Backhoe Attachment

This application covers a steel tube with a hinged lower jaw that is vibrated down into a railway ballast bed, closed by a hydraulic ram to trap the stone inside, lifted out and then opened again to drop the sample into a bucket. It was filed by BNSF Railway Company, one of the largest freight railroads in North America, and names William Dombrow, Michael A. Wnek and Darrell R. Krueger as inventors. The application is a divisional of Australian application 2020352424, and the specification states a priority claim to US application 16/581,915 filed on 25 September 2019. Despite the title, nothing in the claim set is a system in the computing sense: claim 1 is a tool, claim 8 an apparatus and claim 15 a method.

The Problem

Railway track sits on a bed of crushed rock called ballast. The specification describes what it is for: carrying the load off the sleepers, draining water away, suppressing vegetation and stopping the track from creeping sideways under a passing train. The layer is typically six to twenty inches thick and the individual stones are one to two and a half inches across.

Ballast wears out. The stones crack and abrade under repeated loading, and the resulting fines, plus mud pumped up from the subgrade and anything spilled from the wagons above, gradually fill the voids between the remaining stones. Once the voids are full the layer stops draining and stops springing, and the track begins to settle unevenly. Railway engineers call this fouling, and in North America they score it with Selig’s Fouling Index, the mass percentage of material passing a number 4 sieve plus the percentage passing a number 200 sieve. That number is the reason anyone digs a hole in a ballast bed at all.

The specification’s background is brief on this point. It says only that ballast degrades over time and should be replaced, that samples are collected and analysed to evaluate the state of degradation, and that traditional methods such as hand digging pits disturb the surrounding ballast and expose the sample to contamination. It does not describe the laboratory analysis, does not mention any fouling index and gives no target figures for when ballast should be replaced. The tool measures nothing itself. It is a sample getter, and the measurement happens later, off the track, on the contents of the bucket.

The detailed description explains why an off the shelf answer does not exist. Soil samplers rely on the natural cohesion of soil to hold a core together while it is pulled out, and core drills rely on rock being solid. Ballast is neither. It is a loose pile of fist sized stones with no cohesion at all, and the volume needed for a useful gradation test is around a cubic foot, which the specification says is beyond the capacity of existing soil sampling equipment.

What This Invention Does

Claim 1 is a sample confinement tool with four elements. There is a first member whose upper end couples to a driving tool, which is itself coupled to a heavy duty vehicle. There is a second member joined to the first by a fastener, mounted so that it pivots about the axis of that fastener. There is a hydraulic actuator that controls the movement of the second member relative to the first. And the tool as a whole must be operable to insert a predetermined depth below a ballast surface, collect a ballast sample within its second end, and confine that sample by mechanically compressing it.

In plain terms it is a square steel tube that has been sliced lengthwise near the bottom so that one side of it can swing inward like a jaw. The description gives the dimensions: approximately four feet long, approximately eight inches wide and eight inches deep, with walls about half an inch thick. The upper foot is a closed square tube. Below that, the first member becomes a C section for about three feet, and a second C shaped member roughly three feet long overlaps its flanges and is bolted through at the midpoint, about eighteen inches up from the bottom. A hydraulic cylinder sits inside the hollow core. Rotate the second member one way and the depth of the mouth shrinks, squeezing the stone; rotate it the other way and the mouth opens and the stone falls out.

The rest of the rig is ordinary plant. The tool bolts to a driving tool, preferably a vibratory post driver, which bolts to the boom of a backhoe or similar machine, and the tool’s hydraulic cylinder is plumbed into the vehicle’s own hydraulic circuit through vented counterbalance valves. The operator positions the tool between two sleepers, vibrates it down, closes the jaw, lifts, swings over a bucket and opens the jaw again.

The numbers the specification does give are worth quoting. The predetermined insertion depth ranges from six to twenty inches, with eighteen inches named as the example. The confined sample is one cubic foot in volume. A controller can automate two of the steps: if a location sensor reports that the end of the tool is less than eighteen inches below the surface, the controller keeps driving until it is, and if a pressure sensor reports that the confined sample is below 2000 pounds per square inch, the controller raises the pressure until it reaches that figure. The specification is silent on why 2000 psi is the right clamping pressure, and it notes only in passing that the hydraulic pressure may be adjusted according to the type of ballast.

Key Features

• A hinged jaw rather than a closing bucket. The second member pivots about a single fastener at its midpoint, so a hydraulic cylinder inside the tube can narrow the mouth from roughly eight inches deep to less than that, trapping loose stone that has no cohesion of its own.
• Walls that isolate the sample before it is taken. Because the tube is driven down around the sample, its four walls separate the sample from the surrounding ballast during extraction, which the specification says prevents stone external to the tool from falling into the sample area and contaminating it.
• A fixed, repeatable sample geometry. Approximately four feet long, approximately eight inches by eight inches, driven to a stated depth of six to twenty inches, yielding one cubic foot. Controlling depth and area is how the specification proposes to improve sample uniformity between one site and the next.
• Vibratory insertion off the machine’s own hydraulics. A vibratory post driver on the boom drives the tool down, and the tool’s actuator runs from the vehicle’s hydraulic circuit through vented counterbalance valves, so no separate power pack is needed.
• Optional closed loop depth and pressure control. A controller can take location data, for example from a GPS sensor, and drive until the end reaches eighteen inches, and can take pressure data and clamp until the sample reaches 2000 psi.
• Fewer people standing on the track. The specification’s stated advantages include operator safety, on the basis that the tool is mechanically operated from a high visibility vehicle and reduces the number of workers who have to be on the railway to take a sample.

Who Is Behind It

BNSF Railway Company is one of the two large western United States freight railroads, running a network of 32,500 route miles across 28 states and three Canadian provinces, and has been part of Berkshire Hathaway since 2010. A railroad of that size maintains its own engineering research programme, and this filing is part of it rather than a product a supplier is trying to sell.

The context for the filing is public. A 2018 conference paper by BNSF’s Adam Bankston and Stephen Wilk of the Transportation Technology Center, Use of Track Based Inspection Technologies to Improve BNSF’s Ballast Maintenance Planning, describes the railroad’s shift to network scale ballast condition data gathered by ground penetrating radar, and explains the problem that drove it: Selig’s Fouling Index was historically measured by physically removing samples and running gradations, a process the authors call not ideal because it disturbed the track, required track time, was time consuming and was labour intensive. BNSF also runs an automated track inspection programme it describes publicly under the name ODIN. Read against that background, this application is an attempt to fix the four objections to physical sampling rather than to abandon physical sampling, which makes sense given that the radar output still has to be calibrated against real gradations.

On the paper trail, the specification is unusually clear for a divisional. Paragraph [0] states that this is a divisional of Australian application 2020352424, that the disclosures of that application and of its corresponding international application PCT/US2020/048676 are incorporated by reference, and that the application claims priority to US 16/581,915 filed on 25 September 2019. The parent Australian case is published as AU 2020352424 under the same title. The priority country is therefore the United States, on the specification’s own statement rather than on any inference.

Why It Matters

The interesting thing about this filing is how firmly it sits on the unglamorous side of a technology shift. Ground penetrating radar can now be towed over an entire network and will return a Ballast Fouling Index to a depth of about sixteen inches below the track surface without anyone getting out of the vehicle. That is plainly better than digging pits. But a radar estimate of a sieve result is still an estimate of a sieve result, and somebody has to go and get real stone to check it against. Note that the eighteen inch sample depth named in this specification sits just past the sixteen inch depth the radar interpretation covers.

So the value here is in the cost of the check, not in the check itself. If a physical sample takes a crew an hour of track time with shovels, the calibration set will be small and the fouling model will inherit that. If it takes one machine operator a few minutes without anyone leaving the cab, the railroad can afford a lot more ground truth. That is a maintenance planning argument rather than an engineering one, and it explains why a railroad rather than a tool manufacturer is the applicant.

What the specification does not do is claim the thing the title suggests. There are twenty claims, none of them directed to a system in any computing sense, despite a figure and several paragraphs devoted to an example computer system. The controller appears only inside claim 8, where the apparatus includes a controller configured to receive a location of the second end. Claim 1 does not require a controller at all, and claim 15, the method, requires neither a controller nor a hydraulic actuator, only the two members, the fastener and the compress and decompress steps. A reader looking for automated ballast sampling will not find it claimed here. What is claimed is a jaw on a stick.

Related Concepts

• Ballast cleaner – the maintenance machine that sieves fouled ballast out of the track, and the intervention a sample result is usually used to justify.
• Tamping machine – the other standard response to a settling track, which works only while the ballast still has voids to pack into.
• Sieve analysis – the laboratory gradation test performed on the cubic foot of stone this tool delivers to the bucket.
• Ground-penetrating radar – the non contact method now used to estimate fouling at network scale, which still has to be calibrated against physical samples.
• Core sample – the cohesive soil and rock sampling technique that the specification says does not work on loose ballast.
• Permanent way – the track structure as a whole, of which the ballast layer is the component being assessed.

---

AU 2026202232 was published in the Australian Official Journal of Patents on 9 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.