BRENNEN

Industries / Construction and property / Heavy civil

Dredging

Dredging digs sand, silt, clay and rock from the bottoms of harbors, rivers and channels and moves it somewhere else, to keep the water deep enough for ships, to build beaches, marsh and land, and to remove polluted sediment.

1575

Joost Janszoon Bilhamer, a Dutch military engineer, invented the mud mill, a chain of buckets hung between two barges and turned by men walking a treadmill.

the-low-countries.comroyalihc.com

32

Milestones achieved across 6 domains.

7

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Until the mid-1800s most dredging in the Low Countries was done by hand, with a long pole and a leather bag, though Joost Janszoon Bilhamer's mud mill of 1575 put a chain of buckets on a treadmill. Steam drove a bucket dredger on the Thames from 1803. In 1857 a centrifugal pump at Charleston, South Carolina, sucked sand up a pipe, the idea behind today's suction dredgers.

Where it stands

The largest hopper dredgers carry 46,000 m³ in one load, and sediment is pumped through pipelines more than 20 km long to build marsh. IADC, the dredging contractors' association, puts 2024 turnover in the open world market at €7.5 billion, leaving out closed markets such as China and the United States, where federal dredging contracts reached a record $2.4 billion in fiscal 2024 by Michael Gerhardt's count. Dredgers also build beaches, marsh and land, and rules on dumping at sea, drifting mud and wildlife set when and how they may work.

What is open

Most open jobs are measuring and checking. Free survey archives, satellite images and AI coding agents make it practical for one person to score shoaling forecasts, count how much dredged material is put to use, measure how long beach fills last and test plume forecasts against the harm found afterward. Other jobs need only a test tank: reading mud stiffness from echoes, testing density meters that need no radioactive source, and measuring what a clamshell bucket lets out when debris jams it.

Domains

AchievedWatch forOpen

Open in dredging

3 of 12 open deliverables

Open · Surveys and monitoring

Score shoaling forecasts in public

A public test of how well channel shoaling can be forecast. For at least 100 channel reaches with ten or more Corps surveys each, hold back each reach's latest surveys, forecast them from the earlier ones with a simple shoaling-rate method and with a machine-learning model, and publish the errors.

Built with what exists

eHydro survey files from the Corps' open data portal; Python with GDAL or rasterio to grid each survey onto its channel reach; scikit-learn or LightGBM for the learned model; an AI coding agent to write and rerun the pipeline.

Credit · unclaimed

Open · Dredging vessels

A public hopper-loading test bench

An open simulator of a hopper dredger's loading phase, when the drag head sucks up sand and the hopper fills until water and fine sediment spill overboard. It should have a standard interface, so anyone can plug in a controller that sets sailing speed, pump flow and when to stop loading, and a scoring script for payload per hour, sediment lost overboard and energy used.

Built with what exists

Python; the Gymnasium interface used for reinforcement learning tests; Stable-Baselines3 for learned controllers; loading physics from the open TSHD_production_estimation model on GitHub and the TU Delft thesis on loading-phase energy written with Van Oord; an AI coding agent.

Credit · unclaimed

Open · Pumping and placement

Count the 70 percent

A public tally of where the Corps' dredged material went in each of the last five fiscal years: for every dredging job, the volume and whether it was put to use (beaches, marsh, islands, fill) or disposed of (open water, confined sites, upland). Publish the national and district shares with the method and code.

Built with what exists

The Corps' public Dredging Information System, whose web app carries placement types and actual placement quantities for each job; contract solicitations on SAM.gov; a web agent to pull the placement terms from contract documents for a person to check (see the Note "Deleting 80% to start over"); Python and pandas. Web agents that read contract files make a job-by-job tally practical for one person.

Credit · unclaimed

Where it sits