Turbidity That Will Not Settle: Dispersive Clay on a Cut Slope
A highway widening project has exposed 4 acres of cut slope in a stiff, gray, slickensided clay. Runoff from the slope is the color of milk chocolate. The sediment trap holds it for three days and turbidity at the outlet still reads 850–1,200 NTU, against a state benchmark of 280 NTU. Lab results on the clay: 38% clay-size fraction, cation exchange capacity 22 meq/100 g, exchangeable sodium percentage 14, SAR of the extracted pore water 8.5, plasticity index 31. The contractor's response has been to add a second row of silt fence and a line of straw bales below the slope, and to enlarge the sediment trap. A half-inch shower still produces a visible plume 300 ft down Miller Creek.
1. Diagnose why the turbidity does not clear, and explain why more barriers and a bigger trap will not fix it.
Read the worked answer
This is a dispersive soil, not a sediment-control capacity problem.
The numbers say so: ESP of 14 and pore-water SAR of 8.5 mean sodium dominates the exchange complex. Sodium expands the diffuse double layer around each clay particle, so the repulsive force between particles exceeds the attractive van der Waals force and the clay deflocculates - it comes apart into individual particles and stays apart. Those particles are in the clay-size range, where Stokes' law gives settling velocities on the order of hours to days per foot for the finest fraction, and Brownian motion keeps the smallest fraction in suspension indefinitely.
The consequence is diagnostic: the load leaving the site is not bed load that a barrier can strain out or a trap can store; it is wash load. Adding silt fence rows captures a little coarse material and adds more failure points without touching the colloidal fraction. Enlarging the trap increases residence time, and residence time is not the limiting variable - three days already produced 850–1,200 NTU. There is also a feedback loop: rills on the cut indicate the slope itself is eroding, so every storm generates a fresh supply, and rainwater is low in ionic strength, which promotes further dispersion when it contacts the clay.
2. What changes do you require in the plan, and how do you verify the treatment will work before committing the site to it?
Read the worked answer
Two attacks, in this order - source first, then chemistry.
1. Reduce detachment on the 4-acre cut. Chemically treated water downstream of an actively eroding slope is a losing game. Require: a lined or stabilized diversion above the cut to keep upland water off it; a rolled erosion control product or bonded hydromulch with tackifier over the exposed face; slope-length interruption or benching; and prompt permanent stabilization of the topsoil surface.
2. Treat the dispersed clay chemically. Supply the calcium needed to displace sodium and collapse the double layer:
- Gypsum (calcium sulfate) broadcast on the surface ahead of forecast rain, or dosed into the trap inflow. Gypsum is preferred over lime here because lime raises pH and can worsen dispersion of some clays.
- Anionic polyacrylamide as a flocculant aid to bind the colloids into settleable flocs; it complements calcium but does not substitute for it.
- Provide a settling cell and a controlled dosing point, not a manual toss into the ditch.
3. Verify, do not assume. Run a jar test on actual site runoff: a series of doses, 30–60 minutes of settling, measure turbidity against the 280 NTU benchmark. Pick the lowest dose that reliably meets target - overdosing wastes money, raises sulfate and pH, and can re-stabilize the suspension. Design for the verified dose with a safety factor and a dosing log.
4. Test the stockpiled topsoil separately. Sodic subsoil will not revegetate readily; it likely needs gypsum amendment and a suitable species mix, or the slope will keep failing long after construction.
3. The contractor says the creek is already turbid from upstream farms, so the 280 NTU benchmark is unachievable and should be waived. How do you respond, and what happens if the benchmark cannot be met?
Read the worked answer
Background turbidity does not authorize your discharge, and the benchmark is not a waiver mechanism - it is a decision point that triggers action.
How to handle it:
- Isolate the question. Sample the creek upstream and downstream of the outfall at the same time. If the upstream reading is low and the downstream reading is 900 NTU, the 'already turbid' claim is irrelevant to the contribution being measured.
- Know what the benchmark is. Usually it is a monitoring trigger, not an effluent limit: exceeding it obligates the operator to investigate, take corrective action, and document - and repeated exceedances invite the agency to characterize the discharge. Some permits have numeric limits, in which case the obligation is stricter.
- Document the response chain. Each exceedance should show the reading, the identified cause, the action taken, and the follow-up reading. That record is what demonstrates good-faith compliance when an inspector or enforcement officer arrives.
- If treatment cannot get there, say so honestly and escalate. Options: intensify source control, add filtration or a larger treatment cell, reduce exposed area, or notify the permitting authority and seek direction. What is not acceptable is to keep reporting exceedances with no change and call it weather.