12.1 Direct Filtration (Subclass 2)

What DEP Means by Direct Filtration

Pennsylvania defines Subclassification 2 (Direct Filtration) as a series of processes implemented for substantial particulate removal consisting of coagulation and filtration. The definition explicitly includes flocculation after coagulation but explicitly excludes sedimentation. That single omission changes everything about how the plant is run, and the 44-question Subclass 2 examination is built around the consequences.

In a conventional plant the sedimentation basin removes the great majority of the solids mass and the filter polishes what is left. In direct filtration the filter bed does both jobs. All of the coagulated solids must be stored inside the media until backwash.

Source Water Screening: The Go or No-Go Decision

Direct filtration is only defensible when the raw water stays clean. The screening criteria an operator should be able to state are:

• Turbidity generally less than 2 NTU. DEP notes higher values can be treated for short periods but produce very short filter runs, and puts the general limit for raw turbidity at less than 25 NTU.
• True color less than 20 color units. True color indicates colloidal natural organic matter, which needs a higher coagulant dose than suspended solids and therefore loads the filter faster.
• Algal blooms below about 2,000 asu/mL; above that, filter run times suffer. In the source reservoir, algae growth is most commonly controlled with copper sulfate.
• Total organic carbon low, because direct filtration removes far less organic precursor material than conventional treatment, which matters for Stage 2 disinfection byproduct compliance
• Iron and manganese low, since oxidized metal solids consume bed storage capacity

When the source degrades beyond these limits, the plant must either add sedimentation or reduce production. Many Pennsylvania reservoir supplies run direct filtration for most of the year and are hydraulically derated during spring turnover.

Chemistry: Small, Tough Floc Instead of Large, Settleable Floc

This is the single most-tested distinction on the Subclass 2 examination. In conventional treatment the operator builds a large floc that will settle. In direct filtration a large floc is a liability, because it blinds the top of the bed and destroys the filter run. The target is a small, dense, shear-resistant pinpoint floc that penetrates into the media and attaches.

Practical consequences:

• Coagulant dose is typically one third to one half of the conventional dose for the same water.
• Cationic polymer is often used as the primary coagulant or as a coagulant aid because it produces tough floc at low dose without adding much solids mass.
• A nonionic or anionic filter aid polymer may be fed directly ahead of the filters at very low dose (commonly under 0.1 mg/L) to strengthen attachment inside the bed. Overdosing filter aid drives head loss up almost immediately.
• Flocculation time is short, typically 5 to 15 minutes rather than the 20 to 45 minutes used in conventional plants, and gentle enough to avoid growing floc too large.

The Turbidity Rules the Plant Is Judged By

Direct filtration is held to the same performance standards as conventional filtration, and the numbers are exam staples:

Filtration itself is credited with 2-log (99 percent) removal of Giardia, enteric viruses and Cryptosporidium; disinfection supplies the balance of the 3-log Giardia and 4-log virus treatment technique requirement.

Reading the Filter Cycle

Every run has three periods, and each has its own risk:

1. Ripening, immediately after backwash, when effluent quality is at its worst. Tools to shorten it are filter-to-waste (rewash), delayed start or filter rest so the media settles and compacts, slow start so flow ramps up without shearing residual material out of the bed, adding coagulant or polymer to the backwash water left standing in the filter box, and dosing filter aid during ripening only.
2. Steady state, when treatment is optimized and effluent quality is stable.
3. Terminal, when water quality indicators begin trending upward — the signal that the run is ending, whether or not head loss has reached its limit.

Turbidimeters have blind spots the operator must know: they do not detect activated carbon fines well because the black particles do not reflect light, they are insensitive to biological matter whose refractive index resembles water, and they never identify or size the particles producing the reading. That is why particle counting and, in troubleshooting, microscopic particulate analysis supplement turbidity on direct filtration plants.

Operating a Plant With No Safety Net

The sedimentation basin in a conventional plant is a buffer: if coagulation drifts for twenty minutes, the basin absorbs it. Direct filtration has no buffer, so the operating discipline tightens:

1. Continuous coagulation control. Streaming current monitoring, zeta potential or frequent jar testing verifies the dose against changing raw water. A dose set in the morning is not valid after a rain event.
2. Short filter runs are normal. Runs of 8 to 24 hours are common, versus 36 to 72 hours in conventional plants. Rapidly rising head loss usually means coagulant or filter aid overdose, not a dirty source.
3. Filter-to-waste is mandatory practice. With no upstream barrier, the post-backwash ripening spike is the plant's largest single pathogen risk, so initial filtrate is wasted until turbidity settles below the plant goal.
4. Individual filter effluent turbidity is the primary alarm. Because breakthrough is the failure mode rather than sludge carryover, the individual filter turbidimeters, not the basin, tell the operator what is happening.
5. Backwash capacity governs production. Every backwash consumes finished water and takes a filter offline. A plant washing filters every 10 hours needs meaningfully more filter area than one washing every 48 hours.

Trap to avoid: raising coagulant dose in response to rising filter effluent turbidity. In direct filtration an overdose builds floc that is too large, blinds the surface, and drives both head loss and breakthrough. Verify the dose against a jar test or streaming current reading before changing it.