Gas distribution screens for electrostatic precipitators

Flue gas reaches the precipitator through a duct perhaps a tenth of the area of the ESP face, expands through an inlet nozzle, and must then travel through the fields at a low, even velocity. Left alone it will not: the jet from the duct drives a fast core through the middle of the casing while the corners recirculate. Gas distribution screens — perforated plates across the inlet face, fed by splitter and turning vanes in the nozzle — impose the even profile that every other part of the electrostatic precipitator design assumes. The industry acceptance criterion is ICAC standard EP-7: standard deviation of velocity σ ≤ 15 % of mean, measured by traverse across the treatment cross-section.

Uniformity is not cosmetic. The specific collecting area of the design-basis ESP, 75.7 s/m on 3,888 m² of plate, is a statement about residence time; a fast path through the casing is locally an undersized precipitator. Maldistribution also drives secondary damage: high-velocity zones scour collected dust off the collecting electrodes and re-entrain it, while dead zones drop ash in the inlet nozzle and on the screen itself. On bagasse and rice-husk duty the fly ash is high in abrasive silica, so a fast corner also erodes plates and vanes years ahead of schedule.

Distribution is staged. Splitter and turning vanes in the inlet nozzle take the duct jet and spread it through the expansion, doing the coarse work; one to three perforated screens at the ESP face then flatten what remains. Plate open area is the key selection, chosen in the 40–60 % range: lower open area corrects harder but costs pressure drop — which the ID fan pays for continuously — and blinds faster on sticky ash. Where dust drops out in the nozzle, the screen package includes hopper baffles so the inlet hopper is not short-circuited by sneak-by flow beneath the fields.

CFD versus physical flow model. CFD on the full duct-nozzle-casing geometry is the standard design route: it is fast, iterates cheaply, and resolves the vane geometry directly. A reduced-scale physical flow model still earns its cost where the inlet arrangement is unusual — twin inlets, tight duct bends hard against the nozzle — because it measures rather than predicts. Arrow treats neither as acceptance: the contract criterion is the site velocity traverse at the inlet face, σ ≤ 15 % of mean, witnessed before the internals go live.

A retrofit runs in five steps. First, a diagnostic velocity traverse of the existing precipitator, plus inspection of nozzle and screens, during a short outage. Second, CFD of the as-found geometry, calibrated against the traverse. Third, design of the correction package — vanes, screen open areas, baffles — iterated in the model until predicted σ ≤ 15 % of mean. Fourth, fabrication as bolt-in panels sized to pass through existing manways, so no casing cutting is needed. Fifth, installation in a planned outage and a verification traverse before hand-over. Guaranteed performance figures are stated per project after this assessment, on a stated basis (mg/Nm³, reference O₂, dry/wet, load range).

Arrow Energy Co., Ltd. supplies distribution screens and vane packages as internals for other OEMs' casings, not only its own. The compatibility question is purely dimensional — face dimensions, nozzle geometry, manway sizes, existing support steel — so every third-party retrofit begins with a dimensional survey of the casing, from which panels are detailed to suit. Where the survey shows the real problem lies elsewhere, Arrow says so: a precipitator with good distribution and poor plates needs collecting electrodes, not screens.