96.67 %
Design-basis collection (4 fields)
Technology · Dry electrostatic gas cleaning
Electrostatic precipitator (ESP): design, sizing and selection
An electrostatic precipitator (ESP) charges dust in a corona field and drives it onto earthed collecting plates. Sizing turns on specific collecting area: in our bagasse design basis, SCA 75.7 s/m across four fields collects 96.67 % of a 720 mg/Nm³ inlet load, for 24 mg/Nm³ at the stack @ 6 % O₂ dry.
24 mg/Nm³ @ 6 % O₂ dry
Design-basis outlet dust
75.7 s/m
Specific collecting area (SCA)
3,888 m²
Collecting area, design basis
01 — Position in the gas path
Where the electrostatic precipitator sits in the flue-gas train
The last dry stage before the induced-draught fan and stack — and the stage that decides compliance.
An ESP is the final particulate collector in most solid-fuel flue-gas trains. Upstream sit the boiler, heat-recovery surfaces and — on high-dust biomass duty — a mechanical or hybrid pre-collector; downstream sit the induced-draught fan, ash conveying and the stack. Arrow Energy Co., Ltd. designs the ESP as part of that train, not as an isolated box, within our wider gas cleaning and emission control technology range. The selection changes everything behind it: gas-side pressure drop of an ESP is roughly 10–30 mmWC (0.1–0.3 kPa), against 1–2 kPa for a fabric filter, so fan absorbed power and draught margin differ materially between the two.
On bagasse and rice-husk boilers, where raw dust runs 2,000–15,000 mg/Nm³ with abrasive silica ash, we normally place the proprietary Electrocyclone pre-collector ahead of the ESP. It removes about 90 % of the particulate load with near-zero moving-part maintenance, so the ESP is sized for 500–800 mg/Nm³ instead of 6,000 mg/Nm³ — fewer fields, smaller casing, and hopper ash split between a coarse abrasive stream and a fine stream. Where the permit demands below 10 mg/Nm³ regardless of ash resistivity, a bag filter is the alternative final stage; the trade is fan power and bag replacement against insensitivity to resistivity.
02 — Working principle
Charging, collection, rapping — the physics that sets the limits
Three steps, each with a failure mode that shows up as stack opacity.
How does an electrostatic precipitator work?
Discharge electrodes held at negative 45–110 kV ionise the gas in a corona sheath; dust particles crossing the field pick up charge and migrate to earthed collecting plates at an effective velocity of 2–15 cm/s depending on particle size and resistivity. Rapping then dislodges the collected layer into hoppers in coherent sheets.
Corona onset occurs when the local field at the discharge-electrode surface exceeds the breakdown strength of the gas — around 30 kV/cm at the sharp points and edges of the electrode, which is why emitter geometry (spikes, barbs, spiral wire) matters as much as supply voltage. Between onset and spark-over lies the working window; automatic voltage controllers hold each field just below spark-over, because a particle's saturation charge scales linearly with field strength, and migration velocity — which is charging field times collecting field divided by drag — scales roughly with its square.
Particles above about 1 µm charge by field charging (ion bombardment along field lines); submicron particles charge by diffusion. The resulting migration velocity w rises with particle size, which is why the inlet field collects the coarse fraction quickly and the outlet field works almost entirely on fines — and why particle size distribution belongs in every ESP enquiry (our bagasse design basis has inlet d₅₀ of 21 µm after the pre-collector).
The collected layer must conduct its charge to the plate. That is the resistivity window: between roughly 10⁴ and 10¹¹ Ω·cm the layer drains charge at a workable rate. Above 10¹¹ Ω·cm the voltage drop across the dust layer exceeds its breakdown strength and back-corona sets in — the layer itself discharges, injecting positive ions into the gap that neutralise incoming particle charge. Symptoms are high current at depressed voltage and collection that can fall by tens of percentage points. Below 10⁴ Ω·cm the opposite happens: dust loses its charge on contact, the electrical holding force vanishes, and rapped material re-entrains. Moist biomass flue gas (bagasse fires at ~50 % fuel moisture) generally keeps ash comfortably inside the window at 150–180 °C; dry low-sulphur coal ash and dry rice-husk silica push toward the upper limit.
Rapping — magnetic-impulse rappers or tumbling hammers striking the plate and frame suspension — must shear the layer off in sheets a few millimetres thick, not as a re-entrainable cloud. Details, intensities in g and sequencing are covered under ESP rapping systems.
03 — Sizing
Deutsch-Anderson, worked, and why we de-rate it
One equation carries the sizing; one exponent keeps it honest.
How is an electrostatic precipitator sized?
By the Deutsch-Anderson equation, η = 1 − exp(−w·A/Q): collection efficiency η follows from migration velocity w, collecting area A and gas flow Q. The ratio A/Q is the specific collecting area (SCA). Our bagasse design basis uses SCA 75.7 s/m across four fields for 96.67 % collection.
DESIGN BASIS — ILLUSTRATIVE CALCULATION, NOT A GUARANTEE · all concentrations @ 6 % O₂ dry
Take the design basis: a 60 t/h bagasse-fired boiler, raw dust 6,000 mg/Nm³ at the train inlet. The Electrocyclone first stage collects 88 %, leaving 720 mg/Nm³ at the ESP inlet. The four-field ESP provides A = 3,888 m² of collecting area; with A/Q = 75.7 s/m, requiring an outlet of 24 mg/Nm³ means η = 1 − 24/720 = 96.67 %. Inverting Deutsch-Anderson:
w = −ln(1 − η) / SCA = ln(30) / 75.7 s/m ≈ 0.045 m/s = 4.5 cm/s
That effective migration velocity — 4.5 cm/s — is a realistic, conservative figure for bagasse ash at 150–180 °C behind a pre-collector, which is the check that the sizing is buildable. The train collects 99.60 % overall, and the unit absorbs 177 kW.
Deutsch-Anderson assumes every particle migrates at the same w. Real dust is polydisperse: the coarse fraction disappears in the first field, and what reaches the outlet field is progressively finer, slower material. Extrapolating Deutsch with a single w therefore flatters added collecting area. Practice corrects this with the Matts-Öhnfeldt form, η = 1 − exp[−(wk·SCA)k] with k ≈ 0.5: efficiency still rises with SCA, but with the square-root damping that field data across fuels actually shows. We size with k ≈ 0.5 so that the margin quoted for an added field is the margin the stack will see — not a Deutsch fiction.
04 — Availability
The n−1 field rule
Compliance is judged on the bad day, not the commissioning test.
What happens to emissions when one ESP field trips?
Penetration rises sharply but not catastrophically. In our four-field design basis, one field out of service moves outlet dust from 24 to 57 mg/Nm³ @ 6 % O₂ dry — still inside typical biomass permit limits of 80–120 mg/Nm³. That is the n−1 condition, and we quote it on every multi-field ESP.
Fields do trip in service: a broken discharge electrode shorts a bus section, an insulator tracks in wet-season humidity, a TR set faults. The repair usually waits for the next planned outage, so the plant may run days or weeks on n−1. An ESP sized only to pass its performance test at n fields puts the operating permit at the mercy of a single wire. Sizing so that n−1 still clears the limit — and stating the n−1 number in the offer — converts a stack-violation risk into a maintenance-scheduling question. It also disciplines the field count: four smaller fields with independent transformer rectifier sets give better n−1 arithmetic than two large ones, at similar total collecting area.
05 — Selection by fuel
ESP selection across fuels and processes
Inlet load, temperature and resistivity behaviour set the architecture before any drawing is made.
| Fuel / process | Typical inlet dust | Gas temperature | Resistivity behaviour | Design notes |
|---|---|---|---|---|
| Bagasse (sugar mill) | 2,000–8,000 mg/Nm³ | 150–180 °C | Mid-window; ~50 % fuel moisture keeps the layer conductive | High-silica abrasive ash and char carryover — Electrocyclone pre-stage, wear allowances on inlet field |
| Rice husk | 5,000–15,000 mg/Nm³ | 150–180 °C | Ash 85–90 % amorphous silica; resistivity climbs toward 10¹¹ Ω·cm as gas dries | Heaviest biomass dust load; pre-collector effectively mandatory; ash has silica-extraction value |
| Wood chip / wood waste | 1,000–5,000 mg/Nm³ | 140–180 °C | Mid-window; potassium-salt submicron fume in the fines | Outlet field does the work on fume — favour SCA over field strength; watch unburnt-carbon spikes |
| Palm residues (EFB / fibre / shell) | 2,000–6,000 mg/Nm³ | 150–180 °C | Mid-window but K- and Cl-rich ash | Sticky alkali deposits and chloride corrosion — higher rapping intensity, material upgrades on internals |
| Coal, low-sulphur | 5,000–30,000 mg/Nm³ | 120–160 °C | Often above 10¹¹ Ω·cm without SO₃ in the gas — back-corona risk | Generous SCA, wide 400 mm spacing, or flue-gas conditioning; resistivity data before sizing |
| Cement kiln / raw mill | 20,000–80,000 mg/Nm³ | 90–150 °C after conditioning tower | Swings between mill-on (moist, conductive) and mill-off (dry, resistive) | Size for the mill-off case; CO trip interlock mandatory; clinker-cooler duty is a separate hot, abrasive case |
| Kraft recovery boiler | 5,000–30,000 mg/Nm³ | 150–200 °C | Low-resistivity sodium-sulphate salt cake — fine, sticky, prone to re-entrainment | High SCA, heavy continuous rapping, drag-chain hoppers; salt cake is recovered process chemical, not waste |
| Waste-to-energy grate | 1,000–5,000 mg/Nm³ | 150–200 °C | Variable with waste mix; fine fume and volatile metals | ESP as pre-collector ahead of sorbent injection and bag filter where acid-gas limits apply |
Reading the table as a selection guide: bagasse and rice husk are two-stage-train fuels — the load is high and abrasive, so a mechanical-electrostatic pre-stage protects and shrinks the ESP; reference class: sugar mill, Thailand, 170 t/h bagasse. Wood chip and palm residues are moderate-load fuels where fume and ash chemistry, not bulk load, size the box; palm chlorine drives material selection on collecting electrodes. Low-sulphur coal is a resistivity problem first: measure or estimate resistivity before quoting, and if it sits above 10¹¹ Ω·cm, buy area or conditioning, not voltage. Cement is a transient problem — the mill-on/mill-off swing and CO excursions define the controls as much as the sizing. Recovery boilers are a rapping and re-entrainment problem at the opposite, low-resistivity end of the window. Guaranteed figures for any duty are stated per project after the technical assessment, on a stated basis (mg/Nm³, reference O₂, dry/wet, load range).
06 — Enquiry data
What we need to size an ESP
Eight inputs; the first four decide 90 % of the price.
- Gas flow
- Nm³/h dry and actual m³/h at temperature, at design load and maximum continuous rating
- Gas temperature
- Design point and range, °C — the ESP casing is rated ≤ 200 °C; excursions above it are a design case, not a footnote
- Moisture
- Vol % H₂O in flue gas — the strongest single influence on ash resistivity
- Inlet dust load
- mg/Nm³ with reference O₂ stated, dry basis, at design and worst fuel
- Particle size distribution
- d₅₀ and submicron fraction; design basis d₅₀ 21 µm at ESP inlet behind the pre-collector
- Ash chemistry
- SiO₂, alkalis (K, Na), Cl, unburnt carbon — abrasion, stickiness, corrosion and conductivity all live here
- Resistivity
- Measured in-situ where possible, else inferred from ash analysis, moisture and temperature against the 10⁴–10¹¹ Ω·cm window
- Permit limit and basis
- mg/Nm³, reference O₂, averaging period — plus the footprint and draught margin available
07 — Inside the casing
ESP components: what wears, what fails, what upgrades
Every ESP is eight subsystems; performance is set by the weakest of them.
Collecting side
Collecting electrodes
Roll-formed C, sigma and ZT plate profiles, 6–15 m tall at 300/400 mm spacing. Alignment within ±5 mm is the difference between design voltage and constant sparking.
Collecting electrodesEmission side
Discharge electrodes
Rigid frame, spiral and barbed-wire emitters; corona current per metre, material selection, and why they break at the anti-sway frame.
Discharge electrodesCleaning
Rapping systems
MIGI rappers, rapper coils, tumbling hammers and vibrators — intensity in g at the plate, and sequencing that keeps rapping puffs off the opacity trace.
ESP rapping systemsPower
Transformer rectifier sets
45–110 kV secondary classes sized at 0.3–0.7 mA/m² of plate area; conventional versus high-frequency supplies and what lower ripple buys in mean kV.
Transformer rectifier setsControls
Voltage controllers
Automatic voltage control holds each field at the spark-over edge, manages back-corona detection and coordinates power-off rapping.
ESP controllersGas side
Gas distribution screens
Perforated screens at inlet and outlet flatten the velocity profile — poor distribution wastes collecting area the Deutsch equation assumes is working.
Gas distribution screensHV integrity
Insulators and conductors
Support and shaft insulators, heating and purge air, HV bus and through-bushings — the components whose failure takes a whole field down.
Insulators and conductorsMechanical
Drive systems
Rapping-shaft geared drives, hopper dischargers and access interlocks — the rotating hardware that keeps a static machine cleaning itself.
ESP drive systemsExisting ESP underperforming? Most units respond to internals realignment, rapping and TR upgrades before any casing extension is needed — see ESP upgrade and rebuild services and anonymised results under project references.
FAQ
Engineering questions, answered
What outlet dust concentration can an electrostatic precipitator achieve?
A retrofit-basis single ESP typically reaches 30–50 mg/Nm³ at 6 % O₂ dry; with adequate specific collecting area, 10 mg/Nm³ or below is achievable. In a two-stage train behind an Electrocyclone pre-collector, our bagasse design basis reaches 24 mg/Nm³. Guaranteed figures are stated per project after the technical assessment, on a stated basis.
What is specific collecting area (SCA) in ESP design?
SCA is collecting-plate area divided by gas volume flow, in s/m (m² per m³/s). It is the single strongest sizing lever: our bagasse design basis uses 75.7 s/m, giving 96.67 % collection across four fields. Higher SCA means more plate area per unit of gas, lower outlet dust, and a larger, costlier casing.
What is fly-ash resistivity and why does it matter for an ESP?
Resistivity measures how easily collected dust conducts charge to the plate. ESPs work well between roughly 10⁴ and 10¹¹ Ω·cm. Above 10¹¹ Ω·cm the dust layer breaks down electrically — back-corona — injecting positive ions that cancel particle charge and can halve collection. Below 10⁴ Ω·cm dust loses charge and re-entrains.
ESP or bag filter for a biomass boiler — which should I choose?
An ESP suits continuous duty to 200 °C with pressure drop around 10–30 mmWC (0.1–0.3 kPa) and low maintenance; outlet 30–50 mg/Nm³ typical, lower with more collecting area. A bag filter reaches below 10 mg/Nm³ regardless of ash resistivity but adds 1–2 kPa fan load and bag replacement cost. Permit limit, fuel ash and fan margin decide.
How much power does an electrostatic precipitator consume?
Corona power plus rapping and heating is modest: our bagasse design basis absorbs 177 kW for a four-field unit treating roughly 51 m³/s, and a mid-size unit is on the order of 250 kW. Gas-side pressure drop is low, typically 10–30 mmWC (0.1–0.3 kPa), so induced-draught fan penalty is far below that of a bag filter.
Why is an ESP quoted with one field out of service (n−1)?
Fields trip — a broken discharge electrode, a tracking insulator, a TR fault — and repairs wait for an outage. Quoting n−1 shows the stack stays compliant meanwhile: in our design basis, losing one of four fields moves outlet dust from 24 to 57 mg/Nm³, still under typical biomass permit limits of 80–120 mg/Nm³.
Send us your plant data
Fuel, boiler capacity, gas flow, current emission and the limit you must meet. An Arrow engineer replies with a technical assessment basis — not a brochure.