Deutsch-Anderson vs. Matts-Ohnfeldt: Two Ways to Predict the Same ESP
Both formulas take the same three inputs — migration velocity, plate area, gas flow — and give different answers for collection efficiency. Here's why, and why it matters which one a spec sheet used.
Every electrostatic precipitator textbook opens with the same equation. Deutsch, working from Anderson's experimental data in the 1920s, proposed that collection efficiency follows from just three numbers: how fast a charged particle drifts toward the plate (migration velocity, w), how much plate area is available (A), and how much gas flows through (Q):
η = 1 − exp(−wA/Q)
It's an elegant result, and it's still the formula most engineers reach for first. It's also, measurably, too optimistic — real precipitators collect less than Deutsch-Anderson predicts, sometimes by a wide margin on the same nameplate numbers.
The assumption that breaks
Deutsch-Anderson assumes every particle behaves identically — one migration velocity, applied uniformly to a dust cloud that's actually a wide range of sizes. Real fly ash isn't one size. It's a distribution, often spanning submicron fines to tens of microns, and migration velocity depends strongly on particle size (bigger particles hold more charge but also feel more drag; the two don't cancel evenly). A single average w papers over that spread — and because the efficiency equation is exponential, averaging the wrong thing produces a real bias, not just noise.
There's a second gap Deutsch-Anderson doesn't touch: it assumes the gas moves through the precipitator as a single, well-behaved plug. Real flow has some particles re-entrained off the plates, some gas sneaking past through non-uniform velocity profiles, and dust that isn't uniformly distributed across each field's inlet. None of that is in the three-variable equation.
What Matts and Öhnfeldt changed
Working from measured performance on operating precipitators rather than the idealized derivation, Matts and Öhnfeldt (Sweden, early 1960s) proposed keeping the same three inputs but changing the exponent's shape:
η = 1 − exp(−(wA/Q)k)
with k typically fitted around 0.4–0.6 (this site's ESP module uses k=0.5, giving the square-root form 1−exp(−√(wA/Q))). Pushing the exponent below 1 does exactly what the physics calls for: it curbs the runaway optimism of the pure exponential at large wA/Q, without needing to explicitly model a particle-size distribution or a non-ideal flow field. It's a correction fitted to real plant data, not a first-principles rebuild — which is exactly why it works better in practice while staying just as simple to compute.
Why this matters when reading a spec
The two formulas can disagree by several percentage points of efficiency for the same w, A and Q — which, on a 99%-plus target, is the difference between comfortably meeting an emission limit and quietly missing it. If a vendor guarantee or an old design worksheet doesn't say which formula it used, treat the efficiency number as formula-dependent, not as a fixed physical fact. This site's own ESP module uses the Matts-Ohnfeldt form specifically because it tracks real precipitator performance more closely than the textbook-standard Deutsch-Anderson result.
Further reading
H. J. White, Industrial Electrostatic Precipitation, Addison-Wesley, 1963 — the standard English-language reference for the Deutsch-Anderson derivation and its assumptions. S. Matts and P.-O. Öhnfeldt, "Efficient gas cleaning with SF electrostatic precipitators," Flakt Report, 1963/64 — the original modified-exponent correlation.