EBCT and Bed Sizing: Empty Bed Contact Time for Carbon Beds

The US EPA Air Pollution Control Cost Manual sizes the working capacity of a carbon bed at roughly 50 % of its equilibrium capacity — and then says to lower that figure further where short desorption cycles, very high vapour pressure constituents, high moisture contents or significant amounts of impurities are involved. That is the regulator conceding the most common design error in print. A bed sized on an isotherm capacity is sized on a number the bed will never deliver.

Everything below is about the four things that decide what it will deliver: contact time, velocity, the mass transfer zone, and pressure drop. The first of those is empty bed contact time, EBCT — the empty bed volume of carbon divided by the volumetric flow rate through it, quoted in seconds for gas-phase duty and in minutes for liquid-phase.

The short version

• EBCT = empty bed volume ÷ volumetric flow rate. In US units, EBCT (min) = 5.87 × D² × H / gpm. Gas-phase EBCT is conventionally quoted in seconds, liquid-phase in minutes.
• Superficial velocity = Q ÷ cross-sectional area — the velocity the fluid would have in an empty vessel. EPA cites vendor figures of 85, 60 and 65 ft/min and uses 75 ft/min in its worked example.
• Bed depth must exceed the mass transfer zone length. The only published MTZ figure we could verify is 15–30 cm (6–12 in) for sulfur-impregnated carbon in mercury service, at roughly 0.5–1 s residence time.
• Published EBCT by duty: odour control 3 s below 100 ppm H₂S, 10 s above 100 ppm, 20 s above 500 ppm, overall range 1–30 s at a typical 3-foot bed depth.
• Two gaps worth stating: EPA publishes no VOC EBCT at all — it sizes on cycle time, at an optimum regeneration frequency of once every 8–12 hours — and no authoritative published EBCT guidance exists for siloxane duty.

What is EBCT, and how do you calculate it?

Empty bed contact time is the volume of the empty bed divided by the volumetric flow rate through it. It is “empty bed” because the calculation ignores the void fraction: it is the time the fluid would take to traverse the vessel if there were no media in it. That makes it a geometry-and-flow number rather than a physical residence time, which is exactly why it is useful — everyone computes it the same way from the same three inputs.

Two equations set the bed. EBCT fixes how long the gas is in the vessel; superficial velocity fixes how hard it is pushed through, and therefore the pressure drop.

The AWWA-referenced form:

EBCT (min) = Carbon Volume (ft³) / Flow Rate (ft³/min)

The practical US-units form for a cylindrical vessel:

EBCT (min) = 5.87 × D² × H / gpm

where D is bed diameter in feet and H is bed depth in feet. The 5.87 rolls up π/4 and the conversion from cubic feet to gallons.

The metric form:

Contact time (min) = Carbon volume (m³) × 60 / Flow rate (m³/h)

On units. In practice gas-phase EBCT is quoted in seconds and liquid-phase in minutes. Gas-phase design sources deal in fractions of a second to tens of seconds — Jacobi’s 0.2–3 s, odour-control practice at 1–30 s, European practice at 2–7 s — while water-treatment sources deal in tens of minutes, with New York State’s drinking-water recommendations defining GAC10 and GAC20 as 10- and 20-minute empty-bed contact times on average daily flow. No single source states that convention as a rule. It is supported by the aggregate of design sources, and we present it that way.

What is superficial velocity, and what values are published?

Superficial velocity is the volumetric flow rate divided by the cross-sectional area of the bed:

v = Q / A

It is the velocity the fluid would have if the vessel were empty. Actual interstitial velocity is higher by roughly 1/ε, where ε is the void fraction, but every correlation in this field — the Ergun equation included — is written in terms of the superficial value. Do not mix the two.

Published gas-phase ranges, with attribution:

Liquid phase: 2–10 gpm/ft² against AWWA at a 10-minute EBCT, or 5–20 m/h in metric practice.

Jacobi’s minimum-velocity recommendation is the one most often missed. Velocity is normally treated as a ceiling — too fast and you lose contact time and gain pressure drop. But a bed running too slowly on a stream with exothermic chemistry does not shed heat, and hot spots are a real failure mode.

The mass transfer zone

The mass transfer zone is the band of the bed where adsorption is actually happening at a given moment. Upstream of it the carbon is saturated; downstream of it the carbon is still virgin and the fluid has already been cleaned.

The mass transfer zone travels through the bed as it loads. Breakthrough occurs when its leading edge moves beyond the end of the bed.

The MTZ travels from inlet to outlet as the bed loads, and breakthrough occurs when the leading edge of the MTZ moves beyond the end of the bed. From that definition follows the design rule stated explicitly in AWWA-referenced pilot design guidance: bed depth must exceed the MTZ length to prevent early breakthrough.

The one published MTZ length we could verify is from Idaho National Laboratory’s work on sulfur-impregnated carbon in mercury service: about 15–30 cm (6–12 inches), corresponding to roughly 0.5–1 second residence time. That belongs to one adsorbate on one carbon at one set of conditions. MTZ length is not a property of carbon; it is a property of the system.

One honest caveat. The commonly repeated statement that a bed shallower than its MTZ leaks from t = 0 follows directly from the definition, but we could not source that explicit claim to a specific document. We are stating the inference, not citing it.

What governs MTZ length is kinetics. Knaebel’s adsorber design notes put it plainly: fast kinetics produce a sharp breakthrough curve, slow kinetics a distended one, and the compensation is to add adsorbent at the product end or increase the cycle time. He also warns that the length-of-unused-bed method — the standard shortcut for turning a lab column into a full-scale bed — is unreliable for scale-up and when conditions vary widely. If your full-scale conditions differ materially from the column you tested, LUB will mislead you.

Pressure drop

Pressure drop through a packed bed is described by the Ergun equation:

ΔP/L = [150·μ·(1−ε)²·v] / [ε³·d_p²·φ²]  +  [1.75·ρ·(1−ε)·v²] / [ε³·d_p·φ]

• ΔP — pressure drop, Pa · L — bed length, m
• μ — dynamic viscosity, Pa·s · ρ — fluid density, kg/m³
• ε — void fraction, dimensionless · φ — sphericity, dimensionless
• v — superficial velocity, m/s · d_p — mean particle diameter, m

The 150 constant belongs to the viscous, Blake–Kozeny term, dominant below about Re 10; the 1.75 to the inertial, Burke–Plummer term, dominant above about Re 1000. The design consequence is in the exponents: pressure drop scales as 1/d_p² and 1/φ² in the viscous term, 1/d_p and 1/φ in the inertial. Bigger and rounder particles cost less — which is why odour-control practice budgets an extruded pellet bed at about 1 in. w.c. per foot against about 2 in. w.c. per foot for granular carbon, within a typical total design allowance of 6 in. w.c.

For a fitted gas-phase estimate, EPA publishes an empirical correlation for Calgon PCB 4×10 mesh carbon:

ΔP_b / t_b = 0.03679·v_b + 1.107×10⁻⁴·v_b²

with ΔP_b/t_b in inches of water per foot of carbon bed and v_b in ft/min. It is fitted to one mesh cut of one named product. The printed form is also ambiguous across PDF renderings of the EPA chapter — three renderings disagree cosmetically — so verify it against a clean print of EPA Equation 1.30 before designing to it.

Knaebel adds the caveat that keeps this in proportion: pipes, valves and fittings pose as much flow restriction as the bed does. A bed budgeted at 3 in. w.c. inside a system that loses 10 across ductwork and dampers is not the problem.

Published EBCT design values, by duty

The last two rows are the useful ones.

EPA does not publish a VOC EBCT. Its carbon adsorber methodology sizes on cycle time instead, stating that the optimum regeneration frequency for fixed-bed adsorbers treating streams with moderate to high VOC inlet loadings is once every 8 to 12 hours. Working capacity, loading rate and cycle length set the carbon mass; contact time falls out of the geometry rather than driving it. The nearest published proxy is Jacobi’s 2–3 s thick-bed figure, and it is a proxy, not a specification.

No authoritative published EBCT guidance exists for siloxane duty. We looked, and did not find one. Stating that plainly is more useful to a biogas developer than filling the row with a plausible number. Siloxane duty should be sized on a pilot column or a media trial on the actual gas, because siloxane loading varies enormously between digesters and landfills.

Isotherms, and why they mislead

An isotherm is an equilibrium relationship between fluid-phase concentration and adsorbed loading at one fixed temperature. Two forms dominate. Writing n* for equilibrium loading and C for fluid-phase concentration:

Langmuir:    n* = (A·C) / (1 + B·C)
Freundlich:  n* = A·C^(1/B)

with Henry’s law, n* = A·C, as the linear limit at low concentration. EPA writes the gas-phase Freundlich form in terms of partial pressure:

w_e = k · P^m

where w_e is equilibrium adsorptivity in pounds of adsorbate per pound of adsorbent and P is partial pressure in psia. EPA’s worked example is toluene at 77 °F, k = 0.551, m = 0.110, valid over 0.001–0.05 psia. Note the three qualifiers on one line: a named adsorbate, a fixed temperature, a validity range spanning a factor of fifty. Extrapolating a Freundlich fit outside its stated range is not conservative, it is arbitrary.

The deeper problem is what “equilibrium” means in a moving bed. A bed only delivers equilibrium capacity in the limit of zero mass transfer resistance — Knaebel’s point that a realistic breakthrough curve requires knowledge of mass transfer resistances, and that only when those resistances are very small does the front approach a step change. Real beds have real resistances, so the front is spread, and the carbon leaving service is not saturated.

That is why EPA’s 50 % rule exists. Working capacity is the difference, per unit mass of carbon, between the adsorbate on the carbon at the end of the adsorption cycle and the amount remaining at the end of the desorption cycle. EPA estimates it at roughly half the equilibrium capacity at the adsorber inlet — and says to go lower for short desorption cycles, very high vapour pressure constituents, high moisture or significant impurities.

The liquid-phase literature makes the same point about laboratory testing generally: laboratory test methods do not accurately predict the performance of full-scale GAC adsorbers. Only a pilot column field evaluation or an on-site plant test gives an accurate answer. That is not a reason to skip isotherm work — it is a reason to treat it as screening.

A practical sizing checklist

Before anyone quotes you a bed size, these should be on the table:

1. Inlet concentration and its variability. A design point is not a range; swings on a digester or a headworks can be an order of magnitude.
2. Required outlet, with the endpoint definition attached.
3. Flow, and its turndown. Velocity at minimum flow sets hot-spot and channelling risk; velocity at maximum flow sets pressure drop.
4. Gas temperature and humidity. Jacobi recommends below 40 °C for general filter duty, and water competes for micropores.
5. Whether free oxygen is present. It decides which H₂S mechanism is even available.
6. Acceptable pressure drop, against the whole system rather than the bed alone.
7. Change-out access — vessel geometry, manway position, disposal route, and whether the loading is reversible or a one-way chemisorption.

A supplier who quotes a bed size without asking those questions has not sized anything. They have priced a volume of carbon.

Which Suracsh grades suit which geometry

Nothing here changes the arithmetic above, but the media choice interacts with it.

For deep beds at high flow, the pellet grades are the lower-pressure-drop option. The SURSORB PAC family is supplied at 6, 4 and 3 mm — except PAC-1250, which is published in 4 mm only — and PAG, P-ALKA, MAG and PAM at 4, 3 and 2 mm — roughly half the pressure drop per foot of a granular bed, on the published rule of thumb. For shallow beds and cartridges, the granular grades run from 8×30 down to 14×35 mesh cuts, with finer cuts again on the respiratory grades.

And you will need the apparent density figures to convert. Design work is volumetric — EBCT, velocity and bed depth are all geometry — while purchasing and change-out logistics are by mass. Every grade publishes an apparent or bulk density: 480 ± 30, 450 ± 30 and 380 ± 30 g/L across PAC-900, PAC-1050 and PAC-1250, 520 ± 30 kg/m³ on PAM, 420 ± 30 kg/m³ on PAG. Those turn a bed volume into a tonnage, and a tonnage into a change-out interval.

The full specification for every grade named here is published in HTML at suracsh.us, and every datasheet downloads without a form or an NDA. The contaminant-to-grade mapping is at /selection-guide/.

Sources

• ITRC — Empty bed contact time (EBCT), definition
• PNWS-AWWA — Basic principles of water treatment media pilot trial design and evaluation
• P2 InfoHouse — GAC design reference (EBCT formula, mass transfer zone, lab vs full scale)
• Activated carbon columns plant design (metric contact time and linear velocity)
• NYSDOH — Interim recommendations for granular activated carbon installations
• US EPA Air Pollution Control Cost Manual — Carbon Adsorbers, 7th edition
• US EPA — Carbon Adsorbers chapter, pressure drop correlation (Eq. 1.30)
• Jacobi Carbons — Design and selection of air/gas treatment with industrial filters
• PDH Academy course 437 — Activated Carbon Odor Control Systems
• Suez Water Handbook — Odour control
• Knaebel — Adsorber design notes (isotherms, breakthrough, LUB, Ergun)
• Ergun equation, form and constants
• Idaho National Laboratory — Mass transfer zone depth in sulfur-impregnated carbon beds

Questions