Plugholing

Plugholing is the withdrawal of air from below a smoke layer through an opening intended to extract smoke. It occurs when the extraction at one location draws the lower layer into the outlet. Some capacity then removes relatively clean air, reducing the useful smoke removal. The phenomenon can affect natural and mechanical smoke and heat exhaust ventilation. A thin hot gas layer is particularly susceptible because the lower air lies closer to the extraction point.

Plugholing links the extraction rate at each point to the depth and buoyancy of the smoke layer. Increasing total fan capacity does not necessarily give the same increase in smoke removal. Distributing extraction between several points can reduce the local demand on the layer. The onset criterion and the amount of lower-layer air extracted are separate modelling questions. Different sources provide different levels of description, from a limiting flow criterion to a model-specific mixing correlation.

How lower-layer air reaches an exhaust

A smoke exhaust draws gas towards its opening from the surrounding layer. The pressure and velocity field near that opening deforms the interface between the hot and cool layers. As extraction becomes stronger relative to the layer's depth and buoyancy, the deformation becomes larger. Eventually lower-layer air can reach the opening and leave together with smoke.

When local extraction reaches through a smoke layer, some exhaust capacity draws air from below. Based on Van Rossum (2027), §6.8, and Cooper (2002), p. 3-228.

This behaviour explains why plugholing is a local problem. Two systems with the same total extraction can place different demands on each individual opening. Concentrating the extraction at one point can draw air through the interface sooner than distributing it between several points. The relevant comparison therefore includes the flow assigned to each extraction location.

The smoke reservoir supplies the layer from which an exhaust draws. A shallow layer gives little separation between the outlet and the lower air. A deeper layer can sustain a greater local extraction before lower air is drawn through it. This relationship concerns the physical depth beneath the extract point, rather than the total height of the building alone.

Van Rossum uses the draining bathtub as an everyday analogy. As the remaining water becomes shallow, air can enter the outlet while water still remains. The smoke-control analogy concerns withdrawal across an interface between fluids. It does not mean that the geometry or flow rate of a bathtub supplies a quantitative fire-engineering model.

Why nominal extraction can overstate smoke removal

The gas leaving an exhaust can contain contributions from both layers. Before significant plugholing, an idealised upper-layer extraction removes only upper-layer gas. After lower-layer air enters, part of the same total flow no longer removes smoke from that layer. Treating all exhausted volume as smoke would therefore overstate the effective extraction.

Milke relates this loss of effectiveness to an increase in the resulting smoke-layer depth. The plume continues to supply smoke while the exhaust removes less smoke than assumed. The layer can then deepen until a different balance develops. A design calculation that fixes the layer height without accounting for the extraction composition misses this interaction.

The interpretation also depends on whether the quoted rate is a mass flow or a volume flow. The CFAST technical reference weights contributions from the hot and cool layers using their separate densities. Those densities affect the mass carried by the same volume. Its treatment therefore tracks both mass and heat transferred through the opening.

An apparently satisfactory total fan rate does not resolve these questions by itself. The calculation also needs the extraction locations, the layer conditions and the way each opening interacts with the interface. Where the model assumes extraction entirely from the upper layer, that assumption requires examination. The onset calculation provides one way to identify when it may fail.

A dimensionless onset criterion

Cooper describes ceiling extraction using a Froude number, which compares the imposed flow with the stabilising effect of buoyancy. The numerator is the volume flow at the extract point. The denominator combines gravity, the density difference between the layers and the smoke-layer depth. His definition is reproduced below with the volume flow written as V̇.

F = V̇ / {√[g(ρ_a − ρ_s)/ρ_a] h^5/2}

h = z_ceiling − z_interface

Symbol	Description	Unit
F	Froude number in Cooper's ceiling-extraction formulation	Dimensionless
V̇	Volume flow through one extraction opening	m³/s
g	Gravitational acceleration	m/s²
ρa	Density of the lower ambient layer	kg/m³
ρs	Density of the upper smoke layer	kg/m³
h	Smoke-layer depth beneath the ceiling	m
zceiling, zinterface	Ceiling and layer-interface elevations from a common datum	m

For circular vents, Cooper gives a critical value of 1.6. At this limiting condition, the interface reaches the ceiling at the extraction point. Above the critical value, the extracted flow includes some gas from the lower layer. The statement applies to the formulation and configurations discussed in that source.

The exponent on depth makes layer thickness especially influential. At fixed density difference and extraction rate, halving the depth increases the calculated Froude number by approximately 5.7. This follows directly from the power of five halves in the denominator. It explains why a modest-looking change in layer depth can alter the onset assessment substantially.

Cooper considers extending the circular-opening result to square and moderate-aspect-ratio rectangular vents to be reasonable. That statement does not establish an unrestricted rule for every opening shape or surrounding obstruction. The source also distinguishes onset from the detailed interface deformation below the threshold. Its limiting criterion does not supply a complete description of all local flow patterns.

Worked comparison of two layer depths

Consider an illustrative extract point removing 20 m³/s. This rate appears among the mechanical-extraction cases discussed by Milke. Take smoke and ambient temperatures of 393 K and 293 K, respectively, from Cooper's venting example. The comparison keeps temperature and extraction fixed while changing the assumed layer depth.

Cooper's example uses an ambient density of 1.2 kg/m³. With the same gas approximation, the smoke density is about 0.895 kg/m³. Taking gravity as 9.8 m/s² and a layer depth of 3.0 m gives F approximately 0.81. This lies below the stated critical value of 1.6 in Cooper's formulation.

Now reduce the assumed depth to 1.5 m while retaining the other inputs. The calculated Froude number becomes approximately 4.6, exceeding that criterion. The comparison isolates the depth effect; it is not a prediction that a real reservoir retains unchanged temperature as its depth changes. A complete smoke calculation would also update the layer's mass and thermal conditions.

The calculated onset parameter rises sharply as the layer becomes shallower at fixed extraction and density. Based on Cooper (2002), p. 3-228.

The same equation gives a limiting flow of approximately 39 m³/s for the 3.0 m layer under these assumptions. This is a calculated threshold for the stated model, rather than a selected fan duty. It does not include uncertainty in layer conditions, inlet disturbances or the applicability of the opening representation. Those influences remain separate parts of the smoke-control analysis.

If the original 20 m³/s is divided equally between two otherwise equivalent points, each point carries 10 m³/s. With the 3.0 m depth unchanged, its calculated Froude number falls to approximately 0.41. This arithmetic illustrates the benefit of reducing local extraction demand. It does not establish a spacing rule or prove that interacting outlets behave independently.

Measures that change the local flow

The principal response described by both sources is to provide more extraction points with less flow at each. This changes the local demand without necessarily changing the required total exhaust. The distinction matters because total smoke production still has to be addressed. Redistributing an inadequate total capacity does not itself supply the missing capacity.

Increasing the available layer depth is another measure identified by Van Rossum. The onset equation shows why this reduces the dimensionless extraction demand. However, the resulting smoke interface must remain compatible with the intended smoke-control objective. A thicker reservoir and a higher clear layer beneath it are competing geometrical quantities in a space of fixed height.

Van Rossum also describes changing the direction from which an outlet draws. Orientation changes or flow-deflecting construction can encourage withdrawal from the layer horizontally. The purpose is to reduce direct draw from the clean air immediately below. The supplied description establishes the principle but does not give universal dimensions or performance coefficients for such arrangements.

Replacement air affects the smoke layer through a different part of the flow system. Milke places makeup air below the smoke interface and describes avoiding strong velocities near the plume. Air introduced into the layer adds material that must subsequently be exhausted. Air near the interface can also increase mixing between the layers.

These effects can coexist with plugholing but are not identical mechanisms. A supply jet can disturb the plume and increase entrainment before gas reaches the reservoir. Plugholing concerns gas drawn from below the layer at an extraction point. Distinguishing them helps identify whether the difficulty originates at the inlet, the plume or the exhaust.

Representation in smoke models

Milke's 2002 chapter describes a limitation of elementary methods available in that discussion. They assess whether plugholing occurs, without determining the resulting smoke depth after its onset. The source therefore separates the threshold problem from prediction of the smoke-removal loss. This is a statement about those methods, rather than a permanent absence of subsequent modelling developments.

The supplied 2026 CFAST technical reference includes a specific treatment for ceiling and floor vents. It divides the gas removed from a compartment between contributions from the upper and lower layers. A weighting parameter controls those contributions to the mass and heat flows. A deep upper layer tends towards extraction dominated by that layer.

Its weighting correlation was developed using Fire Dynamics Simulator, a computational fluid dynamics model. The correlation decreases the upper-layer contribution as its defined Froude number increases. This is a particular model representation of the effect, supported in the guide by comparison with those simulation results. It does not make every possible extraction arrangement equivalent to the configurations represented by that correlation.

The CFAST guide uses a temperature-based Froude definition that differs from Cooper's density-based expression above. Coefficients and threshold values therefore belong to their respective definitions. Transferring a number from one formulation into the other without checking the normalisation changes the meaning of the calculation. The printed equation and its variable definitions need to remain together.

The same guide describes mechanical ventilation separately using user-specified volume flows. It states that the model does not include ductwork or fan curves in that treatment. The presence of a plugholing correlation in one vent formulation therefore does not establish a complete representation of an installed exhaust system. The relevant model component and the specified boundary conditions determine what has actually been calculated.

Estimating the extracted layer mixture

The CFAST ceiling-vent correlation answers a different question from the onset check. It assigns a continuously varying weighting factor to the upper-layer contribution. The guide expresses this factor using an exponential function of its Froude number. Rewriting its layer depth as h, the two equations are as follows.

Fr = V̇ / √[gh⁵(T_s − T_a)/T_a]

α = exp[−(Fr/2)²]

Symbol	Description	Unit
Fr	Froude number defined for the cited CFAST correlation	Dimensionless
Ts, Ta	Upper-layer and ambient absolute temperatures	K
α	Weighting factor for the upper-layer contribution in the vent equations	Dimensionless

The guide uses the weighting factor in separate terms for the two layer densities. This retains the distinction between a proportion of the withdrawn volume and the resulting mass contribution. In the notation used here, the upward mass flow is represented by the following expression. The corresponding energy expression also includes the temperature of each contributing layer.

ṁ_up = α ρ_sV̇ + (1 − α)ρ_aV̇

A separate ceiling-vent illustration can use the same temperatures and 3.0 m layer depth as the earlier calculation. With 20 m³/s, the CFAST definition gives Fr approximately 0.70 and α approximately 0.88. These values result from the specified correlation. They do not constitute a second circular-vent onset threshold.

Keeping the unrounded values, the upper-layer term contributes approximately 16 kg/s. The lower-layer term contributes approximately 2.8 kg/s. Their sum is approximately 19 kg/s under the illustrative assumptions. These calculated contributions demonstrate why the density weighting remains necessary when the model converts extracted volume into mass. The energy transferred through the vent also changes when cooler lower-layer gas replaces hotter smoke. The model therefore carries each contribution into its heat balance, rather than assigning one smoke temperature to the whole flow.

A layer-weighted vent calculation retains the different mass contributions of hot smoke and cooler air. Based on Peacock et al. (2026), pp. 21–22.

The two approaches do not give identical descriptions below Cooper's critical value. His limiting analysis identifies a threshold for lower-layer withdrawal in the stated configuration. The CFAST correlation instead fits a continuous variation in the contribution from each layer. A comparison therefore needs to consider both the different dimensionless definitions and the different modelling purposes.

This distinction also affects how results are reported. A threshold check states whether the assumed extraction remains within a particular limiting condition. A mixture calculation states how a specified model apportions the extracted flow. Neither result alone supplies the full time history of reservoir depth, plume supply and exhaust-system operation. Those quantities require the surrounding compartment and ventilation calculation.

References

• Cooper, L. Y. (2002). Smoke and heat venting. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 3-219–3-242). National Fire Protection Association.
• Milke, J. A. (2002). Smoke management in covered malls and atria. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 4-292–4-310). National Fire Protection Association.
• Peacock, R. D., McGrattan, K. B., Forney, G. P., & Reneke, P. A. (2026). CFAST: Consolidated fire and smoke transport (version 7). Volume 1: Technical reference guide (NIST Technical Note 1889v1; revision 7.7.7, May 27). National Institute of Standards and Technology.
• Van Rossum, E. (2027). Introduction to Fire Safety Engineering. Fire Safety Academy.