Containment exhaust system design: HEPA housings, bag-in bag-out, dampers, penetrations, and stack discharge

What this covers

• How confinement zones set the pressure and airflow direction of a containment facility.
• How a cleanroom can act as the secondary confinement around a hot cell and still hold its class.
• What a bag-in bag-out housing, its dampers, and its test sections each protect against.
• Principles for ductwork, penetrations, fire protection, fans, and stack discharge.

What changes the result

• The hazard the enclosure handles, which sets the number of testable HEPA stages through safety analysis.
• The breach the exhaust must still hold inflow against, not only normal operation.
• Fire, moisture, and particle loading, which decide the protection placed ahead of the HEPA filters.
• Building geometry, wind, and intake positions, which decide where and how high the stack discharges.

Containment is a direction of airflow that holds under upset conditions

The DOE handbook for air cleaning systems in nuclear facilities describes containment as zones. The primary confinement zone is the inside of the hot cell, glovebox, hood, or exhaust duct where high contamination is expected. The secondary zone is the room and ventilation around it, where contamination could appear through a breach of the primary barrier. The tertiary zone is the rest of the building. Air flows from the tertiary zone toward the primary zone, and the handbook asks for pressure differentials sufficient that no backflow occurs, even under upset conditions.

Two consequences follow for exhaust design. The inside of an exhaust duct carries the hazard classification of the zone it serves, so duct from a primary enclosure is treated as primary confinement all the way to its filters. And recirculation is allowed within a zone through high-efficiency air cleaning, but air is not recirculated from a more contaminated zone to a less contaminated one.

The same logic, at lower hazard, sits behind the negative rooms in pharmacy and health care: the USP 800 ventilation requirements for hazardous drugs and the negative pressure room exhaust offset for isolation rooms are containment zones with smaller consequences and simpler filter trains.

A cleanroom as the secondary zone around a hot cell

Some processes need both containment and particle control, radiopharmaceutical production in shielded hot cells among them. The cleanroom around the cell is then the secondary confinement zone, and it has to be negative to the corridor for containment while it holds an ISO class for the product.

Negative pressure does not prevent a room from being clean. The ISO class is held by HEPA-filtered supply air and enough airflow to dilute what the room releases; positive pressure only keeps unfiltered air from leaking in. A negative cleanroom replaces that protection with a tight envelope and an airlock, so the air it draws in arrives through a controlled space rather than straight from a corridor.

The airlock can be a bubble, positive to both the corridor and the cleanroom, which pushes clean air into the cleanroom and out to the corridor, or a sink, negative to both, which draws from both sides into its own exhaust. A bubble airlock keeps the cleanroom's inward leakage clean. A sink airlock keeps each side's air from reaching the other, and the airlock's exhaust then carries whatever the secondary zone can release, so it joins the filtered exhaust system. The pressure cascade calculation sizes the airflow each arrangement needs.

Sizing exhaust for a breach, not only for normal operation

The DOE handbook asks that primary confinement exhaust be sized to ensure an adequate inflow of air if the confinement is breached, and gives 100 feet per minute as a rule of thumb for capture velocity across a breach between primary and secondary zones. That turns breach scenarios into airflow numbers the exhaust fan must have in reserve.

A glovebox glove lost from an 8 inch port opens π × (4/12)² = 0.35 square feet, and 100 feet per minute across it is 35 cfm. A 2 by 2 foot transfer door left open on a hot cell is 4 square feet, and holding 100 feet per minute across it takes 400 cfm. A 3 by 7 foot room door at the same velocity would take 2,100 cfm, which is why room-scale openings are handled with airlocks rather than exhaust reserve.

The control response matters as much as the reserve. When a breach opens, the enclosure pressure rises toward the room's, and the exhaust control has to open toward its breach airflow rather than reducing flow to restore the old pressure setpoint. Write that response into the failure sequences for the facility.

HEPA housings and bag-in bag-out filter changes

The DOE handbook describes the bag-in bag-out side-access housing as a standard commercially manufactured configuration, and lists what to check when buying one: the filter mounting frame and clamping device, rigidity of the box and its cover, the method of sealing and clamping the cover, access to the installed filter, rigidity of the duct connections, and the materials of every part including the clamping mechanism. It recommends provisions for in-place testing on all filter housings.

For the sealing surface the filter gasket seats against, the handbook recommends a minimum sheet metal thickness of 0.078 inch, No. 14 U.S. gauge, seal-welded so the surface does not warp. A filter that seals perfectly on a flat test fixture leaks on a sealing surface that is not flat.

A bag-in bag-out change removes the loaded filter into a plastic bag sealed to the housing's access port, so the filter and the housing interior are never open to the room. The handbook calls for shutoff dampers upstream and downstream of the filter to isolate the housing during the change and to stop the bag ballooning or collapsing when the access door opens against a pressure difference, and notes that a slight negative pressure in the housing, 0.25 to 0.5 inches of water gauge, helps ensure any leakage is inward.

High-hazard exhaust can use two or more housings in parallel serving the same area and discharging to the same stack. With inlet and outlet isolation dampers on each, one housing can be held in standby, or shut down for maintenance, testing, and emergencies while the other carries the exhaust, which the redundancy design page treats alongside fan redundancy.

In-place testing of installed HEPA stages

A HEPA filter's factory efficiency test says nothing about the gasket, the frame, or the housing it was installed in. The in-place leak test does, and the DOE handbook uses it both for acceptance after installation and for periodic surveillance, with a polydisperse test aerosol injected upstream and sampled downstream of the installed bank.

The procedure the handbook points to is in ASME AG-1, Appendix TA, and it recommends an acceptance criterion of 0.05 percent maximum leakage for the in-place system test in systems designed to its guidance. Designing for that test means an injection point far enough upstream for the aerosol to mix across the duct, an upstream sample point, a downstream sample point after mixing, and a separate set for each stage so each HEPA stage is proven on its own rather than as a pair.

ASME N510 covers the in-service testing program for nuclear air treatment systems: a set of tests the user selects, performed in a stated sequence, at recommended minimum frequencies. It leaves most acceptance criteria to the system's design basis, so the design documents have to state them. The handbook also recommends that filters exposed to smoke from a fire be tested in place within 24 hours and replaced if they fail.

Dampers, ductwork, and penetrations

Isolation dampers on containment housings are available in bubble-tight and low-leakage designs, and the DOE handbook notes they are designed, manufactured, and tested in the same manner as the housings. Choose bubble-tight dampers where the closed damper forms part of the containment boundary during filter change or maintenance, and state the allowable leakage in the specification rather than a damper type alone.

Keep every section of exhaust duct inside occupied space below the pressure around it, by placing the fan at the discharge end, so a leak draws room air inward. ASME AG-1 covers ductwork among the components of nuclear air treatment systems, and in lower-hazard containment the same principle calls for welded or sealed seams on the negative side and no duct joints inside the fan discharge within the building.

Every pipe, cable, and duct that crosses the secondary confinement boundary is a potential leak path. The handbook calls for positive seals at penetrations of the secondary barrier, with air locks or a clothing-change facility at the entrance. Seal penetrations with materials that tolerate the room's cleaning agents and any radiation or chemical exposure, and include them in the envelope leakage the pressure design assumes.

Fire protection of the filter train

HEPA filters collect whatever a fire releases and can be damaged by heat, burning embers, and water. The DOE handbook devotes a chapter to fire protection of air cleaning systems, and describes spark arresters made of coarse glass fibers as reasonable protection at low cost, alongside demisters that keep water droplets off HEPA media where a moist exhaust stream or fire suppression water can reach it.

The fire strategy also decides what the exhaust does during a fire, which is a containment question as much as a fire question. The failure sequences guide covers how that decision is written into the fire alarm response with the authority having jurisdiction.

Fans, backup power, and discharge

Containment exhaust is the airflow whose loss releases material, so it gets redundancy first: a standby fan with automatic changeover, or an array that meets design airflow with one fan out, each fan with its own isolation damper. The DOE handbook calls for backup power for fans, dampers, valves, controls, and humidity-control heaters where the facility's design and safety documentation require it, estimated for both off-normal and accident conditions.

Select the fan at the loaded filter condition for every HEPA stage in series, as the AHU external static pressure calculation shows for a supply unit, and confirm that the pressure reserve for breach airflow is available at that condition too.

Stack discharge: location first, then velocity

The DOE handbook is direct about stacks: high stack velocity is a poor substitute for proper stack location. A stack flush to a roof needs an exit velocity four times the wind velocity to throw effluent out of the roof's recirculation cavity. Stacks should be circular, may use a nozzle at the tip to raise exit velocity, and should not carry caps that deflect effluent downward; drainage provisions keep rain out instead.

For velocity, the handbook recommends at least 1.5 times the wind velocity to minimize downwash, and a minimum exit velocity of 3,000 feet per minute to prevent downwash in winds up to 22 miles per hour, keep rain out, and keep condensation from draining down the stack. The 2012 edition of ANSI/AIHA Z9.5 set laboratory exhaust stacks at least 10 feet above adjacent roof lines and air intakes, discharging vertically at 3,000 feet per minute or more; confirm against the current ANSI/ASSP Z9.5-2022 text.

Stack height and whether the release is elevated, ground level, or mixed mode come from a dispersion analysis against the facility's permit and exposure limits. Intakes are sized separately, with the handbook noting air velocity through louver free area kept below the speed at which water droplets are drawn in, which it puts at less than about 500 feet per minute, and a stack is placed so it cannot reach them.

Questions people ask about this