Cleanroom air handler

Air changes and pressure steps keep the room clean. Filters and fans decide what it costs.

A cleanroom air handler runs all day, every day, and it is often the biggest load in the plant. Clean air comes from air changes and pressure steps from room to room. The twin checks each filter, the fan and every door step, and shows what fewer air changes in empty rooms would save.

20,000 m³/h air handler, drive-fed plug fan, terminal HEPA, Grade C suite, 10–15 Pa steps

Built and run to EU GMP Annex 1, ISO 14644-1

Start with one utility See how we prove results

Air changes in use: 25 per hour
Fan input power: 10.4 kW
Step Grade C to D: 14.2 Pa
Clean-up 100:1: 11 min

Anatomy and nameplate

1Room pressure differentials. One sensor across each door. The step must stay above 10 Pa.
2Supply airflow. Flow station in the supply duct. Sets the air changes.
3Filter pressure drop. Pre-filter, final filter and terminal HEPA. Each loads on its own curve.
4Fan speed and power. From the drive. Power against flow and pressure gives the fan’s efficiency.
5Coil temperatures. Air on and off the cooling and heating coils. A fouled coil needs colder water for the same air.
6Humidity. Room humidity against its range. Coils and humidifier hold it.
7Damper positions. Outside air and return air. A damper that drifts shifts flow and pressure.

Cleanroom air handler nameplate
Air handler
Supply airflow	20,000 m³/h (5.56 m³/s)
Fan	Plug fan, 70% total efficiency, 15 kW motor on a drive
Fan input power, filters at mid-life	10.4 kW at 1,178 Pa total
System pressure, filters clean	1,000 Pa
Coils and humidity	Chilled water cooling coil, heating coil, steam humidifier
Ducts, dampers and diffusers	470 Pa at design flow
Filters
Pre-filter (ISO ePM10)	60 Pa clean, change at 150 Pa, about 4 months
Final filter (ISO ePM1 80%)	110 Pa clean, change at 250 Pa, about 12 months
Terminal HEPA (H14)	180 Pa clean, change at 360 Pa or on a failed leak test
Suite
Rooms	Grade D airlock and Grade C process room, off an unclassified corridor
Volume	800 m³
Air changes	25 per hour in use
Clean-up 100:1	11 minutes at 25 air changes, well mixed
Pressure steps	13.7 Pa airlock to corridor, 14.2 Pa Grade C to airlock
Offset air	160 m³/h from the Grade C room, 50 m³/h from the airlock

The clean utilities behind a batch

Three paths run side by side. Water goes from the still to the hot loop. Steam sterilizes parts, and product moves on to drying and cold storage. Air goes through HEPA filters into the graded rooms.

Water
Air
Steam
Product
Has its own twin page

The physics

The equations behind every curve on this page.

Air changes

ACH = 3600·Q / V_room

Air supplied each hour, divided by the size of the room.

Door-gap leakage

Q = Cd·A·√(2·ΔP/ρ)

Offset air leaks out through the gaps around the closed door. The gap and the air set the pressure step.

Fan laws

Q ∝ N ΔP ∝ N² kW ∝ N³

True when every resistance goes with flow squared. Filters do not, so the real saving is smaller.

Fan power

P = Q·ΔP / η

Air moved times the pressure it is pushed against, divided by efficiency.

Clean-up time

t = ln(100) / ACH

How long a well-mixed room takes to clear particles a hundred to one.

Filters and fan power

Every filter adds pressure the fan has to push against. As filters load, the fan works harder for the same air.

Filter pressure drop over two years, with change-outs

At 20,000 m³/h

Pre-filter (ISO ePM10)
Final filter (ISO ePM1 80%)
Terminal HEPA (H14)
Fan power (right axis)

Fan power, all filters clean: 8.9 kW
Fan power range over two years: 8.9–11.3 kW
Each 100 Pa of filter drop: 0.86 kW

Filters load slowly, then faster as the dust cake builds. Every pascal of filter drop is fan power, every hour the air handler runs.

ΔP(t) = ΔP_clean·exp(t/τ) | P = Q·ΔP / η

What fewer air changes at rest would save

Empty rooms make far fewer particles. Many plants run fewer air changes when rooms are empty, once QA has approved it and the room still recovers in time.

Fan power against air changes

800 m³ suite, filters at mid-life

This air handler (filters go with Q, ducts with Q²)
Cube law from the design point

Clean-up 100:1 at full air changes: 11 min
Clean-up 100:1 at reduced air changes: 18 min
Hours a year with rooms empty: 4,600 h

Fewer air changes in empty rooms cut fan power by more than half. The cube law overstates it, because filter drop falls only in step with flow.

ACH = 3600·Q / V | P = Q·ΔP(Q) / η | t_100:1 = ln(100) / ACH

The pressure cascade

Air must flow from the cleaner room to the less clean one. Each room pushes its offset air out through the door gaps. The step across each door is the check.

Pressure steps from the corridor into the Grade C room

Each step is the room’s offset air leaking through the closed door

Door gaskets sound
Grade C door gasket worn

Step D to corridor: 13.7 Pa
Step C to D, sound gasket: 14.2 Pa
Step C to D, worn gasket: 5.5 Pa

The step goes with the square of offset air over gap area. A worn gasket that opens the gap by 60% takes the step under 10 Pa with the same air.

ΔP = (ρ/2)·(Q / (Cd·A))²

Failure modes

What breaks, what it looks like, what the twin does about it.

Failure mode	The physics	Signals that show it	How the twin catches it
Filter loading	Dust builds a cake on the media. Pressure drop climbs slowly, then faster.	Filter differential pressure up. Fan speed and power up to hold airflow.	Fits each filter to its loading curve and forecasts the change date. Each 100 Pa costs about 0.85 kW at full flow.
Pressure cascade lost	A worn door gasket, a drifting damper or extra exhaust changes the leak path or the offset air.	Step across a door falls toward 10 Pa. Supply or extract flow changes.	Solves for the door gap or offset that explains the step. A gap 60% bigger takes the Grade C step from 14.2 to 5.5 Pa.
Fan belt slip or bearing wear	Belt slip lowers fan speed under load. A worn bearing adds friction and vibration.	Airflow below the fan law for the motor speed. Power up. Vibration at bearing frequencies.	Compares measured airflow with what the fan should move at that speed and pressure.
Humidity out of range	The coil or humidifier cannot hold the room against outside air and people.	Room humidity at the edge of its range. Coil valve wide open. Humidifier at full output.	Runs a moisture balance on the air handler and names the part that has run out of room.
Coil fouling	Dirt on the fins and scale in the tubes cut heat transfer.	Colder water or more flow for the same air temperature. Pressure drop up across the coil.	Tracks the coil’s UA from air and water temperatures, and its air-side pressure drop.

What the twin does not do

Any change to air changes or room pressures goes through your change control and requalification. We show the savings. QA decides.
Your environmental monitoring system stays the record for particles and room pressures. We use its data to find equipment problems.

How we catch it early

A physics twin and a PINN, working on one cleanroom air handler.

The twin is a set of equations for this machine, tuned to its nameplate. Give it the flow, speed and conditions and it returns what every reading should be. It always gives the same answer for the same inputs.

A physics-informed neural network, or PINN, sits on top. It is trained on the plant’s own data and it is penalized whenever it breaks a physics rule. So it can only learn what the equations leave out: wear, fouling, friction. It cannot invent an answer that the machine could never produce.

Step 1
Measure
Flow, pressure, current, temperature, vibration. Whatever the asset already has.
Step 2
Physics twin
The equations of this machine, set to its nameplate. It says what the readings should be right now.
Step 3
Residual
Measured minus expected. Demand, weather and speed are already taken out.
Step 4
PINN
A small neural network that learns the cause of the gap, under the physics rules.
Step 5
Finding
A named cause, a date, and the kWh or hours at stake. An engineer reviews it.

loss = (measured − predicted)² + λ · (physics residual)²

The second term is the rulebook. It is zero when the network obeys the equations.

Why this beats black-box AI

A black-box model learns patterns from history. It needs examples of each failure, and it goes quiet when the pump runs at a point it has not seen. The twin starts from physics, so it works on day one. The PINN adds what physics cannot know about this particular machine.

That is also what makes the savings countable. The baseline is a calculation, not a guess, so the gap between baseline and meter is a number a customer can check.

What the twin checks on this asset

Every pressure step above 10 Pa: now 13.7 and 14.2 Pa
Each filter against its loading curve, with a change date
Fan power against Q·ΔP/η: 10.4 kW expected at 20,000 m³/h
Clean-up time at the air changes in use: 11 min at 25, 18 min at 15
Supply airflow against the air-change target for each room

	Fixed alarm	Black-box AI	Twin + PINN
Needs failure history to learn	No	Yes, lots	No
Says what is wrong	No	Rarely	Yes, by name
Works at a flow it has never seen	No	Often fails	Yes, the equations still hold
Puts a number on the saving	No	No	Yes, in kWh and dollars
Lead time on slow wear	Late	Varies	Weeks to months
Every number can be traced	Yes	No	Yes, to an equation

What the twin puts a number on

Energy and cost, from the equations.

15 air changes when rooms are empty

7.3 kW

10.4 kW in use, 3.1 kW at rest. Only if QA approves the change.

Energy per year

33,626 kWh

4,600 empty hours a year (two shifts, five days)

Cost per year

$4,035

Fan power only, at $0.12 per kWh. Less air to cool and heat saves more.

Final filter changed 100 Pa late

7,477 kWh

0.85 kW more for a year, $897

What we do for you

We watch it. We catch it. We prove it.

Our reliability engineers run the twin on your cleanroom air handler, review every finding, and help your crew fix it. After the repair we compare the meter to the baseline and write the result in the ledger.

Start with one utility See the 90-day pilot

More assets

More from pharmaceutical, then the lead asset of each other industry.