Author: A cleanroom engineer who has been tormented by pressure differentials and bacterial colonies many times

At 1am, in the QC corridor of a biopharmaceutical factory in South China, the project manager threw the differential pressure gauge on the table for the third time.

The fresh air cabinet has been replaced, the FFU is fully open, the HEPA just had a leak test last week, and the middle school efficiency was replaced two weeks in advance - the pressure difference is still stuck at 12Pa, and the gradient is reversed. Not to mention microorganisms, on the A-grade platform with a B-grade background, three consecutive batches of settling bacteria drifted to 1cfu, and the customer's face had already turned black. "

He asked me, 'Teacher, is there a problem with this batch of equipment? Should we use another MAU?'? "

I looked at the floor plan and didn't say anything. The problem is not with the equipment. Three years ago, on that flat surface, the doors of the two buffer rooms were opened upside down, and the return air outlet for the last change was half eaten by the process pipe well - the equipment took all the blame, and the layout was the original sin.

I have seen this too many times.

When there was a deviation in the clean room, the first reaction of Party A was to "adjust the equipment", and the construction party and supplier were also willing to cooperate - after all, replacing HEU, adding MAU, and changing the frequency converter are all budgeted actions. No one wants to say 'the floor plan may have been drawn incorrectly' first, because it was drawn by the design institute three years ago and signed by the owner themselves.

But the truth is:

Pressure difference is a spatial issue, not an air volume issue. No matter how much air you provide, when the door between two adjacent rooms opens and the high-pressure room "spits" towards the low-pressure room, the gradient account will be broken. Whether the door opening direction, door gap, buffer level, and self closing device are installed - these "differential pressure accounts" that should be calculated in the design stage cannot be pulled back by fans in the later stage.

Bacterial colonies are a flow field issue, not a disinfection issue. Smoked once with VHP and dragged three times with disinfectant, the next batch will still float. Why? Because the airflow organization was short circuited during design - the supply air outlet was facing the operating surface, but the return air outlet was blocked directly below the pollution source, or the dirty and clean boundary of the last change of clothes was not locked, and the personnel's bacterial load was not suppressed from the root.

Equipment is the end, layout is the source. Equipment can iterate, and once the walls are built or the layers are sealed, the layout is a one-time decision.

Project background: B+A sterile workshop, producing injections, EU GMP Appendix 1 caliber.

Appearance problem:

Rolling room → B-level corridor → C-level buffer, pressure gradient set at 15/10, often drops to 8/5 in actual testing;

Two out of the three batches of settling bacteria in the A-level laminar flow table reached 1cfu, although they did not exceed the limit, the customer did not accept them;

Everything that is active on site has moved: MAU frequency, FFU gear, and middle school efficiency replacement cycle.

I only did two things on site:

The first thing is to guard the door.

The door leading from the capping room to the B-level corridor is marked with "normally closed+self closing device" in the design, but in actual installation, the self closing device was not installed for forklift access, leaving a 12mm gap in the door. The capping room is the room with the highest positive pressure (+45 relative to the outside). When the door is opened, the wind flows directly into the B-level corridor, and the buffering gradient from B to C is "topped" off. I tested it and found that when the door was open for 3 seconds, the instantaneous positive pressure in the B-level corridor dropped from+25 to+13- this time, it ate up most of the gradient account.

The second thing is to look at changing clothes.

The last buffer room (from C to B) has a design where the return air outlet is located below the side wall. However, during the construction phase, the process pipe well took up space, and the construction unit "took advantage" to move the return air outlet to the ceiling, resulting in the return air becoming "return air". The particles and dander (the main carrier) generated when personnel take off their outerwear should have been taken away by the lower part, but now they are all spinning in the upper part of the buffer room and then "sucked" through the door gap by the B-level area.

The root is clear:

The door is not equipped with a self closing device+the door gap exceeds the standard → the gradient is pseudo unstable

The last form of return air flow is back → personnel carrying bacteria are not suppressed on site → A-level bacterial colonies float

Rectification did not involve replacing a single device: repairing the self closing device, sealing the door gap to within 6mm, replacing the return air outlet with a return air outlet, and installing an audible and visual alarm interlock door for the pressure difference in the buffer room. Two weeks later, retesting showed a stable gradient of 15/10, with five consecutive batches of A-class settling bacteria at 0 cfu. At the beginning of this order, we switched to HEPA and added MAU, which cost us money, but the problem still exists.

This project is more interesting. 12 inch wafer front track, ISO 5 main workshop, FFU fully laid, lower return air duct.

Appearance: The pressure difference in certain areas drops regularly by 3-5 Pa from 2-4 pm every day, while microseismic monitoring shows friction at the threshold edge. OP said, 'Is it because the wind pressure outside has changed?' FAE said, 'FFU aging is uneven.'.

I went to check the mezzanine and found the problem in ten minutes:

The return air duct is occupied by 30% of the process pipeline. Three years ago, when expanding production, two pipes were added to each of the special gas, CDA, and PCW systems. If they couldn't pass through the ceiling, they would be stuffed into the return air duct and stacked against the wall. After the cross-section of the return air was narrowed, during the afternoon, when the process exhaust (scrubber side) increased, the return air resistance suddenly changed, and the static pressure in the main workshop shook - the FFU speed did not change, but the "belly" of the return air was trapped, and the positive pressure naturally dropped.

What about the microseismic line? The fixing clips of the several special air pipes in the return airway were welded to the partition wall of the airway back then, and the other side of the wall is the ISO 5 main workshop. When the pipeline surges, the wall will transmit a microseismic.

The rectification plan is not to replace the FFU, but to move the process pipes in the return air duct out of two, redo the duct separation, and replace the special gas pipe clamp with an independent bracket on the ground. The device didn't move, the pressure difference curve flattened, and the microseismic also went down.

After more than a decade of working in a clean room, I increasingly feel that a good clean room is not created through debugging and adjustment, but through flat drawings. Debugging can fix a deviation of ± 20%, but it cannot fix the -50% pit buried in the design.

Give my colleagues and the client a few personal 'red lines', it is best to check them in the PL review before building or renovating:

The pressure gradient should be counted as the "second the door is open", not the "door is closed".

The pressure difference between adjacent rooms is equal to or greater than the compensation amount for transient backflow when both sides of the door are open. The buffer level, door opening direction (must be towards the high voltage side), door gap limit, and self closing/interlocking of the main pedestrian flow channel must be included in the design specifications and cannot be "self determined during construction".

The boundary between dirt and cleanliness in changing clothes is the first line of defense for controlling bacterial colonies, not the last line.

The last removal (low cleanliness → high cleanliness) must be returned, and the return air outlet position cannot be eaten by the pipe well/cabinet; The airflow in the undressing area should be drawn away from the "carrier side" on the spot, rather than being sucked into the high net area.

3. The return air path is the "vein" of the clean room, so don't let the process pipe occupy it.

Reserve at least 30% of the reserved section for return air ducts/return air ceilings for future production expansion; The process management entering the aisle requires separate approval, and cannot be left unchecked.

4. For the doors of adjacent rooms, don't just draw a door symbol, mark the opening direction and interlocking logic clearly.

I have seen too many floor plans with beautifully drawn door symbols. During construction, for the sake of convenience, they were changed to face towards the low-pressure side, and the pressure gradient was directly removed.

5. Monitoring points (differential pressure gauges, particle counting) should be pre embedded with positions that can represent the real flow field.

Do not install it directly below the air supply outlet (false beauty), and do not install it in a dead corner behind the door (false safety). The monitoring points are for the operation period, not for DQ.

There is a strange phenomenon in the cleanroom industry: the equipment is getting newer and newer, and the floor plan is still the same as ten years ago.

Unstable pressure difference and excessive bacterial growth, people habitually twist the frequency converter, replace HEU, and increase disinfection frequency - these actions are not wrong, they are in the wrong order. First, go back and flip through the floor plan, check the door, check the return air, check the buffer, and check the dirty and clean boundary. You can probably find the answer. Can't find it, move the equipment again.

Design is a one-time decision, equipment is a continuous payment.

Don't let the equipment take the blame for the design for a lifetime.

The author is a professional in cleanroom engineering and has served multiple production lines in biopharmaceuticals, semiconductors, and sterile equipment. Welcome colleagues to share in the comments section the "layout burying mines" cases you have encountered, and choose the typical ones for me to discuss in the next article.