Last week, we (where we is mostly Ron and Gary, a couple of the
guys I worked with) did an interesting test; we performed a leak
test on a clean room that is in the final stages of construction.
The idea was to use the make up air handling systems to do what
amounts to a “blower
door” test on the clean room. In general terms, the steps in
the test where:
Obtain an assessment of the anticipated
leakage rate from the design team.
Make sure all the fab doors and other
openings in the fab envelope are closed.
Seal any openings that have yet to be
closed by final construction.
Verify that the exhaust systems are off and
that the ducts leaving the fans are not leakage
Pressurize the clean room with the make up
Measure the flow rate required to generate
a measured differential pressure.
Correct the measured flow and pressure to
the flow that would exist with the clean room pressurized to the
required set point.
Inspect the envelope construction to verify
the quality and that there are no major breeches that could result
in a higher than anticipated leakage rate.
The following summarizes the results of our test.
Design leakage allowance – 1,888 cfm at
Projected leakage based on measurements –
7,589 cfm at 0.05 in.w.c.
Difference relative to design – 5,701
cfm at 0.05 in.w.c. (over allowance)
Percent of design leakage – 402%
Excess as a percent of design airflow –
We supplemented the test with an inspection of the clean room
envelope construction, including focused inspections of typical
trouble spots like wall to ceiling joints, duct and pipe
penetrations, etc. (This means that Ron crawled around in the
interstitial space getting pretty dirty and cut up by protruding
pieces of metal, etc. and then filled me in on what he saw later
Based on what Ron observed, we concluded that the extra
(unanticipated) leakage likely represents hundreds if not thousands
of small leaks attributable either to the details of the
construction of the envelope elements or random minor deficiencies
rather than one large leak. For instance:
The project includes a large window-wall on one end
o f the clean room. The framing system employs
“weep-holes” that allow air to move through the system
to prevent moisture from becoming trapped inside and eliminate
condensation. Leaks of this type are small in scale on an
individual basis but can add up when they occur repeatedly over a
large area. Plus, they exist for a purpose and thus, they can not
Totally airtight construction is probably just about
impossible to achieve in a the field. For instance, a
common test for verifying that a
BSL-4 facility is airtight is to pull the area down to a
pressure of minus 2.0 in.w.c. and then turn off the fan and
demonstrate that the pressure does not decayed to less than minus
1.0 in.w.c. after 20 minutes with corrections made for any changes
in ambient barometric pressure or temperature that have occurred.
Perfection would mean the pressure did not change at all.
While the results of our measurements are somewhat discouraging
when viewed in the context of the design, we actually were kind of
relieved by them when contrasted with past experience. For one
thing, 5,700 cfm amounts to only a small fraction of the design
flow rate for the make up systems and can likely be accommodated by
For another, the clean room envelope is still in the final
stages of fabrication. The leakage as measured by our test will
likely drop some as construction is completed and some of the
openings we sealed in a temporary manner are sealed
The fact is that many projects in the past have demonstrated
unanticipated leakage rates that are significantly higher than the
design allowance. One of the clean rooms in the wafer fab I worked
at leaked so much that we ended up running the back up fan in the
make-up AHU around the clock. The extra flow necessary to achieve
the desired pressure relationships forced the supply duct system to
run at static pressures in excess of its pressure rating. This of
course, just made the leakage problem worse.
Operating in this mode placed a significant energy burden on the
system in addition to the operating challenges associated with
running a system at pressures above those anticipated by the
design. Specifically, the leakage meant we were handling an outdoor
air load that exceeded design expectations by 25-30%
round-the-clock. This translated to a lot of fan energy to move the
extra air and a lot of process energy to heat, cool and humidify
the extra air.
In contrast, the unanticipated leakage associated with the
project Ron, Gary and I tested last week, while significantly high
relative to the design allowance (402%), is relatively small in
terms of the design make up air capacity (6%). I could be wrong,
but I suspect the systems can be adjusted to handle the leakage
with out undue strain. Toward that end, we have reported our
results to the design team and asked for their advice on how best
to adjust the systems to meet the design intent while dealing with
the realities of the built environment. I’ll let you know how
things turn out in a subsequent post.