ASHRAE Journal - November 2020 - 55

COLUMN ENGINEER'S NOTEBOOK

detect a failure and seamlessly transition to the backup
machinery. Lacking that, things could quickly "unwind."

Setting the Scene

The fab was built on an accelerated timeline to catch
an economic wave sweeping through the industry.
Thus, it was done as a design-build project with a very
basic owners project requirement (OPR), which was
speciﬁc about the conﬁguration of the cleanrooms and
systems serving them, but left many details up to the
contractor.
Figure 1 illustrates the cleanroom. This is a complex
ﬁgure, and that is part of my point; the N + 1 system
design-and in this case the entire cleanroom was the
system-must address the complexities of the system to
be successful. If it focuses on speciﬁc equipment or subsystems rather than the whole system, problems like the
ones I will describe can occur.
The OPR speciﬁed N + 1 for the cleanroom airside with
the intent being that the failure of a fan in the makeup,
recirculation or process exhaust subsystems would not
take the cleanroom out of production. This column will
focus on the N + 1 issues in the process exhaust subsystem. Figure 1 Notes provides additional information on
how N + 1 was achieved (or not) for the MAU and recirculation subsystems.
Ultimately, the contractor's interpretation of "not
taking the cleanroom out of production" and "failure"
ended up being different than what the owner had in
mind. Speciﬁcally, the contractor viewed a brief outage,
for instance, a process exhaust system being ofﬂine for
three or four minutes during a transition, as not being
out of production. But for us, there were production loss
implications beyond the three to four minutes of downtime. And the contractors' perspective considered a failure in the context of the fans, while in our perspective it
was crashing the cleanroom.

N + 1 and Conservation of Mass and Energy

When I arrived on site at the fab, a good two-thirds of
the fab was coming out of the ground (Photo 1). But the
epitaxial (EPI) cleanroom was online in qualiﬁcation
runs (demonstrating that we could consistently make
good product). With part of the site under construction,
the operating team faced a number of challenges. These
included dealing with problems created by systems that
we didn't own and couldn't touch that were not fully

commissioned-but that had a direct impact on operations in the EPI cleanroom.
The EPI fab manager was having a recurring problem
in the EPI quality control (QC) area (see Figure 1 Notes,
Note 7). At random times, the doors would slam against
their frames, and the force across them made it difﬁcult
if not impossible for someone in the room to force them
open and get out. Adding to the excitement, ﬂoor tiles
had also blown out. Given that the ﬂoor tiles were made
of ±40 lb (18 kg) cast-aluminum, this was somewhat of a
concern.
Forty pound (18 kg) projectiles and impossible-to-open
emergency exits aside, the events contaminated the
cleanroom, shutting down production. The root cause of
the problem turned out to be the makeup air unit (MAU)
controller rebooting at random. When that happened, it
de-energized all its outputs and went through an orderly
restart. Even though we were N + 1 regarding fans on
paper, we were not in the context of the subsystem
because the controller failure was shutting both fans
down concurrently.
Because there were no interlocks between the MAU
and process exhaust control system, when the MAU
went ofﬂine, the process exhaust system continued to
operate. Thus, the process exhaust fan did whatever
it took to achieve design conditions. Since the fan was
attempting to provide the QC area with 6,000 cfm
(2832 L/s) of exhaust, with no source of makeup air, it
was pushed up its curve. Since the discharge was referenced to atmosphere, the inlet pressure and cleanroom
went extremely negative. This is why the doors slammed
against their frames and became difﬁcult to open.
The ﬂoor tiles in most of the cleanroom were perforated to provide a recirculation path to the subﬂoor and
maintain the required air change rates. For the QC area,
the required air change rates were achieved without
recirculation due to the high makeup and exhaust ﬂow
rates, thus the tiles in that area were solid.
During the MAU failure, the pressure differential created by the process exhaust fan was imposed across the
ﬂoor tiles. Since the recirculation systems had no direct
access to makeup air, the subﬂoor went negative as the
process exhaust fans tried to pull air directly from the
cleanroom, causing the subﬂoor plenum divider to collapse, blowing out the ﬂoor tiles.
This opened a path to other areas of the cleanroom,
but without makeup air, the entire cleanroom was
N OVEM BER 2020

ashrae.org

ASHRAE JOURNAL

https://www.ashrae.org/

ASHRAE Journal - November 2020

Table of Contents for the Digital Edition of ASHRAE Journal - November 2020