工业厂房工程

David, we're at the most critical installation in the entire assembly hall — the crane runway system. Two 50-ton overhead cranes, span 28 meters, rail length 120 meters. The crane runway beams are welded I-sections, 1,200mm deep, 400mm flange width. Each 12-meter beam section weighs 4.8 tons. The tolerance requirements are merciless: rail straightness ±3mm over any 10-meter length, rail-to-rail span ±5mm across the full 28-meter width, and elevation ±5mm over the entire 120-meter length. If the rails aren't within these tolerances, the crane will skew, wheels will wear prematurely, and in the worst case, the crane can derail.

Understood, Tom. Miguel, how are we installing the runway beams? These are heavy sections at height — 14 meters above the floor at the top of the columns.

We're using a 200-ton mobile crane positioned outside the building. Each beam section is lifted and placed onto the column corbels. The corbels have pre-set shim packs — the shims allow us to adjust the beam elevation before we do the final welding. The beam-to-corbel connection is full-penetration butt weld — it's a moment-resisting connection because the crane's braking and acceleration create significant longitudinal forces in the runway. We'll weld from a mobile elevating work platform positioned under each column. Welder qualification: all position, certified for full-pen welds. NDE: 100% UT on all welds.

(setting up the total station on a fixed pillar) The survey control network was established last week — eight control points around the hall, all tied to the project grid. I'm measuring each column top elevation and the corbel bracket position before we place any beams. If the corbel is out of position, no amount of shimming will fix it. Column C-12, north corbel: design elevation 14,200mm, measured 14,198mm — 2mm low, within the 5mm pre-placement tolerance. OK to proceed.

(after all beams are welded, during final alignment)

Rail alignment verification. West rail: I'm measuring straightness at 2-meter intervals. KP 0-10m: max deviation 1.5mm — good. KP 30-40m: 2.5mm. KP 80-90m: 2.0mm. KP 100-110m: 3.5mm — that's over the 3mm limit. David, there's a problem at KP 104 on the west rail. The rail is bowing inward by 3.5mm over a 10-meter section.

Let me check the beam at KP 104. (climbs up) The beam splice at this location — the bolted flange connection between beam sections — has a 2mm gap on the inner flange. The splice wasn't pulled up tight before the bolts were torqued. We need to release the bolts on this splice, pull the flanges together with a come-along until flush, then re-torque to spec. After that, we recheck the rail straightness.

Dr. Patel, this semiconductor fab cleanroom is 8,000 square meters of ISO Class 4 space — that's maximum 352 particles per cubic meter at ≥0.5μm at rest. We have 840 FFUs in the ceiling grid, each 1,200mm × 600mm, delivering unidirectional downward airflow at 0.40 m/s. Before the fab equipment moves in, we must certify 100% of the FFUs — filter integrity, airflow velocity, and uniformity. What's the test protocol?

Three-phase protocol. Phase 1: filter integrity. We do a DOP/PAO aerosol challenge on 100% of the installed HEPA/ ULPA filters. The upstream concentration is 20 mg/m³ of PAO aerosol. We scan the entire downstream face of each filter — both the media and the gasket seal — with a photometer probe at 50 mm/sec. Any leak above 0.01% of the upstream concentration is a fail. Phase 2: airflow velocity. We measure at working height — 900mm above the raised floor — on a 600mm × 600mm grid. Each measurement is a 60-second average. Target is 0.40 m/s ±20%, meaning 0.32 to 0.48 m/s. Phase 3: airflow uniformity. The relative standard deviation of all velocity measurements must be below 15%.

Li Wei, let's start Phase 1 on Quadrant 1 — 210 FFUs. Set all FFUs to 100% speed. I'll run the aerosol generator upstream of the ceiling plenum.

FFU-001: scanning the filter face... no leak above 0.005%. Gasket seal scan... clean. FFU-002: clean. FFU-003 through FFU-028: all clean. FFU-029 — wait, I'm getting a 0.03% reading on the photometer near the gasket corner. This is a leak — the gasket isn't fully compressed at this corner. Marking with red tape for repair.

Good catch. The fix is to re-torque the FFU hold-down clamps at that corner. If the gasket still leaks, we replace the gasket. This is exactly why we test 100% — a single leaking HEPA in a semiconductor fab can introduce particles that ruin millions of dollars of wafers. Li Wei, after you finish the integrity scan, we'll do the velocity traverse. Let's aim to get Quadrant 1 done today — 210 FFUs, about 6 hours of work including repairs. Three more quadrants to go after that.