“Ah, the killer,” Callahan murmured. “You don’t fix that, tubes will sing for a week, then snap like guitar strings.”

Results: 35% baffle cut dropped pressure drop to 65 kPa (good) but U fell to 235 (bad). 20% baffle cut? Pressure drop: 110 kPa—unsafe for the diesel pump. She needed a different geometry entirely.

Final run: outlet crude temperature: 248°C, U = 291 W/m²·K, pressure drops shell/tube: 58/31 kPa, fouling resistance: 0.00035 m²·K/W. Within all limits.

She clicked to the (shell-and-tube) module. The color-coded flow map showed dead zones near the shell’s center. The baffle spacing was too wide—fluid was meandering, not turbulent. She reduced baffle spacing from 500 mm to 300 mm. Re-ran.

But a new warning blinked red: Vibration potential. Bundle natural frequency close to vortex shedding frequency.

“You’ve got laminar flow in the shell,” Callahan said, peering over her shoulder. “Look at the velocity profile.”

Better. U climbed to 250. But pressure drop on the shell side spiked—from 40 kPa to 95 kPa, exceeding the 70 kPa limit. Trade-off city.

Elena sighed. “What if I change baffle cut from 25% to 35%?” That would reduce cross-flow velocity, lowering pressure drop but also reducing heat transfer. She ran the parametric study in HTRI’s built-in optimizer.

First simulation ran hot. Not good hot— danger hot. The outlet temperature of the crude was 10°C below target. She checked the stream data: shell-side fluid (hot diesel) at 300°C, tube-side fluid (cold crude) at 40°C. Pressure drops were within limits, but the overall heat transfer coefficient, U , was a pathetic 180 W/m²·K. The required was 280.