Working on a small robotics lab where the AHU pushes us past NC-35 around 250 Hz, and I’m trying a mix of crosstalk duct silencers and a 2 in mass-loaded vinyl lined plenum to buy about 8 dB without flattening the fan curve. Has anyone validated a simple two-port transfer matrix against COMSOL or in-situ insertion loss before we order hardware?
Do an ISO 7235-style S21 sweep around ‘250 Hz’; include mean-flow attenuation, else the two-port will overpredict IL.
Fold panel breakout into your two-port — treat the MLV skin as a limp-mass TL and include termination radiation — and you’ll land within about 2 dB of COMSOL around 200–315 Hz in my experience. Before you buy, drop in a 1 m mock plenum and measure S21 with flow on plus a nearfield mic outside; if the outside mic at 250 Hz tracks the in-duct level, it’s a breakout issue, not just IL, so the duct stops doing the bass line. @g_blackman89 is right on flow; what’s your Mach, and is the “2 in MLV” MLV over 2 in mineral wool?
Agree with @g_blackman89 on panel breakout; before you buy, build a 0.6–1 m flanged spool with the 2 in MLV-lined plenum and one crosstalk unit and do an in‑situ two‑mic A/B with flow on to see if the “8 dB” holds without killing the fan curve. Keep mics in the plane‑wave region and log ΔP so you can tune flow‑resistive terms and panel loss. Watch for silencer self‑noise under load; it can eat 2–3 dB in that band.