We trialed an inline torsional rheometer on the blender in Reeves County last month; held apparent viscosity 18–22 cP at 100 s^-1 while trimming HVFR from 1.2 to 0.9 gpt, and saw 200–300 psi lower treating pressure with more uniform cluster take per DAS/DTS. Anyone else seeing similar gains, and what PID tuning or sensor placement kept the signal clean around air entrainment and recirculation?
We kept the signal clean by pulling a small slipstream off blender discharge into a vented mini-deaerator and putting the torsional probe in a flooded vertical run about 4D downstream — ‘keep it flooded and quiet’ and don’t mount within 2D of elbows or the hydration tee. PID that held 18–22 cP for us with HVFR 0.8–1.0 gpt was about P=0.6, I=0.15 s^-1, D=0 with a 1–2 s low-pass and a ±0.5 cP deadband; on rate ramps, clamp I to avoid overshoot.
Ran similar in Loving Co.: we put the torsional probe in a short horizontal spool right after the hydration coil and before the sand tee, added a static mixer about 3D upstream, and kept a small backpressure regulator to stay flooded. PID was P modest, I ≈ 0.06 s^-1 with ‘freeze I when air >1%’ (air from Coriolis density), D on measurement with a 2–3 s filter, plus a 5–7 s first‑order lag on the viscosity signal so you don’t chase bubbles. Small caveat: if tub level swings, gate the loop off for 10–15 s or it’ll hunt.
Quick example: we held a steady signal like your ‘18–22 cP at 100 s^-1’ by low-pass filtering the torque output at about 1.5 s before the PID, which stopped the loop from chasing bubbles off the blender tub. Small caveat: push the filter past about 2 s and you’ll lag on sand step-ups, so HVFR trim reacts late.