Has anyone run a live torque-and-drag model tied to the rig’s WITSML feed at 1 Hz? I’m piloting it on a Delaware Basin pad to trim slide time and stick-slip, aiming to save 6–8 rig-hours per well and cut mud losses — are the ECD forecasts stable enough to meaningfully reduce waste and seep risk?
I’d gate the model with a 7–10 s rolling median and freeze updates during pump ramps so ECD doesn’t chase noise; auto‑tune friction factors per stand from off‑bottom torque/drag, and bound ECD to a per‑interval fracture gradient to flag seep risk. If you’ve got PWD, fuse it with a light Kalman update; without it, blend standpipe dP/SPM/flow with temperature‑corrected mud density so the “ECD forecast is stable enough” and not yo‑yoing. Are you pulling WITSML 1.4.1.1 SPM and hookload quality flags and doing a quick per‑BHA calibration after the first connection?