Nobody on your rig is doing anything wrong. The equipment was specified to remove drilled cuttings, it removes drilled cuttings, and the returns look clean. The problem is that the thing wearing your equipment out is not what any of it was built to catch.
The train is a coarse-to-fine cascade. Shakers screen first, because hydrocyclones plug and erode on unscreened flowline returns, and everything downstream depends on the shakers having gone before it. Each stage is drawn as its own equipment: magnet bars in the header box and the ditch, a sloped screen deck on the shakers, cone banks on the desander and desilter, a drum on the centrifuge.
Watch the counters. Each stage shows what share of the particles reaching it that it actually captures — not a running total, which would only grow with time. The middle stages look weak because by then the coarse fraction is already gone and they are being fed material they were never designed to catch.
All counts normalised to 1,000 particles entering the flowline, so the figures hold regardless of how long this has run. Every particle arriving at the active pit has already survived all seven conventional stages — that population is what reaches your pump with no CleanMud fitted, so the same run gives both answers at once. Press CleanMud: IN SERVICE to bypass the unit and watch the outlet fill.
Judge the train on mass and it scores about 98%. Judge it on the count of particles small enough to enter a running clearance and it scores almost nothing. Neither number is wrong. They are measuring different things, and only one of them is connected to wear.
Mass is dominated by the coarse chips, because mass scales with the cube of diameter. One 100 micron chip outweighs eight million half-micron particles. Shakers, cyclones and centrifuges take nearly all of that mass out. The returns look clean. Everyone concludes the system is working.
But metal is not worn away by mass. It is worn by the number of hard particles that fit into a clearance and get dragged across both faces under load. Almost every one of those is still in the fluid.
Real wear-debris counts rise steeply as size falls; equal-sized bins is the one distribution we know is wrong. The exponent matters and the answer moves materially across the defensible range of 1.5 to 3, so drag it and see. d−1.73 is calibrated so mass removal lands on the 98% figure taken from field experience, not derived independently. Say so if asked.
Every engineer asks this within about ninety seconds, and it is the right question. The answer is that you cannot, and the reason is sitting in your mud already.
Barite is 2 to 74 microns. It is the weighting agent you are paying for and depending on for well control. A mechanical filter fine enough to catch 2 micron wear debris cannot tell the barite from the steel — at that size they are the same object to a screen. It strips both.
What follows is predictable: the medium blinds off in minutes, differential pressure runs away, and your mud weight walks off spec while you are trying to control a well. That is not a filtration problem you can engineer around by going finer. It is a dead end.
Steel wear debris is ferromagnetic. Barium sulphate is not. Bentonite is not. Drilled solids are not. The field reaches into the fluid, takes hold of the ferrous abrasive, and lets everything you actually want flow straight past.
What that buys you:
Zero differential pressure, because nothing is being strained through a medium. No media to blind off, and none to buy, dispose of or change out. No barite loss, so no mud weight excursion. And a capture band that extends below half a micron, where no screen can go at all.
What it does not do:
It will not remove non-magnetic drilled solids — that is what your cyclones and centrifuge are for, and they do it well. Magnetic filtration is not a replacement for solids control. It closes the one gap solids control structurally cannot.
Fine non-ferrous abrasive does come out with it, entrained in the ferrous material captured on the elements. That effect is evidenced by field footage rather than by this model, which simulates ferrous debris only.
Cascade order per Rabia, Well Engineering and Construction (2001) and Hossain & Al-Majed (Wiley, 2015), validated against a generic P&ID. Cut points are industry-typical ranges and vary by make, screen, feed and operating condition.
What fraction of each size class is still in the fluid after all seven conventional stages.
The same seven stages, scored two ways.
A rig can be removing 98% of its ferrous debris by mass and still be circulating essentially all of the damage.