An inspection sheet that serves a small clean-water pump well will miss things on a large slurry pump or a mill discharge pump. One problem is physical: heavy, thick-walled castings that cannot be moved easily, do not fit a standard measuring machine, and can exceed the capacity of a test loop. The other is operational: dense slurry, hard angular particles, and occasional tramp steel from the mill, with a shutdown that costs far more than the pump. This article separates the two and sets out where inspection points belong, what each one checks, and what record it should leave behind.

1. Two kinds of special conditions

Start with the scale. Published product literature puts mill discharge pump suction branches in the 250 to 850 mm range, flows from a few tens up to 20,000 m³/h, and motor ratings from a few hundred to several thousand kilowatts. At that size the questions change. The core of a thick section solidifies slowly, so shrinkage porosity and coarse carbides collect there. A casing several metres across will not fit the travel of a conventional coordinate measuring machine. A complete unit may ask more flow and power than the test rig can deliver. Lifting itself carries a risk of damage, so anything that can be measured in place should be measured in place.

The second condition is the load. These pumps sit at the discharge of ball and semi-autogenous grinding mills and feed cyclones, thickener underflows and tailings lines. Solids concentration is high, particles are sharp, and broken grinding media or liner fragments can arrive with the slurry. A published failure investigation of a cyclone feed impeller recorded roughly 720 hours of service against a rated 2,145 hours; the stripped component showed cavitation, abrasive wear and corrosion acting together, not simple thinning. Such pumps normally run continuously and are serviced when the grinding circuit stops, so a wear part that fails a week early costs lost production rather than a spare part.

Inspection points therefore follow risk rather than process order. Size decides whether coverage is complete; mill duty decides whether the checks are relevant. Four layers form the common base: material, internal soundness, geometry and assembly, and the assembled machine. Each risk axis only adds items inside those layers.

2. Material: from ladle spectrometry to sectioning the casting

Optical emission spectrometry (OES, commonly to ASTM E415) reports what was in the ladle. Carbon and sulphur need a combustion method such as ASTM E1019, and a handheld XRF analyser is handy for material identity checks but cannot measure carbon. The 25 to 30% Cr grades used for slurry pump wet ends correspond to ASTM A532 Class III Type A, with a Cr/C ratio that published specifications place around 7 to 9. Molybdenum matters more as sections get thicker: it suppresses pearlite so the core can still transform to martensite during cooling.

A ladle sample does not describe the structure that actually formed in the heaviest part of the casting. A stricter acceptance practice takes one representative casting from the heat and cuts it transversely on a coolant-fed metallographic saw, then works directly on that section: chemistry of the casting body, a nine-point Rockwell C hardness traverse taken 5 mm below the as-cast surface, core metallography with a 2% nital etch and photographs at 100×, 500× and 1000×, plus quantified retained austenite and eutectic carbide volume fraction. Publicly cited reference values include retained austenite at or below 12%, eutectic carbide at or above 24%, M7C3 carbide microhardness around 1,500 to 1,800 HV, a tempered matrix around 690 to 800 HV, and bulk hardness of HRC 58 to 65. The point of a nine-point traverse is the gradient. A large drop from surface to core says the section did not harden through, and wear in service will be uneven.

Abrasion testing often refers to ASTM G65, the dry sand/rubber wheel method. It supports relative comparison between batches and processes; it does not convert into field life. Keeping the result in the batch file is more useful than printing a multiple in a brochure.

One clause that belongs in every inspection procedure: high-chrome white iron at roughly 58 HRC should not be repair welded to hide a defect. The heat-affected zone is softer than the parent metal and the fusion line becomes a crack initiation site, which coarse mill particles will find quickly. An out-of-specification casting goes back to the furnace. Material certificates follow EN 10204 3.1, and the heat number has to lead back to the heat-treatment batch.

3. Internal soundness: ultrasonics for heavy sections, surface methods for the skin

Radiography, including digital detector imaging (DR), shows volumetric defects such as gas pores, shrinkage cavities and inclusions, and the image is easy to read, but penetration limits it on thick sections. Industrial CT gives three-dimensional location and sizing, with limits set by equipment and component size. For heavy pump castings the main tool is ultrasonic testing. Conventional UT picks up internal reflections, phased array (PAUT) can steer and focus the beam across complex geometry, and time-of-flight diffraction (TOFD) sizes crack height with useful accuracy. All three are sensitive to planar defects and reach deep enough for thick walls. UT is run after heat treatment, once transformation and stress relief are complete, and the procedure states the probes, scan coverage and an acceptance gate based on an 80% back-wall echo.

Surface and near-surface work uses magnetic particle testing (MT, ASTM E709) on ferromagnetic parts and liquid penetrant testing (PT, ASTM E165) for open surface defects on any material. Personnel qualification follows ISO 9712, with Level II interpretation and Level III ownership of the acceptance criteria; the report should show who signed and against which criteria. Hydrostatic testing comes after all machining and NDT and before coating, at 1.5 times design pressure, with hold time and acceptance conditions fixed in the contract.

4. Geometry: taking metrology to the shop floor

A conventional CMM has a limited measuring volume, and a casing several metres across will not fit. The usual alternative combines a laser tracker for large-volume measurement on site with blue or white light scanning to collect a point cloud, then a colour-deviation comparison against the CAD model. Complex flow passages and distortion show up immediately. Scanner acceptance is commonly checked against VDI/VDE 2634 Part 3 and ISO 10360, with reports issued by laboratories operating to ISO/IEC 17025. Temperature discipline still applies: precision work belongs in a 20 ± 1 °C metrology room after the part has soaked, and on-site work logs ambient temperature and applies a correction where needed.

Two further items are easy to omit. One is wall and liner thickness, swept on a grid with an ultrasonic thickness gauge, because wall loss in a heavy casting rarely shows up in drawing tolerances. The other is the assembly datum. Shaft bore, wear ring seats, liner mating faces and flange registers should be produced in a single setup where possible so that coaxiality and face runout are controlled together. Bolt lug positions on liners are checked more reliably with a steel fixture over the whole set than by measuring each dimension; whether a liner set bolts up in one shift or stalls the mill for a day often comes down to this check.

5. The assembled machine: test what you can, declare what you cannot

Unbalance matters more as rotors grow. Balancing follows ISO 21940-11 at G6.3 or G2.5, and the certificate records net rotor weight excluding the mandrel, the upper operating speed, correction planes and the half-key convention. Trimming an impeller diameter or changing a wear ring means another run on the balance machine.

Hydraulic performance testing follows ISO 9906:2012 or the equivalent GB/T 3216, with acceptance grades 1B, 1E, 1U, 2B, 2U and 3B. Tolerances apply at the agreed guarantee point, not across the whole curve. One detail is often left vague: a certificate that says “tested to ISO 9906” without naming the grade is unverifiable, since Grade 1 and Grade 3 accept very different deviations. Measurement uncertainty, instrument calibration records and the measured points belong with the curve; a curve with no points underneath it is a drawing, not a record.

Large pumps regularly meet the limits of a test rig. Two routes are accepted in published practice. One is a test on a geometrically similar model with results scaled by similarity laws, in which case the certificate must state whether a model or the delivered machine was tested. The other is a string test of the complete set, or a field acceptance measurement after installation, with motor, piping and valves included. One boundary should be stated openly: ISO 9906 applies to liquids behaving like clean cold water. Slurry concentration and particle size shift head, efficiency and cavitation behaviour, so a clear-water curve is not a slurry performance guarantee and needs correction for the duty. NPSH tests commonly use a 3% head drop as the criterion.

Vibration follows ISO 20816 or GB/T 29531, measured at the bearing housing in horizontal, vertical and axial directions, with the highest broadband reading used for assessment over 10 to 1000 Hz as RMS velocity, sorted into zones A to D. The often-quoted boundaries of 1.4, 2.8 and 7.1 mm/s for medium machines above 15 kW running between 120 and 30,000 rpm are orientation only; zone lines vary with machine class and support stiffness. The real value of the factory reading is as a baseline. A pump that sits near its baseline for two years says more than one with a flattering single number; a reading that doubles in six months deserves attention even if it still falls inside an acceptable zone.

6. Four extra checks for mill duty

The four layers above are the common base. Mill discharge service adds four questions, and they concern the assembled running clearances rather than whether each individual part is within tolerance.

• Running clearances and the suction-side wear ring. Published heavy mill pump designs leave a generous gap between impeller and liner so a bed of slurry buffers the casing, enlarge the cutwater clearance, and use an adjustable wear ring on the suction side to take up clearance online. The item to inspect is therefore the assembled gap and the full adjustment travel, not the impeller outside diameter alone. The fixture needs to verify the whole travel.
• Liner bond and thickness. Rubber-lined mill pumps need checks on cured thickness, hardness, adhesion and blisters, with the bond between metal frame and elastomer confirmed by ultrasonic thickness measurement plus peel or bond testing. A locally detached liner lets slurry cut through the casing from behind.
• Impact tolerance and defect handling. Coarse particles and tramp iron are absorbed by toughness reserves and generous radii, not by a welded-up surface. This is the same rule as the welding ban above, with consequences that arrive sooner in a mill circuit.
• Spares interchangeability and duty boundaries. Wet-end parts of the same model have to interchange, and parts shared by left- and right-hand symmetrical casings should pass the fixture set by set. Shipping documents should record debris and tramp-metal exclusion arrangements, allowable particle size and concentration range. These are not inspection of the pump itself, but they decide whether the inspection conclusion still holds on site.

7. Join the records, and extend inspection into service

A single report proves little on its own. What a quality system can do is link heat number, heat-treatment batch, chemistry, hardness traverse, metallography, NDT, dimensions, balance, hydrostatic test and performance curve into one record, with EN 10204 3.1 certificates and ISO/IEC 17025 laboratory data where third-party evidence is needed. Process capability tracked through SPC and Cpk trends says more than any one passing result.

Recent international practice pushes inspection past the factory gate. Online measurements of flow, density, speed and energy feed calculations of total developed head and NPSH, while wear trends on liners, impellers and throatbushes support remaining-life alerts. Those systems need a baseline, and the factory vibration and performance readings are exactly that starting point. These are publicly documented technical routes; the actual configuration is a project-level agreement, not a line in a standard product specification.

8. How EXCELLENCE PUMP applies these checkpoints

EXCELLENCE PUMP has built slurry pumps for more than twenty years. Casting and heat treatment of wet-end parts, machining, final assembly and performance testing are carried out within our own system, supported by an independent hydraulic design team, an ISO 9001 quality system, and projects in both domestic and export markets. Our service scope covers selection and scheme design, installation and commissioning guidance, long-term spare parts supply, periodic follow-up and fault response.

On large pump and mill duty orders, inspection items and acceptance grades are settled at contract stage and records are filed by heat number. The following reports can be agreed for delivery with the order: chemistry and nine-point hardness traverse, metallographs, NDT reports, dimensional and running-clearance records, balance certificate, hydrostatic and performance curves, and the vibration baseline. Equipment configuration and inspection capability follow the actual production line; third-party witnessing or a named standard can be written into the technical agreement in advance.