ght elements went into a new filtration skid at a Malaysian petrochemical plant last month. Same part number, same pore grade, same furnace load, one vessel. Six days later three were blocked and maintenance wrote asking for a coarser media. Before quoting anything we asked for three things: the cleaning record after hydrotest, the differential readings from the first shift, and whether anything upstream had been taken out during start-up. The temporary basket strainer had been pulled on day two because it slowed the line. What blinded those three elements was pipe scale they were never meant to see. The other five ran on a loop nobody disturbed, and were still in service eleven months later.
Nothing about that exchange is unusual. Elements that fail early are rarely mis-specified - they are commissioned badly, and week-one damage never appears in a failure report.
Table 1 - What blocks a new element in week one
| Source | Symptom | How it is intercepted | Whose job |
|---|---|---|---|
| Weld slag and mill scale after hydrotest | Step jump on the first day | Flush the loop through a sacrificial screen before porous parts are fitted | Installer |
| Rust bloom from a wet line left standing | Even brown staining, blocked inside three days | Blow dry and purge straight after pressure testing | Installer |
| Temporary strainer removed too early | Two elements fail while their neighbours survive | Keep it until two consecutive inspections come back clean | Operations |
| Sealant tape, thread compound, grease | Patchy film, poor wetting, false low flow | No tape past the last thread; assemble with clean gloves | Installer |
| Too-fast flow ramp on start-up | Cake torn off the wall and re-deposited | Step flow up, hold each step until differential steadies |
Operations |
1. Week one is a plumbing problem before it is a media problem
Our standing advice at the front of a project is unglamorous: spend the first two weeks on something disposable, and keep the porous elements in their bags until the loop is clean.
Step 1
Chemical clean or high-velocity flush, then verify with a screen inspection rather than a sample bottle.
Step 2
A 100–200 µm stainless basket costs less than one porous element and is meant to be thrown away.
Step 3
Circulate at 50–60% of design flow for two shifts, inspecting each shift.
Step 4
Two consecutive clean inspections is the release criterion. One is luck.
Step 5
Unwrap on a bench, not on the floor; new seals torqued to the drawing value.
Step 6
Date of installation, the two readings taken that day, and a signature - years later the only evidence of what those elements were asked to do.
There is also a reason not to intervene early. In most liquid duties the wall first forms a thin layer of its own contaminant, and that layer becomes part of the media: capture improves and differential settles onto a plateau. Strip it off during the first shift and the cycle starts again.
2. Read the shape of the curve, not the number
A single differential figure tells you how blocked an element is. The shape tells you why, and watching that shape through the first week is the cheapest diagnostic on the plant.
Table 2 - Three curve shapes in week one
| Shape | Most likely cause | What to do |
|---|---|---|
| Smooth rise that flattens after 8–20 hours | A working cake forming and stabilising | Leave it running; note the plateau as your reference |
| Step jump above 30% inside one shift | Debris arriving from upstream, often after maintenance | Stop, inspect the upstream screen, backwash gently |
| Sawtooth - recovers after every clean, never settles | Cleaning too early or too hard | Trigger on measured differential, not on the calendar |
The sawtooth is the pattern most often misread as success: every clean works, while each round spends a little wall life.
3. What the furnace does to eight pieces that were supposed to match
The rest of the story is written before dispatch, on our shop floor, and it is what visitors ask about once they have walked the furnace aisle.
A vacuum load is never thermally uniform. Material in the working centre and material nearer the door do not see the same time–temperature history, and because shrinkage governs both density and permeability, that history ends up inside the part. We map each load by tray position, test the extremes, and hold the spread to a defined band.
Table 3 - Permeability spread across one mapped load (indicative)
| Zone | Share of batch | Permeability vs load average | Where the parts go |
|---|---|---|---|
| Working centre | 55–65% | Within ±4% | Installations that must behave identically |
| Periphery | 25–35% | −6% to +7% | Single elements and small lots, graded on their own readings |
| Outside the mapped zone | balance | outside the band | Held back, re-run, or graded into another class |
Two things belong in a specification. Tell your supplier the pieces go into one vessel, so they are spread across the mapped centre rather than one corner of the load. And replace them as a matched set from one load, not from the remnants of two deliveries.
4. How many sets do you actually need on site?
Project teams usually solve this by ordering a comfortable number, then finding it comfortable in all the wrong places. The calculation is short.
Table 4 - Three duty scenarios
| Scenario | T (days) | I (days) | Sets to hold |
|---|---|---|---|
| In-house ultrasonic cleaning | 7–10 | 30 | 2 |
| Contractor clean and return freight | 30–45 | 30 | 3 |
| No cleaning; elements are consumable | - | 60 | 2 |
Where crates stand matters too. Elements dispatched to an open contractor yard have arrived with rusted flanges after one wet season; the same batch sent to a closed store went into service exactly as it left our bench.
What we ask, and what we send
Eleven months of difference between two elements carrying the same part number was never about the media. It was about what came down the pipe on the first shift, and about where those parts sat in one furnace load.
NAVI Titanium builds sintered powder, mesh laminate and fibre felt elements in stainless steel, titanium, bronze, Hastelloy, Inconel, Monel and nickel, as cartridges, discs, tubes, plates and welded assemblies made to your drawing or reverse-engineered from the part you send us. Tell us the fluid, temperature, differential target, what sits upstream, how you intend to clean it, and the date it is due on the line. You get back a structure recommendation, a rated drawing, a realistic lead time, a start-up protection checklist your installer can tick, and a batch report carrying the load map and retained-sample reference.
