JHB Trip · Task Manager
Supreme Poultry, Germiston · 28-31 July with Johan, then 3 August onward · target: software, telemetry and training complete by Wednesday, 19 August · flight booked for Wednesday afternoon, with Wednesday morning deliberately kept clear.
Dashboard
L1
RunningBellows fitted, springs re-tensioned, air supply corrected. Auto top-up repaired by Supreme 14 Aug - started clean, no overflow. Software Sunday, then a wet run.
L2 · bulldog
Software loadedNew software on PLC + HMI, dry-testing. Fryer has never had an auto top-up, and the pipework has no drain valves. Both quoted. Roll axle still missing.
L3 · GEA
RunningSpilling and panel resets fixed. Lead line for software + telemetry - 30 cm cable run to the gateway. Its fryer high-level float is faulty; GEA OEM on site next week.
Telemetry
L3 fine · L1/L2 to testGateway repaired and online, registers defined. L1 and L2 runs are 10-15 m of unscreened cable - testing today, and Golliath told either way.
Documents
v0.4 + cardsSOP and Maintenance Manual at v0.4 with revision histories, plus printable training cards and a dry-test record sheet. All in the Docs tab.
Claude's estimate, weighted by remaining effort. Moves live as tasks are pushed, parked or dropped.
*not associated with F1
Summary
Filters
Key
Target: complete by Wednesday, 19 August - which in practice means finished by the end of Tuesday, because Wednesday morning is being kept clear and the flight is Wednesday afternoon. The standard is functional software, not perfect software - everything testable is tested and recorded, and where Supreme's own equipment prevents a test, that is written down with the reason rather than skipped quietly.
One piece of sequencing matters more than the rest: tell Golliath the cable result today, whatever it says. If L1/L2 need new cable, the plant needs lead time - Monday is too late.
Both documents are the material. The session only has to land four things, and the first is the one that changes behaviour.
Demonstrate a roll change and a start/stop rather than describing them, and get the register signed in the room.
Everything below is the live version. Superseded copies in Drive are renamed SUPERSEDED so a stale link is obvious at a glance.
The report for Morné. Covers both visits, 28-31 July and 3-15 August. Opens on production's own words about oil life, then the per-line diagnosis, day-by-day bullets, what remains before departure, and what we need from Supreme. Days worked, no hour counts.
Word original: Supreme Chicken JHB / Orbit Filters - Progress Report - Aug 2026.docxWhat an operator does and, just as importantly, what is never an operator duty. §2A explains why the fryer level drops when the filter starts - the part that actually changes behaviour. v0.4 settles what actually causes an overflow: overfilling by hand while the filter is holding its 50-60 L, or water flashing to steam. Carries the 2022 original-issue date and a revision history.
Source: Supreme Chicken JHB / Training & SOP / OPERATOR_SOP_DRAFT.mdThe water purge as the critical procedure, daily readiness, per-line oil flow, the level-sensor wiring standard, and a written bellows-replacement procedure that did not exist anywhere before. Warns that L2 has no drain valves, so the purge and the interlock do not apply there yet.
Source: Supreme Chicken JHB / Training & SOP / MAINTENANCE_MANUAL_DRAFT.mdWhere the three documents above live. Anything added later lands here.
These are meant to be printed and carried, so they live as Word files rather than links. Say the word and they go up to Drive too.
Two pages, one per audience. Short enough to read standing at a running line, and to leave at the machine afterwards. This is what gets handed out Monday and Tuesday.
Supreme Chicken JHB / Training & SOP / Orbit Filter - Training Cards (operator + maintenance).docxThe running order for testing and the evidence at the same time. Eleven sections from screen sanity to the drain interlock, including the cable-tie high-pressure method. Has a box for what was not tested and why - a test skipped for a stated reason is worth as much as one that passed.
Supreme Chicken JHB / Analysis / DRY_TEST_RECORD_SHEET.docxThe operator SOPs issued with commissioning, located by Johan on 13 August. Retained by Orbit Filters and available on request - which is what makes the document-history section of the report a statement of fact rather than a defence.
Held by Johan SmitMy working area - the detail behind the plan rather than the plan itself. Addresses, register maps, per-line build order, the reasoning I need at the machine. It is here so it lives in one place and does not clutter the tabs Dad and Supreme actually read. Nothing here is a commitment; the commitments are on Deliverables.
The routine I am holding to for the rest of the trip. It is not a preference - a fixed sleep pattern is what keeps me sharp enough to be trusted around a live fryer, and it is why the plan is built around day-shift work.
Night shift: only where it is genuinely needed for training. The night crew has to see the new software in their own session, so those visits are real and planned - but they are the exception, not a habit. Every night shift attended costs the following day, so it has to buy something that cannot be bought in daylight.
✓Latest software pulled from each line's HMI and PLC, and backed up.
✓L2 built and downloaded - PLC and HMI, verified in the export.
▸Now: L2 dry-testing; L3 build to finish, then load and stress-test it.
○Sunday 16 Aug: L1 window - load, dry-cycle check, then the wet run.
○By end of Tuesday 18 Aug: all three loaded and tested, training done - Wednesday morning stays clear.
Do not read D620 / D630 / D650 / D690 / D700 / D710. They look like live values but they are the HMI display multiplex - D620 selects which parameter is showing and D630/D650 are reloaded from a different source in each phase. Telemetry reading those would log silently wrong numbers.
Worth doing before wiring: D10 and D12 are internal working registers, so renumbering anything during the rollout would break telemetry silently. A mirror block at D900+ - free in this program - is nine unconditional MOVs and gives a stable contract, identical on all three lines. Modbus base to confirm: Delta DVP normally maps D0 to 4096 and M0 to 2048.
L3 as it stands now, read out of its own program export rather than copied from L2 - the two are not the same machine. This is the set to hand over so the gateway can be proved on L3 while the L2 wiring goes in. Same warning as L2: do not read D620 / D630 / D650. L3 runs the identical HMI display multiplex (MOV C0 D630, MOV K1 D620), so those registers change meaning with the screen.
Hand over as: address, what it is, its scaling, and its update rate. The three that matter for a first proving run are D12, D10 and D600 - if those move sensibly while the line runs a cycle, the path from PLC to gateway to Cape Town is proven and the rest is just more of the same. Modbus base still to confirm: D0 to 4096, M0 to 2048 on Delta DVP.
I put it to him straight - that Dad says an SOP was originally issued. First answer: "No." Then, asked whether it was before his time: "Might be before my time... by the time when I arrived here I was just a supervisor, not a foreman." I asked if the foreman might have received it and not passed it to him - "Yeah." Then twice, unprompted: "So tell him to resend it, 'cause we don't have it."
The SOP question is closed and it is not an OEM failure. Jeffrey was a supervisor when the machines went in, not the foreman, so the document never reached him - exactly Dad's archiving point. Nobody in that meeting was lying; the plant simply lost it in a handover.
It also changes the shape of Dad's email: resending is now answering a direct request from their maintenance manager, not us defending ourselves. Send the original with its date, then the new version on top of it.
He asked whether I had been through the GEA manual - I am on it now. That is the one that should show a better take-off position than the hole cut at install.
Reliability, not safety. Morné asked what the issue was and Jeffrey said "It was not working... It starts sometimes, sometimes." - one cycle fine, the next one spilling oil over the front of the filter as the chamber opened after the dry cycle. Phone support was not resolving it, so he asked for someone on site: "I said, Johan, please come down... Otherwise if you can't help us, it's better you take them out." He also said Sydney, the artisan Dad trained, had become scared to work on them - "sometimes it's just boom while it was running."
Read it as escalation language from before I arrived, not a live threat: nothing in the meeting pointed at removal, and it closed with a directive to commission Line 3 first. My read of the room is that the filters are here to stay and the whole subject is control of the safety around them.
His own read, before any of my feedback: "the problem or the issue that we sit with is, since they were installed, there was no proper SOPs, and people understand what they're doing. Am I correct in saying that?" then "Artisans and operators, they didn't have that. Do we have that now, or are we going to implement that now?" Jeffrey confirmed it - "Yeah, there was no that." So the premise was put to the room and agreed, not challenged. I said the drafts are already in motion.
Settled the next morning (see the 13 Aug note): Jeffrey was a supervisor when the machines went in, not the foreman, so the SOP never reached him - the plant lost it in a handover, which is Dad's archiving point exactly. He asked us to resend it. The new documents carry ORIGINAL ISSUE DATE and LATEST VERSION on the front page so this cannot be argued again.
Lily set out Clayton's position: he is comfortable with his operators changing the rolls and starting and stopping the oil cycles, but not with them handling blockages, faults or valves. Golliath backed it as the right arrangement - "that will be the safe way and it's actually the correct way" - with an artisan called to the machine for anything beyond normal running. Nobody in the room argued otherwise, so this is now settled with management rather than being Clayton's request alone.
I took them through both. Splatter is water left in the pipework after cleaning flashing to steam the moment it meets 180 °C oil - that one is solved and proven on Line 3, and the fix is a purge procedure that belongs to maintenance. Overflow is a different problem: the fryer's automatic top-up being bypassed. Jeffrey then raised the software side with Morné himself - after cleaning the filter will not allow a start until the water has been drained, and the screen will say why. Morné's question was simply whether that is really what it will do.
Line 3's works - Jeffrey confirmed it. Line 1 has a selector switch under the tank: left on automatic it tops up from the dirty-oil tank, but it is sitting on manual, so an operator fills by eye. Line 2's Bulldog fryer is hard-wired with no PLC and never received the automatic top-up Dad was told was being commissioned at install - a project that predates both Morné and Lily. Line 2 does already have the level sensor and two tanks, so a retrofit is realistic.
Because L3's top-up works, Morné's call was to start there, and I agreed. L3 becomes the first line commissioned, run and trained on; L1 and L2 follow once their top-ups are sorted. I still want to confirm L3's oil routing on the machine before its live run.
After Wednesday morning's spill during my own test I stopped rather than put anyone at risk. Until the top-up works, the purge procedure is in place and the training is done, I am staying hands-off the machines: software work and dry signal-only tests. I committed to Morné that the remaining software and telemetry tasks finish this weekend. Lily added the rule for when live testing resumes - Clayton close by for any in-line test in production, so if something goes wrong he is there rather than supervisors scattering.
Air supply is short across both plants - supplied at 6 bar originally, most machines now see 4-5, so it is worth Supreme checking the compressor and reticulation. The fryer-side shut-off needed for the purge sits on the operation side under the machine and should be moved so every maintenance valve is together on the filter side. I also asked for the GEA manual, to check whether there is a better take-off position than the one cut at install. On departure Morné gets the report: fittings to replace with metal, the air supply, and the outstanding mechanical items. Worth recording that this trip was the machines' first major service since installation, on all three lines. Morné closed by thanking us for the effort.
"Where you have the knowledge and the way you try to explain, it catches the attention of people to listen and understand. Your wording... it's unbelievable. You know, I've worked with a lot of contractors and what-what. You are just starting in the market, but it's like you have been in the field for many years. Keep it up."
"That's why I had to stay, and I'm going to even tell your father about it. It just blew my mind... technically when a person is saying things which have got sense, you just pick this, this, this. And the wording, the language, you just attract attention for people to listen."
He stayed behind to say it, unprompted, and he is the one who has to live with these machines after I leave. Coming from the head of maintenance - the person whose team carries the purge procedure and the wiring - that is the relationship the rest of this depends on.
Morné had asked Clayton and Selato about safety and about who is trained to operate the filters, ahead of the GM meeting. Behind that: the maintenance burn incident and the report Lily has to compile on how a repeat gets prevented. Clayton's problem is evidential - he has to be able to show that the person operating the machine was competent to. Plus a short follow-up chat with Jeffrey afterwards.
Splatter = water left after weekend cleaning flashing to steam in 180 °C+ oil (L3's problem - fixed and proven: L3 ran full production clean after a proper purge). Overflow = the fryer's auto top-up bypassed to a manual pipe (L1/L2): the filter draws ~50-60 L to fill its pipes at start, an eyeballed-full fryer then overflows on the return. Working top-ups (L3, L4) never overflow.
Operators keep what they have safely done since the machines went in: start/stop, roll change, recipe select. The water purge becomes a maintenance procedure (~5-10 min at the start of the week). Jeffrey's shape for it: one SOP for operators and a separate one for the maintenance side. Clayton's condition is that safety officer Lily has to be able to sign his operators off against that SOP - he cannot certify them on his own.
Jeffrey's ask, twice: after cleaning, production cannot start until the water drain has run - auto-detected, no honesty button. "The machine doesn't want to start - then they realise: we need to drain it."
In the follow-up chat Jeffrey offered his team for it - he has fixed it before. His interim answer: leave the manual valve open and put the selector on automatic, and it tops up on its own all shift. I put it to Clayton that our position to the GM would be that L1's filter should not run until that is restored; he did not take a view on the messaging - his two requirements are simply that the filter runs and that it is safe to use.
L3's drying-return path (tank vs fryer) - trace the actuators with Clayton, he wants to see it. L1's original routing (Jeffrey recalls filter-return → heated dirty tank → level-controlled top-up). Oil-flow documentation for the manual - Selato asked for exactly this.
Operators attend training too - Clayton schedules who, per shift; Jeffrey joins the training WhatsApp group. Clayton's numbers: filter off = oil lasts ~1.5 days; filter on = 3-4 days. "The oil saver needs to run. It's a must."
The recipe engine and its interlock (the machine refuses to start without a recipe selected), the drying intelligence, a restored water drain from filter button on the CIP screen, and a pressure drying lamp on the Production screen so the drying phase is visible while it runs.
It now watches the chamber pressure: drying can end early once the pressure has decayed, or hold a dwell at a set pressure, instead of always running the clock out. That is the piece that stops a wet, blocked chamber being opened.
Every edit was checked against the program export rung by rung, the program compiled clean, and it was run on a simulated PLC to prove the interlock fires. Then HMI first, PLC second, same order as the July download.
The live proving cycles. Those wait on L2's fryer top-up and on the training - Line 3 goes first for commissioning. The pressure threshold also still has to be read off the machine itself; it cannot be copied from another line.