On a Tuesday morning in March 2023, I stood in the receiving bay of our Cikarang facility in West Java with a digital caliper in one hand and a supplier inspection report in the other. The report said the crossover subs for an upcoming completion string were "within API tolerance." The caliper disagreed.
Not by much. Two of twelve thread connections measured 0.003 inches over the high limit on pitch diameter. Small enough that a tired inspector could have signed the sheet and moved on. Big enough that I didn't.
That decision delayed a drilling program by eleven days and, by our internal estimate, added about $47,000 in standby and rescheduling costs. It also produced the 14-point receiving checklist our team now runs on every completion component, at every location we operate. I've come to think of that as a $47,000 education — expensive, but cheaper than the alternative.
How the batch got to my bay
The parts came from a regional machine shop we'd used for three years. Their paperwork was clean: API Spec Q1 certificate, material test reports, dimensional inspection sheets signed by their QC lead. On paper, everything matched the drawings we sent.
Here's what took me longer to understand than I'd like to admit: an API Spec Q1 certificate tells you that a supplier's quality management system has been audited. It does not tell you that the specific parts in this specific crate match the specific revision of the drawing you approved on this specific purchase order. Those are two separate claims. For most of my first four years in this role, I treated them as one.
A certificate is evidence of a process. It is not evidence of a part.
I didn't fully get that until the caliper readings started coming back outside the band.
The pressure to approve
The rig schedule was tight. The completion string was slotted for a well in South Sumatra with bottomhole temperatures north of 300°F and expected pressures that put us squarely in the high-pressure, high-temperature envelope. The supplier's rep was on the phone within an hour of my first email, and his argument was reasonable: two connections slightly over on one dimension, within the broader "industry practice" band, no functional risk.
I almost went with it. That's the honest version. My project manager had already started talking about "conditional release" and "monitor in service," and I remember thinking that if I rejected the batch, I'd be the person who delayed a multi-million-dollar program over three-thousandths of an inch.
Then I ran magnetic particle inspection on the two suspect connections. Both showed linear indications at the thread root. Not huge. But present, and not on the supplier's inspection report.
The supplier's story changed after that. Their threading tool had drifted, they said, and their final inspection had sampled only the first and last connections on the pallet. They offered to rework or replace, at their cost. I declined the rework and asked for a full replacement batch with 100% MPI on every connection. That's when the eleven-day delay became real.
What nobody warns you about
The hardest part wasn't the technical call. It was the conversations. Two of them, specifically. The first was with our drilling engineer, who wanted to know why I was "adding scope" to a receiving inspection. The second was with the supplier's regional manager, who suggested — politely — that I was interpreting API 7-1 more strictly than the rest of the industry.
Maybe I was. But the well we were completing wasn't a standard well, and the consequence of a thread failure at temperature isn't a reprint or a refund. It's a trip out of the hole, a fishing job, and, depending on the failure mode, a well control event. I'd rather be the person who over-rejected a batch than the person who signed off on a failure that put a crew at risk.
We shipped the replacement batch on time for the revised schedule. The well was completed without incident. But that's not the part that stuck with me.
The protocol that came out of it
After the batch cleared, I sat down with our two receiving inspectors and wrote a checklist. Fourteen points, built around the specific failure modes we'd either seen or could reasonably anticipate on completion components: thread pitch diameter, root profile, taper, coating thickness at the sealing surface, hardness verification against the material test report, traceability of heat numbers, and a handful of documentation checks that take ninety seconds each.
We've refined it since. The version we run now includes a step where the receiving inspector photographs every connection under raking light and uploads it to the supplier's file. That one addition came from a later near-miss in 2024 on a different product line, and it's caught three supplier errors since.
The numbers, for anyone who cares about the business case: in the twelve months after we implemented the protocol, our completion-component reject rate at receiving went from roughly 8% to 2.5%. That's not because suppliers got better. It's because we started catching problems before they left the crate. Receiving time per batch went up by about 45 minutes. Rework and schedule impact went down enough to pay for that time several times over.
Five minutes of verification really does beat five days of correction. Simple. Not easy, but simple.
What I'd tell a new quality manager
Three things, and I'll keep them short because the story above already makes the argument.
First, separate the certificate question from the part question. Every time. A supplier can have a perfect quality system and still ship you a nonconforming component, because systems produce parts and people inspect parts.
Second, document the decision, not just the outcome. My conditional-release memo from that March is still in my files. I keep it because reading it reminds me how close I came to signing off.
Third, build the checklist before the incident, not after. I got lucky — the failure I was worried about didn't happen on my watch. Most people I know in this role who built the checklist afterward can tell you exactly what the incident cost.
One clarification, because the name causes confusion: the Schlumberger discussed here is the oilfield services company (now SLB). It is not related to the jewelry designer Jean Schlumberger, whose work — including pieces like a 16-stone ring — shows up in a very different search context. Different family branches, different industries, same famous name. If you arrived looking for a piece of jewelry, this is the wrong page. If you arrived looking for a receiving inspection protocol for completion components, the protocol is described above.
Note: API Spec Q1 and API 7-1 requirements referenced here reflect our internal interpretation as of 2024. Verify the current edition and scope at api.org before applying them to your own operations. Cost figures are internal estimates from a single project, not industry averages.
Continue The Conversation
If this topic connects to an active project or a planned technology transition, use the inquiry form below and our team will route the discussion to the right engineering contact.
Share the operating context behind your power requirement
Tell us about your site profile, control priorities, and energy transition targets so our team can respond with a more relevant configuration path.