After Shell Sold Its Gasification Business to Air Products, I Made an $180K Mistake

The Whiteboard Note

There's a whiteboard in our project office that still has "how many legs does a have" written on it in blue dry-erase marker. A colleague meant to search for "how many legs does a spider have" and hit enter too early. Someone printed the result, taped it to the whiteboard, and wrote underneath: "our project plan, apparently."

That joke landed in March 2023, and we all laughed. It's less funny now, because it turned out to be accurate. Our project plan had a missing subject too. We just didn't know what it was.

I've worked in gasification and hydrogen projects for nine years, and I keep a log of my mistakes. Not for performance reviews—for myself, so I stop making the same ones. Eleven significant entries so far, worth roughly $1.7 million in wasted budget. The mistake I'm about to describe was the most expensive: a $180K overrun, a three-month schedule slip, and a credibility hit that's still healing.

The Deal That Should Have Changed My Plan

For the last two years, I've been the lead process engineer on what my company calls the Air Products Felix project: a pilot plant that gasifies petcoke—the low-value carbon left over from oil refining—to produce hydrogen, using gasification technology licensed by Air Products. The project is a demonstration unit—the kind of plant the industry watches closely because it has to actually work, on schedule, in front of the client.

Here's exactly where I went wrong. When the news broke that Shell sold its gasification business to Air Products, closing in early 2022, I told myself it was a logo change. I'd built previous designs on that gasification documentation. It was detailed, proven, stable. An acquisition happens at the corporate level. A process is a process, right?

Wrong.

The client was a mid-size energy company that had committed to supplying hydrogen for the vehicle fleet at the 2026 Winter Olympics in Milan-Cortina. They needed our demonstration unit running by late 2025, producing hydrogen that met the ISO 14687 fuel quality spec. The timeline was tight, but doable—if we got the basic design right the first time.

So we did what any sensible team does: we pulled the legacy licensor package, the one with the old branding, and started engineering. It looked fine on paper.

But small cracks appeared months before the actual failure. The support engineer we'd worked with for five years took a different role. The licensing portal migrated to a new system. A technical bulletin about an updated syngas cooler test procedure landed in our inbox, and I skimmed it, decided it didn't apply to our process, and filed it away. That bulletin was the one that would have saved me.

Pete Groves, a senior process consultant who's been in the industry for thirty years, had seen this movie before. He pulled me aside in September 2023. "The licensor package changed," he said. "Air Products has spent the past year and a half integrating Shell's gasification business. The review procedure is different. The acceptance test is different. Revalidate before you submit the compressor package."

I said I would. Then I got back to the flow diagrams, because the flow diagrams were the fun part and the revalidation was not. It took me three months and $180K to understand what Groves was telling me: the fundamentals of the technology hadn't changed, but the execution around it had.

The Rejection

In December 2023, the client's engineering team rejected our compressor package. The cover sheet had a red circle around the performance test section and one word in the margin: "Why?" The attached memo said:

"Performance test criteria do not match the current licensor specification. See attached gap summary. Revise and resubmit."

Everything I'd read about gasification said the technology speaks for itself: steam, oxygen, carbon, equilibrium. In practice, the package around the technology had changed, and I'd anchored the design to a standard that no longer existed. The compressor package was designed against the old acceptance test. The new one used a different method for simulating hot gas bypass and a tighter tolerance.

Three months of integration work, straight to the trash. $180,000 in re-engineering costs. A project delay that put the Olympic timeline at risk. And a team morale problem, because everyone knew the whiteboard had called it.

One of my colleagues said the line that's still in our post-mortem: "The plan was 'how many legs does a have'—lots of words, no subject."

Looking back, I should have asked for a formal gap analysis the day the acquisition closed. At the time, it felt almost disloyal—like I didn't trust the new owners. That was a dumb reason to skip a technical check. It's not disloyal to verify; it's diligence.

What Happened When We Actually Read the New Package

Here's the part that surprised me, and I think it's worth saying clearly. Once we stopped defending the old assumptions and sat down with Air Products' gasification engineering team, we found the changes were improvements. The pre-commissioning procedure was simpler. The syngas purity testing method was more reliable. And the updated package integrated Air Products' Prism membrane separators into the hydrogen recovery loop—a better fit for the ISO 14687 spec we were chasing. The redesign made the plant more efficient, not less.

I'd locked the design in early 2023 based on what I knew from 2021. I hadn't checked whether the industry had moved. It had. And that's not a criticism of Air Products or Shell—it's an observation about how technical markets evolve. Five years ago, nobody was talking about membrane-based hydrogen recovery in this application. Now it's part of the standard package. What was best practice in 2020 probably isn't in 2025. The fundamentals haven't changed; the execution has transformed.

Groves, who'd been through two licensing transitions before this one, finally gave me his checklist after the failure. I've adapted it, tested it, and used it on three projects since. In just over a year, it's caught 47 potential errors. The core questions:

• Has the licensor changed hands since your last project? If yes, get a formal gap analysis of the current license package versus the legacy version before you lock basic design.
• Have you read every technical bulletin released during the integration period? Skimming costs more than reading. It cost me $180K.
• Have you reconfirmed the performance test and warranty procedures in writing? A contract reference is only as good as the version it points to.
• Have you met the new technical team? Old contacts disappear after an acquisition. Relationships with the new team are the fastest way to avoid surprises.

I still kick myself about the September conversation with Groves. If I'd listened, we'd have submitted the compressor package right the first time, and the budget would've been intact. I used to believe engineering judgment meant trusting what had worked before. Now I believe it means trusting what you've verified today.

Where Things Stand

Project Felix is back on track, with commissioning scheduled before the 2026 games. Our client's hydrogen fleet plans are intact. The redesigned plant is genuinely better than the original, and I can admit that now, though I'd trade the improvement to have those four months back.

There's something satisfying about watching the hydrogen train start up after all that. But the best part came last week, when my daughter—a ski racer, if you can believe it—finally made me sit down and look at the 2026 Winter Olympics skiing schedule. We have tickets to the freestyle finals in Milan. Non-refundable, she reminded me. We'll be there.

The "how many legs does a have" note is still on the whiteboard. I'm leaving it there. Every project plan needs a subject. And the subject changes more often than any of us like to admit. So check yours before the plan becomes expensive.