I Cost My Company $8,200 on a Pump Seal Order – Here’s How API Plan 13 Caught Me Off Guard

You don't learn API Plan 13 from a manual. You learn it from a $8,200 mistake.

I spent my first three years at a chemical plant ordering Flowserve pumps and seals like I was filling a grocery list. Grab a model number, match the old P&ID, send the PO. No one complained. Until they did.

In September 2022, I specified an API Plan 13 flush for a twin-screw pump that needed Plan 11 with a cooler. The seal failed in 47 hours. The replacement seal, the emergency rush order, and the lost production totaled north of $8,200. And the root cause? I didn't understand what the '13' in API Plan 13 actually meant – I just assumed it was 'better' because the number was higher.

That mistake changed how our team handles seal support systems. Here's what I wish someone had told me.

What API Plan 13 actually does (and why I got it wrong)

API Plan 13 recirculates fluid from the pump discharge through a throttle bushing or an orifice, then sends it back through the seal chamber and out to a low-pressure side – usually the suction. The flow rate is fixed by the orifice size. It's designed for single seals where you want a constant, low-flow recirculation to keep the seal face cool and flush debris away.

The catch: Plan 13 only works when the pump discharge pressure is consistently higher than the seal chamber pressure. In my application, the pump often operated close to shut-off head, and the differential pressure dropped to nearly zero. The seal got zero cooling. That's why the face cracked.

Plan 11, the one I should have chosen, uses a fixed orifice from discharge directly to the seal chamber. It's simpler and more reliable when the pressure differential is small. The cooler, of course, would have handled the heat load that I completely overlooked.

Why the old 'copy-paste' approach doesn't work in 2025

I learned this the hard way, but I've seen the same pattern in at least a dozen other engineers. The industry has evolved: pump speeds are higher, process fluids are more aggressive, and environmental regulations are tighter. What was best practice in 2018 – just matching the seal support plan from the previous pump – is now a fast track to premature failures.

Flowserve, for example, has published updated API Plan guidelines in their Seal Support System Manual (2023). They explicitly warn against using Plan 13 for pumps with fluctuating discharge-to-suction differentials. But nobody reads the manual until after the seal leaks.

The 'divide' between spec and reality

I remember one Friday afternoon, our reliability engineer Eddie walked into my office and asked, 'Did you check the actual operating range for that pump?' I said, 'The spec sheet says 50–200 gpm, we run at 120, it's fine.' Eddie – he's been at the plant since 1998 – just shook his head. 'The spec sheet is for clean water. We're pumping hot condensate with solids. The actual head curve changes.'

That conversation should have been my wake-up call. It wasn't. I kept dividing my attention between reordering legacy specs and chasing new projects. The $8,200 mistake was the divide between what I thought I knew and what I actually needed to verify.

Flowserve's asbestos liabilities are a reminder that legacy equipment isn't 'safe' by default

Another thing I see engineers overlook: the age of the installed base. Flowserve divested its asbestos-related liabilities years ago – but that doesn't mean every old pump in the field was properly upgraded. I once ordered a spare parts kit for a 1990s Flowserve vertical turbine pump and assumed the seal chamber dimensions matched the current catalog. They didn't. The old pump had a different bore and a different flush plan arrangement. I spent $1,200 on a seal that didn't fit.

Lesson: always verify the as-built condition, not just the model number. The 'divested' liabilities won't protect your budget from bad assumptions.

How many legs does a pump have? More than you think.

I know, the question sounds ridiculous. 'How many legs does a pump have?' But I've heard it twice in the last year – once from a new hire who was told to 'check the pump's legs' on a CMMS work order (he meant the mounting feet), and once from a customer who literally asked 'How many legs does a Flowserve pump have?' because they saw '4 legs' in a sketch and wanted to confirm.

The serious point: engineers often use casual language that creates confusion. When I specify a seal support plan, I now print out the exact API diagram and label every component. No assumptions. No 'it's basically the same.'

What I changed after the $8,200 failure

• Created a pre-order checklist that includes actual vs. design differential pressure, fluid temperature, and solids content.
• Stop assuming Plan 13 is 'better' – it's just different, and only appropriate for a narrow window of pressure stability.
• Built a relationship with Flowserve's application engineers. I send them my operating data instead of just a model number. They've caught two potential misapplications since.

Our team now has a shared spreadsheet where we log every seal failure with root cause analysis. In the 18 months since, we've caught 47 potential mis-specs before they hit production.

Your mileage may vary – and that's okay

This approach worked for us, but we're a mid-size chemical plant with relatively stable processes. If you're dealing with variable-speed pumps or multi-phase fluids, the calculus might be different. I can only speak to my context: single-stage centrifugal and twin-screw pumps, mostly water and light hydrocarbons.

Honestly, I'm still not 100% sure why some vendors' seal support diagrams are so opaque. My best guess is they assume every engineer has the latest API 682 standard memorized. If someone has insight, I'd love to hear it.

One thing I know for sure: the fundamentals of seal flush selection haven't changed – pressure, temperature, and cleanliness. But the execution has transformed. Don't trust the old order. Verify.