Nobody thinks about cables when they're working. The wire inside a blood pressure cuff. The charger for a Verizon 2660 Flip. If you just bought that flip phone and you're looking up how to turn it on, the cable in the box is the last thing on your mind. Nobody thinks about cables. When a small one dies, you replace it and move on.
A 35kV feeder cable doesn't get that luxury. When a medium-voltage cable fails, it's not a trip to the store. It's a root-cause investigation. Contract clauses. Insurance claims. Schedules slipping by weeks. I'm a quality engineer at Prysmian, reviewing roughly 200 medium-voltage cable orders a year. In 2024, I rejected about 12% of first deliveries for spec mismatches or documentation gaps. After four years in this role, I can tell you: a lot of what gets called a “cable failure” is anything but.
The Surface Problem: Everyone Blames the Cable
A feeder trips during commissioning. A termination flashes over. The field test fails. The first reaction is almost always the same: bad cable. Get the manufacturer to replace it.
I've read that report. I've written the response to it. And here's the uncomfortable truth: the cable that left the factory passed every test we ran. Conductor resistance. Insulation thickness. Partial discharge. Full voltage withstand. It was still in spec when it hit your site.
So why did it fail? Let me walk you through three root causes I see on a regular basis—none of them involve a defective cable leaving the factory.
The “35kV” Spec Mistake Nobody Notices
People treat “35kV cable” like it's a voltage limit. 35,000 volts max, and beyond that it burns. Not quite.
The 35kV rating is the phase-to-phase voltage of the system the cable is designed for. Per ICEA S-93-639—the standard for 5 to 46 kV shielded power cables—a 35kV cable comes in different insulation levels. The most common is the 100% level, with a phase-to-ground rating of 21.1 kV. In normal operation, the insulation actually sees phase-to-ground voltage, not the 35kV nameplate number.
Put another way: “35kV cable” is a system designation, not a material property.
The spec mistakes I see come from treating it like a material property. Someone orders “35kV cable” for a system that isn't effectively grounded, where a ground fault can push phase-to-ground voltage toward the full phase-to-phase voltage before the fault is cleared. That's when a 100% level cable gets stressed beyond its design. In a Q1 2024 audit of customer specs, I flagged nine out of roughly fifty orders where the cable's insulation level didn't match the system's grounding design. The cable wasn't the problem. The system design was.
Accessories Fail. Cables Rarely Do.
This is the one that surprises almost everyone I talk to: the cable usually isn't the weak point. The termination is. Or the joint. Or the splice.
Partial discharge in medium-voltage systems almost always starts at an accessory—at a stress cone trimmed unevenly, a semicon layer not cut back exactly per the drawing, or a contaminated interface between insulation and termination. IEEE 48 requires that terminations be tested as part of a complete cable system, because that's how they work: as a system.
Here's the thing: factory testing proves the cable. It proves the accessory design—but only if the accessory was assembled correctly. The cable leaves our plant tested and certified. The termination is assembled in a bucket truck or a manhole by a crew working against the clock. Not ideal, but workable. The problem is when workmanship shortcuts get baked into the routine.
We've had customers ship “failed” cables back for analysis. When we dissect them, the failure point is almost always in the first few inches past the termination. That's not a manufacturing defect. That's an installation issue.
I'm not an installation supervisor, so I can't speak to every crew's training. What I can tell you from the manufacturing side: if the cable passed factory test but fails in the field, inspect the accessories before blaming the cable.
The Testing Gap Between Factory and Field
Let me walk through what quality control actually looks like.
At Prysmian's Scottsville, TX facility, every 35kV cable order gets tested before it ships. Per ICEA S-93-639, that includes conductor resistance, insulation thickness, eccentricity, partial discharge, and a voltage withstand test. Every reel. Every order.
Then the cable ships. It gets pulled through conduits. Bent around corners. Possibly dragged across sharp edges. Terminated. Spliced. And tested again on-site, usually with a VLF or resonant test per IEEE 400 at 2 to 3 times U0 for 15 to 30 minutes. That field test stresses the whole system: cable, terminations, joints, and the quality of the installation workmanship. (Should mention: we've seen cables damaged by pulls that exceeded the maximum allowable tension—the pull line looked fine, but the cable inside the conduit was stretched beyond what the spec allowed.)
The factory test and the field test are testing different things. The factory test verifies the cable as manufactured, on test reels, in a controlled environment. The field test verifies the cable as installed, after everything construction threw at it. When the field test fails, it's easy to blame the manufacturer. But the field test tests the whole installation. No factory test could have caught a cable damaged during installation.
The Real Cost of Misdiagnosing a “Cable Problem”
Here's what I see happen when a 35kV cable “fails” but the real cause is never identified.
The schedule slips. A feeder replacement that should take days becomes weeks of testing, finger-pointing, and re-engineering. Meanwhile, the crew waits. The substation stays offline.
The costs compound. If I remember correctly, there was a project in late 2023 where the same feeder cable was ordered three times because the spec kept changing. Every replacement required different termination kits and new field tests. The rework ended up costing about three times the original cable order. And the original cable was never the problem.
The underlying issue never gets fixed. If the root cause is a grounding design error or an installation practice, the next feeder will fail the same way. And the one after that.
I reviewed our returns data in early 2024—roughly 200 items analyzed over twelve months. About one in three “defective” cables that customers sent back showed no manufacturing defect at all. They were damaged during installation or failed at an accessory.
Let me say that again, because it matters: about 35% of the “bad cables” we received had no defect at the point of manufacture. The problem wasn't the cable. It was the cable system not being treated as one.
What to Do Instead (the Short Version)
I've spent most of this article on the problem. Here's the solution, and it's not “buy better cable.”
- Define the system, not just the voltage. Grounding method. Available fault current. Fault clearing time. Share these with your cable supplier before finalizing the spec. It costs nothing, and it prevents the most expensive mistake in medium-voltage procurement.
- Specify accessories with the cable. Termination and joint kits from the same manufacturer as the cable are designed as a system. Mixing brands can work, but it transfers the compatibility risk to you.
- Review factory test reports. Don't assume the cable was tested—ask for the reports and read them. A manufacturer that won't share full test data is a red flag, regardless of brand.
- Watch the first termination. The most dangerous step in any medium-voltage installation is the first accessory assembly. Have a qualified supervisor on site. Check the pull tension. Verify the bend radius. The first joint tells you a lot about how the rest of the job will go.
- Do root-cause analysis before reordering cable. Cut the cable back. Examine the failure point. If it failed at a termination, that's a different problem than if it failed mid-span. Ordering new cable doesn't fix a termination issue.
I'll add one more thing. People think premium cable costs more because the materials cost more—copper, insulation, jackets. Actually, the material difference is marginal. What you're really paying for is testing, traceability, and documentation. That's the expensive part.
Full disclosure: I work for Prysmian, and I'm biased. I believe the Prysmian 35kV cable we make at our Scottsville, TX plant—and across our other North American facilities—is thoroughly tested and documented before it ships. That's my job, and I take it seriously.
But that's not the point of this article. The point is: an informed customer asks better questions and gets better outcomes. If you understand the difference between a cable problem and a system problem, you'll save yourself time, money, and a lot of finger-pointing.
Next time a 35kV cable “fails,” ask one question first: where exactly did it fail?
The answer will tell you more than any test report.