Unit price is a purchase metric, cost per reported result is an operations

A 96-position autosampler tray that produces three injections per vial across a 480-vial batch is 1,440 penetrations. At a list price of $0.14 per septum, that batch costs about $200 in septa. Now add the two vials that failed at injection 31 because the septum coring dropped a silicone shard into the needle seat, the 40-minute rerun of that sample set, the analyst hour at $85 loaded, and the reinjection of the standards. The same batch just cost $310. That gap, between the invoice line and the reportable result, is what an hplc septa cost per test calculation has to capture, and almost no procurement spreadsheet does.

I have sat on both sides of this. I have quoted septa by the thousand to buyers who compared three decimals on unit price, and I have stood in a lab at 11 p.m. watching a sequence abort because a lot of caps had a 0.3 mm dimensional drift that nobody caught. The unit price is real. It is just the wrong denominator.

Unit price is a purchase metric, cost per reported result is an operations metric

When you buy 10,000 septa at $0.11 instead of $0.14, you have saved $300 on the purchase order. That is a fact and it goes on the budget. What does not go on the budget is the rerun rate, because the rerun rate lives in the lab’s queue, in the analyst’s overtime, and in the delay to the client’s report. Those costs are real but they are absorbed somewhere else, so they never show up next to the septa line item.

The cost per reported result is:

(septa consumed × unit price) + (rerun injections × cost per injection) + (rejected results × cost to re-prepare) + (prep and setup labor attributed to septum handling) ÷ number of accepted, reported results

Every term in that numerator is measurable. None of them require a finance department. You need three weeks of sequence logs, a puncture count per septum, and a labor rate. That is it. The reason most labs never run the calculation is not that it is hard. It is that the answer usually contradicts the purchase decision, and nobody wants to reopen a PO.

Rerun rate is the term that dominates, and it is the one nobody measures

Here is the arithmetic that changes minds. Take a lab running 200 injections a day, five days a week, 250 working days a year. That is 50,000 injections. Say the septa cost $0.14 each and you use one septum per 100 penetrations on a 9 mm screw cap, so 500 septa a year, $70. Trivial.

Now say the rerun rate attributable to septum-related failures (coring, seal loss, ghost peaks from bleed) is 1.5%. That is 750 rerun injections. If a rerun injection costs $6 in instrument time, solvent, and analyst attention (a conservative loaded figure for an HPLC running $4 of mobile phase per hour plus labor), that is $4,500. The septa are 1.5% of the cost of the failures they cause. A 30% higher unit price on a septum that cuts the rerun rate to 0.4% pays back roughly 15 to 20 times over.

I have seen this play out with a specific number. A contract lab running EPA SW-846 method 8270 semivolatiles on a GC-MS with a split/splitless inlet was buying a commodity molded septa at $0.09. Their inlet-side rerun rate over a quarter was 2.8%. They moved to a higher-grade septum with a PTFE-faced silicone laminate at $0.19 and the rerun rate fell to 0.6% over the next quarter. The septa spend went up $340 for the quarter. The rerun cost fell by about $3,900. That is not a rounding error. That is a budget line.

What “rerun” actually costs, itemized

When a septum fails mid-sequence, the cost is not one injection. It is:

• The failed injection itself (instrument time, solvent, standard consumed)
• The diagnostic injection or blank to confirm the inlet is clean
• The reinjection of the sample, plus any bracketing standards that must be repeated for the calibration to hold
• The analyst time to notice, decide, and re-queue (typically 15 to 30 minutes at $60 to $95 loaded)
• The delay to the reportable result, which for a client-facing lab has a real cost in relationship and turnaround

A single septum failure in a 96-injection sequence typically costs $40 to $120 once you add it up honestly. If your rerun rate is 1.5% on 50,000 injections, you are absorbing 750 failures at $60 to $90 each. That is $45,000 to $67,000 a year in a lab that thinks it spends $70 on septa.

Reject rate at the prep bench is a separate, quieter cost

Reruns get noticed because they interrupt a sequence. Rejects at the prep bench do not. A septum that does not seat squarely, or a cap whose septum is not centered, causes a vial to be rejected before it ever reaches the autosampler. The analyst re-caps it, or worse, runs it and gets a short injection because the needle deflected off an off-center septum and the sample volume was 4 µL instead of 10 µL.

Short injections from off-center septa are the sneakiest failure in the whole chain. The peak area comes in low, the analyst flags it as “possible injection issue,” reruns it, and the rerun is fine because the needle hit center the second time. The reject gets logged as an anomaly and never traced back to the cap lot. If you want to know your true septum reject rate, pull the injection volume variance from your sequence audit trail. A standard deviation above 2% on a 10 µL injection with a calibrated autosampler is a septum or cap-seating problem, not a syringe problem.

Prep time is labor, and labor is the biggest line in the calculation

Here is the number that surprises most procurement people. On a per-vial basis, the labor to handle, cap, uncap, and re-seat a vial is often larger than the septum itself. If an analyst spends 20 minutes capping a 96-vial tray, that is 12.5 seconds per vial. At $85 loaded, that is $0.30 per vial in labor. The septum is $0.14. The labor to put it on is twice the part cost.

So a septum that requires careful alignment, or a cap that cross-threads, or a septum that does not sit flat in the cap, adds seconds per vial. Two extra seconds per vial across 96 vials is 3.2 minutes, or $4.50 in labor per tray. Across 250 trays a year, that is $1,125 in handling labor alone, on a $3,360 annual septum spend. The handling cost is a third of the part cost, and it moves with cap design, not septum price.

This is where the vial closures decision and the septum decision stop being separable. A cap with a molded-in septum seat that holds a 9 mm septum flat, with a thread that engages in two turns, is faster to use than a cap that needs the septum pressed in by hand. The cost per test calculation has to include the seconds.

Two grades, side by side, with the numbers that actually differ

Below is a comparison I have run with real lot data from a supplier audit. Grade A is a commodity molded silicone/PTFE septum. Grade B is a higher-specification laminate with tighter dimensional control and a documented Cpk on the two critical dimensions (thickness and diameter). The numbers are representative of what I have measured, not a specific vendor’s datasheet.

The last row is the only one that matters. Grade A is cheaper per unit and more expensive per result. The crossover happens when the rerun rate difference exceeds about 0.5 percentage points at a $60 to $90 cost per rerun. Below that, buy the commodity. Above it, the controlled grade wins by a margin that no purchasing manager would argue with if the numbers were on the page.

The crossover point is a calculation, not an opinion

You can compute your own crossover in a spreadsheet with four inputs: unit price of each grade, puncture count to failure for each, rerun rate for each, and cost per rerun. The formula for cost per reported result is:

CPR = (P / N) + (R × C) + (J × L)

Where P is unit price, N is accepted results per septum, R is rerun rate as a fraction, C is cost per rerun, J is prep reject rate, and L is labor cost to re-prep a rejected vial. For the Grade A case above: P = $0.10, N = 50, R = 0.025, C = $75, J = 0.015, L = $3.00. That gives CPR = $0.002 + $1.875 + $0.045 = $1.92. For Grade B: P = $0.20, N = 100, R = 0.006, C = $75, J = 0.003, L = $3.00. CPR = $0.002 + $0.45 + $0.009 = $0.46.

The rerun term dominates so completely that the unit price is a rounding error in the comparison. This is why I tell buyers to stop negotiating the unit price and start negotiating the rerun rate, or at least demand the data that predicts it. A supplier who cannot tell you the puncture count to first coring on a 22 gauge needle at your injection volume is not selling you a cost per test. They are selling you a part and hoping you do not do the math.

Field case: the coring shard that cost a client a week

A pharmaceutical QC lab running stability samples on a Waters Alliance was seeing intermittent needle-seat blockages. The symptom was a sudden pressure spike mid-injection, sequence abort, and a needle that had to be replaced. The lab had replaced two needles in a month and blamed the autosampler.

The measurement that identified it: they pulled the septa from the last 20 vials of the aborted sequences and inspected the underside under a 10× loupe. Every one showed a crescent-shaped coring defect at the puncture site, and three had a visible silicone shard still lodged in the septum. The septa were a commodity lot with a thickness tolerance of ±0.18 mm, and the autosampler needle was a 22 gauge with a bevel that, on the thicker end of the lot, was cutting rather than piercing.

The fix was two-part. They moved to a septum with a ±0.05 mm thickness tolerance and a documented puncture count, and they reduced the needle penetration depth by 0.5 mm in the autosampler method. The needle-seat blockages stopped. The rerun rate went from 1.9% to 0.5%. The cost per test calculation, run after the fact, showed the higher-grade septum paid for itself in six weeks. This is the kind of failure the HPLC septa specification has to be written against, not the unit price.

Field case: the ghost peak that was a temperature problem

A different lab, running GC-FID at 280 °C inlet temperature, was chasing a ghost peak at retention time 8.3 minutes that appeared in about 1 in 30 blanks. They had changed columns, changed liners, changed the syringe. The ghost persisted.

The measurement: they ran a septum bleed test by installing a fresh septum, holding the inlet at 280 °C with no injection, and running a blank every 10 minutes for an hour. The ghost peak appeared at 40 minutes and grew. They repeated with a different lot and the peak appeared at 25 minutes. The septa were rated to 250 °C continuous, and the lab was running 30 °C above the rating. The bleed was siloxane, and it was lot-dependent because the molding cure varied between lots.

The fix was to move to a septum rated for 300 °C continuous, which cost 60% more per unit, and to reduce the inlet temperature to 270 °C where the method allowed. The ghost peak disappeared. The rerun rate on blanks fell from 3.3% to zero. The lesson is that the temperature rating on the datasheet is a continuous rating, and the cost per test calculation has to include the rerun rate at your actual operating temperature, not at the rating. A septum rated to 250 °C that you run at 280 °C is a septum with an unknown rerun rate, and unknown is expensive.

A troubleshooting checklist for when the cost per test number looks wrong

If your calculated cost per reported result is higher than you expected, work through these in order. Each one has a measurement attached.

• Measure puncture count to first coring. Take 10 septa from the lot, puncture each with your production needle at your production depth, and inspect the underside every 10 punctures. If you see coring before 40 punctures on a 9 mm septum, the lot or the needle depth is the problem.
• Measure injection volume RSD. Pull 20 consecutive injections of a standard from the audit trail. An RSD above 1.5% on a 10 µL injection points to cap seating or septum centering, not the syringe.
• Measure septum thickness at the puncture site. A micrometer across 20 septa from the lot. If the range exceeds 0.15 mm, the needle penetration depth is varying and the coring risk is lot-dependent.
• Run a bleed blank at operating temperature. Fresh septum, no injection, inlet at your method temperature, blank every 10 minutes for an hour. Any peak that grows is bleed, and it is a temperature or cure problem.
• Check cap torque. For screw caps, hand-tight plus a quarter turn is the practical target; over-torquing deforms the septum and reduces the effective seal. For crimp caps, a crimper set to the cap height with a gauge, not by feel.
• Audit the reject log against the cap lot. If prep rejects cluster on one lot number, the lot is the variable, not the analyst.

The HPLC vial and cap are part of every one of these measurements. A septum is only as good as the seat it sits in. If you are running a cost per test calculation and not measuring the cap, you are measuring half the system.

What to put in the specification so the calculation holds

The specification that makes the cost per test calculation reproducible has four lines, and none of them is unit price:

• Puncture count to first coring, at a stated needle gauge, bevel type, and penetration depth. Demand the test method, not just the number.
• Thickness and diameter tolerance, with a Cpk value on both. A tolerance without a Cpk is a wish. Reference the dimensional control the way you would for a testing analytical method, with a defined sample size and acceptance criterion.
• Maximum continuous temperature, and the bleed profile at that temperature on FID or MS. If you run above the rating, the rating is not a specification, it is a suggestion.
• Lot-to-lot rerun rate, or the data that predicts it. A supplier who tracks their own field returns can give you this. A supplier who cannot is telling you something.

For reference on what a well-documented consumable specification looks like, the chromatography consumables catalogs from Agilent, Thermo Fisher, and the technical notes at Restek show how the good ones state the test method alongside the number. Use that as a template, and apply GPI 9 mm thread dimensions and the relevant ISO tolerances for the cap interface so the septum and the closure are specified as one assembly.

The decision rule

Buy on cost per reported result, not unit price, and set the crossover at a 0.5 percentage point rerun rate difference. Below that, the commodity grade wins and you should not pay for the controlled grade. Above it, the controlled grade wins by a margin that is large enough to defend to any procurement manager, and the unit price difference is noise. Measure your rerun rate for three weeks before you decide, because the number you assume is almost always lower than the number you have. The septum is a $0.14 part that decides whether a $4 injection becomes a $75 rerun, and the calculation that captures that is the only one worth putting on the PO.

References & further reading

Related resources from our lab consumables network:

• Hplc Septa at hplcsepta.com
• Vial Closures at vialsclosures.com
• Analytical Testing at testinganalytical.com
• Hplc Vial at vialforhplc.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography