Buying perfusion and flow control hardware: shear stress is a property of the channel, not the pump

The buying problem here is a category error that costs laboratories five figures. Perfusion hardware looks like a ladder, rocker at the bottom, syringe pump in the middle, pressure controller at the top, and buyers place themselves on it according to budget. It is not a ladder. A rocker and a pressure controller do different jobs, and the question that decides which one you need is not how much flow you want but whether shear stress is an independent variable in your experiment or merely a culture condition.

If it is a culture condition, you want medium exchange and some movement across the cells, a rocker does it, and buying a pressure controller adds tubing, connectors, priming volume and three new failure modes to solve a problem you did not have. If it is an independent variable, you intend to write “cells were exposed to 10 dyn/cm² of unidirectional laminar shear stress” and defend it, then you need a pressure controller, a channel with a published shear relationship, and a budget in the tens of thousands. There is no useful middle.

The expensive mistake is buying the second thing and specifying it like the first: choosing a flow rate as though it were a dose. It is not. Shear stress is a property of the channel, and the same flow rate through two different channels produces wildly different numbers. The published data below shows how wildly.

What the hardware costs, where it is published

ibidi is the vendor in this category that publishes USD list prices per SKU. All four figures below were live-verified against ibidi.com on 2026-09-01. These are list prices, not derived figures and not quotes, and they are pre-tax and pre-discount.

The arithmetic that follows from those figures is the honest entry price for defined shear:

• Single channel: pump plus one Fluidic Unit is $23,985.00 before any consumables. Call it around $24,000.
• Four channels: pump plus a Fluidic Unit Quad is $33,800.00. Call it around $34,000.

Worth noticing: the Quad at $13,085.00 is five dollars more than four single Fluidic Units at $3,270.00 each. There is no volume discount inside the product line, so the Quad is a packaging and footprint decision rather than an economic one. Note also that the pump drives up to four Fluidic Units, so four channels is the ceiling on one pump and a fifth channel means a second pump.

MIMETAS publishes OrganoFlow specifications, programmable tilt from 0 to 25 degrees, rocking intervals from 5 seconds to 999 minutes, up to 16 OrganoPlates on the L variant, and a UniFlow option for unidirectional perfusion, and publishes no price and no shear stress figure on the pages we cite. We therefore do not state either. Peristaltic and syringe pump prices are also absent from this page: the cross-vendor range is too wide for one figure to be informative and we did not verify one.

The fifteen-fold problem

This is the load-bearing technical fact in the whole category, and it comes from ibidi’s own published Application Note 11 rather than from us.

Wall shear stress in a rectangular channel is calculated as:

τ = η × k × φ

where τ is the wall shear stress in dyn/cm², η is the dynamic viscosity of the medium, φ is the flow rate, and k is a coefficient specific to the channel geometry. ibidi’s published convention for the viscosity term is η = 0.0072 dyn·s/cm².

Here is what k does across a single product family:

Two consequences, both of which routinely wreck experiments.

The same 1 mL/min produces roughly fifteen times more shear in the 0.2 mm channel than in the 0.8 mm one. A protocol ported from one paper to another laboratory that used a different channel height, at the same flow rate, is not the same experiment. If you are trying to reproduce a published shear condition, the flow rate is useless information without the channel.

Swapping to a glass bottom for imaging changes the answer by 20 percent. On the 0.4 Luer, k moves from 131.6 on polymer to 104.7 on glass. That is a substantive change in the mechanical stimulus, caused by a decision made for optical reasons and usually not modelled as a biological one.

This is why a laboratory can spend $24,000 on a defined-shear system and still report a shear stress its cells never experienced. The pump is accurate to ±1 mbar. The error is upstream of the pump, in the specification.

The specification line that fixes it

For a catalogue channel, the coefficient is published and the problem is only one of reading it carefully. For a custom geometry, nobody has computed it, and this is where procurement has to do work that is usually left as homework.

Put this line in the RFQ, verbatim if you like: supply the wall shear stress as a function of flow rate for the as-built channel, and state the assumed medium viscosity.

That single sentence converts an ambiguity into a deliverable. It forces the fabricator to either run a computational fluid dynamics calculation on the geometry they are actually going to build, or to measure it on a first article, or to tell you they will not, which is itself useful, because then you know the derivation is your job and you can budget for it.

Two details make the difference between a useful deliverable and a decorative one:

• As-built, not as-designed. A channel moulded 15 percent shallower than the drawing has a different coefficient. If the derivation is against the CAD file rather than the part, it is a design document, not a calibration.
• State the viscosity assumption. Medium viscosity varies with temperature and serum concentration. ibidi’s published convention is η = 0.0072 dyn·s/cm²; if your medium is 10 percent serum at 37 °C and the derivation assumed water at 20 °C, the number is wrong in a direction nobody will notice.

Choosing between the routes

The strongest argument for the rocker route is not price, it is failure modes. A rocker has no tubing, no connectors and no priming step, which removes the three commonest causes of contamination, bubbles and unattended overnight failure. A perfusion rig that fails at 2am costs the culture, and the pressure-driven route accumulates more ways to do that than the tilting route does. If you do not need the number, do not buy the failure modes.

Recirculation is usually compulsory, and nobody says so

At 10 dyn/cm² in a typical channel geometry, single-pass perfusion consumes medium at a rate that makes an overnight experiment impractical. In practice, defined-shear work at physiological magnitudes is recirculating work, which is what the ibidi Fluidic Unit provides.

That has two consequences buyers do not price:

Metabolite accumulation. A recirculating loop is a closed system. Whatever the cells secrete stays in the medium, and whatever they consume depletes from it. Over a long experiment the medium composition drifts in a way that static culture with scheduled changes does not.

Reservoir volume becomes an experimental parameter. The µ-Slide I Luer holds 60 µL per reservoir. The total loop volume, not the channel volume, sets how quickly a secreted factor reaches a given concentration. If your endpoint is anything paracrine, that volume belongs in the methods section.

Bubbles are the failure mode

Every perfusion programme loses cultures to bubbles, and the mechanisms are documented rather than mysterious. Gas comes out of solution when medium warms, when pressure drops across a constriction, or when it is drawn through a connector that was not primed. Gas-permeable device materials both help and hurt: they let small bubbles dissolve away, and they let gas in.

The mitigations worth specifying at purchase rather than improvising later:

• Degas the medium before it enters the loop, or run an in-line degasser.
• Fit a bubble trap upstream of the channel. Retrofitting one into a rig that was specified without it usually means new tubing runs and a new priming protocol.
• Prime deliberately, with a written procedure, because priming errors are the largest single source of bubbles and the easiest to eliminate.

There is a measurement consequence too, if you are combining perfusion with barrier readouts. The Holzreuter and Segerink review of TEER in organ-chips notes that impedance measurements are sensitive to air bubbles in channels, and that in long channels with small cross-section the medium resistance can overshadow the biological signal. A bubble does not merely risk the culture; it corrupts the number before it kills anything.

Imaging integration, which is a hardware decision

If the perfused device has to sit on a microscope, three physical constraints decide whether the system works, and all three are cheaper to solve at purchase than afterwards:

1. Does the umbilical fit? Tubing has a bend radius, and a stage-top incubator has a lid. A rig that works on the bench and not on the stage is a common and avoidable outcome.
2. Bottom thickness and material. The µ-Slide I Luer is offered with a #1.5 polymer or #1.5H glass bottom, and as shown above, that choice changes the shear coefficient by 20 percent. Decide it once, with both consequences visible.
3. Where the pump lives. Pressure controllers are not usually incubator-rated, so the pressure line crosses the incubator boundary. Plan the port before you plan the experiment.

What to put in the RFQ

1. Is shear stress an independent variable, or a culture condition? Answer this first; it selects the route.
2. Target wall shear stress in dyn/cm², and the medium viscosity at your temperature and serum concentration.
3. Unidirectional or bidirectional. Gravity levelling is inherently bidirectional.
4. Number of channels perfused in parallel, and whether that exceeds four per pump.
5. Recirculating or single-pass, and the total loop volume if recirculating.
6. Channel geometry and bottom material, since both change the coefficient.
7. Whether the device is a catalogue channel slide, a plate-format chip, or a custom geometry needing its own derivation.
8. Imaging requirement, and whether the umbilical fits the stage or stage-top incubator.
9. Bubble mitigation: degassing, in-line degasser, bubble trap.
10. Whether the fabricator must deliver a CFD result or a measured flow-to-shear calibration with the part.

Line 10 is the one that is nearly always missing and nearly always the one that matters.

What this page does not establish

No claim is made here that perfused culture produces better biology than static culture; that depends entirely on the assay. No shear figure is asserted for any MIMETAS product, because none is published on the pages cited. The ibidi coefficients apply to ibidi geometries and to nothing else. And the prices are one vendor’s USD list on one date, a genuine anchor for what defined shear costs, and not a market survey.

How we can help

We are useful in two places. First, at specification: writing the shear-stress derivation into the RFQ as a deliverable, so the number you eventually publish is one somebody calculated for the part you actually own. Second, at the junction between catalogue and custom, when the geometry you need does not exist as a slide, and the question becomes whether a fabricator will bond, characterise and calibrate a channel to your dimensions.

The paired datasheet, with the full specification field list and supplier evidence notes, is at perfusion and flow control hardware. The route comparison, with the four perfusion methods set against each other, is at perfusion and flow control for cell culture. Send us the specification.