Infusion Pump Flow Accuracy Verification Project

This project produces a verification and release package for infusion pump flow accuracy. The goal is not only to show that the pump motor runs or that the display accepts a programmed rate. The goal is to prove that the complete pump, disposable set, fluid path, software configuration, calibration state and test method deliver the commanded volume within defined limits.

The example uses a volumetric infusion pump and saline test fluid. The workflow also applies to syringe pumps, ambulatory pumps, enteral pumps and service verification of installed pump fleets. Real medical-device work must use the applicable intended use, risk management file, manufacturer instructions, recognized test methods, clinical workflow, accessories, software version, and regulatory evidence.

Project objective

Verify flow accuracy for an infusion pump after a software-library update and preventive maintenance campaign. The final deliverable is a reviewable engineering package containing:

• intended-use boundary and pump configuration;
• disposable-set and fluid basis;
• gravimetric test method;
• rate-profile test matrix;
• startup delivery and bolus checks;
• uncertainty budget and guard-band rule;
• risk-control traceability;
• release decision and retest triggers.

This project is different from an occlusion-alarm review. Occlusion testing asks whether the pump detects a blocked downstream line fast enough. Flow-accuracy verification asks whether delivered volume matches the programmed delivery profile when the line is not occluded.

Verification basis

Use this simplified pump basis.

Item	Project value
Pump type	volumetric infusion pump
Test fluid	saline substitute
Fluid density at test temperature	\rho=0.998\ \text{g/mL}
Main continuous rate	5.0\ \text{mL/h}
Low-rate therapy profile	0.50\ \text{mL/h}
Bolus command	1.00\ \text{mL}
Main-rate test duration	2.0\ \text{h}
Low-rate test duration	4.0\ \text{h}
Flow accuracy requirement	\pm 5.0\% after guard band
Startup accuracy requirement	\pm 10.0\% over first 30\ \text{min} after guard band
Bolus accuracy requirement	\pm 5.0\% after guard band
Maximum uncontrolled post-test residual volume	0.10\ \text{mL}

The test uses a calibrated balance, a collection vessel, evaporation blank, stopwatch or synchronized data logger, controlled tubing height, specified backpressure condition, and the intended disposable set.

Acceptance criteria

Use these project acceptance criteria.

Requirement	Acceptance value
Main-rate delivery error	inside \pm5.0\% with uncertainty guard band
Low-rate delivery error	inside \pm5.0\% with uncertainty guard band
Startup delivery error	inside \pm10.0\% with uncertainty guard band
Bolus delivery error	inside \pm5.0\% with uncertainty guard band
Test method uncertainty	stated and included in pass/fail decision
Disposable set	traceable lot and configuration
Software and calibration state	recorded and locked for release
Usability and alarm state	no test bypass left active after verification
Release evidence	raw data, calculations, deviations and disposition complete

Acceptance should be defined before testing. Changing criteria after seeing results weakens the evidence and should trigger engineering review.

Step 1: Convert mass to delivered volume

The gravimetric method estimates delivered volume from collected mass:

where:

• m_{collected} is the measured net mass in the receiving vessel;
• m_{evap} is the evaporation correction from a blank vessel;
• \rho is fluid density at the test temperature.

For the main-rate test, the pump runs at:

for:

Expected volume:

Measured net mass:

Evaporation blank:

Corrected delivered volume:

Flow error:

Engineering comment

The measured flow is slightly low, but the raw result is well within the nominal \pm5\% requirement. The release decision still needs uncertainty and guard banding because a measurement near the limit could be falsely accepted.

Step 2: Check uncertainty and guard band

Use this simplified uncertainty budget for the main-rate test.

Contributor	Standard uncertainty
balance repeatability and calibration	0.004\ \text{mL}
evaporation correction	0.010\ \text{mL}
density correction	0.006\ \text{mL}
timing	0.002\ \text{mL}
test repeatability	0.015\ \text{mL}

Combined standard uncertainty:

Use coverage factor:

Expanded uncertainty:

Relative expanded uncertainty for the 10.0\ \text{mL} test:

Guarded acceptance band:

The measured main-rate error is:

Because:

the main-rate result passes the guarded decision rule.

Engineering comment

Guard banding prevents a result near the acceptance limit from being accepted without enough measurement confidence. The guard band should be defined by the verification plan, not invented after testing.

Step 3: Verify low-rate delivery

Low-rate therapies are often more sensitive to startup delay, mechanism backlash, disposable-set compliance and quantization of pump movement.

Programmed rate:

Test duration:

Expected volume:

Measured net mass:

Evaporation correction:

Delivered volume:

Error:

Assume the expanded uncertainty for this smaller delivered volume is:

Relative uncertainty:

Guarded acceptance band:

Because:

the low-rate result passes, but with less margin than the main-rate test.

Engineering comment

Low-rate verification usually has less measurement margin because the delivered volume is small. A longer test, lower evaporation setup, better balance, or repeated runs may be necessary when the result is near the guard band.

Step 4: Check startup delivery

Startup delivery is reviewed over the first 30\ \text{min} at:

Expected volume:

Measured net mass:

Evaporation correction:

Delivered startup volume:

Startup error:

Assume expanded uncertainty:

Relative uncertainty:

Guarded startup band:

The measured error:

does not pass the guarded startup criterion.

Engineering comment

This is an engineering finding, not a mathematical inconvenience. The pump may pass average flow after two hours while underdelivering during startup. The release package should either improve startup compensation, restrict the affected profile, extend priming or stabilization instructions, or justify why the startup behaviour is clinically acceptable for the intended use.

Step 5: Verify bolus delivery

Bolus command:

Measured net mass:

Evaporation correction:

Delivered bolus:

Bolus error:

Assume expanded uncertainty:

Relative uncertainty:

Guarded bolus band:

Because:

the bolus result passes.

Engineering comment

Bolus testing should confirm delivered volume, command logging, user confirmation, cancellation behaviour, dose limit enforcement and post-bolus return to the programmed basal rate. A bolus feature is both a flow-control function and a usability-critical workflow.

Step 6: Trace findings to risk controls

Use the verification results to update the risk-control table.

Hazardous situation	Evidence from this project	Control decision
underdelivery during low-rate therapy	low-rate test passes with limited guard-band margin	retain profile but monitor manufacturing and service variation
startup underdelivery	guarded startup test fails	require corrective action or profile restriction before release
unintended bolus error	bolus test passes	retain bolus function with dose-limit checks
wrong disposable set used	test is set-specific	label, lockout or compatibility controls must remain effective
occlusion alarm delay	not evaluated by flow test	rely on separate occlusion-alarm evidence
service release after software update	test tied to software version	release only for recorded software and calibration state

Engineering comment

A failed startup check does not necessarily mean the entire pump design is unsafe, but it does mean release is not complete. The engineering action must be explicit: fix, retest, restrict, justify with clinical validation, or hold release.

Verification matrix

The final package should include:

Evidence item	Required record
intended-use boundary	therapy profiles, patient population, clinical environment and disposable set
pump configuration	model, serial number, software version, calibration state and service history
test method	balance certificate, fluid density, evaporation blank, timing basis and setup geometry
main-rate raw data	mass readings, time stamps, calculated volume, error and uncertainty
low-rate raw data	mass readings, correction, guard-band decision and repeatability notes
startup test	first 30\ \text{min} result, failure disposition and retest plan
bolus test	command, collected mass, delivered volume, dose-limit confirmation
risk traceability	hazards, controls, verification evidence and residual decision
release decision	pass, conditional release, restriction, or hold

Release decision

The project result is mixed:

Test	Result
main-rate delivery	pass
low-rate delivery	pass with limited margin
bolus delivery	pass
startup delivery	fail under guarded criterion

A suitable release statement is:

The infusion pump is not released for unrestricted use under the tested software and disposable-set configuration because startup delivery at the reviewed therapy profile does not meet the guarded acceptance criterion. Main-rate, low-rate and bolus accuracy results support the measurement method, but release requires corrective action, retest evidence, or a documented restriction that prevents use of the affected startup profile.

This statement is useful because it prevents a common error: averaging good long-duration flow results and ignoring a clinically important short-duration startup deficiency.

Common project mistakes

Common mistakes include:

• verifying only one nominal flow rate;
• ignoring startup behaviour and small-volume delivery;
• reporting mass without density, evaporation and timing corrections;
• accepting results near the limit without an uncertainty rule;
• testing a disposable set that is not the intended clinical configuration;
• treating occlusion-alarm evidence as proof of delivery accuracy;
• failing to tie a software update to retest scope;
• releasing the device after a failed subtest without a documented restriction or corrective action.

The practical rule is that flow accuracy is a system property. Pump mechanism, software, disposable set, fluid path, accessories, calibration and test method all affect the delivered therapy.