Prototype Evidence Pack · UCL MECH0073 Group 22

Portable Ultraclean Airflow Canopy

A portable, H14-filtered device that delivers a uniform, downward sheet of ultraclean air directly over the surgical field — supplementing conventional operating-theatre ventilation.

13×
More Even Airflow
>1000×
Cleaner at the Edges
100%
Wound-Zone Coverage
10.4 kg
One-Person Portable
Explore the Design ↓ View Poster
The Problem

Why This Matters

Surgical site infection (SSI) affects 2–5% of inpatient surgical procedures in high-income countries (WHO 2018). Airborne bacteria-carrying skin squames shed by theatre staff are a key transmission route.

Room-level ventilation alone cannot guarantee clean conditions at the wound — lamps, drapes, and staff create local stagnant zones right where protection matters most.

Commercial mobile devices exist, but no study has compared internal flow-conditioning architectures on the same platform. That is our gap.

  • ◆
    Same-Platform ComparisonFirst controlled comparison of open-cavity vs conditioned-cavity on identical fan, filter, ducting, and outlet.
  • ■
    H14 HEPA FiltrationEN 1822-4, 99.995% efficiency at MPPS. Medical-grade filtration in a portable form factor.
  • ▲
    Experiment-DrivenAll data from measured results: testo 405i anemometer, Fluke 985 particle counter, Extech 407732 SLM.
  • ●
    Three Iterative PrototypesP1 (feasibility) → P2 (diagnostic baseline) → P3 (conditioned cavity). Each informed by measured gaps.
💡

In plain terms

Operating theatres already filter the whole room, but the air directly above the wound can still go stagnant where lamps, drapes and staff get in the way — exactly where cleanliness matters most. This device stands over the operating area and pushes a gentle, even “curtain” of medical-grade filtered air straight down onto it, keeping that spot cleaner than the room around it.

We built three prototypes to answer one question: what shape of airflow protects best? The key finding is that you don’t need a stronger fan — you need a more even flow. Our final design adds two simple plates inside the housing that spread the air out, turning a narrow central jet into a wide, uniform sheet with almost no extra weight or power.

▶ Key terms used on this page — tap to expand
Unidirectional ultraclean air
Medical-grade filtered air pushed in one steady downward direction, so contaminants are swept away from the wound instead of swirling back over it.
H14 HEPA filter
The top HEPA grade (EN 1822-4). Traps 99.995% of the hardest-to-catch particle size — the “MPPS” (Most Penetrating Particle Size).
CV (Coefficient of Variation)
How even the airflow speed is across the outlet. Lower is better — 0 would be perfectly uniform. P2 scored 0.92 (very uneven); P3 scored 0.07.
Protected zone
The area where the downward airflow stays at least 0.3 m/s — fast enough to hold back dirtier room air from drifting in.
β (open-area ratio)
The fraction of the perforated baffle plate that is actually holes (12.4% here). It sets how evenly the plate spreads the flow.
ISO Class 5
A point on the ISO 14644-1 cleanroom scale — the “ultraclean” cleanliness target used for surgery.
SR1–SR8
The eight measurable success requirements the design was judged against: airflow speed, filter grade, uniformity, noise, footprint, weight, and serviceability.
SSI (Surgical Site Infection)
An infection at the surgical wound. Airborne bacteria riding on shed skin flakes are one route — the route this device targets.
Spatial CV vs distance
Uniformity is measured at several heights below the outlet (5, 15, 30 cm…). Near-field (≤30 cm) is the primary evidence; further out is supporting trend.
Prototype 1

P1 — Feasibility

Prototype 1
Feasibility Prototype

Can it work at all?

Built to answer one question: can a portable, fan-driven, HEPA-filtered canopy produce measurable airflow?

P1 used a 5-inch mixed-flow hobby fan and an H13 filter — failing the H14 specification requirement (SR3). Noise at 60 dB(A) exceeded the SR7 ceiling. The concept was validated, but critical hardware limitations were identified.

H13
Filter grade (below H14 target)
60 dB(A)
Noise (above 53 dB(A) target)
PLA
Housing material (taped)
Proved concept
Outcome

P1 CAD & Assembly

Exploded CAD view
Physical assembly
P1 — tapered outlet section

P1 Test Results

Velocity heatmaps
Wind speed statistics
Velocity profiles
Prototype 2

P2 — Open Cavity (Diagnostic Baseline)

Prototype 2
Diagnostic Baseline

The central-plume problem

Upgraded to the ebm-papst D3G146-HQ13-34 centrifugal blower and SPCB Bespoke H14 filter (EN 1822-4). 3D-printed PLA housing designed in Fusion 360.

The open cavity produced a strong central jet with weak stagnant edges. At 5 cm below the outlet, spatial CV was 0.92 (highly non-uniform — fast in the middle, almost still at the edges). Edge particle counts at 30 cm reached 11,500 — nearly the same as untreated room air.

This central-plume failure directly motivated the P3 conditioned cavity.

H14
HEPA filter (EN 1822-4)
D3G146
ebm-papst EC blower
CV 0.92
Non-uniformity at 5 cm
66.1 dB(A)
Noise at 1 m
10.4 kg
Total mass
0.14 m²
Footprint

P2 CAD Views

Isometric view
Front view
Right view
Top view
P2 — full assembly render (shared hardware)

P2 Velocity Heatmaps — All 6 Distances

d = 5 cm — strong central jet
d = 15 cm
d = 30 cm
d = 60 cm
d = 100 cm
d = 150 cm

P2 Composite Analysis

6-panel heatmap composite
4-panel heatmap (near field)
Velocity profiles (6-panel)
Velocity distribution histograms
Active regions (≥0.3 m/s)
Turbulence intensity
▶ P2 Per-Distance Detailed Plots (profiles, histograms, active regions, turbulence)

d = 5 cm

Profiles
Histogram
Active region
Turbulence

d = 15 cm

Profiles
Histogram
Active region
Turbulence

d = 30 cm

Profiles
Histogram
Active region
Turbulence

d = 60 cm

Profiles
Histogram
Active region
Turbulence

d = 100 cm

Profiles
Histogram
Active region
Turbulence

d = 150 cm

Profiles
Histogram
Active region
Turbulence

P2 Cross-Distance Analysis

Centerline velocity decay
CV vs distance
Peak / mean / min vs distance
Effective jet width
Protected-zone fraction
Turbulence decay

P2 Noise Measurements

SPL vs distance
Noise decay (log-log)

P2 Particle Count Analysis

Concentration vs distance
Reduction efficiency
ISO 14644-1 class matrix
Centre vs edge particle counts
Protective envelope
Prototype 3 — Final

P3 — Conditioned Cavity

P3 labeled
Conditioned Cavity — Final Prototype

Redistribution, not addition

Same fan, same H14 filter, same outlet. The only change: a two-stage flow conditioner inside the cavity. Spatial CV dropped from 0.92 to 0.07 — a 13× improvement.

Edge particles at 30 cm: 11,500 (P2) to ∼8 (P3). Protected-zone coverage: 100% at all near-field distances — the clean sheet now reaches the edges, not just the centre.

CV 0.07
Spatial uniformity at 5 cm
100%
Protected-zone coverage
10.4 kg
Total mass
0.14 m²
Footprint
β = 12.4%
Baffle open-area ratio
~64 dB(A)
Noise at 1 m

P3 Photos

P3 device — front view
P3 assembly in laboratory
Noise measurement setup

P3 CAD & Internal Architecture

Cross-section: two-stage flow conditioner
CAD cross-section
Exploded assembly
Section cut
Cutaway view
Front elevation
Perforated baffle plate
CAD exploded view

P3 Velocity Heatmaps

d = 5 cm — uniform distribution
d = 15 cm
d = 30 cm
Ambient reference
6-panel composite
Near-field composite (5, 15, 30 cm)
▶ P3 Per-Distance Detailed Plots (d=5, 15, 30 cm)

d = 5 cm

Profiles
Histogram
Active region
Turbulence

d = 15 cm

Profiles
Histogram
Active region
Turbulence

d = 30 cm

Profiles
Histogram
Active region
Turbulence
As-Built Hardware

Build & Test Evidence

Footage and photographs of the finished P3 hardware — the fastening, filter seating, blower coupling, and the calibrated instruments behind every measurement on this page.

P3 — assembled device, full pan: filter face → bolted housing → centrifugal blower
  • ▶
    H14 filter face & sealed housingPleated H14 media seated in the 3D-printed front frame; bolted side panels clamp the filter and close the cavity.
  • ▶
    Bolted panel seamsThrough-bolted joints along every edge — the fastening that holds the housing rigid and limits edge leakage.
  • ▶
    Blower couplingebm-papst centrifugal blower joined to the housing through a 90° duct; impeller and wiring visible.
  • ▶
    Tool-accessible filter (SR8)The front frame unbolts with hand tools so the H14 cartridge can be replaced in the field.
  • ▶
    Calibrated instrumentsEvery velocity, particle and noise figure traces to the testo 405i, Fluke 985 and Extech 407732 shown below.

Build & Instrumentation Photos

P3 — H14 filter face & bolted housing
P3 — blower coupling (side profile)
P3 — bolted panel seams (top view)
P2 — honeycomb outlet & shared blower
Test instruments — sound meter, anemometer, particle counter

Assembly Session — Timelapse

Hand-assembly of the housing, blower and ducting on the bench — sped up ~24×.

Interactive 3D Models — the printed parts

The actual STL files printed for P3. Drag to rotate, scroll to zoom — including the two-stage flow conditioner that is the project’s key contribution.

Drag to rotate · scroll to zoom
↓ Download this STL

All 7 parts are downloadable. Models are the as-printed P3 geometry (Fusion 360 → STL).

The Key Contribution

Two-Stage In-Cavity Flow Conditioner

Same fan, same filter, same outlet. The only change: what happens inside the cavity before air reaches the HEPA face.

How it works

Stage 1 — Solid Impingement Plate

140 × 140 mm, 175 mm from outlet. Breaks the concentrated axial fan jet and deflects flow radially, distributing momentum across the cavity.

Stage 2 — Perforated Baffle Plate

240 × 240 mm, 3 mm thick. 365 holes, 5 mm diameter. β = 12.4%. 135 mm from outlet. Creates uniform pressure drop across the HEPA face.

ΔPpp ≈ 1–3 Pa on a ∼60 Pa baseline. Added mass: 0.05 kg.

The Result

P2’s fast 2.51 m/s jet over the central ∼60% becomes a calm 1.71 m/s spread across 100% of the outlet. Momentum redistributed, not added — same total airflow, just spread evenly instead of piled into the middle.

Measured Performance

Key Results

All from physical measurements. Velocity: testo 405i (8×8 grid). Particles: Fluke 985. Noise: Extech 407732.

13×

More Uniform Outlet

CV: 0.92 → 0.07 at d=5 cm

>1000×

Cleaner Edges

11,500 → ~8 at d=30 cm edge

100%

Protected Zone

≥0.3 m/s at d=5, 15, 30 cm

5/8

Requirements Met

P3: 5 met, 1 partial, 2 unmet — the two gaps share one fix

Uniformity Data (CV by Distance)

DistanceP2 CVP3 CVReductionP2 ProtectedP3 Protected
5 cm0.920.0713.1×58%100%
15 cm0.560.096.2×78%100%
30 cm0.510.114.6×92%100%
60 cm0.380.251.5×94%100%
100 cm0.350.201.8×100%97%
150 cm0.310.211.5×100%89%

Near-field (d≤30 cm) primary evidence; far-field (dimmed) supporting trend. SR4 target: CV<0.25.

Outlet velocity: P2 vs P3
CV vs distance

SR1–SR8 Verification Matrix

SRRequirementP1P2P3
SR1Outlet mean 0.3–0.5 m/s at d=30 cm×××
SR2≈ ISO Class 5 at centre, d≤15 cm◔◔◔
SR3H14 installed (EN 1822-4)×✓✓
SR4Spatial CV < 0.25××✓
SR5Footprint ≤ 0.5 m²✓✓✓
SR6Mass ≤ 15 kg✓✓✓
SR7SPL ≤ 53 dB(A) at 1 m×××
SR8Filter accessible, hand tools✓✓✓

P3: 5 met, 1 partial, 2 not met. SR1+SR7 share one root cause — fix: higher β (18–25%) + 30–40% slower fan.

Head-to-Head

P2 vs P3 Comparison

Same fan, same filter, same ducting, same outlet. The only variable: in-cavity flow conditioning.

Heatmap Comparison

Near-field 2×3 grid
Full 2×6 grid (all distances)
Report figure
Before/after poster figure

Uniformity & Velocity

CV comparison
Mean velocity
Velocity range
Protected-zone fraction
Report: CV comparison
Report: centerline decay
Report: protected zone
Report: mean velocity

Noise & Particles

Acoustic comparison
Report: noise
Particle comparison
Presentation: acoustic

Cleanliness (Fluke 985, ≥0.3 µm)

LocationP2P3
d=5 cm, centre60
d=5 cm, edge mean62~3
d=15 cm, centre260
d=30 cm, edge11,500~8
Room ambient27,113
Honest Assessment

Known Limitations

P3 met 5 of 8 success requirements. Three remain open — stated plainly here, with their shared root cause and the validation still needed. This is a research prototype, not a certified device.

SR1 — Not met

Outlet velocity too high

Target (at 30 cm)0.3–0.5 m/s
Measured (P3)≈ 0.98 m/s

Uniform across the field, but roughly double the target speed — the fan is over-driven for the current baffle.

SR7 — Not met

Noise above ceiling

Target (at 1 m)≤ 53 dB(A)
Measured (P3)≈ 64 dB(A)

Driven by the same over-speed fan plus flow noise through the baffle — not a separate problem.

SR2 — Partial

Cleanliness claim unproven

Evidence1× 1-min sample
Needed≥3 repeats/point

Particle counts are encouraging (∼8 vs 27,113 ambient), but single samples can’t support a formal ISO Class 5 claim.

One root cause, one path to close the gaps

SR1 and SR7 are the same problem. Both come from running the fan harder than the design needs. The fix is to raise the perforated baffle’s open-area ratio β from 12.4% to ~18–25% and slow the fan by 30–40%. That drops the 30 cm velocity into the 0.3–0.5 m/s band and cuts the noise, while the two-stage conditioner keeps the uniformity gain that P3 already proved.

Next validation (P3.1). Re-run the 8×8 velocity grid at 30 cm and the 1 m sound level at the lower fan speed; repeat particle counts at the centre and four edges across 5/15/30 cm with ≥3 samples per point and ambient logged each time. This directly retests the two open requirements without redesigning the device.

Technical Details

Hardware & Materials

All components student-designed in Fusion 360. Audited project spend: £1,901.81.

⚙

Centrifugal Fan

ebm-papst D3G146-HQ13-34
EC centrifugal blower
Mass: 3.9 kg
Cost: £369.60

■

HEPA Filter

SPCB Bespoke H14
EN 1822-4, 99.995% at MPPS
110 mm depth, 3.0 kg
Cost: £323.96

▲

Housing

3D-printed PLA
Designed in Fusion 360
Mass: 3.5 kg
Bio-based thermoplastic

◆

Perforated Baffle

240×240×3 mm
365 holes, 5 mm ø
β = 12.4%
Added mass: 0.05 kg

◉

Outlet

203 × 203 mm
Hex honeycomb diffuser
Footprint: 0.14 m²

▣

System Total

10.4 kg
Single-person portable
Tool-free filter access
£1,901.81 total

Instruments (all loaned by Prof. Eames)

testo 405i

Hot-wire anemometer
8×8 grid, 6 distances

Fluke 985

Particle counter (≥0.3 µm)
2.83 L sample volume

Extech 407732

Sound-level meter
dB(A), 5–150 cm range

For the Next Team / Assessor

Handover Package

Everything needed to verify, rebuild, or continue the project. Items marked Available are linked here; the rest are in the MECH0073 supplementary submission.

📑

Final Design Report

40-page report: PDS, SR1–SR8 verification, uncertainty, CFD cross-validation.

In supplementary submission
📄

Project Poster

Summer Session 2026 exhibition poster.

Available — view
🧾

Bill of Materials

Full component list, suppliers and audited spend (£1,901.81).

Available — download .xlsx
⚠

Risk Assessment

Hazards, mitigations and sign-off for build and testing.

Available — open PDF
✅

SR1–SR8 Verification

Pass / partial / fail matrix for all three prototypes.

Available — on this page
📷

Build & Test Evidence

As-built photos, device pan, and the calibrated instrument set.

Available — on this page
🔧

CAD & STL Models

All 7 P3 parts as STL (rotate live or download), plus the P2 Fusion 360 source.

P3 3D viewer · P2 .f3d
📊

Raw Data & Scripts

Anemometer / particle / noise logs and the Python analysis pipeline.

In supplementary submission
📝

Data-Collection Tool

The custom web app used to log velocity / particle / noise on the measurement grid, with CSV & JSON export.

Available — open tool
♻

Maintenance & End-of-Life

Filter-change steps, HEPA disposal and material recycling plan.

Planned for handover
Exhibition

Project Poster

UCL Mechanical Engineering Summer Session 2026.

MECH0073 Group 22 Poster
The Team

Group 22

UCL Mechanical Engineering • MECH0073 • 2025–2026

YZ

Yifei Zhang

HG

Haotian Gao

JZ

Jiahe Zhou

QY

Qi Yang

RW

RuiFan Wang

Supervisor: Prof. Ian Eames • Module Lead: Dr Andrea Grech La Rosa

UCL Department of Mechanical Engineering • London, UK