Engineering Impact in Practice

Applications

Four projects across EV battery module assembly, implantable cardiac monitor production, surgical implant compliance recovery, and battery backfill process optimization. Each defined by a clear problem, engineered solution, and measurable outcome.

EV Battery Module Assembly

Machine Vision System Rebuild: Cycle-Time Reduction & Yield Recovery

S Situation
Before After
  • Six EV battery module assembly lines running at capacity
  • Sub A/B: adhesive dispense + profilometer inspection via overhead vision gantry
  • One AutoAssembly cycle = one module; any Sub A/B failure loses one unit
  • 6 lines × failure rate = 240 cars/week exposure
T Task

Dual Bottleneck Identified

1.

Cycle-Time

Sub A/B CT exceeded line bottleneck, starving downstream

2.

Yield

Components failing inspection reduced available throughput

Gantry Path: Cycle Time Split

Manual Ingress
4%
Conveyance To Dispense
12%
Adhesive Dispense
43%
Profilometer Inspect
32%
Conveyance From Dispense
9%

Adhesive Dispense + Profilometer = 75% of path time. Inspection travel reducible by 2/3.

Required Outcome

Rebuild the system ground-up: eliminate 100% of starved time, recover 1% yield, integrate all 6 line variants without halting production

A Action

Root Cause Analysis: 8 Identified

×Image Validation Issues
×Inspection Programming is Convoluted and Ineffective
×Visualization is Ineffective
×Data Meaningless
×Mechanical Issues With Gantry
×Fixtures Maintenance
×Subcomponent Quality Issues
×Troubleshooting Requires Engineering to Deep Dive each failure

Ground-up rebuild eliminates these failure modes and enables structured categorization for reaction planning, recovering 1% process yield.

System Integration: 14 Interdependencies

Vision programming
Simplified movements
Conveyance control interlocks
MES API integration
Mechanical gantry changes
Useful data collection
Feedback Control Loops
Clear failure mode definitions
Effective visualization
SPC alert
Python bot alerts
Useful reaction plans
Effective troubleshooting documentation
Effective training

Fixturing Design & Gantry Improvements

1.Custom fixturing for 6 line variants with differing mechanical configurations
2.Duplicate system built to run parallel to live production line without interrupting output
3.Developed gantry motion, image tuning, triggering sequencing, storage, networking

Inspection Programming

1. Define Failure Modes and Severities

Critical to Quality of Product

Critical to Yield of Station

2. Develop Inspection Techniques

1. Fundamental Program Architecture, PLC Inputs, Vision Tools, Variable Maintenance

2. Data Collection Logic

3. Data Storage Array Logic

4. Data Conditioning Logic

5. Communication and Storage Logic (PLC and Global)

6. Part Disposition Logic

7. Visualization and SPC Logic

15-Step Script Architecture

1 Locate Part Location by Edge
2 Locate Inner Fiducial Markers
3 Calculate Geometries
4 Generate Inspection Regions
5 Profile Validation
6 Profile Measure
7 Special Feature Validation
8 Special Feature Measure
9 Foreign Object Measure
10 Inspection and Image Loop
11 Array Variable Data Storage
12 Data Conditioning
13 Data Judgement
14 Communication to PLC
15 Communication to FTP

System Validation / Process Control / Management Tools

Gauge R&RFault InjectionProcess QualificationControl PIDSPC and Data VisualizationHMIAutomated Python Alerts
R Result

100%

Starved Time Eliminated

240

cars / week

6 lines

+1%

Process Yield

Vehicle Sale Value

$631.4M

12,480 carsets / yr

91%

CT contrib.

9%

Yield contrib.

Implantable Cardiac Monitor

5× Production Rate Planning: OEE Modelling & Automation Strategy

S Situation
1× → 5×
  • 10-step implantable cardiac monitor battery manufacturing process
  • OEE model: Availability × Performance × Quality; rolling throughput = cumulative FPY × downtime losses
  • Current rolled throughput: 219 units/day
  • 5× demand increase with existing process infrastructure
T Task

Scale to 1,096 units/day. VSM revealed two root capacity failures:

1.

Process Capacity

7 of 10 processes below target; Laser Weld Can Lid at 228/day is the critical constraint

2.

First-Pass Yield

Powder Metering FPY at 45%; loss compounds across all downstream steps

Process Cap/Day Rolled T
Powder Metering 1,021 1,021
Cathode Forming 657 657
Sealing and Excision 1,131 624
Inspection 694 562
Stack Assembly 720 556
Interior Pin Weld 2,328 551
Laser Weld Can Lid 228 228
Electrolyte Fill 268 224
Close and Terminal Weld 613 221
Laser Completion Weld 1,263 219
Current final throughput 219
A Action

Strategy Evaluation

Short-Term BN Management

1,096 /day

  • 3 shifts required
  • +11 headcount, 6 CAPEX
  • Process duplication only
  • Does not address root yield causes

Automation: Selected ✓

1,113 /day

  • 2 shifts only
  • CT reduction to takt
  • Yield + labor optimization
  • Solves root causes

Cathode Takt

28.15s

Battery Takt

21.89s

Automation Plan: CT & Yield Optimization Per Process

C Powder Metering Yield 45% → 95% via automated dispense control; eliminate manual rework loop 28.15s
C Cathode Forming CT reduction to cathode takt via process parameter optimization 28.15s
C Sealing and Excision Automated sealing control; CT and DT reduction to cathode takt 28.15s
C Inspection Automated inspection; CT reduction to cathode takt 28.15s
B Stack Assembly Automated stack feed; CT reduction to battery takt 21.89s
B Interior Pin Weld Weld process optimization; CT reduction to battery takt 21.89s
B Laser Weld Can Lid Laser parameter optimization; CT reduction: eliminates 3rd shift requirement 21.89s
B Electrolyte Fill Precision fill automation; DT reduction; CT reduction to battery takt 21.89s
B Close and Terminal Weld Automated weld scheduling; CT reduction to battery takt 21.89s
B Laser Completion Weld CT reduction via parameter optimization to battery takt 21.89s
R Result

1,113

units / day

5× demand achieved

2

Shifts

−1 vs plan A

10/10

Processes

≥ 1,096/day

Key Yield Recovery

Powder Metering

45% 95% FPY

vs. Baseline

Throughput 219 1,113
Bottleneck 228 ≥1,113

Rolling Throughput / Day by Phase

2100 1500 1096 600 0 PM CF SE IN SA IPW LWC EF CTW LCW Current BN Mgmt Auto
Bovine-Derived Surgical Implant

Bacterial Endotoxin Compliance Failure: FDA Warning Letter & Facility Recovery

S Situation

Product: Bovine to Implant

Bovine

Processing

Implant

What is Bacterial Endotoxin?

A component of the outer bacterial cell membrane. Sterilization kills bacteria but does not destroy residual endotoxin; it must be quantified directly in the finished implant.

Measured via LAL assay: enzymatic reaction induces clotting; scattered light density measures reaction time.

  • Bovine-derived surgical implant; $68M manufacturing facility
  • FDA Warning Letter invalidated prior qualification methodologies
  • No finished-goods Bacterial Endotoxin Test method existed for the product
  • Facility shut down 1 month; product withdrawn from distribution
T Task

Crisis Conditions

Facility shut down for 1 month. Product withdrawn from distribution. Outsourcing of the test initiated.

Recovery plan created and implemented with FDA inspectors on-site.

Required Outcomes

Develop and validate a Bacterial Endotoxin Test method
Release quarantined inventory
Restore production capability
All under active FDA observation

Target Timeline

3 Days

to facility back online

Recovery Framework

1

Patient Safety

2

Method Validated

3

Facility Online

4

Compliance/Cost

A Action

1. Patient Safety: Post-Market Surveillance

Post-market surveillance: Occurrence rate in complaint history
% spike recovery: spike sample with known Endo concentration
95% confidence limit implies 80% 'interference factor' is possible
Applied correction factor to historic population

2. Test Method Validation

Strategy to limit scope to 'Worst-Case' by Product Family
Factors: Age of Animal, Processing Method, Volume, Surface Area
Reduced unknown interference: Increase dilution
Rewrote test method
Corrected direct to FDA observations

Capability Progression

Initial

Cpk 0.32

Ppk 0.33

Triage

Cpk 0.90

Ppk 0.87

Final

Cpk 2.7

Ppk 2.34

BE Validation Criteria

Accuracy

Sample in Duplicate

Linearity

CV%<10%, CC≥0.980

Precision

CV%<10%

Specificity

% Spike Recovery = 50%–200%, PPC CV%<10%

Detection Limit

<0.005 EU/mL

3. Facility Back Online

Quarantine and sustaining lots cleared for release. Second shift added to clear production backlog.

3 Days

Facility Back Online

4. Compliance / Cost Reduction

BET Validation Decision Flowchart

BET Test Dilution Time Volume Temp Analyst Performance Pipetting · Reagent Prep Accuracy Linearity Precision Specifi- city Detection Limit ? NO ? NO ? NO ? NO ? NO Y Y Y Y Y Validated Process

Monitoring Reworks Per Day

0 3 5 8 Initial Triage Final

Spike Concentration Recovered

USL 0.5 LSL 0 1.0 Initial Triage Final

Saved 2 hours of test time per day. Second shift QC implemented, doubling throughput. Facility prepared for FDA audit.

R Result

3

Days to Facility Online

$1.2M

Released

$3.6M

Saved

Revenue Preserved

$5.1M

Final Capability

Cpk 0.32 2.7
Ppk 0.33 2.34

Rework Reduction

Per day 4–7 ≈ 0
Implantable Battery Backfill Process

Battery Electrolyte Backfill Optimization: DMAIC Yield & Downtime Recovery

S Situation
DefineMeasureAnalyzeImproveControl

Problem Statement

Yield 95.4%
Scrap cost $1.045M/yr
Downtime 24 hr/mo
Downtime cost $741K/yr
  • Implantable battery electrolyte backfill; 0.020 mg process window
  • Weight Low NC = 89.6% of all nonconformances
  • Goal: +3.6% yield, reduce downtime to 3hr planned PM, save $692K scrap + $648K throughput
  • Timeline: DMAIC executed to April 2025
T Task

Process Map: Backfill Sequence

Fixing turing Vacuum 1 Electro lyte 1 Vacuum 2 Electro lyte 2 Ctrl: Elec Pressure · Spool Valve · Vac Pressure · Elec Batch

Nonconformance Pareto

100 75 50 25 0 Wt Low 89.6% Equip Dn 5.4% Wt High 2.5% Other 2.5% n=713
A Action

Measure: MSA & Process Capability

Gauge R&R

Part-to-part exceeds within-operator → measurement system adequate

Special cause: Operator 3 introducing variability

I-MR Control Chart

Special causes identified · Clear process shift · Bimodal distribution

Window: 0.020 mg

Analyze: Root Causes (Multi-Vari & DoE)

×Batch variation: leaking electrolyte tank (removed)
×Vacuum pressure variation: leaks identified and fixed
×Spool valve misalignment: corrected
×Nozzle alignment inconsistency: pneumatic cylinder redesigned
×Airflow variability on spool valve: standardized with fixed orifice
×Operator 3 special cause: addressed via training & cable mgmt overhaul

DoE Top Factors (Pareto of Effects)

B
Spool Elec→Vac
A
Electrolyte Pressure

Improve

Redesigned pneumatic cylinder for nozzle alignment
Fixed orifice on spool valve: standardized airflow
Standardized vacuum & electrolyte pressure setpoints
Cable management overhaul: eliminated interference

Control

Troubleshooting guide + reaction plan
Standardized PM kits
Trailing 50 measurements returned to PLC for auto-SPC
Cathode forming improved to reduce incoming mass variation
R Result
95.4% 99.6%

Process Yield

+4.2%

vs 3.6% goal

$770K

Cost savings

Final Capability

Cpk (within) 1.01
Pp 0.97
Ppk 0.90

Failure Rate

4.6% 0.4%

Downtime

Per month 24 hr 3 hr PM