5+
Granted patents
Named inventor
~ 7000
Workers
Worker monitoring scale
85/100
SUS score
Usability evaluation at site
Over six years at Honeywell, I designed and later led UX for mission-critical industrial safety systems where decisions carried real-world consequences. My work addressed two challenges: helping operators monitor thousands of workers despite approximate location data, and unifying fragmented safety events and dashboards across acquired products.
My contribution:
This case study is shared for portfolio evaluation only. Selected screens have been reconstructed to respect confidentiality. Product names, trademarks and underlying IP belong to their respective owners.
Workers in large oil & gas refineries face:
Hazardous gas exposure (H₂S, SO₂, CO, Methane etc.,)
Emergency evacuation scenarios
Compliance requirements
Large-scale operational risks
Workers carry portable gas detectors that monitor hazardous gas exposure and safety conditions. With newer connected detectors and site-wide Wi-Fi infrastructure, worker location and device status could also be relayed in real time to the control room.
Delayed or inaccurate information can impact the workers’ life or the productivity of the plant.
“It all started with the canaries”
“Wear yellow. Work safe”
Real-time worker awareness platform
Problem:
Safety teams need immediate visibility into
Worker location
Gas exposure
Incident zones
Evacuation opportunities
Solution:
Build a map-based real-time situational awareness platform.
Role:
Sole UX designer & researcher
Learning the safety-critical domain before designing
What I needed to understand:
• Emergency response procedures
• Operator workflows
• Refinery safety protocols
• Communication chains during incidents
What constraints existed:
• No access to domain experts
• Engg & PM have only surface knowledge
• Site visit allowed, but no interview
• Only 15 mins with site safety leads
I created a set of illustrated scenario cards depicting potential emergency situations and printed them to show to the users. The goal was to validate my understanding of the domain in their environment.
These represented my assumptions about:
Gas exposure incidents
Emergency evacuation workflows
Compliance monitoring
Correcting assumptions in the proposed workflows
Explaining SOPs during emergencies
Informing what data is actually required in such times
The visual scenarios created a shared language with the operators, shifted from answering our questions to teaching us how the system needed to work.
3 refinery visits
Sweeny, Texas
Port Arthur, Texas
Garyville, Louisiana
9 scenario cards
Critical event detection - 4
Non critical events - 3
Compliance monitoring - 2
15 interactions
Safety leads - 4
Workers & operators - 7
Management & others - 4
I synthesized recurring field observations into principles that shaped attention, navigation, spatial context and the representation of uncertain data.
Attention is Scarce & Fragmented → Make critical state changes impossible to miss
Operators managed multiple processes and monitored safety intermittently. High-contrast visual and audible alerts surfaced only when conditions required attention.
Glanceable UI & High Contrast State Changes
Site
Zone
Worker
Emergencies cause Tunnel Vision → Keep interaction in single screen progressive flow
The earlier multi-screen proof of concept (PoC) was cumbersome under alarm stress. Progressive disclosure with increasing visual contrast kept operators in one context while narrowing attention.
Operators rely on Visual Memory → Use recognizable site imagery
Operators struggled with abstract blueprints while testing the PoC created by Engineering team, but could recognize even obscure plant locations from satellite or aerial imagery of their premises.
Active Incidents
Alarm
Location
Alarm
Location
Device Fault
Location
Warning
Location
Active Alerts are Few → Keep incidents persistently visible
Medium-sized plants typically had fewer than ~15 active alerts, making a persistent incident list practical without overwhelming the screen.
False Man-down alerts are Many → Use location context before escalation
Detectors left in parking or office areas could trigger man-down alerts by inactivity. Location context helped to distinguish these as potentially lower-risk events to verify further.
Location info is Inherently Approximate
→ Show uncertainty, not false precision
Wi-Fi triangulation was inconsistent and GPS degraded around metallic infrastructure. Designed multiple ways to show approximate location as that is still actionable as teams are familiar with site.
2D Location Misses Vertical Risk → Explore elevation awareness
Floor-level information wasn't enough around tanks and elevated structures. We experimented with barometric sensing for Z-axis data along with gas-detector, but the reliability was insufficient.
All that led to
➤
New LocaXion Manager
38 seconds duration
With voice over
Scope:
LocaXion Manager focused on real-time awareness and response initiation. Continuous worker tracking to a safe zone was explored but not productized because location remained approximate; post-incident exposure analysis was handled separately.
Validated in the Operating Environment
Tested a version of LocaXion Manager with five users spanning safety leads, supervisor, operators and automation. Users completed representative alarm and navigation tasks with minimal onboarding.
Testing also exposed issues around location uncertainty, reporting expectations and interaction affordances that informed subsequent refinement.
Location: A midstream Gas processing plant, Bridgeport, Texas
Participants
HSE focal
Control room operators - 2
Plant foreman
Automation specialist
Test objectives
Using, comprehending alarm notifications
Accessing incident & device details
Navigating across site, zone and worker
Accessing and using historical records
Result: Average SUS score of 85 / 100
Solving for Extremes
During a pre-launch demo, a major oil & gas customer in Kazakhstan told us they had an approximate 7000 workers on site at once, each carrying a gas detector.
That number broke every assumption the design had been built on.
Standard map clustering groups worker icons by screen proximity. As users zoom, clusters split and merge based on what is visible on screen but not the workers' physical proximity on site.
Problem: Why standard pixel-based clustering fails here
For safety-critical monitoring, changing clusters with zoom can obscure which workers are actually near one another and make alarmed workers harder to identify.
Same area of a refinery shown at two zoom levels
Small circle
:
A worker
Big circle
:
A screen based cluster representing multiple workers
Pixel-based clusters split while zooming, misrepresenting physical proximity
Solution: Clustering by physical proximity
I designed clustering around real-world worker proximity rather than screen pixels, using configurable physical footprints that remained stable across zoom levels.
The cluster footprint was configurable, not fixed.
The cluster footprint depended on the site map’s resolution, usable zoom, operational needs and worker-location accuracy.
For this site, a 5m × 5m footprint provided sufficient spatial distinction. Four or more workers within that area formed a cluster.
Once configured, zooming doesn’t change the physical grouping.
➤
27 seconds duration
With voice over
Outcome
The interaction architecture and geospatial method resulted in a granted US utility patent. I am the only inventor for this patent which is active until 2037.
Strategic Evolution
From Real-Time Awareness to Unified Safety Intelligence
Field research revealed an opportunity beyond monitoring the live incidents. Safety teams were managing fragmented device systems, manual compliance workflows and disconnected reporting.
The opportunity evolved from understanding real-time incidents to managing safety, compliance and productivity as one connected operational system.
For Honeywell’s hazardous gas detection business, the acquisitions (BW & RAE) had left the functions of ‘Device management’ and ‘Real-time monitoring’ fragmented across four tools and two device families, each with different information structures, terminology and interaction patterns.
The challenge wasn't simply merging four tools; it was creating a common operational model across them.
Role :
Focus :
Same safety conditions, different language:
Design decision:
I created a unified taxonomy covering 50+ safety events, alarms, faults and compliance conditions across the platform.
One event model across the platform
Consistent interpretation and notification behaviour, enabling easier integration and future scalability.
Monitoring tells you what happened. Investigation explains why.
Unlike LocaXion Manager's real-time incident response, Sotera also had to help safety teams diagnose exposure and device issues after they occurred.
I designed event details modal around three questions:
What happened?
How severe was it?
What else happened around it?
A single event became an investigation path rather than an isolated alert.
Today's compliance issue can become tomorrow's safety problem.
I led the dashboard design to connect real-time worker safety with forward-looking device compliance and availability.
Safety intelligence
Alarm durations
Exposure severity
Worker health by these two
Operational readiness
Compliance workload
Devices availability
Planning
The dashboard shifted the platform from reacting to incidents toward preventing conditions that could contribute to them.
Outcome
The Safety + Productivity dashboard resulted in an EU registered design. I am one of five named designers on the registration which is active until January 2028.
EU Registered Design:
Sotera Dashboard - 004595312-0002
Validation of Sotera
Concepts were validated across three prospective customer sites in the US by a UXR colleague; findings shaped inventory filtering, information density and monitoring presentation.
Impact & Legacy
Intellectual Property across US, EU & China
Utility & GUI Design Protection
Longevity of the work done for Safety Suite
Taxonomy retained for 2.0
Product scale proof point
Scaled for ~ 7000 workers
Deployment breadth
US, Canada & UAE
LocaXion Manager usability
85/100 score on System Usability Scale (SUS)
Recognition
3 Partner awards & 1 for Coaching
What Remained
Location Precision
Real-time worker location remained approximate rather than precise. The underlying positioning technology had not matured enough to close this gap before I left Honeywell.
Vertical Position
Reliable Z-axis positioning in multi-level hazardous environments also remained unresolved. Our barometric sensing experiment did not achieve the reliability required for deployment.
Unlike the taxonomy and architecture problems we could solve through design, location precision remained constrained by hardware and physics. We could design around those limitations, but not eliminate them.