Duration
130 weeks
Case Study 01 · Industrial IoT
Energy Asset Monitoring Platform
Designing proactive power systems intelligence for high-stakes enterprise operations.
- Role
- Lead Designer — End-to-End Design
- Domain
- Industrial IoT · Energy & Utilities
- Duration
- 130 weeks
- Tools
- Figma, Adobe XD, Miro, Google Form, MS Teams
Overview
The strategic frame
I designed the interface where a grid engineer decides whether to trust what an automated diagnosis just told them about a transformer they cannot see — and is held accountable for that decision either way.
Scope
216 screens and states
Roles served
4 distinct operator types
Research
B2B, via proxy users
Problem
The problem and its stakes
Energy operators managing transformer fleets across geographies had no unified view of asset health. Processes were siloed, raw sensor data was hard to interpret, and the consequences compounded: performance issues were detected late, maintenance stayed reactive, compliance reporting consumed time, and there was no way to model what-if scenarios for risk planning.
What was at stake
Unplanned downtime costs large industrial facilities an average of $129M a year (Siemens, True Cost of Downtime, 2022). That is the industry context this platform was built to reduce — not a measured result of this project.
Why it was hard
Four roles, one product
Operators need instant situational awareness. Maintenance technicians need diagnostics and history. Managers need fleet-level strategy. Engineers need deep technical analytics. A single generic view would serve none of them; four separate products would fragment the shared truth operators most need to trust.
The system advises about physical equipment, and it can be wrong
The platform computes suggested fault diagnoses from dissolved-gas ratios against published standards. An engineer who follows a wrong recommendation damages equipment; one who dismisses a correct one risks failure. The interface sits directly on that judgement.
No direct access to end users
As a B2B platform, research ran through Product and Project Managers and domain experts who worked with customers. That is a real limitation, and it shaped how much weight to put on early assumptions versus what testing later confirmed.
Research
Evidence under constraints
Quantitative
Structured questionnaires distributed to selected customers on functionality and expectations, coordinated through the PMs who held the customer relationships.
Qualitative
Discussions with PMs and domain experts on recurring pain points, usability challenges and feature-adoption issues — proxies for users I could not reach directly.
Usability testing
Remote moderated sessions with proxy users across four task scenarios: check equipment health, respond to an alarm, review historical analytics, generate a compliance report.
What worked
- Central dashboard and real-time multi-device view were valued
- Alarm acknowledgment flow was straightforward and reduced response time
- Reporting and export were highly valued for audits
What didn't
- Users struggled to locate specific device data — navigation needed clearer information architecture
- Analytics graphs were too data-heavy and needed simplification
- Report customisation was too limited — users wanted filtering by date and asset
Personas
User landscape
Power Systems Operators
Real-time- Context
- Control room environments, 24/7 operations
- Needs
- Instant situational awareness, critical alert prioritisation
- Goals
- Quickly identify high-risk assets requiring immediate attention
- Challenges
- Information overload, time-critical decisions
- Success metrics
- Response time, situational awareness accuracy
Maintenance Teams
Predictive- Context
- Field operations, mobile access, power electronics expertise
- Needs
- Early fault detection, maintenance planning optimisation
- Goals
- Shift from reactive to predictive maintenance strategies
- Challenges
- Resource allocation, service planning complexity
- Success metrics
- Maintenance efficiency, equipment uptime
Managers & Planners
Strategic- Context
- Executive environments, strategic planning sessions
- Needs
- Fleet-wide health insights, environmental impact understanding
- Goals
- Optimise asset utilisation, allocate resources strategically
- Challenges
- Balancing operational cost with performance
- Success metrics
- Asset utilisation, strategic goal achievement
Power Electronics Engineers
Analytical- Context
- Engineering workstations, complex analysis workflows
- Needs
- Load flow simulation, fault location, weather risk assessment
- Goals
- Energy distribution optimisation and system balancing
- Challenges
- Complex data analysis, multi-variable optimisation
- Success metrics
- Grid stability, distribution efficiency
Design judgement
Decisions under constraint
01
Accountable Alarm Dismissal
Problem
Operators face high alarm volume, so dismissal must stay fast — but a silently dismissed alarm on critical infrastructure is an accountability gap.
Solution
Notifications move through New → Noted → Disregarded, and marking one disregarded requires a written reason. The record then shows who disregarded it, why, and when.
Result
Dismissal stays quick but stops being invisible — the system captures human judgement rather than discarding it.
02
Advisory Diagnostics, Not Automated Verdicts
Problem
Dissolved-gas analysis can reveal developing faults, but interpretation is expert work grounded in published standards. Automating the decision would be inappropriate; showing only raw ratios would waste the analysis.
Solution
The platform computes gas ratios and presents a Suggested Fault Diagnosis alongside the underlying ratio values and the standard it derives from (IEEE Std C57.104-2019, ICDL 1999).
Result
Engineers can evaluate the reasoning rather than accept a verdict — transparency about why is what makes the recommendation trustworthy.
03
Expert Override With Stated Risk
Problem
Siemens recommends monitoring thresholds, but site conditions vary and engineers legitimately need to adjust them. Locking them would be wrong; allowing silent changes would be unsafe.
Solution
Thresholds are adjustable with explicit upper and lower values, and deviating from recommended values surfaces a direct warning that doing so, or relying on monitoring with inaccurate parameters, may negatively affect the equipment.
Result
Expert authority is preserved and the consequence is stated at the moment of decision rather than buried in documentation.
04
Role-Based Dashboards
Problem
Four stakeholder groups had different priorities competing within one interface.
Solution
Designed four dashboard variants — Operator, Maintenance, Manager, and Engineer.
Result
Reduced cognitive load by showing each role the information most relevant to its work.
05
Geospatial Visualization
Problem
Distributed asset health was difficult to understand quickly.
Solution
Introduced an interactive map with status color-coding and heat maps.
Result
Faster situational awareness across the fleet.
06
Alert Prioritization
Problem
Teams faced 100+ daily alerts without a clear way to focus.
Solution
Introduced severity levels and smart filtering.
Result
Critical alerts became easier to identify and act on.
07
Customizable Dashboards
Problem
Fixed dashboards could not support the priorities of every team.
Solution
Added user-configurable widgets and layouts per team.
Result
Each team could tailor monitoring to its workflows without fragmenting the platform.
08
Real-Time Updates Without Disruption
Problem
Live data refreshes could interrupt attention and create jarring interface changes.
Solution
Used a 5–10 second refresh with smooth animation.
Result
Information stayed current without disrupting users' focus.
Impact
Business outcomes
- 4 distinct roles served from one information architecture
- Maintenance model shifted from reactive to predictive across the fleet
- Accountability built into overrides — every dismissal carries a reason and an author
- Advisory, standards-based diagnostics rather than opaque automated verdicts
- 3 siloed dashboards consolidated into 1 unified platform
Downtime cost figures cited above are industry benchmarks (Siemens, 2022), not measured project results.
Reflection
What I would carry forward
The strongest thing about this project is not the dashboards — it is that the design took a position on where automation stops. The platform computes, suggests and warns; it never decides. Every consequential action leaves a human name attached to it. That principle came from the constraint rather than in spite of it: without direct access to end users, I could not design on assumption about how operators would behave. Making the system's reasoning visible and every override attributable was the safer architecture — and the one that held up in testing.
What came next
- Navigation and IA simplification following usability findings
- Simplified analytics visualisation to reduce data density
- Expanded report filtering by date and asset
Strategic impact
Design Leadership
- Led design end-to-end as the lead designer, partnering with product, business, and development teams
- Aligned conflicting stakeholder priorities into a shared strategy
- Established design standards adopted across multiple platform initiatives
Strategic impact
Executive Influence
- Presented platform strategy to C-level executives and business unit leaders
- Translated UX requirements into business value propositions
- Influenced product roadmap and strategic investment decisions
Strategic impact
Innovation Leadership
- Pioneered a geospatial, real-time visualization approach for enterprise power systems monitoring
- Established design patterns influencing future platform development
- Pioneered a geospatial analytics approach later referenced in adjacent platform work.
Next case study
SIMIT Rapid Tester — Engineering Automation Platform
Get in touch