Operational Snapshot & Impact
High-stakes systems integration demands real-world reliability, sub-second latency, and deterministic execution under peak production load. Here is the operational profile:
The Partnership: Domain Expertise Meets Software Engineering
Maintenance In Balance (MIB) was developed within CMServices Global Ltd as a commercial machinery-health venture. Simon Edmondson led the vibration analysis side of the business, supplying the sensor data and deep domain knowledge of industrial machine health. Warren Basterfield was the technical partner responsible for the complete software architecture, mathematics, and product engineering.
The success of the venture came from this cross-disciplinary collaboration: translating specialist mechanical vibration expertise into algorithmic signal processing and scalable software.
The Challenge: Unstructured Raw Waveforms
The machinery was monitored via Prüftechnik condition monitoring sensors, which recorded periodic vibration measurements. The objective was to evolve beyond static condition reporting (i.e., "is this machine shaking excessively right now?") to true prognostic maintenance: When will this specific bearing fail?
The technical challenge was that the monitoring system did not produce pre-digested feature vectors. The data was stored as raw binary byte arrays representing time-domain waveforms. Warren had to engineer the mathematical pipeline to extract meaningful harmonic signatures from these raw signals and track their evolution over time.
FFT Signal Processing Pipeline
Warren engineered a signal-processing pipeline using Fast Fourier Transforms (FFT) to convert time-domain waveform byte streams into frequency spectra:
Different bearing failure modes (outer race spalling, inner race defects, ball wear) manifest at characteristic frequencies governed by bearing dimensions and shaft RPM. By isolating these specific frequency bands, the software tracked the subtle energetic growth of early defects long before they caused noticeable machine vibration.
Time-Series Forecasting
With clean frequency-band energy levels extracted over time, Warren trained time-series forecasting models against the historical trend lines.
- Trajectory Modeling: Instead of waiting for a hard threshold breach, the system projected the wear slope forward to estimate the operational window remaining before critical failure.
- Continuous Model Updating: As new vibration measurements arrived from the field, forecasts automatically recalibrated their trajectory and confidence bounds.
- Actionable Engineering Output: Reports provided plain-English diagnostic guidance: identifying which specific bearing was deteriorating, the nature of the developing defect, and the recommended intervention schedule.
The Web Reporting Platform
Warren engineered the entire client-facing web application in C#, providing plant reliability engineers with instant access to their machinery portfolio:
- Secure, role-based dashboards presenting high-level fleet health heatmaps down to individual bearing harmonic graphs.
- Automated alert generation prioritizing assets exhibiting accelerating wear patterns.
- Exportable engineering compliance documentation supporting scheduled overhaul planning.
Scale & Industry Recognition
The platform proved that complex mathematical signal processing could be operationalized into a dependable commercial product:
- Operational Scale: Actively monitored over 1,000 bearings across 200 heavy machines for 10 corporate industrial clients.
- Proven Outage Prevention: Successfully alerted plant managers to impending bearing breakdowns weeks in advance, preventing catastrophic unplanned downtime and collateral motor damage.
- Industry Award: In 2023, CMServices Global Ltd / Simon Edmondson received an Award of Excellence recognizing the business as a “Global Provider of Innovation Services in Machinery Health 2023.”