Leading productive & meaningful lives

[dibyendu]

Pipe Schedule in Lubrication Systems — What Really Matters

1. First Principle: Velocity Depends on Internal Diameter, Not Nominal Size Two pipes with the same nominal size but different schedules have different internal diameters. Example (for illustration): PipeOD (same)ScheduleID (approx)Effect2” pipesameSch 40larger IDlower velocity2” pipesameSch 80smaller IDhigher velocity Since: V = {4Q}/{pi D^2} A small reduction in ID → large increase in…

[dibyendu]

Beyond Sensors. Beyond AI.

Introducing a New Era of Engineering Intelligence with RAPID AI Most industries today are trapped in a false choice: Invest heavily in sensors and infrastructureOROperate blind and reactive But what if you didn’t have to choose? RAPID AI: Dual Intelligence by Design RAPID AI is built on a fundamentally different philosophy: 1. Sensor-Driven Intelligence Leverages vibration, acoustic…

[dibyendu]

Lubricant Line Velocities — Why They Matter in Reliability Engineering?

Lubricant velocity in oil circulation systems is not just a piping-design number. It controls pump suction stability, oil delivery, heat removal, drain-back behaviour, foaming tendency, pressure loss, bearing temperature, filter performance, and long-term reliability. The practical ranges you quoted are broadly consistent with common hydraulic/lube-oil guidance: Line typePractical velocity…

[dibyendu]

Electrical Bearing Failures in VFD Motors: The Hidden Reliability Killer

Introduction The adoption of Variable Frequency Drives (VFDs) has transformed industrial operations—improving efficiency, control, and flexibility. However, it has also introduced a new class of failure mechanisms, often misunderstood and misdiagnosed. One of the most critical among them is: Electrical erosion of motor bearings The bearings fail. Not randomly. Not mysteriously.But silently……

[dibyendu]

Machines Don’t Fail — Systems Do.

When Machines Don’t Fail — Systems Do.
A Hard Lesson from the Recent Boiler Blast. The recent boiler explosion at a Vedanta power plant is not just a tragic industrial accident. It is a stark reminder of something we often overlook: Catastrophic failures are rarely caused by a single component.They are caused by interdependencies going unstable. Initial findings point toward a sequence many…

[dibyendu]

The Biggest Problem with Industrial AI is NOT Technical. It’s Human.

We don’t have a data problem. We don’t even have an AI problem. We have a trust problem. Think about this. You are a maintenance planner. A system tells you: “Anomaly probability: 87%.” Now what? Do you shut down a unit losing $500,000 per day? Do you wait? Do you ignore it? Because that one decision… can make or break your month, your plant performance, even your credibility. This…

[dibyendu]

The Scale of Opportunity in Maintenance

Every year, industry quietly burns the equivalent of 12 aircraft carriers… or 2,000 F-35 fighter jets… simply because we fail to convert data into engineering decisions. Let that sink in. The global industrial maintenance spend is about $450 billion annually. Out of this: $157 billion is lost in reactive maintenance → ~12 aircraft carriers → ~2,000 F-35 jets $180 billion is spent on…

[dibyendu]

Prevention of failures requires more than detection. It requires understanding the physics of why energy cannot flow freely through a machine’s elements, classifying the dynamics of how failures initiate and accelerate, and prescribing actions at the right level — from immediate intervention to permanent design-out. This understanding organizes into three tracks of increasing leverage:

[dibyendu]

The Data Paradox

Modern industrial plants do not lack data. They drown in it. A single gas turbine generates 500+ sensor readings per second. A medium refinery with 2,000 monitored assets produces terabytes of vibration, temperature, pressure, and process data every month. The cost of sensor hardware has dropped 90% in the last decade. The cost of data storage has dropped even further. Organizations that once…

[dibyendu]

The four generations share a common limitation: they all focus on the wrong question. Reactive asks: “What broke?” Preventive asks: “When should we replace it?” CBM asks: “Has a threshold been exceeded?” Pattern AI asks: “Does this look like something we’ve seen before?” None of

[dibyendu]

Four Generations of Maintenance — And Why Each Falls Short

Industrial maintenance has evolved through four generations, each an improvement, none sufficient. Generation 1: Reactive Maintenance Fix it after it breaks. This was the norm for most of the twentieth century. It is still practiced in plants that cannot afford better, or that have not yet paid the price of a catastrophic failure. It is the most expensive strategy per unit of uptime and the…

[dibyendu]

Why we miss detecting failures? And the IAR method

Machines do not fail suddenly. They whisper first. A bearing running toward failure radiates faint high-frequency energy weeks before it seizes. A misaligned shaft produces a subtle 2x harmonic that hides beneath the noise floor until it doesn’t. A pump starved of net positive suction head cavitates with a characteristic broadband high frequency hiss that an experienced ear can catch, but that…

[dibyendu]

Cost of Failures -- Direct and Hidden

The Cost of Failure — Direct Every minute a machine stands still, money burns. Unplanned downtime costs industrial plants more than $50 billion annually worldwide. A single gas turbine trip can cost $500,000 per day in lost generation. A pump failure cascade in a refinery can halt an entire process unit, burning through $1 million before the root cause is even identified. In mining, a SAG mill…

[dibyendu]

What a Stable System Looks Like (When Nothing Is Wrong)

Most of my professional life has been spent looking at machines only when something was wrong. Today, I want to talk about the opposite — what a stable system looks like when nothing is asking for attention. Stability is usually invisible — A stable system rarely announces itself. There are no alarms, no urgency, no dramatic signals. In fact, stability often looks boring — and that’s precisely…

[dibyendu]

What Exactly Do We Measure & Infer in Vibration? - Part 1

At first glance, vibration measurement appears simple—we measure acceleration, velocity, or displacement — the parameters of any dynamic movement, which causes a machine to either perform or fail. But in reality, what we are observing and infering something far more profound, which are the following: 1. The Energy — where Strain Energy is proportional to the square of displacement; Kinetic Energy…

[dibyendu]

Weibull, Hazard, RUL, Bayesian & Monte Carlo in an IIOT environment -- a simple understanding

Imagine You Have Many Toy Cars You have many toy cars Some break early Some run for a long time Some suddenly stop working Now you want to answer: “When will THIS car break?” Step 1: Learn from all old toy cars (Weibull) You look at all your old broken cars and ask: When did they break? Which ones lasted long? Which ones broke quickly? From this, you learn a pattern: Some cars break…

[dibyendu]

The Nature of Failure