The Illusion of 95% Availability
Most industrial operators celebrate when a plant achieves 95% availability. But here's the truth: High availability does not guarantee high value.
A compressor can be "available" 98% of the time, yet underperform due to degraded efficiency, poor configuration, or suboptimal scheduling — all while costing millions in lost production.
At Knar, we don't measure success by uptime. We measure it by Expected Value: the probabilistic forecast of financial performance under real-world uncertainty.
The Problem with Traditional RAM Metrics
Reliability, Availability, Maintainability (RAM) studies are essential — but they often stop at physical metrics:
- 1"Mean Time Between Failures"
- 2"System Availability"
- 3"Downtime Hours"
These tell you what happened, not what it cost.
Worse, they create a false sense of security. A system may appear reliable on paper, but if its failures occur during peak pricing windows, the business impact is catastrophic.
Expected Value: The Missing Link Between Physics and Finance
Expected Value integrates:
Technical Reality
Failure modes, degradation, maintenance logistics
Operational Constraints
Crew availability, spare parts, feed variability
Market Dynamics
Commodity prices, demand cycles, contractual penalties
It answers the question: "What is this asset expected to earn over the next year, given all sources of uncertainty?"
This is not speculation — it's simulation. Using tools like AspenTech Fidelis and PDEL®, we build models where every failure mode has a dollar value.
Case Example: Compressor Performance Beyond Uptime
In one project, a client reported 96.2% availability for their gas compression system. Sounds good — until we modeled Expected Value.
We discovered:
- •Degraded compressors were operating inefficiently 40% of the time
- •Maintenance was scheduled during high-demand periods
- •Feed variability caused frequent derates not captured in standard RAM reports
Result:
Despite high "availability," the system delivered only 78% of potential EBITDA.
By optimizing configurations and maintenance timing using stochastic programming, we increased Expected Value by 22%.
How We Build Expected Value Models
Our approach follows five steps:
Map the Causal Chain
From equipment failure → system unavailability → production loss → revenue impact (using PDEL®)
Integrate Probabilistic Scenarios
Simulate thousands of operational futures
Link to Market Data
Apply price curves, contract terms, and risk profiles
Validate with Historical Performance
Ensure model fidelity
Institutionalize the Metric
Embed Expected Value into decision-making systems (iDSS)
This transforms RAM from a compliance exercise into a strategic tool.
Conclusion: Measure What Matters
In complex industrial environments, simplistic metrics fail. If you're managing assets based on availability alone, you're flying blind.
At Knar, we engineer clarity. We don't report uptime — we forecast value.
And we do it with precision, traceability, and no tolerance for oversimplification.