Why it's needed

Manual setpoints don't match real-world meter behavior.

FlowCal is powerful — but its validation assumes your meter data is normalized. Real meter data is often skewed, multi-modal, and messy. NDL models the distribution each meter actually has, so you get defensible limits for every meter.

01

Assumes normal distributions

FlowCal validation assumes your data behaves like a normal curve. Many meters don’t — especially with operational changes and seasonal effects.

02

Percent-bad rules drive noise

Assuming a fixed percent of “bad” points floods analysts with fake exceptions and trains teams to ignore alarms.

03

Manual limits are slow and brittle

Tuning high and low limits one meter at a time is slow, hard to repeat, and breaks the moment behavior shifts.

04

Standard stats miss real issues

Basic min/max or standard-deviation rules can’t see complex patterns. NDL models the full distribution so you keep the real exceptions.

What that costs a measurement team

Every exception an analyst clears that wasn't really an exception is time taken from the one that was. At scale, the noise is the problem.

Analysts clearing noise

A few minutes per exception, thousands of times a month, is a full-time role spent on alarms that were never real.

Alarm fatigue

When most exceptions turn out to be nothing, the team stops trusting the queue — and the real problem waits.

Unexplainable limits

When nobody can say why a limit is what it is, every audit and every handoff starts from scratch.

What measurement teams get instead

Fewer false positives

Cut exception noise so analysts spend their time on real issues instead of clearing nuisance alarms.

Analyst time back

Automate the heavy lifting of setpoint tuning and recurring reviews — without adding headcount.

Defensible limits

Every recommendation is backed by the underlying meter data and model, giving you an audit trail for reviews.

Scales across assets

Apply one consistent, explainable methodology to meters across basins, systems, and operating regions.