Maintenance Mindset: Sensor placement and why chickens see better than your plant

Most sensor placement strategies fall into one of three patterns:

• Legacy placement: “That’s where we’ve always put it.”
• Convenience placement: “That’s where we can physically mount it.”
• Overcompensation placement: “Add more sensors and we’ll be safe.”

Each creates the same outcome: activity that resembles reliability, without actually improving it. You can have 98% of assets instrumented, 100% of routes completed, and thousands of data points collected and still miss the one signal that mattered because placement is not about quantity, it’s about distribution quality.

A lesson from biology: the exclusion zone

Researchers from Princeton University and Washington University in St. Louis studied how chickens arrange the cone cells in their retinas. Chickens have five types of color-sensing cones. These cells are not arranged in a grid. They are not random either. They follow a structure known as disordered hyperuniformity.

At first glance, the pattern looks irregular, but when measured across distance, it reveals something much more interesting. There is no clustering, no large gaps, and uniform coverage without rigidity.

Each cone maintains a small “exclusion zone,” preventing others of the same type from getting too close. The result is a system that maximizes coverage and resolution without redundancy. In other words, chickens solved a problem most plants have not and that’s how to distribute sensors so nothing important is missed.

The hyperuniform approach to sensor placement: 4 principles

Instead of thinking in terms of “more sensors,” think in terms of controlled distribution and the following four principles.

Principle 1: exclusion zones

Each sensor should “own” a region of detection. In vibration monitoring, this means not stacking sensors along the same transmission path unless justified. Also:

• Avoid placing similar sensors too close together.
• Define minimum spacing based on signal propagation and failure modes.

Principle 2: multi-type distribution

Different sensor types should interleave, not overlap. This mirrors the multi-cone system in the chicken retina, where each sensor type contributes uniquely to the whole. Also:

• Temperature, vibration, and acoustic sensors should complement one another.
• Each captures a different dimension of failure.

Principle 3: coverage uniformity over distance

Step back and evaluate your system at scale. Uniformity is not visual, it’s statistical. A plant can look well-instrumented and still have poor distribution. Also:

• Are there regions with no monitoring?
• Are some areas oversampled?

Principle 4: signal relevance over accessibility

Stop placing sensors where it is easy. Start placing them where failure initiates, such as:

• bearing load zones
• lubrication entry points
• thermal gradients
• flow disturbances.

If installation is difficult, that is often a signal that the measurement is valuable.

Reliability is a sampling problem

At its core, condition monitoring is not about sensors, it’s about sampling reality correctly. If your sampling is flawed, then:

• Data quality degrades.
• Diagnostics become uncertain.
• Decisions become delayed or incorrect.

This ties directly into what we see across industry:

• Data collected but not trusted.
• Trends identified but not acted upon.
• Failures labeled as “unexpected” when signals were simply missed.

The system did not fail. The sampling did.

From data collection to decision confidence

Better sensor placement does not just improve detection. It improves decision confidence. When distribution is correct, signals appear earlier, trends are clearer, false positives decrease, and intervention timing improves. This is the difference between, “We think something is wrong” and “We know what is happening and when to act.”

Like the chicken retina, the system does not draw attention to itself, it simply works.

Nature does not optimize for appearances; it optimizes for function. The chicken does not see well because it has more sensors; it sees well because those sensors are arranged correctly. Your plant is no different.

Because in the end, reliability is not about what you measure, it’s about whether you are measuring the right things, in the right places, at the right time.