Build your own metrics: from the signals you already have

Almost every number a business is managed by is calculated rather than measured directly. Energy per tonne, overall equipment effectiveness, yield per square meter, and cost per batch are all derived from data the plant already collects.

In many plants, those calculations still happen in spreadsheets: export the data, paste it into a file, apply the formulas, and repeat the process next week.

DataPilot’s formula engine moves those calculations onto the live data. You define the formula once, and the result becomes a metric like any other. It can be plotted on a chart, used in an alarm, and included in daily or weekly reports.

A spreadsheet formula, connected to live data

If you can write a formula in Excel, the idea will feel familiar. Select the inputs by the names they already use, combine them with arithmetic or functions, and give the result a name.

The difference is that the formula remains connected to the underlying signals.

A spreadsheet usually works on a snapshot: data exported at one point in time. A calculated metric in DataPilot updates as new readings arrive. It can also be calculated over historical data, so a new metric can immediately be viewed across the history of its input signals.

Alongside standard arithmetic, DataPilot includes functions designed for time-series data:

• daily aggregates a signal from midnight and resets at the start of the next day.
• MovAvg calculates a rolling average over a period such as 1h or 24h, helping distinguish sustained changes from short-term fluctuations.
• CumSum calculates a running total over a selected period, while CumSum_From can restart that total from an event such as the beginning of a production batch.
• STDDEV and TREND help describe how variable a signal is and how it is changing over time.
• IF, AND, MIN, and MAX provide familiar spreadsheet-style logic.
• VPD and DLI provide common greenhouse calculations without requiring users to recreate the formulas themselves.

Time periods are written directly as values such as 30m, 12h, or 7d. There is no need to convert a time window into a number of samples.

A greenhouse example: vapor pressure deficit

Growers measure temperature and relative humidity, but plants respond to the combined effect of both.

Vapor pressure deficit, or VPD, describes the difference between the amount of water vapor the air could hold and the amount it currently holds. It is widely used as an indicator of the evaporative demand experienced by the crop and is closely related to plant transpiration.

VPD changes with both temperature and humidity, and leaf temperature can make the value experienced by the plant different from one calculated from air temperature alone.

DataPilot has the calculation built in:

VPD({Leaf temperature}, {Air temperature}, {Relative humidity})

The result is expressed in kilopascals. When leaf temperature is available, it can be used to better represent the conditions at the leaf surface. Without a leaf-temperature measurement, air temperature can be used for both temperature inputs.

Two other common greenhouse metrics can be calculated just as easily: 24-hour average temperature and daily light integral.

MovAvg({Air temperature}, 24h)

DLI({PAR at canopy})

A manufacturing example: energy per tonne

On a pellet line, the raw signals may include production output from a belt scale and energy consumption from meters on major equipment.

Neither signal alone answers the question a plant manager usually cares about:

How much energy did it take to produce one ton of product?

That metric can be calculated by dividing energy consumption by production over the same period.

Each input is selected by the name it already has in DataPilot.

Before production begins, there may be no valid denominator, so the metric remains blank until enough data is available to calculate it.

The same signals can answer different questions by changing the time window.

For example:

CumSum({Dryer kWh}, 1h)

shows accumulated dryer energy over the last hour.

CumSum_From can restart the calculation at the beginning of a production batch, allowing the same approach to calculate energy per tonne for each batch instead of for the entire day.

That makes it possible to compare process efficiency across products, suppliers, moisture levels, or other production conditions.

Where a function requires a period, select the metric and specify the time window directly:

OEE, and why plants use it

A production line can appear busy while still losing a significant part of its productive capacity.

Overall equipment effectiveness, or OEE, separates that loss into three components:

• Availability: Was the equipment running when it was supposed to?
• Performance: When it was running, was it operating at the expected rate?
• Quality: How much of the output met the required quality standard?

OEE combines the three:

OEE = Availability × Performance × Quality

The overall number shows how effectively planned production time was used. The three components help explain why performance changed.

Each component can be calculated from information the plant already records.

For example, a run signal may come from the control system, with 1 representing running and 0 representing stopped. Reject quantities may come from a quality check entered by an operator. Planned production time and rated equipment capacity can be stored as reference values.

For a system where the run signal is sampled every five minutes:

daily({Press running}) * 5 / {Planned minutes}

daily({Pellets out}) / (daily({Press running}) * 5 / 60 * {Rated tons per hour})

(daily({Pellets out}) - daily({Rejects})) / daily({Pellets out})

These can be saved as Availability, Performance, and Quality, then combined:

{Availability} * {Performance} * {Quality}

An OEE decline caused by two hours of downtime requires a different response from one caused by lower production rate or increased rejects.

Plotting the three components together makes the cause visible before someone has to investigate the overall number manually.

What a calculated metric can do

A calculated metric behaves like any other metric in DataPilot.

It can be charted alongside its inputs. If energy per tonne increases, you can view it beside the energy meter, production output, and other related signals on the same timeline.

It can trigger an alarm. Alarm conditions use the same metrics, so a calculated value can be monitored just like a directly measured signal.

It can appear in reports. Daily and weekly reports can include calculated metrics alongside production, energy, downtime, and other operating data.

The point

Many of the metrics a business actually manages are not measured directly by an instrument. They are derived from other measurements: a ratio, a daily total, a rolling average, or a calculation across several signals.

Traditionally, much of that work has been done in spreadsheets using exported data.

The formula engine moves those calculations onto the live operational record.

Define the metric once using the names your team already knows, and from then on it can update automatically, appear on charts, trigger alarms, and feed reports.

If you would like to see your own metrics built from your plant data, book a demo.