From Signal to Standard Work, Part 2 of 4. One upstream operator, one year of multi-sensor emissions data.
Part 1 followed one upstream operator through the unglamorous first stage of a multi-sensor detection program: reconciling 1,039 detections against a drifting picture of their own assets. Here is what they found once the picture was true.
You’re probably chasing the wrong thing
When ranked by total count, fugitive emissions look like the biggest culprit, with 268 events, roughly a third of all confirmed emissions.
It has to be taken in context, because counting how often something happens and measuring how much gas has been released are two different rankings. Fugitives had the lowest median rate of any category, at 73 kg/hr, while compressor events and blowdowns ran 162 to 176 kg/hr, more than double. Ranked by how often they happen, fugitives win. Ranked by how much gas actually escapes, compressors do.
Chase the tally of leaks and you spend the year on smaller, more frequent ones. Follow the rate instead, and the priority becomes clear quickly. More than half of the blowdown events traced back to compressors, and 62% of those compressor events sat in one operating region. What the operator thought was an emissions tracking program was, largely, a compressor reliability program too. That made environmental compliance an operations problem, with owners, budgets, and equipment classes to go with it.
The finding
When you can connect your emissions data to operations, your environmental compliance becomes an operations program.
66% fewer detections in a year
The rollup showed something else too. Detections fell 66% over the year, for two different reasons. Repeat sites got fixed, so the real leaks stopped coming back. And as the asset records got reconciled, the false alarms that used to send a crew chasing gas that wasn’t theirs, or a site that had already been sold, stopped being logged against them in the first place.
The physical fixes and the data fixes showed up the same way: fewer crew dispatches, and fewer hours burned visiting the same site twice. Detection connected to historical and production data, integrated with the field work, led to compounding rewards. That is the payoff every operator running detection is trying to reach.
A slide can’t dispatch a crew
“Most of the compressor emission volumes sit in one region” is the kind of line that ends up on a quarterly-review slide. No one in the field can act on it. It doesn’t tell management whether there is a reliability problem or simply more compressors in that region. For it to mean something, someone has to connect the data to the macro view, then connect back to the field, assigning work and crews to tackle it compressor by compressor.
That step, from the insight to the instruction, is where most operators stall, and where most analytic tools declare their job done.
This operator didn’t stop at the slide. If you are staring at the same gap between the insight and the instruction, talk to an engineer about what it takes to close it. Part 3 is where the analysis leaves the spreadsheet.