Air Emissions Compliance Software: Why Do You Find Permit Deviations Six Weeks Late?

The six week gap between the deviation and the discovery

A heater's fuel gas sulphur content drifts up over a weekend. The permit limit is evaluated as a rolling average. The data exists, in the historian, second by second, and has existed the whole time. Nobody looks at it against the limit until an environmental engineer opens the quarterly workbook and starts building the semiannual monitoring report.

By then the deviation is six weeks old. You now have three problems instead of one. The exceedance itself, which might have been correctable in an hour if anyone had known. The reporting obligation, which in most permits requires prompt deviation reporting on a defined timeline that started when the deviation occurred, not when you found it. And the annual compliance certification, which a responsible official signs personally, and which now has to describe a deviation you cannot explain because the board operator who was on shift has rotated out and the process conditions that caused it were never captured.

This is the operating reality at most refineries and power plants we have worked with. The data is excellent and the evaluation is manual. Every compliance manager knows this. The reason it persists is that the calculation methods are written into the permit in prose, unit by unit, and translating that prose into a running calculation is exactly the work that never gets prioritised until an agency inspection or a consent decree forces it.

Your permit is a specification, and it is unlike anyone else's

The reason generic environmental software struggles here is not sophistication, it is specificity. A Title V permit for a refinery is a document of hundreds of pages that says, for this heater, this limit applies, evaluated over this averaging period, calculated by this method, using this monitoring, with these exemptions during startup, shutdown and malfunction, and reported in this way.

The methods vary wildly across units in the same plant. One unit is governed by a continuous emission monitor with data availability and substitution rules. Another uses a parametric monitoring approach where you demonstrate compliance by holding process parameters inside an operating envelope established during a stack test. Another uses published emission factors multiplied by throughput. Another uses a mass balance. Storage tanks use their own estimation routines. Flares carry combustion zone gas requirements evaluated on short blocks. Fugitive emissions come from a leak detection and repair programme with its own monitoring frequencies and repair clocks.

Then there is time. A twelve month rolling total for a limit that determines whether you remain a synthetic minor for a pollutant is a completely different calculation from a three hour block average, and both may apply to the same unit for different pollutants. A rolling calculation also has to be recomputed backwards when a historian value is corrected or a monitor is invalidated after a failed quality assurance test, which is the part spreadsheets handle worst.

Where Sphera, Intelex, Cority and Envirosuite stop

• Sphera has real depth in industrial environmental and process safety and is a credible enterprise choice. The air module works as a configurable calculation engine, which means your permit methods have to be expressible in their framework, and where they are not you get services work or a workbook beside the system. Historian integration is a project in every implementation we have seen, not a connector you switch on.
• Intelex is a broad, well built EHS and quality platform, strongest in incidents, audits, actions and management systems. Air emissions calculation at unit level against permit specific methods sits at the edge of its natural strength, and the configuration effort to get there is substantial.
• Cority comes from an occupational health heritage and covers environmental modules competently. Same structural point: the engine is generic by necessity and your permit is not.
• Envirosuite is genuinely good at a different problem, namely ambient monitoring and dispersion around a site, including community facing odour, dust and noise. If your issue is neighbours and ambient impact, look at it seriously. It is not a permit limit compliance engine.

The recurring outcome is a suite that holds your obligations, your audits and your actions, with the actual emission calculations still running in workbooks because that is the only place the permit specific method could be expressed. If that describes you, a targeted build that owns calculation and deviation detection, feeding your existing suite for action management, is often the right answer rather than a rip and replace.

What a custom air compliance system must include

Start with the permit as structured data. Every limit becomes a record: the unit it applies to, the pollutant or parameter, the numeric value and units, the averaging period and how it is computed, the monitoring basis, the applicable exemptions, the citation in the permit and the effective date range. This register is the spine of the system, and building it is a serious exercise involving your environmental engineers reading the permit line by line. Expect three to six weeks of that work on a large site. It is also the most valuable artefact the project produces, because half the time it surfaces limits nobody was actively tracking.

Version everything by effective date. Permits get renewed, modified and amended. A limit that changed on renewal must not retroactively rewrite whether you complied last year. Every evaluation records the limit version it ran against.

Then connect to the historian. AVEVA PI, Honeywell PHD, Aspen IP.21 and Wonderware are the common ones, and each has a supported access path. The engineering discipline that matters is tag mapping: each calculation input names a specific tag, with unit conversion, expected range, and a rule for what happens when the tag is bad or missing. Tag mapping done casually is the single biggest source of wrong numbers in this category, because a tag renamed during a control system upgrade will silently feed a calculation with nothing and produce a comfortable zero.

The calculation engine has to be transparent rather than clever. Every evaluated value should be reproducible: here is the limit, here is the averaging window, here are the input values that were used, here is the arithmetic, here is the result, here is whether an exemption applied. When an inspector asks how a number was derived, you show the record. When a value is corrected upstream, the affected evaluations recompute and the change is logged rather than overwritten.

Deviation detection then becomes continuous instead of quarterly. A limit approached triggers a warning to the unit engineer while there is still time to act. A limit exceeded creates a deviation record with the timestamp, the data, the operating context pulled from the historian around that window, and a workflow with the prompt reporting clock visible. That last detail is what changes behaviour: the clock starts at occurrence and the responsible official can see it.

Monitor quality assurance belongs in the same system. Relative accuracy test audits, calibration and linearity checks, and data availability all have schedules and consequences. A missed check can invalidate data, which reaches back into every rolling calculation that consumed it. Systems that treat quality assurance as a separate calendar miss that dependency.

Finally, reporting. Semiannual monitoring reports, the annual compliance certification, the annual emission inventory to your state, greenhouse gas reporting where applicable, and any consent decree specific reports should assemble from the same evaluated data rather than from a parallel spreadsheet exercise. The certification in particular deserves a defensible assembly trail, because a named individual signs it.

Cost, timeline and what moves the number

A first release with the permit limit register, historian integration, calculation and continuous evaluation for your highest risk limits, and deviation workflow runs $65,000 to $130,000 over 12 to 16 weeks. Scoping that first release to twenty or thirty limits that carry the real risk is the right move, not attempting the whole permit at once. The full platform adding the site emission inventory, monitor quality assurance scheduling, leak detection and repair, all statutory reports and greenhouse gas reporting runs $170,000 to $380,000 phased over 6 to 12 months.

What raises the cost: the number of distinct calculation methods, which correlates with unit count but not linearly, since twenty identical heaters are one method. Historian access complexity, especially where the control network is segmented and data has to cross a boundary under your cyber security rules, which is a legitimate constraint and adds architecture work. Leak detection and repair, which is effectively a separate field application with its own monitoring and repair clocks. Multiple sites under different permits, where each permit is a fresh register. And consent decree obligations, which are bespoke by nature.

What lowers it: pick the limits where a deviation would actually hurt. Most sites have a handful that carry most of the exposure, and everything else can join later on the same engine.

When you should not build this

Do not build if you hold a minor source permit with a few units and annual reporting. The calculation load fits a workbook and your risk profile does not justify the engineering.

Do not build a full platform if you are already standardised on Sphera, Intelex or Cority and using them well for obligations, actions and audits. Build the calculation and deviation engine only, and feed the suite. That is a smaller project with most of the value.

Build when you hold a Title V permit with many units and varied calculation methods, when your rolling averages are computed after the fact, when a deviation has been discovered during report preparation in the last two years, when you operate under a consent decree, or when the person who understands the spreadsheets is one retirement away from taking the method knowledge with them.

How to choose a developer for emissions work

Ask them to walk through how they would implement one twelve month rolling limit, including what happens when a historian value from four months ago is corrected. If recomputation and an audit trail are not in the answer, the system will produce numbers you cannot defend.

Ask how they handle startup, shutdown and malfunction periods and other permit exemptions. If exemptions are not modelled explicitly as part of the evaluation record, your deviation list will be full of false positives and your engineers will stop trusting it within a month, which is worse than no system.

Ask specifically which historians they have connected to, in which security architecture. Reading tags across a segmented control network is a real constraint and the answer should reflect that they have done it rather than assumed a database connection.

Ask who owns the code, the database and the cloud accounts, in writing, before kickoff. Compliance calculations underpin certifications signed by a named official and may be examined years later. At Digital Heroes the client owns the repository from the first commit. A good starting point is to send us your permit and a tag list for one complex unit, and we will show you exactly how that unit's limits would be modelled and evaluated.