CEMS Data Acquisition Software: Build Custom, Buy StackVision, or Build the Compliance Layer Above It

When is off the shelf genuinely the right call here?

ESC Spectrum StackVision, CMC Solutions and the ABB emissions packages already do the regulated arithmetic. Averaging, calibration drift application, missing data substitution, out of control tracking and the electronic quarterly file in the format the regulator accepts are solved problems, and rebuilding them buys no reduction in risk. If you run one or two monitored stacks under a single programme with modern analysers, one of these products plus a disciplined environmental team is the right answer, and we say so before quoting.

Buy, and stop reading here, if this describes you:

• One or two monitored stacks, all reporting under a single regulatory programme.
• Analysers of one or two vintages with a data acquisition system in front of them that already exposes validated values.
• A monitoring plan that has not been amended in years, and no permit renewal in progress.
• One state agency, so one set of permit conditions to hold in your head.
• An environmental team who can assemble a quarter by hand in a day when they need to.

At that shape the annual licence for a certified data acquisition and handling system (DAHS) is far cheaper than owning a rule engine, and the vendor carries the regulatory tracking burden on your behalf. That burden is real. Submission schemas get revised, subpart requirements change, and somebody has to notice and act inside a window that is rarely generous. Paying a vendor to do that at two stacks is a good trade.

Two further cases where buying is right even at larger scale. If nobody can state today which subparts and permit conditions apply to which stack, software will only make that gap visible in an audit finding, so close it first. And if your only motivation is that licence fees feel high, price the alternative honestly, because a rule engine you own still has to be maintained every time a schema moves.

When does a custom build actually pay off?

The certified products own the calculation. What they do not own is context: why a monitor was out of service, what maintenance was performed, when it requalified, and what the resulting numbers mean to the people running the units. That context lives in a maintenance system, a plant historian, an operator log and somebody's memory, and it is exactly what an agency asks for when a submission is questioned.

The gap widens with programme mix rather than with stack count. Every quote request opens with a stack count and it is the least predictive number in the conversation. Five stacks under one programme with identical analyser configurations is a smaller build than two stacks where one reports under an acid rain programme and the other carries a state permit with a rolling limit on a different averaging basis. In Digital Heroes delivery experience each additional regulatory programme adds $20,000 to $40,000, while an additional stack under an existing programme adds far less.

Build when two or more of these are true:

• Several facilities, so nobody can see fleet emissions position without a person assembling it from exports.
• A mix of federal programmes and state permit conditions on the same units, so no single vendor configuration covers everything without workarounds.
• An incumbent DAHS being sunset, where migration cost is approaching build cost anyway.
• A submission rejected or corrected in the last two years where reconstructing the cause took days.
• Emissions data that has to feed operational or allowance accounting systems a closed compliance product will not open up.

The first of these moves organisations most often. A single site tolerates manual assembly at quarter end. Eight stacks across three sites under three programmes does not, and the failure is a slow tax on the few people who understand the rules.

How do they compare on the things that matter in this industry?

Substitution and availability. This is the regulated core and the certified products handle it properly. It is also the single most expensive thing to build correctly, at $18,000 to $35,000 for the engine alone, because the applicable substitute depends on the monitor's recent availability history rather than on the reading itself. If you build, this is the last line to trim and the first to test.

Programme coverage. Packaged products are configured around a compliance model, and the configuration ceiling shows when a stack carries obligations from more than one rule family at once. Adding a state permit limit with an averaging period neither federal programme knows about typically means a vendor change request with a lead time. Owning the rule model removes the lead time and adds the maintenance.

Operator visibility. None of these products is trying to be an operations tool and it is unfair to criticise them for that. The practical consequence stands: operators watch analyser readings on the control system while compliance watches rolling averages under permit specific averaging periods, and those are different numbers. Sites discover a thirty day rolling exceedance at day thirty, when the only remaining option is a deviation report.

Release cadence. Submission formats and reporting requirements are revised on the regulator's timetable, and packaged vendors ship connectors on theirs. That is a reasonable commercial decision and an awkward operational one when the change window is short. Budget $8,000 to $20,000 a year for format changes whichever side of the decision you land on.

Evidence retrieval. An inspection picks specific periods and asks you to prove them. If that means joining a DAHS export to a maintenance report to a paper binder in the control room, you will spend days per sample. Storing downtime periods as records with a cause, a linked work order, the corrective action and the requalification test turns the same request into a query.

Data portability. Emissions records have to survive an audit years after the engineer who built the system left. Ask how the complete validated history leaves the product, including quality assurance events and not just hourly values, and get the answer into the contract.

What does total cost of ownership look like at your scale?

On the build side, from Digital Heroes delivery experience, a compliance visibility layer over an existing DAHS runs $50,000 to $85,000. A first production release that owns averaging, substitution and quarterly reporting for a site runs $90,000 to $170,000 and ships in 14 to 20 weeks. A fleet platform across several facilities and programmes runs $220,000 to $500,000 phased over 9 to 14 months.

Component by component, so you can fund only what you need: analyser and data acquisition interfacing at $5,000 to $12,000 per interface, the averaging and substitution engine at $18,000 to $35,000, each additional regulatory programme at $20,000 to $40,000, quarterly report generation in submission format at roughly $22,000, quality assurance event scheduling and relative accuracy test audit (RATA) records at roughly $16,000, and a parallel run against the incumbent for a full quarter at $12,000 to $25,000.

Annually after go live, plan on support at 15 to 20 percent of build cost, regulatory format changes at $8,000 to $20,000, hosting and records retention at $4,000 to $12,000, $6,000 to $15,000 per permit change that introduces a new limit or averaging period, and $5,000 to $12,000 each time an analyser is replaced, because a new analyser is a new interface and a new set of calibration behaviours.

On the buy side, the licence is only part of it. Add the staff hours spent assembling explanations the product does not hold, the vendor change requests when a permit is modified, and whatever the state agency requires if you ever want to retire the certified system. Then set both figures against the exposure nobody prices, which is a submitted quarter that has to be corrected in public.

What does the hybrid look like, and when is it the honest answer?

For most multi site operators this is the answer, and we give it whether or not it wins us work. Keep the certified DAHS as the system of record for the regulated calculation and build the visibility and evidence layer above it, at $50,000 to $85,000, leaving the recertification exposure alone.

In practice that layer is three pieces:

• Downtime and quality assurance records. Every gap in valid data becomes an object with a cause, a linked work order, the corrective action, the requalification test and the substitution applied, assembled as the outage happens rather than reconstructed under deadline pressure.
• Continuous averaging with forward projection. Every permit averaging period computed alongside the certified system and pushed to operations, so a rolling limit becomes an operating decision days before it becomes a deviation report.
• Fleet reconciliation and reporting. Your own calculation compared against the certified output, with every divergence visible and explained, plus a single fleet position nobody has to assemble by hand.

Keeping the certified path narrow is the largest cost saving available in this category, and it also tells you which reports genuinely hurt before you commit to owning a rule engine. Anyone offering to replace your DAHS with a custom build on day one is selling you regulatory risk you are not being paid to take.

Which should you choose, by operator size and stage?

Find your row and act on it.

• One or two stacks, single programme, modern analysers. Buy StackVision or a comparable certified product. Spend the difference on analyser maintenance.
• Three to five stacks, one or two sites, still a single programme. Still buy, and add the visibility layer only if quarter end is consuming more than a week of your environmental team's time.
• Several stacks with a mix of federal programmes and state permit conditions. This is the decision point. Keep the certified system and build the downtime, averaging and reconciliation layer above it, roughly $50,000 to $85,000 depending on site count and analyser diversity.
• Multi site fleet, incumbent DAHS being sunset, or emissions data needed by operational systems. Build properly and phase it, but keep the certified path for the first release and run a full parallel quarter before retiring anything.

Two conditions apply to every build row. Cut over at a quarter boundary rather than mid quarter, and accept slipping a full quarter rather than rushing a partial one. And develop the rule engine against written test cases derived from the monitoring plan and the permit, not from a developer reading of the regulation, because those test cases are also the artefact that survives the retirement of the one engineer who understands substitution.

If you would rather someone argued with your brief than agreed with it, Digital Heroes has delivered more than 2,000 projects with a named team you can speak to before you sign, rather than a bench you meet in month two. You keep the specification either way.