SmartCEM-75 models match NOx analyzers on eight Houston turbines

EPA found that SmartCEM-75 models trained on eight NRG turbines in Houston matched NOx analyzers over 720 hours each. NRG has reported NOx with them since 2009 for six units and 2011 for two.

  • 1 an organization

    NRG has reported the eight turbines' NOx with the software since 2009 and 2011, after EPA found that it matched the analyzers.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported benefit.
Evidence

foundstudiedreportedclaimed

Two independent regulatory documents support the benefit.

Compared with

Hourly readings from certified NOx analyzers on the same units over 720 operating hours each, judged by EPA's three statistical tests. Measured.

Without the system
A linear correlation coefficient (r) above 0.8 required
With it
r = 0.936 to r = 0.999 across the eight units
The system’s part
Not isolated. EPA measured each model's accuracy against a certified analyzer on the same unit over 720 hours. It did not compare monitoring with and without the software in use. EPA describes statistical hybrid models developed from each unit's training data and does not call them neural networks.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Persisted, 21 May 2025. EPA wrote in May 2025 that NRG had used the PEMS to report NOx since 2009 for six units and since 2011 for the other two. It cut the required audits from monthly to quarterly.

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
SmartCEM-75 predictive emission monitoring systems (PEMS), one model per turbine
Made by
CMC Solutions
Used by
NRG Texas Power (NRG Energy), T.H. Wharton Generating Station Units THW31 to THW34 and THW41 to THW44
Where
Houston, Texas, United States
Stage
In routine use
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Estimating the Quantifiable Characteristics of Products, Events, or Information, Monitoring Processes, Materials, or Surroundings
Collections
Scrutiny and findings
Last revised
31 October 2026

Sources

  1. Regulatory document, by authors with no tie to the system.

    U.S. Environmental Protection Agency, Clean Air Markets Division. Updated Approval of the Predictive Emission Monitoring Systems Installed on Units THW31 through THW34 and Units THW41 through THW44 at NRG Texas Power's T.H. Wharton Generating Station [Facility ID (ORISPL) 3469]. Letter to NRG Texas Power, 27 September 2010.

    The PEMS NOx lb/mmBtu output on each unit passed each of the three statistical tests for all unit operations.

  2. Regulatory document, by authors with no tie to the system.

    U.S. Environmental Protection Agency, Office of Air and Radiation. Request for Changes to the PEMS Approval for Units THW31 through THW34 and THW41 through THW44 at T.H. Wharton (Facility ID (ORISPL) 3469). Letter to NRG Energy, 21 May 2025.

    Based on the very high observed passing rate for RAAs by all units with approved PEMS, EPA has determined that the monthly RAA frequency may be reduced to quarterly while continuing to ensure that the reliability of PEMS data is comparable to the reliability of data obtained using CEMS, consistent with the purposes of part 75, subpart E.

Details

AI system

Workflow

Temporary certified analyzers gave hourly reference data while each model was trained and tested. CMC Solutions built a separate model for each turbine from that unit's training data. NRG later submitted expanded training data and new operating ranges for all eight units.

Follow-up

Conditions for replication

Inputs must stay within the range of the training data, or NRG must report the unit's maximum potential NOx rate. A unit whose combustion system changes must have its PEMS recertified before its data count.

Entry details

Sector

Environment and climate.

Purpose

CMC Solutions built SmartCEM-75 to predict a unit's NOx emissions from its sensor readings so that the prediction can serve as the unit's emissions monitor.

Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). SmartCEM-75 models match NOx analyzers on eight Houston turbines (AIBD-0845). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0845.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions