Predictive Emissions Monitoring System (PEMS) - ladco

Software CEM ® 

Predictive Emissions Monitoring System 

(PEMS) 

Copyright © 2009 Rockwell Automation, Inc. All rights reserved.

Copyright © 2009 Rockwell Software. All rights reserved. 

LADCO Workshop 


Predictive Emissions Monitoring System 

(PEMS)

(Confidential – For Internal Use Only) 

Agenda 

1. Introduction to Software CEM 

2. Software vs. Hardware 

3. Software vs. Software 

4. Summary 

5. Case Studies 

6. Discussion


Industry Challenges 

With regulations continually become more stringent 

many industries and processes now have the 

responsibility to account for their impact on the 

environment by replacing an aging system or 

implementing a new monitoring system that: 

– meets all regulatory demands 

– provides the highest levels of accuracy and reliability 

– ensures a sustainable long-term solution 

– minimizes the total cost of ownership


Solution – Software CEM ® 

A patented, nonlinear hybrid model-based, Predictive Emission 

Monitoring System (PEMS) that provides an alternative to hardwarebased 

Continuous Emission Monitoring System (CEMS). 

– Meets all performance criteria for regulatory certification 

– Cost effective solution powered by Pavilion8 software platform 

– Same platform for Model Predictive Control & Optimization 

– Rockwell Software is an active expert in 

technical support to regulatory 

agencies worldwide. 

• Patented - 1 st introduced in 1993 

• Over 250 installations in 

• Diverse applications

Solution - Software CEM ® 

• Software CEM uses a nonlinear hybrid modeling methodology 

– Models can be empirical, 1 st Principles based, or a combination of both. 

• Rockwell Software interfaces to all major DCSs, PLCs, and historians 

– Works with any recording and reporting system 

• Have interfaced with software developed in-house by end users 

– Pavilion OPC interface makes this especially easy by allowing users to interface 

with our Software CEM technology without the need to modify source code for 

over 1000 applications 

Copyright © 2009 Rockwell Software. All rights reserved.

Software CEM ® Applications 

Sources 

• Boilers 

• Turbines 

• Duct Burners 

• Dryers 

• Thermal Oxidizers 

• Reciprocating Engines 

• Regenerative Thermal 

Oxidizers 

• Process Heaters 

• Olefin Furnaces 

• Crude Heaters 

Industry Independent 

Fuel Types 

• Natural Gas 

• No. 2 & 6 Fuel Oils 

• Other Fuel Oils 

• Refinery Gas 

• Process Gas 

• Combined Fuels 



Software CEM ® Advantages 

Cost Effective Monitoring & Compliance 

• A proven, model-based, software solution that leverages existing 

process instrumentation and data 

• Does not require additional capital equipment or real estate 

• Scalable solution that supports multiple Software CEMs in a 

standard office environment PC 

Highly Accurate & Reliable Emissions Monitoring & Reporting 

• Detects & alerts operator to sensor failures 

• Meets strict continuous emissions monitoring data availability 

requirements of 95% or more 

• Continues to operate in the event of sensor/instrument failure 

• Accurate emissions modeling with robust prediction capabilities 

Environmental and Industry Experience & Expertise 

• Proven solutions that meet unique, local regulatory requirements 

• Provides added confidence and reduced risk exposure

The Pavilion Software CEM ® 

Solution 


Consists of an emission model that 

reflects the relationship between 

process operations and ambient 

conditions to emissions. 

A patented, sensor validation 

system to ensure the accuracy of 

the predicted emissions data.

Software CEM ® - 

Sensor Validation 


• Definition 

– An algorithm imbedded in the emissions 

model to: 

• check raw inputs for validity 

• reconstruct variables that are out of 

specification or missing 

• predict what the failed sensor value should be. 

• Method 

– Development of a model for each key input 

– Model inputs accuracy with R 2 >95% 

• Results 

– Sensor Validation is a unique patented process that provides consistent 

achievement of the USEPA mandated 95% uptime


Sensor Validation Model 


This graph represents the real 

sensor (fuel flow) and the 

reconstructed value. 

This graph illustrates the other 

sensor values being used during 

the reconstruction period

Software CEM ® – 

Sensor Validation & PS-16 


• Sensor Validation is a requirement of PS-16: As defined in section 6.1.8 and 9.2: 

– “6.1.8 Sensor Evaluation System. Your PEMS must be designed to perform automatic 

or manual determination of defective sensors on at least a daily basis. This sensor 

evaluation system may consist of a sensor validation sub-model, a comparison of 

redundant sensors, a spot check of sensor input readings at a reference value, 

operation, or emission level, or other procedure that detects faulty or failed sensors. 

Some sensor evaluation systems generate substitute values (reconciled data) that are 

used when a sensor is perceived to have failed. You must obtain prior approval before 

using reconciled data.” 

– “9.2 Daily Sensor Evaluation Check. Your sensor evaluation system must check the 

integrity of each PEMS input at least daily.” 

• Rockwell Software provides a patented, real-time continuous sensor validation 

route with the ability to reconstruct data by using an independent model for each 

sensor. This virtually eliminates any missing data.


Model Validation 

• Model Validation 

– A standard feature on Software CEM that provide a daily check on the model’s 

integrity; verifying that no unauthorized changes have occurred 

– Note: the model does not change only the process changes with time 

• Definition 

– A routine that applies known sensor input values and verifies the results against 

known output values 

• Method 

– Predetermined input values are applied for low, medium, & high ranges 

– Output values are then calculated by the model 

– Calculated values are then compared to the known output values developed during 

modeling and RATA 

• Results 

– An equivalency to an EPA mandated Quarterly audit. 


Software CEM ® RATA Accuracy 

within + 7.5% Scale 



Certification Rate – 100% 

• Based on a USEPA Review of Software CEM Installations 

– Initial Certification Tests 

• Average Relative Accuracy (20% or less is goal): 

– NOx –6.4%, O2 –4.9% 

• Less than 7.5% Relative Accuracy 

– NOx – 97 (66%), O2 – 105 (72%) 

– Ongoing Relative Accuracy Tests 

• Average Relative Accuracy: 

– NOx –3.2%,O2–3.9% 

• Less than 7.5% Relative Accuracy 

– NOx – 132 (95%) 

– O2 – 119 (86%) 

– No failures 

– Demonstrated as accurate as CEMS over tougher test 


Sustaining Software CEM ® 

• Retaining accuracy over long periods of time is a function of physical 

equipment change and not the software application 

– Greater than1,600 stack years 

– Greater than 2,000 RATA (without failure) 

• Greater than 100 of our installed Software CEMs are customer or 

independently maintained and supported 

– Greater than 40% of all installations 

100% data availability lowers the risk of inaccurate reporting with the use of continuous, 

real-time sensor validation 


Software CEM 

vs. 

Hardware CEM 



Lower Total Cost of Ownership 

Cost Analysis Based on USEPA CEM Model 


Increasing Cost of 

Hardware Validation 

• Hardware CEM (analyzers and system) accuracy is verified daily due to 

equipment drift and aging 

– Verification is performed with costly Calibration Gas 

• Typical 2% certified gas is used 

– Uptime as required by agencies is >95% 

• Continuous routine maintenance is required for operation 

• Yearly sustainable operating costs are more then double that of Software CEM 

• Hardware CEM Linearity is verified quarterly via a Calibration Gas Audit 

(CGA) 

– Verification is performed with multiple Calibration Gases that are different then the 

gases used for Daily Cal-Check gas 

• 3 level (low, mid, & high range) gases are used for linearity 

– 1% Protocol gas is used 


Software CEM ® vs. 

Hardware CEM Validation 

• Software CEM sensor validation routine is equivalent to the daily Cal-gas 

Check for hardware CEMS 

– Software CEM validates sensors once per minute versus hardware CEM once every 

24 hours 

• 1440 times to 1 in a 24 hour period. 

– Uptime as required by agencies is >95% 

• Hardware CEM targets 95% (98% for Part 75) 

• Software CEM is near 100% 

• Software CEM Model Validation routine is equivalent to the CGA for a 

hardware CEM 

– Software CEM validates the model once per day versus hardware CEM once per 

every calendar quarter 

• 90 times to 1 in a calendar quarter 


Copyright © 2009 Rockwell Automation, Inc. All rights reserved. 

Model-based Software 

CEM 

vs. 

Other Software 

Methodologies

Software CEM ® Methodologies 

• Nonlinear hybrid modeling methodology versus a 

statistical methodology (i.e. table look-up) 

– A true Software CEM model is robust for a wider range outside of 

the RATA range 

– The table look-up is similar to a classic data substitution 

method and is more dependent upon the amount and diversity 

of data collected 

• A table look-up method cannot interpolate or 

extrapolate operating conditions that are not already in 

the table 

• Both methodologies can be built directly from historical 

data (provided hardware-CEMS data is available) and 

require no additional inputs 


Software CEM ® Accuracy 

• Nonlinear Hybrid modeling methodology employs extrapolation (an 

accepted practice by PS-16) that allows initial testing to be held at a 

minimum by reducing the number of test points required during model 

development 

– Table Look-up systems must include the entire range of the measurement envelope 

in order not to have “holes” in the data table. These “holes” will result in unsupported 

data in the table and result in a missing data flag. 

Look-up table data does not extrapolate thus requiring more testing than an empirical 

equation-based models 


Ambient Conditions – 

Compliance Accuracy 


• The Pavilion modeling methodology does not require additional input 

sensors such as ambient humidity 

– Uses ambient sensors used for high accuracy because: 

– As NOx levels come down, ambient conditions have a more dramatic 

affect on NOx calculations 

– Omitting or using surrogates for this input places the system at risk of 

misrepresenting emissions 

• Ambient sensors not used with: 

– Steam injection, water misting, flue gas recirculation 

– PS-16 (6.1.4 Ambient Conditions) requires ambient condition affects 

PS-16 states that if ambient sensors are not used then sufficient data must be provided to 

show that ambient conditions do not affect the model accuracy


Impact of Ambient Conditions 

39.89ppm NOx 

Weather and seasonal 

changes can result in 

humidity changes as 

much as 50%. 

75% 

A 25% change in Humidity 

results in a 12% change in 

the NOx output. 

50% 

46.11ppm NOx


Summary 



How is Software CEM ® Different? 

Cost Effective Monitoring & Compliance 

• Up to 50% less cost to install and maintain than hardware CEMS 

• No specially trained technicians/dangerous environments/hazardous materials 

• No consumable products, thus no increase in material operating costs 

Highly Accurate & Reliable Emissions Monitoring & Reporting 

• Continuous, real time sensor validation & emission performance 

• Meets or exceeds regulatory uptime requirements (~99% - 100%) 

• 100% data validation lowers the risk of inaccurate reporting 

Environmental and Industry Experience & Expertise 

• 1 st to patent & introduce (1993) a software-based alternative 

• 100% Certification Record with 250+ PEMS installations 

• Proven global applications in major industries 

• Active member of environmental quality technical support groups


Software CEM ® Highlights 

• Nonlinear Hybrid Modeling Methodology 

– Reliability based on real-time continuous sensor validation 

– High accuracy by design using non-linear process modeling 

• Data set is not the model 

– Once the model is validated it no longer uses the test data 

• Meets all USEPA requirements for PEMS 

– 40 CFR Part75; Subpart E for “Alternative Monitoring Systems” 

– 40 CFR Part 60; Appendix A, Performance Specification 16 (PS-16) 

• Over 250 applications of diverse emission sources and types as well 

as local, state, national and global regulatory experience 

– Greater than 40% of installed Software CEMs are customer or independently 

maintained and supported


Case Studies 



Challenges 

Netherlands NO x Emissions Credit Trading 

• Continuous NO x Emissions Monitoring if > 100 MWth 

• CEMS (hardware) or PEMS (software) 

Pernis, The Netherlands 

Performance Requirements 

• En-14181: QA standards for continuous emissions 

monitoring at stationary sources 

– NO x accuracy ± 20% 

250 

– Availability > 97% 

Gas Turbine 

• 160 MWth, 19 burners, dry low NOx 

Extremely Short Project Schedule 

• Urgently replace poor performing monitoring system 

NOx [ppmv] 

NO x profile relative to burners in operation 

200 

150 

100 

5 

15 

25 

35 

45 

55 

65 

75 

85 

95 

50 

0 

Gas turbine 

Load [%] 

20 

18 

16 

14 

12 

10 

8 

6 

4 

2 

0 

Burners [#] 

NOx [ppm] 

Burners

The Benefits 

• Performance 

– Accurately and reliably predicts NOx emissions on complex sources such as 

the dry low NOx gas turbine: 

• NOx emissions accuracy +/- 9.2% (exceeds the requirement of +/- 20%) 

• Patented Sensor Validation provides 100% availability 

• Fully automated quality assurance testing 

– Minimal maintenance required (lower TCO) 

• Met Tight Project Schedule 

– Project duration: < 5 weeks including: 

• P.O. and contract completion 

• Model development and deployment 

• Stack testing 

• Certification in accordance with EN-14181 

• Fast time to value 

– Quicker ROI 


Boilers with 

Ultra Low NOx Burner Technology 

Challenges 

• Subpart Db Boiler (total 3) 

– Industrial Powerhouse 

– 3 – 180MMBTU 

• New Air Permit required NOx outputs below 10ppm 

– New permit required reconfiguration of boilers 

– Installation of Ultra Low NOx burners and 

– Firebox extension 

• Retrain model for new configuration 

– Time for all 3 boilers under 8 weeks 

• First Ultra Low NOx burner application for a PEM 

• Passed USEPA and TCEQ certification 

– Better then 7.5% relative accuracy 


Boilers with 

Ultra Low NOx Burner Technology 

Benefits 

• Existing Software CEM reconfigured for low cost 

• No replacement with Hardware CEM 

• Cost savings 

– Over $1MM in saving over 10 year operating period 

– No additional personnel needed for HWCEM maintenance 

– No calibration gases required 


An Gas Compressor Station 

Ultra Low NOx SCR Application 

Challenges 

• Subpart Db Boiler (total 1) 

– Oil & Gas Compressor and Stripping Station 

– 1 – 280MMBTU 

• New Air Permit required NOx outputs below 10ppm 

– New permit required reconfiguration of boilers 

– Installation of SCR with ammonia injection 

– New boiler installation 

• Retrain model for new configuration 

– Model Boiler and SCR outputs 

• First Ultra Low NOx SCR application for a PEMS 

• Passed USEPA and TCEQ certification 

– Better then 6% relative accuracy 


An Gas Compressor Station 

Ultra Low NOx SCR Application 

Benefits 

• Existing Software CEM reconfigured for low cost 

• No replacement with Hardware CEM 

• Cost savings 

– Over $300K in saving over 10 year operating period 

– No additional personnel needed for HWCEM maintenance 

– No calibration gases required

Predictive Emissions Monitoring System &#40;PEMS&#41; - ladco

Create successful ePaper yourself

Delete template?

Save as template?

Predictive Emissions Monitoring System (PEMS) - ladco