Author: N.Bonavita, G.Ciarlo Date: February 25 , 2014 MENA … · 2018-05-09 · §PEMS, Predictive...
Transcript of Author: N.Bonavita, G.Ciarlo Date: February 25 , 2014 MENA … · 2018-05-09 · §PEMS, Predictive...
© ABB GroupNovember 18, 2014 | Slide 1
MENA Offshore Platforms 2014Combining analytics and process automationfor enhanced emission control strategies:PEMS at work
Author: N.Bonavita, G.Ciarlo Date: February 25th, 2014
Emission Monitoring Systems
§ EMS encompasses the followingfunctions:§ Sampling
§ Analysis
§ Data Acquisition
§ Data Validation
§ Data Processing and Archiving
The Traditional Approach
© ABB Inc.November 18, 2014 | Slide 2
COCO
SO2SO2
NOxNOxHClHCl
HFHF O2O2
NH3NH3HgHg
TOCTOC
H2OH2O
DustsDusts
DioxinsDioxins
MetalsMetals§ The core of traditional Emission
Monitoring System are HardwareAnalyzer, integrated with IT andsoftware infrastructure for emissionstorage and reporting
IntroductionZooming on CEMS
© ABB Inc.November 18, 2014 | Slide 3
Source: “EPA Handbook – Continuous Emission Monitoring Systems for Non-criteria Pollutants”, 1997
§ PEMS, Predictive Emission Monitoring Systems, represent analternative technology for emission monitoring based on software
§ PEMS provide emission estimation by means of sophisticatedModels based on mathematical and/or statistical techniques.
§ Models are able to exploit the inherent correlation existingbetween process variables (flow, temperature and pressure) andemission properties (NOx, SO2, CO, CO2).
§ Correlation are assessed and evaluated through fieldmeasurement campaigns:
§ Emission Data collected from existing HW analyzers orwhen not available temporarily installed at plant stack
§ Process Data from plant Control System (DCS, PLC) orProcess Historians
IntroductionWhat are PEMS?
© ABB Inc.November 18, 2014 | Slide 4
§ A SYSTEM is defined as a collection of objects among which itexists a set of cause-effect relationships
§ A MODEL is defined as a set of rules by which, knowing theinputs, it is possible to derive the outputs behaviors
It is possible to quantitatively represent a system by a model:
Yi = Li [x1, x2, …, xs] i=1, …, r
Model-based Emission MonitoringWhere does it come from?
Let’s move from Definitions …
§ It is possible to classify modeling technique in two main groups:
§ Fundamental (or First-Principle) Models, where the descriptiveequations are derived from the basic physical and chemical laws
§ Empirical (or Data-driven) Models, where the models are built througha fitting-like procedure over the plant live data
Momentum Balance: gvvvd
vd rtrr-
¶¶
+¶¶
-¶¶
-= 2
2
xxxP
t
UA
UB
YA
YB
U Y++ +Y = a1UA+ a2UB+…
+UA UB YA
15.23 1.37 6.5416.02 1.39 6.5814.89 1.43 6.4115.75 1.47 6.38
Model-based Emission MonitoringWhere does it come from?
Wide Spread of Process Computers, Distributed ControlSystems (DCS) and SCADA
Historical data are a valuable asset for bettercontrol, management and optimization
Data Collection and Storage is easy, cheap andstraightforward
Model-based Emission MonitoringWhere does it come from?
§ Process Automation Trends in the last 25 years:
DRAFTStrictly Confidential
© ABB Inc.November 18, 2014 | Slide 8
§ Leveraging and exploiting Process Automation, Analytics andInformation Technology to improve environmental monitoring andperformances
§ Exploiting the wealth of meaningful data available in modernindustrial plants to accurately estimate emission levels and deeplyunderstand how they develop and can be reduced
Model-based Emission MonitoringRationale for PEMS
PEMSMain Criticalities
§ PEMS successful implementation requires three mainingredients:§ Technology: i.e. SW environment able to handle
data from different sources, build and validateefficient models and endowed with proper facilitiesfor on-line deployment and monitoring
§ Know-how: i.e. highly skilled engineering resourcesable to blend instrumentation and analyticscompetences with process automation capabilitiesand modelling building expertize
§ Local Presence: in order to provide competitivesolutions (and efficient maintenance) serviceengineers should be next-door
© ABB Inc.November 18, 2014 | Slide 9
§ A unique software platform, ABB proprietary, fordevelopment and deployment of empirical processmodels, featuring different technologies:§ Neural Networks
§ Statistical Regressions
§ Multivariate Statistical Analysis§ Equation-based models
§ Custom-based models (DLLs)§ It is composed by two different environments:
§ The Model Builder for data analysis, model buildingand validation
§ The On-line environment for effective and quickmodel deployment, execution and monitoring
PEMS: TechnologyABB Inferential Modeling Platform - IMP
© ABB Inc.November 18, 2014 | Slide 10
PEMS: TechnologyIMP Model Builder – Features
© ABB Inc.November 18, 2014 | Slide 11
§ Data Import from common sources like .xls,.csv, .txt etc.
§ Data pre-processing: Automatic OutlierRemoval to clean dataset for modelling fromanomalous process data
§ Correlation and PCA: specific advancedstatistical techniques to identify best processvariables influencing emissions or qualityvariables
§ Modeling functions are provided by proven,field-tested, latest generation routines
§ Model development executed throughWizards, to optimize manpower allocation
§ Model Explorer extends use of modelsbeyond on-line utilization (“What-if” analysis)
© ABB Inc.November 18, 2014 | Slide 12
§ Quick and effective real-time implementationon different DCS through OPC
§ Supports multi OPC connection to gatherdata from different sources
§ Configurable filtering of inputs and outlierremoval strategies to verify and evidenceabnormal process condition and bad qualitydata
§ Emission estimation visualization, trendingand storing capabilities
§ Direct connection to Laboratory InformationManagement Systems
§ Direct connection with Data AcquisitionSystem.
§ Built-in functions for periodic recalibration(Bias calculation)
PEMS: TechnologyIMP On-line: – Features
Inferential Modeling PlatformInstallations
Hawaii
IMP© ABB GroupNovember 18, 2014 | Slide 13
PEMSApplication Fields
Two Main Possible Uses
© ABB Inc.November 18, 2014 | Slide 14
Primary Source ofEmission Monitoring
Back-up/Validation ofExisting HW Analyzers
§ Coupled withtraditional CEMS toincrease theavailability of theEmissionMonitoring Systemup to 99,5%
§ Where CEMS arenot feasible or cost-effective
§ PEMS typical general requirements are:§ Plant has to be instrumented (T,F,P sensors) and
automated (PLC or DCS presence)§ Quality of combustion fuel has not to vary too much
or has to be monitored§ PEMS can be ideally applied to every combustion
process across the industry sectors from O&G, Chemical& Petrochemical to P&P, Cement, Metal, Power Utilities.
§ PEMS perfectly fit in Gas Turbine applications (Power,Compressor Drivers), Large Industrial CombustionEngines, Boilers and Furnaces.
PEMSApplication Fields
© ABB Inc.November 18, 2014 | Slide 15
IMP
PTTTFT
NOxSO2CO
© ABB Inc.November 18, 2014 | Slide 16
PEMS ApplicationIntegrated Architecture (Example)
HMI
EmissionManagementSystem
OPCServers
Controllers
Field
Ethernet TCP/IP
Ethernet TCP/IP
CEMS
PEMS ApplicationTypical SW Architecture (Back-up to HW CEMS)
© ABB Inc.November 18, 2014 | Slide 17
TC FCFTTTFT PC
FIELD
+ -
IMP
ProcessVariables
AnalyzerValues
PredictedValues
S
“maintenancetrigger”
discrepancybetweenmeasured andpredictedvalues
Indication of maintenance need§ Physical sensor drift from model: by detecting the drift, maintenance needs
can be anticipated; maintenance activity can be validated by using the model
© ABB Inc.November 18, 2014 | Slide 18
Physical Sensor
SW Sensor
40
45
50
55
60
65
70
75
80
85
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
AnalyzerModel
Alarm sent tomaintenance station
Maintenancesignal
Post maintenancevalidation
PEMS ApplicationExample: Maintenance trigger
PEMS ApplicationProject Execution
OfficeActivities
ImplementationSteps
DataCollection
Final Assessment &Commissioning
Model Validation
Inferential Model Buildingand Testing
Data Processing
On SiteActivities
© ABB Inc.November 18, 2014 | Slide 19
PEMS Success Story #1Gas injection compressor units
© ABB Inc.November 18, 2014 | Slide 20
§ Twin trains of gas compression§ Compressors gas turbine driven, each GT emission to a
dedicated chimney§ Two stage for each train (high and low pressure)§ PEMS installed at each stack (4 system)
PEMS Success Story #1Plant EMS: Hardware/Software architecture
© ABB Inc.November 18, 2014 | Slide 21
Plant LAN
ABBIMP OnlineABB PGP
DCS Ring- Infi 90 Bailey
EMS Server-DuragNew Cabinet-
Control room
EMS Server: Third partyEMS Client: ABB PGP
EMS Client:§ Collects data from DCS and
make those available to IMP On-Line trough OPC Protocol.
§ Collects Prediction from IMP On-Line trough OPC and troughModbus send to EMS Server
EMS Server:§ Collects data from EMS client
and CEMS analyzer at stack§ Send data to Company HQDCS System
Connection to DCSConnection to EMS
Bailey Infi90ICI03Modbus
PEMS Success Story #1Conclusions
§ Commissioning successful completedby May 2008
§ It has been achieved the internationalcertificate EPA (US EnvironmentalProtection Agency)
§ First ABB commercial solution for anemission monitoring system based onSoftware Sensors, EPA Certified, inO&G field in Middle East Region.
§ Fully Integrated in the plant baseautomation system
§ A paper published on ABB review
§ 2013: major maintenance andrecalibration completed
© ABB Inc.November 18, 2014 | Slide 22
§ Project includes the supply of PEMS for three stacks in a refineryin Mediterranean and inclusion in the existing EMS
§ Customer wanted to§ Increase CEMS service time to meet tighter regulation§ Prove the viability of PEMS technology to local environmental
legislator, pioneering the first application in the country
§ Site Selection§ Refinery is one of the biggest in Europe and has a production capacity
of 180000 bbl/day.§ PEMS are integrated in the existing EMS and act as back-up to
increase traditional CEMS availability for the three below plant units§ Sulfur Recovery Units (SRU)§ Fluid Catalytic Cracking Unit (FCCU)§ Belco Plant
§ Emissions and properties estimated:§ Flue Gas flow, CO, O2 ,SO2, NO, Particulate
© ABB Inc.November 18, 2014 | Slide 23
PEMS Success Story #2Mediterranean Region Major Refinery
PEMS Success Story #2Sulfur recovery Units (SRU)
© ABB Inc.November 18, 2014 | Slide 24
Characteristics :§ 3 parallel trains with different layout§ 3 feed gas with variable composition, coming from several
different plant units:§ SO2 rich gas (SRG)§ Sour water gas (SWS)§ Amine acid gas (AAG)
Plant configuration of the three sulfur recovery units
System includes:
§ 3 EMS client collecting emission datafrom stack analyzers
§ 1 EMS server for emission datareporting and storage for the wholerefinery
§ 1 PEMS server
§ 1 PIMS server collecting data fromdifferent DCS
PEMS Success Story #2EMS Hardware Architecture
© ABB Inc.November 18, 2014 | Slide 25
PEMS Success Story #2EMS Human Machine Interface
© ABB Inc.November 18, 2014 | Slide 26
PEMS Success Story #2PEMS Results: FCC
© ABB Inc.November 18, 2014 | Slide 27
PEMSCEMS
PEMS Success Story #2PEMS Results: FCC
© ABB Inc.November 18, 2014 | Slide 28
PEMSCEMS
NO Predictions and measured values in a 1-day span for the FCC unit
PEMSConclusions
§ PEMS are a cost effective solution to complete andenhance emission monitoring strategies.
§ PEMS features distinct advantages in a number ofapplications with or without HW analysers
§ ABB has the experience, the technology and the qualifiedlocal presence to provide the best solution no matter it is aHW-based CEMS, a PEMS or a smart and flexiblecombination
§ ABB owns IMP, a software tool recognized for PEMSdevelopment and deployment which has already a fewapproved (and certified!) installations.
Source: WW Emission Monitoring report- ARC 08/12