GER-4198 - Intelligent Optimization of Coal Burning to Meet … · 2019-10-04 · coal of a...

GE Power Systems

Intelligent Optimizationof Coal Burning to MeetDemanding Power Loads,Emission Requirements,and Cost Objectives

R. SehgalPraxis Engineers, Inc.

P.J. MaroldaGE Power SystemsSchenectady, NY

GER-4198

g

Contents

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Silo Flow™ Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Three Key Applications of the CBAS Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Application 1: Matching Coal Quality to Load Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Application 2: Using the CBAS Product to Manage Emission Requirements . . . . . . . . . . . . . . . 5Application 3: Using the CBAS Product to Deliver Premium Coals When Rates Are High . . . . . 6Additional CBAS Product Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

CBAS Product Architecture and Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Intelligent Optimization of Coal Burning to Meet Demanding Power Loads, Emission Requirements, and Cost Objectives

GE Power Systems ■ GER-4198 ■ (10/00) i


GE Power Systems ■ GER-4198 ■ (10/00) ii

AbstractCoal is the most important fuel in the genera-tion of electricity, accounting for between 55%and 60% of the power produced in the UnitedStates and Canada. Available in great abun-dance from indigenous sources, there are noforeseen issues associated with the long-termsupply of the fuel resource of coal. In the past,because of the regulated business environmentin which utilities operated, coal was viewed as anecessary evil and was always treated as a "passthrough" cost. Although the quality of coalimpacts boiler operations and subsequentpower produced and emissions, utilities had noeconomic interest in monitoring or managingcoal quality.

This perspective has started to change with ini-tial legislative requirements for the control ofSO2 emissions. The legislative agenda has nowadvanced to include NOx emissions and neces-sitates a clear understanding of the interactionsbetween coal quality and the process of com-bustion. Further, the economic aspects of coalutilization are also changing. The main drivingforce behind this change is the increasing ten-dency in the United States to allow utilities tokeep any savings they can achieve in the cost oftheir coal. Since coal usage accounts for nearlyseventy percent of operation and maintenance(O&M) costs of electricity generation, coal pro-vides a maj s utilities to use the most economicblend of coals under varying conditions of load,emissions, and operational requirements. Thisform of blending, wherein the blend of coalsreaching the burners is changed to accommo-date the changing conditions in the boiler, isdefined as “dynamic blending.” This is a signifi-cant departure from the practice of using afixed blend of coal under all conditions. TheCBAS product was developed mainly with fund-ing from the U.S. Department of Energy and

TransAlta Utilities and is currently installed at26 boilers at power plants around the world.

Please note that the terms “silo” and “bunker”are used interchangeably in this paper, andrefer to the coal storage devices used at powerplants. Silos and bunkers have been modeledsuccessfully in CBAS product installations.

IntroductionFuels cost reductions are achieved by deliveringcoal of a predetermined quality to the burnersat all times. If a power plant adopts a dynamicapproach to coal blending (i.e., the as-firedblends change continuously to adjust to chang-ing conditions such as load, emissions, opacity,coal availability, and expected coal deliveries),the power plant can reduce fuel-related costssignificantly.

Dynamic blending of coals can improve effi-ciency and decrease costs in a number of ways.Plants blending their boiler feed from a num-ber of coals can maximize the use of the leastexpensive coal whenever conditions permit;utilities can rationalize the use of coals to avoidover-using premium coals and under-usingproblem coals; and utilities can deliberatelyincrease purchases of better spot-market coalsknowing that they can mitigate the attendantproblems by controlled dynamic blending. Theeconomic benefits of this approach are consid-erable, as plants are able to reduce operatingcosts significantly by optimizing the coal blendburned.

Figure 1 shows a typical utility coal yard wherecoal is received from a number of sources andstored in semi-segregated piles. The coal isremoved from these piles and loaded into themill silos. Each silo feeds a pulverizer, and usu-ally has a storage capacity of about eight hours.

The main goal for a power plant is make surethat the most economic coal of precisely the


GE Power Systems ■ GER-4198 ■ (10/00) 1

right quality gets to the burners to meet theboiler requirements. For example, a boiler canbe fed coals of lower quality at low-loads andonly needs high-quality coals at full-load opera-tions. The ability to deliver the right coals to theburners consistently can reduce fuel costs forthe plant significantly.

To achieve this goal, a power plant must addressthree main issues. The first and the easiest toaddress is to know the quality of all the coalspurchased and stockpiled at the site. In general,the quality data provided by the coal supplier isconsidered sufficient for use in the CBAS prod-uct. The second key issue, operational innature, is for the power plant to keep track of allcoals as they arrive and to stockpile them in seg-regated or semi-segregated piles. This can bedone through electronic data acquisition inmost cases, and no manual data entry isrequired. The third key issue is the ability toknow what coals are being loaded into the millsilo when and in what quantity. From this infor-mation, the CBAS product can calculate thequality of the coal being discharged to the boil-er from each of the silos in real time. Since the

coals in the silos rarely come out in the samesequence as they were loaded in, this aspect ofthe CBAS product is extremely complex andrequires the modeling of coal flows through thesilos. This is handled by a special component ofthe CBAS product called the Silo Flow™ Model,which models the coal flow in three dimensionsand is able to accurately correlate the inputquality sequence with the output qualitysequence. The model can also be used for simi-lar purposes in modeling flow through varioustypes of stockpiles.

Once the flow of coal in silos and stockpiles canbe accurately modeled, the tracking of all coalsin a yard, from receipt to combustion, becomesrelatively easy. In addition, the same capabilitycan also be used to perform controlled loadingsequences into the silos to assure that coal of aprescribed quality is discharged at prescribedtimes. The CBAS product uses this capability todeliver the “right coal to the burners at theright time,” saving users significant fuel costs.

The Silo Flow Model is described below ingreater detail.



Ship

BOILER

Truck

Train

Figure 1. Schematic showing typical power plant coal yard

Silo Flow™ ModelThe approach adopted by Praxis Engineers tocalculate the flow patterns in the silos combinescoal quality with silo geometries. The three-dimensional Silo Flow Model uses the sequenceof input quality to predict the sequence of out-put quality, thus making it possible to know fuelquality at the burners in real time for manyhours in the future. A typical silo is shown inFigure 2 and its active and dead zones are iden-tified. Four different coals have been loadedinto the silo in sequence and the problem is topredict the quality of the output stream. TheSilo Flow Model is able to do this for all silodesigns.

The silo represented in this instance has onecylindrical section and two sloping conical sec-tions. This silo design was used as the prototypeto develop the first Praxis Silo Flow Model in1987. Note that this design is somewhat morecomplex than many silos normally used to feedutility boilers, which have a single conical sec-tion below the cylindrical one.

Please note that regardless of the silo geometry,the CBAS product is individually configured foreach power plant. A team of engineers visits the

site and records the flow properties (e.g. anglesof friction, moisture contents, size distribution,bulk density) that are required to customize themodel for the silos at each site. Currently, allsilo and bunker types generally found at powerplants have been modeled.

The sequence of input and output coal qualitiesfor the silo in Figure 2 is shown in Figure 3. Forillustrative purposes, it is assumed that the fourcoals loaded into the silo had different ash con-tents of 10, 12, 14, and 16% respectively. As thediagram shows, the outflow coal quality (interms of ash content) follows neither a first-in-first-out (FIFO) pattern, nor an average value,as might be assumed. The actual pattern is con-siderably more complex and difficult to predictwithout intensive computation and accuratemodeling. In this particular case, it can bedemonstrated that a considerable quantity ofthe second coal (12% ash) came out at the veryend, in blends with Coal 3 and possibly Coal 4as well.

Once the ability to predict and control coalquality at a plant has been codified, most usersemploy the CBAS product to get blend or coalloading recommendations in order to achieve



Figure 2. Schematic of typical expanded flow silo

their particular goals. The CBAS product exam-ines all available coals and provides one to fiveranked coal loading recommendations toachieve all of the plant’s goals. These goals mayinclude matching coal to load, SO2 compli-ance, NOx emissions, opacity control, or others.Please note that a user can specify a main crite-rion, such as ash, and any number of secondaryconstraints such as sulfur, moisture, and BTUvalue.

A typical manner in which recommendationsare presented to the user is shown in Figure 4.The plant wants to match the ash content of thecoal to the expected load demand over the nextperiod of hours, such as 24. The operationalconstraint in this case is that the plant onlyloads coal into each silo twice a day. Its goals areto use high-ash, low-price coal blends when theload is low, and low-ash coals when the load ishigh. If done successfully, the implementationof this strategy leads to the use of the lowest-price coal consistent with all load conditions.

Three graded recommendations, characterizedby a “Fitness Function,” are presented to theusers for each of the two loadings. Their respec-tive value to the plant is indicated by the fitnessfunction (the lower the better), and the yardoperators can select any one of the suggestedapproaches, with each able to meet its objec-tives, though at somewhat different costs. Thisfeature allows the plant operators to take into

consideration any unforeseen events or equip-ment breakdowns that the CBAS product maynot be collecting data for and thus cannot con-sider in making the recommendations.

Three Key Applications of the CBAS Product

Application 1: Matching Coal Quality toLoad ProfileFigure 5 shows the ability of the CBAS product tomatch coal quality to the load requirements.The expected load demand curve is eitherentered into the CBAS product by the plantoperators (or obtained on-line from a plantdata source). Next the CBAS product makesrecommendations for coal loading and blend-ing to assure that the right coal is available atthe burners to meet the load requirements.

The example shown in Figure 5 is a sample oftest data from a utility that measured the accu-racy of the model. The model’s predictionswere always within 0.5% of the actual coal blendexiting the mills. The load profile is shown atthe top of the graph. As the load increased, thepercent ash in the coal decreased, demonstrat-ing the load matching capability of the CBASproduct.



Figure 3. Silo model prediction of silo outputcomposite quality vs. time

Figure 4. The CBAS product blend recommen-dation table

Application 2: Using the CBAS Product toManage Emission RequirementsFigure 6 shows the ability of the CBAS productto blend coals to avoid generation losses that aplant was incurring due to uncontrolled blend-ing/usage of its coals. The plant obtains all of itscoals from four or five seams in a captive mine.Each of these coals has a different “opacitypotential” and if the opacity at the stacksexceeds a certain value, the plant has to shedload in order to remain in compliance. TheCBAS product, by managing all coal receipts

and blending the coals appropriately, was ableto avoid nearly all instances of exceeding theseopacity values. As a result, the plant producedan additional 70,000 MW-hr in one year, with asubsequent increase in its revenues.

Remarkably, the CBAS product required nochange in the mining practice and only a mini-mal change in the yard operations. No newequipment was added, and the plant continuedto use all the same coals. The CBAS productsimply rationalized the use of available coals.However, this made a difference of approxi-mately $2 million (USD) in added revenues.



CBAS Predictions are Accurate

4

6

8

10

12

14

16

18

20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Hours From Start of Test

%Ash

17 JUNE 18 JUNE

500 MW

400 MW

300 MW

Load

Projected Load

Actual Ash Projected Ash

Figure 5. Least expensive coal to meet power load

CBAS Reduces Derates

0

1000

2000

3000

4000

5000

6000

13 14 15 16 17Average Opacity Potential (Indicator of Coal Quality)

With CBAS

Without CBAS

Figure 6. Opacity derate history with and without the CBAS product vs. varying coal quality

MW

h/M

onth

Der

ate

CBAS Predictions Are Accurate

CBAS Reduces Derates

500 MW

400 MW

300 MW

Load

Application 3: Using the CBAS Product to Deliver Premium Coals when Rates Are HighFigure 7 illustrates the use of the CBAS productto deliver a coal of extra-high quality to theburners during the hours of expected peakdemand when the price of power is at its high-est. By using the best coal only when needed,and using less expensive coals at other times,the CBAS product assists the plant in two ways:lower fuel costs and increased revenues.

The graph in Figure 7 shows how the CBASproduct can be used to deliver the highest-grade coals (highest Btu/lb) to the burnersonly during peak periods, allowing the plant togenerate extra MW only during the peak peri-ods. Lower-grade coals were delivered beforeand after the peak periods, minimizing theusage of expensive coals.

Additional CBAS Product ApplicationsThe system can also be used to optimize FlueGas Desulfurization systems. The CBAS prod-uct can track each type of coal and the associat-

ed sulfur content. The CBAS product recom-mends blends to meet regulatory requirements,and also recommends a bypass operation whenconditions permit this. This type of operationcan save millions of dollars per year in energyand reagent costs, and optimize the usage orgeneration of SOx credits.

The CBAS product can also be used to manageany other emission issues that are related tocoal. For instance, fuel NOx (the NOx relatedto the nitrogen content in the fuel) producesapproximately 70% of the total NOx generatedin the boiler. As such, fuel NOx is a very impor-tant variable in the control of NOx. The CBASproduct can track the nitrogen content in thecoal, and notify the boiler/NOx system ofupcoming changes in fuel nitrogen content.This allows the boiler optimization system tomake much better decisions in the manage-ment of NOx. In addition, the CBAS productcould even be used to blend coals based ontheir nitrogen content, significantly improvinga plant’s ability to manage NOx emissions overa daily or weekly period.



450

Load(MW)

0 4 8 12 16 20Time (hours)

150

200

250

300

350

400

Load Target

Max Load @Best Fixed Blend

CBAS blend useslower price fuelwhen load is low

CBAS blend meetspeak load requirement

Load

(M

W)

Figure 7. The CBAS product matching of coal blend to load curve

CBAS Product Architecture andInstallationThe CBAS architecture is shown in Figure 8.The architecture consists of a number of mod-ules such as the CBAS Client (the graphical userinterface), DASMod, EventMan (EventManager), CoalTrack, and OptiMix. The mod-ule functions are outlined below.

The CBAS Client is the module for thehuman/user interface. The CBAS Client mod-ule displays the current status of the coal yard,including the coal properties in the piles,bunkers and silos, and the coal entering theboiler. The CBAS Client also displays the resultsof the OptiMix calculations. It is the vehicleused to set any manually entered data, such asequipment status and OptiMix targets andoptions.

DASMod is the data acquisition module whichconnects the CBAS product to the plant data

sources. A number of modules for the com-monly used DCSs are available and others canbe developed as and when needed. TheEventMan module manages all the informationtraffic flow between the other modules of theCBAS product.

The CoalTrack module contains the Praxis sim-ulation modules for coal handling equipment,silos, bunkers and coal piles. This module usesthe real-time operations data for weigh scales,conveyor status, flop gate positions and the likealong with the Silo Flow Model which modelsbulk materials flow in piles and silos to track theflow of coal from receipt, through piles, convey-ors, and silos all the way to the burners.

The OptiMix module implements the ModelPredictive Control (MPC) strategy. TheOptiMix module uses several types of models,either alone or in conjunction with one anoth-er, such as mathematical models, fuzzy-neuralmodels, and the Silo Flow Model, and con-



Figure 8. The CBAS product architecture

verges at an optimum solution for a case-specif-ic prediction horizon. The strength of this mod-ule lies in its accuracy and flexibility, both inmodeling and in optimization. The optimiza-tion techniques in the CBAS product use con-ventional methods, fuzzy logic-based algo-rithms, and Genetic Algorithms (GA). Based onadjustable constraints and requirements,OptiMix computes the best pile management,coal blending, and bunker feed actions for thecoal yard and boiler operators.

All other CBAS product computations andfunctions are performed on the CBAS HostComputer, a dual-processor Pentium® (or bet-ter) PC running Windows NT®. Please note thatinstallation of the CBAS product at a plant isgenerally simple, as the host computer onlyneeds to be connected to the plant data system.In general, the implementation of the CBASproduct is completed within twenty weeks.

SummaryThe CBAS product is a powerful tool that canimprove the economics of power plant opera-tions significantly. Based on the economic ben-efits that may be achieved, the payback of aCBAS product can be achieved in months andresults in millions of dollars in coal savings forthe power plant. Some ways the CBAS productcan improve plant profitability include:

■ Using the least expensive coal to meetpower load demands

■ Minimizing derates through control ofcoal quality burned

■ Limiting the burn of expensive high-quality coal to only peak generationtimes

■ Managing boiler emissions throughappropriate coal blending at theburners

■ Minimizing double handling.



List of FiguresFigure 1. Schematic showing typical power plant coal yard

Figure 2. Schematic of typical expanded flow silo

Figure 3. Silo model prediction of ilo output composite quality vs. time

Figure 4. The CBAS product blend recommendation table

Figure 5. Least expensive coal to meet power load

Figure 6. Opacity derate history with and without the CBAS product vs. varying coal quality

Figure 7. The CBAS product matching of coal blend to load curve

Figure 8. The CBAS product architecture



GER-4198 - Intelligent Optimization of Coal Burning to Meet … · 2019-10-04 · coal of a...

Documents

Transcript of GER-4198 - Intelligent Optimization of Coal Burning to Meet … · 2019-10-04 · coal of a...