THE

,

INDUSTRY

A Method forImprovingRegenerative

Furnace Efficiency

A Method forImproving

RegenerativeFurnace

EfficiencyBy R.J. NAVEAUX and J.J. SHEA

Toledo Engineering Co., Inc.Toledo, Ohio

THE regenerative furnace wasfIrst invented by Robert Stirlingin 1816and was applied to glass

melting by Fredrick Siemens.When wecompare the details of the original in-vention with the furnace design prac-tices of today, we must conclude thatlittle change has been made in the origi-nal concept.

It has been only since the late 1960s,when fuel costs started to climb, thateffIciency became a priority concern.Then, with the oilembargo in the 19705and the resulting energy crisis in theUnited States, intensive efforts weremade to improve the effIciency of re-generative furnaces.

This report on an effectivesystemof combustion control

and heat recoverywaspresented at the 42nd Annual

Conferenceon GlassProblemsheld

at the Univ.of Illinois

This paper will focus on an econom-ic improvement that can be adapted toexisting furnace installations or incor-porated into new furnaces. We willconcentrate on large sideport-type fur-naces since all of our studies to datehave been centered around them.

The Basic Regenerative Furnace

Fig. 1 shows a plan-view of a typicalregenerative furnace. The flue connec-tion is on the end of each regenerator,and both the waste gas and the com-bustion air pass through this connec-tion. These regenerators come in manydifferent sizes and shapes, but all ofthem have as their main function thepreheating of the combustion air tosave fuel. They do this by transferringthe heat in the exiting waste gas to re-fractory bricks stacked in a checker-board fashion inside each regenerator.The bricks, in turn, giveup their heat tothe incoming combustion air on abouta 15-minute reversing cycle. This pro-cess goes on night and day, every day,for 5,6, or7years, and evenup to 10or12years.

Most regenerators could stand im-provement, especially from an effI-ciency point of view. It wasn't until re-cently that many segments of the glassindustry in the United States addressedthe fuel shortages and rising fuel pricesand concentrated on fuel economy intheir melting operations. We became"fuel economy conscious" very quick-ly as fuel costs started to escalate, andwe can be assured that fuel costs willcontinue the upward trend. Some fore-casters say that natural gas could go to$12per thousand cubic feet by the year1990.

A most prominent cause of low effi-ciency and high fuel usage in a regener-ative furnace is related to the unevendistribution of both the hot waste gasand the combustion air across andthrough the checker pack, thereby notutilizing all the checkers uniformly.This creates overheating of somecheckers and underheating of others.The furnace is forced to operate athigher levelsof excess oxygen in orderto get the air required to the properports for efficient combustion. Theelement in the system that creates thesepoor efficiency conditions is the flueconnection at the end of each regenera-tor. This fact can be explained by thefluid dynamics of the exiting waste gas

and entering combustion air move-ments through these end connections.

Fig. 2 shows, in elevation, the ex-treme lack of gas flow symmetry thatexists in this type of system and clearlyindicates that the waste gas tends tofavor the flue connection as it is beingpulled by the furnace draft system tothe flue. This causes overheating ofcheckers on the flue end and under-heating of checkers on the oppositeend. Then, when the system is re-versed, the cold combustion air rushesin through the end connection and theair, because of its momentum, tends tofavor the end opposite the flue, makingit difficult to get the air needed forgood combustion at all ports.

In addition to contributing to lowenergy efficiency, these conditionshave been known to shorten campaignlife by overheating some rider arches,causing them to collapse.

Various schemes have been tried tocreate uniform gas flow through thecheckers; e.g., sloping the floor underthe rider arches and varying the fluesize within the checker pack down thelength of the regenerator. Otherschemes have been tried with varyingdegrees of success. In most cases, theimprovement favored either the wastegas or the combustion air distributionand rarely satisfied both flow condi-tions.

At some point in time, the compart-

FurnaceEfficiency

mented regenerator scheme of individ-ual chambers within the checker packwas devised. Each regenerator is di-vided into separate chambers at eachport. Fig. 3 shows a plan-view of thistype regenerative system. In this de-sign, the waste gas and the combustionair leave and enter the regeneratorthrough sideconnections. This type re-generator design has been more preva-lent in Europe than in the UnitedStates. This can probably be attributedto the fact that the Europeans encoun-tered high fuel cost problems long be-fore the 1973embargo. Also, this typedesign may have been developed to en-able a section of the checkers to be re-placed "on the run" when a checkerplugging problem occurred.

The compartmented design solvedthe distribution problem as long as thefurnace was fIred close to the designfuel gradient, but it created other prob-lems. First, it ismore expensiveto buildthan a conventional regenerator.Second, if a compartment becomesplugged, that compartment and portare out of commission until the plug-gingproblem iscorrected. Third, if onelooks closelyat the division walls, it canbe seen that these walls occupy valu-able space that could contain addi-tional checkers which would add about100/0to 15% to the heat transfer sur-face. This added checker surface couldprovide an increase of about 50OFto75 OFto the air preheat temperature.Fourth, the amount of the regeneratorthat serves each port is fIxed by theoriginal design, and it is not uncom-mon to fire a furnace with a differentfuel input gradient than originallyplanned; this results in an ineffIcientoperation. However, while this designsolved the fuel/air ratio distributionproblem by totally controlling the airand fuel to each port, it added greatly

to the initial cost and to the costs ofequipment and instrumentation and ofupkeep.

To summarize, most regenerativefurnaces of the end connection varietycan stand some improvement in effI-ciency.If existingfurnaces are to be im-proved or new ones are to be built, theyshould probably be provided with animproved regenerative system or theregenerators should be compart-mented. However, the compart-mented-type regenerator is relativelyexpensive and has need for improve-ment; therefore, it is only natural thatan effort be made to come up withsomething new.

The SUPERFLUE System

Design Features-This brings us tothe SUPERFLUE system, which is a

patentedI schemeforimprovingregen-erative furnace performance. Fig. 4shows a plan view of this type system.The waste gas exits the regeneratorchamber and the combustion air entersthe regenerator chamber through sideconnections in a similar manner as inthe compartmented design, but with-out division walls to separate the gasesat each port. Each side connection hasits own manually controlled damper,and by positioning these dampers, evendistribution of both the exiting wastegas and the incoming combustion aircan be achieved. An important advan-tage of this scheme is that once the in-dividual dampers are adjusted, it isnot

(I) u.s. Patent No. 4,174,948 - "Manifold Inputsand Outputs for Furnace Regenerators," by R.O.Bradley, H.J. Knighton and R.J. Naveaux, Nov. 20,1979,assignor to Toledo Engineering Co., Inc.

necessary for them to be moved untilmajor changes in pull rate, fIringdistri-bution, etc., are encountered.

Another important advantage ofthis scheme is that each damper, bothcombustion air and waste gas, is, in ef-fect, a "one way" damper. The com-bustion air and waste gas settings forfIringto the right have absolutely no ef-fect on the operation when fIringto theleft.

Many believethat it is not possible tosignifIcantly bias the amount of com-bustion air to each port of a box-typeregenerator without division walls bycontrolling the amount of air that en-ters each connection below the riderarches. However, test data have shownthat it can be done very successfully.

The SUPERFLUE system encom-passes the total furnace fIring and ex-haust system. A proper design musttake into consideration all facets; e.g.,port design and shape, burner arrange-ment, combustion space, checkers andair preheat temperature, temperaturelosses, pressure losses, etc.

When this sytemisincorporated intoa furnace design, the conditions thatmay have been considered as some-what typical will change. Air preheatsof 2300"F and 2400OFand abovecanbe achieved and should be consideredin the design.

The increased utilization of the heattransfer surface brought about byproper positioning of the dampersmakes it possible to add additionalcheckers to attain these high preheattemperatures. There have been in-stances where the No.6 port in a six-port furnace was blocked off and fullutilization of the checker pack was stillobtained. These higher preheat tem-peratures, along with the properburner/port geometry, have resulted innatural gas flames that develop fullluminosity.

Port sizing relationships, althoughvery important, are not as critical in aSUPERFLUE system. This systemcanforce the air and waste gas to achievefull utilization of the regenerators overa range of fuel input gradients. A con-ventional regenerator must depend onport size alone and, therefore, cannotaccomplish this. A compartmentedregenerator is restricted by the size ofthe compartment and, therefore, canaccomplish this for only a limited fIringrange.

OperatingEconomics-A comput-er program that allows us to make aquick analysis of the improvementsthat could be expected with this systemhas been developed. The data requiredto make this analysis isshown in Fig. 5.

Capital costs and operating costs arealways key considerations in any fur-

FurnaceEfficiencynace design. When this system is in-stalled in an existing furnace, capitalcosts will vary considerably dependingupon existing conditions. Of course,on a new furnace, only the premiumcapital costs need be considered. Inmost cases, the additional capital costcan be justified by the resulting reduc-tion in operating costs. To get an ideaof the cost savings potential, an in~crease in air preheat of 100"F repre-sents a fuel saving of about 50/0,and adecrease of 1% excess oxygen on amass average basis represents a fuelsaving of about 2%.

Fig. 6 shows the magnitude of thecost savings available. If a containerfurnace operating at 5mill BTU/tonwhen melting 200ton/day could be im-proved to operate at 4.25 mill BTU /ton, a savings of about $420,000 peryear could be realized. Similarly, if afloat furnace operating at 8.5mill BTU/ ton when melting 450 ton/day couldbe improved to operate at 7.25 millBTU/ton, a savings of $1,650,000 peryear is possible. To develop thesefigures, we used a gas cost of $7.50perthousand cubic feet, which could bethe average fuel cost for the next cam-paign.

Actual installation of the SUPER-FLUE system on an existing furnacehas resulted in a savingsof about 12%.On new installations, where checker

depth and length can be increased, ithas been possible to reduce fuel con-sumption by about 17%. Savings likethese must be given due consideration.

Operating Data-Six SUPERFLUEsystems have been installed between1977and 1981.Each system has been alittle different due to existing condi-tions, method of firing, tonnagepulled, size of furnace, and size ofchecker pack.

Fig. 7 shows a seriesof curves repre-senting the data that has been taken onthe furnaces with and without SU-PERFLUE systems. We have dividedthe furnaces studied into three groupsto represent the data more accurately.Temperature readings were taken witha modified Land SU4 Suction Pyrom-

eter to obtain temperature readingsthat were not affected by radiationfrom surrounding surfaces.

Group A shows the conventionalend flue-type regenerators with sin-gle span rider arches. Group Bshows the conventional end flue-type regenerators with open centerwalls below the double rider arches.Group C shows the SUPERFLUEsystem, two of which are new and

Fig. 12

two are retrofit. All four have singlespan rider arches.Unfortunately, we do not have any

data on compartmented regeneratorsthat we can publish.

The top curve "W" shows the aver-age waste gas temperatures at the tar-get walls of Groups A, B, and C and isa representation of the temperature ofthe waste gases as they enter thechecker pack.

The "X" group of curves shows theaverage temperature of the combus-tion air after it has been preheated bythe checkers. Note the increase in tem-peratures and uniformity of the tem-peratures in Group C.

Since these are average tempera-tures, it should be pointed out that sys-tems with more checkers achieve pre-heat temperatures of about 2400"P,and with these preheats, optical tem-perature readings taken at the top ofthe checkers exceed 2500oF.

The "Y" group of curves shows theaverage temperature of the exitingwaste gases at the rider arch levelaftergiving up their heat to the checkers.Again note the uniformity of the tem-peratures in Group C.

The lower curve "Z" is an averageof the incoming combustion air tem-perature for all groups studied.

Fig. 8 shows the average percent O2

(excess oxygen) present in the groupsof furnaces studied. Note the flatterprofIle of Group C.

To enable us to rate regenerators andassist us in determining the best systemdesign, wehave developed a method ofanalyzing the suction pyrometer data.Fig. 9 shows, in tabular form, the re-sults of this analysis for the furnacesstudied. These numbers are only a cal-culated representation of what reallyoccurs in a regenerator, but they can beused to show the imbalance presentand to compare or rate the systems.

The column titled "Checker PackUtilized" is, in reality, the percent offuel per port and approximates the

amount of the regenerator that serveseach port. Ideally, to achieve the fullbenefit of the checkers, the portions ofthe regenerator used by the waste gasand by the combustion air for eachport should be equal to the amount ofenergy transferred each way.

Think of a regenerator with a SU-PERFLUE system as having imagi-nary division walls that separate thesections of the regenerator served byeach port. These imaginary walls havethe ability to move in order to obtainthe best possible utilization of thechecker pack for any given operatingconditions. We do not really know

FurnaceEfficiencywhere these imaginary walls are, butwe do know that they can be movedwith the SUPERFLUE dampers to ob-tain the best possible utilization of theregenerator.

The column in Fig. 9 titled "Calcu-lated Waste Gas" showsthe theoreticalportion of the waste gas that is handledby the portion of the regenerator beingutilized. A similar condition pertains tothe "Calculated Combustion Air" col-umn and the combustion air. A num-ber higher than the "Checker PackUtilized" indicates an overloaded con-dition, while a number lower indicatesan underloaded condition. For exam-ple, Group A-Port No.1 indicates that23.2070of the total waste gas is using20.1% of the regenerator and, cor-respondingly, 15.4% of the total com-bustion air is using 20.1% of theregenerator. This represents anoverheating condition at port No.1.

Although these calculations only ap-proximate the actual conditions, theydo provide a useful tool for comparingand evaluating systems.

Fig. 10is a graphic representation ofthe valueslisted in Fig. No.9. Each pairof curves, waste gas and combustionair, has been plotted to show the devia-tion from the "Checker PackUtilized" values.

Fig. 11is another graphic represen-tation of these same values. In thischart, weplotted the total deviation be-tween the calculated waste gas andcombustion air. Note the uniformity ofGroup C.

Fig. 12 is a photograph that showsthe balanced conditions that exist inSUPERFLUE-type regenerators. Thispicture was taken under the riderarches of a six-port float glass furnacewhere No.6 port is blocked off. Thepurpose of presenting this picture is todemonstrate the uniformity of color(temperature). Those who are familiarwith this sizeof regenerator are used toseeing changes in color (yellow toblack) from end to end, and this pic-ture shows what this system can do.Smaller regenerators usually do notshow such extreme color difference,but even a temperature difference ofonly 300 of indicates a distributionproblem that could be improved.

Conclusion

In conclusion, the SUPERFLUE

system has made significant efficiencyimprovements to large capacity, sideport, regenerative furnaces. Exper-ience with the installations to date indi-cate that approximately a 17% fuel re-duction has been achieved, and wehope to improve upon this on futureinstallations. It has also provided otheroperational benefits, such as uniformrider arch temperatures, thus reducingthe possibility of failure due to over-heating. In addition to saving fuel, fur-naces with these systemshave providedimproved glass quality and a betterpercent pack brought about by themore even and constant operation.

As fuel prices continue to increase,the SUPERFLUE system should provebeneficial to medium-sized side portregenerative furnaces and should evenprove beneficial to end port furnaces.We hope to report on this aspect as wellas provide an update on results fromsubsequent system installations in PartII, which will be presented in a futurepaper.

(Editor's Note: This article is based ona paper presented by Mr. Naveaux atthe 42nd Annual Conference on GlassProblems held Nov. 10-11,1981,at theUniversity of Illinois, Champaign-Urbana.)

TEeDTOLEDOENGIN~G co.3400 Executive ParkwayToledo, Ohio 43606(419) 537-9711FAX: (419) 537-1369

Reprinted fromTHE GLASSINDUSTRY

February, 1982