ger-4203 - ge generator technology update

GE Power Systems

GE GeneratorTechnologyUpdate

Christian L. VandervortEdward L. KudlacikGE Power SystemsSchenectady, NY

GER-4203

g

Contents

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Design for Six Sigma Process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Identify . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Optimize . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Validate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Evolution to the 7FH2 Model 761/763 Generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Technology Development and Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Electromagnetic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Ventilation Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Structural Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Rotor Design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Insulation Systems and Dielectric Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

GE Generator Technology Update

GE Power Systems ■ GER-4203 ■ (04/01) i


GE Power Systems ■ GER-4203 ■ (04/01) ii

AbstractGlobal deregulation is causing major changesin the market requirements for the power gen-eration industry. Generator designs for today’sapplications must be cost effective while achiev-ing high reliability and improved efficiency. Tomeet these challenges, GE Power Systems hasintroduced three new products during the past18 months: the 7A7 and the 9A4 air-cooled gen-erators, and the 390H hydrogen-cooled genera-tors. The first 9A4 unit entered service in thesecond half of 1999. The first 7A7 and 390Hgenerators will achieve commercial operationby the end of 2000. Factory tests for each ofthese new products were completed with highlypositive results.

The focus of this paper is the design activity ona fourth new generator, the 7FH2 model761/763, to match the 7FB gas turbine plannedfor introduction in the second half of 2001.This design joins upon the highly proven fami-ly of 7FH2 designs, and provides the increasedoutput and improved efficiency needed for the7FB. Gas turbine (“leads up”) applications willuse the Model 761, steam turbine (“leadsdown”) applications will use the Model 763. GEPower Systems has established a history ofdeveloping new products by applying state-of-the-art design tools and incorporating fieldtests. Most recently, Design for Six Sigma(DFSS) tools and methodologies have facilitat-ed designing high quality and reliability directlyinto the new product development process.

The 7FH2 Models 761 and 763 generators willleverage GE’s proven track record andadvanced DFSS tools to meet rising market-place expectations. Some areas benefiting fromthese advanced techniques include insulationimprovement, design automation, electromag-netics, structural modeling, and ventilationanalysis. The resulting design will satisfy cus-tomer Critical to Quality (CTQ) characteristics

while providing design flexibility to meet vary-ing customer needs.

IntroductionGE’s development of turbine generators forpower generation applications began in the late1800s when GE leadership became interested inCharles G. Curtis’ impulse turbine. The initialgenerator designs coupled to this new productutilized air as the cooling media. Significanteffort was invested in optimizing these air-cooled machines before the capability limit ofthe air-cooled platform was exceeded. Air-cooled platform ratings reached surprisinglyhigh levels of up to 200 MVA at 1800 rpm (therating achieved by a unit installed in Brooklyn,NY, in 1932). As the need for increasedmachine rating continued, hydrogen coolingwas introduced in the early 1930s. A period ofrapid technological/capability growth followed,with GE’s first hydrogen-cooled generatorentering service in 1937.

Hydrogen-cooled generators were originallyintroduced because the combination of the lowdensity and high specific heat of hydrogenmade it an ideal cooling medium. The opera-tion, installation and maintenance of an air-cooled generator are generally simpler thanthat of a hydrogen-cooled generator.Technological improvements in generatordesign have increased the rating breakpointsbetween air-cooled and hydrogen-cooledmachines. However, the lower effectiveness ofair as a cooling medium requires a 20% to 30%size increase over a hydrogen-cooled machineof a similar rating. There is also a decrease ofapproximately 0.3% in generator efficiency foran air-cooled vs. a hydrogen-cooled generator.Hydrogen-cooled machines also can bedesigned with a higher electrical loading thanair-cooled machines due to the better cooling,and tend to have a larger subtransient reac-


GE Power Systems ■ GER-4203 ■ (04/01) 1

tance than an air-cooled machine. These factorsmake hydrogen-cooled generators preferablefor 7FH2 applications. The smaller size reducescivil engineering costs in the plant design, andthe higher efficiency provides a lifetime plantoutput gain. The higher subtransient reactancealso limits fault currents, which increases equip-ment reliability and reduces the interrupt capa-bility needed from the generator breakers.

The 7FH2 generator was introduced in 1990 foruse with the newly introduced 7F gas turbine.Throughout its history the reliability, availabili-ty and maintainability of the 7FH2 have beenexcellent. Currently the 7FH2 is being appliedto gas and steam turbines at a rating of 230MVA, with power factor of 0.85 and class “B”insulation temperature rises. As of April 2000,over 80 units have been shipped. Of these, 50units are in service with combined operatinghours of over 500,000 hours. The fleet leaderhas achieved over 48,000 hours.

As we enter the 2000s, customer requirementsare focusing on total installed cost, includinginstallation, operation, and maintenance.Designers and producers of turbine generatorshave to reduce equipment costs, improve quali-ty, and simplify operations. At the same timethey must achieve higher levels of plant effi-ciency in order to reduce power plant operatingcosts. The 7FH2 Model 761/763 generator NPIdesign program was initiated to meet thesegoals. The new design will accompany the 7FBgas turbine in the same manner that the 7FH2Model 741/743 is consistently matched with the7F/7FA line.

Design for Six Sigma ProcessDesign for Six Sigma (DFSS) tools are beingapplied to the design of this newest 7FH2model. The fundamental concept is to designquality directly into the product by developing

transfer functions to predict overall quality andfacilitate design trade-offs. Elements of thisprocess are grouped into four phases: Identify,Design, Optimize, and Validate (IDOV). Animportant aspect of this process is the ability toincorporate experience from more than 6,400GE generators currently in service around theworld. DFSS dovetails with established GE toll-gate processes that include formal structuresfor design reviews from conceptual designthrough introduction into service.

The four phases of the DFSS process used formajor generator development are describedbelow.

Identify This phase of the design process focuses ondefining the overall product requirements.Customer feedback and marketing data arereviewed to quantify and rank benefits to thecustomer, and then translate them into criticalproduct features. These features are document-ed in a formal product specification that identi-fies technical requirements, performance tar-gets, and specification limits. A “House ofQuality” is constructed by ranking the impor-tance of features that impact delivery of the per-formance specifications. This House of Qualitymethod identifies the features that are mostimportant to meeting customer requirements,or those that are Critical to Quality (CTQ).

Top level CTQs for the power plant include: (1)responsiveness; (2) on time, accurate, and com-plete delivery; (3) product technical perform-ance; and (4) price/market value. The genera-tor has the greatest impact on technical per-formance, which is most affected by plant effi-ciency, reliability/availability, and total poweroutput. The CTQs at the power plant level canbe further broken down to identify CTQs at thegenerator level. The resulting rankings fromthis “flowdown” process clarify the features most



important to satisfying customer needs. Thegenerator CTQ flowdown is shown in Figure 1.

A formal Risk Management process is incorpo-rated into the design process to significantlyreduce potential issues with the design evolu-tion. This process consists of a detailed descrip-tion of the generator’s operational characteris-tics and features, followed by brainstorming ses-sions to identify possible risks. Risk items arescored based on their probability of occur-rence, and the resulting impact on machineperformance. Remediation actions are definedto reduce the risk. An action item list is createdto insure that these follow-up actions are fullycompleted, and that risk is reduced to accept-ably low levels.

Design During this phase, transfer functions areapplied to develop the overall layout and prod-uct geometry. Transfer functions establish rela-tionships between various parameters, and canapproximate overall performance. These vari-ables include power output and MVA rating,generator efficiency, reliability, availability andmaintainability (RAM), short circuit ratio, andsubtransient reactance. A CTQ flowdown forcomponents and sub-components is performedand the results are entered into baselineProduct Quality Scorecards (PQS) as mean out-

puts and statistical variations. These values arecompared with the process specification limits

to develop the standard statistical variable, “Z”,or, in the short term, “Zst”. The baseline PQSare then populated to estimate the quality ofthe design concept and provide the basis for astatistical, quality-focused approach to thedesign process.

Entries into the scorecards are generatedthrough use of transfer functions that aredeveloped and refined by:

■ Benchmarking historical transferfunctions

■ Deriving mathematically-based closedform solutions

■ Performing analytical simulations, orconducting experiments or tests.

DFSS tools such as Design of Experiments(DOE), Finite Element Analysis (FEA),Computational Fluid Dynamics (CFD), andHypothesis Testing can be applied effectively inthis process step.

Optimize Optimization studies are performed to mini-mize the sensitivity of generator performanceto CTQ design features. These optimizationstudies identify those processes (e.g., manufac-turing or design) most in need of improved



Figure 1. Generator Critical to Quality (CTQ) flowdown process

capability. Trade-offs are assessed among theelectrical, mechanical, and power plant systemsto optimize the overall generator quality esti-mate (Zst). DFSS tools, including Monte Carlosimulations and Robust Design, can beemployed during this step. The selection andapplication of these tools typically is prioritizedby the relative Zst scores of each line item andtheir contribution to defects per unit (DPU).Optimization is an iterative process that caninclude multiple prototype hardware tests toassess the performance of each optimizationattempt.

Process capability data are collected from man-ufacturing and sub-component vendors for theprinciple processes used to build the bill ofmaterials (BOM) of the concept generator.Manufacturing process capability is assessed byusing the Product Quality Scorecards to list thequality score (Zst) for each BOM item.

The PQS and the process capability assessmentresults are used to determine tolerances fordrawing details. Some tolerances can be statisti-cally expressed for sub-component or assemblyfeatures critical to the quality of the generator'sperformance. Tolerance values for these criticalfeatures are best determined by reviewing thePQS and the process capability data availablefrom manufacturing and engineering. As toler-ances for each drawing are established andresults entered into the PQS, the results arecomputed to obtain quality ratings (Zst) foreach generator performance line item. Theseline item ratings are used to calculate an overallquality rating for the generator system. The ele-ments and results of this “flow-up” of the quali-ty scores are typically used as a basis for discus-sion in the detailed design review.

Validate The validation of design tools, transfer func-

tions, and resulting designs is of critical impor-tance. This validation often can be obtained bycomparisons to historical results or perform-ance of sub-scale performance tests. Actualcomponent or subsystem operation is, ofcourse, the best measure of design success. Thisemphasizes the importance of using an evolu-tionary approach to product design, whereexisting, proven components are applied to thegreatest extent possible.

Finally, factory testing of all NPI generators isperformed to validate the overall performanceand verify that the performance specificationshave been satisfied. DOE techniques areapplied to create appropriate test plans and theresults are used to generate transfer functionsfor performance assessment. PQS are used todemonstrate that the required level of qualityhas been met. Further validation data are col-lected during startup and subsequent commer-cial operation of the generator. These data andobservations can be used to further confirmthat the product meets the required perform-ance specifications.

Figure 2 provides a summary of all four DFSSphases (Identify, Design, Optimize, andValidate).

Evolution to the 7FH2 Model 761 / 763GeneratorThe overall approach to design of the Model761/763 is to apply experience and, whenappropriate, existing hardware componentsfrom the present 7FH2 models. Common con-cept allows carryover of production methodsthat were tailored to the 7FH2. Existing designsof major parts (such as end shields, coolers, andbushings) and small parts (such as fasteners)can be reused. CAD/CAM packages allow forthese small parts to be stored as three-dimen-



sional models for importing into the new designmodels. Larger part models can be copied froma previous design and modified for a newdesign, significantly reducing design and draft-ing time.

These techniques facilitate design for assembly,further enhancing the quality of the finishedproduct. Modern software can animate 3Dgraphic models and determine if a path existsfor assembly or removal of a component in thepresence of other parts. Potential interfaces canbe quickly identified and resolved without cre-ating scale models or manipulating hard copiesof drawings.

The frame of the 7FH2 Model 761/763 is verysimilar to existing 7FH2 models, and canaccommodate the electrical design geometrywith few changes. Efficiency for the new modelsis increased by approximately 0.1% while main-taining the same basic footprint. The designteam will capitalize on proven features andcomponents to best utilize existing drawings.Existing interfaces from the 7FH2 will be car-ried forward whenever possible to make thetransition and introduction of the 761/763transparent. Overall risk is substantially reduced

because both performance and cost structureof the current 7FH2 family are well established.Of particular note, the Model 741/743 endshields will be applied to the Model 761/763.Table 1 provides a summary of the Model741/743 and 761/763 performance parame-ters. Figure 3 shows the actual Model 761 gen-erator configuration.

To increase overall efficiency of the 7FB, a“once-through” ventilation system has beenchosen. This system reduces the required flowand corresponding windage loss, and also sim-plifies frame construction. The flow configura-tion employs an axial fan on each end that sup-plies the overall cooling flow. There are threemain parallel flow splits from the fan discharge:the rotor, the stator end turn region, and the airgap.

Rotor (field) coils are cooled through uniform-ly spaced radial ducts supplied through theaxial inlet manifold. Gas enters the rotor underthe retaining ring and exits through a series ofradial ducts that are machined in the field wind-ing conductors, removing heat from the cop-per. The gas distribution and heat transfer inthe slot region of the rotor are controlled by the



Figure 2. DFSS IDOV phases

design of the subslot and the radial ducts. Thesize of the subslot controls the quantity of gasentering the radial ducts. Detailed DFSS analy-sis helps achieve the optimum design of the sub-slot, given the mechanical and electromagneticdesign constraints. Modeling and analysis of theradial ducts allow the selection of a suitableduct profile that can be efficiently manufac-tured. The rotor end region receives indirect

cooling as gas is scooped and circulated withinthe end-strap cavities.

The stator core also receives cooling via radialducts. However, non-uniform package sizing isspecified to optimize overall cooling efficiency.All outer radial sections are employed as exitregions from the stator core and provideplenums to feed the four coolers.



7FH2 Model 741 / 743 7FH2 Model 761 / 763

Hydrogen directly cooled rotor

Cooling and conventionally cooled stator same

Configuration Single-end drive, end shield mounted same

Rated Speed 3600 rpm/60 Hz same

Output 195.5 MW/60 Hz 212.5 MW/60 Hz

Power Factor 0.85 lag Same

MVA Rating 230 MVA/60 Hz 250 MVA/60 Hz

Terminal Voltage 18 kV Same

Temperature Rises Allowable Class B per IEC/50 Hz

And ANSI/60 Hz Standards Same

Insulation Class Rotor - Class F; Stator - Better than Class F Same

Excitation System Bus Fed, Static Excitation Same

Table 1. 7FH2 Model 741/743 and Model 761/763 performance comparison

Figure 3. Three-dimensional view of the 7FH2 Model 761/763 generator

The air gap is the exit manifold for the rotor,and the inlet manifold for the stator. The flow iscomprised of swirling axial flow mixing fromthe fan and rotating radial jets from the dis-charge of the radial ducts. The multiple pathsof the stator end region receive flow from thefan and discharge directly into the stator outletradial sections. The rotor end region receivesindirect cooling as gas is scooped and circulatedwithin the end-strap cavities. Figure 4 provides arepresentation of the Model 761/763 coolingconfiguration.

Technology Development andApplicationGenerator design requires skill in a broad rangeof electrical, mechanical and material technolo-gies, at the heart of which is the development ofelectromagnetic design tools. Ventilation andheat transfer, structural design, electrical lossevaluation, and rotor dynamics are all criticaltechnologies for hydrogen-cooled generators.For the machine to perform as predicted, theanalytical tools used in the design have to be val-idated.

Insulation systems applied to hydrogen-cooledunits have to meet many of the same require-

ments as those used in air-cooled machines,including low thermal resistance and thermalcycling capability. Larger hydrogen-cooled unitstend to be designed for higher terminal volt-ages and have greater electromagnetic forcesthan air-cooled designs. They require a some-what greater level of dielectric capability andmechanical toughness than air-cooledmachines, but the differences are not so greatthat the same insulation systems cannot be usedin both applications.

Electromagnetic Analysis The electromagnetic design of large generatorslies at the leading edge of the art. Energy den-sities are higher than in almost any other elec-trical apparatus. Flux densities, currents andvoltages, and electromagnetic forces are high.Conventionally-cooled designs require adetailed knowledge of the loss distribution inthe machine, so the correct amount of coolingcan be employed at each loss location.

Electromagnetic analysis of electrical machinesis greatly facilitated by the increasing refine-ment and ease of use of finite element electro-magnetic analytical tools. Three dimensionalelectromagnetic field solutions are derived for



Figure 4. 7FH2 Model 761/763 “once-through” cooling configuration

both the stator and rotor. They allow the design-er to accurately predict losses in these compo-nents, to adequately cool them, and to mini-mize design complexity. An example is shownby Figure 5 example where losses are calculatedin the copper flux shield at the end of the core,enabling the designer to confidently choose thecorrect design for the particular application.

Electromagnetic loss calculations are also usedto analyze the excitation requirements of thedesign to assure adequate excitation systemmargins. When an application has an unusualoperating requirement, such as additional lead-ing power factor capability, these tools are usedto design the ends of the stator core, taking intoaccount the leakage fields of the stator androtor end windings.

Ventilation Design The ventilation pattern chosen for a givendesign depends on the length of the machineand the temperature distribution in its variouscomponents, particularly the stator and fieldwindings. Gains in product efficiency can beachieved by optimizing the ventilation circuit tominimize pressure drop while maximizing cool-

ing effectiveness. Advances in ComputationalFluid Dynamics (CFD) analysis allow detailedevaluation and prediction of the flow in variousparts of the ventilation circuit. FLUENT/UNS,a general-purpose computational package forsolving a variety of heat transfer and fluiddynamics problems, is used extensively in gen-erator ventilation and cooling analyses.Numerous models are generated, with each onefocused upon a given element of the ventilationcircuit. Results of an analysis are often appliedas boundary conditions for the adjacent model.

One such case is shown by Figure 6, where theflow path and pressure drop is modeled at thestator-rotor gap. The ventilating gas is forced toflow into the gap by a ventilating fan mountedon the rotor. A “bottleneck” is formed betweenthe retaining ring nose and the core end taperat the gap entrance. In addition, the retainingring nose and the rotor core form a backward-facing step. As cooling gas passes the stator-rotor entrance, it generates a large flow circula-tion at the retaining ring nose. This results in alarge pressure drop at the gap entrance. For theModel 761/763, techniques will be evaluatedfor reducing these losses along the axial cooling



Figure 5. Core end heating analysis

flow. Theoretically, it is feasible to reduce thestator-rotor gap entrance pressure drop byapproximately 60%.

Cooling of the end winding region and the slot(or body) region are interrelated due to the ris-ing temperature of gas as it exits the end turns,and the axial heat conduction in the copper.The gas flow patterns under the retaining ringare particularly difficult to understand andquantify. Variations in peripheral velocity andgas density, along with the Coriolis effect due torotation, affect the movement of the coolinggas. CFD analysis is helping to develop a greaterunderstanding of the cooling effects, and lead-ing to enhanced cooling of the end turns.

The design of the fans that circulate the coolinggas through the generator is critical to the over-all ventilation performance. The fan needs tobe efficient and producible. CFD is used to pre-dict fan performance accurately and determinethe impact of changes in fan geometry. Figure 7provides a representation of calculated fan per-formance and overall system resistance. By uti-lizing CFD, new fan designs can be rapidlydeveloped and optimized in conjunction withthe overall ventilation circuit.

Structural Design Generator structural design requires a thor-ough understanding of structural interactions



Stator Core Region

Rotor Retaining Ring

GasFlowFromFan

Figure 6. CFD - Cooling gas flow across the retaining ring

Flow (cfm)

Pre

ssu

re (

inch

es H

2O)

17 Deg Blade

18 Deg Blade

7FB Resistance

Figure 7. 7FH2 Model 761/763 fan performance

for all internal generator components as well asthe interactions between the generator, founda-tion and drive train. Coupled with the design ofcomponent interactions is the requirement tooptimize material usage to meet design limitsand minimize generator costs. The design of anew generator makes extensive use of FiniteElement Analysis. All major components of thegenerator are modeled and assembled into acomplete generator assembly model, as shownin Figure 8.

The assembly model is used to analyze the gen-erator vibration due to electromagnetic forcesand rotor unbalance. It also evaluates acousticnoise performance and performs stress analysisfor all loads to which the generator will be sub-jected. These loads include normal loads suchas generator lifting, shipping and handling,internal operating pressure and rated operatingtorque. Emergency transient loads are also con-sidered, such as over-pressurization and syn-chronizing out-of-phase torques.

Robust design tools and techniques can opti-mize the generator design and ensure adequate

performance over a wide range of noise param-eters. A parametric geometry-driven programnamed AMEA (Automated Mesh AnalysisInterface) has been developed to allow quickand efficient assembly of Finite Element Models(FEM) for numerous components. AMEAallows concurrent design optimization, model-ing and analysis of the various generator com-ponents while continuously maintaining theconnections to the complete generator assem-bly model. Tools have been developed to effi-

ciently perform DOE design optimization andperformance evaluation over a wide range ofuncontrollable noise parameters and externalinfluences.

The Finite Element Model is exercised to inter-rogate the generator assembly for all loadsencountered during assembly and operation.Electromagnetic force is the most likely force tocause structural issues. This force is cyclic andacts in the radial direction at the inside diame-ter of the stator core with a magnitude of onemillion pounds. The resulting stator core vibra-



Figure 8. 7FH2 Model 761/763 Finite Element Model

tion and the transmission to the generatorstructure and foundation is a significant designconsideration. Forced harmonic responseanalyses are performed to ensure that the elec-tromagnetic forces cannot excite the machine’snatural frequencies.

The structural design of the stationary compo-nents also must be considered when calculatingthe dynamic behavior of the rotor, since therotor is supported on bearings located in theend shields of the machine. In this load config-uration the structural vibratory loads caused bythe rotor, and the loading caused by statorvibration that drives rotor behavior, are interro-gated. Once again, a forced harmonic analysis isperformed to understand and optimize theinteractions. Lastly, the structural design has amajor impact on the overall producibility andserviceability of the generator. The complexityof the fabrication determines the unit’s machin-ing cycles as well as its accessibility for thor-oughly cleaning the inner cavities of themachine before shipping.

Rotor Design As with the structural design, the rotor design istightly linked to the electromagnetic design.The electromagnetic design determines the sizeand proportions of the rotor, which, in turn,determines the dynamic behavior of the rotor.The designers will explore a number of possiblesolutions to the overall machine design, lookingfor the best combination of rotor critical speedsand overall machine performance. Smoothrunning generators have been achieved byimplementing sophisticated FEM-based rotordynamics tools and applying rigorous high-speed modal balance procedures. The vibrationperformance demonstrated in service is excel-lent and well within ISO Standards require-ments (see Refs. 1, 2, 3, 4, and 5). The mathe-matical models that were used accurately repre-

sent the dynamic properties of the rotor and itssupport system, with particular attention toaccurate stiffness and damping modeling of thebearing oil film.

Modern design philosophy permits rotor-lateralcritical speeds in the vicinity of operation speed,as long as the modes are sufficiently damped.Bearing selection ranges from elliptical pad totilting pad, which each have application-specificadvantages. The rotor dynamics optimizationprocess determines selection of the appropriatebearing type. Torsional vibration design studiesensure that the turbine generator rotor systemis robust to transmission network electrical dis-turbances and malsynchronization accidents.They also address the impact of continuouslyacting stimuli arising from unbalanced loads(see Ref. 6 and 7). The modal damping isextremely low for torsional vibration, so it isvital to demonstrate acceptable frequency sepa-ration margins in the design process, particu-larly in regard to the second harmonic of sys-tem electrical frequency.

As gas turbine designs move to higher firingtemperatures and higher ratings, the torquesrequired to rotate the turbine during startingincrease rapidly. This is particularly true for sin-gle-shaft steam turbine and gas turbine (STAG)combined-cycle units. The inertia of both thesteam turbine and the gas turbine, along withtheir aerodynamic losses, have to be supplied bythe starting means. This has led to the use ofthe generator as a starting motor for the gas tur-bine. Careful design analysis is required toensure that the effects of the current harmonicsfrom the power supply, and the application ofexcitation at low speed, are accommodated reli-ably.

A key design goal for the Model 761/763 gen-erator is increased efficiency. One technique toraise efficiency is to increase the cross-sectional



area of the field winding turns, which reducesthe resistive losses in the field winding. Rotorwedges are components that hold the fieldwinding turns in the rotor slot while the rotor isspinning. A thinner rotor wedge allows morespace for field winding copper and increasesgenerator efficiency. A DFSS project was initi-ated to optimize the rotor wedge design. Thegoal was to minimize wedge thickness, whileensuring that the design had sufficient strengthto provide high generator reliability.Optimization technology and statistical designmethods were applied to achieve this key cus-tomer CTQ.

Numerical optimization techniques wereapplied to create a candidate wedge design.The Design of Experiment approach was thenused to create response surfaces that character-ized wedge stresses as the dimensions of thecandidate design were varied over the toleranceranges. A response surface example for the VonMises stress at the top of the wedge dovetail isshown by Figure 9.

To ensure adequate stress margin (or high reli-ability), statistical material strength data frommaterial tests was included with the calculatedstresses in a Monte Carlo analysis. This analysis

calculated the probability of positive stress mar-gin (the difference between the wedge stressand the design strength limit). Positive stressmargin means the wedge has more strengththan required. An example of the stress marginprobability distribution is shown in Figure 10.

The Monte Carlo analysis shows how applica-tion of optimization techniques and statisticaldesign methods can directly improve customervalue.

Insulation Systems and Dielectric Design The insulating materials in a high voltage gen-erator occupy valuable space, and must be capa-ble of conducting heat from the winding to thecooling gas. As a result, the insulation systemdesigner is continually challenged to developinsulation systems that occupy minimum space,are capable of handling higher electrical stress-es, and have maximum thermal conductivity.Figure 11 shows how the thickness of the statorgroundwall insulation might affect the size of atypical conventionally cooled machine.

The mechanical forces on the insulation systemcan be very high. Forces on the stator windingare especially high during the time that an elec-trical fault is applied to the machine. The statorinsulation must be a chemically stable, rigidstructure that can accommodate modest bend-



0.015

0.01023600

0.08

23700

DWidthClr

23800

23900

24000

0.09

24100

0.100.005

VM

0.11Depth3

Figure 9. Response surface for rotor wedge optimization

Frequency Chart

psi

.000

.016

.032

.049

.065

0

6492

0.00 11250.00 22500.00 33750.00 45000.00

100,000 Trials 0 Outliers

Forecast: DoveTopVonMisesMar

Figure 10. Monte Carlo analysis of rotor wedgestress

ing at winding without developing cracks orvoids in the ground wall. The rotor insulationsystems are exposed to very high “g” loadingsdue to the centrifugal forces on the copperwinding, ranging up to 9000 g on a large 3600-rpm rotor.

Rotor insulation systems are as crucial to thereliability of the machine as the stator windinginsulation systems. Due to lower voltages, therotor insulation is not as thick as the stator insu-lation. However, the mechanical duty is highduring assembly of the windings and duringoperation. As a result of these operatingrequirements, the mechanical properties of theinsulating materials and the mechanical designof the insulation systems are as crucial as theirelectrical performance. An additional factor inevaluating the rotor insulation system is thepresence of voltage spikes generated by theexcitation system. High-response static excita-tion systems frequently operate with full ceilingvoltage applied to their AC or source side. Theappropriate level of DC voltage to be applied tothe field is determined by controlling the firingangle of the power thyristors in the bridges. The

field winding will be exposed to full AC voltagelevels and switching spikes that reach the fieldceiling voltage several times per cycle.

In addition to the mechanical forces and dielec-tric duty, thermal expansion of the stator androtor windings are among one of the major con-tributors to loss of life of the insulation systems.Restriction of the relative motion of systemcomponents produces unacceptable strain, andrepeated thermal cycling causes abrasion, bothof which can reduce the life expectancy of theinsulation. Successful operation over theexpected life of the machine requires a ther-mally stable insulation system in both the statorwinding and the field. Thermal cycling andexposure to the hot conductor throughout thelife of the generator can break down a poorquality insulation system (materials and/orapplication of the materials). The properties ofthe insulation being designed must considerelectrical and mechanical performance charac-teristics at operating temperatures.

The same Six Sigma approach and statisticalmethods used to design the rest of the machineare used to optimize the combination of mate-rials and processing parameters. DFSS method-ology organizes the development process andDOEs are performed to screen material combi-nations. The Product Quality Scorecards areused to compare performance and select theappropriate insulation system design.

Testing Validation of the machine design is obtained bycombining component, assembly, and machinetesting. GE’s philosophy is to test a prototypemachine whenever a significant design changeis made. It is not possible to factory test themachine under full-load conditions. However, itis possible to perform no-load open and shortcircuit tests to confirm the electromagnetic andthermal characteristics per ANSI and IEC stan-



Figure 11. Relationship between machine outputand insulation thickness

dards. Sudden short circuit tests are also per-formed to confirm winding mechanical designand measure reactances. Unbalanced loadingtests are performed to confirm negativesequence capability. Special tests to confirmspecific design features are performed asrequired. In addition to the overall machinetests, a number of special tests are performedon prototype machines during assembly to con-firm structural frequencies, stator end windingvibration and other important design features.

ConclusionsThe newest additions to the 7FH2 family, theModel 761/763 generator design will provideequally high levels of customer satisfaction as dothe current models. Their introduction followsclosely on the heels of the successful 9A4, 7A7,and 390H product releases. Each of these threedesigns was thoroughly validated by factory testsand the 9A4 has entered commercial operation.

The 7FH2 Models 761/763 will provideincreased efficiency and output beyond theModel 741/743, while incorporating many ofthe key features and components that havebeen thoroughly validated through extensiveusage. Critical elements such as reliability, avail-ability, and maintainability will be carried for-ward. Cost effectiveness will be achievedthrough close coupling of the Model 761/763program with ongoing product improvementactivities for the Model 741/743. The Model761/763 generator will satisfy all CTQ customerrequirements for utilization with the 7FB gasturbine combined-cycle power plant.

References1. “Advances in Turbogenerator Technology,”

B.E.B. Gott, IEEE Electrical InsulationMagazine, Vol. 12, No. 4, July/August 1996,pp. 28-38.

2. “Cycle Power Train,” T. Bonner, B.E.B. Gott,R.E. Fenton, and B. Sherras, CIGREConference, 1997.

3. “New Hydrogen-(NH1)-Cooled Generatorfor Single-Shaft Combined-Stator WindingSystems with Reduced Vibratory Forcesfor Large Turbine-Generators,” C.H. Holleyand D.M. Willyoung, IEEE Transactions onPower Systems, Vol. PAS 89, No. 8,November/December 1970, pp. 1922-1934.

4. “Experience and Recent Development withGas Directly Cooled Rotors for Large SteamTurbine Generators,” B.E.B. Gott, C.A.Kaminski, A.C. Shartrand, IEEETransactions on Power Systems, Vol. PAS103M, No. 10, October 1984, pp. 2974-2981.

5. “Static Starting of Gas Turbine GeneratorsMechanical and Electrical Considerations,”R.E. Fenton, G. Ghanime, E.E. Kazmierczak,R.D. Nold, presented at the 1996 CIGREMeeting, Paris.

6. “Advances in Design Practices to ImproveRotor Dynamics Performance of MediumSized Generators,” R.E. Fenton, D.R. Ulery,D.N. Walker, CIGRE Paper 11-205, 1994Session.

7. “Advances in Design Practices to ImproveRotor Dynamics Performance of Medium-Sized Generators,” R.E. Fenton, D.R. Ulery,D.N. Walker (General Electric Company),presented at the 1994 Biennial Meeting ofCIGRE, Paris.

8. “Turbine-Generator Shaft Torsional Fatigueand Monitoring,” D.N. Walker, R. Placek,C.E.J. Bowler (General Electric Company),J.C. White, J.S. Edmonds (EPRI), presentedat the 1984 Biennial Meeting of CIGRE,Paris.



List of FiguresFigure 1. Generator Critical to Quality (CTQ) flowdown process

Figure 2. DFSS IDOV phases

Figure 3. Three-dimensional view of the 7FH2 Model 761/763 Generator

Figure 4. 7FH2 Model 761/763 “Once-through” cooling configuration

Figure 5. Core end heating analysis

Figure 6. CFD – cooling gas flow across retaining ring

Figure 7. 7FH2 Model 761/763 fan performance

Figure 8. 7FH2 Model 761/763 Finite Element Model

Figure 9. Response surface for rotor wedge optimization

Figure 10. Monte Carlo analysis of rotor wedge stress

Figure 11. Relationship between machine output and thickness

List of TablesTable 1. 7FH2 Model 741/743 and 7FH2 Model 761/763 performance comparison



ger-4203 - ge generator technology update

Documents