Assessment of the Assemblability of Aero Engines on the

ASSESSMENT OF THE ASSEMBLABILITY OF AERO ENGINES ON THEBASIS OF A LPT MODULE

J. Rendle (GSaME), S. StaudacherInstitute of Aircraft Propulsion Systems

University of StuttgartPfaffenwaldring 6, 70569 Stuttgart, Germany

AbstractThe aviation industry and air traffic have been growing constantly for the last decades and an end of thistrend is not predicted for the next decades. As a result of this the pressure on aero engine manufacturers isincreasing to evaluate life-cycle related costs of future products as early as possible. One aspect of this isthe assemblability of the product which is initially assembled at the OEM’s production facility but later on it isalso subject to Maintenance, Repair and Overhaul activities. Since assemblability has a significant influenceon life-cycle-related cost a systematic method for the evaluation of assemblability during preliminary designis proposed in this paper. The developed method considers the boundary conditions that are necessary toevaluate the assemblability of an aero engine during preliminary design, which only exists at a low level of detailin this phase. The method considers requirements of the preliminary design concept, the assembly process,and the assembly system in order to derive characteristic inter-dependencies and assembly time allocations ofthe chosen concept. Subsequently, the developed method will be applied to a Low Pressure Turbine module todiscuss the effect of the Level of Detail on the assembly time estimation.

NOMENCLATURE

CAD Computer Aided DesignLoD Level of DetailLPT Low Pressure TurbineMTM Methods Time MeasurementOEM Original Equipment ManufacturerPD Preliminary Design

1 INTRODUCTION

As a result of the constant growth of global air trafficthe demand for new airplanes and hence new aero en-gines is increasing [2]. Since the production capacityfor aero engines is not growing as fast as the projecteddemand of new aircrafts the backlog of orders mightrise [2,6,14,16]. In order to accept this challenge forfuture and today’s products, the throughput time fromorder to delivery within the companies has to be re-duced. The overall producibility of aero engines has tobe improved significantly to achieve this. Therefore, atransition process from a manufacture-like small seriesproduction to a series production, which can handlethe growing demand is needed. Enablers in terms ofthroughput time of this transition from a technologicallyfeasible product towards an economically producibleproduct are specifically the assemblability of the prod-uct and its sub assemblies.Previous research in the field of preliminary design ofaero engines mainly focused on aerodynamics, per-formance, engine weight, and dimensions of the aero

engine concept [4]. In order to increase competitive-ness, it is essential to additionally evaluate a futureproduct and its assemblability as early as possible.Hence, an approach for the assessment of the assem-blability of aero engines during preliminary design willbe presented in this work.

2 PRODUCT DEVELOPMENT OF MODERN AEROENGINES

The product development process of new aero en-gines typically is organized in different developmentstages. As of the large investment cost and significantdevelopment time , this formalized process is feasible.Nevertheless, estimation of the costs of a new engineproject is difficult because of increasing material andlabor costs as well as increasing costs for qualificationand certification [9,10,12,15].A typical Product Development Process for aero en-gines can be broken down into five major phases, thefirst phase being Preliminary Design. [10]. The im-portance of preliminary design in the overall productdesign process because of its impact on architecture,weight and dimensions, as well as life-cycle relatedcosts is undoubted. To aid in the process of gener-ating as much knowledge of the product as early aspossible, numerous computer based tools such asGasTurb, NPSS, MOPEDS, PMDO, Genesis etc. weredeveloped by research institutions and aero engineOEMs [4,8,12,13,17].In order to increase the knowledge of the product dur-ing preliminary design even further, assembly informa-

Deutscher Luft- und Raumfahrtkongress 2015DocumentID: 370009

1

Cum

ulat

ive

Life

-Cyc

leC

ost

Kno

wle

dge

ofth

ede

sign

Previous PD-Tools

Assembly Evaluationduring PD

Time0 %

25 %

50 %

75 %

100 %

0 %

25 %

50 %

75 %

100 %

PreliminaryDesign

DetailDesign

Development& Qualification

Production(Assembly)

ProductUse

1%

7%

18%

50%

Incurred Cost

70%

85%95%

Determined Cost

5%

20%

60%

Knowledge of the design

Fig. 1: The cost of design vs. knowledge and the advantages of PD-Tools

tion need to be considered to evaluate the assembla-bility. However, the knowledge of the design is linkedwith Life-Cycle Cost because more knowledge of thedesign allows more profound decisions, which influ-ence the downstream design activities.Fig. 1 displays the typical progressions of costs andknowledge for an aero engine project along its devel-opment cycle. During preliminary design, the amountof incurred costs is very low with 1% as well as theknowledge of the design with 5%, whereas 70% ofthe costs are determined through design decisions inpreliminary design. Aided by computerized preliminarydesign tools, the amount of knowledge about the prod-uct can be increased early on, which is displayed bythe thick dashed line in Fig. 1. [4,7,10]. Assembly eval-uation is typically conducted in later phases, such asdevelopment & qualification and production [3,5]. How-ever, the incorporation of assembly information as wellas information about tools, fixtures, and workers intopreliminary design activities increases the knowledgeof the product even further, as indicated by the boldline. Eventually, this leads to time and monetary ad-vantages because the knowledge about the product aswell as knowledge about the corresponding assemblysystem is generated simultaneously, and therefore thelatter can be analyzed and specified at an early point oftime in the product development process [10,12,18].

3 ASSEMBLABILITY

In order to conduct assembly evaluations in early de-sign stages, a detailed analysis of the overall productstructure is necessary. In this context, the conceptsneeds to be analyzed to understand the interactionsof design features, product architecture, and assemblyprocesses on the assemblability.Assembly refers to all processes of combining differ-ent manufactured and geometrically defined parts toform a final product. A product in this context mayonly consist of parts or of different sub assembliesand parts. Typically, assembly is the last value addedstep of production [11]. The method of assembly canbe defined as the way how parts, sub assembliesand fasteners are assembled to form a final productwith regard to different technological constraints andtherefore characterizes the assemblability of the prod-uct [3,11,18,20]. In general, a limited number of subtasks of assembly can be defined, as displayed inFig. 2. These sub tasks are linked to the superordinateassembly system as well as to product elements, suchas parts and fasteners. However, the assembly systemis an integral part of production and the overall productdevelopment process and therefore influences the as-sembly process as a whole. As seen in Fig. 2, certainproduct elements influence the assembly process andvice versa. To assess the assemblability of a productdifferent considerations of interdependencies betweenassembly process and product elements need to beconducted, especially in a development phase with

Deutscher Luft- und Raumfahrtkongress 2015

2

ProductionEngineering

PurchasingManufacturing Assembly Logistics

material

Information

sub tasksof assembly

handling joining checking

positioning grasping storing traction interlock adhesive position quality

parts fasteners

level of detail geometry tolerances materials type number

assembly system

AssemblyProcess Tools Fixtures capacity time

Fig. 2: Inter-dependencies of product elements and the assembly process

limited knowledge about the product and the eventualassembly system.

Digital assemblability evaluationA method for the assessment of assemblability in earlyproduct development phases is the digital product val-idation method proposed by WACK et al. [18], whichutilizes digital models for the assembly evaluation atdifferent levels of detail, as seen in Fig. 3.

production-relatedproduct validation

product-relatedprocess validation

production-relatedprocess validation

resourcevalidation

e.g. assembly precedencegraph, de-/mounting paths,etc.

e.g. joining technology, ac-cessability, tolerances, etc.

e.g. ergonomics and timeevaluation, etc.

e.g. handling tools, logisticsequipment, etc.

Fig. 3: Assemblability evaluation through digital prod-uct validation [18]

The different stages of the design process are exe-cuted in CAD systems. During this stages a lot of digi-tal data is generated, which can be used for digital as-semblability evaluation in an early stage to validate the

product and future assembly systems. Through suchan approach, cost and time advantages can be utilized,such as a reduction of physical assembly ramp-uptests, earlier knowledge of assembly equipment, andearly training of assembly personnel [18, 19]. WACKet al. propose a 4-stage model for the digital productvalidation prior to a production ramp-up, as displayedin Fig. 3. Based on a production-related product vali-dation, the designed product is evaluated in terms ofbuildabilty. For instance, collision analyses of partsalong the mounting and demounting paths are exe-cuted as well as different assembly sequences, whichcan be described in assembly precedence graphs toevaluate the product during this stage. During the sub-sequent stage of product-related process validationthe value added processes for the assembly of theproduct are analyzed. This includes for example dif-ferent joining technologies, tolerances of the parts, aswell as the accessability of joining elements. In thefollowing stage, a production-related process valida-tion will be conducted, which also includes non-valueadded aspects of assembly such as ergonomics andworkstation layout. In addition to that, time estimationsand evaluation can be conducted as well. To concludethe digital product validation, a resource validationanalysis is conducted at last, which mainly includes lo-gistics equipment (e.g. carts, boxes, etc.) and aspectsof the assembly system as well as handling tools [18].

Deutscher Luft- und Raumfahrtkongress 2015

3

4 METHOD FOR THE ASSESSMENT OF ASSEM-BLABILITY DURING PRELIMINARY DESIGN

Given limited resolution of predicted details duringthe preliminary design of aero engines, only a certainamount of accuracy can be expected from an evalua-tion of assemblability at this stage.Based on the method of WACK et al. [18] only thestage production-related product validation can beevaluated in a satisfactory manner during preliminarydesign. Depending on the level of detail of the prelim-inary design concept the stages product-related pro-cess validation and production-related process vali-dation can be carried out partially. In order to achievefeasible results for further design considerations thesystematic method for the assessment of assemblabil-ity during preliminary design is derived as displayed inFig. 4.

Geometry Analysis

Preliminary Design Concept

Evaluation of Assemblability

Estimation of theAssembly time

Evaluation of theLevels of Detail

Data of previous PD activities

Data of Production and factory systems

To next Stage

no

yes

plausible

refinement

Fig. 4: Systematic procedure of the deduced method

Geometry AnalysisThe starting point is a geometry analysis similar tothe consideration displayed in Fig. 2. The proposeddata of the preliminary design tools and activities aswell data of the production and involved factory sys-tems (e.g. fixtures and tools, assembly process expe-rience, capacity) are analyzed. The focus lays on thedetection of inter-dependencies of product elementsand technological aspects of the assembly system.

Preliminary Design ConceptA simplified, fully parametric 3D CAD model is setup to assess the preliminary design concept interms of assemblability. The model is based on aparameter-table, which includes all geometric infor-mation of the parts and therefore allows direct linksbetween the dimensions of adjacent parts. In addi-tion, the number of parts, specific design features,as well as the spatial position of parts and sub-assemblies can be specified in the parameter-table.

Evaluation of AssemblabilityDifferent mounting and demounting paths are eval-uated manually in the CAD system. In ad-dition, an assembly precedence graph is gen-erated manually. This graph displays the se-quence of tasks in which the different compo-nents are combined to form the LPT module [5].

Estimation of the Assembly timeBased on the assembly precedence graph, the differ-ent assembly tasks can be evaluated with differentmethods to estimate the required time for the execu-tion of the specific workloads. The manual assemblyprocesses are evaluated with the MTM method, whichutilizes standardized work step elements for the es-timation of the required assembly time [1]. In thiscontext, MTM is used as a remote analysis tool basedon the geometry of the CAD model. The evaluationsteps are performed without previous experience ofmanual assembly steps in the factory environment.

Evaluation of the Level of DetailThe generated simplified 3D CAD model serves as thebasis for the calculation of the Level of Detail of theconcept. This figure of merit is calculated through aseparation of the LPT module in different subgroupsand features. The features are rated with 0 (not con-sidered), 1 (sketched/estimated), and 2 (designed withcommon rules). The average score of all the con-sidered subgroups eventually resembles the Levelof Detail. Depending on the focus of the concept,different subgroups may be defined and evaluated.

Plausibility checkIn the final stage of the method, weight and part countestimations are conducted as well as an assembly timeestimation. Through these evaluations the explorationof the remaining design space and further design ac-tivities can be conducted more target oriented. If theresults are not plausible, a refinement loop may beinitialized.

Deutscher Luft- und Raumfahrtkongress 2015

4

5 APPLICATION OF THE METHOD AND DISCUS-SION OF THE RESULTS

In order to test the suitability of the introduced methodit is applied to a 5-stage LPT module of a civil aeroengine at two different Levels of Detail (LoD). A 3DCAD model of the LPT served as a reference.As a means to classify the different parts of a LPTmodule, the following distribution is proposed. Partsor components, that are dependent on the chosen ar-chitecture are classified as A-parts (e.g. case, discs,blades, vanes). Parts that are necessary to fixate A-parts are classified as B-parts (e.g. fasteners, sealplates). Additional parts that are subject to later de-sign activities are classified as C-parts (e.g. coolingmanifolds, blade dampers, bearing structures).

5.1 Parametric 3D CAD model

In order to achieve a suitable parameter set for theset up of the parametric CAD model, different simpli-fications were made. As an example, circumferentialgrooves are displayed in Fig. 5, which serve as theinterface of the case for vane segments as well astip seals. Both, vane and seals, engage the grooveutilizing a key with the respective geometric values.

Case

VanesStage 1

VanesStage 2

BladesStage 1

Tip Seal Stage 1

Grooves and Keys

Fig. 5: Simplified interface of turbine case

The first evaluation considers only A-parts of the5-stage LPT module. According to the introducedmethod for the calculation of the Level of Detail insec. 4 this results in a LoD of 55% (LoD A). Followingthe geometry analysis, the simplified, fully parametric3D CAD model is set up as displayed in Fig. 6. Themodel is shown on a simplified assembly fixture andcomprises one case, five discs, a total of 277 blades,vane segments for each turbine stage, as well as tipseal segments. The numbers of blades and vane seg-ments are derived from the reference model whereasthe numbers of tip seal segments are estimated. Ex-

Fig. 6: 3D CAD model of evaluation with LoD A - onlyA-parts

ternal accessories, as well as the LPT shaft are notconsidered at this point.The second evaluation additionally considers fasteners(F) and seal plates (S) as representatives for B-parts,as displayed in Fig. 7. The quantities of seal platesand fasteners between the different turbine discs areestimated. The Level of Detail is calculated to 61%(LoD B).

Vanes Stage 1

Vanes Stage 2

Blades Stage 1

Blades Stage 2Seal

Plates (S)

SealPlates (S)

fasteners (F)

Fig. 7: 3D CAD model of evaluation with LoD B -A-parts with fasteners (F) and seal plates (S)

5.2 Discussion of systematic relations

In general, certain dependencies can be derived,which influence the evaluation of the assemblabilityof a preliminary design concept more than others (e.g.number of blades on a rotor). In addition, different de-sign features show a direct dependency on the chosenarchitecture of the concept. In order to minimize thenecessary efforts for an evaluation of the assembla-bility, different interfaces within the LPT module suchas flanges or firtrees of blades and discs can be esti-mated on the basis of standard times.Based on the generated 3D CAD models, differentmounting and dismounting evaluations are conducted

Deutscher Luft- und Raumfahrtkongress 2015

5

Rotor Sub-Assembly

PT n = Process time ofBalancing processof Stage n

61 6’ 17”

62 3’ 25”

63 12’ 06”

PT 1

101 6’ 17”

102 3’ 04”

103 12’ 06”

PT 2

141 6’ 17”

142 3’ 04”

143 12’ 06”

PT 3

181 6’ 17”

182 2’ 49”

183 12’ 06”

PT 4

221 6’ 17”

222 2’ 46”

223 12’ 06”

PT 5

LPT Final Assembly

10

9’50

”

20

3’28

”

30

3’05

”

40

2’50

”

50

2’50

”

70

9’25

”

80

3’42

”

90

2’50

”

110

9’25

”

120

3’35

”

130

2’50

”

150

9’25

”

160

3’20

”

170

2’50

”

190

9’25

”

200

2’50

”

210

2’50

”

230

9’25

”

Stage 1

Stage 2

Stage 3

Stage 4

Stage 5

Fig. 8: Assembly Precedence graph with estimated manual assembly times for LoD B

manually, which provide findings for the Evaluation ofAssemblability and the set up of the assembly prece-dence graph. The assembly precedence graph of LoDB of 61% in Fig. 8 displays the different tasks in aconsecutive order that are necessary to assemble theconsidered product concept. In addition to that, dif-ferent tasks are displayed in a sub-assembly section,which can be conducted parallel or detached from thefinal assembly process. These parallel tasks comprisethe task for pre-assembling the rotors of the differentstages. In the final assembly section, the characteristicconsecutive sequence of the different turbine stagesis displayed. Subsequently, the different manual as-sembly tasks, which were defined in the assemblyprecedence graph are analyzed with the MTM methodto estimate the assembly time. The results of theseestimations are also displayed in Fig. 8.In general, any LPT of an axial turbo machinery fol-lows an assembly structure similar to Fig. 8. Although,based on the chosen architecture of the concept (e.g.number of stages, interfaces, etc.), this structure variesin the quantity of the stages whereas the overall struc-

ture of the assembly precedence graph remains thesame.

5.3 Results

The results of two evaluations with different Levels ofDetail are displayed in Fig. 9. The first LoD A with 55%comprises only the A-parts and manual assembly worksteps while the LoD B of 61% also considers B-partssuch as fasteners and seal plates as well as the nec-essary assembly steps complemented with estimatedprocess times for the balancing of the rotors.

The results of the design related weight estimationsand part count for LoD A show a distribution which issimilar to a Pareto distribution. This means that 30%of the parts account for 68% of the weight of the LPTmodule. The estimated assembly time for A-parts islow with 6% compared to similar product in production.The results for LoD B with 61% display only a minorincrease in the estimated weight of the LPT moduleto 70%, whereas the part count increases significantly

Deutscher Luft- und Raumfahrtkongress 2015

6

0 %

20 %

40 %

60 %

80 %

100 %

68 % 70 %

30 %

57 %

6 %

18 %

Detail Design

Weight [kg] Part count [-]AssemblyTime [h]

LoD A = 55%

LoD B = 61%

Fig. 9: Results of two assembly evaluation iterations

to 57% due to the fact that a large number of fasten-ers is necessary to connect the different stages. Theassembly time increases significantly to 18% with theconsideration of B-parts. One of the reasons is theincreased part count as well as the more complex op-erations of installing fasteners in limited space insidethe LPT module. The results for the estimated assem-bly times display the trend, that the large number ofB-parts accounts for a large amount of the assemblytime compared to fewer but heavier A-parts.

6 CONCLUSION AND FUTURE WORK

The method introduced in this work displays a system-atic evaluation of assemblability of LPT modules withlimited knowledge during Preliminary Design basedon a digital model of the concept. Mounting and dis-mounting paths for axial symmetrical as well as unsym-metrical components can be evaluated in a simplifiedfully parametric 3D CAD model. These findings aidthe set up of an assembly precedence graph, whichillustrates the sequence of the tasks to assemble theconsidered concept. This is complemented by an esti-mation of the required manual assembly time utilizingthe MTM method. The results of two evaluation itera-tions display a significant dependency of B-parts suchas fasteners and seal plates on the assembly time,whereas the quantities of A-parts such as blades havelittle influence on the assembly time.In future work, the parametric 3D model will be en-hanced to depict more design characteristics for LPTmodules. Furthermore, the possibility of a rule basedgeneration of the quantities for fasteners, blades, vanesegments, etc. will be evaluated. Additionally, the re-sults of the MTM method will be validated for typicalwork steps of aero engine assembly.

REFERENCES

[1] Basic MTM: Studentische Ausbildung, 2011.Deutsche MTM-Vereinigung e.V.

[2] BOEING: Current Market Oulook: 2013 - 2032.Tech. rep., Boeing Commercial Airplanes, 2013.

[3] BOOTHROYD, G.; DEWHURST, P.; KNIGHT, W.:Product design for manufacture and assembly.CRC Press, 2011.

[4] BRETSCHNEIDER, S.: Knowledge-Based Pre-liminary Design of Aero-Engine Gas-Generators.Ph.D. thesis, University of Stuttgart, 2010.

[5] BULLINGER, H.J.: Systematische Montagepla-nung: Handbuch für die Praxis. Hanser, 1986.

[6] DEAGEL.COM: Turbofan Engines Report be-tween 1996 and 2015, 2015 (accessed June 29,2015).

[7] GAIROLA, A.: Montagegerechtes Konstruieren:Ein Beitrag zur Konstruktionsmethodik. Ph.D.thesis, Technische Hochschule Darmstadt, 1981.

[8] JESCHKE, P.; KURZKE, J.; SCHABER, R.;RIEGLER, C.: Preliminary Gas Turbine Design Us-ing the Multidisciplinary Design System MOPEDS.Journal of Engineering for Gas Turbines andPower, vol. 126 (2004)(2), pp. 258–264.

[9] JINKS, S.; WISEALL, S.: Utilizing dynamic factorysimulation to improve unit cost estimation and aiddesign decisions. Proceeding of the 2010 WinterSimulation Conference, (2010), pp. 1758–1766.

[10] JONES, M.; BRADBROOK, S.; NURNEY, K.: APreliminary Engine Design Process for an Af-fordable Capability. Paper presented at the RTOAVT Symposium on "Reduction of Military Vehi-cle Acquisition Time and Cost through AdvancedModelling and Virtual Simulation", held in Paris,France, and published in RTO-MP-089, (2002).

[11] LOTTER, B.; WIENDAHL, H.P.: Montage in derindustriellen Produktion. Springer Vieweg, 2011.

[12] LYTLE, J.: The Numerical Propulsion SystemSimulation: A Multidisciplinary Design Systemfor Aerospace Vehicles. In: ISABE Conference1999. Florence, Italy, 1999.

[13] LYTLE, J.: Multi-fidelity Simulations of AirBreating Propulsion Systems. In: 42ndAIAA/ASME/SAE/ASEE Joint Propulsion Con-ference & Exhibit. Sacramento, California, 2006.

[14] RED, C.: Neuro-fuzzy comprehensive assembla-bility and assembly sequence evaluation. Com-posite World, vol. 1 (2015)(1), pp. 32–39.

Deutscher Luft- und Raumfahrtkongress 2015

7

[15] ROLLS-ROYCE: The Jet Engine, 5th Edition.Rolls-Royce plc, 2005.

[16] ROLLS-ROYCE: Annual Report 2014. Tech. rep.,Rolls-Royce Holdings plc, 2014.

[17] TONG, M.; HALLIWELL, I.; GHOSN, L.: A Com-puter Code for Gas Turbine Engine Weight andDisk Life Estimation. Proceeding of ASMETURBO EXPO, GT-2002-30500, (2002).

[18] WACK, K.J.; BAER, T.; STRASSBURGER, S.:Grenzen einer digitalen Absicherung des Produk-

tionsanlaufs. Integrationsaspekte der Simulation:Technik, Organisation und Personal - KIT Scien-tific Publishing, (2010), pp. 45–52.

[19] WESTKÄMPER, E.; SPATH, D.; CONSTANTINESCU,C.; LENTES, J.: Digitale Produktion. SpringerVieweg, 2011.

[20] ZHA, X.: Neuro-fuzzy comprehensive assem-blability and assembly sequence evaluation. AIEDAM, vol. 15 (2001)(5), pp. 367–384.

Deutscher Luft- und Raumfahrtkongress 2015

8