Cost Modelling of Composite Components for Helicopter …919302/... · 2016. 4. 13. · Composites...

Cost Modelling of Composite Componentsfor Helicopter Applications

PATRIK IDSTAM

Master Thesis in Lightweight StructuresStockholm, Sweden 2015

Supervisor: Mathilda Karlsson HagnellTRITA: AVE 2015:75, ISSN: 1651-7660

AbstractComposite materials are becoming more popular than ever. The increasing envi-ronmental concerns results in new challenges, where one of the biggest is to reducethe emission of greenhouse gases. Both the aerospace industry and the automotiveindustry work hard to become more environmental friendly. To reduce the emissionsof vehicles such as cars, planes and helicopters reduction of weight is important aslower weight reduces fuel consumptions. To save weight more of the structural andload-bearing components in such vehicles are manufactured out of composites.

To meet the requirements are a lot of new manufacturing techniques developing.To make the new methods competitive with the techniques used today it is importantto make the new methods as cost-efficient as possible.

The purpose of this master thesis is to investigate, describe and analyze themanufacturing cost of composite components manufactured by 3D weaving, whenthe technique is widely used in the industry. The goal is to determine cost drivingparameters and investigate how to make 3D weaving cost competitive. This is donethrough out the design of a basic technical cost model and comparison betweendifferent cases and with other manufacturing methods. The component evaluated isa stringer for a side shell panel of a helicopter. The method used as comparison inthis thesis is a hand lay-up preform consolidated by High pressure RTM (HP-RTM).

A fully automated manufacturing process is chosen for each method in order tomanufacture large annual manufacturing volumes. The cost model is designed usinga bottom-up perspective where each step of the manufacturing process is evaluatedseparately. The lead-time and the cost corresponding to each sub-step is calculatedand passed on to the final sum of all steps. The calculated cost are investmentcost, operator cost, material cost and fixed cost corresponding to plant cost such aselectricity cost.

The cost model shows that the cost is decreasing with increasing manufacturingvolumes and can be divided in to two regions. The first region corresponds to lowermanufacturing volumes and is highly dependent on the investment cost, The secondregion corresponds to higher manufacturing volumes and the cost driving parameterfor this region is the manufacturing time.

The model also shows that it should be possible to manufacture cost competitivestringers out of 3D woven preforms. To make the technique cost competitive it isimportant to manufacture 3D weaving machines with high weaving speed, since thespeed is the factor influencing the total component cost mostly.

i

AcknowledgementsI would like to express my gratitude to my supervisor Mathilda Karlsson Hagnell forthe support, useful comment, remarks and engagement throughout this master the-sis. I also want to thank Associate Professor Malin Åkermo and Associate ProfessorStefan Hallström for accepting me as a master thesis student. Furthermore I wouldlike to thank Fredrik Winberg at Biteam AB for introducing me to the topic of 3Dweaving as well for the support on the way. I also want to thank Frank Weiland atAirbus Helicopters for supplying data of the component used in the thesis. FinallyI would like to thank my family, who have supported me throughout entire process,both by keeping me harmonious and helping me putting pieces together.

ii

ContentsAbstract i

Acknowledgements ii

List of Figures vi

List of Tables vi

I Scope of work 1

1 Background 1

2 Objectives 1

3 Delimitations 23.1 Manufacturing methods . . . . . . . . . . . . . . . . . . . . . . . . . 23.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.3 Component properties . . . . . . . . . . . . . . . . . . . . . . . . . . 2

II Frame of reference 3

4 Technical cost modeling 3

5 Manufacturing 45.1 Preforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

5.1.1 3D woven preforms . . . . . . . . . . . . . . . . . . . . . . . . 4Conventional weaving . . . . . . . . . . . . . . . . . . . . . . . 43D-weaving . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . 6

5.2 Main process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65.2.1 Liquid composite moulding . . . . . . . . . . . . . . . . . . . . 6

Resin Transfer Moulding . . . . . . . . . . . . . . . . . . . . . 65.3 Post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

6 Moulding theory 86.1 Liquid resin flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86.2 Preform permeability . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.3 Resin viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.4 Consolidation theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

6.4.1 Curing of thermosets . . . . . . . . . . . . . . . . . . . . . . . 10

III Analysis 11

7 The component 117.1 Quality requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . 127.2 Redesign for 3D weaving . . . . . . . . . . . . . . . . . . . . . . . . . 13

iii

8 Developed cost model 138.1 Cost model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

8.1.1 Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158.1.2 Installation- and adaption-fees . . . . . . . . . . . . . . . . . . 158.1.3 Labour cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158.1.4 Machine and tooling lifetime . . . . . . . . . . . . . . . . . . . 15

8.2 Manufacturing method . . . . . . . . . . . . . . . . . . . . . . . . . . 158.2.1 Machines and tooling . . . . . . . . . . . . . . . . . . . . . . . 15

Material handling robot . . . . . . . . . . . . . . . . . . . . . 16Weaving machine . . . . . . . . . . . . . . . . . . . . . . . . . 16CNC-cutter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Hydraulic press . . . . . . . . . . . . . . . . . . . . . . . . . . 17HP-RTM-injection equipment . . . . . . . . . . . . . . . . . . 18Hydraulic press mould . . . . . . . . . . . . . . . . . . . . . . 18Milling machine . . . . . . . . . . . . . . . . . . . . . . . . . . 18Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

8.3 Material cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198.4 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

9 3D weaving 199.1 Preforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209.2 RTM station . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219.3 Post-processing station . . . . . . . . . . . . . . . . . . . . . . . . . . 219.4 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

10 HP-RTM 22

11 Adaption 2211.1 Dedication degree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2311.2 Number of plies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2311.3 Curing time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2311.4 Filling time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2311.5 3D weaving machine . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

11.5.1 Weaving speed . . . . . . . . . . . . . . . . . . . . . . . . . . 2411.5.2 Investment cost . . . . . . . . . . . . . . . . . . . . . . . . . . 2411.5.3 Electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2511.5.4 Floor size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

IV Results and discussion 26

12 Best and worst case scenarios 2612.1 Comparative analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 2612.2 Detailed analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

12.2.1 Cost per component . . . . . . . . . . . . . . . . . . . . . . . 27Cycle time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

12.3 Cost driving parameters . . . . . . . . . . . . . . . . . . . . . . . . . 3012.3.1 Investment for 3D weaving machine . . . . . . . . . . . . . . . 3012.3.2 Weaving speed . . . . . . . . . . . . . . . . . . . . . . . . . . 31

iv

13 Sensitivity analysis 3213.1 Number of plies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3213.2 Filling time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3313.3 Curing time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

14 Conclusions 35

15 Future work 36

16 References 37

Appendix A - Machine data 40

Appendix B - Weaving speed for 2D machine 41

Appendix C - Hydraulic press investment costs 42

Appendix D - Data for closed moulds 43

Appendix E - Detailed analysis of HP-RTM 44

v

List of Figures4.1 Division of cost models according to some research . . . . . . . . . . 35.1 The interlacing of 3D woven 3D fabrics [1] . . . . . . . . . . . . . . . 55.2 Dual-directional shedding procedure [2] . . . . . . . . . . . . . . . . . 65.3 Schematic picture of resin transfer moulding (RTM) . . . . . . . . . . 76.1 Time-temperature-transformation (TTT) cure diagram [3] . . . . . . 10

7.1 The side shell of a helicopter, given by Airbus Helicopters [4] . . . . . 117.2 The Z-shaped cross-section of the stringer, with areas for facilitating

drape highlighted in green . . . . . . . . . . . . . . . . . . . . . . . . 127.3 Representation of the changes in reinforcement orientation when drap-

ing the component in Interactive Drape . . . . . . . . . . . . . . . . . 127.4 a) Original cross section b) Redesigned Z-profile . . . . . . . . . . . . 138.1 Main structure of the developed cost model . . . . . . . . . . . . . . 138.2 Conceptual graph for cost per part over the development time. . . . 179.1 Industry layout of the 3D weaving process . . . . . . . . . . . . . . . 209.2 Schematic picture of the preforming cell in the manufacturing process 219.3 Schematic picture of the RTM station in the manufacturing process . 219.4 Schematic picture of the Post-processing station in the manufacturing

process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

12.1 Comparison between 3D weaving and HPRTM . . . . . . . . . . . . . 2612.2 Cost per part in detail for the best case 3D-weaving scenario . . . . . 2812.3 Cost per part in detail for the worst case 3D-weaving scenario . . . . 2812.4 Time for manufacturing one component during best case . . . . . . . 2912.5 Time for manufacturing one component during worst case . . . . . . 3012.6 Cost difference between low and high investment cost for 3D weaving

machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3112.7 Cost difference between low and high weaving speed . . . . . . . . . . 3213.1 HPRTM comparision with different number of plies . . . . . . . . . . 3313.2 Sensitivity analysis of the filling time . . . . . . . . . . . . . . . . . . 3413.3 Sensitivity analysis of the curing time . . . . . . . . . . . . . . . . . . 35E.1 Part cost for the HP-RTM . . . . . . . . . . . . . . . . . . . . . . . . 44E.2 Manufacturing time for the HP-RTM . . . . . . . . . . . . . . . . . . 45

List of Tables8.1 User controlled input for the developed cost model . . . . . . . . . . . 149.1 Material data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2211.1 General input data used in the thesis . . . . . . . . . . . . . . . . . . 2311.2 Assumptions used for 3D weaving machine . . . . . . . . . . . . . . . 24

A1 Machine data used in developed cost model . . . . . . . . . . . . . . 40B1 Data for carbon fabrics woven by 2D machines . . . . . . . . . . . . . 41C1 Hydraulic press data . . . . . . . . . . . . . . . . . . . . . . . . . . . 42D1 Mould data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

vi

PART ISCOPE OF WORK

BackgroundComposite materials are becoming more popular than ever. The increasing envi-ronmental concerns results in new challenges, where one of the biggest is to reducethe emission of greenhouse gases. Both the aerospace industry and the automotiveindustry work hard to become more environmental friendly. To reduce the emissionsof vehicles such as cars, planes and helicopters reduction of weight is important aslower weight reduces fuel consumptions. To save weight more of the structural andload-bearing components in such vehicles are manufactured out of composites.

Composites are materials made out of two different constituent materials withdifferent properties, usually high modulus fibre and a plastic matrix. One benefitwith composites is the possibilities to design the mechanical properties of the ma-terial, by changing the direction of fibers, volume fraction of fibers or the materialof the matrix or fibre. Composites are usually manufactured by stacking layers offibers on top of each other and consolidate the material with matrix. The majordisadvantage with this method are the low strength between the layers out-of-plane.Some applications requires higher out-of-plane properties.

To solve the problem of low out-of-plane properties, a lot of new manufacturingtechniques that adds fibre in the thickness direction has been developed during thelast decades. Some of these techniques are today commercial, such as stitching andbraiding. Techniques such as 3D-weaving and noobing are more immature and ontheir way to be commercial. To be fully commercial and used in all types of industriesit is important to understand every aspect of the new manufacturing methods. Withgood knowledge of the methods it is possible to lower the manufacturing cost as wellas get the methods certified to use for manufacturing of structural components.

To be able to decide which manufacturing method that is suitable for certainapplications it is important to investigate the cost driving parameters of the moreimmature techniques. According to some researches [5,6] over 70% of the manufac-turing cost is being set during the design phase. Making wrong decisions in the de-sign phase can be extremely costly further down the manufacturing process [7],whichmakes it important to be able to make cost estimations during the design phase.This thesis addresses the problem of estimate the manufacturing cost of 3D wovencomposites in the early design phase through out the development of a technical costmodel. The costs is also compared with other manufacturing methods, to identifyhow to make 3D woven composites price competitive.

ObjectivesThe main goal of this master thesis is to investigate, describe, analyse and evalu-ate the manufacturing cost of composites manufactured by 3D-weaving. The cost

1

is evaluated against manufacturing volume and compared to other manufacturingmethods to highlight important and cost driving factors for 3D weaving. This isdone by development of a technical cost model for a manufacturing method contain-ing 3D weaving and compare the results with already existing cost models for othermanufacturing methods.

DelimitationsThe cost model developed in this thesis is aimed to be used during the early con-ceptual design phase when the data known is limited. The model is not completeand can be further developed with more variables active. The focus in this thesis ishigh quality components to be used in helicopter applications.

3.1 Manufacturing methodsThe manufacturing method treated in this thesis is Resin Transfer Moulding (RTM)of 3D woven composite preforms. The method is investigated in this study due tothe repeatability and the possible to manufacture high quality components withhigh production rate. RTM is also the most common manufacturing method usedfor 3D woven preforms.

3.2 MaterialsThe thesis is focused on components for helicopter applications. That leads torequirements of high quality products with great mechanical properties both in longand short series. In aerospace industry carbon fibre/epoxy components are usuallyused to fulfill these requirements. The focus in this thesis will therefore be on highstrength (HS) carbon fibre yarns together with epoxy resin.

3.3 Component propertiesAn important factor for the total cost of producing a component is the mechanicalrequirements of the product. The component evaluated in this thesis is a componentcarrying secondary loads. The main load is transferring of shear forces. 3D weavingare today mostly considered for such parts to avoid the high certifications cost aswell as the development costs.

The component manufactured by 3D woven composite should also be comparedto components made by laminates. To be able to perform such comparison it is im-portant to have knowledge about the relationship between the mechanical propertiesfor laminates and 3D woven composites. The mechanical properties are dependenton the manufacturing method, but will not be fully evaluated in this thesis. Theproperties need to be determined elsewhere and considered together with this modelto fully be able to make decision on manufacturing methods and production costs.

2

PART IIFRAME OF REFERENCE

Technical cost modelingThere are many different approaches and models existing today. It is possible todivide all different cost models into two main branches; quantitative and qualitativemethods. The qualitative methods estimates the cost by expert judgement andonly indicates whether an alternative is better or worse. No absolute values areobtained [8], which makes it a unsuitable method in this project. The quantitativemethods can be further classified into three subgroups; statistical, analogous andgenerative and analytical. Some more recent research adds one more subgroup calledfeature based cost modeling [8, 9]. The subgroups can be seen in the tree-structurein Figure 4.1.

Figure 4.1: Division of cost models according to some research

In statistical modeling large quantities of historical data are used to find casuallinks between costs and product characteristics and determine a parametric function,through statistical models and criteria. A disadvantage with statistical models arethat large amounts of qualitative data is necessary to find the relations [9]. Sincelimited data is available regarding 3D woven composites a pure statistical methodis rejected for this thesis.

Analogous models are based on previous manufacturing of functional and geo-metrical similar components. Since the resulting cost model of this thesis shouldbe as general as possible and data for 3D woven composites are very limited is theanalogous models not interested in this thesis.

Feature based modeling is a method where the design features of a product areused to estimate the production cost. This means that the features, such as cut-outsor ribs are corresponding to a certain cost. The total production cost is the sum ofthe cost of all features [8]. This results in a very general model which is interestingin this thesis. The disadvantage is that all costs can not be connected to certainfeatures of a product which leads to uncertain values.

Generative and analytical models are based on the manufacturing process. Theprocess is decomposed to sub-processes and the cost is calculated for each sub-

3

process. The total cost is the sum over all subprocesses. The advantages of thisapproach is the accuracy since it is possible to depict the actual manufacturingprocesses and make estimations for each sub-group. The disadvantages are thatin-depth knowledge of each sub-process and the overall process is needed [8]. Theanalytical methods often implement parts of the other quantitative sub-groups, suchas feature based methods or statistical methods. The generative models are used asthe structure of the cost modeling while other methods or models are used for eachsub-process. Activity-based-costing (ABC) is one model classified as analytical [10].ABC is used frequently in both aerospace and automobile industry [11, 12] and isthe model to be used in this thesis, since it is possible to construct a very generalmodel with many different variables. In calculating cost for some sub-process willparametric cost modeling be used to make the model as general as possible. Ifpossible statistical methods can also be used to make the model as accurate aspossible.

ManufacturingThere are a lot of different manufacturing methods available for manufacturing ofcomposites. The methods differs very much but it is possible to divide all manufac-turing methods into three main parts: preforming, main process and post-processing.

5.1 PreformingThe preforming process is one very important step in the manufacturing process.The impact on the total manufacturing cost differs depending on the main process,but can in some case be up towards 50%. The process of preforming can vary verymuch, containing everything from cutting fabrics to manufacture 3D preforms. Inthis thesis 3D woven preforms are in focus, and the preforming process correspondingto that will be outlined in the next subsection.

5.1.1 3D woven preformsWeaving has been used in over five millennia and are today a very common manu-facturing process in the industry. Weaving produces the majority of the single-layerfabrics used as reinforcement material for composite components today [13]. Overthe years a lot of new methods has been evolved from the conventional weavingto be able to manufacture three dimensional woven fabrics. Some of this methodhas been called 3D weaving even though the lack of some weaving characteristicoperations [1]. To be able to understand the main principles of 3D weaving it isimportant to look at the 2D weaving process.

Conventional weaving

The 2D weaving process has been around for thousands of years and is a high-speedeconomical process [14]. The main principles are that two set of yarns, called warpand weft, are interlaced as a fabric. That is done by a shedding operation and bya picking operation. The shedding operation displaces the warp yarn, which runsin the 0◦-direction, in a crossed manner to produce a shed in the direction of the

4

fabric-width. The picking operation is inserting wefts into the sheds which producesthe interlacing of two orthogonal set of yarns. The shedding and picking operationsare crucial to describe weaving methods [2].

It is also possible to use multiple warp-layers to produce 3D fabrics. The multiplewarp-layers are arranged in layers and interlaced by wefts in fabric-width direction.Since this method only interlaces two orthogonal sets of yarns is not technically 3Dweaving. So it is possible to talk about 2D woven 3D fabrics and 3D woven 3Dfabrics [15].

3D-weaving

As said previous it is possible to produce 3D fabrics with conventional weaving.The difference between 2D woven 3D fabrics and 3D woven 3D fabrics are that the3D technic interlaces three orthogonal set of yarns.There are three requirements toproduce a 3D woven 3D fabric [2] and the fully interlaced fabric can be seen inFigure 5.1:

• A grid-like multiple layer-warp (Z).

• A shedding operation able to produce sheds in two directions.

• Two orthogonal sets of wefts (X and Y).

Figure 5.1: The interlacing of 3D woven 3D fabrics [1]

To fulfill the second requirement, producing sheds in two directions, a sheddingoperation that are called Dual-directional shedding is used. The operation producesthe sheds row-wise or column-wise in the grid-like warp, the procedure can be seen inFigure 5.2. These sheds are produced in a successive manner but not simultaneouslyduring a weaving cycle [1].

5

Figure 5.2: Dual-directional shedding procedure [2]

Mechanical Properties

Some of the advantages with 3D woven preforms are the possibility to customize theload-bearing capacity and higher delimitation resistance [16]. The manufacturingmethod is also much more precise than other manufacturing methods such as handlay-up. Since the preforming is automized and no manual labour is needed is therepeatability high which assure consistent quality of the manufactured preform [17].

5.2 Main processThe manufacturing of 3D preform is, as said previous, the first step in the processof manufacturing 3D fibre reinforced composite materials. The fibre must also beconsolidated with polymer resin to form the finished composite component. Thiscan be done through a lot of different methods.

Due to the complex fibre structure in the preform a lot of traditional consolida-tion methods are not appropriate to use since they might cause distortions in thepreform architecture. Such methods are hand impregnation, which will also leavevoids of air inside the material, and pultrusion. The only way to consolidate highquality components from the 3D woven preforms are liquid composite moulding [13].

5.2.1 Liquid composite mouldingLiquid composite moulding is a term that contains a range of processes.The mainfunction is that liquid resin is transferred into a closed mold, containing fibre, inwhich it is cross-linking before being demoulded. Resin transfer moulding (RTM) isone process corresponding to liquid composite moulding and is the one used in thisthesis [18].

Resin Transfer Moulding

Resin Transfer Moulding (RTM) is the most common liquid moulding method. Itis widely used since it is possible to produce large, complex and high-quality com-ponents with great surfaces. With this method it is possible to achieve up to 60% fibre volume fraction in the components. The method is also very flexible which

6

makes it possible to include inserts and fasteners with the reinforcement before con-solidation [18]. It is possible to speed up the process by injecting the resin withhigher pressure which fills up the mould quicker, this method is called High pressureRTM (HP-RTM) and gives the advantage of being able to produce larger quantitieswith the same numbers of equipment. Larger quantities will result in lower cost perpart [19].

Figure 5.3: Schematic picture of resin transfer moulding (RTM)

In Figure 5.3 a schematic picture of the process is seen. The dry reinforcementpreform is placed in the cavity of a closed mould. Before the reinforcement is placedin the cavity the mould is prepared with release agent and often gelcoat. Thatis done to avoid cracks when demoulding as well as produce great surfaces. Theresin is injected to impregnate the reinforcement, while the mould is subjected to apressure and heated to proper temperature [20]. The temperature will initiate thecuring process. The component can be demoulded either after fully being cured orafter the gel point has been reached. If the component are demoulded when thegel point is reached it is necessary to further cure the component within a separateoven. When the component is demoulded it is ready for post-processing [18].

5.3 Post-processingJust as preforming post-processing is highly dependent on the complexity of the com-ponent as well as which main process is used. Post-processing of composites includeoperations like machining, joining and surface treatment. The post-processing stepsfor composites are more or less the same steps used when manufacturing metal com-ponents. It can be machining (drill holes and trim edges), joining parts togetheror surface treatment. Even if the same steps are present the differences betweencomposites and metals are large regarding post-processing. Composites breaks eas-ier than metals and it is harder to predict the behavior of composites due to thegreater number of variable dictating the response. The fibre orientation at contact,

7

matrix hardness and the heat sensitivity are just a few of the variables taken intoaccount. The machining processes need to be specially developed for composites,water jet cutting or rotary machining are often used [21].

The closed mould of the RTM process leads to the possibility to produce twogood surfaces, the quality of the mould as well as the gelcoat are dictating thequality of the surfaces. With good surfaces the need for polishing and other surfacetreatments is reduced. It is also possible to carry out the painting process while thecomponent are curing inside the mould, which is called in-mould decoration (IMD),where the paint is working as gelcoat [22].

Moulding theoryTo develop a cost model it is important to understand the process modeled andthe cost driving parameters. By understanding the physics that govern the man-ufacturing methods it is simpler to evaluate which parameters that are influencingthe cost mostly. In this section is the moulding theory described. The cost drivingparameters in this section are mostly parameters influencing the cycle time, whichin the end will influence the final cost.

6.1 Liquid resin flowA key issue when manufacturing composite through RTM is the resin infiltration ofthe fibrous preform. The resin must impregnate the entire preform before the geltime otherwise voids of trapped air and dry spots will be present in the componentand the component will be of very low quality. To make sure the mould is filledbefore gelation it is important to be able to model the mould filling process toestimate the mould filling time tfill [23]. The mould filling time depends on thevolume of the component and the resin flow as seen in equation 6.1

tfill = V

Q(6.1)

where V is the component volume and Q is the volumetric flow rate. For a RTMprocess the flow of the resin through the preform is described by Darcy’s law whichcan be seen in equation 6.2

u = −Kµ∇P (6.2)

where u is the volume average velocity, µ is the viscosity of the resin, K is thepermeability and ∇P is the pressure gradient. The pressure can be controlled butboth the permeability and the viscosity are hard to determine during the manufac-turing process. The permeability contains all interaction between resin and fibersand the viscosity varies over the whole process. To model the process accurately itis necessary to have good models of resin viscosity and the permeability [23]. Thetwo parameters will be discussed further in the sections below.

8

6.2 Preform permeabilityThe permeability of a material describes the ability of a porous material to allowfluids to flow through it. The permeability is dependent by a lot of factors suchas the fiber arrangement, the volume fraction and will be varying in the differentdirections of the material. Since the permeability is dependent on a lot of differentfactors it is hard to analytically determine. A number of different models exists suchas the often used Kozeny Carman model [24,25] which can be seen in equation 6.3

K =r2fibre(1− Vf )3

4kV 2f

(6.3)

in which rfibre is the radius of the fibers, Vf is the fibre volume fraction and k isKozeny constant. The constant is usually determined experimentally for every fibrearchitecture.

Most models are developed to predict the permeability of laminate composites,where the in-plane properties are the most important. Due to fibre architecturewith interlaced fibre in three directions the prediction of permeability in 3D wovenpreforms are harder to estimate with the existing models [23]. New models arebeing developed by researchers [26] and estimations of the permeability of 3D wovenpreform is possible by CFD analysis of a unit cell of the 3D woven composite.

6.3 Resin viscosityThe viscosity of a fluid is a measurement of the internal friction in a flowing fluid.During the RTM process will the resin viscosity will vary, since it is dependent onthe degree of cure as well as the temperature [25]. The resin viscosity can often bedescribed by equation 6.4.

µ = µ∞e(U/RT+κα) (6.4)

In equation 6.4 are µ∞ and κ constants, U the resin activation energy, T theabsolute temperature, R the universal gas constant and α the degree of cure. Todetermine the constants and the resin activation energy a lot of experiments areneeded. In most cases are these experiments carried out by the resin-manufacturingcompany and supplied to the composite manufacturer.

The temperature and the degree of cure will vary during the manufacturingprocess and to estimate the resin flow as well as the filling time it is importantcontrol the temperature and also to model the degree of cure.

6.4 Consolidation theoryThere are two types of resin, thermosets and thermoplastics. Thermosets consoli-dates through curing and creation of cross-links. Thermoplastic solidifies throughcooling. Thermoplastics can also be reused through melting. Since this thesis willfocus on epoxy as resin the consolidation of thermoplastics will not be discussedfurther. The curing of thermosets will be discussed in the following section.

9

6.4.1 Curing of thermosetsDuring consolidation of a thermoset the resin is transfered from liquid state to asolid state, through gelation. The solidification occurs through cross-linking. Thetime for the composite component to cure depends on material system properties,temperature as well as desired degree of cure. It is possible to speed up the curingtime by adding extra curing agents, which affects the resin reactivity. By optimizingthe parameters and add extra curing agents it is possible to reduce the time to acouple of minutes. In Figure 6.1 the different phases during the curing of a thermosetresin and the influence of the cure temperature, Tcure can be seen [27].

Figure 6.1: Time-temperature-transformation (TTT) cure diagram [3]

In Figure 6.1 Tg0 is the glass transition temperature for the reactions, Tg thetemperature at which gelation and vitrification occurs at the same time and Tg∞

the glass transition temperature for the fully cured system. To obtain a fully curedcomponent the cure temperature must reach Tg∞ , since vitrification stops the curing.If the temperature is higher than Tg∞ for a long time the resin will degradate. Asseen it is very important to be able to control the curing temperature to reachthe desired degree of cure as fast as possible. The total curing time, tcure, can bedescribed by

tcure = tgel + tpost cure (6.5)

where tgel is the time it takes for the resin to reach the semi solid state and tpost cureis the time after gelation necessary to reach the wanted degree of cure. There aresome different methods to model the curing cycle of thermoset resins. Usually thecuring cycle is determined by data supplied by the resin-manufacturer.

10

PART IIIANALYSIS

The componentThe component used for evaluation in this thesis is a curved stringer. The cross-section of the stringer has a S-shape and the stringers are parts of the side-shell ofa helicopter. A CAD-model of the side shell panel can be seen in Figure 7.1 wherethe stringers are highlighted in green and the blue part is a support frame. Thedimensions of the panel are 2 and 1.5 meters, the panel is also curved. The exactdimensions and curvature depends on the helicopter model. On the real part thereare stringers every 150 mm over the height of the panel and support frames every300mm over the length to stiffening the whole shell in a grid structure. The sideshell is mainly subjected to shear and does not carry primary loads [4].

Figure 7.1: The side shell of a helicopter, given by Airbus Helicopters [4]

The S-shaped cross-section can be seen in Figure 7.2 where a change of the cross-section at the end of the stringer also is displayed. The stringer preforming processtoday is hand lay-up. To facilitate the draping of the component the corners ofthe S-profiled are curved, these areas are highlighted in green in Figure 7.2. Thecomponent is also curved over the length.

11

Figure 7.2: The Z-shaped cross-section of the stringer, with areas for facilitatingdrape highlighted in green

7.1 Quality requirementsTo investigate the quality requirements of the component a draping simulationwas performed. The simulation was performed with the free software InteractiveDrape [28]. Three draping strategies, with different starting points, were used. Thestrategy with smallest changes in reinforcement orientation is shown in Figure 7.3.The maximum allowed deformation were set to 25◦. As seen below the maximumdeformation obtained varies between 17◦and 22◦but most part of the stringer hasno or very small deformations.

Figure 7.3: Representation of the changes in reinforcement orientation when drapingthe component in Interactive Drape

12

7.2 Redesign for 3D weavingDue to the big changes in reinforcement orientation around the changed cross-sectionat the end of the stringer it is logical to assume small loads at that point. Varyingcross sections make the weaving more complex, to fully take advantage of the 3Dweaving technology a small redesign of the stringer is necessary.

It is also not necessary to include the curved corners of the cross-section [4],highlighted in Figure 7.2, since weaving makes it possible to produce 90◦cornerswhich can not be obtained with draping. The new stringer will have a constant Zshaped cross-section and still be curved over the length. The original cross sectionand redesigned cross section can be seen in Figure 7.4, where the dimensions of thecross sections are indicated.

Figure 7.4: a) Original cross section b) Redesigned Z-profile

Developed cost modelThe cost model in this thesis follows the tree structure presented in [29]. Thestructure is illustrated in Figure 8.1 and is used to facilitate comparison of thedifferent manufacturing methods. The model contains multiple function files andgets user based input and geometry input based on the CAD file. The functionsfiles will be described further in the following section.

Figure 8.1: Main structure of the developed cost model

13

8.1 Cost modelThe master function named Cost model is governed by two types of input, inputprovided by the user and input based on the CAD-file of the component. The inputvariables provided by the user can be seen in Table 8.1. The variable correspondingto the annual production volume, n, can be a vector to make it possible to evaluatethe cost against the manufacturing volume and plot the results.

Table 8.1: User controlled input for the developed cost model

Input variable Description Extra notes

M Manufacturing method The function name of themanufacturing method

m Material system The function name of thematerial system

n Yearly manufacturing volume Input usually a vectornr_workdays Workdays per year Machine and tooling main-

tenance cost is consideredby decreasing the totalnumber of workdays peryear

nr_workhours Workhorse per dayd Dedication degree [%] Percentage of total

processing time attended toby manual operators.

margin Installation and adaption-fees Extra percentage to coverfor machine and toolinginstallation-and adaption fees. Defaultvalue is set to 0.2

salary Manual workcost per hour [AC/h]kvm_cost Cost per square meter [AC/m2] Rent and maintenance costkWh_cost Electricity cost [AC/kWh]fill_time Mould filling time [min]press_cure_time Component time in mould [min] Gel time if post-curing is

used, else total cure time.post_cure_time Cure time outside of mouldweave_time Weaving time [min] The time for the machine

to weave the component. Ifthe value is set to None adefault weaving velocitywill be assumed.

r Fibre volume fraction [%]

The output from the master function contains both graphical and structuraloutput. The graphical output are plots of lead-time per station and cost per partas a function of annual manufacturing volume. The following sections will describehow important parameters, such as maintenance and labour costs, are taken into

14

account in this cost model.

8.1.1 MaintenanceMaintenance of machines and tooling is necessary to obtain a effective productioncycle. In the developed cost model maintenance is taken into account in the in-put of workdays per year as well as work hours per day. By decreasing these themaintenance needed is taken into account. A three-shift day, 24 hours per day and200 days per years of production accounts for the adequate maintenance in manycases [29].

8.1.2 Installation- and adaption-feesInstallation and adaption-fees are usually cost not covered by the price of the ma-chines, even the interest rate for the investment is needed to be taken into account.To take these extra costs in to account in the developed cost model a extra percent-age margin is added to the total final cost. The default value is set to 20%.

8.1.3 Labour costEven highly automated process such as the manufacturing methods evaluated in thisthesis need manual labour to assure quality and safety. The labour cost is calculatedwith Equation 8.1

Costlabour = Cl · d · ttotal (8.1)

Where Cl is the hourly labour cost, d is the dedication degree listed in Table8.1 and ttotal is the total production time given in hours. The dedication degree is apercentage of the degree of manual labour needed. The hourly labour cost is basedon workers employed in Sweden [30]

8.1.4 Machine and tooling lifetimeThe lifetime of a machine or tool is set in the machine data rather than governed bythe user input. This is to make it possible to decide separate lifetimes for differentmachines as well as decrease the number of user governed inputs. In this model thedeprecitation considers a zero scrap value.

8.2 Manufacturing methodThis function controls the machine park and the tools needed for a certain manufac-turing method. The manufacturing methods evaluated in this thesis are premade.If other manufacturing methods are of interest it is possible for the user to createany manufacturing method as long as the machines and tools needed are defined.

8.2.1 Machines and toolingThe machines that are defined in the program are: binder dispersal robot, CNC-cutter, conveyor oven (260◦capacity), 3D weaving machine, HP-RTM injection equip-

15

ment, hydraulic draping press, hydraulic press, milling machine and material han-dling robot. Tooling possible to add are drape mould (2 MPa pressure capacity) andpress mould (20 MPa pressure capacity). There are also possibilities to add storage,both room-temperature and freezer storage are predefined.

Each machine and tool contains five different calculation categories: lead-time,investment, floor, electricity and tooling. The output from the functions correspondsto the different categories. In the lead-time category the total time and the time percomponent are calculated. In similar way are the total cost and the cost for eachcomponent returned from the investment and tooling categories. Depending on ifthe function is a machine or a tool the tooling output or the investment output isset to zero. The calculations for the floor cost and the electricity is repeated in eachfunction. The electricity calculations returns both the total fixed cost, calculated asseen in equation 8.2, and the total variable cost, calculated as seen in equation 8.3.

Eltotal,fix = Fuse · nmachines1200 · 70000 (8.2)

Equation 8.2 is based on the annual cost of AC70000 for a 1200 Ampere fuse [31].In that equation is Fuse the necessery electrical input in Ampere and nmachines thenumber of the particular machine.

Eltotal,variable = kWh · AC/kWh (8.3)

The floor space cost is calculated using a cost per square meter that includes rentaland maintenance [29, 32]. The machines and tools used in this thesis are describedin the following subsections.

Material handling robot

The material handling robots are robots transporting the preform as well as thecured component. Reference data is taken from IRB 2400 from ABB [33], see Ap-pendix A. The robot need to be able to place the preform into the mould and todemould the cured component. The diversity of task leads to differences in cycletime, in order to take that into account is the time supplied as input data for eachrobot in the process. The number of robots is also required to be known since therobots serves other machines, which is the larger bottle neck of the process.

Weaving machine

The 3D weaving technique is developing and there are today no commercially avail-able 3D weaving machines, which makes it hard to determine the data to be usedin the modeling. Due to that the data in this thesis will, for this type of machine,be assumptions based on data for conventional 2D weaving machines as well as dis-cussion with the 3D weaving industry [17]. In Figure 8.2 a conceptual curve overthe development of manufacturing techniques regarding cost per component anddevelopment time is shown. The thesis will aim to model the performance of a 3Dweaving machine when the technique is widely used in the industry which is symbol-ized in Figure 8.2 with a red cross. The development time until that point dependson the capital invested in development of the manufacturing technique. Referencedata for conventional weaving machines is from Lindauer DORNIER GmbH [34,35],see Appendix B. The assumptions made are discussed further in section 11.5.

16

Figure 8.2: Conceptual graph for cost per part over the development time.

CNC-cutter

To cut the woven preform in to the desired length a CNC-cutting machine is used.Reference data for the cutter is from Haas Automation Inc [36], see Appendix A.The cutter has a maximum cutting speed of 0.21 m/s but the actual speed dependsof different parameters, such as thickness and stiffness of the material. A estimationof the cutting speed, vcut, used in this thesis is 0.16 m/s. The cutting time for onecomponent is then calculated with equation 8.4 where P is the perimeter of thecross-section.

tcut = P

vcut(8.4)

Hydraulic press

The hydraulic press is in this manufacturing method used to keep the mould halvesshut during injection of resin. Presses from Omera S.r.l are used for reference values,see Appendix C for further information. The hydraulic press used in the model foreach manufacturing method is chosen according to the necessary moulding pressure,which is supplied by the user, when designing their manufacturing method withinthe cost model. When the hydraulic press is chosen the function proceed to calculatethe five cost categories.

The necessary moulding pressure for a manufacturing method requires a certainclamping pressure to keep the mould halves shut during the injection process. Forhydraulic presses the required clamping pressure is on of the governing cost factors.For calculation of the investment it is important to know the required clampingforce. The force cannot be calculated exactly but can be estimated with equation8.5

Fclamp = Ap · P (8.5)where Ap is the projected area and P is the moulding pressure, in Pascal. Thecalculated clamping force is then converted to american ton-force, the unit mostcommonly used for sizing of hydraulic presses, and a safety factor of 1.25 [37] is alsoadded to produce the necessary clamping force. This estimation is not true if thecomponent contains narrow cavities that are more than 4 cm deep [38]. If this is

17

the case higher clamping force is needed to force the material into the deep regions.In this thesis equation 8.5 is used.

The lead time for the hydraulic press is calculated through summation of the timefor the several stages the press is used. The different stages are loading of preform,closing of press, press, opening of press and unloading. The loading time, tload,and the unloading time, tunload, depends on the material handling robot serving thepress, these times are know from previous knowledge or calculations. The openingtime and the closing time, tclose and topen, is calculated through the possible pressram velocity and the daylight opening as seen in equation 8.6.

tclose = topen = daylight opening

vpress(8.6)

The pressing time contains the filling time and the curing time. Both these param-eters are user defined input, see Table 8.1. Regarding the electricity category is thepower consumption corresponding to the heating of the press surfaces neglected.

HP-RTM-injection equipment

The injection of resin in to the mould under high pressure is done by high-pressureinjection equipment mounted on the hydraulic press. Reference data for this equip-ment is taken from KraussMaffei Technologies GmbH [29], see Appendix A. Themachine is considered to have zero lead-time, since the filling time is attributed tohydraulic press. The needed number of injection machines are determined by thenumber of presses in the process line.

Hydraulic press mould

The mould used in the HP-RTM process is a steel mould with 20 MPa-pressurecapacity. Reference data for the mould is presented in [29], see appendix D forfurther information. The mould is chosen according to the size of the component.The mould is not considered to have lead-time. The component time in mould isinstead attributed to the press, in which the mould is mounted.

Milling machine

For the post-processing of the manufacturing a milling machine is used. The machinecan trim flash and do similar secondary machining operation. The reference dataused in this model is supplied by the industry [29], see Appendix A. Due to thedifference tasks the milling machine is able to perform and to the different needscorresponding to different manufacturing method and component is the necessarymachine-time user supplied when designing the manufacturing method within thecost model.

Storage

To keep flow in the process cycle it is necessary with intermediate waiting steps toaccount for the different lead-times for each step of the process. These waiting stepsare represented by storages. Reference data is supplied by the industry [29], seeAppendix A.

Storage time at every waiting step vary and depends on several factors such asthe process throughput. To allow different storage times in each step is the storage

18

time user supplied when designing the manufacturing method withn the cost model.The storage in this thesis has no particular climate-control, but simply kept to theclimate of the rest of the factory. Due to this the electricity consumption of thestorage is considered to be a part of the floor cost. The size of the storage is set to100 m2 to allow for plenty of storage space as well as material handling spaces.

8.3 Material costThe Material cost function govern the costs for the material used in the manufac-turing process and is provided with material data from the sub function Materialdata. The scrap of material during the manufacturing process is taken into accountby adding extra percentage of the costs. For material systems that contains dryfibre and resin are two different scrap values considered, one representing the scrapof fibre and one representing the wasted resin.

Since this thesis aims to model the future of 3D woven composites it is hard toestimate the price of carbon fibre. The price is predicted to decrease during thecoming years. Researchers of MAI Carbon Cluster Management GmbH is makingprogress towards reduction of the price for carbon fibre with up to 90 % [39] but it ishard to predict the actual price development. Since the prediction is very uncertainthe benchmark case will be with the price of carbon fibre today. A sensitivity analysisof the results regarding the material price will also be performed to investigate theinfluence of the price uncertainty.

8.4 GeometryThe geometry function reads a CAD-file of the component to evaluate in the thesis.Data for the component such as area, perimeter and volume are provided to themaster function and is used as input for the other sub functions.

3D weavingThe flow chart in Figure 9.1 displays the high-degree automated industry layout ofthe 3D weaving manufacturing process [11, 40]. The different shapes correspondsto different types of stations. Triangles corresponds to storage spaces, ellipses tomaterial handling robots and squares corresponds to machines.

19

Figure 9.1: Industry layout of the 3D weaving process

9.1 PreformingThe preforming section in the manufacturing method is schematically described inFigure 9.2. The section contains a 3D weaving machine (1), a CNC cutter (2)and material handling robots symbolized by the red octagons. The yellow trianglessymbolizes storage space. Due to the length of the component and the deformabilityof the preform two material handling robots are needed. Continuos weaving isassumed, where the CNC cutter cuts the preform in to the required lengths. Thecomponent is woven straight and curved to the right shape in the mould at the RTMstation. Depending on the manufacturing rate it is possible to store the preforms instorage before being sent to the mould.

20

Figure 9.2: Schematic picture of the preforming cell in the manufacturing process

9.2 RTM stationThe RTM station of the manufacturing process is schematically described in Figure9.3. The station contains a mould mounted in a hydraulic press (3) and injectionequipment (4). As mentioned in the previous section the preform is curved in tothe desired shape in the mould by the two material robots. The mould is then pressshut as the resin is injected by the injection equipment. After the mould is filled themould is kept shut during the whole curing process. After the curing is completedthe mould is opened and the component is moved in to storage.

Figure 9.3: Schematic picture of the RTM station in the manufacturing process

9.3 Post-processing stationThe post-processing station is simple, it contains of a milling-machine (5) and mate-rial handling robots serving it, see Figure 9.4. The milling-machine trim flash beforethe completed component is moved to storage. Due to the big size of the millingmachine two material handling robots are needed.

21

Figure 9.4: Schematic picture of the Post-processing station in the manufacturingprocess

9.4 MaterialThe material system used in this manufacturing process is carbon fibre yarn andepoxy resin. The material data can be seen in Table 9.1.

Table 9.1: Material data

Volume fibre fraction 55[%]Fibre density 1.8[kg/m3]Resin density 1.2[kg/m3]Fibre cost per kg 20[AC/kg]Resin cost per kg 5[AC/kg]Resin scrap 2[%]Fibre scrap 2[%]

HP-RTMThe hand lay-up and high pressure resin transfer moulding technique (HP-RTM)that is used as comparison in this thesis is described and can be read about in [29].The method has been updated to include the geometry function, which added thepossibility to obtain data from CAD-files.

AdaptionThe cost model in this thesis is used to evaluate different scenarios of productionthrough 3D weaving and also compare 3D weaving with High pressure RTM. Thegeneral data that is constant for all the different scenarios and methods can be seenin Table 11.1. Inputs depending on the different methods and scenarios are discussedin the following sections.

22

Table 11.1: General input data used in the thesis

Input variable Values

n from 5 000 up to 200 000nr_workdays 200nr_workhours 24 [hours/day]

margin 20 [%]salary 27 [AC/hour]

kWh_cost 0.081 [AC/kWh]kvm_cost 137 [(AC/m2)/year]

11.1 Dedication degreeThe machinery in the evaluated manufacturing methods are expensive and the qual-ity of the components is crucial in the aerospace industry. To be able to assurequality and safety is the manufacturing methods often watched at all times. Thatleads to a dedication degree that is 100%, and that assumptions will be used for allmanufacturing methods evaluated in the thesis

11.2 Number of pliesComponent manufactured by conventional manufacturing methods are as describedprevious manufacturing by stacking plies on top of each other. The number of pliesneeded is a factor influencing the cycle time since each ply needs to be treatedseparately.

To be able to evaluate different manufacturing methods against each other it isimportant to compare methods with similar level of automation. The 3D weavingmethod in this thesis i highly automated and designed for high annual productionvolumes. To design the HPRTM process in similar way it is assumed that thematerial system used is a non-crimp fabric (NCF) with epoxy resin. A non-crimpfabric is a fabric with a number of plies stitched together to form one thick fabric.In this thesis it is assumed that this fabric have the necessary thickness to producethe evaluated component, which result in the number of plies needed to handle inthe HPRTM manufacturing method can be set to 1.

11.3 Curing timeThe same epoxy resin is used in all manufacturing methods evaluated in this thesis.The curing time of epoxy can vary from few minutes up to 3-4 hours [41]. In thiswork a curing time of 10 minutes for every manufacturing method is evaluated.

11.4 Filling timeThe filling time is one user supplied input variable needed to define. The fillingtime is dependent on the preform permeability, which varies with different fibrearrangements. Preforms manufactured by 3D weaving have more complex fibre

23

arrangement than preforms manufactured by hand-lay up. It is possible to reducethe filling time by adding filling inlets in the mould. In this thesis the benchmarkscenarios will be evaluated with a filling time set to 5 minutes for all manufacturingmethods.

11.5 3D weaving machineThe data needed in the thesis are weaving speed, investment cost, necessary fuse,power consumption and the size of the machine. Due to the uncertainty and toinvestigate the influences on the cost per part a span will be assumed for eachcategory. Some values will be based on commercial machines used in the industrytoday and some values are set to create spans wide enough to identify the influences.All values used in the model can be seen in Table 11.2. Each category is discussedfurther in the following subsections.

Table 11.2: Assumptions used for 3D weaving machine

Assumption Values

Weaving speed from 0.000085 to 0.0017 [kg/s]Investment cost from 210 000 to 500 000 [AC]

Fuse 84 [A]Power consumption 32 [kW ]

Floor size from 10 to 20 [m2]

11.5.1 Weaving speedA usual weaving speed for a conventional 2D weaving machine is about 30 m2/hour[34], with an areal density around 200 g/m2 a weaving speed of 0.0017 kg/s isobtained. Due to the more complex shedding and picking operations in 3D weavingand the maturity of conventional weaving it is reasonable to believe that it is notpossible to obtain 3D weaving machines that are faster than conventional weavingmachine. So 0.0017 kg/s will be the maximum speed investigated for 3D weavingmachines in this thesis.

The minimum speed is harder to estimate due to the lack of data of the per-formance from the 3D weaving machines that exist today, but the actual value ofthe minimum weaving speed is not that important since the focus in this thesis isto highlight important factors of 3D weaving and investigate their influence on thecost. To do that a wide span of the weaving speeds are important. The minimumvalue is set to a twentieth of the maximum speed, to provide a wide span.

11.5.2 Investment costThe investment cost for a conventional 2D weaving machine specially designed forweaving of carbon fibre is £150 000, which is converted to approximately 210 000AC[42]. As in previous section it is reasonable to believe that more complex sheddingand picking operations and the maturity of 2D weaving will result in 3D weavingmachines that are more expensive than conventional 2D weaving machines. The

24

cost of a 2D weaving machine will be used as the minimum investment cost for a3D weaving machine in this cost model.

It is not possible to set a specific value to the maximum investment cost. Thereare no reference data to base the assumption on, the maximum value in this modelwill be set to 500 000 AC. That leads to a wide span which should make it possibleto see the influence of the investment cost on the final cost per part.

11.5.3 ElectricityDue to the low price per kilowatt hour it is reasonable to assume that the powerconsumption of the weaving machine will influence the total cost per part very little.Since the contribution to the total cost is assumed to be small is no span createdfor the electricity costs. To be conservative values corresponding to machines withhigh power consumption, such as ovens, are used for the 3D weaving machine. Thenecessary fuse is assumed to 84 A and the power consumption is set to 32 Kw

11.5.4 Floor sizeThe size of the machine is also assumed to contribute very little to the total costper part. The sizes of conventional 2D weaving machines will be used even for the3D weaving machine. Depending on the width of the fabric produced in the 2Dmachines the floorspace needed is varied between 10 to 20 square meter [35].

25

PART IVRESULTS AND DISCUSSION

Best and worst case scenariosThe data discussed in the previous chapter is used to form a best case scenario anda worst case scenario. The scenarios are used to create an upper and lower limit onthe possible cost per part for composite components manufactured out of 3D wovenpreforms.

12.1 Comparative analysisThe best and worst case scenarios are compared to components manufactured byHPRTM over the annual production volume in Figure 12.1. The cost per finishedcomponent decreases with the annual manufacturing volume in all three cases andthe curves can be divided in to two regions. The first region is when the cost decreaselogarithmic and the second region when the cost is almost stable. The cost in thefirst region is highly volume dependent and is decreasing rapidly with increasingmanufacturing volume. The reoccurring peaks in the three curves corresponds toreinvestment points. Reinvestment points are points when the machine park needsto be upgraded to be able to produce the desired annual production volume.

0 50000 100000 150000 200000Annual manufacturing volume [Components/year]

0

20

40

60

80

100

120

Cost per finished component [€]

3D weaving, Best case

3D weaving, Worst case

HPRTM

Figure 12.1: Comparison between 3D weaving and HPRTM

26

As seen in Figure 12.1 is 3D weaving more expensive than HPRTM. The dif-ference in cost between the best case scenario and HPRTM is however very small.It should also be noted that the manufacturing method influences the mechanicalproperties and to fully compare different manufacturing methods against each otherit is important to investigate the mechanical properties, which is not done in thisthesis.

The reinvestment points in the volume dependent part have a large impact onthe part cost, while in the stable part the impact is very low. It is then reasonableto draw the conclusion that in the volume dependent part the total investment costis a major contribution to the total cost per part and in the stable region there areother cost categories that influences the cost more, a direct result of the machinedepreciation rate. To investigate the influences of the different cost categories furtherthe costs are presented and analyzed more in detail in the following sections.

12.2 Detailed analysisIn this section the different cost categories contribution to the total cost per part forthe 3D weaving cases is investigated and discussed. The cost is analyzed at threedifferent annual manufacturing volumes and the cycle time is also analyzed. Themanufacturing volumes evaluated are 5000, 50000 and 200000 components per year.5000 components per year is in the volume dependent region, 50000 is close to thecrossing of the two regions and 200000 is in the stable region. The detailed cost forthe HP-RTM case can be seen in Appendix E.

12.2.1 Cost per componentIn Figure 12.2 and 12.3 the cost for the best and worst case 3D-weaving scenariopresented. The electricity cost contributes very little to the total cost which corre-sponds to the assumptions used in section 11.5.3. Even the material and materialwaste cost contributes very little to the total cost for both cases.

27

5000 50000 2000000

20

40

60

80

100

120

[Euro

]

Z-stringer

FloorCost

InvestCost

ToolingCost

OperatorCost

VarElCost

FixElCost

MtrlCost

MtrlWasteCost

Figure 12.2: Cost per part in detail for the best case 3D-weaving scenario

5000 50000 2000000

20

40

60

80

100

120

140

[Euro

]

Z-stringer

FloorCost

InvestCost

ToolingCost

OperatorCost

VarElCost

FixElCost

MtrlCost

MtrlWasteCost

Figure 12.3: Cost per part in detail for the worst case 3D-weaving scenario

The investment cost is for low manufacturing volume, the volume dependentregion, the cost category contributing mostly, but even tooling and floor cost influ-

28

ences majorly. For the worst case is even the operator cost a category that majorlycontributes to the total cost per part. The big contribution from the investmentcost corresponds well to the conclusions drawn in the previous section. To minimizethe contribution from the investment,tooling and floor costs it is important to usethe maximum capacity of the machines and tools, since the investment cost andfloor size for each machine and tool are independent of the manufacturing volume.Unused capacity will only increase the contribution to the cost per part,which is thereason that this region is highly manufacturing volume dependent.

At higher manufacturing volumes, the volume independent region, there are twocategories dominating the cost, investment cost and operator cost. In this regionthe investment points have small impact on the total cost, as seen in Figure 12.1.This means that investment cost has reach a stable level and is hard to decreasefurther without changing the manufacturing line. The operator cost is dependenton the cycle time and the automation level of the process and it is reasonable todraw the conclusion that this region is more cycle time dependent.

Cycle time

The cycle time for the two different cases can be seen in Figure 12.4 and 12.5. InFigure 12.5,the worst case, it is clear that the weaving time, which is dependenton the weaving speed, is influencing the cycle time majorly. An increase cycle timewill result in higher operator costs since more workers are needed to watch themanufacturing during the whole process. It will also result in higher investmentcost since longer cycle time makes more machines and tools necessary to be able tomeet the annual production volume. Since these are the two main cost categoriesin the volume independent region it is reasonable to believe that the weaving speedinfluences the component cost majorly in this region.

.

50000

5

10

15

20

Tim

e[min

]

Z-stringer

S1 Storage

S2 weaving1

S3 CNC cutter [Haas VF-6/50TF]

S4 Mtrl robot [IRB 2400]

S5 Storage


S7 Hydraulic press [50.0 ton]

S8 Mould, [20 MPa], A=0.0625

S9 HPRTM-injector


S11 Storage


S13 Milling machine


S15 Storage

Figure 12.4: Time for manufacturing one component during best case

29

50000

5

10

15

20

25

30

35

40Tim

e[min

] Z-stringer

S1 Storage

S2 weaving2

S3 CNC cutter [Haas VF-6/50TF]


S5 Storage



S8 Mould, [20 MPa], A=0.0625

S9 HPRTM-injector


S11 Storage


S13 Milling machine


S15 Storage

Figure 12.5: Time for manufacturing one component during worst case

12.3 Cost driving parametersIn this section the parameters identified in the previous section are investigated.The investment cost for the 3D weaving machine and the weaving speed are theparameters investigated. The best case scenario is used as a benchmark case andone parameter at a time is changed and the cost per part is plotted against themanufacturing volume to be able to determine the influence in the different regions.

12.3.1 Investment for 3D weaving machineIn Figure 12.6 is the investment cost for the 3D weaving machine is set to 210000ACforthe first case and to 500000ACfor the second case. It is clear that the cost for the3D weaving machine does not influence the total price per component in the vol-ume independent region of the curves. For very low manufacturing volumes theinfluence is higher, but the difference between the two cases decreases rapidly withincreasing manufacturing volume. One reason for the low impact on the total costis that investment cost for the other machines and tools needed is very high, sincea lot of different and expensive machines are needed in the production line. Thehigher investment cost for the 3D weaving machine percentage increases the totalinvestment cost very little in the volume independent region.

It is reasonable to believe that the manufacturing volume in the aerospace indus-try for this type of component is fairly high, since more than 10 stringers are neededper side shell, and is in the volume independent region. With high manufacturingvolumes the investment cost for the 3D weaving machine is not a parameter thatinfluences the total cost per component massively.

30


0

20

40

60

80

100

120Cost per finished component [€]

3D weaving, Low investment

3D weaving, High investment

Figure 12.6: Cost difference between low and high investment cost for 3D weavingmachine

12.3.2 Weaving speedIn Figure 12.7 the weaving speed is set to 0.0017 kg/s as the high weaving speedand 0.000085 kg/s as the low weaving speed while all other parameters are constant.It is clear that weaving speed influences the cost per component massively in thevolume independent region and very little in the first part of the volume dependentregion. In the volume independent region the weaving speed is the major parameterinfluencing the difference between the best and the worst case seen in Figure 12.1.

To make 3D weaving competitive, in the aerospace industry, with the moremature manufacturing techniques used in the industry today it is important todevelop a weaving machine with high weaving speed even if that results in higherinvestment costs.

31


0

20

40

60

80

100


3D weaving, High weaving speed

3D weaving, Low weaving speed

Figure 12.7: Cost difference between low and high weaving speed

Sensitivity analysisIn this chapter the influences of the assumption made are investigated. Each as-sumption is tested with a couple of different values to evaluate and investigate howeach parameter is influencing the component cost.

13.1 Number of pliesIt is assumed that a Non-crimp fabric of the desired thickness is used for the HPRTMmethod in this thesis. If instead a couple of carbon fibre plies were used and stackedon top of each other would the material handling times and the cutting of the pliesincrease the cycle time, which it seen as one important parameter. In Figure 13.1 istwo cases of such scenarios, 20 plies and 40 plies, plotted and compared to the bestand worst case of 3D weaving. It is clearly seen that the number of plies influencesthe cycle time and the cost per component for the HPRTM-manufacturing method.It coulld make it possible that it is cheaper to produce components through 3Dweaving than HPRTM.

32


0

20

40

60

80

100


3D weaving, Best case

3D weaving, Worst case

HPRTM, 20 plies

HPRTM, 40 plies

Figure 13.1: HPRTM comparision with different number of plies

13.2 Filling timeThe assumed filling time in this thesis is set to 5 minutes for all scenarios. In Figure13.2 is the best case plotted with three different filling times. The total curing time,which consist of the filling time and the curing time, is kept constant. It is clear thatthe filling time do not influences the total cost per component. That is connectedto the fact that the total curing time is constant, which is the cycle time for thissubprocess.

33


0

20

40

60

80

100

120Cost

per finis

hed c

om

ponent [€

]

3D weaving, fill time = 5 min, total cure time=10 min


3D weaving, fill time = 10 min,total cure time= 10 min

Figure 13.2: Sensitivity analysis of the filling time

13.3 Curing timeIn this section is then the curing time investigated, the filling time is kept constant.The total curing time is assumed to 10 minutes in this thesis and is in Figure 13.3plotted together with two other scenarios. The decreasing curing times is the costper component also decreasing. It shows that the important factor is to lower thetotal curing time. To do that it is important to be able to fill the mould fast sincethe mould must be fully filled before the gel time, and with decreasing curing timesis also the gel time decreasing.

34


0

20

40

60

80

100




3D weaving, fill time = 5 min,total cure time= 15 min

Figure 13.3: Sensitivity analysis of the curing time

ConclusionsThe results presented in this thesis can be used to understand the process and thecost driving parameters when manufacturing Z-stringer with 3D weaving. The modelis flexible and it is possible to investigate many different scenarios and assumptions.To fully understand the process more cases and scenarios are needed to be testedwithin the model. A general conclusion drawn from this thesis is that 3D weavingmachines development is one key factor to make 3D weaving competitive in theindustry, where the weaving speed is the parameter needed to be focused on. Theother major conclusions drawn are:

• The component cost decreases with increasing manufacturing volume and canbe divided in to two regions.

– Region 1 is highly volume dependent and the reinvestment points affectsthe component cost. Investment cost is the category contributing mostly.

– Region 2 is stable and the reinvestment points are relatively unimpor-tant. Operator cost and investment cost is the main contributions to thecomponent cost.

• It should possible to achieve Z-stringers manufactured by 3D weaving compet-itive with similar components manufactured by other manufacturing methods.

– The component cost in the best case scenario is, in region 2, more orless the same as the cost for the component manufactured by non-crimpfabric (NCF) with HPRTM.

35

– 3D weaving has lower cost compared to HPRTM if single layer plies areused instead of NCF.

• The main cost driving parameters are the cycle time and the investment cost

– In the first sector of region 1 is the investment cost the major cost drivingparameter.

– In region 2 is the cycle time the parameter controlling the componentcost. The major contributions to the cycle time is the total curing timeand the weaving time.

Future workFurther work should be done on developing the cost model, where more data for 3Dweaving machines should be implemented to more accurate estimations. To fullyunderstand at which markets 3D weaving is competing in and to be able to predictthe future for 3D weaving should more components be modeled and more assump-tions investigated. The focus should be on the cost driving parameters identifiedin this thesis and to compare the mechanical properties corresponding to differentmanufacturing methods.

36

References[1] N. Gokarneshan and R. Alagirusamy. Weaving of 3d fabrics: A critical appre-

ciation of the developments. Textile progress, 41(1):1–58, 2009.

[2] N. Khokar. 3d-weaving: Theory and practice. The Journal of The TextileInstitute, 92(2), 2001.

[3] John B. Enns and John K. Gilham. Time-temperature-transformation (ttt)cure diagram: Modeling the cure behavior of thermosets. Journal of AppliedPolymer Science, 28:2567–2591, 1983.

[4] Frank Weiland. Personal contact. Airbus Helicopters, 2015.

[5] Rajkumar. Roy. Cost engineering: Why, what and how? Technical report,Cranfield University, 2003.

[6] E. M. Shehab and H. S. Abdalla. Manufacturing cost modelling for concurrentproduct development. Robotics and Computer Integrated Manufacturing, 2001.

[7] David P. Hoult, C. Lawrence Meador, John Deyst, and Maresi Dennis. Costawareness in design: The role of data commonality. SAE Technical Paper, 1996.

[8] K. Agyapong-Kodua, B. M. Wahid, and R. H. Weston. Towards the derivationof an integrated process cost-modelling technique for complex manufacturingsystems. International Journal of Production Research, 2011.

[9] Alexander Layer, Erik ten Brinke, Fred van Houten, Hubert Kals, and SiegmarHaasis. Recent and future trends in cost estimation. International Journal ofComputer Integrated Manufacturing, 15(6):499–510, 2010.

[10] Stephan Langmaak, Stephan Wiseall, Christophe Bru, Russel Adkins, JamesScanlan, and András Sóbestar. An activity-based-parametric hybrid cost modelto estimate the unit cost of a novel gas turbine component. InternationalJournal of Production Economics, 142(1):74–88, 2013.

[11] J. Verrey, M.D. Wakeman, V. Michaud, and J.A.E Månson. Manufacturingcost comparison of thermoplastic and thermoset rtm for an automotive floorpan. Composites Part A: Applied Science and Manufacturing, 37(1):9–22, 2006.

[12] Than Lin, Jae-Woo Lee, and E.L.J Bohez. New integrated model to estimatethe manufacturing cost and production system performance at the conceptdesign stage of helicopter blade assembly. International Journal of ProductionResearch, 50(24), 2012.

[13] Liyong Tong, Adrian P. Mouritz, and Michael K. Bannister. 3D Fibre ReinforcedPolymer Composites. Elsevier Science Ltd, 2002.

[14] M.H. Mohamed and A. E. Bogdanovich. Comparative analysis of different 3dweaving processes, machines and products. In ICCM17 Conference, 2009.

[15] B.K. Behera and R. Mishra. 3-dimensional weaving. Indian Journal of Fibreand Textile Research, 2008.

37

[16] Biteam AB. Biteam - 3d woven reinforcements. http://biteam.com/profiled-3d-woven-reinforcements.htm.

[17] Fredrik Winberg. Personal contact. Biteam, [email protected], 2015.

[18] B. Tomas Åström. Manufacturing of Polymer Composites. Nelson Thornes,2002.

[19] Klaus Ritter. Rtm advances facilitate mass production in the automotive mar-ket. http://www.reinforcedplastics.com/view/27188/rtm-advances-facilitate-mass-production-in-the-automotive-market/, Retrived at: 2015-02-24.

[20] M. Balasubramanian. Composite Materials and Processing. CRC Press, 2013.

[21] Jeff Sloan. Machining carbon composites: Risky business. High-performanceComposites, 2010.

[22] Karen Wood. In-mold alternatives to postmold decoration. Composites Tech-nology, 2009.

[23] N. D. Ngo and K. K. Tamma. Complex three-dimensional microstructual per-meability predictions of porous fibrous media with and without compaction.International Journal for Numerical Methods in Engineering, 2004.

[24] Suresh G. Advani and E. Murat Sozer. Process Modeling in Composite Manu-facturing. CRC Press, 2002.

[25] Zhong Cai and Timothy Gutowski. Handbook of Composites (2nd Edition):Chapter 26 Consolidation Techniques and Cure Control. Springer, 1998.

[26] Mohammad Waseem Tahir. Dual Scale Porosity and Interlaminar Properties ofComposite Materials. PhD thesis, KTH- Royal Institute of Technology, 2014.

[27] Raju S. Davé and Alfred C. Loos. Processing of Composites. Hanser Publishers,2000.

[28] Interactive Prototyping. Interactive drape. http://interprot.com/interactive-drape/, Retrived at: 2015-05-05.

[29] Mathilda Karlsson. The development of a technical cost model for composites- adapted to the automotive industry. Master’s thesis, KTH- Royal Institute ofTechnology, 2013.

[30] Mats Larsson. Löner för industriarbetare i olika länder. LO- Landsorganisatio-nen i Sverige, 2010.

[31] Malin Åkermo and B. Tomas Åström. Modelling component cost in compressionmoulding of thermoplastic composite and sandwich components. CompositesPart A: Applied Science and Manufacturing, 2000.

[32] NAI Svefa. Svensk fastighetsmarknad- fokus 24 orter.http://www.naisvefa.se/globalassets/svensk-fastighetsmarknad/svensk-fastighetsmarknad-hosten-2012.pdf, Retrived at: 2015-05-25.

38

[33] ABB. Irb 2400 industrial robot-product sheet. www.abb.com, Retrived at:2015-05-22.

[34] Daniela Hagg. Personal contact. Lindauer DORNIER GmbH, 2015.

[35] Lindauer DORNIER GmbH. P1 rapier machine - product brochure.https://www.lindauerdornier.com/global/mediathek/brochures/weaving-machine/dornier-rapier-type-p1-e.pdf, Retrived at: 2015-05-22.

[36] Haas Automation Inc. Vf-6/50tr information site. http://int.haascnc.com, Re-trived at: 2015-06-23.

[37] Robert A. Tatara. Applied Plastics Engineerig Handbook: 17 - CompressionMolding. William Andrew, 2011.

[38] A. Brent Strong. Plastics - material and processing - 3rd edition. PearsonPrentice Hall, 2006.

[39] Elisabeth Behrmann. Bmw-backed researchers closing in on cheaper carbonfiber. www.bloomberg.com, 2014.

[40] Dieffenbacher. Preform technology - advancements in the fully automated pre-form process for complex parts. PDF published on www.speautomotive.com,Retrived at: 2015-05-04.

[41] Hexion. Technical data sheet - epikote resin mgs 135.https://www.hexion.com/Products/TechnicalDataSheet.aspx?id=8246, Re-trived at: 2015-06-24.

[42] CompositeWorld. Amrc installs dornier rapier loom for carbon fiberweaving. http://www.compositesworld.com/news/amrc-installs-dornier-rapier-loom-for-carbon-fiber-weaving, Retrived at: 2015-06-01.

[43] OECD. Ppps and exchange rates. http://stats.oecd.org, Retrieved: 2015-06-26.

39

A Machine dataEquipment, tooling and storage data are presented in Table A1. Data marked witha star is assumed, further information in Chapter 11. The fuse is in most casescalculated from known power consumption and the necessary supply voltage.

Table A1: Machine data used in developed cost model

Machine Cost[kAC]

Fuse[A]

Electricityconsumption[kW]

Floorspace[m2]

Material-handlingRobot:IRB 2400 [33]

60 1.34 0.67 0.4338

2D weavingmachine:P1 Rapier [34, 35,42]

210 84* 32* 10-20

CNC-cutter :VF-6/50TR [36]

165 97 22.4 16

HP-RTMequipment:

300 57* 65 7.78

Milling Machine: 300 100* 50 6.25

40

B Weaving speed for 2D machineThe weaving speed for 2D weaving machines is provided by Lindauer DORNIERGmbH [34]. The speed is based on calculations of the most woven carbon fabrics.The data for the most common fabrics is presented in Table B1. With 250 weftinsertions per minute and 6,5 ends per cm a weaving speed of 38.5 cm per minuteis obtained. With a fabric width of 1300 mm is the speed transformed to 30 squaremeters per hour.

Table B1: Data for carbon fabrics woven by 2D machines

Filling density 6,5 [ends/cm]Fabric width 1300 [mm]

Weft insertion speed 250 [rpm]

41

C Hydraulic press investment costsOne of the factors influencing the investment cost for a hydraulic press is the requiredtonnage capacity. The investment cost could be approximated with Equation 1.

Cpress = 1.25 · 1000 · tonnage (1)

where tonnage is the required press force in ton-force. Equation 1 is used togetherwith data for 14 different Omera presses, presented in Table C1 [29]. The necessaryfuse for the presses is extrapolated from knowing a 630 ton press need 400 volt andapproximately a 120 Ampere fuse.

Table C1: Hydraulic press data

Press ca-pacity[tonnage]

Max ram[m ·m]

Daylightopening[m]

Open/closespeed [m/s]

Electricityconsumption[kW ]

50 0.75 x 0.6 0.8 0.045 7.580 1.2 x 1.1 1 0.06 15125 1.2 x 1.1 1 0.105 2 x 22160 1.2 x 1.1 1 0.078 55200 1.4 x 1.2 1 0.063 55250 1.4 x 1.2 1 0.065 110315 1.6 x 1.2 1.3 0.052 110400 2.5 x 1.5 1.3 0.083 2x75500 2.5 x 1.5 1.3 0.066 2x75630 2.5 x 1.5 1.3 0.054 2x75800 2.5 x 1.5 1.3 0.043 2x751250 4.5 x 1.5 1.3 0.044 2x1101600 4.5 x 1.5 2 0.033 2x1102000 5 x 2.5 2 0.035 4x74

42

D Data for closed mouldsTwo types of mould are available to use in the developed cost model, aluminummoulds and steel moulds. Steel moulds withstand higher pressures than aluminummoulds and requires less maintenance. The aluminum moulds are however lessexpensive. The reference data used in this thesis originates from previous work [31].The cost is updated according to [43] and presented in Table D1.

Table D1: Mould data

Mould type Mould area [m ·m] Cost [kAC]

Aluminium 0.25 x 0.25 110.5 x 0.5 190.65 x 0.85 481 x 1.2 116

Steel 0.25 x 0.25 310.5 x 0.5 550.65 x 0.85 661 x 1.2 158

43

E Detailed analysis of HP-RTMThe detailed analysis of the cost corresponding to the hand lay-up and high pressureRTM technique (HP-RTM) is presented in the following figures. In Figure E.1 is thecost presented in detail for some different manufacturing volumes. In Figure E.2 isthe manufacturing time presented in detail.

5000 50000 2000000

10

20

30

40

50

60

70

80

90

[Euro

]

Z-stringer

FloorCost

InvestCost

ToolingCost

OperatorCost

VarElCost

FixElCost

MtrlCost

MtrlWasteCost

Figure E.1: Part cost for the HP-RTM

44

50000

5

10

15

20Tim

e[min

] Z-stringer

S1 Storage

S2 Cnc cutter [zund G3]


S4 Storage



S7 Mould, [2 MPa], A=0.0625


S9 Storage



S12 Mould, [20 MPa], A=0.0625

S13 HPRTM-injector


S15 Storage


S17 Milling machine


S19 Storage

Figure E.2: Manufacturing time for the HP-RTM

45

Cost Modelling of Composite Components for Helicopter …919302/... · 2016. 4. 13. · Composites...

Documents

Transcript of Cost Modelling of Composite Components for Helicopter …919302/... · 2016. 4. 13. · Composites...