24 Design and Analysis of Composite Structures Technology A misc. Technology skins doors frames decks Technology Resulting cost savings distri- bution for given technology fittings stringers mix Figure 2.11 Combining different technologies to an airframe and expected cost distribution Given the data in Tables 2.2 and 2.3,one can combine different technologies to make a part family.Doing that over all part families results in a technology mix.This technology mix has an overall mean cost savings and variance associated with it that can be calculated using the data from Tables 2.2 and 2.3 and using the percentages of how much of each part family is made by each technology [12,13].This process is shown in Figure 2.11.Obviously,some technology mixes are better than others because they have lower recurring cost and/or lower risk.An optimization scheme can then be set up[13]that aims at determining the technology mix that minimizes the overall recurring cost savings(below the SMT baseline)keeping the associated variance(and thus the risk)below a preselected value.By changing that preselected value from very small(low risk) to very high(high risk)different optimum mixes can be obtained.A typical result of this process is shown in Figure 2.12 for the case of a fuselage and wing of a 20-passenger commuter plane. 35 50%probability of lower svgs 25 25%probability of lower svgs 20 一10% 15 5% 0 2.5% 1% 0 0.05 0.1 0.15 0.2 0.25 Risk(std.dev.of savings) Figure 2.12 Recurring cost savings as a function of risk
Given the data in Tables 2.2 and 2.3, one can combine different technologies to make a part family. Doing that over all part families results in a technology mix. This technology mix has an overall mean cost savings and variance associated with it that can be calculated using the data from Tables 2.2 and 2.3 and using the percentages of how much of each part family is made by each technology [12, 13]. This process is shown in Figure 2.11. Obviously, some technology mixes are better than others because they have lower recurring cost and/or lower risk. An optimization scheme can then be set up [13] that aims at determining the technology mix that minimizes the overall recurring cost savings (below the SMT baseline) keeping the associated variance (and thus the risk) below a preselected value. By changing that preselected value from very small (low risk) to very high (high risk) different optimum mixes can be obtained. A typical result of this process is shown in Figure 2.12 for the case of a fuselage and wing of a 20-passenger commuter plane. skins frames … fittings stringers decks … doors … misc. Technology A Technology B Technology C Resultingcostsavingsdistributionforgiventechnology mix Figure 2.11 Combining different technologies to an airframe and expected cost distribution 0 5 10 15 20 25 30 35 0 0.05 0.1 0.15 0.2 0.25 Risk (std. dev. of savings) Recurring cost savings over baseline (%) 50% probability of lower svgs 25% probability of lower svgs 10% 5% 2.5% 1% Figure 2.12 Recurring cost savings as a function of risk 24 Design and Analysis of Composite Structures
Cost of Composites:a Qualitative Discussion 25 The risk is shown in Figure 2.12 on the x axis as the square root of the variance,or standard deviation of the cost savings of the resulting technology mix.For each value of risk,the optimization process results in a technology mix that maximizes cost savings.Assuming that the cost savings of each technology mix is normally distributed,the corresponding probabili- ties that the cost savings will be lower than a specified value can be determined [13].These different probabilities trace the different curves shown in Figure 2.12.For example,if the risk is set at 0.05 on the x axis,the resulting optimum mix has 1%probability of not achieving 11.5% savings,2.5%probability of not achieving 13.5%savings,5%probability of not achieving 15% savings and so on.Note that all curves,except the 50%probability curve go through a maximum.This maximum can be used for selecting the optimum technology mix to be used. For example,if a specific factory/management team is risk averse it would probably go with the 1%curve which goes through a maximum at a risk value slightly less than 0.05.The team would expect savings of at least 11.5%.A more aggressive team might be comfortable with 25% probability that the cost savings is lower and would use the 25%curve.This has a maximum at a risk value of 0.09 with corresponding savings of 22.5%.However,there is a 25%probability that this level of savings will not be met.That is,if this technology mix were to be implemented a large number of times,it would meet or exceed the 22.5%savings target only 75%of the time. It is up to the management team and factory to decide which risk level and curve they should use.It should be noted that for very high risk values,beyond 0.1,the cost savings curves eventually become negative.For example the 1%curve becomes negative at a risk value of 0.13.This means that the technology mix corresponding to a risk value of 0.13 has so much uncertainty that there is 99%probability that the cost savings will be negative,i.e.the cost will be higher than the SMT baseline. Once a risk level is selected from Figure 2.12,the corresponding technology mix is known from the optimization process.Examples for low and high risk values are shown in Figures 2.13 and2.14. Misc. SMT 8% 15% ALP 7% PLT1% HLP 41% RTM 12% 3% AFP 13% HSM Figure 2.13 Optimum mix of technologies for small airplane (low risk)
The risk is shown in Figure 2.12 on the x axis as the square root of the variance, or standard deviation of the cost savings of the resulting technology mix. For each value of risk, the optimization process results in a technology mix that maximizes cost savings. Assuming that the cost savings of each technology mix is normally distributed, the corresponding probabilities that the cost savings will be lower than a specified value can be determined [13]. These different probabilities trace the different curves shown in Figure 2.12. For example, if the risk is set at 0.05 on the x axis, the resulting optimum mix has 1% probability of not achieving 11.5% savings, 2.5% probability of not achieving 13.5% savings, 5% probability of not achieving 15% savings and so on. Note that all curves, except the 50% probability curve go through a maximum. This maximum can be used for selecting the optimum technology mix to be used. For example, if a specific factory/management team is risk averse it would probably go with the 1% curve which goes through a maximum at a risk value slightly less than 0.05. The team would expect savings of at least 11.5%. A more aggressive team might be comfortable with 25% probability that the cost savings is lower and would use the 25% curve. This has a maximum at a risk value of 0.09 with corresponding savings of 22.5%. However, there is a 25% probability that this level of savings will not be met. That is, if this technology mix were to be implemented a large number of times, it would meet or exceed the 22.5% savings target only 75% of the time. It is up to the management team and factory to decide which risk level and curve they should use. It should be noted that for very high risk values, beyond 0.1, the cost savings curves eventually become negative. For example the 1% curve becomes negative at a risk value of 0.13. This means that the technology mix corresponding to a risk value of 0.13 has so much uncertainty that there is 99% probability that the cost savings will be negative, i.e. the cost will be higher than the SMT baseline. Once a risk level is selected from Figure 2.12, the corresponding technology mix is known from the optimization process. Examples for low and high risk values are shown in Figures 2.13 and 2.14. 8% 41% 13% 3% 12% 1% 7% 15% SMT HLP HSM AFP RTM PLT ALP Misc. Figure 2.13 Optimum mix of technologies for small airplane (low risk) Cost of Composites: a Qualitative Discussion 25
26 Design and Analysis of Composite Structures Misc. 洲 7% HLP 15% ALP 12% PLT 2% 22% 6% HSM RTM 20% AFP Figure 2.14 Optimum mix of technologies for small airplane (high risk) For the low risk optimum mix of Figure 2.13,there is a 10%probability of not achieving 12.5%cost savings.For the high risk optimum mix of Figure 2.14 there is a 10%probability of not achieving 7%cost savings.The only reason to go with the high risk optimum mix is that,at higher probability values(greater than 25%)it exceeds the cost savings of the low-risk optimum mix. A comparison of Figures 2.14 and 2.13 shows that as the risk increases,the percentage usage of baseline SMT and low-risk low-return HLP and RTM decreases while the usage of higher- risk high-return AFP and ALP increases.ALP usage doubles from 6 to 12%and AFP usage increases by a factor of almost 7,from 3 to 20%.The amount of PLT also increases(in fact doubles)but since PLT is only limited to stringers in this example,the overall impact of using PLT is quite small.It should be noted that there is a portion of the airframe denoted by 'Misc'. These are miscellaneous parts such as seals,or parts for which applicability is unclear,and mixing technologies(for example pultruded stringers co-bonded on fiber-placed skins)might be a better option,but no data were available for generating predictions. Finally,the breakdown by part family for one of the cases,the low-risk optimum mix of Figure 2.13 is shown in Table 2.4.For example,21.1%of the frames are made by HLP,32.4%by Table 2.4 Low-risk technology mix by part family and technology %SMT %HLP %HSM %AFP %RTM %PLT %ALP Skins +.. 0 81 0 4.9 3.6 0 10.5 Frames +.. 0 21.1 32.4 0 40.5 0 6 Stringers 79.3 0 0 0 0 20.7 0 Fittings 44.1 0 55.5 0 0 0 0.4 Decks +.. 0 76.5 0 8.4 3.1 0 12 Doors+... 0 81 0 10 0 0 9
For the low risk optimum mix of Figure 2.13, there is a 10% probability of not achieving 12.5% cost savings. For the high risk optimum mix of Figure 2.14 there is a 10% probability of not achieving 7% cost savings. The only reason to go with the high risk optimum mix is that, at higher probability values (greater than 25%) it exceeds the cost savings of the low-risk optimum mix. A comparison of Figures 2.14 and 2.13 shows that as the risk increases, the percentage usage of baseline SMT and low-risk low-return HLP and RTM decreases while the usage of higherrisk high-return AFP and ALP increases. ALP usage doubles from 6 to 12% and AFP usage increases by a factor of almost 7, from 3 to 20%. The amount of PLT also increases (in fact doubles) but since PLT is only limited to stringers in this example, the overall impact of using PLT is quite small. It should be noted that there is a portion of the airframe denoted by ‘Misc’. These are miscellaneous parts such as seals, or parts for which applicability is unclear, and mixing technologies (for example pultruded stringers co-bonded on fiber-placed skins) might be a better option, but no data were available for generating predictions. Finally, the breakdown by part family for one of the cases, the low-risk optimum mix of Figure 2.13 is shown in Table 2.4. For example, 21.1% of the frames are made by HLP, 32.4% by 6% 15% 22% 20% 6% 2% 12% 17% HLP SMT RTM AFP PLT HSM ALP Misc. Figure 2.14 Optimum mix of technologies for small airplane (high risk) Table 2.4 Low-risk technology mix by part family and technology %SMT %HLP %HSM %AFP %RTM %PLT %ALP Skins þ ... 0 81 0 4.9 3.6 0 10.5 Frames þ ... 0 21.1 32.4 0 40.5 0 6 Stringers 79.3 0 0 0 0 20.7 0 Fittings 44.1 0 55.5 0 0 0 0.4 Decks þ ... 0 76.5 0 8.4 3.1 0 12 Doors þ ... 0 81 0 10 0 0 9 26 Design and Analysis of Composite Structures
Cost of Composites:a Qualitative Discussion 27 HSM,40.5%by RTM and 6%by ALP.Note that SMT is only used for three quarters of the stringers and almost half the fittings.Note that these percentages are the results of the optimization mentioned earlier and do not exactly determine which parts will be made with what process,only that a certain percentage of parts for each part family is made by a certain process. It is up to the designers and manufacturing personnel to decide how these percentages can be achieved or,if not possible,determine what the best compromise will be.For example,6%of the frames and bulkheads made by ALP would probably correspond to the pressure bulkheads and any frames with deep webs where automated layup can be used effectively. The above discussion focused on recurring cost as a driver.The optimum technology mixes determined have a certain weight and nonrecurring cost associated with them.If weight or nonrecurring cost were the drivers different optimum technology mixes would be obtained. Also,the optimized results are frozen in time in the sense that the applicabilities of Table 2.2 and the cost figures of Table 2.3 are assumed constant.Over time,as technologies improve, these data will change and the associated optimum technology mixes will change.Results for the time-dependent problem with different drivers such as nonrecurring cost or optimum return on investment can be found in the references [11,14]. It should be kept in mind that some of the data used in this section are subjective or based on expectations of what certain technologies will deliver in the future.As such,the results should be viewed as trends that will change with different input data.What is important here is that an approach has been developed that can be used to trade weight,recurring cost and nonrecurring cost and determine the optimum mix of technologies given certain cost data.The interested user of the approach can use his/her own data and degree of comfort in coming up with the optimum mix of technologies for his/her application. 2.4 Summary and Conclusions An attempt to summarize the above discussion by fabrication process and collect some of the qualitative considerations that should be taken into account during the design and analysis phases of a program using composite materials is shown in Table 2.5.For reference,sheet metal built-up structure and high-speed-machining (aluminum or titanium) are also included.This table is meant to be a rough set of guidelines and it is expected that different applications and manufacturing experiences can deviate significantly from its conclusions. As shown in the previous section,there is no single process that can be applied to all types of parts and result in the lowest recurring and/or nonrecurring costs.A combination of processes is necessary.In many cases,combining two or more processes in fabricating a single part,thus creating a hybrid process (for example automated fiber-placed skins with staged pultruded stiffeners,all co-cured in one cure cycle)appears to be the most efficient approach.In general,co-curing as large parts as possible and combining with as much automation as possible seems to have the most promise for parts of low cost,high quality and consistency.Of course,the degree to which this can be done depends on how much risk is considered acceptable in a specific application and to what extent the investment required to implement more than one fabrication processes is justified by the size of the production run.These combinations of processes and process improvements have already started to pay off and,for certain applications [15]the cost of composite airframe is comparable if not lower than that of equivalent metal structure
HSM, 40.5% by RTM and 6% by ALP. Note that SMT is only used for three quarters of the stringers and almost half the fittings. Note that these percentages are the results of the optimization mentioned earlier and do not exactly determine which parts will be made with what process, only that a certain percentage of parts for each part family is made by a certain process. It is up to the designers and manufacturing personnel to decide how these percentages can be achieved or, if not possible, determine what the best compromise will be. For example, 6% of the frames and bulkheads made by ALP would probably correspond to the pressure bulkheads and any frames with deep webs where automated layup can be used effectively. The above discussion focused on recurring cost as a driver. The optimum technology mixes determined have a certain weight and nonrecurring cost associated with them. If weight or nonrecurring cost were the drivers different optimum technology mixes would be obtained. Also, the optimized results are frozen in time in the sense that the applicabilities of Table 2.2 and the cost figures of Table 2.3 are assumed constant. Over time, as technologies improve, these data will change and the associated optimum technology mixes will change. Results for the time-dependent problem with different drivers such as nonrecurring cost or optimum return on investment can be found in the references [11, 14]. It should be kept in mind that some of the data used in this section are subjective or based on expectations of what certain technologies will deliver in the future. As such, the results should be viewed as trends that will change with different input data. What is important here is that an approach has been developed that can be used to trade weight, recurring cost and nonrecurring cost and determine the optimum mix of technologies given certain cost data. The interested user of the approach can use his/her own data and degree of comfort in coming up with the optimum mix of technologies for his/her application. 2.4 Summary and Conclusions An attempt to summarize the above discussion by fabrication process and collect some of the qualitative considerations that should be taken into account during the design and analysis phases of a program using composite materials is shown in Table 2.5. For reference, sheet metal built-up structure and high-speed-machining (aluminum or titanium) are also included. This table is meant to be a rough set of guidelines and it is expected that different applications and manufacturing experiences can deviate significantly from its conclusions. As shown in the previous section, there is no single process that can be applied to all types of parts and result in the lowest recurring and/or nonrecurring costs. A combination of processes is necessary. In many cases, combining two or more processes in fabricating a single part, thus creating a hybrid process (for example automated fiber-placed skins with staged pultruded stiffeners, all co-cured in one cure cycle) appears to be the most efficient approach. In general, co-curing as large parts as possible and combining with as much automation as possible seems to have the most promise for parts of low cost, high quality and consistency. Of course, the degree to which this can be done depends on how much risk is considered acceptable in a specific application and to what extent the investment required to implement more than one fabrication processes is justified by the size of the production run. These combinations of processes and process improvements have already started to pay off and, for certain applications [15] the cost of composite airframe is comparable if not lower than that of equivalent metal structure. Cost of Composites: a Qualitative Discussion 27
Table 2.5 Qualitative cost considerations affecting design/analysis decisions Process Application Comments Sheet metal All airframe structure Assembly intensive,relatively heavy.Moderate tooling costs including fit-out and assembly jigs High-speed machining Frames,bulkheads,ribs, Very low tooling cost.Very low recurring cost.Can generate any desired thickness beams,decks and greater than 0.6 mm.Moderate raw material cost due to the use of special alloys. floors.In general,parts Extremely high scrap rate(more than 99%of the raw material ends up recycled with one flat surface as machined chips).Limited due to vibrations to part thicknesses greater than that can be created via 0.6-0.7 mm.Issues with damage tolerance (no built-in crack stoppers)repair machining methods,and low damping:Size of billet limits size of part that can be fabricated Hand Layup All airframe structure Weight reductions over equivalent metal of at least 15%.Recurring cost competitive with sheet metal when large amount of co-curing is used.Moderate scrap.High raw material cost.High tooling cost.Hard to fabricate 3-D fittings. Reduced out-of-plane strength(important in fittings and parts with out-of-plane loading) Automated fiber/tow placement Skins,decks,floors,doors, Weight reductions similar to hand layup.Recurring cost can be less than metal fairings,bulkheads, baseline if the number of starts and stops for the machine are minimized(few Desi large ribs and beams. cutouts,plydrops,etc.).Less scrap than hand layup..High tooling cost.For parts In general,parts with made on concave tools,limited by size of robotic head (interference with tool). large surface area Fiber steering is promising for additional weight savings but is limited by maximum radius of curvature the machine can turn without buckling Analy the tows RTM All airframe structure Weight reductions somewhat less than hand layup due to decreased fiber volume for complex parts.Combined with automated preparation of fiber performs it can result in low recurring fabrication cost.Relatively low scrap rate.Very high tooling cost if matched metal tooling is used.Less so for vacuum-assisted RTM omposite (half of the tool is a semi-rigid caul plate)or resin film infusion.To use unidirectional plies,some carrier or tackifier is needed for the fibers,increasing the recurring cost somewhat Structures
Table 2.5 Qualitative cost considerations affecting design/analysis decisions Process Application Comments Sheet metal All airframe structure Assembly intensive, relatively heavy. Moderate tooling costs including fit-out and assembly jigs High-speed machining Frames, bulkheads, ribs, beams, decks and floors. In general, parts with one flat surface that can be created via machining Very low tooling cost. Very low recurring cost. Can generate any desired thickness greater than 0.6 mm. Moderate raw material cost due to the use of special alloys. Extremely high scrap rate (more than 99% of the raw material ends up recycled as machined chips). Limited due to vibrations to part thicknesses greater than 0.6–0.7 mm. Issues with damage tolerance (no built-in crack stoppers) repair methods, and low damping; Size of billet limits size of part that can be fabricated Hand Layup All airframe structure Weight reductions over equivalent metal of at least 15%. Recurring cost competitive with sheet metal when large amount of co-curing is used. Moderate scrap. High raw material cost. High tooling cost. Hard to fabricate 3-D fittings. Reduced out-of-plane strength (important in fittings and parts with out-of-plane loading) Automated fiber/tow placement Skins, decks, floors, doors, fairings, bulkheads, large ribs and beams. In general, parts with large surface area Weight reductions similar to hand layup. Recurring cost can be less than metal baseline if the number of starts and stops for the machine are minimized (few cutouts, plydrops, etc.). Less scrap than hand layup,. High tooling cost. For parts made on concave tools, limited by size of robotic head (interference with tool). Fiber steering is promising for additional weight savings but is limited by maximum radius of curvature the machine can turn without buckling the tows RTM All airframe structure Weight reductions somewhat less than hand layup due to decreased fiber volume for complex parts. Combined with automated preparation of fiber performs it can result in low recurring fabrication cost. Relatively low scrap rate. Very high tooling cost if matched metal tooling is used. Less so for vacuum-assisted RTM (half of the tool is a semi-rigid caul plate) or resin film infusion. To use unidirectional plies, some carrier or tackifier is needed for the fibers, increasing the recurring cost somewhat 28 Design and Analysis of Composite Structures