~~u;g( mm Hot rulling : ,Dit:cll~ting(Al) Shell casting (sleel) Cold rolling Forglng Isteel} Sand casurtg tsteel) 34 Manufacturing Analysis; Some Basic Ouesttona for a Start-Up Company Chap, 2 Forging(AI.Mg);~~ (AI,Cllstiron) Thermoplastic polymers 0.4 i 0.3 ~ ~ 1 0.2 .1 ~ 0.1 Minimum dimension of web w (in.) FifW"lil2.7 Process capabilities related so part geometry. Very thin sections tevor rolling and thermotorrmng: "cDunky"s<:ctiQusfavor machining and injection molding (from fmroductivlIlIJ Manufacturing Processes by J, A Schey, if) 1987. Reprinted with permission of the McGraw-Hill Companies), The thermoforming of plastic sheets is slightly above cold rolling in the graph. This also creates sections that are relatively thin, and thus it competes with cold rolled metal products for many common items that require less structural rigidity, The middle part of the graph relates to processes that create more "chunky" looking parts of greater thickness (the y axis in the figure). Finally, note that the mold making procedures in sand casting prevent it from being selected if one of the dimensions is less than 5 millimeters (0.2 inch), 2.3.7 Accuracy, Tolerances, and FideUty between CAD and CAM In all fabrication processes-semiconductors, plastics, metals, textiles, or other- wise-the physical limitations of each process have a major impact on the echiev- able accuracy. Each processing operation comes with a bounding envelope of performance that is constrained by the physical and/or chemical processes that, during fabrication, are imposed on the original work material. This begs the fol- lowing question: How much fidelity will there be between (a) the specified CAD geometry, tolerances, and desired strength and (b) the final physical object that is manufactured? In the best case scenario, the CAD geometry will be perfectly trans- lated into the fabricated geometry. Also, the properties of the original piece of work material stock will be either unchanged or possibly work-hardened into an even more preferred state. 2.3 Question 2: How Much Will the Product Cost to Manufacture (e)? 35 Accuracy microns TABlE 2.3 Routine Accuracies for Mechanical Processes (One "Thou" Approximately = 25 Micronsl Accuracy inches Hot, open die forging Hot, closed die forging Investment casting Cold, closed die forging Machining Eleetrodischarge machining Lapping and polishing +f-1250microns + f - 500 microns +f-75-250microns +1- 50 125microns +/-25-125microns +/-12.5microns +1-0.25 microns +/-0.05 inch +/-0.02mch +/-0.003-0.01 inch +/- 0,002-0.005 inch +/- 0.001-0.005 inch +/- 0.0005 inch +/- 0.ססOO1inch In the worst case situation, a poorly controlled process will damage a perfectly good work material. Examples of tbis were widespread in the early days of welding, where beat-affected zones reduced the fracture toughness of materials. Controlling this envelope for each process is quite complex and relies on a number of factors, which include: •The properties of the work materials that are being formed/machined/ deposited •The properties of the tooling/masking/forming media •The characteristics of the basic processing machinery and its control structure • The number of parameters in the physics or chemistry of the process • Sensitivity of tbe process to external disturbances such as dirt, friction, and humidity Table 2.3 and Figure 2.8 convey the typical tolerances that can be obtained. Note that even witbin one particular process there can be subtle differences in performance, resulting in a range of tolerance. The darkest bars in the center of each process are the normally anticipated values. This range is given the name natural tol- erance (NT) of the process and is crucially important in both design and manufac- turing work. It cannot be emphasized enough that the cost of manufacturing, and the sub- sequent cost of any consumer product, is related to the designer's selection of part accuracy and dimensional tolerance. Once the design and its related tolerances reach a factory floor, the manufac- turers will be obliged to choose processes that deliver the accuracy and NT implicit in the decisions made by the designer. Quite clearly, costs will rise rapidly if the designer has been overdemanding or just thoughtless. Poor design decisions could result in the obligatory choice of an inherently expensive manufacturing process. The next concept to emphasize is that of process chains within a particular family of manufacturing processes. Examples of these are also shown on the Website <cybercut.berkeley.edu>. In general, several processes are used sequentially to gradually achieve a highly accurate, smooth surface. A common chain in mechanical manufacturing is to start with a flame-cut plate. a casting, or a forging to obtain the Process 36 Manufacturing Analysis: Some Basic Ouestions for a Start-Up Company Chap, 2 in.X 10-3 100 50 Process Traditional Flame cutting Hand grinding Disk grinding or filmg Turning. shaping, or milling Drilling Boring Reaming or broaching Grinding Honing, lapping, buffing, or polishing Nontraditional Plasma beam machining Electrical discharge machining Chemical machining Electrochemical machining Laser beam or electron beam machiru Electrochemical grinding Electropolishing c:::=J Less frequent application _Averageappllcation 2.0 0.5 0.2 0.05 0,02 0.005 0.002 z Tolerance frnrn] F1guu ZJI Natural tolerances (NT) ~ Ihe darker bands, for a variety of common mechanical manufacturing processes. Variations = the lighter bands (from MClI1ufacrurmg Processes for Engineering Materials by Kalpakjian, © 1997. Reprinted by permission of Prentice-Hall, Inc., Upper Saddle River, NJ). bulk shape. Flame cutting could then be followed by a series of machining operations to obtain further accuracy. These can then be followed by grinding and polishing if high accuracy and finish are desired by the designer. In Figure 2.8, the NTs of flame cutting, machining, and grinding are shown, moving across from left to right with finer accuracy. Several points should be made: •The designer should realize that these process chains exist, as summarized in the simple diagram of Figure 2.9. • Each additional process is needed after a certain transitional tolerance. If the designer is unaware of these transitions, unnecessary finishing costs may be created, as shown in Figure 2.10. The other side of this coin is that manufac- turing costs can he saved if the designer is willing to loosen desired tolerances. • The manufacturing quality assurance at one step in the process chain must be carefully executed before moving on to the next process. If a "parent" process is "ended too early," the next "child" process may have too much or an impos- sible amount of work to do. (Imagine cleaning a rusty garden tool; heavy 2.3 Question 2: How Much Will the Product Cost to Manufacture (e)7 37 015 015 Secondary process flat capability FiJUre2.' Process chains with levelsof tojerance grinding or heavy abrasive papers are needed before moving on to the final polishing steps.) 2.3.8 Product Life Expectancy Recall that part strength is listed as the third criterion in Table 2.2. It is related to the design geometry, tolerances, material selected, and chosen manufacturing method. These factors also have a coupled influence on the long-term in-service life. Aero- space and structural engineers are probably the designers who are most concerned with these long-term properties. Hertzberg (1996) and Dowling (1993) describe the fatigue properties of metals and polymers. The influences of material composition and local-geometry effects are also described. A fatigue failure always begins at a stress concentration. A sharp corner, a small hole, a rapid transition in diameter are examples of danger zones for crack initiation. Designers in such fields will specify high integrity grades of steel and aluminum, will choose processes like forging and forming (rather than casting) to maintain a homogeneous grain structure, and will specify additional final finishing operations such as grinding and lapping. These I Drill IEDM I Broach Ream IBm",l Honing I Hole hierarchy Flat hierarchy c;ingl IFinegrind Broach IEDMI I Mill !Roughgrjnd Secondary process hole capability Surt rougjrin Itr-e tnche Dtm acc in Hr-s mcnes Dim ace in 10 _1 inches Surf rough in Iu-e Inches 38 Manufacturing Analysis: Some Basic Questions for a Start-Up Company Chap. 2 400 Figure2.10 Finishingcostsincreaseasa part moves from a rough casting.to a finish-machined part, to fine-honed final product (from Manufacturing Processes for Engineering Materials by Kalpakjian, © 1997. Reprinted by permission of Prentice-Hail, Inc" Upper Saddle River,NJ). #" 300 i ~ 200 i ~ 100 additional operations lead to very smooth surfaces that give dramatically improved long-term fatigue life. Figure 2.10 illustrates the costs of these additional fine finishing operations. The additional grind and hone operations add 400% more cost over the as-forged, or as-cast, surfaces. Even in comparison with turning on a lathe, they add 200 to 300% more cost. It is not surprising that carefully manufactured aircraft components, or the surface of a production quality plastic injection mold, are so very expensive. 2.3.9 Lead lime Lead time is defined for this book as "the number of weeks between the release of detailed CAD files to the fabrication facility and the actual production of the part." It is a small subcomponent of the total time-to-market. This broader topic will be reviewed in greater depth in Section 2.5.For this overview, the important point isthat lead time is very dependent on the designer's decisions, which then have direct impli- cations on the choice of manufacturing process. The desired batch size, part geom- etry, and accuracy are the main factors. As a benchmark, a small batch of medium complexity metal parts with +/~50 microns (+1- 0.002 inch) accuracy can be obtained from a production machine shop with a two- to three-week turnaround time, obviously depending on normal business conditions. However, several weeks of lead time will be experienced as soon as a serious mold or die is needed. For the processes like forging, sheet metal forming, and high- volume plastic injection molding, the die making involves many extra steps. During die design, factors such as springback for metals and shrinkage for plastics need to be incorporated. Since the deformation stresses that build up during manufacturing are high, the die designer also has to create supporting blocks and pressure plates. The designer will also need to consider parting planes and the draft angles that give slight tapers to any vertical walls: these are needed to ensure that the part can be ejected after forming. Unfortunately, perfect analytical models do not exist yet for 2.3 Question 2: How Much Will the Product Cost to Manufacture Ie)? 39 predicting the precise amounts of springback or the best draft angle. This usually means that handcrafting is needed in the production of the first die. Subsequent trial- and-error adjustments and iterations to the die surfaces are always needed. The above paragraph still pertains only to one machine and one process. Several months of lead time are needed to set up the large-scale FMS systems and high-volume batches indicated at the bottom right of Figure 2.6.These contain a large number of manufacturing processes, linked together and scheduled to make complex subassem- blies.And of course as product complexity and scope increase, the lead time increases proportionally. At the extreme, for a completely new model of aircraft or automobile, the lead time from design to first product will run into years rather than months. 2.3.10 Cost Factors Especially Related to Adjoining Parts The following example shows how design and manufacturing keep changing to suit a rather complex interaction between (a) the availability of innovative manufacturing techniques and (b) new economic conditions. In other words, recalling Ayres and Miller's (1983) quotation in Chapter 1, "elM is the confluence of the supply elements and the demand elements." TWentyor even ten years ago, it would not have seemed reasonable to machine very large structures from a solid monolithic slab. However, innovative machining pro- grams at BoeingAircraft are proceeding in that direction. Inside the ceilingof the plane, structural members that resemble giant coat hangers are spaced across the plane at intervals to giveit torsional stability.Today,most are made from many conjoined pieces. This arrangement is shown in the upper photograph of Figure 2.11. However, newer designs are favoring machining from one very large solid slab, as shown in the lower photograph. This eliminates costly and unpredictable joining operations in the factory. Thomas (1994) has observed that such manufacturing innovations flowing back into the design phase must be the new way of organizing the relationship between design and manufacturing. Using Ayres and Miller's definition of CIM we can observe that the new innovations, or the new supply elements, include: •Improved cutting tool technology and an understanding of how to control the accuracy of very high speed machining processes •The availability of stiffer machine tools and very high speed spindles •More homogeneous microstructures that give uniformity in large forging slabs •The ability to carry out comprehensive testing and show that these one-piece structures are at least if not more reliable than multiple-piece structures. Meanwhile the new demand elements include: • Escalating costs of joining and riveting operations, which can only be partially automated. Specifically these operations often require manual fixturing of the workpieces •A preference fOT eliminating multiple fabrication steps, which always demand more setup, fixturing, documentation, and quality assurance. • General pressures on the whole airline industry, since deregulation, to cut costs and yet improve the safety and the integrity of the aircraft. 40 Manufacturing Analysis: Some Basic Questions for a Start-Up Company Chap. 2 FIgure 2.11 Integrated product and process design allows this aerospace component to be completely machined from the solid as shown in the lower photograph (counesy of Dr. Donald Sandstrom, The Boeing Company). These trends introduce a great deal of complexity into the design and manu- facturing process, but on the other hand, creative companies can exploit them to their advantage. The conclusion to be drawn is that no single component should be ana- lyzed and optimized in isolation. There will always be something that can be improved, simplified, or made cheaper if design and manufacturing are viewed from a slightly wider system perspective. 2.3.11 Analyzing Costs in Terms of the Profit Potential Hewlett-Packard's return map (RM) is another method for analyzing design and manufacturing costs. However, it focuses not just on cost but on these survival questions: • How much profit. aP, will be made at any given time? • How long, T b , will it take to make any profit? Fignre 212 plots the costs or revenues against logarithmic time expelled. The key curves on the chart (modeled on House and Price, 1991) are: 2.3 Question 2: How Much Will the Product Cost to Manufacture (en 4' 1,000 Total sales RF", return factor at AT after T m '" total investment + AP atl:1T after T m total investment T. ~gure 2.12 Hewlett-Packard's return map (diagram based on House and Price, 1991;Magrab,1997). • The total investment of dollars starting from the first instant (Te) that engi- neers start dreaming up the project (see the top of Figure 2.1 at the beginning of the chapter). •The total sales that begin as soon as possible after the first product is manu- factured and sold, T m' Note that setting up and debugging the manufacturing line generate no sales. •The total profit that starts to be gained at Tb• The key points on the time axis are: • Te~the project initiation point. This is followed by product definition, product development, process planning for manufacturing, setting up machines, debug- ging the assembly line, and first launch into manufactured product around point T m• •Tm-the point where real manufacturing begins and products get sold. • Tb~the break-even lime from the very beginning uf the "conceptual product definition" to the point where a positive profit occurs (T b - T e ), Note that the chart also shows the break-even-after-release time, which measures (T b ~ T m) and focuses more on manufacturing productivity. Obviously, fast production and high volumes of product are desirable. The goal is to quickly amortize all the development costs. • li.T and li.P-any arbitrary point (AT,AP) beyond the break-even-after-release time (Tm)where Hewlett-Packard's return factor (RF) is calculated. The RFis ( Product definitior I Product development Manufacturing and sales Total investment -Break-even time Total operating profit <, -B'reak-evenafter I release 42 Manufacturing Analysis: Some BasicQuestions for a Start-Up Company Chap. 2 calculated by dividing total profits by total investments. The goal is to maxi- mize RF in the shortest time after T m' More importantly, it is also possible to measure an RF' from the break-even time, T e• In terms of profitability for the whole company,break-even time ismore crucial than break-even-after-release time. New companies with limited cash flow should focus more on the former measure. What is the income stream from the product? The following definitions are often used: •Sales price = estimated sales price of one unit from company to distributor (not retail) •Net sales = individual sales price x number of products sold •Cumulative net sales = integrated net sales over several consecutive years What are the costs of being in business and producing that particular product? The following definitions are often used: •Unit cost = prime manufacturing and related manufacturing overhead costs of a single unit of product (see the cost of goods manufactured on the right of Figure 2.5) •Cost of the product = unit cost x number of products sold •Development costs = conceptual and detailed design + launch + support •Marketing costs = a percentage of net sales (Magrab, 1997,uses 13%) •Other promotional and running costs = a percentage of net sales (Magrab, 1997,uses 8%) What is the potential profit or loss? The following definitions are often used: •Gross margin = net sales - cost of product •Percentage gross margin = gross margin I net sales x 100% •Pretax profit = gross margin - development costs - marketing costs - other •Cumulative profit = integrated profits (or losses) on a year-by-year basis Table 2.4 has been reproduced from Magrab (1997) to show some specificfig- ures. In that example, the first two yean; have no sales. However, the design and development costs are running up all the time showing a bottom line, temporary loss of $1.6 million. This particular illustration shows that by the year 2005,the product makes an impressive profit. But the risks of the first two to three years cannot be emphasized enough. And what if the customer does not like the product when itis released to the market? What if the development time is too long and another company launches a similar product first? Or a better product a few weeks later? The risks of a company are far too evident here. Also it is useful to ask, Where will the 1.6 million come from? Obviously from a loan of some kind (new company) or a strategic investment (larger, existing com- pany).At what effective interest rate? 8%? 1O%?12%1 What other products might Year 1997 1998 1999' 2lJlJO 2001 2002 2003 2004 2005 j = 1 j = 2 I = 3 J =4 j =5 j =6 j =7 j = 8 j =9 A Sales price $65.90 $65.90 $67.90 $67.90 $67.90 $67.90 $67.90 B Number of units sold 100,000 250,000 3OO,lJlJO 350,000 250,000 200,000 150,000 C Net sales [=AB] $6,5lXl,000 $16,475,000 $20,370,000 $23,765,000 $16,975,000 $13,580,000 $10,185,000 D Cumulative net sales [=SUMC(j)] $6,590,000 $23,065,000 $43,435,000 $67,200,000 $84,175,000 $97,755,00 $107,940,000 E Unit cost ttarget} $34.00 $33.50 $33.00 $33.00 $33.50 $34.00 $34.50 F Cost of product sold [=BEJ $3,400,000 $8,375,000 $9,900,000 $11,550,000 $8,375,000 $6,800,000 $5,175,000 G Gross margin ($) [=C-F] $3,190,000 $8,100,000 $10,470,000 $12,215,000 $8,600,000 $6,780,000 ~,010,00J H %grossmargin[=IOOGICJ 48.41% 49.17% 51.40% 51.40% 50.66% 49.93% 49.19% I Development cost $8OO,lJlJO $800,000 $4OO,lJlJO $50,000 $50,000 $50,000 $50,lXXl $50,000 $50,000 J Marketing(13%nelsales) [=O.13CJ $856,700 $2,141,750 $2,648,100 $3,089,450 $2,206,750 $1,765,400 $1,324,050 K Other(8%ofnetsales) [=0,08CJ $52'1,200 $1,318,000 $1,629,600 $1,901,200 $1,358,000 $1,086,400 $814,800 L Total operating expense {",I+J+KJ $800,000 sscicco $1,783,900 $3,509,750 $4,327,700 $5,040,650 $3,614,750 $2,901,800 $2,188,850 M Pretax profit [=G-LJ ($800,000) ($800,000) $1,406,100 $4,590,250 $6,142,300 $7,174,350 $4.985,250 $3,878,200 $2,821,150 N %profit[=l00MlCJ 21.34% 27.86% 30,15% 30.19% 29.37% 28.56% 27.70% o Cumulative profit [=SUMMUlJ ($800,000) ($1,600,000) ($193,900) $4.396,350 $10,538,650 $17,713,000 $22,698,250 $26,576,450 $29.397,600 "Product enters market midyear. TABLE 2.4 An Example of Magrab's Baseline Hypothetical Profit Model. (Reprinted with permission from Integrated Product and Process Design by E. 8. Magrab. Copyright CRC Press, Boca Raton, Florida.) [...]... Start-Up Company I Aim: 3. 4 parts per million quoted as Chap 2 00 .1 Process on target 3. 4 pa~ts 1 part -'per milli.onper billion I part 3. 4 parts per~illionpe[/million_ 1. 5" Offset 1, 35 0parts: oer mtnton I 3. 4 parts per million Conclusion' When centered, the 4.50"lines give 3. 4 parts/million on each side When offset 1. 50-,the 6.00-lines give 3. 4 parts/million total Figure2 .15 Motorola's 6-sigma quality... In manufacturing, C should be ~1. 00 for the process capability to be acceptable Again following the example of Devor, Chang, and Sutherland (19 92), consider that the process mean has drifted and is located at 13 0,somewhat away from the nominal of 14 5,with a standard deviation of IJ"x = 10 .The calculations proceed as follows: pk ZUSL ~ 19 0 ~ 13 0 ~6 ZLSL ~ 10 0 ~ 13 0 ~3 Zmm ~ min [16 , 0'(- (- 31 1 1J ~3 c.;... of (USL-LSL) should be wider than the achievable +1- 3rr (or NT = Sc} of the manufacturing process 48 Manufacturing Analysis: Some Basic Questions for a Start-Up Company 10 12 .95 13 .05 13 .00 Diameter of shafts (mm) (a) Chap 2 F'ipre1 . 13 lheupperdiagram(a)is from the manufacturina prouss itself showing data clustered around a mean shaft diameter of 13 mm In standard statistical quality control (SOC)... of spec The Cpk evaluation considers this drift (courtesy of DeVor, Chang, and Sutherland, 19 92) 2.4 Question 3: How Much Quality 51 (Q)? However, if the process mean were recentered at the nominal of 14 5, then: 19 0 - 14 5 ZUSL= I0~ 4.5 10 0 - 14 5 ZLSL= 10 ~ 2m," ~ - 4.5 min[!4.5, 0 '1- (- 4.5)) 11 ~ 4.5 Cpk=) 4.5 ~ 1. 50 The example shows that by recentering the process, the value of increased by 50% 2.4.4... be [12 .950 to 12 .955J, [12 .955 :0 12 .960], and so forth Next, the histogram plots the number of shafts in each band all across to the right side tail • The mean ~X' • If the company accepts all the shafts of diameter within (6u = +1- So}, then the rejection rate will be 27 out of 10 ,000 parts If the company accepted (12 0" = + 1Sc}, then the rejection rate would be 2 out of 1 billion parts The desired... bell-shaped curve Standard SQC uses + 1- 30 ' (namely, i t- 3 standard deviations) This measure is known as Cp: USL - LSL C ~ -6 . $3, 878,200 $2,8 21, 150 N %profit[=l00MlCJ 21. 34 % 27.86% 30 ,15 % 30 .19 % 29 .37 % 28.56% 27.70% o Cumulative profit [=SUMMUlJ ($800,000) ( $1, 600,000) ( $19 3, 900) $4 .39 6 ,35 0 $10 , 538 ,650 $17 , 7 13 ,000 $22,698,250. $6,590,000 $ 23, 065,000 $ 43, 435 ,000 $67,200,000 $84 ,17 5,000 $97,755,00 $10 7,940,000 E Unit cost ttarget} $34 .00 $33 .50 $33 .00 $33 .00 $33 .50 $34 .00 $34 .50 F Cost of product sold [=BEJ $3, 400,000 $8 ,37 5,000 $9,900,000 $11 ,550,000. expense {",I+J+KJ $800,000 sscicco $1, 7 83, 900 $3, 509,750 $4 ,32 7,700 $5,040,650 $3, 614 ,750 $2,9 01, 800 $2 ,18 8,850 M Pretax profit [=G-LJ ($800,000) ($800,000) $1, 406 ,10 0 $4,590,250 $6 ,14 2 ,30 0 $7 ,17 4 ,35 0 $4.985,250 $3, 878,200 $2,8 21, 150 N