Automated manufacturing costing system and method5249120Abstract An automated manufacturing cost estimating method and system with indirect cost allocation which identifies a material in an initial state for manufacturing a product in a final state; defines an operation for converting the material from its initial state into the final state of the product; calculates the direct labor cost for each operation based on direct labor hours and calculates the indirect cost specifically allocated for each operation based on consumption of overhead cost elements by the operation; and sums the cost calculated for each operation to determine the total cost of manufacture of a part. Also disclosed is a cost analyzing method and system which utilizes the cost estimating method and system by identifying a plurality of different materials in an initial state for manufacturing a product in a final state and defining a number of manufacturing operations for converting each of the different materials from its initial state into the final state of the product; generating a plurality of paths through the interconnected operations between the different materials in the initial state and the product in the final state; determining a predetermined cost parameter associated with each operation; and accumulating the determined cost parameters along each path. Claims What is claimed is: Description FIELD OF INVENTION
TABLE I
______________________________________
Raw Material Cost ($/lb)
______________________________________
12.increment. Unidirectional Prepreg Tape
$75.00
4.increment. Unidirectional Prepreg Tape
75.00
Unidirectional Prepreg Broadgoods
150.00
Biaxial Broadgoods 140.00
Pre-consolidated flat sheet
357.00
______________________________________
Unidirectional broadgoods consist of unidirectional tape that has been spliced together by the material supplier. Biaxial broadgoods are produced by slitting tape into smaller strips and weaving into a broadgoods fabric. Preconsolidated sheets are laid up and consolidated to use specifications by the supplier and can be shipped like plywood, complete with C-scan inspections. In the case of the latter three materials, the material supplier has added value (hence cost) to the material by providing it in a form which is in some way more desirable to the end user. This desirability usually relates directly to ease of use. Materials that are easier to use reduce the manufacturing labor content and therefore lower labor costs. In the subsequent analysis of program results, tradeoffs between material cost and labor costs and how these affect total fabrication costs are examined for the various materials at different production volumes. The total cost for manufacturing thermoplastic parts in a particular production program is a function of the number of different parts per shipset and the number of shipsets per year, as shown in the following equation:
______________________________________
C.sub.total =
NPV (Cost.sub.-- per.sub.-- partj) *
Shipsets.sub.-- per.sub.-- year) (1)
where:
C.sub.Total =
Total cost over entire production period
i = year (1 to N) of an N year production
period
j = number of discrete parts (1 to M) per
shipset during production period
NPV( ) = Net Present Value Function
Cost.sub.-- per.sub.-- part =
Cost per discrete part in current $
Shipset.sub.-- per.sub.-- year =
Number of Shipsets in a given year i
______________________________________
The total Cost.sub.-- per.sub.-- part for a particular part is calculated using the following equation:
______________________________________
Cost.sub.-- per.sub.-- part = C.sub.E + C.sub.M + C.sub.L + C.sub.OHD +
C.sub.T + C.sub.F (2)
where:
C.sub.E =
Cost of Equipment
C.sub.M =
Cost of Materials
C.sub.L =
Cost of Labor (Based on direct and identifiable
indirect labor hours)
C.sub.OHD =
Cost of Labor Overhead (Overhead Burden Rate and
Benefits Rate charged to all identifiable labor hours)
C.sub.T =
Cost of Tooling
C.sub.F =
Cost of Floorspace
______________________________________
The individual cost components (C.sub.E, C.sub.M, etc.) in Equation 2 are calculated taking into account the considerations in Table 2.
TABLE 2
______________________________________
Considerations in Cost Calculations
Cost Component Considerations
______________________________________
Cost of Equipment, C.sub.E
Initial cost of equipment
Depreciation cost over
anticipated production period as
% of total useful life of
equipment
Cost of power to run equipment.sctn.
Cost of equipment maintenance
required
Cost of equipment downtime
Cycle time per part*
Cost of Materials, C.sub.M
Cost of raw materials (in $/lb.)
Total material in each completed
part
Cumulative scrap rate for total
process used to produce
each part (adjusted for
learning curve effects*)
Cumulative process yield for
process path (adjusted for
learning curve effect*)
Cost of Labor, C.sub.L
Direct labor hours per part
Indentifiable indirect labor hours
per part (supervisors,
programmers, inspectors, etc.)
Labor rates (in $/hr.) .sctn.
Productivity loss due to idle time,
personal time, fatigue
Cumulative process yield for
process path
Adjustment for learning curve
effects*
Cost of Labor Overhead, C.sub.OHD
Direct and identifiable indirect
labor hours per part
Overhead Burden Rate
(including OHD, General and
Admin. Expenses, Support
Services, etc.)*
Benefits Rate
Cost of Tooling, C.sub.T
Tool Material Cost.sctn.
Tool life (in parts made/tool)
Tool Fabrication Time
Mold Material Cost.sctn.
Mold life (in tools/mold)
Mold Fabrication Time
Cleaning Time
Tool Repair Cost
Cost of disposable parts (peel
plies, vacuum bags)
Cost of Floorspace, C.sub.F
Floorspace required
Cost of floorspace (in $/sq. ft.)
Maintenance cost of floorspace
(in $/sq. ft.).sctn.
______________________________________
*Cycle time calculations, overhead burden rates, and learning curve
effects are discussed elsewhere
There is a provision so that delineations can be made for the following
skill levels: skilled, unskilled, supervisor, engineer
.sctn.Adjusted for inflation. In the case of Tool and Mold materials, the
cost of materials such as steel which are expected to follow inflation wa
adjusted while the cost of materials such as graphite/epoxy were not.
How one accounts for manufacturing overhead costs plays a key role in how to go about analyzing and comparing the various manufacturing options available from a cost estimation viewpoint. From a machine designer's point of view, it is important to accurately understand where manufacturing costs originate in order to design cost effective automated systems. In a CD/CE environment, part designers can also benefit from manufacturing cost tradeoff studies since this gives them the opportunity to design the parts with an a priori knowledge of the manufacturing costs involved. For these reasons, it is beneficial to separately identify and consider as many components of the total manufacturing cost as possible. A traditional cost accounting scheme is to separate the cost of a finished part into three items: ##EQU1## The Overhead Application Base can be direct labor hours, direct labor cost, machine hours, material cost, etc. Selection of the base is critical to accurate cost accounting. While labor intensive manufacturing operations are best served with a direct labor overhead base, a Flexible Manufacturing System (FMS) operation might choose a machinery related overhead base. The size of the cost pool to which the Overhead Burden Rate is applied plays a key role in how useful it will be to someone in accurately assessing product costs and in making manufacturing or design strategy decisions. One study cited the fact that 59% of U.S. manufacturing operations surveyed used an individual plant or multiple plants as a cost pool to which they applied their overhead rate. Using this cost accounting method, all facilities, tooling, equipment, and all employees in the entire manufacturing operation who do not charge a direct labor hour to a part must be absorbed in an ever-growing burden rate. For aerospace manufacturers, a set of average values for the Overhead Burden Rate and other components of the total Wrap Rate is given in Table 3.
TABLE 3
__________________________________________________________________________
Average Wrap Rates for a Manufacturing Facility in the Aerospace
Industry
Wrap Rate Cost as % of Cost in $/hr
Cost Component Direct Labor (assuming 22$/hr. Direct
__________________________________________________________________________
Pay)
Direct Labor N.A. $22.00
Overhead Burden Rate 114% $25.00
Facilities/Equipment/Tooling
Capital Investment
Maintenance of Facilities and Equipment
Workers not charging Direct Labor Hours (DLH)
Personnel Dept.
Guards
Expeditors
Part Runners
Industrial Eng.
Programming
Q/A
Low level management
Floor supervision not charging DLH
Other support services
etc.
Fringe Benefits 41% $9.00
Accrual for Holiday/Vacation
Accrual for Retirement/Savings
Life/Health Insurance
Morale/Welfare/Misc.
Gen and Admin. 27% $6.00
Other Charges 14% $3.00
Training
Some other materials
Misc.
Wrap Rate 196% TOTAL $65.00/hr.
Multiplier WRAP RATE
__________________________________________________________________________
Some or all of each of these categories are considered separately in cos
model
The fact that over 50% of the total Wrap Rate cost is "hidden" in one lump sum (the Overhead Burden Rate) makes it very difficult to decompose any cost accounting data represented in this manner. Analysis and comparison of different manufacturing processes is therefore hindered using this accounting method. For this reason, the individual components of the Overhead Burden Rate portion of the total wrap rate listed in Table 3 are separated out. This is why Equation (2) explicitly includes capital investments in Equipment, Floorspace, and Tooling as separate components of the total manufacturing cost. Equipment maintenance, maintenance of the floorspace, and tool cleaning and repair are also considered as subcomponents of each of these new categories, respectively (see Table 2). Also, in computing labor costs, an attempt was made to separately identify as many as possible components of the labor associated with making each part. For this reason, even though machine programming time and supervisory labor might not normally be considered as "direct labor hours" in a traditional sense, they are included as separate items in the "identifiable labor hours" component of Equation (2). Correspondingly, the Overhead Burden Rate used in the cost model was reduced to account for these provisions. Table 4 shows the default Overhead Burden Rate and other components of the total Wrap Rate used in the model. The Overhead Burden Rate is charged against a base of labor hours since most cost components of the Wrap Rate relate to labor.
TABLE 4
__________________________________________________________________________
Wrap Rate Data Used in Thermoplastic Cost Estimator Model
Wrap Rate Cost as % of Cost in $/hr
Cost Component Direct Labor (assuming 22$/hr. Direct
__________________________________________________________________________
Pay)
Direct Labor N.A. $22.00
Overhead Burden Rate 29% $6.00
Workers not charging Direct Labor Hours
Guards
Expeditors
Part Runners
Support Services
Management
etc.
Fringe Benefits 41% $9.00
Gen and Admin. 27% $6.00
Other Charges 14% $3.00
Wrap Rate 111% TOTAL $46.00/hr.
Multiplier WRAP RATE
__________________________________________________________________________
A number of databases contain information which is used as input data for the cost calculations. Tables 5-9 list such data.
TABLE 5
__________________________________________________________________________
Default Equipment Data Used in Cost Study
Power Non-recurring
Recurring
Cost
Req'd
Programming
Setup Setup
Machinery ($)
(KW)
(hrs./discrete part)
(hrs/part)
(hrs/part)
Productivity
__________________________________________________________________________
Tape Layer.sup.1
1.8M
37.5
160 1 1 10 lbs/hr
Press Former.sup.2
150K
54 0 4 1 1 part/hour
(consolidating, 2D to 3D)
Press Former.sup.2
150K
54 0 4 1 1.1 parts/hour
(consolidating, 3D to 3D)
Press Former.sup.2
150K
45 0 4 1 1.4 parts/hour
(non-consolidating)
5 .times. 14 autoclave.sup.3
220K
60 0 0 1 1 part/hour
12 .times. 30 autoclave.sup.3
1.2M
100 0 0 0.75 5 parts/hour
knife cutter.sup.4
120K
34 4 0.25 0.75 800 in/min
ultrasonic cutter.sup.5
380K
45 4 0.25 0.75 200 in/min
__________________________________________________________________________
This maximum flat out rate was adjusted for machine acceleration and
turnaround time
Sources: .sup.1 Cincinatti Millicron; .sup.2 Average values from
Wabash,PHI, OEM; .sup.3 Autoclave Systems, for our part we assumed 5
.times. 14 has 5 part capacity, 12 .times. 30 has 25 part capacity; .sup.
4 Gerber Garment Technologies; .sup.5 American GFM
NOTES: Setup times, programming times, and productivity figures are based
on equipment manufacturer's specifications, and in some cases were
adjusted to match empirical data gathered from industry. Times listed her
are projected as theoretical best values, and will be adjusted for
learning curve effects.
These values were gathered from research publications from industry data, and from communications with composites material suppliers, parts fabricators, equipment designers, and industry consultants. Several data points relating to materials utilization, human productivity, and labor cycle times represent information which is company proprietary or sensitive from a national security standpoint. In the case of company proprietary information, empirical data and parametric models (obtained through plant visits and telephone communications) were averaged to provide a "industry average" value. A number of reports available to qualified government contractors contain parametric models for thermoset fabrication processes which were used when applicable. In addition, empirical data for thermoplastic fabrication is also available. By modifying the `production data` data base, different production scenarios can be evaluated. A production run consists of a some number of parts per shipset and some number of shipsets per year, as represented in the examples in Table 10.
TABLE 10
__________________________________________________________________________
Examples of Production Data Used in Analysis
Parts per
Total #
Total #
Shipsets per
Shipset
Years
Shipsets
Year
__________________________________________________________________________
Production Run 1
10 7 151 [1, 4, 8, 16, 32, 45, 45]
Production Run 2
17 7 1400 [200, 200, 200, 200, 200, 200, 200]
__________________________________________________________________________
Production Run 1 in Table 10 might be a fighter aircraft program beginning in pre-production and ramping up to full scale production. Production Run 2 might be a program which replaces aluminum parts with composite parts on a commercial aircraft. In the Composites Processing Tree in FIG. 5, there is a high level representation of a number of fabrication processes. Each of the "nodes" in FIG. 5 can be treated as a separate operation performed at a discrete station on a factory floor. Therefore, the tree representation is a simplistic way to simulate the flow of materials through a factory. Important factors in this type of simulation are the number of stations required to meet the production schedule, the transfer time between stations, and the actual cycle time (including setup, loading, unloading, etc.) at each station. Note that effects relating to optimal queuing to maximize equipment utilization or to minimize work in process inventory are not included in this model. The minimum number of machine stations required is a function of the total hours available in a working year and the machine hours needed to meet the production schedule, as shown in Equations (5) and (6): ##EQU2## Combining Equations (5) and (6) yields the expression for the minimum number of stations required ##EQU3## Therefore, if there are 4000 labor hours in a year and 5000 machine hours are required at Station X to complete the expected orders for that year, a second machine station must be incorporated. All equipment that is required to meet the expected needs of a given program year is purchased at the beginning of that year and is depreciated for the remainder of the program. Transfer time of work in progress from station to station in the factory was considered as well. Transfer time was nominally set at 15 minutes per transfer, with several exceptions. For example, a station with automated transfer (e.g. layup robot which removes parts from cutter) or a station virtually connected to another station (e.g. a forming press that was linked indirectly to a preheating oven) would have a shorter transfer time, and any transfer to the autoclave would take longer due to the fact that autoclaves are typically located at an end of a facility or in a separate building. Total cycle times at each station included several components, summarized in Equation (8): ##EQU4## Part cycle time was the time spent actually working on the part, which was based on the actual productivity measure for the human or machinery at each station. These are summarized in Tables 5 and 6. Total cycle time also includes recurring and non-recurring setup times for the individual stations. A non-recurring setup time is one that would only occur once for a given run of a discrete part (e.g. a change in tooling) and is therefore charged as a fixed cost which is spread over all parts of the particular run. A recurring setup time, such as loading or unloading a machine or cleaning a tool, is added to the cycle time for any part through the station at any time. Total cycle time was used to calculate direct labor hours and subsequently, the Cost of Labor and Cost of Labor Overhead components of Equation (2). As was mentioned earlier, several other labor cost components were identified as separate items. They include the cost of programming equipment, the cost of supervision on the factory floor, and in-process quality inspection. Programming labor cost was treated as a fixed cost per discrete part, a percentage of which was absorbed in each part made of that type. Programming times are listed in Table 5. An engineering skill level was required for programming. One supervisor was required for every 18 workers to oversee laborers on the factory floor. To account for the in-process inspection, Q/A inspectors were incorporated into the hand layup and bagging stations. An inspector was present during the layup and bagging operation 25% of the time, checking ply orientations and accuracy tolerances on the layup work in progress or checking vacuum seals for leaks, with one exception: in the case of 3D layup of 12 inch tape the inspector was required 50% of the time. Direct labor hours and material scrap rates were adjusted to include learning curve effects. The learning curve is a standard way to account for increases in worker skill and efficiency over time. As a brief example, learning curve effects can be seen in the following phenomena which occur over time as production increases: --reduction of worker cycle times --an increase in overall process efficiency * reduced scrap rates * decrease in transfer time from station to station in a factory The rate an amount of skill increase depends on the task the worker is performing. A fully manual task increases a greater amount at a faster rate than an automated one, as shown in FIG. 6. This is due to the fact that less "learning" occurs when a machine is involved Machine "learning" is a phenomenon which reflects optimization of control programs generated by humans and in human improvement in setup, loading and unloading of the machine. A continuous "two-kneed" learning curve, FIG. 7, illustrates learning curve effects The first knee occurs at full rate production, the point at which maximum learning on a new project or task has occurred. This occurs in the aerospace industry somewhere between 120 units and 200 or 250 units. Between the first knee and the second knee, increases in learning will theoretically occur only through process optimization, and not through discrete human beings learning to perform their jobs better The second knee occurs when learning curve effects level off. This is to provide realism in the illustration; without the second knee the average cycle time per part would approach zero as the number of parts produced continued to increase. Estimates on where the second knee occur range from 500 units to 1200 units. In this study, knees were placed at 120 units and 500 units. These numbers were chosen because they are conservative from a machine designer's point of view in the sense that they make human labor appear in its best light relative to industry data. It was assumed that learning in manual processes increased at a rate of 80% until the first knee, then at 90% until the second knee, where learning levels off. Automated processes increase at 95% and then go to 97% after the first knee. When cost analysis program 10 is used to evaluate the fabrication costs involved in manufacturing a thermoplastic skin 50 of simple curvature, FIG. 4, a large number of process paths are considered, FIG. 5. The result of any given run of the program 10 is a listing of the various components of the total part cost summarized in Equation (2) for all of the process paths considered. By varying input data, the user can generate cost breakdowns which simulate a wide variety of situations: the sensitivity of part cost to part size, unique part count, production ramp up rate, burden rate and other factors are considered. The high and low volume baseline parameters described below in Table 11 are used unless otherwise noted.
TABLE 11
______________________________________
Baseline Parameters for Cost Study
Parameter Value (Units)
______________________________________
Program length 7 years
Unique Parts/Shipset 10
Shipsets/Year "low volume" 10
"high volume"
100
Total Parts/Program
"low volume" 700
"high volume"
7000
Part Dimensions length 2 feet
width 3 feet
thickness 16 plies
Overhead Rate (see Table 3) 70 %
Benefits 40 %
Layup 2D
Cutting Manual
______________________________________
The final step for all of the processing paths represented by FIG. 5 is either autoclave consolidation or press forming with consolidation. First, considering these alternative consolidation techniques from a cost standpoint, FIG. 8, compares the total part cost at the low volume production rate for autoclave consolidation and press forming with consolidation. It is clear that a significant cost penalty is incurred when using an autoclave for consolidation. The cost increase is approximately $1300/part, independent of the material type. The resultant percentage increase varies from 72% for 12 inch tape to 114% for the preconsolidated flat sheet. There are situations where the equipment costs for an autoclave may not be an issue; for example, a company may already have a fully depreciated autoclave from theromset manufacturing operations. Under the assumption that a "no-cost" autoclave is used, the cost penalty for using the autoclave is still $1100/part. This indicates that equipment purchase cost is a small percentage of the overall costs associated with autoclave operation. FIG. 9 is a breakdown of all other costs incurred during the bagging and autoclave process. Tooling contributes approximately $600/part or 61% to overall autoclave costs. The high tool cost is a result of using a graphite tool in autoclave operations. It is assumed that graphite tooling was a necessary result of using a graphite tool in autoclave operations. It is assumed that graphite tooling was necessary since Coefficient of Thermal Expansion (CTE) is a concern when a part must be temperature cycled on a tool. Lower cost tooling materials such as steel and electroformed nickel were also evaluated Even with these low cost tools, the total cost of the autoclave process was still higher than the cost of the press forming process. In a large part, this was due to the cost of bagging ($285/part). This indicates that press forming with consolidation should be employed whenever the desired part geometry can be achieved through pressing Consider next the actual forming operation. In preparation for forming, a part can be laid up in the flat or laid directly on the tooling in near net shape. It is reasonable to assume that the three-dimensional layup rate will be some percentage slower than the two-dimensional rate, since extra care is required to assure accurate fiber placement on a contoured surface. FIG. 10 illustrates this effect; here it was assumed that three-dimensional layup took 33% longer than two-dimensional layup. The cost difference between two-dimensional and three-dimensional layups are accentuated with materials in tape form, which require more layup labor than broad goods. The conclusion is that for the simple curvature skin, the part should be laid up in two-dimensional and formed int net shape It is important to note that for more complex geometries, three-dimensional layup may be a necessity. Since three-dimensional layup is more expensive, it should only be used if the desired part geometry cannot be achieved with two-dimensional layup and press forming. Another advantage of three-dimensional layup is the possibility of reduced forming times. FIG. 11 shows empirical data of temperatures and pressure cycles versus time during the forming of a thermoplastic part. In this case, since forming time is only about 5% of the total cycle time, no significant advantage can be gained from forming three-dimensional laminates as opposed to two-dimensional laminates. In this example, it is thermal inertia (i.e. heating up the oven and laminate that sets the limits on overall forming cycle time. For non-automated processes, four basic types of material were considered: 12" unidirectional tape, biaxial broadgoods, unidirectional broadgoods and the preconsolidated flat sheet. The combination of laying up in two-dimensional (as opposed to three-dimensional) and press forming with consolidation (as opposed to autoclaving) is the most economical way to make the thermoplastic skin. This process path will therefore be used in the material comparison. The total cost and cost mix for parts made from each material is shown in FIG. 12. Equipment and other costs are equivalent in all cases since the same processing path was used. Therefore it is the mix between material and labor costs which determines the total part costs in FIG. 12. This example clearly illustrates how paying different amounts for a raw material can affect downstream manufacturing costs. Although 12-inch tape is the lowest-cost raw material, the high labor content at the 12-inch tape layup station makes this the most costly process overall at $1834/part. All three of the other materials are in broadgoods form, and since broadgood are easier to layup, processing labor is reduced. The least labor intensive of all processing paths uses the preconsolidated flat sheet, which eliminates layup labor all together. Even though the preconsolidated flat sheet is the most expensive raw material, the reduced labor requirement results in the lowest total cost part ($1167). The biaxial material is the lowest cost broadgoods material and results in a part cost of $1247, slightly higher than the preconsolidated flat sheet. Unidirectional broadgoods has a raw material cost approximately equal to the biaxial material. Since it is not interlaced, it requires twice the layup time of the biaxial material (see Table 6). It is evident that labor expense contributes significantly to total part cost when low-cost materials such as twelve-inch tape are used. If automation techniques can be developed to reduce labor content then it is possible that low cost materials can become cost effective. The potential cost saving which can be achieved through implementation of labor-saving automation can be evaluated by considering only the non-labor expenses of a process. By ignoring the labor portion of each cost bar in FIG. 12, it can be seen that twelve-inch tape has the lowest non-labor cost content and consequently stands to benefit mos from labor-reducing automation. The preconsolidated flat sheet with little labor content stands to benefit much less from this type of automation. If cost-effective automation were available which allowed a part manufacturer to work with lower cost materials, the manufacturer could add value to the material in-house instead of paying a material supplier fee plus profit to do it for him. And added advantage of working with materials nearer their raw state is that the part manufacturer gains greater control of final part quality and has more flexibility in tailoring the final properties of the part. In order to target the labor-reducing automation effort, labor content must first be studied. FIG. 13 shows the labor split for cutting, layup and press forming for each material. Buildup labor comprises the largest portion of the labor content for all materials other than the preconsolidated sheet, where all labor is in press forming. The labor percentage attributed to buildup varies from 52% for the biaxial material to 77% for twelve-inch tape. Total labor hours for the buildup process is made up of actual layup, part transfer, recurring and non-recurring setup, quality control and supervision. FIG. 14 shows how buildup is divided among each of these components. An automated layup system has the potential to reduce direct layup labor significantly. Assuming that supervision and quality inspection costs can be reduced through the application of layup automation as well, automation can affect a total of 70% of all buildup labor for the biaxial material, up to 92% for twelve-inch tape. This indicates that automation of actual layup has strong potential for cost reduction for all of the materials being studied except the preconsolidated flat sheet. Although not as potentially significant as buildup labor reduction, labor cuts can also be achieved by automating cutting and press forming. Automated cutters do exist, but little or no automation is available for the press forming process. FIG. 15 illustrates the cost centers in the press forming operation. Tooling, forming and setup labor are the major cost components. Tooling, at 24% of total cost, has always been a significant contributor to total costs, especially when many unique parts are required. When production volumes are high, tool costs can be reduced by increasing tool life. When production volumes are low, a flexible tool that is easily reconfigurable to new geometries would greatly reduce costs. Mold cleaning and setup would be a good candidate task for automation. Based on the conclusions in FIG. 15, which suggest that mold preparation and setup comprise 37% of all press forming costs (regardless of what material is used), these operations could potentially be the next best automation target after layup. Thermoplastic tapelayers are available, and are making parts in a factory environment They are currently capable of achieving 80% and therefore require a subsequent consolidation operation. Since press forming is the most cost effective consolidation operation, this will be used for the tapelayer analysis. FIG. 16 shows total cost versus part volume for an automated tapelayer and the four baseline processes discussed above. The shape of the cost curves for the baseline processes are all dominated by learning curve effects. Consequently the relative cost differences between each of these materials remains fairly constant as volume increases. The shape of the tapelayer cost curve is not dominated by the learning curve but rather by equipment costs which are being amortized over total part volume. As a result, the tapelayer becomes cost effective for volumes over 2800 parts (7700 lbs) per year. The components which drive the shape of the curve are better illustrated by looking at the change in cost mix as production volume changes. FIG. 17A shows a dramatic change in the cost as production volume increases for a part made with a tapelayer. The cost decrease as volume increases is mainly attributable to a large drop in equipment cost on a per part basis. FIG. 17B shows the cost mix at the same volumes for biaxial broadgoods. Here, since equipment cost is a much smaller percentage of total costs, total cost is less sensitive to volume changes. Tapelayer programming labor costs are also sensitive to part volume. Since the programming time for each unique part is spread out over all parts made of that type, increased part volume reduces the per part programming costs. As total volume increases from 700 to 7000 total parts, the portion of labor attributable to programming drops off from 49% to 9.1%, FIG. 18. For this same volume increase, since programming costs are distributed over a greater number of parts, total labor cost per unit drops by 40%. This example illustrates the importance of an efficient and user friendly programming system, especially for high unique part count or low volume. Since cutting comprises a small percentage of overall part cost, FIG. 13, the cost impact of automated cutting is moderate Reciprocating knife and ultrasonic cutters, which cut two to eight times faster than human operators, have long been used with thermosets and can also be applied to thermoplastics. However, for the part considered in this study, very little cutting is actually needed. At the human and machine cutting rates listed in Tables 5 and 6, an entire part can be cut in about 1.5 minutes by the automated cutter and twelve minutes by a human. Most of the cost of cutting is in spreading the material, removing scrap, and kitting the cut part The amount of labor saved by using automated cutters roughly offsets the capital invested in the equipment, FIG. 19. FIG. 19 illustrates that the difference in cost between processes using automated and manual cutting is minimal, especially at higher volumes. A change in layup rates can have an effect on the tapelayer break-even point. Increasing the unidirectional layup rate by 33% and the broadgoods layup rates by 210%, Table 6, yields the total cost curve of FIG. 22. First, the increase in the layup ratio has raised the tapelayer break-even point from 2800 to 3500 total parts. Second, biaxial broadgoods are now the most cost-effective baseline material. This resulted from raising the broadgoods layup rate which reduced labor costs for the biaxial material but not for the preconsolidated sheet process. These results indicate that layup rate can affect process break-even points. The effect of part size on total cost is shown in FIG. 23. As the part size increases, a corresponding increase in unit part cost occurs. Part size increases have the greatest effect on material costs and layup time. As part size is increased from 16 plies to 128, material cost and layup labor increase by a factor of eight. As a result, the processes which have a relatively high combination of material and layup costs, such as twelve-inch tape, are most sensitive to changes in part size. On the other hand, processes such as the automated tapelayer process (which uses the lowest cost material and has the least amount of layup cost) are much less sensitive to part size changes. This indicates that automated systems that reduce labor and utilize low cost raw materials are desirable when part size varies a great deal since they are least sensitive to variances in this parameter. See Table 12.
TABLE 12
______________________________________
Material and Layup Cost Comparison
Layup
Material Labor Material +
Material/process
Cost Cost Layup
______________________________________
12" tape $346 $876 $1222
Preconsolidated sheet
$812 $812 $812
Biaxial broadgoods
$571 $155 $726
Auto Tapelayer 4" tape
$198 $56 $254
______________________________________
Sensitivity to changes in the number of unique parts can be analyzed as well by varying the number of unique parts while holding part size and the total part count constant. In this case, material costs and layup costs will not vary since the total number of parts and the amount of material used in each part is held constant. Instead, variations in total cost will be used by changes in setup, programming and tooling expenses. The tapelayer is the most sensitive to changes in unique part count, FIG. 24. A unique part count is raised from 10 to 100, total cost increases $1622, most of which is attributable to increased programming labor. The autoclave process, which uses expensive graphite tooling, is also highly sensitive to increased part count. This is especially evident when comparing the slope of this curve with the slope of the press forming curves, which all use steel tooling. This analysis indicates two things. First, as was mentioned earlier, tooling is a major cost factor in fabricating thermoplastic parts The is a need for low cost and/or flexible tooling if the number of unique parts is high. Second, automated equipment can be quite sensitive to changes in unique part count. The issue is one of flexibility. Flexibility comes from hardware and from software. We can see in this example that it is software that limits the automated system's flexibility. This indicates that when new automated equipment is designed, careful thought must go into the development of the programming system to insure that the full cost cutting capabilities of the system can be utilized. In all analyses conducted to this point it was assumed that the production volume was constant over the life of the program. In actuality, however, many programs will gradually build up volume to a full production rate. A gradual volume ramp-up can reduce total costs by delaying capital expenses to later years in a program. The two production runs listed in Table 13 will be used to analyze the decrease in unit part cost that occurs when a gradual ramp-up rate is used.
TABLE 13
______________________________________
Production Runs Used in Sensitivity to
Ramp-Up Rate Analysis
Number of Shipsets Per Year
Year Year Year Year Year Year Year
1 2 3 4 5 6 7 Total
______________________________________
Constant
28 28 28 28 28 28 28 196
Rate
Gradual 4 5 8 12 24 60 84 196
Buildup
______________________________________
When switching from the constant to the ramp-up production schedule shown in Table 14, a 20% reduction in material cost occurs. This reduction in cost results from delaying the purchase of material to later years in the program. Since all future purchases are adjusted to account for the time value of money and much of the material is purchased in the final years, a net savings of 20% occurs. In the case of labor costs, in addition to taking into account the time value of labor expenses, labor rates were adjusted for inflation. For this reason, going to the ramp-up schedule only reduced labor costs by 14%. The total effect of the savings generated through purchasing labor and material in the future depends on the ratio of materials cost to labor cost in the process Total decrease in unit cost as a result of gradual volume ramp is shown in Table 15 for the four processes analyzed. Different cost accounting techniques account for overhead costs in different ways. In the traditional scheme, all indirect costs are lumped together into an overhead burden rate that is charged against an application base such as labor. It is important to understand the effect that this accounting technique can have on break-even analysis which may be used to make a go/no go buy decision for automated equipment. FIG. 23 shows total cost for the baseline processes and the automated tapelayer based on traditional accounting procedures. Direct labor was selected as the overhead application base and a $65 wrap rate was assumed, Table 3. What is immediately noticeable is that all cost curves have taken on the characteristic shape of the labor learning curve. The shape of equipment intensive processes can no longer be distinguished from the shape of processes which are labor intensive. When indirect costs such as equipment were identified separately the shape of the tapelayer cost curve was dominated by the amortized equipment cost, FIG. 15. As a result a clean break-even point could be found above which the tapelayer is the most cost effective processing choice. Using traditional accounting practice the tapelayer is economical regardless of production volume. To account for the purchase of a $1.8 million piece of equipment, the burden rate could be adjusted. However, burden rate adjustments would only shift the cost curve up or down without modifying its characteristic shape. As a result, the true cost characteristics of various processing options are not revealed. This analysis indicates that traditional accounting methods which lump indirect costs into an overhead rate can lead to incorrect decisions when trying to justify new equipment expenditures Traditional techniques should be replaced with methods that isolate cost centers and allocate costs appropriately. Improved methods such as these can then serve as better guidelines for future expenditures as well as research efforts into automated system development. At each step 24-36, FIG. 2, in accordance with the novel approach of this invention, a cost estimation is done which takes into consideration the unbundled elements of the cost including labor including indirect and direct labor and learning curve effects, tooling costs, material costs, equipment costs, and the like. At each step a complete workup is done by the system 10 using process knowledge module 14, process path generator module 20, process cost account module 18, and the input database 12, as shown in the flow chart of FIGS. 24A, B and C. Initially, in step 150 the part geometry including length, width, ply, thickness, number of plies, part volume, surface area, perimeter, are called from the database. Then the number of production years is called in step 152 and the cycle time is computed in step 154. The cycle time is equal to the transfer time plus recurring setup time plus nonrecurring setup time divided by the production volume plus the process rate divided by the process yield. After this the cycle time is adjusted for productivity loss in step 156 and then the learning curve calculations are computed in step 158 using the equations and parameters shown. The total cycle time is then computed in step 160 from the adjusted cycle time and the learning curve effect coefficient developed in step 158. After this the labor hours are calculated, including both the touch such as unskilled and skilled labor, and the non-touch, such as engineer and supervisory labor, in step 162. The number of stations is then calculated in step 164 by multiplying the INT times the adjusted cycle time divided by work hours per year multiplied by the production volume plus 1. In step 166, the maintenance cost is calculated by multiplying initial cost times maintenance cost percent, depreciation costs are calculated by multiplying initial cost times one minus the total number of years times the useful life, all of which is divided by the total number of years. Finally, the power cost is determined from the cycle time multiplied by the power required times the electricity cost times an inflation factor. In step 168, the equipment costs are calculated as shown followed by the calculation in step 170 of the material used and the adjusted cumulative scrap factor. Material costs are determined in step 172 for the material used, the adjusted cumulative scrap factor and raw material costs. The elemental labor costs are calculated in step 174, the facilities costs in step 176, and the full labor costs in step 178. The costs of the number of tools required is calculated in step 180 from the number of stations multiplied by the number of tools per station, and the tooling costs are calculated in step 182 from the number of tools required and the cost of tools in conjunction with the production volume, disposable tooling costs, tool repair costs and inflation factors. Finally, the total cost per part is calculated in step 184 from the equipment costs, material costs, labor costs, tooling costs and the facility costs. Thus the specific elements of these costs in step 184 are calculated for each step in the process so that there are no hidden or bundled overhead burdens which will distort the true costs of a part or a particular operation in the making of a part. An automated manufacturing cost analyzing system for finding the best path through the processing tree 14, FIG. 5, begins with defining each node consisting of an operation, its input and output states in step 200, FIG. 25. Then the determination is made of the initial states and the final operation states in step 202. The tree logic algorithm is then executed in step 204 backwardly through the nodes beginning with the final state and ending with the initial, state. This is done in order to minimize the combinatorial explosion that occurs when more than a few initial states and multiple interconnected paths are possible. Because the program stores partially completed paths as it branches backwards from the final state to the initial state, the total number of steps needed to find all paths in the tree is dramatically reduced since the program does not need to rediscover connections it already found. The execution begins in step 204 with the finding of the next previous node going backward from the final state toward the initial state. Inquiry is then made in step 206 as to whether this node is in a unique path, that is, one that has not been traveled before by the system in this cycle of operation. If the answer is yes, the system goes to that node in step 208 and calls the data for that node in step 210 and calculates the cost for that node and stores it in step 212. Inquiry is then made in step 214 as to whether this present node contains an initial state. If the answer is yes, then the total cost for all the nodes on the path are totalled in step 216, and the system then goes back to the last previous node in step 218 to see if there is another branch to be explored. This cycles the system back to step 204 to find the next previous node. When again the inquiry is made as to whether this node is in a unique path, if the answer is no then in step 220 inquiry is made as to whether there are any other possible paths. If there are, the system cycles back to step 204. If there are not, further inquiry is made as to whether this node contains a final state. If it does, the system simply exits the routine in step 222. If this node does not contain a final state, then in step 224 the system backs up to a node with the same final state and returns to step 204 to find another unexplored branch. In step 214, if in response to the inquiry as to whether the present node contains an initial state the answer is negative, the system recycles back to step 204 to attempt to find the next previous node. In this manner the system most efficiently steps through all the possible paths and in each path, as indicated with respect to FIG. 2, a cost estimation calculation is made at each step in accordance with the sequence shown in FIGS. 24A, B and C. That is, the sequence of FIGS. 24A, B and C is applied to each step in the sequence shown in FIG. 2, and the routine in FIG. 25 generates a plurality of sequences similar to that in FIG. 2 as indicated by the multiple paths depicted in FIG. 5. Thus not only does this invention provide a system and method for automatically generating reliable cost estimation for automated manufacturing which exposes all the true costs allocated to the proper elements, but it can also automatically analyze the multiplicity of manufacturing paths available by which a part can be fabricated to reveal which is truly the least expensive path. Although the example herein deals with fabrication of a part made from composite materials, the invention is equally applicable to the manufacture of any part in accordance with the same inventive principles. Although specific features of the invention are shown in some drawings and not others, this is for convenience only as each feature may be combined with any or all of the other features in accordance with the invention. Other embodiments will occur to those skilled in the art and are within the following claims:
|
Same subclass Same class Consider this |
||||||||||