IT-390 Preliminary and Detail Methods

1. IT-390 Preliminary and Detail Methods Ch 6

2. Design and Evaluation • Preliminary Estimate requested during the initial evaluation for several reasons: • 1) Investigate the validity of the project • 2) Clarify direction or change direction • 3) To allow proper preparation

3. Design and Evaluation • Preliminary Estimates - limited facts specifics • ROM (Rough Order of Magnitude) • Methods used are: • 1) Rules of thumb (heuristic approach) • 2) Simple calculations

4. Design and Evaluation • Preliminary Estimates • less than accurate • Dollarless in value • No substantial foundation • Invaluable to continuing or terminating the operation, project, product, or system estimate in whole or part.

5. Design and Evaluation • Detail Estimates are normally a "re-estimate” • Re-estimate of the Preliminary Estimate • More discrete information • Increased accuracy • Greater legitimacy of analysis (Mathematical/Factual)

6. Opinion • Opinion - judgment or belief stronger than a perception, but weaker than positive knowledge. • 1) Natural part, but not uneducated ones • 2) Capable, experienced engineer - good working knowledge • 3) Natural/Planned experiences • common sense, judgment, observations, gut feelings, all of which can also be done collectively (with others).

7. Conference • Conference is formulating an educated opinion collectively with "others". • A consensus (agreement) reached • "others" - knowledge in specific areas • engineers only - individual designs (model of estimation) • Downfall - lack of adequate analysis and verifiable facts

8. Comparison • Similar to "Conference" but uses formal "Logic” • Compare to similar item(s)

9. Unit • Most popular • Average • Rough Order of magnitude (ROM) • Lump sum • Module estimating • Examples of unit estimates include: • $/sq. foot (the / means “per”) • $/pound • $/machine shop man hour

10. Learning • Idea - more often an operation is performed, less time will be required to complete it • Repetition results in less time or effort expended • The improved performance is called "Learning"

11. Learning • Identifying Situations for Use of Learning • Not used in all situations • Must have • opportunity for improvement • reduction in labor hours per unit

12. Learning • Use of the improvement curve should be considered in situations where there is: • A high proportion of manual labor • Uninterrupted production • Production of complex items • No major technological change • Continuous pressure to improve

13. Learning • Factors that Support Improvement • Job Familiarization by Workers • Improved Production Procedures • Improved Tooling and Tool Coordination • Improved WorkFlow Organization • Improved Product Producibility • Improved Engineering Support • Improved Parts Support

14. Learning • First application - manufacture of airframes • Costs lowered with increasing quantity of production or experience • Rate of improvement is 20% between doubled quantities • Therefore, learning curve is 80% or, 2nd unit will take 80% labor of what it took on 1st unit

15. Learning • Other names for learning include: • Manufacturing progress function • Experience or dynamic curve

16. Learning • The learning model is based on 3 assumptions: • 1) Amount of time or cost required to complete a unit of product is less each time the task is undertaken • 2) Unit time will decrease at a decreasing rate (in other words, the time reduction will slow as time moves on) • 3) Reduction in unit time follows the model y = axb (curvilinear from Ch 5)

17. Learning • The underlying hypothesis says: "The direct labor man hours necessary to complete a unit of product will decrease by a constant percentage each time the production quantity is doubled."

18. Learning Curve Example

19. Learning • Several equations in this section are useful to us: • 1) Eq. • 2) Eq. 6.9, pg. 259 • 3) Eq. 6.13, pg. 261 • 4) Eq.

20. (Ti,Ni) T Effort. Labor Rate, etc. (Tj,Nj) N Unit # Learning • Where for each, • T = effort per unit of production • N = unit number • S = slope of improvement rate, a constant • K = Constant, for unit 1, dimensions compatible to T •  = the percent learning as a decimal • These all assume that (Ni , Ti) and (Nj , Tj) are two points on a log - log curve (straight line) of "N" vs "T"

21. Learning • Problem 6.27, pg 294, • Supplemental Problem • For the following problems, assume that the unit line is linear. • Find the first unit value when the 100th unit is 60 hours with 81% learning. • Find the value for unit 6 when unit 3 is 1000 hours with 74% learning. • If the unit value at number 1 $2000, then find the unit dollars for units 20 and 40 with learning rates of 93% and 100%. • If the cumulative average time at unit 100 is 100, then find the unit, cumulative, and average time at unit 101 for a learning rate of 92%.

22. Learning • Calculated by regression models since data of many points are required • There are two approaches to learning curves: • 1) The Wright (cumulative average) system • 2) The Boeing (unit) system

23. Learning • Ostwald refers to the Crawford system as the "Boeing" system • In Fig. 6.4 (pg. 258) • Two systems so it's important to know what data has been collected • Is it for unit cost or cumulative average cost? • Each requires a different system and a different set of equations

24. Learning • Wright SystemBoeing System • cumulative average per unitcumulative average • Tá = KNs Ta = KNs /(1+s) • cumulative total effort from unit 1 - Ncumulative total • Tć= KNs+1 Tc = Tu or Tc = (Ta)(N) • unit effortunit effort • Tú = KNs+1 -K(N-1) s+1 Tu = KNs • Notes: Plot of Tá & Tu are assumed to be linear. • The “T” in Tá is for time. For cost we could have used a “C” as in Cá, but for simplicity, we will leave the equations as “T”s even when dealing with cost. N = unit number S = slope of improvement rate, a constant K= constant, for unit one

25. Learning • Values in Appendix 5, pg. 553 • Assumption - "cumulative average" data have been collected and plotted to form the straight line in regression analysis (Wright system)

26. Learning • Example of Wright and Boeing Methods

27. Learning • Slope - determined from historical/predicted from experience • Minimal improvement in "hard" tooling VS manual labor

28. Learning • Learning curve values typical of general industry groups as of 1995 were: • Aerospace 85% • Shipbuilding 80-85% • Complex machine tools for new models 75-85% • Repetitive electronics mfg. 90-95% • Repetitive machining or punch-press oper. 90-95% • Repetitive clerical operations 75-85% • Repetitive welding operations 90% • Construction operations 70-90% • Raw materials 93-96% • Purchased parts 85-88%

29. Learning • Theoretical First Unit (TFU) Cost + an estimate of the learning curve slope = a cost estimate in (Ch 8) • 1) Risk Method - Both the TFU and learning curve slope are estimates • 2) If slope of the curve is off only +/-5%, at the 1000th unit, estimate is off by 68% • 3) The TFU cost is the cost (in dollars or man hours) to produce unit #1 • 4) "theoretical" since rarely will the cost of the first unit produced match this figure • 5) Why? Because we typically build several prototypes before reaching the 1st unit that have the effect of lowering its cost