1. Estimating – Methods and Practise
A discussion paper
Ó Galorath Incorporated 2003

2. es·ti·mate (es′ti mit), n.
Estimate defined
es·ti·mate (es′ti mit), n.
an approximate judgment or calculation, as of the value or amount of something
a prediction that is equally likely to be above or below the actual result (Tom DeMarco)
Galorath Inc. 2003
All Rights Reserved

3. Estimating – why ? Conceptual design Which way
Feature / function implications
Budget setting
Feature / function trade offs
Bid no / bid evaluation
System / assembly level
Trade studies
What if
Detail design
Target cost modelling
Design to cost
Value analysis
Part level
Should cost models
Supplier cost modelling
Make buy decisions
Process selection
Material implications

4. The estimating environment

5. Time available to generate the estimate
Estimate continuum
Assumptions
High
Low
Domain experience
Low
High
Time available to generate the estimate
Low
High

6. Rough Order of Magnitude Percentage expected error
Classes of estimates 20 40 60 80
100
-20
-40
-60
Class 1
Class 2
Class 3
Class 4
Class 5
Rough Order of Magnitude
Feasibility studies
Preliminary estimate
Definitive estimate
Detailed Estimates
Worst range of expected accuracy
Best range of expected accuracy
Project Phases
Percentage expected error
Calendar Time (No Scale) A B C D E F
Source: WOODWARD, C. & CHEN, M. Cost Estimating Basics. Skills and Knowledge of Cost Engineering, 4th edition, 1999.

7. Information availability
Part Knowledge
Estimate assumptions
Conceptual design
Detail design
Part level

8. Effort vs. accuracy

9. Classes of estimates Class 1 0 - 20% Concept Screening 1 Class 2
Primary Characteristic
Secondary Characteristic
LEVEL OF PROJECT DEFINITION
Expressed as % of complete definition
END USAGE
Typical purpose of estimate
METHODOLOGY
Typical estimating method
EXPECTED ACCURACY RANGE
Typical variation in low and high ranges
PREPARATION EFFORT
Typical degree of effort relative to least cost index
ESTIMATE CLASS
Class 1
0 - 20%
Concept Screening
Parametric Models, Judgment, or Analogy
Low = -20 to -50%
High = +30 to +100% 1
Class 2
1 - 15%
Study or feasibility
Equipment factored or Parametric Models
Low = -15 to -30%
High = +20 to +50%
2 - 4
Class 3 %
Budget, Authorisation control
Semi-detailed unit cost with assembly level line items
Low = -10 to -20%
High = +10 to +30%
3 - 10
Class 4 %
Control or Bid/Tender
Detailed Unit costs
Low = -5 to -15%
High = +5 to +20%
4 - 20
Class 5 %
Check Estimate
Detailed Unit costs
Low = -3 to -10%
High = +3 to +15%
Source: WOODWARD, C. & CHEN, M. Cost Estimating Basics. Skills and Knowledge of Cost Engineering, 4th edition, 1999.

10. How do we estimate

11. Types of estimate Domain experience driven Guess Comparison
Commodity parametric
Domain value
General parametric
Macro level
Process level
Feature based
Generative
Variant process
Generic plan with variables
Measured
Time study
MTM

12. Guess the time - 1
20mm
10.00mm
32mm
16mm
Material is Aluminium

13. Guess the time - 1 - Results
Time dependent on domain knowledge
Accuracy
Volume
Process
Time dependant on level of detail
Does this change your estimate?
Add some more information

14. Guess the time - 2

15. Comparison
20mm
10.00mm
32mm
16mm
20mm
10.00mm
100mm
16mm

16. Commodity Parametric What is it
Cost estimate relationship built for a specific commodity within a specific industrial instance
What's it based on
Current supply costs and trends
Common part attributes
Lots of assumptions
Example
Need to estimate the cost of a die casting for use within the aerospace industry
Review history of cost against parts
Plot number of features, weight, accuracy, volume, application against cost
Look for correlation of key cost drivers
Derive CER
Test CER

17. Macro Level Parametric Estimating
Little knowledge of details / high assumptions
Estimates based on high level information
Weight
Boards
Complexity
Quicker than manual methods
Able to estimate without cost data
Should be calibrated to local environment
Can/Should include Development, Production, Logistics, Operations, & Support Costs all in one model
Should include sensitivity and risk analysis

18. Macro level parameter examples
Electronics circuitry can be accurately described
- Number of Printed Circuit Boards - Number of Discretes per PCB - Operating Environment
- Circuitry Composition - Number of Integrated Circuits per PCB - IC Technology
- Packaging Density - Number of I/O Pins per PCB - Fault Isolation
Electronic Classification - Clock Speed (Frequency) - Fault Detection
Note: Weight to board conversion available for those dealing with weight statements only
Mechanical subsystem aspects tailor estimate to user situation
- Weight - Material Composition - Hardware Classification
- Volume - Operating Environment - Internal Pressure
- Complexity of Form - Construction Process - Operating Service Life
- Complexity of Fit
Program attributes are easily defined (for both Electronics & Mechanical)
- New Design - Certification Level - Dev/Prod Tools & Practices
- Design Replication - Hardware Integration Level - Production Qty’s Prototypes
- Requirements Volatility - Dev/Prod Experience & Capability - Purchased Parts
- Schedule - Labor Rates - Wraps & Fees
G A SEER Technologies 16

19. Process based parametric estimating
Based on mathematically derived CER’s
Estimates based on generic manufacturing details
Production methods evaluated
Should be calibrated to local environment
Includes sensitivity and risk analysis
Should produce an acceptable range for the items / assembly
Process knowledge but no time
Good part data available but no time
Need to run multiple trade studies

20. Generative estimating
Deterministic
Base on formulas
Detailed process plan
Speeds
Feeds
Precise removal rates
Scrap rates
Virtual factory model for suppliers
Can add new process models
Tends to be in-house verified data

21. Parametric vs. Generative - 1
Parametric Benefits
Speed
Level of data required
Learning curves
Design as well as production
Operation and support costs
Three value input indicates level of uncertainty
Generative Benefits
Detail
Accuracy
Flexibility
“open” data source

22. Expressing Uncertainty
Estimates of Size and Technology expressed as single point values don’t tell the whole story:
How confident am I in this value; i.e., what is the probability of not exceeding this value?
How certain am I in this value; i.e., how wide is the probability distribution?
Three-point estimates are better:
LEAST: 1% Probability; “I can’t imagine the result being any smaller than this.”
LIKELY: Best Guess; “If I were forced to pick one value, this would be it.”
MOST: 99% Probability; “I can’t imagine the result being any larger than this.”
Pert mean
Galorath Inc. 2003
All Rights Reserved

23. Parametric vs. Generative - 2
Problems with Parametric
Too generic
Need experience to understand results
“black box”
Too Good to be True!
Problems with Generative
Most data is at T250 + and may be unknown or differ between process types
Hard to determine risk as mono input
Typical systems have no learning curves
Takes a long time to build and maintain the system

24. What is learning
Simply the effect that experience with a process has on the time taken to complete the process
Two main types
Unit (Crawford)
Cumulative Average (Wright)
Effects low unit volumes and manual work more than automated processes
Hand lay-up
Complex assembly

25. Learning Curves Time Units T1 = 388 Represents Learning with 95% slope
400
B Learning Step 2
A Learning Step 1
End Production
1000
T1B = 388
Represents Learning with 85% slope

26. Risk for short programs
If your project runs at lower rates than your data generated from you could risk losing money as the learning curve is not taken into account
Opportunities for reporting real cost reduction via process improvements are lost

27. What should you be using?

28. Use several estimating tools
Macro level
Risk
Cost
Features / functions
Conceptual design data
Low data requirements
Process based
High risk elements
Detailed design data
Should cost models
Purchase target range
Process variants
Quick
High volume
Generative
Models
Parts outside expected range
Detailed supplier data
Supplier cost models
Accepted new base line for product type
Slow
Low volume

29. Estimating solution overlap
Macro Level
Over-lap – large sub-systems, single component costing
Overlap allows for calibration and sanity checks
Process Level
Overlap – mid value, low volume, spares, tooling estimates.
Generative Models

30. Estimating toolbox for the integrated enterprise
Model at process level
Set reference
Output target cost range
Macro level model
baseline
Assess risk
Detail Level
Apply supplier
cost model
negotiations
negotiations

31. Benefits of multiple tool approach
Use appropriate estimating technology at each stage of the product life cycle
Top-down Parametric tool can be used with the minimum of process knowledge
Bottom-up parametric tool allows fast accurate ranges to be established for family groups
Generative modelling will establish base lines for supplier modelling
Pyramid approach supports ALL the company cost engineering needs
Use for sanity check and calibration between models
Increased confidence and overall capability

32. Estimates are always wrong!
Last but not least
Remember
No matter how long you spend
How much you discuss with your colleagues
Who you involve
How experienced you are
Estimates are always wrong!
Our task is to understand how wrong and to make sure our organisation is wise to the risks and assumptions