estimating the costs of extremely large telescopes · • facility-class instruments for 8-10 m...
TRANSCRIPT
![Page 1: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/1.jpg)
4840-72 1LMS, 8/27/02
Estimating the Costsof
Extremely Large Telescopes
Larry Stepp, Larry Daggert and Paul Gillett
![Page 2: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/2.jpg)
Scaling Laws
Aperture (m)1 10 100
Proj
ect C
ost
106
107
108
109
105
D2.
7
0.1
![Page 3: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/3.jpg)
Scaling Laws
Aperture (m)1 10 100
Proj
ect C
ost
106
107
108
109
105
D2.
7
0.1
$10.6 M -- 1973$33.7 M -- 1992
$110 M -- 1992
10m ~ $400 M$141 M -- 2002
$2.5 B ~ 2002
350 tonnes 270 tonnes
![Page 4: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/4.jpg)
4840-72 4LMS, 8/27/02
Life-cycle Costs
Typical yearly expenditures of large observatories:– Operations: 3% - 6% of construction cost– Upgrades: 3% - 5% of construction cost
• New instruments• Adaptive optics• Etc.
Over ~ 30-year lifetime, = 2 to 3 times construction cost
![Page 5: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/5.jpg)
4840-72 5LMS, 8/27/02
Cost Control
Control of life-cycle cost is the most important factor in success of an ELT – But, we need to control costs in a manner that maximizes
scientific performance– Best way to do this is Design to Cost
![Page 6: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/6.jpg)
4840-72 6LMS, 8/27/02
Design to Cost
• Cost is considered a design variable in the trade space that includes performance and schedule– “Cost As an Independent Variable” (CAIV)
• Value Engineering– Discipline of reducing LCC while preserving essential performance
• Emphasize primary functions • Consider many alternative designs• Simplify designs and operational modes
– Requires understanding of what constitutes value to the customer
![Page 7: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/7.jpg)
4840-72 7LMS, 8/27/02
Quantitative Measures of Performance and Cost
1. Need to quantify science value:– Scientific merit functions
2. Need to predict telescope performance in presence of disturbances:– Integrated modeling
3. Need way to predict cost of multiple design options:– ?
![Page 8: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/8.jpg)
4840-72 8LMS, 8/27/02
Scientific Merit Functions
• Traditional approach to defining science requirements:– Performance requirements for each observing mode are set in
wish-list manner– Requirements are collected from all the observing modes– Specifications are based on most demanding requirements
• This is essentially a “contractual” approach; descoping takes semi-formal approval
• This doesn’t lend itself to the performance/cost trades of Design to Cost approach
![Page 9: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/9.jpg)
4840-72 9LMS, 8/27/02
Scientific Merit Functions
• To evaluate design trades in terms of performance/cost requires a quantitative expression of scientific value
• For each observing mode / instrument capability define a scientific merit function (SMF), based on:– SNR– Multiplexing– Timeliness– etc.
• Scientific value is summation of SMFs x weighting functions x time allocation:
Scientific value = ΣWi * Ti * SMFi
![Page 10: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/10.jpg)
4840-72 10LMS, 8/27/02
Integrated Modeling
• The distinguishing feature of “integrated” modeling is that it incorporates active control– Can include:
• Structural models• Optical models• Thermal models• Fluid dynamics models
– Incorporates:• Actuators• Feedback sensors• Disturbances
![Page 11: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/11.jpg)
4840-72 11LMS, 8/27/02
Integrated Modeling
• Used to predict the performance of a given design in the presence of disturbances– Performance can be evaluated statistically, in terms of
probability of occurrence• Each SMF is evaluated based on the calculated system
performance• The effect of a design trade on the SMFs can then be
determined
![Page 12: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/12.jpg)
Applying Design to Cost
Need to implement Design to Cost as early as possible
![Page 13: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/13.jpg)
Applying Design to Cost
Need to implement Design to Cost as early as possible
Adapted from presentation by Joseph Hamaker
Cost Incurred
Knowledge
Potential for Cost Change
Life-cycle Cost Committed
![Page 14: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/14.jpg)
Applying Design to Cost
Need to implement Design to Cost as early as possible
Adapted from presentation by Joseph Hamaker
Cost Incurred
Knowledge
Potential for Cost Change
Life-cycle Cost Committed
Applying Design to Cost requires more effort in the conceptual design phase (typically ~ 20%) but since D&D phase is typically ~ 5% of LCC, the increase in total cost is ~ 1%, whilethe savings can be large
![Page 15: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/15.jpg)
4840-72 15LMS, 8/27/02
Dilemma:
How do you make key Design-to-Cost decisions early in the project, before you have detailed designs on which to base cost estimates?
![Page 16: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/16.jpg)
4840-72 16LMS, 8/27/02
Cost-Estimating Methods
Top-down methods Bottom-up methods• Expert opinion
• Analogy
• Parametric
• Also called a WAG
• Used when no other method is available
• Credibility depends on experience and independence of expert
• Industrial Engineering
• Parametric
![Page 17: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/17.jpg)
4840-72 17LMS, 8/27/02
Cost-Estimating Methods
Top-down methods Bottom-up methods• Expert opinion • Industrial Engineering
• Analogy
• Parametric
• Based on actual cost of similar programs
• Usually require some adjustment
• Accuracy depends on similarity between programs
• Parametric
![Page 18: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/18.jpg)
4840-72 18LMS, 8/27/02
Cost-Estimating Methods
Top-down methods Bottom-up methods• Expert opinion • Industrial Engineering
• Analogy • Parametric
• Parametric
• Based on historical data
• Relationships between cost and measurable parameters are determined by regression analysis
• Can estimate project cost based on key parameters
![Page 19: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/19.jpg)
4840-72 19LMS, 8/27/02
Cost-Estimating Methods
Top-down methods Bottom-up methods• Industrial Engineering
• Parametric
• Based on detailed design:
• Catalog prices
• Vendor quotes
• Material costs
• Accurate but time-consuming
• Expert opinion
• Analogy
• Parametric
![Page 20: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/20.jpg)
4840-72 20LMS, 8/27/02
Cost-Estimating Methods
Top-down methods Bottom-up methods• Expert opinion • Industrial Engineering
• Parametric
• Can also be used to estimate cost of main component systems
• Well suited to evaluate effects of design changes
• Analogy
• Parametric
![Page 21: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/21.jpg)
4840-72 21LMS, 8/27/02
Instruments
• Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study:– Average cost overrun 100%– Average schedule overrun 160%
• Instruments 10% - 20% of telescope construction costs• Instrument upgrades ~ 3% of TCC per year• Need improved project management techniques
– Instrument-building organizations are collecting historical data to improve cost estimating
Paper describes instrument cost-estimating lessons learned and bottom-up parametric estimating methods
![Page 22: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/22.jpg)
4840-72 22LMS, 8/27/02
Quantitative Measures of Performance and Cost
1. Need to quantify science value:– Scientific merit functions
2. Need to predict telescope performance in presence of disturbances:– Integrated modeling
3. Need way to predict cost of multiple design options:– Parametric Cost EstimatingTo assess the performance and cost impact of the design
trades without delaying the project, these analytical tools need to be developed in advance.
![Page 23: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/23.jpg)
4840-72 23LMS, 8/27/02
Cost-Estimating Capability
• Design to Cost must be a key part of systems engineering for any ELT project
• We are entering a new regime, where auditable cost estimates will be required to attract funding– Aerospace companies, DoD and NASA have full-time cost
estimating staffs– Ground-based astronomy has none
• We need to develop our cost estimating capability– Train / recruit staff– Collect historical data– Develop CERs
![Page 24: Estimating the Costs of Extremely Large Telescopes · • Facility-class instruments for 8-10 m telescopes have had problems with cost and schedule -- Adrian Russell study: – Average](https://reader033.vdocuments.net/reader033/viewer/2022042807/5f7d8cf6b050d03cb12cf2ff/html5/thumbnails/24.jpg)
4840-72 24LMS, 8/27/02
Collaboration
This is an excellent area for community-wide collaboration.