NASA Instrument Cost Model NICM Presented to the NASA Cost Symposium

1
NASA Instrument Cost Model(NICM)Presented to
the NASA Cost Symposium

• Hamid Habib-Agahi
• Eric Kwan
• April 13, 2005

2
Objective
Develop NASA Instrument Cost Model (NICM) at the
System Subsystem Levels
Approach
CER Verification Validation
User Req Development
Database Development
Data Collection
Model Development
Data Verification Validation
Model Maintenance Updates
3
High Level Schedule
4
User Requirement Development
CER Verification Validation
User Req Development
Database Development
Data Collection
Model Development
Data Verification Validation
Model Maintenance Updates
5
User Requirements Development

• Formed core instrument development team
• JPL Section 312 - Systems Analysis Model
  Development Group
• Technical division experts (instruments)
• Users (Team X, Team I, Cost Office, Div. 38)
• JPL Procurement/Contract department
• Non-JPL consultants
• Analyzed current available instrument cost models
• Solicited capability briefing from experienced
  industry Systems Engineering/Technical Assistant
  (SETA) contractors
• Developed a comprehensive NICM instrument list
• Included all instrument built from 1985 to 2004
• Grouped by JPL vs. non-JPL led instruments
• Total of 120 instruments

6
Current Instrument Cost Models
7
NICM Instrument List (1985 2004)By Lead
Organization
Note JPL Led 42 Non-JPL Led 58 No DoD
Instruments
8
User Requirements Development

• Categorized the instruments by class
• Optical
• Active Micro/Sub-millimeter Wave
• Passive Micro/Sub-millimeter Wave
• Particle Fields
• Others/Radio Science
• Developed standard instrument WBS
• Identified cost and technical parameters by
  subsystem
• Developed a data collection questionnaire to
  capture required cost and technical parameters

9
Data Collection Questionnaire (1 of 3)
Cost Modeling Attributes
10
Data Collection Questionnaire (2 of 3)
Cost Modeling Attributes
11
Data Collection Questionnaire (3 of 3)
Comments Assumptions Review with Dave Swenson
(Jan 13, 2005) 1) Management and Systems Eng
costs were costed together at 117,542.
D.Swenson estimates 27M for Management, 13M
Prod Assurance, the rest for Systems Eng. 2)
Design Inheritance New 3) Mission Class B 4)
TRLs Optics-5, Mech-6, Thermal-5, Electronics-5,
Detectors-5 5) Detector mass is empty because
it's so light, only a few ounces. Cost
breakdown and TRLs from ESSA database. See chart
below. NASA New Start Inflation Index (Actuals
Thru 1998) From http//modis.gsfc.nasa.gov/about/
design.html 2) 4 bands defined The optical
system consists of a two-mirror off-axis afocal
telescope, which directs energy to four
refractive objective assemblies one for each of
the VIS, NIR, SWIR/MWIR and LWIR spectral regions
to cover a total spectral range of 0.4 to 14.4
µm. 3) The combined weight of the Optical Bench
Assembly is 22 kg. Scan mirror is 4.3 kg,
requires 1.9 W power 4) Each MODIS instrument has
three doors that protect the internal
components from contamination, damage, and in
some cases help the instruments self-calibration
processes. NAD (Nadir looking door) is 3 kg, SVD
(Space Viewing Door) is 7 kg, and SDD (Solar
Diffuser Door) is 1.9 kg. 5) Mainframe, main
structural component, supports the internal
instrument components. Mainframe is 43 kg. 6)
Passive Radiative Cooler is 11 kg 7) Optical
System includes Afocal telescope, scan mirror,
fold mirror, afocal Gregorian telescope, field
stop, dichroic housing, secondary mirro, 3
dichroic beamsplitters, 4 FPAs 8) Design life,
mass, data rate from this website. From MODIS
Technical Description Document (Final), August
1997, prepared by Hughes - Optics Scan mirror
assembly (10.9 kg), Afocal telescope assembly
(11.5 kg), Aft-optics assembly (10.3 kg), -
Mechanical Mainframe (43.8 kg), Doors (12.9
kg) - Thermal Radiative cooler (10.0 kg),
Sunshade (0.8 kg), Blankets (4.1 kg) -
Electronics Main Electronics Module (61.7 kg),
Analog Electronics Modules (28.0 kg), Solar
diffuser monitor (1.5 kg), Spectral response,
radiometric calibration, band-to band
registration (12.4 kg), solar diffuser (1.5 kg),
Blackbody (5.6 kg) - Other Mischellaneous (13.2
kg) - TOTAL 228.7 kg - POWER MEM (70W), AEMs
(60W), Scan Motor (3 W), SRCA (44 W). Average
177W, Peak 205 W. (G.Chen subsystem power is
listed for average power, use their ratios for
peak power situation). - 15 year mission
duration, G.Chen 180 month design life - 10.5
Mbps (day mode), 3.2 Mbps (night mode)
12
Data Collection Questionnaire - Block Diagram
Radar Launched in Oct 97
ASI
Dish Antenna w/ Feeds
JPL
JPL also led Management Systems Engineering
13
Data Collection
CER Verification Validation
User Req Development
Database Development
Data Collection
Model Development
Data Verification Validation
Model Maintenance Updates
14
NICM Data Collection Activities
Task Start Oct 2003 Requirements
Development/Data Collection
Non-JPL Instruments
Statement of Work/RFP Development
Instrument Contract April 04
JPL
Lessons-Learned from JPL Instrument Data
Collection
Mar 05
FY05 Goal
Feb 04
Mar 04
JPL Instruments
Pathfinder test case
15
Data Collection Status

• Collected 81 instruments at system and
  subsystem levels
• Questionnaires were stored electronically in an
  Excel-based database
• Provided quick search and comparisons of the
  raw data
• The raw data were further reviewed and
  normalized by JPL technical experts in the
  related fields
• Captured outliers and anomalies
• 30 reviewed and normalized
• Lessons-learned for Non-JPL data
• Availabilities of instrument cost data at
  subsystem level
• Especially instruments built by the Universities
• Issues of Non Disclosure Agreements (NDA) among
  NASA centers
• Full cost accounting policies and practices were
  not consistent
• Aerospace Survey indicated lack of support from
  non-JPL institutions

16
NICM Database Summary
Total instruments 81
17
Model Development
CER Verification Validation
User Req Development
Database Development
Data Collection
Model Development
Data Verification Validation
Model Maintenance Updates
18
Initial Cost Modeling

• Apply modern data mining and econometric
  techniques to develop and validate CER
• Cluster Analysis (CA)
• CA is a multivariate statistical method to
  partition a population into homogeneous subsets
  by grouping cases based on similarities within a
  database of descriptive parameters.
• Principal Component Analysis (PCA)
• PCA is a multivariate statistical technique for
  uncovering quantitative relationships between
  multiple attributes of observational/experimental
  units.

19
Cluster Analysis - Optical Instruments

• Derived from simple distance metric (Euclidean
  metric of log (attribute) differences) of
• 24 programmatic and technical variables
• 15 optical instruments.
• Euclidean Metric
• D(i,j) (Sk(ln(Aik/Ajk))2)1/2
• Where Aik is the value of attribute k for
  instrument i.
• A table of distances is created between all pairs
  of instruments
• Complete Clustering is then performed using the
  table.
• Output results in an unrooted tree.

20
Example - Cluster Tree
21
Principal Component Analysis (PCA)

• From the PCA mathematical equations, we can
  express Cost as a Product of Factors
• C C0 Pk(Xk / X0k)ek
• Where Xk is the value of instrument attribute k.
• C0 and X0k are the reference values of cost and
  instrument attribute k.
• The ek are parameters derived from the PCA

22
Examples PCA Cost Functional Form
23
Next Steps (FY05 FY06)

• Continue data collection, evaluation
  normalization
• Develop, verify and validate CERs
• Develop the instrument cost model architecture
• Validate and test the cost model
• Integrate NICM to NAFCOM
• Documentation
• Maintenance/Updates