Life Cycle Analyses of Manned Exploration Architectures

1
Life Cycle Analyses of Manned Exploration
Architectures

• J.D. Reeves
• National Institute of Aerospace
• Exploration Concepts Branch (ECB)
• NASA Langley Research Center, Hampton, VA

2
Traditional Role of Life Cycle Analysis

• Performance and schedule constraints have
  traditionally dominated conceptual design
• Operations, reliability, and maintainability are
  usually at the end of design phase
• Design changes later in design phase can have
  significant effects on both life cycle and
  development costs
• Life cycle considerations need to be made
  concurrent with performance and schedule they
  are just as important in terms of affordability!

3
Overview

• Concept and Life Cycle Analysis
• Life Cycle Analysis Tools
• Recent Life Cycle Analysis studies
• Example 1 Affordability study of a lunar
  architecture (PCAT)
• Example 2 Modularity study of a lunar
  architecture using Discrete Event Simulation
  (study for AIAA Space 2005 conference)
• Lessons Learned discussion of significant
  discipline drivers

4
Operational Analysis and Life Cycle Analysis
STS
DoD Aircraft
Atlas/Titan
Operations Support Cost
Logistics Reliability Maintainability
Concept of Operations
Operational Analysis and Life Cycle Cost
Discrete Event Simulation (DES)
Concept Reference Vehicle, Technologies, and
Mission
Development Acquisition Cost
System Reliability
5
Life Cycle Analysis Tools

• Reliability and Maintainability Assessment Tool
  (RMAT)
• Estimates reliability and maintainability
  requirements for new concepts
• Based on comparability to historical aircraft and
  Shuttle reliability and maintainability data
• Tailored to work with products related to
  conceptual design studies
• Estimates are generated at subsystem level
• Provides estimates for a specific flight rate
• Maintenance burden
• Manpower
• Turnaround time
• Optimal fleet size
• Developed by ECB (Doug Morris and Nancy White)

6
Life Cycle Analysis Tools

• Logistics Cost Model (LCM)
• Defines support costs in terms of logistics
  (training, spares, transportation, etc.)
• Based on historical Shuttle and aircraft
  logistics support parameters
• Based on vehicle and mission definitions
• Matches to RMAT fidelity by examining cost at the
  subsystem level
• Currently being updated (Dr. Charles Ebeling
  University of Dayton)
• Operations Cost Model (OCM)
• Generates top-level operations and support cost
  for reusable and expendable launch systems
• Outputs include ground support and mission costs,
  facilities cost, and engineering support cost for
  up to four specified flight rates
• Under contract through MSFC to KT Engineering
  (Richard Webb)

7
Life Cycle Analysis Tools

• Space Launch and Transportation System (SLaTS)
• Uses parametric cost estimating techniques to
  estimate DDTE and TFU cost
• Cost Estimating Relationships (CERs) are at the
  group level
• Structures
• Propulsion
• Avionics, etc.
• CERs developed using data from NASAs REDSTAR
  database
• NASA / Air Force Cost Model (NAFCOM)
• Similar to SLaTS generates DDTE and TFU
  estimates using CERs
• Estimates generated at subsystem level (e.g.,
  structures, thermal, etc.)
• System Integration costs also generated (SEI,
  GSE, etc.)

8
Life Cycle Analysis Tools

• Mission Reliability
• Estimates reliability and risk based on system
  configuration and mission requirements
• Inputs include configuration definition, element
  reliability, operational scenario, and
  performance requirements
• Outputs include loss of crew (LOC), loss of
  vehicles (LOV), and loss of mission (LOM)
  probability estimates along with statistical
  descriptor metrics
• Tools utilized
• QRAS
• Monte Carlo Simulation (Excel with Crystal Ball)
• Discrete Event Simulation (Arena)

9
Life Cycle Analysis Tools

• Economic Analysis Tool
• Derived from SAIC (Blake Putneys) Screening
  Program for Architecture and Capability
  Evaluation (SPACE)
• Allows for life cycle (development, production,
  operations, etc.) costs to be phased over
  campaign timeline with various learning curve,
  batch buy, and cost ramp-up assumptions
• Inputs include DDTE, TFU, learning curve
  factors, traffic model, and development spreads
• Outputs include funding profile of campaign
  timeline, total spending, and budget overruns

10
Life Cycle Analysis Tools

• Discrete Event Simulation (Arena version 9.00)
• Ability to model complex systems in terms of
  operational characteristics with demands on
  limited resources. Models are tailored to address
  specific problems in life cycle analysis.
• Has been used for in-house studies regarding
  launch processes, maintenance actions,
  transportation studies, and reliability
  evaluations
• Allows for all input variables/factors to be
  stochastically addressed
• Launch vehicle processing times
• Vehicle reliability
• Architecture technology assumptions (e.g., ARD
  reliability)
• Uncertainty associated with cost estimates

11
Recent Life Cycle Analysis Studies

• Economics analysis for the Phased Capability
  Advanced Technology (PCAT) Architecture
• Generated cost estimates using NAFCOM (DDTE,
  TFU, etc.)
• Investigated cost phasing over campaign lifetime
• Lunar Architecture Modularity and ARD
  Reliability study
• Initial study investigated launch vehicle
  selection options
• Recent study investigated the general effects of
  modularity and ARD reliability
• Utilized Discrete Event Simulation to generate
  figures of merit
• Economics/Affordability analysis for the Robotic
  Lunar Exploration Program (RLEP) proposal
• Utilized NAFCOM and various programmatic cost
  wraps to generate expected program costs
• Programs costs were translated into workforce
  labor metrics to demonstrate LaRC personnel
  utilization

12
Ex. 1 Economic Analysis PCAT

• Develop a sustainable lunar architecture that
  fits within the Vision for Space Exploration cost
  and schedule guidelines
• Evolve architecture capabilities as resources
  allow and apply advanced technologies to reduce
  long-term operations costs

• Phased Capability
• Mission capabilities evolve
• Initial crew to lunar surface 3
• Initial missions are lunar equatorial
• Limited application of advanced technology
• Crew to LEO on 8MT launcher
• Cargo on 25 MT launcher

• Advanced Technology
• Mission capabilities
• Greater than 3 on lunar surface
• Global access
• Reusable elements and highly efficient
  pre-positioning of propellants to reduce launch
  rates
• Crew to LEO on 8MT launcher or commercial
• Cargo on 25 MT launcher, Option for commercial
  delivery of propellant

13
Ex. 1 Economic Analysis PCAT
14
Ex. 1 Economic Analysis PCAT
15
Ex. 1 Economic Analysis PCAT
Traffic Model Requirements Phase 1 2014 - 2016
Crewed flight of CLEM in LEO 2017
Crewed LEO test of CLEM/RTM
docking 2018 Crewed LEO test of all
elements 2019 Full crewed mission to check out
all elements both in LEO and LLO,
with LM demonstrating precision
landing and ascent without crew Phase
2 2020 First crewed landing on the Moon 2021-
2023 One crewed mission per year Phase 3 2024
40 t cargo to surface one crewed
mission 2025 and beyond One crew one 10 t
cargo per year
16
Ex. 1 Economic Analysis PCAT
17
Ex. 2 Modularity Study Problem Statement

• Many lunar mission architectural options exist
  which one would result in the lowest life-cycle
  costs while maximizing crew safety?

• Problem is influenced by many parameters
• Launch vehicle characteristics (cost, capability,
  reliability, etc.)
• Technology utilized to achieve workable design
  (e.g., ARD reliability)
• Life-cycle costs of individual hardware elements
• Reliability inherent in hardware elements
• The costs associated with the reliability Cost
  of unreliability
• All of which are related to one of the biggest
  drivers of architectural life cycle costs
  Hardware Element Modularity

18
Ex. 2 Definition of Modularity

• Specifies the number of of physically disparate
  elements required by an architecture
• Differentiation/allocation of required
  capabilities (e.g., life support or propulsive
  burns) amongst hardware elements
• Potential impacts to architecture life-cycle
  metrics
• Cost of unreliability (costs associated with
  failures) will be a function of the number of
  elements, their reliability, their associated
  costs, and launch vehicle costs

• Advantages (higher level)
• Negates need for larger (gt25mt) launch vehicles
• Allows for easier launch vehicle allocation
• Improved survivability of single-fault failures

• Disadvantages (higher level)
• Requires more launches
• Lower architectural reliability due to increased
  ARD operations
• Heavier reliance on ARD technology development

19
Ex. 2 Methodology Overview
Input Parameters
Generic Concept of Operations
Generic Launch Vehicle Options

• Output Metrics
• Affordability
• Reliability
• Crew Safety

Range of Modularity Options
Manifesting Routine
Discrete Event Simulation Model

• Excel/ VBA-based tool

• Arena 9.00 model

20
Ex. 2 Architecture Concept of Operations
21
Ex. 2 Architecture Cost Assumptions

• Nominal 7-day visit to lunar surface for crew of
  four
• Sized for each level of modularity using a
  top-level patched conic orbital analysis with
  standard rocket equation
• Modularity reductions achieved by combining
  crewed and propulsive stages result in a 15 mass
  savings
• Baseline (8 elements) concept requires 136mt to
  be delivered to Earth orbit
• All dollars are FY 2005
• Parametric cost estimates generated using
  top-level CERs based on element and element
  engine dry weights with the inclusion of
  appropriate complexity factors
• CERs from TRANSCOST 7.1, TransCostSystems
• Launch vehicle production and development costs
  generated include ground facility modification
  costs
• Propellant launches allowed with a 5 mass penalty

22
Ex. 2 Modularity Options
23
Ex. 2 Architecture Sizing

• Steps from level 11 to 8 and from 5 to 4 result
  in an increase in IMLEO since decreasing the
  number of TLIs causes an increase in aggregate
  TLI mass since staging effects are incrementally
  removed
• Steps from 8 to 5 result in a decrease in IMLEO
  due to the 15 mass savings assumption for crewed
  and propulsive element combinations
• Mass savings carry over into propulsive element
  sizing, resulting in further mass savings

24
Ex. 2 Architecture Cost Estimates

• Production costs monotonically decrease with
  modularity level
• Due to element count reduction
• Development costs follow IMLEO trend
• Increase from 11 ? 8 and 5 ? 4
• Due to removal of TLI staging effects
• Decrease from 8 ? 5
• Due to reduction in element count
• Reduction in TLI element count does not reduce
  development costs because TLI stages are
  developed only once
• Resulting TLI stages are larger higher
  development costs

25
Ex. 2 Launch Vehicle Options

• Fictitious representations derived from existing
  concepts with generic life-cycle characteristics
• Production cost, development cost, LOV
  reliability, and processing times
• Some vehicles costed on a fixed basis, others on
  a per day basis
• Three vehicles (10mt, 25mt, and 40mt) are
  EELV-based
• Two vehicles (70mt and 100mt) are shuttle-derived

• Grouped into seven suite options
• Options 1 through 4 utilize a 10mt single-stick
  crew delivery vehicle
• Options 5 through 7 launch crew with cargo on the
  larger launch vehicles

26
Ex. 2 Manifesting Routine

• Excel/VBA-based tool used to automatically
  generate launch manifests for each
  modularity/launch vehicle suite option
• Inputs per modularity/launch vehicle suite
  combination
• Element count, mass, and cost characteristics
• Launch vehicle information (capacity, cost,
  processing times, and number of launch pads)
• Flags identifying crew support elements (EV and
  TEI) to ensure that they are always launched
  together and last
• Outputs per combination
• Element count and description after propellant
  breakout evaluation
• Vehicle launch schedule containing element
  allocations and aggregate mass and cost data for
  each of the launches
• Algorithm actually generates two manifests and
  selects the optimal option
• First marches forward through element list
  assigning elements to launch vehicles with
  available capacity
• Algorithm then marches similarly through element
  list in reverse order
• Multiple elements assigned to a launch vehicle
  are integrated in batches up to three
• Original attempt at algorithm was a Genetic
  Algorithm that randomly investigated the option
  space
• Did not provide meaningful results

27
Ex. 2 Manifesting Complexity Example
Element Listing
Launch Manifest
Modularity Level 10
10 Elements, 7 Launches
10 Elements
Element Listing
Launch Manifest
Modularity Level 9
12 Elements, 8 Launches
9 Elements

• Smaller launch vehicles often necessitate
  propellant breakout
• Leads to counterintuitive manifest results (e.g.,
  less elements ? more launches)
• A 5 mass penalty is also added to propellant
  elements to take into account tankage

28
Ex. 2 Element Count Results
29
Ex. 2 DES Model Overview

• Input
• Manifest
• Launch schedule
• Element allocation
• Element production costs
• Element and launch vehicle development costs
• Element ARD reliability
• Launch Vehicle characteristics
• Vehicle costs
• Vehicle processing times
• LOV reliability
• Output
• Average costs (FY05 M)
• Launch costs
• Element costs
• Development costs
• Reflight element costs
• Reflight launch vehicle costs
• Supporting metrics

• Arena version 9.00, Rockwell Software, Inc.

• Model also provides descriptive statistic metrics
  for all outputs
• Standard deviations
• 95 confidence interval half-widths
• Minimum and maximum values

30
Ex. 2 DES Model Assumptions

• Run Parameters
• Each modularity/launch vehicle suite combination
  replicated 500 times
• Generates adequate number of data points for
  statistical evaluation
• Allows for three decimal places in resulting data
• Model simulates a single mission (integrated
  stack on trans-lunar transfer)
• Replications end prematurely with a single loss
  of crew (LOC) event
• Model Capability
• Lunar window, micro-meteoroid/orbital debris, and
  boil-off constraints present in model, but are
  not utilized for this study
• Launch vehicle or on-orbit integration failures
  not resulting in a LOC scenario are rescheduled
  without an assigned downtime, and take precedence
  to any subsequent launches that have not started
  integration
• On-orbit elements, replacement elements, and
  launch vehicle hardware are always assumed to be
  readily available (no schedule delays due to
  inventory or production delays)
• ARD reliability treated parametrically, ranging
  from 75-100
• All crewed launch vehicles have a launch abort
  escape system success reliability of 90
• All integration and pad operations are
  stochastically modeled using triangular
  distributions

31
Ex. 2 DES Model Decision Points
Lunar Surface
LEGEND Ascent Stage (AS) Descent Stage
(DS) Entry Vehicle (EV) Crew stay, 2-3
day Orbital Module (OM) 2-3 day habitat Surface
Habitat (SH) 7 day Trans-Earth Injection (TEI)
stage Trans-Lunar Injection (TLI) stage

• All missions must be scheduled, processed and
  launched to minimize element on-orbit time
• Include metrics for
• LOV Loss of Vehicle
• PDF Payload Delivery Failure failure to
  properly deliver payload (not utilized)
• Failure of either initiates a re-launch

• LV Options
• EELVs
• EELV-derived
• STS-derived

Earth Orbit
SH,DS
TLI 1
TLI 2
TEI
EV
OM
AS
Lunar Launch Window
On-orbit Integration Window

• This window requires full integration and launch
  prior to orbit degradation threshold (not
  utilized)
• FTI Failure to Integrate (on-orbit) ARD
  success rate

Launch Window
32
Ex. 2 Cost Totals (100 ARD Reliability)

• Total Cost Comparisons
• Smaller launch vehicle options (10mt-40mt) seem
  to be optimal
• High development costs for the larger vehicles
  seem to create a disadvantage
• Cost Without Development Comparison
• Removal of development costs allows modularity
  effects to become more pronounced
• 70mt cases become lowest cost options

33
Ex. 2 Cost Totals (80 ARD Reliability)

• Same general trends apparent for 80 ARD
  reliability case
• Total cost comparison favors smaller launch
  vehicle options
• Total cost without development comparison favors
  70mt case
• Overall increase in cost (with and without
  development) due to the cost of unreliability
  associated with ARD
• Modularity effects to cost become slightly more
  apparent with lower reliability, particularly
  when development costs are not present

34
Ex. 2 Differential Cost Plot (Total Cost)
35
Ex. 2 Differential Cost Plot (w/o Development)
36
Ex. 2 Observations/Conclusions

• The assignment of hardware elements to launch
  vehicles has been demonstrated to be complex,
  with counterintuitive results
• Element allocation cannot be assumed to be
  straightforward
• Launch vehicle development costs dwarf the
  effects of modularity and ARD reliability
• Launch vehicle developments costs, if
  substantial, are the biggest cost drivers of an
  architecture
• The model demonstrates that stochastic analyses
  are crucial when evaluating the cost of
  unreliability associated with complex systems
• Discrete Event Simulation has been shown to be
  highly effective in capturing such complexity and
  uncertainty

37
Lessons Learned Major Discipline Drivers

• Although a majority of cost and economic analyses
  are generated using CERs, actual hardware costs
  are not the primary drivers
• Changes in programmatic philosophy required to
  lower life cycle costs
• Many programmatic cost estimates are baselined
  off of development costs
• Development costs must be minimized while keeping
  operational considerations in mind
• ARD technology investment and modularity effects
  can be inconsequential when development and
  programmatic costs are high

Robotic Lander Mission to Lunar Surface Cost
Estimate
38
Lessons Learned Major Discipline Drivers

• Operational considerations must be taken into
  account throughout the entire design phase,
  despite monetary constraints
• Operations generally one of the first areas to
  compromise when faced with restricted development
  budgets
• Lower operations cost usually conflicts with
  performance requirements
• Requirements must be realistic and economically
  adaptable

• STS Phase B Program-Level Requirements (1970)
• Fully Reusable Vehicle
• 25-75 launches per year (with two orbiters)
• Turnaround time less than two weeks

• Demonstrated Shuttle Capability
• Partially Reusable Vehicle (expendable tank)
• 9 launches max (1985), 5.26 average per year
• Average turnaround around 89 days

These requirements were not adaptable from an
affordability perspective. The result is a
system that requires 500 Million per launch!!
39
Lessons Learned Major Discipline Drivers

• Computational models are just that models
• ..a theoretical construct that represents
  physical, biological, or social processes, with a
  set of variables and a set of logical and
  quantitative relationships between them.
  (www.wikipedia.org)
• theoretical construct implies that they are not
  entirely based on truth
• This is especially true in life cycle analysis
  due to a dearth of useful historical data points
  to baseline models
• Many life cycle analysis tools use just as much
  subject matter expert estimation (e.g.,
  complexity factors) as they do actual
  regression-based calculations

• Demonstration of the effects of the New Design
  complexity factor in NAFCOM (Version 2004)
• Structures Mechanisms subsystem
• Unplanned Planetary CER methodology
• All other inputs set to Mars Pathfinder
  defaults
• 1 Existing flight proven design requiring
  mods
• 5 Existing design with greater than half of
  components requiring mods
• 8 New design. Components validated in lab
  environment or relevant environment

40
The Battle will continue..

• Life Cycle Analysis has to be considered an
  iterative component of the conceptual design
  process
• It is not a single answer that is generated once
  the design has closed
• Must be considered part of the iterative loop
  along with other disciplines
• It is a dynamic, non-physics based, highly
  complex discipline that provides the most crucial
  figures of merit safety and affordability
• Life Cycle Analysis provides estimates, which
  should be considered as such
• Life Cycle Analysis estimations have to be
  considered within the proper context, with
  properly documented assumptions
• What is included in the estimate, or more
  importantly what is not?