STEP Protocol Development and Converter Implementation Processes at ESA

1
STEP Protocol Development and Converter
Implementation Processes at ESA

• Hans Peter de Koning - ESA/ESTEC - The
  Netherlands
• Hans-Peter.de.Koning_at_esa.int
• Sandrine Fagot - Simulog - France
• Sandrine.Fagot_at_simulog.fr
• STEP for Aerospace Workshop, NASA-JPL, 16-19
  January 2001

2
Topics

• What is Europe doing w.r.t. standardisation of
  product data exchange archival for space
  applications?
• How did we develop STEP based solutions?
• What successes?
• What lessons learned?
• Road ahead

3
What is Europe doing?
w.r.t. standardisation of product data exchange
archival for space applications

• ECSS E-10-07 WG Exchange of Product Data
• European Cooperation for Space Standardization
• High level meta-standard like NASA-STD-2817
• www.estec.esa.int/ecss
• ftp//ftp.estec.esa.nl/pub/ecss-e-10-07-wg/index.h
  tml
• Specific standards
• STEP-TAS Thermal Analysis for Space (ESA)
• STEP-NRF Network-model Results Format (ESA)
• STEP-PRP Propulsion (CNES) (CNES
  French Space Agency)
• Active role AP232 Technical Data Package (CNES
  Industry)
• Active role AP233 System Engineering (Aerospace
  Industry)

4
Focus of this presentation

• Development process of STEP-TAS and STEP-NRF
• Achievements, some metrics and lessons learned
• Objective To provide useful feedback for others
  contemplating to develop similar niche/specific
  purpose standards

5
Main characteristics NRF (1)Network-model
Results Format

• Targets engineering-discipline independent
  exchange of bulk results data from analysis, test
  or operation
• Representation of engineering objects by network
  models consisting of discrete nodes and
  node-relationships
• Hierarchical tree of network models / submodels
• Definition of properties
• Quantitative, descriptive and functional
  properties
• Scalar, vector and tensor property values
• Property values only at discrete locations /
  discrete states
• Full annotation of analysis / test / operation
  context
• Campaign, case, phase, run
• Facility/tool, environment, date and time,
  organisation, person, ...

6
Main characteristics NRF (2)NRF dataspace
Each gridpoint in the 3D dataspace is a property
value Each can be scalar, vector, tensor Data
model and implementation designed to handle
sparsely populated dataspace efficiently
model_component (model, node, node_relationship)
state
property_class
abscissa_property
7
Main characteristics NRF (3)Proposed NRF/HDF
architecture
STEP-NRF protocol (discipline- independent)
discipline-specific dictionaries
User Application (Reading or Writing Results Data)
NRF Dictionary Structural
NRF Dictionary Thermal
NRF Dictionary Electrical
NRF Dictionary ...
STEP-NRF SDAI programming library
HDF5 Hierachical Data Format v5 - Public domain
from NCSA - Efficient portable binary storage
format - Full C, Fortran, Java libraries on gt10
platforms - Standard for all NASA EOS missions
and some ESA earth observation - Info/downloads
at hdf.ncsa.uiuc.edu - Many COTS / PD tools with
HDF interface
reference to product (-part), shape, location
NRF HDF mapping
reference to FEM / FE
HDF 5 programming library
AP203 STEP file
HDF5 as an efficient alternative for Part21 for
large amounts of similar data
AP209 STEP file (future)
HDF5 binary results file
COTS or PD Data Analysis / Visualisation
Application with HDF i/f (also via WWW)
8
Main characteristics TAS (1)Thermal Analysis for
Space

• Self contained, complete Application Protocol
• AAM, ARM, Mapping Table, AIM, Express-G
• Conforms to TC184/SC4 methods and guidelines
• Geometry defined conform AP203 CC4 surface model
• Thermal-radiative model faces added as associated
  features
• Including possibility to support hierarchical
  submodel tree
• Associated notional thickness, surface material
  and bulk material
• Thermo-optical, thermo-physical properties for
  named material
• Concept of material property environment (Part
  45)
• Kinematic model conform STEP Part 105
• For articulated rigid bodies (e.g. rotating solar
  arrays, ...)

9
Main characteristics TAS (2)

• Space mission aspects
• orbit arc (Keplerian and discrete ephemeris)
• space co-ordinate system, celestial bodies
• orientation, general and named pointing,
  spinning, linear rotation rates
• space thermal environment, including constant or
  lat/long dependent albedo / planetshine tables
• Boolean construction surfaces available for
  advanced tools
• STEP-TAS CC1 Abstract Test Suite
• conform STEP Part 3xx series
• test suite used in validation of TAS processors

10
Main characteristics TAS (3)STEP-TAS Conformance
Classes
thermal-radiative model with basic geometry
kinematic model
constructive geometry
space mission aspects
CC-1
?
CC-2
?
?
CC-3
?
?
CC-4
?
?
?
CC-5
?
?
?
CC-6
?
?
?
?
11
Brief history TAS and NRF developments
12
Challenge

• Find an effective development process
• Within modest budget constraints
• TAS for a niche market
• Between 20 to 300 real users per space thermal
  analysis tool
• Make sure that user requirements are well
  captured
• Ensure implementations can be, and are, validated
• Facilitate efficient and maintainable converters
• Attempt to obtain global consensus
• Ensure robustness and ease of use for end-users

13
Incremental development scheme
Define / Refine / Extend Protocol (ARM, AIM,
Express)
Develop / Upgrade Libraries
Develop / Upgrade Converters Test Suites
Validate Converters with Test Suites
14
Stable Team with Clear Roles

• Simulog (F, software engineering services)
• lead since 1998, implementation libraries and
  some converters
• Fokker Space (NL, space subsystems contractor)
• initial lead, user requirements, ARM, ATS
• GOSET (F, STEP standardisation experts)
• mapping table, AIM, EXPRESS/p21 verification
• Epsilon Ingénierie (F, thermal analysis services)
• check user requirements, converter implementation
• ALSTOM Power (UK, thermal tool vendor)
• converter implementation, distribution to
  end-user
• CNES (F, French Space Agency)
• Review requirements and model, generation of test
  suites

15
Approach to Developing with STEP

• Be as precise as possible on semantics
• Regular feedback between protocol editor and
  implementors
• Minimise ARM - AIM mapping need as much as
  possible
• Use IR contructs where possible direct in ARM
• Automate generation of library APIs as much as
  possible
• Directly from EXPRESS model
• Design for simplicity, implementability and
  testability
• E.g. do not use EXPRESS ANDOR
• Use a high level programming interface to build
  converters
• Use an SDAI toolkit for low level interface

16
STEP-TAS High Level Libraries Toolkit

• High-level API to support efficient converter
  development
• Use ARM constructs - close to thermal developers
  data model
• Hide STEP complexity
• Full set of reading / writing functions in ANSI-C
  and F77
• Associated documentation, examples and test suite
• BagheraView (CNES) included to perform
  independent visual inspection and model reporting
  on PC/Windows
• Distribute toolkit at nominal cost to attract
  tool vendors

API Application Programming Interface
17
BagheraView Screenshot
18
Benefits of High Level Libraries Approach

• Already ported and tested with target compilers
• PC/Windows, Sun/Solaris, HP/HP-UX, Compaq/Tru64,
  SGI/Irix
• Enables to jumpstart converter implementation
• Reduces converter validation / verification
  effort
• All converters share reading/writing interface
• implementation mismatch is minimised
• increased reliability
• Extensibility at affordable cost
• e.g. to add HDF or XML encoding

19
STEP-TAS Converter Architecture
Native Tool Format
Thermal Analysis Tool (ESARAD, THERMICA, TRASYS,
)
STEP-TAS High Level Libraries C and Fortran
API (Simulog, France)
STEP-TAS EXPRESS data model
STEP-TAS physical file (ISO 10303-21)
SDAI C Library (ISO 10303-24) ST-Developer (STEP
Tools Inc., USA)
20
Successes

• Full implementation of converters at reasonable
  cost
• ESARAD/STEP-TAS-CC1 since mid 1999
• Thermica/STEP-TAS-CC1 since Dec 2000
• Broad awareness and (prototype) implementation in
  US
• JPL STEP-TAS demonstrator in 5 US thermal tools
• Full bi-directional implementation STEP-TAS-CC1
  in Thermal Desktop 3.3Beta (Cullimore Ring
  sponsored by NASA-LaRC)
• Reached almost 100 of space thermal tool vendors
• Very close to robust and reliable quality
  exchange for STEP-TAS-CC1 in production tools

21
Problems with STEP

• Bulkiness and incomprehensibility for humans of
  AIM
• Makes validation cumbersome and costly
• Causes large overheads
• Additional in-core storage
• Addition interface layer in library
• Part21 files much larger than really really
  necessary
• ARM to AIM mapping is not computer interpretable
• EXPRESS-X was not available for TAS a lot of
  handwork for high level library interface mapping
• Conformance checking / Abstract Test Suite (Parts
  3x)
• Very cumbersome and costly to develop and
  maintain
• Not well designed for automation of regression
  testing

22
Lessons Learned

• Iterative, incremental development necessary
• Can not get things right in one go
• High level programming library very useful
• Greatly reduces converter development effort
• Speeds up standard penetration increases
  exchange reliability
• Chosen API close to old SET-ATS interface needs
  reconsideration
• Continuous resource / funding level is needed
• To meet vendor expectations by providing timely
  support and library updates resolving SPRs
• Communication must be improved
• Start a STEP-TAS implementors forum ?

23
Metrics

• STEP-TAS took 5 years from conception to CC1
  industrial converter release
• Up to now funded with roughly 1 million Euro /
  USD
• Excluding US converter developments
• Appears to be a quite good price/performance
  ratio
• Approximately 20 different people have worked on
  different parts of STEP-TAS and STEP-NRF

24
Why was STEP-TAS not yet submitted as an ISO
TC184/SC4 work item?

• Valid question
• Started as a European effort to follow up SET-ATS
• Limited resources prevented us to bear the
  additional effort of formal ISO standardisation -
  no added value yet
• With the good co-operation with NASA in
  place,nearly 100 of the space thermal tool
  vendors in the world were reached
• However it is agreed that a formal way of
  publishing and maintaining the standard is now
  needed imminently

25
Road Ahead (1)

• Web-based collaboration on further development
• ftp//ftp.estec.esa.nl/step-tas/index.html as a
  start
• Publication of protocol, SPR list, test suite
• FTP distribution of STEP-TAS toolkit releases
• E-mail tech support
• Take European / US collaboration to a next stage
• Action ESA and NASA together with Simulog
• Clarify/establish intellectual property rights
  and support

26
Road Ahead (2)

• Currently trade-off for Return on Investment
  next 5 years
• Appropriate level of formal publishing (ISO?,
  ECSS?, NASA?)
• Select AIM or ARM based exchange?
• Upgrade high level API to support all STEP-TAS
  constructs
• e.g. submodelling, non-uniform meshing and node
  numbering
• Upgrade libraries to resolve all reported issues
  (SPRs)
• Upgrade converter options
• Consistently support length unit selection /
  conversion
• Meta data entry

27
Road Ahead (3)

• TRASYS/STEP-TAS-CC1 bi-directional converter
• Stand-alone tool by Simulog on ESA funding
• Will be made available at nominal cost or
  preferrably free
• Upgrade BagheraView capabilities
• By Simulog sponsored by CNES
• PATRAN/STEP-TAS-CC1 interface
• By MSC sponsored by NASA-LaRC
• Continue NRF developments
• HDF5 binding (co-operation with EDF, NCSA and
  others)
• Assess links with Engineering Analysis modules,
  ISO 10303-50
• Pilot web-based remote consultation of structural
  test results

28
Acknowledgements

• STEP-TAS and STEP-NRF were developed under ESA
  contract by an industrial consortium, consisting
  of
• Simulog (F, prime)
• Fokker Space (NL)
• Association GOSET (F)
• Epsilon Ingénierie (F)
• Alstom Power (UK)
• and supported by CNES (F)
• Special thanks to Georg Siebes (NASA-JPL) who is
  and has been the driving force for the
  development and acceptance of STEP-TAS in the US
• Also acknowledgements to Ruth Amundsen
  (NASA-LaRC)for actively promoting STEP-TAS
  implementations