ISO/TS 10303-26:2011(E)

Annex B
(informative)

Technical discussions

B.1 General

This annex contains discussions concerning the relationship of the EXPRESS language and binary representation of EXPRESS-driven data. Example use cases supported by this part of ISO 10303 are also described.

B.2 HDF5 overview

For those less familiar with HDF5, it's perhaps simplest to start ignoring some of the complexity and focusing on a simple use of its basic architecture as follows:

HDF5 was chosen as the underlying technology for this part of ISO 10303 after performing a state of the art survey of existing binary technology. The existing technologies had two main areas of focus:

Given that the ISO TC 184/SC 4 Industrial data community is based in the manufacturing and engineering industries, the highest priority requirements were for support of large volume datasets. Those priorities aligned well with the capabilities and aims of HDF5 making it the choice for the underlying technology of this part of ISO 10303. However, there is nothing precluding future editions of this part of ISO 10303 from including binary data representation capabilities based on other underlying technologies.

A key resource for the understanding of the HDF5 technology and how it relates to this part of ISO 10303 is the HDF5 Users Guide: Chapter 1 The HDF5 Data Model and File Structure. Implementers and reviewers are encouraged to read that document before continuing with this part of ISO 10303.

Another key resource is the HDF5 Tutorial [3].

The EXPRESS/Binary project [5] development site maintains project requirements, overview presentations and notes/minutes of project meetings.

B.3 Use case: External HDF5 files

HDF5 datasets can represent data defined by an EXPRESS schema as specified by this part of ISO 10303. One benefit of this approach is that no modification is required to the EXPRESS schema.

In addition to the approach of mapping EXPRESS-driven data directly into HDF5 files, it is possible to use HDF5 files as part of a data exchange set where the HDF5 file is referenced from an EXPRESS-driven data representation based on other ISO 10303 implementation methods.

The benefit of the second approach is that HDF5 is used only for the large volume of data in the table for which efficient handling is required. The data in the table can be created by an application that knows nothing about EXPRESS. The data defined by EXPRESS is a wrapper that adds semantics to the HDF5 dataset.

In this use case, an HDF5 File can be considered to contain a table of data. A table of data can also be modelled as an entity type within an EXPRESS schema. An example of such an entity is explicit_table_function defined in ISO 10303-50. An entity which is a table of data can also be an instance of the entity type externally_defined_item defined in ISO 10303-41, which references an HDF5 dataset for its definition.

The way in which a table of data is used to represent the variation of a property with respect to a variable is described in ISO 10303-51. ISO 10303-51 also assigns a representation_context as defined in ISO 10303-43.

B.4 Use case: Thermal balance verification and validation

It is standard practice to perform a Verification and Validation effort during the product development cycle. A specific instance of this activity in the Aerospace Industry is the Thermal Balance (TB) testing, which is performed to validate that the actual product performs as intended.

During a TB test a spacecraft is exposed to bounding mission environments and operated in such a way to most closely approximate actual space flight conditions. These operating points are maintained for extended periods of time to permit the hardware to thermally equilibrate. Once steady state conditions have been achieved, the resulting data are used to compare test performance against analytical predictions for these conditions. Typically, some adjustments have to be made to the simulation model obtain satisfactory agreement between the observed behaviour and the model predictions. At this point the simulation model is considered 'correlated' and can be used to make predictions of the spacecraft during actual flight that are regarded as close to reality as possible.

In a typical spacecraft TB test the number of test sensors (typically temperature sensors) can be range from a few hundred to several hundreds. In addition, the hardware under test has flight sensors that count upward from 50. These sensors are queried frequently to get early warning of impending undesired conditions and to give control circuits inputs at a suitable frequency. Most of these sensors are recorded at least once per minute.

The duration of TB tests can range from a few days to a few weeks, resulting in millions of unique records and data files in the order Gigabytes. It is not uncommon, that several of these data originate in different location, generated by subcontractors using different tools.

The short description above points out two, often tedious and error prone, activities:

EXPRESS BINARY will significantly reduce the time required to perform these function, with the additional benefit of long-term data retention and retrieval. EXPRESS BINARY is able to do so because it combines the ability to use a data model, which can be custom developed and expressed in a number of languages, e.g., UML or EXPRESS, with a highly efficient data archiving and retrieval mechanism (HDF5).

A taxonomy of information related to TB testing is depicted in B.1. It can be used to develop a custom data model or mapped to an existing data model like STEP-NRF. Data for an instantiation of any of these models can be obtained if an actual demonstration of EXPRESS BINARY is pursued.


Taxonomy for Thermal Balance use case

Figure B.1 — Taxonomy for Thermal Balance use case

B.5 Random access and navigation

As part of the requirements gathering activity for this part of ISO 10303, the need to for support of random access and efficient navigation of very large datasets was identified. A technical analysis of the existing capabilities for binary very large dataset management shows that this requirement is not currently satisfied by any of the available technologies. Research into support for this requirement continues and it is hoped that a future edition of this part of ISO 10303 will address the requirements once it is available in an underlying technology.


© ISO 2011 — All rights reserved