Task: Analyze effectiveness of the items architecture to cope with random hardware failures
Analyze how well the system architecture can deal with random HW failures by comparing to the architecture metric requirement targets. This task is applicable to ASIL (B), C and D.
Purpose

To determine how well the HW architecture of the item meets the requirements for fault handling as represented by the HW architectural metrics

Relationships
Main Description

The architecture metrics to be analyzed are the single point and the latent fault metrics that were specified in the HW specification.

The scope of this analysis is limited to the electrical and electronic parts, only electrical failure modes and failure rates are considered for electromechanical parts.

Compliance with the HW architectural metrics specified shall be achieved for each safety goal the item is involved in.

If combining applicable safety goals then the target metric of the safety goal with the highest metric shall be considered.

The concepts of diagnostic coverage, latent fault metric and single point metric as detailed in ISO 26262, vol 5 annex C are applied to the requirements in this section.

Relevant latent and residual faults shall drive the estimation of the diagnostic coverage of safety related hardware elements with their safety mechanisms.

The estimate failure rates for hardware components used in the analyses shall be determined in a number of ways

Using part failure rates of hardware from recognized industry sources

Statistics based on tests and field returns, it is expected that these results should have an adequate confidence level

Expert judgment founded on a engineering approach based on quantitative and qualitative arguments, that uses predetermined structured criteria.

If there is insufficient evidence for the calculated failure rate of a single-point or latent fault, then alternative means of preventing this fault from violating safety goals should be employed.

It is advised that the user refers to ISO 26262 Vol 5, clause 8 for details of the objectives, requirements and methods relating to architecture metrics.

The calculation of the single-point fault metric is required for items of ASIL C or D, and recommended for items of ASIL B.

For each safety goal, a quantative target value for the "single-point fault metric" must be determined, this can be derived from the HW architectural metrics calculation as applied on well trusted design princples or derived based upon the following information.

For ASIL B the single point fault metric is >= 90%

For ASIL C the single point fault metric is >= 97%

For ASIL B the single point fault metric is >= 99%

This information was summarized from ISO-26262, Vol 5, clause 8.4.5, Table 4, Possible source for the derivation of the target "single point fault metric" value.

The calculation of the latent fault metric is required for items of ASIL D, and recommended for items of ASIL B or C

For each safety goal, a quantative target value for the "latent fault metric" must be determined, this can be derived from the HW architectural metrics calculation as applied on well trusted design princples or derived based upon the following information.

For ASIL B the latent fault metric is >= 60%

For ASIL C the latent fault metric is >= 80%

For ASIL B the latent fault metric is >= 90%

This information was summarized from ISO-26262, Vol 5, clause 8.4.6, Table 5, Possible source for the derivation of the target "latent fault metric" value.

For each safety goal the whole Hardware of the item shall comply with one of the following alternatives.

To meet the latent-fault metric value

To work back from the fault targets applied at the hardware element level to comply with the overall latent fault metric for the complete hardware item, and to provide the rationale for compliance with these targets.

To meet the diagnostic coverage (which will have the same value as the latent fault metrics, above), for each hardware element with faults that can lead to the unavailability of a safety mechanism.

This section is not exhaustive and it is recommended that users refer to ISO 26262, vol 5 annex C for further information regarding HW architectural metrics and annex D for evaluation of the diagnostic coverage (this annex has a large number of tables that help with identification of possible random HW failures). Appendix E and F are also very informative regarding the use of architectural metrics.