Synthèse comparative : ISR versus Analyse Fonctionnelle
Typage méthodologique
Analyse Fonctionnelle
Functional Analysis relies on “common sense”. Derivative applications pin around the concept of “function”, stemming from mathematics.
Nevertheless, Functional Analysis does not endow “function” concept with any objectifiable precise meaning.
Functional Analysis does not define any consistent and comprehensive semantic to be shared between its different applications, such as:
All functional associated developments are based on the set Theory, which is inappropriate to physical modelling, as quantum physic puts it forth.
As the consequence, Functional Analysis cannot be regarded as a valid starting ground to set up modelling methods conforming to scientific criteria.
Ingénierie Système Relativisée
RSE is a constructive, systematically relativized and formal method, which relies on RS (Relativized Systemic) mathematical algebra.
Any RSE model, at any stage of development process, meets scientific criteria, so that it may be used as a reference to evaluate early prototyping for exemple,
and from any viewpoint.
RS algebra is a non-commutative algebra stemming from a deep analysis conducted by Pr. Mugur-Schächter,
of the cognitive situation entailed by quantum physic.
Deux façons antinomiques de concevoir le Réel
Analyse Fonctionnelle : un credo réaliste
Material reality does exist and may be known, such as it is.
Genetically, Functional Analysis comes as a Safety Analysis tool to qualify something already designed, not as a design tool.
A material entity may be broken down into other material entities, and so on. Reality organization may be fully described.
Ingénierie Système Relativisée : la connaissance comme construction relativisée
Reality is a substratum we are inner part of, both physically and psychologically. It cannot be known on its own.
Any description results from a finality driven building process,
out of traces attributed to the interactions between the entity to be described and the measure devices or our biopsychic perceptions.
What is respectively ascribed as properties to the first one and to the last one is a matter of convention resulting from established knowledge and theories.
We devise verifiable organizations of Reality according to commodity and efficiency criteria. This characterizes so much innovation as knowledge building processes.
Deux concepts différents de description
Analyse Fonctionnelle
Any physical entity is described as a “component” to which “functions” allocated.
This “component” may be broken down into other “components” and “functions” further refined independently from any context.
“Functional states” may be used to specify different ways of producing “outputs” from the same “inputs”.
Even when some “component” is dedicated to a specific “service function”,
it is not possible to objectify the corresponding viewpoint on the “component” in terms of correlations of physical interactions.
What is the impact of some “input” on the “outputs” in a given situation that is not explicitly defined?
Ingénierie Système Relativisée
Any physical entity “exists” relatively to a context, just as a causal node of the stable correlations in space and time which can be registered or publically testified,
between different observables.
A life phase (a specialization of the concept of physical state) is regarded as a lasting way of “being” of such an entity,
relative to a given viewpoint and a given future, such as it can be conceived.
By construction such a life phase is physically reproducible.
A life phase wholly describes the associated physical entity from this viewpoint.
It stands as the physical model of causal node that associates a view with the external parameters which condition its values in a deterministic or a statistical way.
The organization of the different life phases makes it possible to describe the whole dynamic of the entity under study.
The RSE conceptualization levels make it possible to focus on the semantic level while connecting it explicitly to the physical interactions level
(transfer level) to ensure objectification.
A model which specifies the different possible chaining of the life phases relative to a viewpoint, depending on context and events.
Conception Système
Analyse Fonctionnelle
System design boils down to two different kinds of breakdown structure without any rule to ensure consistency:
-
As many “functional” breakdown structures as “top functions” (e.g. “service function”);
-
A “material” breakdown structure endowed with specific “functions” (e.g. “constraint function”).
“Functional” breakdown is supposed to formalize the original analysis framework used in Safety Analysis consisting in “external functional analysis” on one hand and
“internal functional analysis” on the other.
Nevertheless all the “functional” breakdown levels between top level “functions” supposed to describe the Product and allocated “functions” remain deprived of any meaning.
As a corollary top “material” breakdown structure, above the level at which “functions” are allocated to “components”, remain irrelevant.
System design eventually consists in allocating “functions” to “components” but without any rule managing further refinement of the design:
no generally specified constraints ruling “allocated functions” breakdown and associated “component” breakdown.
Ingénierie Système Relativisée
System design results from a viewpoint on the set of Statements of Work (SoW) which motivates the Product development.
They are as many systems as spheres of operation defined by the Organization, each one in charge of a finite set of Product features.
SoW formalizes the stakes associated with Product usages either from Human viewpoint (bio psychological description: man is the measure instrument) or
from technical viewpoint (technical and scientific description relying on standardized measuring devices).
RSE manages explicitly the conventional dependencies between both kinds of physical description.
Each system design tackles the Product development in an autonomous way and it is up to the architecture design to define and optimize shared material resources.
This organized one-on-one debate between needs driven design processes and resources optimization, results in compromises and arbitrations.
This convergence of autonomous viewpoints toward a single Product is made possible by the RSE concepts of relativized description,
system specification and system design.
They make possible to combine or to isolate at will standalone physical descriptions, at different conceptualization levels, at any stage of development process.
Complétude
Analyse Fonctionnelle
Functional Analysis considers it is possible to fully describe the expected behavior of the Product.
To meet this goal, it allocates to the same “component” and consolidates “functions” resulting from different "functional" breakdown structures,
each of this ones corresponding to a given viewpoint.
As a result, the different stakes get mixed up and can neither be individually traced in the design nor tested or evaluated during IVVQ.
Any modification or new need resulting in new allocations entail a tedious reworking of the “component” functional structure with no straightforward measurable impact.
Ingénierie Système Relativisée
An entity description explicitly results from a finite set of viewpoints aiming not at exhaustiveness but at relevance against the goals.
Any entity may be wholly described at design time and evaluated during IVVQ from one or several viewpoints without any mixing.
Any change as an impact limited to the relevant viewpoints.
The retro engineering of above mentioned example within RSE framework has led to free autonomous physical description of the same ECU usable for testing (IVVQ).
These concurrent relativized descriptions of the same physical entity were made possible by the combination of key RSE concepts:
-
Event algebra to implement the semantic of a given viewpoint;
-
Protocol to bridge the gap between physical transferred level of description and these semantics;
-
Physical state relative to some view involved in at least one viewpoint.
Vérifiabilité
Analyse Fonctionnelle
Functional analysis doesn’t deal with the crucial problem of the relationship between models and facts which determines nevertheless the assessment
of the Product eventually realized against the needs and constraints which have motivated its development.
Operationally such a connection is made possible only when:
-
A basic “function” corresponds to a software or a physical part;
-
There is no dependency whatsoever between this “function” and other “functions” allocated to the same “component”.
But:
-
This may be possibly achieved only at the bottom level of “functional” breakdown;
-
If so, it closely links “functional” analysis and design and results in the impossibility to define clear areas of responsibility and in poor design work,
especially software design.
As a corollary:
-
“Functional” requirements are usually not physicaly verifiable during IVVQ;
-
A “component” at the bottom of the material breakdown structure is verifiable only at the price of a costly, tedious and approximate retro engineering
of its “functional” description to identify the correlations to be observed at its physical boundary.
Ingénierie Système Relativisée
Any RSE concept (entity, system, state) may be physically described.
A requirement is defined as a specific generative and qualification process which stresses a dynamic or static feature of the entity to be described.
Requirements are organized according to the life phases structures which ensure consistency and make clear impacts and dependencies.
This together with formal definition of the requirement concept make possible automated documentation generation out of models.
RSE puts forth a threefold approach to testing:
-
The must-testing scenarios which correspond to very sensitive situations, safety imposed requirements …;
-
The automated generation of scenarios out of statistical models of context that may reflect, for example, the chaining of events more likely to happen;
-
The computation of trusty indicators as to the conformity of the physically realized to its models:
a purely statistical approach based on the computation of probability laws relative to the different viewpoints, out of the model of a stable entity
(http://www.theses.fr/2014VALE0020 - chapter VIII p213,(2014)).
Complexité de mise en œuvre
Analyse Fonctionnelle
When dealing with “big systems”, functional analysis leads to black-hole-like-models, which spin out of the designers’ control,
impossible to understand and without identifiable links with factuality: a conceptual chaos, a primitive stew of “functions” and “flows” never achieved.
Ingénierie Système Relativisée
Each viewpoint remains autonomous and different viewpoints may be combined at will to highlight interferences.
Together with conceptualization levels, this feature makes it possible to master the modeling of so called “complex” situations.
Based on a brand new physical interpretation of probability and entropy,
a complexity indicator may be computed relatively to a finite set of views on the same entity.
This may be used so much to qualify an existing resource against a set of goals as to evaluate a design to keep development process under control.
The RSE complexity measure is relative to:
-
The respective entropies of the views at stake (stability, salient aspects);
-
The interdependence of these views relatively to the context (does an event that changes an aspect value also impact on over aspects?).