1) Summary In the presentation, I discuss some aspects on the concept of flow and Flow Analysis, which have emerged primarily from considerations of flows in computer science and from the component concept used in Component Based Software Engineering. I discuss the development of basic concepts, in particular the difference between the structural and operational concept of a system, as elaborated by Shchedrovitsky, for example. The similarities between Flow and Functional Analysis, which Logvonov and Lebedyev have elaborated, are relativised in this way, since Functional Analysis is based on a structural system concept, whereas Flow Analysis belongs to the operational part. In the presentation I can outline the concepts only examplarily. In more detail I discuss an application to flows of water and dirt in a dishwasher, which was analysed by Uwe Schaumann applying Flow Analysis. Due to time constraints, I cannot touch the relations to computer science, which are presented in more detail in the paper. ------------------ 2) TRIZ Flow Analysis - State of the Art After 2000 major works on Flow Analysis were published by Lyubomirsky, Logvinov and Lebedyev. In particular the Flow Development Patterns by Lyubomirsky 2006 should be mentioned. However, the 60 examples compiled there have more to do with flow improvement than with flow development. This is understandable from the perspective of the step-change approach of a practically oriented TRIZ, but it is based on flows in TS that are of specific structure and thus already have the character of a TC. Alex stated in his tutorial that flows go through components and thus the concept of flow is in a certain sense orthogonal to the concept of component. I will give more arguments in that direction, but insofar flows provide active functionality they are also to be considered as TC, since TC is the central TRIZ concept to encapsulate functionality. In my opiniion, this is also the deep cause why the mentioned papers emphasise the strong parallels to the well-developed TRIZ concepts of Functional Analysis. So what is a flow and should the term be narrowed down further? The word "potok", which is used in a largely uniform way in Russian texts, appears in English translations in several forms as flow, flux, stream ... and thus points to a semantic overloading of the terminology that hardly played a role in the analyses discussed so far. There are more questions: Do the parts source, sink, flow, carrier and channel, which Lebedyev and Logvinov consider constitutive for flows, belong to every flow? Does a control flow in computer science have a carrier or a channel? Is a bit stream a flow? What about the flow of events? Event channels and event schedulers certainly play a role in computer science, but in the context of the more general Observer Pattern and in a dynamic structure which is orthogonal to the control flow structures. ------------------ 3) Flows and Systemic Throughput My central objection to previous flow concepts, however, is that flow analysis is essentially concerned with throughput issues and hence not so much with functional conditions, but with viability of systems. Lyubomirsky shows that flow analysis has historically arisen from the development law of minimum energy throughput, which Altshuller formulated as his second law and counted it among the "static" ones, although a highly dynamic process is involved here. Altshuller formulates in detail "A necessary condition for the viability of a technical system is the flow of energy through all its parts." Altshuller thus implicitly postulates a technical system to be an Open System that only builds up its internal structure under the required *throughput of energy, substance and information*. The stability of these external throughput conditions is, as context, the condition for the stability of the system. This also applies beyond a "minimal" energy throughput – minimality, which is essential and discussed as a matter of course by Lyubomirsky, does nevertheless not occur in Altshuller's original text. At this point, one of the fundamental contradictions of any systemic approach is already present – the decomposition of the indecomposable. For a functional analysis, the decomposition of the system into its components is indispensable. However, the system can only be operated in its assembled state. Petrov emphasises that no part of an aeroplane can fly, not even the sum of all parts. Only the assembled aeroplane can fly. This indecomposability does not end at the boundaries of the system, because the operation of a system is bound to certain operating conditions. On the moon, even an assembled aeroplane does not fly. ------------------ Shchedrovitsky devotes a large part of his system concepts to precisely this contradiction and distinguishes system models of first and second kind, which build on each other. He emphasises: The systemic-structural approach (of a system of first kind) does not capture processes as such. It are precisely these questions of a complex descriptive structure of systems that have to be addressed in flow analysis. In this respect Flow Analysis is fundamentally different from Functional Analysis and much closer to Process Analysis. Organisational Informatics distinguishes in this respect between structural and operational organisation of a socio-technical system. ------------------ 4) Flow of workpieces A technical system (TS) consists of a number of technical components (TC), which – in the simplest case – transform the object (workpiece) several times in a well-defined sequence of operations until it becomes the useful product. For example, the bodywork department of a car manufacturer consists (structurally) of two sub-departments (components) – the press subdepartment and the coloring subdepartment – which are involved with different functions in the transformation of the raw material into the bodywork as useful product. ------------------ The *flow of workpieces* through this chain of functions leads to a sequence of targeted state changes of the workpiece. For this to happen, however, the workpiece must fulfill well-defined specifications at input and output of each TC, because the output of a TC on the path of the flow of workpieces is followed by the input of the next TC. This structure is typical for a simple assembly line system, showing that the flow of workpieces, as the prototype of any "useful" flow of substance, is *orthogonal* to the energy flow to the tool which carries these state changes of the workpiece. ------------------ 5) Schaumann's Example Are there also functions assigned to that orthogonal flow of workpieces? Schaumann records this functional aspect of flows in a separate column with heading *function type* and values *transport function*, *correction function* and *production function*. However, this subdivision does not seem to be very helpful in his modelling, since all three function types are always present in the depicted parts of the tables. Nevertheless, such a *Flow Function Analysis* based on a clear classification of flow function types seems to be quite useful in order better to understand the difference and interplay between a *flow as an active tool* and a *flow as a simple means of transport of passive workpieces*. ------------------ The transport function seems to be one of the basic functions of flows. Here, however, a distinction must first be made between the carrier and the carried as proposed e.g. by Lebedyev. He tries to unite both under the term flow. But is that appropriate on its own? The conveyor belt in an assembly line production as a carrier of the workpieces has clearly other independent functions as a TC – in addition to the actual transport function, the speed of the conveyor belt determines the work rhythms and also the reaction possibilities in problematic situations – triggering the "red button" stops the conveyor belt and thus also the flow of workpieces in order to concentrate on eliminating a problem (as provided for in the Toyota Production Model, for example). In Schaumann's example the water flow fulfils its transport function by transporting the "workpiece" dirt from the plates to the collection sieves. The water flow is thereby modelled as a carrier of three other flows (coarse, fine and micro dirt). We are dealing with a situation similar to that of the conveyor belt: the carrier as flow is meaningless ("parasitic") without the carried material, but on the other hand, as a mixture of carrier and carried material, it may change the properties and thus the behaviour of the flow. Hence the properties of the transport function of the flow do not result solely from those of the carrier. ------------------ 7) Altshuller's Development Laws and Open Systems Altshuller's Law of minimum energy conductivity of a system may be read as requirement to supply its components with an energy throughput to "awake" them and keep them "alive". This is a fundamental constitutive principle of Open Systems in general, especially of living and social systems driven by metabolism. A specific internal structure only gets formed and maintained if a defined throughput is guaranteed. The classic example in the Theory of Dynamical Systems are the Bénard convection cells in a heated fluid. In Altshuller's law this throughput is seen solely under increasing conductivity, hence as quantitative optimisation. The structuring effect of the energy throughput on a component depends, however, in most cases also very strongly on qualitative parameters of the energy flow. Both the type and composition of the energy and the supply regime play an important role, for example, in the technical exploitation of resonances and dissonances. ------------------ This, however, brings a number of other Altshuller's development laws into the focus of consideration – the Law of Adjusting the Rhythms (nowadays also called Trend of Increasing Coordination), the Law of Uneven Development of System Components (caused by the respective specific absorption capacity of energy of a component also in the demarcation and competition with other components) and the Law of Transition to the Supersystem (which means in many situations the shift of control of this energy flow, which is external for the component but internal for the system, to the system level). We see that the effects of several of Altshuller's laws meet in a reasonably comprehensively understood Flow Analysis. ------------------ 8) Flows and Transmission Another connection, which has not been considered much in the previous explanations, is the implicit appearance of flows in the (extended) TRIZ model of a technical system with the components energy source, engine, transmission, tool, processed object and control. Indeed, it is about an energy flow with source, transformation into energy useful for the tool (the engine), flow of energy to this tool (the transmission) and transformation of the energy into a state-changing effect on the object (by the tool). The flow of energy ends here at the tools and is orthogonal to the flow of workpieces. ------------------ The same considerations about the interaction of quantitative and qualitative parameters as developed above for the energy flow also apply to the flow of substance. However, the target of this flow is not the tool, but the place (a central notion in Shchedrovitsky's system concept of first kind) that the workpiece occupies as object of transformation. ------------------ Picture The same applies to the flow of information, understood as a data stream that provides relevant formally structured information as description of the state transformation to be performed on the object. ------------------ Adding the control to the technical system, the energy flow is joined with the flow and functional transformation of information. The close interlacing of both flows is expressed in the fact that in the abstract TRIZ model of a complete technical system the control not only affects the tool, but also (potentially) the energy source, engine and transmission. While energy and information flows are considered as active agents in the process of transformation of the workpiece by the tool, in the TRIZ concept of that functional transformation the workpiece as an object remains peculiarly passive. The corresponding flow of substance is limited in its function in that methodological model solely to transport into and out of the operative zone. Of course, such a passivity of the workpiece is by no means appropriate for a description of many technological processes, especially in chemical and biological contexts. ------------------ 8) Conclusion For many flows, it is difficult to identify concepts such as carrier, channel, source and sink which are constitutive for Lyubomirsky and Lebedyev. Flows with such additional structural properties are already highly technically *enclosed flows* and more or less elaborated technical systems with very specific functional properties. In the case of the propagation of thermal or acoustic fields in the SF analysis, as well as in the case of the propagation of liquids and gases through diffusion or similar phenomena that are widespread in technical applications, a flow analysis can hardly be carried out in a targeted manner with the conceptual tools developed so far. This does not devalue this work in any way, but raises the question how to frame the target of a flow analysis considered there in an appropriate way. ------------------ It is suggested to take more into account the distinction between system concepts of first and second kind in the sense of Shchedrovitsky and thus a distinction between structural and operational organisation. In the transition to a system model of second kind, the functional properties of the components that appear as bundles of functions in the system are only one of the ingredients which turn the relationship between tool and workpiece (object) into a real-world state-changing action. Additionally, their interaction with the flows of energy, substance and information is required. ------------------ The clear separation of the terms *component* as a stateless functional unit and the flow of *objects* as state-carrying units of instantiation as in Component Based Software Engineering is suitable for further terminological clarity here. It is proposed to qualify the previous considerations on flow analysis as Flow Functional Analysis and to consider it as part of a more complex field of Flow Analysis, which in turn is a part of Process Analysis within a system model of second kind. ------------------ ------------------ 4d) Let us take a closer look at the flows of dirt in Schaumann's application. We apply another TRIZ tool for their analysis, the Smart Little People (SLM) or Miniature Dwarf's method, in order to emphasise the workpiece character of the dirt particles more clearly. At the beginning the dwarfs cling to the plates everywhere. The water flow – initially in a productive function – detaches the dwarfs from the plates and then transports them – in its transport function – to the collection point, the filter system. The focus of Schaumann's analysis and innovation is on the functioning of this filter system. In this case, however, the operational zone includes not only the filter system itself, but also the feed mechanisms to it, since the dwarfs have different sizes and thus resist to be "captured" by the filter system to different degrees. The subdivision into three flows (or streams?) (coarse, fine and micro dirt) therefore has less to do with the subdivision into flows than with different technico-functional properties of the objects (the dwarfs) to be processed with respect to the tool (the filter system) in the cleaning process, which are brought to the collection point several times via the carrier flow of water. When enough dwarfs are "arrested", they are "released" into the wastewater via a functional inversion in the filter system. 5) Due to time restrictions I skip the part of my paper on flows in Computer Science and the relation of control flows to component concepts, in particular to component models and COTS components. This are components produced by independent third parties and being available as Commercial Off The Shelf. This part is quite important since such a composition of TS using components is very common also in other engineering areas. 8) Flows and SF Modelling The prominent role of the energy flow compared to the flows of substance and information is also constitutive for generalisations of functional analysis, especially around elements of a substance-field (SF) analysis. At a first glance, such a generalisation seems to have little to do with flow analysis. However, fields mediate effects between two substances, which can often be interpreted as tool and workpiece of a functional model, but due to the more symmetrical structure of an SF model the clear division into active and passive parts that is typical for functional models are avoided. Fields, at least in the classical understanding in which the field concept is not overstretched as in some TRIZ publications, are always introduced descriptively as potentiality for action, which generate the energy flows required in the operational mode, for example, via differences of potentials. This is why gradient methods play an important role in Lyubomirsky's flow development patterns directly or in a disguised form (for example by shortening the flow length).