SIMULATOR, AN OPEN SOURCE SIMULATION PLATFORM DEDICATED TO AND EMERGY STUDIES

Raphaël Valyi a,1, Enrique Ortega b, 2 a Candidate for Master of and Master of Materials at the Ecole Centrale de Lyon, 36 Av Guy de Collongue 69134 Ecully, France; b Professor at the Engineering Faculty of the Unicamp University 6121 CEP 13.083-862 Campinas - SP - Brazil

ABSTRACT To develop more objective and more efficient criteria for judging , the interactions between human activities and the environment are being simulated more often using various forms of systemic modeling. H.T. Odum was a pioneer in applying modeling concepts from electrical circuitry and irreversible to systems ecologyusually dedicated to electronics andor . The Emergy Simulator (EmSim) project is a computer implementation of the main concepts of Odum’s Systems Language a.k.a. the Energy Circuit Language. First, with EmSim you can share models through the Internet or export them as drawings. Second, EmSim is able to translate Odum’s Energy Systems diagrams directly into a set of ordinary differential equations that it can then integrate and plot. Finally, EmSim can correctly compute emergies and transformities. If we admit that the emergy of a product is the sum, on a common basis, of all the available energy (eg. ) contributions required to make that product (from sun light for solar emergy or from fertile for ) then EmSim is able propagate the emergy in very complex production networks involving loops and co-productions. If you are able to enter an exergy weigthed causal diagram, then EmSim can compute the emergy of the various products and also infer their transformity! You are welcome to try it on line, learn more and participate at: h ttp://emsim.sourceforge.net a democratic project open for unlimited improvement by otherss. See also: http://emsim.sourceforge.net.

1 Phone : 00 334 72 85 94 11 ; e-mail adress: [email protected]

2 Phone : 00 5519 37 88 40 35; Fax: 00 5519 37 88 40 27 e-mail address: [email protected] 1. INTRODUCTION Why we created Emergy Simulator [1] the way it is because we wish it could overcome the following current limitations in the future: 1. We think that systems ecology models can only make consensus if they are widely discussed and taught. But sharing models requires standards. In every scientific field, the trend is that standard exchange formats closely stick with informatics file formats. Odum's energy diagrams is already a powerful standard but no suitable informatics implementation was available so far. A specific drawing tool was also required to communicate about models without depending on commercial softwares. 2. Moreover, when Odum's energy diagrams stands for dynamical systems, researchers are always compelled to translate themselves graphs into differential systems (not always with the same rules!) and program themselves the temporal integration using commercial software like Matlab or Excell. Others simply don't know how to do the conversion because of a lack of communication about the methodology. 3. When dealing with the emergy, very few researchers are able compute themselves the transformity of a product within a complex production process. That's because people is used to abuse of precomputed transformity tables and thus miss the original definition of emergy as well as the real power of this concept. Let's highlight briefly the problem: abusing of transformities tables can lead to questionable accuracy of emergy indicators! Indeed, a transformity is always dependent of the specific you are speaking. Picking a transformity from one system and injecting it in an other implicitly assumes that the products you are speaking are made and exactly the same way and that they are independent. If not then you will start forgetting or double accounting some emergy. Remember the rules stated by H.T. Odum in [5] chapter 6. A simple example is that 1 Kg of Brazilian bananas will have a much more greater emergy if transported in Europa than it has in Brazil because of the transportation energy. If you miss this concept then emergy won't defend anymore the need of economy re-localization. If you would account for the transportation, then you would make a step in the right direction. But is that always sufficient and credible as such? Imagine that you have to compute the emergy of a cake made of those bananas and the emergy of Brazilian soy. But let's imagine you can to it fully because emergy data are correctly collected: imagine that you can deeply investigate at the specificity of your case study because models of many products or services are available on the Internet. Transporting 1 Kg of bananas require a very small part of the boat technology. When you'll later add the emergy of that banana to make a cake, you could have to add the emergy of the bananas with that one of the

Also the basis of the the most relevant algorithms used by the pioneer H.T. Odum for systems simulations have already been coded during the 10 months of intensive programming in 2004 as it will be discussed later. But in order the project to keep improving using others academic research projects (you are very welcome to contact the authors to participate), we created EmSim as an extensible democratic project: so it's very modular and which includesing the most relevant algorithms used by the pioneer H.T. Odum for systems ecology simulations. Because the task is huge and poorly funded, EmSim has been created as a democratic project: its source code is open (GNU GPL license) and accessible on the Internet through a Concurrent Version System (CVS) while providing all the communication facilities of modern software development like forums mailing lists and bug tracking... Moreover, it is programmed in Java [2] so that it can run on line and be easily coded on any computer under any ; for free.

This article deals with the main features of EmSim. To explain the algorithms, we will often refer to the traditional modeling done in systems ecology. However, we are unable to teach all the basics in a short paper. Also for further details, we advise the reader to look at [3] and [4] for causal modeling, energy language and its dynamical meaning, and at [5], chapter 6 for the basics of emergy algebra.

EmSim allows different levels of use :that directly reflects the main steps of ecological/economical modeling. Those steps can be summed up as: searching laws, representing the model, mapping available information into the model, simulating the kinetic of the model, qualifying the thermodynamical efficiency of the model. So this means respectively:first, in order to deduce laws in phenomena, you have to tell the spatial and temporal boundaries of your system as well as stating the granularity level at which you start accounting things (eg. accuracy). You should also find a pertinent basis of parameters qualifying the state of your system (eg. state vector). better perform inferential statistics Then, you can investigate both qualitatively (what is required to make what resource) and quantitatively (the steady state energy intensity or the dynamic) the causal links between phenomena by means of intuition, observation and even inferential statistics.he domain limits as well as the degree of aggregation to be used for system structures and functions.level. Of course EmSim can't help yet there since this analysis work is quite subjective and intuitive remains the responsibility of the user job. 1. Then it becomes possible to draw a causal network as an energy systems diagram that explicitly documents qualitatively and also possibly quantitatively (using classical thermodynamics) the causal pathways so that specialists from various fields can discuss it. EmSim is a diagram editor that performs this function, see part 2. 2. Models should be easy to share and remain open for further modifications, improvements and review to make it possible to achieve consensus within the scientific change so that they are catch credibleility.. This can only be achieved by means of standardized data, that's why EmSim store every model as XMLexchange models efficiently, we needs useful standards. To learn know how EmSim structures the systemic knowledge and stores it as XML, , read the part 3. 3. Once the network of causality is established, you can study both the kinetic and the thermodynamics of the model. The kinetic of such a system is well defined by a set exactly closely the system under variable conditions and try to detail the functioning of the causality by means of detailed mathematical expressions within a set of simultaneous differential equations. In [3] and [4] Odum gives various examples of how the “” can be translated into a system of differential equations. EmSim automates this translation and makes it possible to plot the evolution of the system given an initial state, see part 4. 4. Concerning the thermodynamics, the idea is quite simple: at any given time, whether at steady state or not, it can be interesting to check how much emergy a resource requires to be maintained in a given system, what's its transformity in that system or even what is the structural quality (the level of organization might be qualified by the empower) of the system emergy storages and flows.analysis. For instance, you can compute the empower flux through the network or the energetic cost of maintaining an emergy storage of aone resource.. EmSim is able to propagate the emergy in large and complex networks, see part 5. 5. Finally, EmSim is incomplete and has some drawbacks, but it can be perfected with additional work: see part 6.

2. EMSIM AS A DEDICATED DRAWING TOOL TO COMMUNICATE WITH ENERGY SYSTEMS DIAGRAMS

Basically, EmSim is an Odum Energy Diagram editor allowing to save diagrams with plenty of structured informations or even export them as pictures. Instead of spending time re-inventing the wheel, our philosophy has been to make the most of object oriented programming and to derivate the graphic EmSim features from existing standard tools, see figure 1. For instance, the graphical user interface isn't fully specific so we made it by deriving tools for graph theory. Thus, EmSim uses objects from the JGraph library and invokes actions from the JGraphpad interface, see [6]. We mainly sub-classed some objects and we also included a toolbar for the “energy systems language symbols”, cell rotation ability, split (un)aggregated view, Portuguese translation and so on. Finally, commercial softwares like MS Visio are no longer required to draw energy systems diagrams of the topmost quality, instead these model illustrations can be created intuitively using the EmSim diagramming tools.

Thus, when you open EmSim, you get a multi-document user interface that allows you to intuitively draw complex and elegant graphs simply by dragging and dropping the required components from a dedicated toolbar. We can mention the following special features among others: a multilingual interface, group / ungroup, copy / paste, unlimited undo / redo, curved (spline) connectors, unlimited color choice, transparency in gif pictures, rotations… full advantage of from the object oriented language (Java?). EmSim uses objects from the JGraph graph handling library and invokes actions from the JGraphpad interface, see [6]. We sub- classed some objects and we also included a toolbar for the “energy systems language symbols”, cell rotations ability, split (un) aggregated view, Portuguese translation and so on. Finally, commercial software’s like MS Visio is no longer aren’t required anymore forto design and drawing intuitively energy systems diagrams of the topmost quality, instead these model illustrations can be created intuitively using the EmSim diagramming tools.with the topmost quality.

3. EMSIM STRUCTURES THE KNOWLEDGE OF SYSTEMS ECOLOGY STUDIES, MAKING IT THE IDEAL USER INTERFACE OF A WIDER

3.1 Ecology researchers need an information file format to communicate more efficiently

Today lots of researchers using systems ecology are spread all over the world and are often working separately. People using emergy seldom reuse the work already done on input/output exergy, cycle assessment or ecological footprints. However, lots of data could be more efficiently revised by sharing information among these various methods (see [7] and [15]).

It seems clear that systems ecology researchers lack a communication format through which they can share their data across the Internet. Indeed, often published papers don’t provide all the elements required to get the results! For instance, the emergy methodology is based on performing energy accounting upon a causal weighted energy graph. Many diagrams are drawn in a manner that makes it difficult to determine whether an outflow is a split, a co-product or even an hybrid case. Consequently, peer review, reuse, and improvements of existing work are very low.

What if people could exchange standardized extensible data on energy systems that everyone could share just like people already do in other fields like Computer Aided Design? What if the knowledge about emergy was collected in a data bank for use in deducing statistical laws for the considered systems? Just imagine how this could improve the emergy methodology, its accuracy and its credibility… After drawing an energy diagram with EmSim, you are free to enter all the information you judge necessary exactly where it should be thanks to the XML formalism! Associate the energy or matter flow directly with the corresponding pathways. Explicitly mention the allocation rule you use at each divergence of flow. If a pathway has a flow proportional to the gradient of the forces provided at its two extremities, then simply choose the corresponding sub-model for that pathway! If a parameter that you enter has a Gaussian distribution, then associate such a distribution with the parameter to allow someone else to perform a stochastic simulation in the future…

3.2 The data structure handled by EmSim The data structure handled by EmSim essentially consists of a graph made of nodes and edges that respectively stand for products and energy pathways. The user interface allows to associate information with each element of the graph. Thus, each cell or node keeps track of the information about its view, but also about the graph structure and about its mathematical model; see Figures 2 and 3.

3.3 Storing efficiently systemic data as XML The data structure generated by the user has to be persistently saved (serialized). The key strategy is to use a special file format relying on: • Hierarchically organize each piece of information, • Describing with markups all piece of information you store, • Not being dependent of the application you use, • Being human readable, • Being expendable while keeping compatibility with all applications. So EmSim rigorously saves the graph structure using the jgx format, that’s in fact an XML application (see [9]). This means that the way we write the file as a text respects some standard conventions making it XML and thus very easy to plug in to other applications like a data bank or data . In this way, you could perform statistical analysis on emergy data and you could also make emergy data inter operable with widely collected life cycle assessment data! The XML files are based on two powerful standards: the graph structure is encoded with respect to the GXL (Graph eXchange Language) standard while cells models and parameters are persisted and retrieved using the Java XMLEncoder and XMLDecoder objects.

None of the information you enter (either by means of the graphical user interface or with a text editor) will be lost. Indeed, even if the file format of EmSim files were to change a bit, then an XSLT transformation will bring back to live your old data. XSLT could also be used for interoperability with other simulators!

4. EMSIM TRANSLATES CAUSAL ENERGY DIAGRAMS INTO ORDINARY DIFFERENTIAL EQUATIONS (ODE) SYSTEMS

4.1 Statement of the problem

Systems Ecology [3] and Computer Mini models [4] are two of H.T. Odum’s books that include lots of “energy diagrams” representing the structure of the differential equations used to model real systems. H.T. Odum explains that he created this graphical language to allow specialists from various fields, ranging from sociology to ecology, to collaborate more efficiently in modeling human activities. However, very few people seem to understand how this graphical language works and we could not find any synthetic explanation!

By carefully reading the two books mentioned above, we found that the Odum graphical language for dynamical systems was something very powerful and very universal, because the diagrams are directly translatable into a mathematical structure. We also found that a similar language (bond graphs, see [10]) is already widely used in other scientific fields. For example, car-makers use bond graphs to get one simple model when dealing with various technologies.

First, by analyzing lots of examples, we intuited the laws that make up the kernel of the language. The difficulty here was to re-formalize the rules so that they can be applied in the most general manner possible without requiring many specific cases. Second, the task was to implement that kernel inside EmSim.

Finally, the Odum energy language is suitable for modeling dynamic systems where you know the functioning (meaning the differential equations) and a particular state (the initial conditions) and you want to simulate the evolution of the system with time or space. Such systems are called ordinary differential systems. They are often nonlinear. You can only model the variation of one variable (usually time) by opposition of partial derivative systems that require more advanced mathematics, like finite element integrations.

4.2 How EmSim converts causal energy diagrams into ODE equations

4.2.1 General equivalence between dynamical equations and graphical symbols in EmSim: Given a causal energy diagram, it is possible to get the equivalent differential equations describing the evolution of the system. Like other simulation software, EmSim is able to perform such a translation. Here are the basic equivalences:

• real <-> “energy diagram” model • set of state variables <-> set of storage symbols • first order variation of a state variable <-> set of edges connected to a storage • one term of the variation <-> the flow inside one pathway

When EmSim recursively rebuilds the differential system by “recursively reading” your diagram:

• For each capacitive storage symbol, it writes: . (flow conservation) • For each inductive storage J (little used in ecology), it writes: . (force equilibrium) Then, the simulation will consist of estimating at each time, t, the quantity for each storage, and plotting the evolution. This is achieved using a classical 4th order Runge- Kutta method; see [11].

Source symbols behave like storages but their variations are neglected compared to other variations within the temporal scope of the simulation.

4.2.2 How flows along pathways are computed in dynamical energy language: The flows into and out of a node component) determine the first order variation of any state variable. So to get the freedom of modeling any ODE system, the user should have the freedom to generate any flow inside a connection. This is possible because the flow inside an edge is computed in one of several ways, depending on the following cases: • Flow imposed by the source component of the pathway • Flow demanded by the target component of the pathway • Classical pathway: • Sensor: ; but the information is transmitted if required • Gradient pathway: • Gradient valve pathway:

4.2.3 How forces at the component connections are computed: In the last section, we saw that flows inside connections determine the terms of the expression for the first order variation of any state variable. However, it appeared that lots of flows are driven by forces. So once again, to model any ODE system, the user should have the freedom to generate any function of forces. This is possible because some nodes other than a storage allow building interactions. Basically, a storage provides a force proportional to its quantity at the time t. Examples are a tank were the bottom pressure is proportional to the height of the water; other storages are electrical capacitors, mechanical springs, pressure and so on. Then, you can modulate and combine those forces using additional graph components: • Regulations (2 ports, unary operator) 3

3 Inforce stands for the force provided by the component from where the flow is coming whereas Outforce is the force provided by the current component. • Interactions (3 ports, binary operator) • Composite: a complex component. For instance, in a transaction component, one flow is the slave and the other is the master.

One of the great advantages of EmSim is that it allows the user to specify the functions used by customizing regulation and interaction components. EmSim uses the parser of the FrAid mathematical library to evaluate customized expressions made of standard mathematical formulae (e.g. polynoms, trigonometry, logarithms, boolean expressions...)

4.3 Additional remarks about dynamic causal modeling When a storage has to experience a discontinuous variation, we can still model it with EmSim, using the switch component that will cause the slope of the variation to be “almost infinite” at the discontinuity.

Contrary to behavioral modeling, causal modeling lets the user accomplish the entire job of decomposing a real system to get a causal description of that system using only differential equations of first order (see Figure 4). But non-causal modeling algorithms (inferring the causality depending on the solicitations) are too complicated for an open source perspective. And moreover, the causal modeling is much more suitable to perform emergy analysis latter.

In order to have each component to have its own behavior (for instance when the force at a connection is asked by another component to supply a flow), we use the polymorphism of object oriented languages. This means that each cell becomes specialized along an inheritance tree. Thus, for instance, a variable source inherits from a constant source but adds the possibility of entering as a function of the time.

4.4 Examples

To make it easier to understand how EmSim works and also for a pedagogical application at the LEIA (you can of course also use it), we provided a huge set of exsamples. Almost all simulation models from the Computer Mini model book have been entered as examples and can be used to teach either the models or EmSim. These examples also validate the ability of EmSim to represent dynamical systems since we found the same results as those given in the book. See Figure 54 for instance.

5. EMSIM PROPAGATES EMERGY ACROSS LARGE AND COMPLEX NETWORKS

5.1 Statement of the problem

An other very important issue when dealing with sustainable development is whether or not an alternative production process requires less energy. The emergy methodology is a great modeling tool to study it. Concerning the emergy methodology, EmSim mainly focuses on the fact that: • An arrow signifies: that one component is required to produce another component, at least within the considered system (causality). • Flows of can split to support various processes. In this case the required emergy for each process is split in the same proportion as the energy (or often matter) flow. An example would be a model where the oil production is partitionned to support various activities. • Some reactions are structurally co-productive. An example is a process breaking a molecule in two parts. In this case, once you determine the emergy (by properly summing all the vailable energy inputs) needed to provide one co-product, the other is an automatic consequence: it would be a double accounting error to say that this other co-product also require some more emergy to be provide count double account some more the required emergy again to provide the second co-product.d. Thus, whenever the production of a resource requires several products that are in turn co-products of the same process from sometime before, then only the largest emergy contribution should be counted (see [2] and [8] for more details). This issue explain why a computer aided monitoring is absolutely essential.

Considering these simple rules of accounting for emergy (the last one is however widely misunderstood because it makes the emergy not conservative), As usual, EmSim build up the formal expressions by propagating the emergy across very complex networks. Then it asks the user to fill in the minimum required inputs (independent emergy sources and splits ratios). Moreover it highligths the different pathway contributions on the graph for those who don’t believe!

Compared to the matrix approach [13], EmSim produces the same results but is much more user friendly an information system oriented.

5.2 The emergy track summing algorithm By performing iterations on the graph and by using tree data structures (for the energy paths), EmSim performs the following algorithm:

Obtaining Emk,i: First, we apply a superposition principle, assuming that each independent source contributes independently to the emergy of each product. If S is the ensemble of all the N sources, and xi, i ∈ {1, ... ,m} are the nodes of the ensemble X, we have:

[Fra Emk, ∀ xi∈ X, Em(xi)= me1] i

indep where m is the number of simple paths and where Emk,i,j are the independent emergies from Sk propagated to xi through the path pj. The following section explains

indep how to calculate Emk,i,j :

Single path of emergy propagation: obtaining Emk,i,j: along each path pj, we propagate the Emergy brought by the source Sk from node to node while respecting the allocation rules (see section 5.1). Also notice that for a single path, the emergy to a node xm is simply equal to emergy of Sk brought by the inflowing pathway a | target(a)

=xm.

Pathway historic bookkeeping : At the same time, an history of the nodes encountered by an inflow Sk and the emergy from Sk remaining at each node is kept.

indep Obtaining Emk,i,j , avoiding double counting: When reaching the node xi, we may

indep have to filter out some Emk,i,j to obtain the independent emergy contributions Emk,i,j . Indeed, when we were faced with a node that produced co-products, we propagated the emergy normally for each simple path pj. But now we should look further at the logical implications of this action.

The first time the energy from Sk crosses a co-productive node --- let's call it xp --- on the way to the target node, xi, the following conditions exist: (1) the emergy from xp is required to support the xi resource and (2) the current emergy at xi from Sk is the emergy that was required from Sk to support xp. Once the energy from xi --- equivalent to emergy at such an elementary level --- is provided to xp the resource xp exists (assuming that this algorithm will also be done for the other independent sources) and thus all co-products resulting form xp exist. Thus, no more emergy from Sk is required to support them!

So then, each time we cross the co-productive xp node when propagating the emergy to xi through other paths pj, we should assume that the required energy from Sk to support xp is already counted!

Filtering out double counted energy: The next question is: when several simple paths from Sk to xi cross the xp node, which one carries the right emergy required to support the xi resource? Each of the simple paths carries the emergy from Sk to support xp. But after xp, depending on the paths of the co-products, more or less emergy will support xi.

• If we only take into account a path not bringing the maximum of those emergies

Emk,j, then it will lack the energy needed to support xi in (indeed another path requiring more energy should be supported).

• If we sum several of co-productive paths, then it's clear that some energy is double counted.

• On the contrary, if we only account for the maximum energy required by any of the

co-productive paths, then all co-products resulting from the xp node exist and no inflow is double counted. So that's the solution!

indep So finally the ensemble of the Emk,i,j is built form the computed ensemble of Emk,i,j by filtering out double counted emergy: when two paths bring emergy from Sk to xi while passing through the same co-productive node, then only the maximum emergy of

indep the two enters into the ensemble Emk,i,j .

5.3 Examples

Again, to demonstrate the ability of EmSim to propogate emergy through a network, some examples are also provided as demo files. But very few studies are complex enough to show the full range of EmSim’s abilities, so we only validated our algorithm on a few famous complex examples. Especially, Figure 65 is a screenshot of an example from H.T. Odum ([5], page 100), also mentioned by C. Giannantoni ((14] page 30).

5.4 EmSim experiments could be used to check the consistency of the maximum empower theory.

The work of C. Giannantoni [14], leads us to mention that empower studies could overcome the limitations of current emergy based indices. Indeed emergy indicators often suffer the followings limitations: • Low credibility because they take as inputs many sources that are de facto assumed independent and whose emergy simply come from transformities precomputed tables without special investigations to see if those transformities can indeed be injected in the considered case study and without serious care about confidence intervals. • source independentnce verification (but information systems like EmSim could pay off!)Long run postive or negative feedbacks of the investigated product upon their own production is barely modeled nor considered by those indicators. • Huge sensitivity to the structure of the chosen model (see [7] and [15]). Especially, saying that products are fully co-produced is often a huge and common approximation since often only small parts of those products are really co-products. Not to mention this is also affected by our knowledge of the process. • Narrow validity scope. Indeed, emergy indices are relative measures. They aren’t absolute because they are issued from computations performed on conceptual models and are impossible to verify most of the time. So before comparing emergy- based indicators for two alternatives, one should ensure that the two models are comparable (it should even be the same model but connected differently in the ideal case). • Some people have made a theory of value based on emergy. We find it naive to apply it directly to human activities, because only relatively old have faced an optimization through evolutionary laws that leads to the highest efficiencies at maximum emergy flow (see Odum [3] page 253 and [7]). Human society however has plenty of optimized processes that waste the emergy!

While it could inherit some of those limitations, the empower of a system seems to be a more objective measure. Nevertheless, it is still dependent on the granularity level of the modeling. For further investigations, EmSim could be the ideal platform to automate empower computations according to the discrete formulation made by C. Giannantoni in [14].

Moreover, since it was quite easy, we made EmSim capable of propagating emergy in a network under dynamical conditions, i.e., where energy or matter flows are driven by the differential system associated to the causal energy diagram. C. Giannantoni suggested that the results could be wrong and it would be necessary to introduce emergy accumulation terms in the equations as he did in [14]. We don’t really know about this issue but some research could be undertaken using EmSim.

Finally, C. Giannantoni suggested that dynamic empower studies should involve fractional calculus and “incipient derivative operators”. EmSim could build fractional derivative equations by reading graphs using similar recursive programming as that used for integer order derivatives differential equations generations. However, concerning the numerical simulation of those equations, it would be quite hard because we would need to implement the Grünwald formula [16]. Moreover, we never saw any application of this method in empower studies so far… 6. EMSIM IS FULLY OPEN FOR FURTHER DEVELOPMENT BUT REQUIRES NEW RESEARCH PROJECTS

Commercial simulators, while they’re handy, aren’t a long run suitable alternative, because they their code is not available for modification by the emergy community. Even if companies use the bond graph paradigm which can be derived into the Odum Energy Language behavior, as long as companies won’t use the emergy concept, its specific and complex algorithms won’t be available in any commercial software. Not to mention the Brazil’sian lowest annual income for instance…the public money saved or better allocated when using free alternatives...

In contrast, EmSim has lots of advantages, foremost is that its code is open and freely available at: http://emsim.sourceforge.net. Sourceforge also provides a discussion forum for EmSim at: http://sourceforge.net/forum/?group_id=102093. And there are also many tools for international team development: CVS, bug , feature trackers and so on. If you are a researcher, then you could probably tutor a Master or even a PhD to improve a part of EmSim. Else you can simply correct a bug, provide a new translation, or provide more examples. Desired tasks should be listed here: http://sourceforge.net/pm/?group_id=102093.

And finally, when emergy reaches its maturity, EmSim could become the perfect user interface to collect and share and inter-connect systems ecology data. For this to work, systems ecology researchers would have to publish their studies as EmSim files and an appropriate server would have to collect them before data could be handled and transformed into knowledge though data mining techniques (see [17]). Currently those files can be sent to the authors, they would be put on line as demonstration files.

References [1] Sun Microsystems. Java, a platform independent and object oriented programming technology. http://java.sun.com. [2] Raphaël Valyi. the emergy simulator open source code project. http://emsim.sourceforge.net, 2004. [3] Howard T. Odum. Systems Ecology An Introduction. John Wiley & Sons, 1983. [4] Howard T. Odum Elisabeth C.Odum. Computers Minimodels and Simulations Exercices for Science and Social Science. Center for - University of Florida, 1994, extracts provided as samples file with EmSim at. http://emsim.sourceforge.net, [5] Howard T. Odum. EMERGY and Environmental Decision Making. John Wiley & Sons, 1996. extracts at http://www.unicamp.br/fea/ortega/htodum/emergyaccount.htm and http://www.unicamp.br/fea/ortega/agroecol/emergy.htm. [6] The jgraph team. the jgraph and jgraphpad open source projects. http://www.sourceforge.net/projects/jgraph, 2004. [7] Jorge L. Hau Bhavik R. Bakshi. Promises and problems of emergy analysis. Ecological Modelling, 2004. [8] Jorge L. Hau Bhavik R. Bakshi. Expanding Exergy Analysis to Account for Ecosystem Products and Services. Ecological Modelling, 2004. [9] XML.org. XML formalism and tools. http://www.xml.org. [10] bondgraphs.com. About bond graphs. http://www.bondgraphs.com, 2004. [11] William H. Press Saul A. Teukolsky William T. Vetterling Brian P. Flannery. Numerical Recipes in C. Cambridge University Press, second edition, 1992. [12] Ivaylo I. Iliev. Fractal aid, a java language to process and plot mathematical data. http://www.fraid.org, 2004. [13] Dennis Collins Howard T. Odum. Calculating transformities with an eigenvector method, 2000. [14] Corrado Giannantoni. The Maximum Em-Power Principle as a basis for Thermodynamics of quality. Servizi Grafici Editoriali snc, 2002. [15] Robert. A. Herendeen. Energy analysis and emergy analysis: a comparison. Ecological , 2004. [16] Lubomír Dorèák, Ivo Petráš, Ján Terpák and Martin Zborovjan1. Comparison of the methods for discrete approximation of the fractional-order operator. Acta Montanistica Slovaca 2003. http://actamont.tuke.sk/pdf/2003/n4/25dorcakpetrasterpakzbo.pdf [17] Weka: Data Mining Software in Java. http://www.cs.waikato.ac.nz/ml/weka/

Figure 1: obtaining the required EmSim specific features by subclassing standard java open source objects. Figure 2: information stored in an EmSim cell

Figure 3: The graph topology using the JGraph [6] open library

Figure 4: Differences between behavioral and causal modeling (here with Simulink) Figure 5: EmSim successfully simulates the WORLD model from the Odum book[4] Figure 6: EmSim successfully computes the emergy (here 28250 to A) on this famous Odum example from [5], page 100.