EPC vs. BPMN - the perfect flamewar

I also think so. And Bruce Silver says;

... In fact, BPMN barely touches on what the activity is or

who performs it. ...

in BPMN Method and Style.

I sometimes need BPMN which is embedded in ARIS House.

I tried it last year.

blog.livedoor.jp/hikiko_mori/archives/1238708.html

(Sorry. It's Japanese only)

Hi Yoshinori,

the sad thing is that many business users were hoping BPMN 2 will bring this kind of linking to other elements of an enterprise architecture. Unfortunately, the development of BPMN 2 mostly focused on making BPMN executable. This is an important topic, but probably not what business users were hoping for.

Hi Sebastian,

I agree with your remarks. I think the design of both language has been strongly influenced by the issues you have point out (ARIS is more suitable for business communication, while BPMN is strongly oriented to implementation issues). Although they have been developed for different purposes, i think it is a valid effort to compare them so that one can know (in advance) which language to use in each situation.



Particularly, in my experience with ARIS and BPMN (with respect to the behavioral perspective), i've noticed that learning ARIS was much easier than BPMN. I believe that this difficulty can be explained by the high amount of concepts of BPMN which makes it harder to learn (BPMN is too expressive!). In the counterpart, although ARIS has been easier to learn, in some situations, i had the need to model complex behaviour and the language had limited me :(

(to solve this problem, i had to extend the language in the ARIS filter with the introduction of some concepts borrowed from BPMN, such as multi-instance activity).



With respect to the linkage with the other viewpoints, EPCs are really more expressive than BPMN (in particular, the way how the ARIS toolset organizes the other information about enterprises around the EPCs, such as, the assignment relation of the activities with its respective FADs, significantly improves the navigation among the models).

I also agree with your conclusion that, in modeling activities, one must have in mind why he/she is producing some model. In particular for process modeling languages, i believe that there is a trade-off between the expressiveness (in terms of the level of granularity of information that one can capture, or in rigor of capturing complex behavior) and the ease of use/learn. In fact, the most important (in my opinion) is to make the criteria for comparing both languages explicit. For instance, some works in research investigate the quality of process languages in terms of rich real-world semantics (what includes the assessment of these languages in terms of a reference ontology, as in this paper - SANTOS JR., P. S. ; ALMEIDA, J. P. A. ; GUIZZARDI, G. An Ontology-Based Semantic Foundation for ARIS EPCs. In: 25th ACM Symposium on Applied Computing (Enterprise Engineering Track), 2010, Sierre, Suíça. 25th ACM Symposium on Applied Computing (Enterprise Engineering Track), 2010). An analysis in terms of these expressiveness criteria would certainly produce different results about the better language for process modeling than your analysis. Then, besides concluding that ARIS is better than BPMN (or vice-versa), before, the criteria for judgment must be explicit!



Well, i write too much :) I hope that my opinion had contributed to enrich the discussion :)

Ps: If it could be possible, I would like you to post some remarks about the relation between ontologies and EPCs.

 

Hi Sebastian, thanks for this highly useful explanation.

You mentioned that your BPMN vs EPC diagram is based in BPMN 1, and also that BPMN 2 mostly focuses on making BPMN executable, and not in a better linking to other elements of an enterprise architecture. Anyway, how would your BPMN vs EPC diagram change if updated to reflect BPMN 2?

Thanks.

@Evellin: The term "ontology" is a beast, because everyone has a different understanding of it. I have done research in the past to transfer the content of EPCs in the sEPC ontology. So in that sense the "ontology" is just a tool to make the content of an EPC accessible to machine reasoning. There is also an ontology for BPMN. So EPC/BPMN models contain a lot of semantics, but those semantics are not explicit. Here, ontologies are a great tool.

Hi David,

so far I had no time to do a detailed analysis about how the picture would change for BPMN 2. Therefore, I wrote that the analysis was only done for BPMN 1. As far as I have seen I don't think the picture will change. BPMN 2 has not introduced any new elements, which you could use to express additional links to other EA dimensions.

Interesting discussion, but as Evellin seems to suggest the criteria under which a particular notation is 'best' depends entirely on the objectives of a modeling program or project. So for some the conflict of choice between EPC and BPMN does not exist as their relevance to modeling objectives has been defined. For example we have modeled workflows using BPMN as detailed children to high level EPC business processes that conform to our general EA guidelines.

I have recently moved from a complex IDEF0/BPMN modeling environment to a simplistic EPC only methodology and I’m in the process of promoting BPMN as a valuable asset to our architectural strategy.

Hi Marcus,

that is exactly my point, too. There is no best language at all but instead it depends on what you are going to do. I would say if you are targeting process automation, go with BPMN. But if you are just interested in conceptual business process modeling, consider EPC.

Sebastian,

I like this analysis, clearly people should understand what BPMN is best used for and when other modelling notations (e.g.EPC) are stronger. The point about BPMN and linking (or not) to the other dimensions was very apparent to us from a simulation standpoint....there was no information on who performed the task, or even reference to the piece of work that flows through and its characteristics. Modelling notations also often omit the 'time dimension'...which of course is crucial for our simulation analysis.

Knowing what you are building the model for is so obvious in hindsight, but often gets missed in my experience.

I found this article a while ago and liked the point it made about the use of BPMN.

http://www.it-director.com/blogs/Mark_McGregor/2009/3/free_mapping_modelling_survey_repo_.html 

Hi Olaf,

could you please share the key points of the report, because not everyone would be willing to sign up to get access?

I agree with Marcus in the form that he use every kind of model. EPC for high process views and BPMN for detail.



Sometimes we need custom  BPMN docs, we need explain it from business viewpoint, and many documentation about it it's mixed with a lot of IT/SOA vocabulary,

Igor

Hi Igor,

I guess your first point is based on the old papers about EPC. It is true that those old papers don't have many extra symbols defined. But today nobody is using this old kind of EPC anymore. Not just ARIS, but also other tools have extended the EPC notation heavily. I don't see a point why I should compare this old definition of EPC to BPMN if nobody is using it.

About the other points, I think we share the same view. EPC and BPMN shall be used for different modeling tasks and you will need a clear guideline when to use which notation. Still, it is important to also do a direct comparison, because there are many vendors trying to fool their users by saying they now got BPMN, which is much better than EPC.

I totally agree.

I've communicated on this since several years and I found BPMN supporters often very defensive on this subject thinking that we can do anything with BPMN.

With BPMN you can model almost anything, but the same can be said for EPC. It's not that one method is better than the other or that one suits a higher level of process modeling than the other. It's that an architectural group or team needs to define a modeling hierarchy that best fits their and their customer's objectives and preferences.

For example we've found our engineers prefer to see their work modeled in BPMN because they can track tasks against application (lane) and domain (pool).

Our project managers prefer EPC as it's easier (for them) to identify task/function against responsibile individuals and kpi's.

This has led to a hierarchy of EPC/BPMN based on customer preferences.

I'm looking forward to an increase in translation capabilities between EPC and BPMN in ARIS (hint, hint).

Hi Michael,

> Not BPMN supporters but BPM vendors are very defensive.

Well, at least at IDS we are not trying to push any notation. Instead, we try to focus on what is really needed in a modeling situation. If this means using BPMN, fine for us. The same goes for EPC and other process notations such as UML activity diagrams or flow charts. Notations are just tools and no religions.

When a methodology is assessed base on what it was originally created to solve, then only an accurate view and use of it make sense. Why use BPMN to model a concept? The motive is invalid in the first place, of course in this sense the underlying methodology will serve no purpose.

BPM/BPMN vs ARIS/EPC, we find that both serves well when come to express business LOGIC. We believe at the end of the day, it is the correct LOGIC that is matter and not really what physical notation is it.

Frankly, in many cases, a basic flowchart + quick note can do a lot more and that was how COBOL system developed 40 years ago still out run many fancy system today. Rather, vendor should spend efforts in ACCURATELY automate and integrate the output (regardless of BPMN/EPC) to produce desired OUTCOME - keep the process neat, clear and sharp. Hell at the end of the day, it is the software component that make the actual link, not circle on a paper!

Hi Igor,

Thank you very much for the post. I hope that in the near future you can further detail on how BPMN should be used in conjunction with EPC.

That will be really useful.

I would like to post you a question that would help shed light on a subject that is critical on Business Process Modeling.

I hope you could help answer this question if you believe it is also helpful to other members of the community.

How would you compare/relate EPC and BPMN with the methodology proposed by the MIT Process HandBook, that introduced concepts of generalization/specialization of business processes (analog to concept of inheritance of object-oriented programming) and managing dependencies from coordination theory.

In the MIT Process Handbook there are some dependencies established for resources and activities and proposed coordination mechanisms.

The three major types of dependencies : Flow, Sharing and Fit.

According with the paper: "Tools for inventing organizations: Toward a handbook of organizational processes", Published in Management Science 45(3) pp 425-443, March, 1999, published by Center for Coordination Science Massachusetts Institute of Technology: 

 - Flow dependencies "arise whenever one activity produces a resource that is used by another activity.";

- Sharing dependencies "occur whenever multiple activities all use the same resource. For example, this kind of dependency arises when two activities need to be done by the same person, when they need to use the same machine on a factory floor";

- fit dependencies "arise when multiple activities collectively produce a single resource. For example, when several different engineers are designing different parts of a car (such as the engine, the transmission, and the body) there is a dependency between their activities that results from the fact that the pieces they are each designing need to fit together in the completed car."

The dependencies of Flow can also be subcategorized as :

• Prerequisite ("right time") 
• Accessibility ("right place") 
• Usability ("right thing")

Sets of coordination mechanisms could be used to effective managing those dependencies.

Please advise,

Thank you very much,

Carlos Oliveira

Very usefull discussion. I just know that EPC and BPMN is different. In my organization, we map our target operating model (TOM) using EPC.

But lately, they slowly change to BPMN

Deceptive simplicity,

It is often happen that a flow diagram that looks clear is only a use case, showing only one possible way of process execution scenario.  For sure a flowchart that is covering all possible cases will look unclear.

BPMN (if it is an executable model and not a poor drawing) will most probably cover all possible ways of process execution. I say most probably, as model is thorough not because we use BPMN and not EPC notation but as a result of development methodology. BPM design includes a stage called verification where we can find others not yet found scenarios. Of cause with EPC one can also make a thorough model, but have never seen such a quality.

I have my consideration of EPC process simplicity. Let us discuss a term FUNCTION. We all talk that process approach is better than a functional one. Why do we use this term and not WORK, ACTIVITY, TASK, etc.  I have a strong feeling that this term is more than just wrong wording. Let me explain.

We often use a process modeling to develop a “functional” IT system. Functional mean that a man play an active role at the execution phase and system allows him perform his functions. To develop such IT one need a list of all tasks of a participant. From my point of view term FUNCTION means a collection of works within a process, organizational unit, company that are performed by a user. For example, one participant must do several works within one process. So we call this set of works his function. Same for OU and company. As a result we make a system that can perform all works that a participant is allowed to do, but a man decide the order of operations and time to do a work. Let me call such IT “Allowing System”. Easy to see that in “Allowing System”, where a man decide, it is not necessary to model all possible execution paths but only functions to perform.

With a help of BPM we build a system that play an active role, that guide a participant through a process, forcing him perform an action in time.  Let me call such IT “Guiding System”. In “Guiding Systems” we have to have all works to be performed as well as all routes of process execution. That makes BPMN more comlex.

I am far from saying that BPMN is better than EPC. Both are more or less equal. What I want to declare: Let us be careful when talking about a simplicity, it is deceptive.

Regards,

Igor

Hi everyony and special thank to  Dr. Sebastian Stein. I know you may see this article, but I find it very useful too. Check it out:

http://bpt.hpi.uni-potsdam.de/pub/Public/OryxResearch/TransformEPC2BPMN.pdf

Have a nice day.

Dear all,

speaking of literature on EPC-to-BPMN, in my dissertation (soon to be published by Logos  )   I found that "there are several papers implicitly referring to a EPC-to-BPMN transformation (Recker & Mendling 2006; Ouyang et al. 2008; Tscheschner 2008; Ouyang et al. 2009). However, they do not provide any detailed mapping rule as if considering this a trivial task. Some approaches shift transformation to XML-based presentations to avoid the informality of graphical models. They presuppose a meta model mapping (Theling et al. 2005), which cannot be conducted since OMG has not yet provided an official BPMN meta model (originally expected for BPMN2.0) (Allweyer 2008, p. 9)."

Best regards,

Katrina

Hello community,

I am a distant student of the university of applied science in Merseburg. I would like to raise a survey within the scope of my study module process organization. The results should flow in my seminar work which deals with the subject of the concept ARIS.

In particular I would like to compare both models of the EPC and BPMN and light them up more exactly. 

For this reason I would appreciate if you take part in my survey.

http://www.q-set.de/q-set.php?sCode=QHAKUHHJFGUB

Thanks a lot for your time.

best regards

Yvonne Pietrowsky

I would like to thank all members who have already taken part in my survey. To those they have not taken part yet, the survey is not concluded yet and the link is still active.

 

Of course I will publish the results in this community

Yours sincerely

Yvonne Pietrowsky

This article is very interesting and useful. I come from the BPMN community and I  would like to know if there is posible to model exeptions or programming language elements with eEPC. Right now I am writing a comparison of five BPMLs and I need to consider this criteria as well.

Thanks in advance!

Hello Dr. Stein and everybody,

a very interesting discussion for sure. I am wondering, whether you know something of an existing set of  BPMN constructs for those concepts belonging to the extended EPC notation such as risks,. Do you have any references on this? I appreciate your help on this matter. Regards,

Cheers,

Alexis Ocampo

Hello Dr. Ocampo,

it is not possible to extend the BPMN 2 notation itself in ARIS (like adding risk objects to the diagram). However, you can assign to each BPMN activity a function allocation diagram, where you can model those extra artefacts. This has the advantage that your main BPMN diagrams are standard conform and easy to read, while you don't have to give up on the expressiveness needed for enterprise modelling.

A big thank-you to all who have taken part in my survey.

Of course I would like to share the results with you.

As supposed, there is no absolute number one. EPK and BPMN are still on the same level.

For sure that will relight discussions again.. 

 In what kind of environment are you working with process modeling?

How old are you?

How long are you working with process modeling?

 How complex were your models?

 What tool are you using for process modeling?

 Is the tool you are using easy to handle regarding the given points?

1 means very easy, 5 means very complicated

Can you handle all given problems with the tool you are using?

When will BPMN be the major tool for process modeling?

Hi pals!

In my understanding BPMN shows current (any) situation and ECP shows business rule. The bad news is that most non-proffesionals looking at BPMN and cant understand anything. But with EPC i could explain only 2 things about rule and activity.. thats it!

So i found out that EPC is better 4 understanding of simple rules and BPMN could draw a diffilult situation "as is" but prefer prepared listeners.

Best regards

  Hello!

That's a really useful comparison that you've done. I'm curious to know if it is possible to integrate elements from BPMN and EPC models in the same project using Aris Business Architect?

For instance, to link BPMN elements with the EPC product dimension?

Thank you in advance!

Best regards,

Kenia Sousa

The usage of object types in model types is defined in the ARIS method which is not freely configurable by a user/admin (you can only "hide" objects by using a filter, or add derived object types/symbols to a filter). The only way to do what you suggest is to assign a FAD (Function Allocation Diagram) to each BPMN task object that includes the wanted object types, as Sebastian suggested above.

In any case it might be a good idea to check the Method Manual (in the "Documentation" folder) and/or to visit the Method Help in the Help menu. There you can see what is allowed in the various model types.

Hi all,

Thanks for this valuable information. I am new to ARIS and I have to design a job description. What is the best model type for this purpose.

Thanks

Hello everyone,

Would you mind visiting the below link and answering my question which is related to this post.

http://www.ariscommunity.com/users/mohammad64/2011-01-08-which-bpm-language

Thanks in advance

Make sure to check the latest post by Ivo adding more details on BPMN vs. EPC!

Very good and useful article and good discussion! Too bad, I already spent some time to solve this out myself. I definitely need to start spending more time here. The picture (the ARIS house) is just super!

Best regards, Anna

Very Good discussion Sebastian Stein  but i'm not agreed on ur slide because we can use end events in BPMN diagrams instead of product/services in EPC modeling standards it italks about same objects. we can use stereotypes in BPMN modeler sof twars to classify output types

i have worked with 2 main softwrae ARIS7.1(EPC ) STANDARDS & Visual paradigm (BPMN2) STANDARDS

I think EPC standards has more complete objetcs but it's software is too much and many companies can't buy it .

Hello Sebastian,

It's quite interesting this confusion about trying to compare two models which are actually not possible to be compared;

it's like trying to compare a car and a truck! Which is better ? But for what ! And that is, in my opinion, the real question to answer; while EPC was meant to describe business, BPMN was meant to automate processes; you may need to describe a Business process without needing to automate it and  trying to do this with BPMN is quite impossible and difficult to be read by a Business User; actually, you can describe a Business process with 5 or 6 Graphic Symbols butt may need many more to describe its Automation, which does'nt mean that BPMN is more powerful!

So, we obviously need both , each one with its own purpose.

Hello,

Is there anyway i can use pools/lanes in flowcharts???