BP Variability Case Studies Development using different Modeling Approaches

1 Informe Técnico / Technical Report Ref. #: ProS- TR - 2011 - 03 Title: BP Variability Case Studies Development using different Modeling Approaches Author (s): Clara Ayora, Victoria Torres and Vicente Pelechano Corresponding author (s): cayora@pros.upv.es vtorres@pros.upv.es pele@pros.upv.es Document version number: 1. 2 Final version: No Pages: 30 Release date: February 2011 Ke y words : Business Process Modeling, Model Variability BP Variability Case Studies Development using different Mode ling Approach es Clara Ay ora, Victoria Torres, Vicen te Pelechano Universidad P olitécnica de Va lencia · Camino d e Vera s/n · Edific io 1F · 46022 Va lencia Spain · T . +34 96 387 70 07 Ext. 83530 · M . + 34 619 543 623 · F. +34 96 387 73 59 · info@p ros.upv.es · w ww.pros.upv.es 2 BP Variability Ca se Studies Development u sing different Modeling Appr oaches Clara Ayora, Victor ia Torres and Vice nte Pelechano Centro de Investigación en Métodos de Producción de Software Universidad P olitécnica de Valencia Camino de Vera s/n, 46022, Valencia, Spain {cayora, vtorres, pele }@pros.upv.es January, 20 11 3 Index 1. Introduction ………… ………………………………… ………………………….. 4 2. Approaches overv iew ………………………………………………… ………. 4 3. Case Studies Vehicle Repair Process ……………………………………………………………………….. 6 Approach: Configurable Event-driven Process Chains (C-EPC) .. 7 Approach: Rich BPMN (PESOA) …………………………………………….. 8 Approach: Process Variants by Options (PROVOP) ………………. .. 9 Approach: Worklets (Yawl+RDR) …………………………………………… 10 Healthcare process …………………………………………………………………………….. 12 Approach: Configurable Event-driven Process Chains (C-EPC) 13 Approach: Rich BPMN (PESOA) …………………………………………….. 14 Approach: Process Variants by Options (PROVOP) ………………… 15 Approach: Worklets (Yawl+RDR) …………………………………………… 16 e-business shop ………………………………………………………………………………….. 19 Approach: Configurable Event-driven Process Chains (C-EPC) 20 Approach: Rich BPMN (PESOA) …………………………………………….. 21 Approach: Process Variants by Options (PROVOP) ………………. .. 23 Approach: Worklets (Yawl+RDR) …………………………………………... 24 4. Conclusions …………… ……………………………………………………… . ……. 28 5. References …………… ………………………………………………………… ….. 31 4 1. Introductio n Variability in Business Process mo deling has already been faced by different authors from the literature . Depending on the context in which each author faces the modeling problem, we find different approaches (C -EPC [1], C-YAWL [2], FEATURE-EPC [3], PESOA [4], PROVOP [5], or WORKLETS [6]). In this report we present f our of the most representative approaches (C -EPC, PESOA, PROVOP and WORKLETS) which are presented by means of the different case studies found in the li terature . The remainder of this t echnical report is o rganized as follows. Next section presents each of the case studies used i n this report and, for each of them, applies the four different approaches selected from literature. 2. Approac hes Overvie w  Approach: Configurable Event-driven Process Chains (C-EPC) C-EPC is an ext ension of EPC (Eve nt -driven Process Cha in) that incl udes new constructs to represent variability in reference process models. The main idea of C -EPC is to represent differently commonalities from individualities in order to configure the process model according to its context. This differentiation is possible by comb ining the use of (1) configurable nodes (functions and connectors), which allow specifying different behavior depending on the context of use with (2) configuration requirements and guidelines , w hich state, by mea ns of logical p redicates, the valid configurations of the model. The complete description of the approach is presented i n [1].  Approach: Rich BPMN (PESOA) PESOA (www.pesoa.org) is a cooperative project carried out by a group of companies (DaimlerChrysler AG, Delta Software Technology GmbH, ehotel AG and Fraunhofer IESE) and academics from th e Hasso-Plattner-Institute and the University of Leipzig who se main objective was the investigation of a proposal for the development and cust omization of f amilies of process oriented software. As a result, focused at the design level and t aking in to account the relevance of the reusability aspect, a set of basic and compo site variability mechanisms was identified. The basic set includes (1) Encapsulation of Varying Subprocesses, (2) Addition, Replacement, Omission of E ncapsu lated Subprocesses, (3) Pa rameterization, a nd (4) Variability in Data Types, while the composite in clude (5) Inheritance, (6) Design Patterns, and (7) Extensions/Extension Points. This set was transferred to d ifferent languages such as UML Activity D iagrams, UM L State M achines, BPMN, and Matlab/Simulink. Sin ce we are focused on th e concepts and not in th e particularities of any language, we h ave taken the transference made to BPMN to evalua te the PESOA proposal. 5 The complete description of the approach can be found in [4].  Approach: Process Variants by Options (PR OVOP) Provop is an operational approach for managing large collections of process variants during the process life cycle. It has been motivated by the f act that a process variant can be created by adjusting (configuring) a given process model to a given context. These variant-specific adjustments are expres sed by means of a set of h igh -level change operations (INSERT, DELETE, MOVE and MODIFY). Furthermore, Provop allows more com plex process ad justments by grouping multiple chan ge operations into so called options . Thus, a particular process vari ant is specified (configured) by applying one or more o ptions to the respective base process. Th e options used for a process variants are selected when evalu ating the given cont ext. Provop provides a model for capturing this process context by means of context variables , which represent different domain dimensions of it. A complete description of the approach is presented in [5].  Approach: Worklets (YAWL+RDR) This proposal is an appro ach for dynamic flexibility, evolution and exception handling i n workflows through th e support o f flexible work practices. It was not conceived to be targeted to any notation (language independence) which means that it can be applied to any BPML. A worklet is def ined as a small, complete and re -usable workflow specification which handles o ne specific t ask in a composite parent process. In this parent pro cess, an extensible repertoire (or catalogue) of worklets is maintain ed for ea ch nominated t ask. Each time a worklet is need ed, an intelligently choice is made from this repertoire u sing a set of associated selection rules (Ripple Down Rules, RDR). These rules determine the most appropriate substitution. Then, the selected worklet is launched as a separate case and, when it h as completed, the control is returned to the original (parent ) process, which continues normally. T hus, d ynamic ad -hoc change and process evolu tion are provid ed without having to modifica te the original process specification and/or to resort to off-system intervention. The full description of the approach can be found in [6]. 6 3. Case Stu dies This secti on presents a complete description of each case study followed by the different models obtained after applying each one of the evalua ted approaches. Vehicle Repair Proc ess This case study is taken from t he work developed by Hallerbach et al. in [7] to present their a pp roach Provop. T his c ase study is developed in the context of the automobile industry. The process starts with t he reception of a vehicle. After a diagnosis is made, the vehicle is repaired if necessary. During its diagnosis and repair the vehicle is maintained ; e.g. oil and wiper fluid are checked an d refilled if necessary. The p rocess ends whe n handing over t he repaired and maintained vehicle t o the customer. Depending on t he process context, different variants o f this process are required, whereas the context is described by country-specific, garage-specific, and vehicle specific variables. Variant 1 : Assumes that the damaged vehicle requires a checklist of “Type2” to perform th e diagnosis. Therefore, activities diagnosis and repair ar e adapted by modifying their attribute ch ecklist to value “type2”. A dd itionally, the garage omits maintenance of the vehicle as this is con sidered as special service not offered conjointly with the repair process. Variant 2 : Due to country- specific legal regulations, a final security check is required before handing over the vehicle b ack to the customer. Regarding this variant, n ew activity final check has to be added when compared to the standard process. Va riant 3 : If a checklist of “type 2” is required for vehicle diagnosis and repair, the garage does not link maintenance to the repair process, and there are legal regulations requiring a final security check. 7  Approach: Configurable Event-driven Process Chains (C-EPC) 8  Approach : Rich BPMN (PESOA) 9  Approach: Process Variants by Options (PROVOP ) 10  Approach: Worklets (Yawl+RDR) Worklet-Enabled process: Selection Rule Tree for the “Repair process” task: Selection Rule Tree for the “Final check” task: 11 Worklet Repair1: Worklet Repair2: Worklet Maintenance: Worklet Non Final check: 12 Healthcare process This c ase study is take n from [8]. It has being developed w ith in the heal th care domain and shows a simplified version of a healthcare process representing a cruciate rupture treatment. It is modeled in BPMN. The process is started by the Ad mission of the patient . Then, the Anamnesis and Clinical Examination is performed to the patient. After this examination, different tests ( X-ray , MRT , an d Sonography ) can be processed in p arallel in any arb itrary order. On ly in the case t hat the patient is suffering from a cruciate rupture the activities Initial Treatment and Operation planning and Operative Treatment will be performed. Variant 1 : When patients with cardiac pacemaker skip MRT test. Variant 2 : When patients suffer from an effusion in a knee, a puncture has to be done. Variant 3 : Due to legal regulat ions, it would be n ecessary to inform the patients about the treatment. 13  Approach: Configurable Event-driven Process Chains (C-EPC) 14  Approach : Rich BPMN (PESOA) 15  Approach: Process Variants by Options (PROVOP) 16  Approach: Worklets (Yawl+RDR) Worklet-Enabled process: 17 Selection Rule Tree for the “MRT” task: Selection Rule Tree for the “Puncture” task: Selection Rule Tree for the “Inform patients” task: 18 Worklet Non MRT: Worklet Non Puncture: Worklet Non Inform patients: 19 e-business shop This case study is taken f rom the PESOA report [ 9] developed in the context of th e BMBF-Project. Th is case study is developed in the context of th e e -commerce domain (B2C market) using the BPMN notation. Two different roles interact: the customer and the shop. The process is started by the customer who explores the products that t he shop off ers to its clients. D uring this exploration, t he shop de livers to the customer the appro priate i nformation of the products. T hen , once t he customer has chosen some of the products, t he shop composes the customer shopping-cart. Du ring the next step, the customer decides to buy t he selected products. At t his point, the shop start s with the checkout process and then with t he delivery of the selected products. If t he customer takes a long time to perform the purchase, then the process is finished. If not, the products are received by the customer. Variant 1 : The information provided to the cust omer regarding th e product consist of a textual description and optionally pictures and reviews. Variant 2 : The shopping cart can be made persistent optionally. Variant 3 : The shop could support personaliz ed shopping carts. This means that the 10% of the p urch ase is given to the customer to buy what she/he wants. Another option is t o have an anonymous shopping cart. In this case, the identity of the customer is hidden. The shop should only support one of these two types of shopping carts. Variant 4 : The checkout task shou ld support the payment by credit card . However, if a personalized shopping cart is selected, then the invoice paymen t is offered optionally. Variant 5 : If the customer takes a lon g t ime to perform the p urchase, th en the process is finished. 20  Approach: Configurable Event-driven Process Chains (C-EPC) 21  Approach : Rich BPMN (PESOA) 22 23  Approach: Process Variants by Options (PROVOP ) 24  Approach: Worklets (Yawl+RDR) Worklet-Enabled process: Selection Rule Tree for the “Deliver information” task: Selection Rule Tree for the “Compose shopping cart” task: 25 Selection Rule Tree for the “Save shopping cart” task: Selection Rule Tree for the “Checkout” task: Selection Rule Tree for the “Deliver” task: 26 Show TextInf: Show PicsReview: Compose AnonymousShoppingCart: ComposePersonalizedShoppingCart: Give10%Purchase: Non SavedShoppingCart: 27 CheckoutByCreditCard: Show CheckoutInvoice: Non Deliver: 28 4. Conclusions Conducting these three case studies has helped us, not only t o im prove our knowledge and expertise of the selected approaches, but also t o identify t heir strengths and weaknesses. Th e conclusions of our modelin g exp erience is presented below. 4.1 Configurable Event-driven Process Chains (C-EPC) C-EPC is an approach to exp licitly capture variability in pro cess models by extended the EPC notation. An easy task when modeling using C -EPC is t he ident ification of those places within the model that may vary, t he variation points. However, in C-EPC the resolution of these variation points is attached to other previous decisions of t he same mo del, instead of being based on context asp ects . But process mo deling is concerned with context relat ed aspects as well , a good example of it can be found i n the Variant 5 of the e- business shop example: “if the customer t akes a long time to perform the purchase, t hen the process is f inished”. C-EPC does n ot provide any technique to solve this situation , lacking on support context modeling concepts . Thu s, it is no t possible to model/consider in C-EPC variability that occur on ly during t he execution of the process (run time). Since C-EPC models are integrated representations, all the process variants are defined t ogether within the model . In ord er to being able to combine all of them in one un ique model, modelers need to h ave a cl ear prior idea of the entire process , which is very d ifficult in models that include a high level of variability , e.g. e-business shop . As a consequence, the mo dels tend to get b ig and co mplex ve ry fas t which implies that its simplicity disappears. To restrict variant combination is n ecessary to define configuration requirements and guidelines u sing logical predicates, but C-EPC does n ot provide any technique to check t he consistency of th e m. Thu s, it needs to be done manually, which is very time - consuming. 4.2 Rich BPMN (PESOA) The aim of the approach is to improve re -usability and customization of those systems that are developed from the specification of process models. For such purpose and taking into accou nt variability aspects, different stereotypes h ave been defined in order to cover different variant behaviors. D espite this, t he main p roblem of the PESOA approach is that it is n ot specified how variation points should be solved. When 29 modelers detect the places that the pro cess may vary, h ow they should tra nsf orm them into to those stereotypes th at suit better f or them is not clear ly defined. This leaves the variant def inition for modeler interpretation which m ay lead to correctness issues in t he resulting models , requiring to the modelers a high level of expertise to use this approach. On t he contrary, a positive aspect of the PESOA approach is the comb inat ion of feature diagrams and BPMN. This combination facilitates the configuration of process models by capturing context aspects. 4.3 Process Variant by Options (PROVOP) Provop provides an operational approach for managing process variants. In particular, the process variants can be configured by applying a set of h igh -level change o perations to a common base process. T he i ntroduction of these change operations allows modelers to d efine the common process separately from individualities as well as to distinguish them, making the m odel s simpl er and more intuitive. A goo d example of that is the resulting model of the case study e -business shop where, d espite having a high level of variability, the model is only one workflow in which tasks may be added, delete or modify i f necess ary. Provop also provides support for context-aware p rocess configuration by means of the context rules. Nevertheless this rules do not exp licitly specify t he time in which variation points are solved (design time or run time), which is important in order to configure the process before the deployment. 4.4 Worklets (Yawl+RDR) The Worklet approach en ables late b inding o f process fragments to process activities at run time. Thu s, at design time, the activity is merely mo deled as placeholder and, at run time, the appropriate process fragment is selected to bound to the process activities. This late binding allows a better activity re -using as well as model understanding. The problem ap pears in h ow activities that are going to be bounded (variation points) are identified within the model. There is not provid ed any special mark to distinguish commonalities from individualities. Another drawback id entified is h ow to model these activities that may be skipped during the execution . For instance, the MRT activity of the health care process case study should b e skipped if the patient has a pacemaker in his/her h ear t. How to model this pacemaker cond ition is not clearly sp ecified in the related literature. M oreover, how the activity should be replaced in order to be skipped (i.e. should it be replaced by 30 an empty activity?) or how the RDR should define this replacem ent are neither clarify in the documentation. 31 5. Referen ces 1. Rosemann, M., van der Aalst, W.M.P.: A con fi gurable reference modelling lan - guage. Inf. Syst. 32(1) (2007) 1- 23 2. F. Gottschalk and M. La Rosa. Process Configuration in YAWL. QUT ePrints 15718, Queensland University of Technology, Bris ban e, Australia. 3. Vervu urt, M .:Modeling business p rocess variability: a search for innovative solutions to business process variability modeling problems (October 2007). 4. Pu hlman, F., Schjeders, A., WeilandJ., Weske, M.: Variability mechanisms f or process models, Technical report, BMBF-Project (2006). 5. Hallerb ach, A.; Bauer, T. & Reichert, M. Captu ring Variability in Business Process Models: The Pr ovop Approach. In: Software Process: Improvement and Practice, Wiley InterScience, 2010 6. Ad ams, M., t er Hofstede, A.H.M., Edmon, D ., van d er Aalst, W.M.P.: Facilitating fl exibility and dynamic excep tion handling in workflows thorugh worklets. In: CAiSE Short Paper Proceedings. (2005) 7. Hallerb ach, A.; Bauer, T., M.R.:Int ernational Handbook on Business Process Management. In: Configuration and Management of Process Variants. Springer (2010). 8. Web er, Barbara and Sadiq, Shazia and Reichert, Manfred. Beyond rigid ity – dynamic process lifecycle support. Springer-Verlag 2009. 9. F. Puhlmann. Modeling Workflows in the E -Business Domain. PESOA Report No. 08/2004, Hasso-Plattner-Institute, 2004.