Scenario Development Approach to Management Simulation Games

Transcription

1 do: /tms / 17 Scenaro Development Approach to Management Smulaton Games Jana Bkovska, Rga Techncal Unversty Abstract The paper ntroduces a scenaro development approach to management smulaton games. Nowadays the hgh effcency of smulaton games as a tranng method s obvous due to ther ablty to help students understand the complexty of decson makng wthn a dynamc envronment. Modern computer technologes allow creatng complex games wth an adaptable envronment that requres development of approprate game scenaros. It s not a trval problem that requres elaboratng of a formal approach, whch wll be descrbed n the paper. As a result a general scheme for scenaro development and modellng for a smulaton game s proposed n the paper. Some sample scenaros for logstcs and supply chan games are provded n the paper as well. Keywords Game scenaros, logstcs and supply chan game, management smulaton game, scenaro management. I. INTRODUCTION Smulaton games n educaton are wdely used as they provde realstc, experental learnng envronments, thus, provde a more comprehensve way to teach concepts and theores than tradtonal lecturng [1]. Ths paper focuses on management smulaton games, a subset of smulaton games that provde a dynamc, uncertan and compettve busness envronment to tran learners for management decson makng. The players of such games make management-type decsons n an offered stuaton, and ther decson choces generally affect the envronmental condtons under whch the subsequent decsons must be made. In a very bref tme, ths gves the game partcpants a chance to study both long and short-term results of ther decsons. Ths provdes a possblty to react to these effects and make decsons n the lght of altered crcumstances as n the real busness. As a result, the game partcpants are able to gan knowledge and practce sklls n the game by repeatng experments wth the smulated realty through the smulaton game. In addton, such games help developng such sklls as teambuldng, socal networkng, collaboratng, etc. Recent advances n technology allow dentfyng several research and development drectons n management smulaton gamng. As ndcated n [2], n the past two decades the number of Internet-based management smulaton games has grown as t allows runnng them through a central server wth admnstrator-selected parameters and enterng partcpants decsons drectly to the server. Thus, management smulaton games use computers-servers to run the models and to process partcpants decsons. Ths allows creatng more realstc and complex game scenaros wth respect to the number of decson varables ncluded, the number of products and markets defned, and the amount of feedback avalable to the partcpants. Software agents emboded n management smulaton games facltate modellng and explorng possble scenaros n the game whle controllng the game process, thus, affectng the envronment and drecton of the smulaton [3]. Modern management smulaton games should possess the followng features: realsm, accessblty, compatblty, flexblty and scale, smplcty of usage, decson support systems, and communcaton [2]. Ths paper focuses on enhancng realsm, flexblty and scale of management smulaton game by proposng a general scheme for the game scenaro development. The objectves of ths paper are: (1) to ntroduce a scenaro concept for management smulaton games; (2) to propose a general procedure for management smulaton game scenaro development; (3) to provde examples of scenaro development for ILMG and ECLIPS management smulaton games. II. GAME SCENARIO CONCEPT Scenaro approach s wdely used n dfferent felds, for nstance, forecastng, optmsaton etc. There are dfferent vews on the defnton of scenaro, dependng on the context n whch the term s used. Ths term appears n a wde varety of felds, rangng from state admnstraton, nformaton systems, requrement engneerng to human computer nteracton. Interpretaton of scenaros seems to depend on ther usage and how they are generated. In general, a scenaro can be defned as a set of possble sequences of future events [4], [5]. There are dfferent approaches to scenaro development: formal, semformal, and nformal, based on experts opnon [6]. The paper concentrates on a semformal approach as t ncludes automatc procedure of scenaro generaton adjusted by an expert. The game smulaton could be performed accordng to a predefned scenaro chosen (or specally elaborated) by the game manager (nstructor). Scenaro plays a very mportant role wthn the game structure and t has to be formalzed for further mplementaton. Wth a respect to management smulaton games, a scenaro can be conceved as a means of transformng the game ntal state to a fnal state, followng man development trends, whch are nfluenced by both nternal events and external actvtes. It could be presented by the 5-tuple [2] as t s shown n (1): <SI, T, E, A, SF > (1) where: SI the ntal state of the game; 144

2 2014 / 17 T predefned game busness envronment development trends; E predefned nternal events; A performed external actvtes; SF game fnal state once the scenaro has termnated. The generc scenaro structure s shown below (see Fg. 1). III. SCENARIO DEVELOPMENT APPROACH The scenaro development for management smulaton games becomes a complex process as the scenaros have to reflect real lfe stuatons [2], n other words, the number of a scenaro parameters that defne game ntal state, game busness envronment development trends, nternal and external events can be very large. In ths stuaton, manual scenaro development would be not so effcent. Ths makes t necessary to propose an advanced scenaro development procedure (see Fg. 2). Fg.1. Generc scenaro structure of management smulaton game. Trends T ntroduced to a game scenaro dynamcs over tme t are defned by dfferent functons, e.g. (2): T f ( t), T T (2) Patterns of these trends depend on the type of functon and ts parameters whch can be changed by the game manager. All nternal and external nfluences depend on the current game state and they are amed at achevng a partcular result of the game functonng or predefned learnng objectves. Any nternal and external nfluences lead to short-term or long-term changes n the pattern of trends,.e., (3). E A f : E T and f : A T, (3) where: E { e1, e2,..., en} s a set of nternal events; A a, a,..., a } s a set of external actvtes. { 1 2 n Game ntal state s defned by the proposed busness envronment characterstcs and structure as well as by value of the trends functons (4) at the moment when the game sesson starts: SI T ( t 0 ), (4) where t 0 defnes the ntal tme nstant. Thus, a scenaro provdes a flexblty of management smulaton games as allows a game manager to defne the ntal state of game busness envronment, to determne the development of the game stuaton through tme accordng to predefned trends, as well as preset the lst of mportant events takng place durng smulaton. Events can affect trends. Partcpants make ther decsons (or n other words, take actons) that can affect both trends and events. Trends may be confgured accordng to the learnng objectves by changng approprate parameters avalable n the game. Once all of the game scenaro elements are defned, the scenaro development procedure should be appled. Fg. 2. Scenaro development procedure for management smulaton games. Here, a scenaro base represents a set of prevously developed scenaros that can be used as a base for further scenaro development. Game manager chooses a default scenaro and sets basc scenaro characterstcs, thus defnng ntal state of the game. Then specally elaborated tools are appled to generate new values of scenaro parameters that defne game busness envronment development trends, nternal and external events. After the scenaro s generated, t s necessary to smulate t wth the game. It can be done manually or automatcally. As the manual smulaton supposes real players to test the game scenaro, t would take more tme and efforts. Automatc scenaro smulaton has several advantages: (1) the scenaro can be smulated n a very short tme perod; (2) several replcatons can be made for more effcent scenaro evaluaton. To smulate the developed game scenaro, specal software agents are proposed [7], [8]. They act as the game partcpants by performng actvtes related to the company development under the proposed scenaro condtons [4]. 145

3 2014 / 17 Dependng on the type of the game, several agents can be runnng smultaneously. When the game results are obtaned, a game manager has to evaluate a new scenaro. If t matches learnng objectves, and t s proved that certan game results can be acheved, then the scenaro could be appled to the regular game sesson and added to a scenaro base. Scenaro from the scenaro base can also be appled to the game wthout any changes. The proposed procedure can be appled to scenaro generaton of dfferent management smulaton game types. IV. ILLUSTRATIVE EXAMPLES To provde the examples of scenaro development, two management smulaton games were selected, namely, Internatonal Logstcs Management Game and ECLIPS game. A. Internatonal Logstcs Management Game As an example of a management smulaton game for ntroducng scenaro development and modellng approach, the Internatonal Logstcs Management Game (ILMG) was selected [9], [10]. The game s amed at provdng a vrtual busness envronment for tranng decson-makng sklls n the area of nternatonal logstcs, producton management, fnances etc. The game can be classfed as a total enterprse (or general management), nteractng, computer-aded and nternet-based, real-tme processed busness smulaton. ILMG software functonng accordng to the structure presented n Fg. 3. followng goals: to earn proft, to capture substantal market share and to acheve hgh customer satsfacton. The ILMG scenaro s generated by the game manager accordng to specfc learnng objectves. Intal condtons of the game and corporatons state are set up by changng dfferent parameters that can be dvded nto several groups (see Fg. 4): regonal parameters (e. g., unt projectng cost, basc wage level), regonal-related producton parameters (e. g., productvty parameters, overhead parameters), techncal coeffcents of products (e. g., materal per unt, setup cost), market related parameters (e. g., prce effect parameters), fnancal parameters (e. g., prme rate, tax rate) etc. Fg. 4. ILMG scenaro development screen. Most of the parameters have to be defned before the game starts and they cannot be changed afterwards. However, some parameters (. e., prme rate, busness cycle ndex of the regon, regonal productvty ndex, and regonal wage level) can be easly edted durng the game that allows changng a busness envronment to smulate dfferent stuatons. In ILMG, the game manager has to set over than a hundred dfferent parameters for the smplest scenaro (one regon, one product, two bank accounts). To support ths task, specal stand-alone applcaton GameEdtorPro was developed (see Fg. 5). It generates a full set of the game scenaro parameters accordng to gudelnes provded by the game manager. Fg. 3. ILMG software structure. Dfferent scenaros n the game are ntroduced to smulate real world stuatons and force tranees to acqure sklls and experence n managng dfferent functons of the company n varous stuatons. Durng the game, partcpants operate as dfferent corporatons wthn an nternatonal market, producng goods or servces (. e., makng strategc and operatonal decsons) and competng wth each other n order to mprove ther corporaton performance and to acheve the Fg. 5. GameEdtorPro for ILMG nterface example. 146

4 2014 / 17 The game manager decdes only on a number of regons (R) and products (P) and ther characterstcs; then the software automatcally generates all other scenaro parameters. Regons are defned wth qualtatve characterstcs as regons wth advanced, less advanced or developng economy. Products are classfed as h-tech, medum-tech and lowtech products. For example, the game manager sets up a new scenaro wth three regons R1= advanced, R2= developng and R3= advanced, and three dfferent products such as P1= low-tech, P2= h-tech and P3= h-tech. Then an automatc generaton of scenaro parameters s performed by the followng algorthm: the total number of parameters s calculated; the full range of parameters values s defned; each defned range s dvded nto three equal subsets; each subset of parameters characterzes ether one regon type or one product type; wthn each subset, values of parameters are generated randomly to exclude the possblty obtanng dentcal regons or products. The parameters of the scenaro receved from the GameEdtPro module can be exported to the ILMG software. It s used to smulate and test a specfc scenaro. In ths case, a player agent s nvolved and t makes strategc, marketng, producton, purchasng, transport, nvestment and fnancal decsons. Fnally, the game manager manually evaluates the key ndcators receved by the player agent and decdes f they meet learnng objectves such as a postve Net proft, a customer s fll rate s more than 50 %, and a lqudty rato s greater than 1. If the evaluaton s negatve, the procedure of scenaro generaton s repeated and new values of game parameters are generated. Otherwse, the generated scenaro s approved and ntroduced n the game regular sesson. For the testng purpose, the game wth 3 player agents was run for 8 perods (or 2 years). These agents substtuted real players of the game. The key ndcators receved by the agents players are gven n Table I and presented n Fg. 6. As far as the obtaned results meet the above-mentoned requrements, the tested scenaro can be ntroduced n the game regular sesson. TABLE I KEY INDICATORS OBTAINED FROM SCENARIO MODELLING Agent 1 Agent 2 Agent 3 Net Proft (EUR) 80,749,580 81,865,660 90,920,780 Servce Level 64% 60% 63% Then the scenaro s mplemented and the game manager contnuously revews the key performance ndcators of corporatons represented by the human players. If they are out of range as defned by the teachng objectves, the game smulaton tme s fxed and a new scenaro model s generated, e. g., such parameters as prme rate, busness cycle ndex of the regon, regonal productvty ndex, or regonal wage level could be adjusted. A two-year perod s smulated durng the regular sesson of a game. B. ECLIPS Game ECLIPS game s a board game that consders an nventory management problem n a mult-echelon supply chan, n partcular, detaled workngs of dfferent nventory replenshment polces. It provdes an nsght nto organzaton and functonng of mult-echelon supply chan based on contnuous or perodc nventory revew plannng as well as demonstrates benefts and shortcomngs of ther mplementaton under dfferent condtons [11], [12]. ECLIPS game provdes flexble supply chan network modellng capablty. To support actvtes of the game manager and players, a specal computer-aded tool was developed. It s amed at provdng envronment that allows smulatng dfferent scenaros of the ECLIPS game [12]. The example of the smulated supply chan s represented n Fg. 7. Fg. 7. Three echelon supply chan of the ECLIPS game. The current verson of the software allows testng scenaros by changng dfferent parameters of the supply chan (see Fg. 8), e. g., end customer demand, nventory holdng cost, order cost, new producton cost, lead tmes between stock ponts and servce level coeffcent. Here, the number of stock ponts or supply chan echelons s lmted by three as well as the end customer demand can be modelled ether by one or two dce wth sdes defned by the game manager. Lqudty Rato Fg. 6. Lqudty rato of tested scenaro. Agent 1 Agent 2 Agent 3 Perods Fg. 8. Scenaro parameters of the ECLIPS game. 147

5 2014 / 17 After settng all parameters, the scenaro s smulated to obtan game results. To evaluate reproducblty of the acheved results, the confdence nterval method s proposed. Scenaro s evaluated by three key ndcators: servce level (cumulatve), average cost and average nventory. The smulaton results from 10 replcatons of 52 game perods for average nventory under a contnuous replenshment polcy are aggregated and represented n Table II. The cumulatve mean and 95 % confdence ntervals of the smulaton results are gven n Table II, as well. Smlar calculatons were performed for servce level and average cost. Replcaton TABLE II CONFIDENCE INTERVAL CALCULATION FOR AVERAGE INVENTORY Average Cumulatve nventory mean Standard devaton Confdence Level Lower nterval Upper nterval % devaton n/a n/a n/a n/a n/a % % % % % % % % % The cumulatve mean and confdence ntervals for average cost, average nventory and servce level are graphcally represented n Fg. 9 to Fg. 11. Here, the cumulatve mean n all the graphs s reasonably flat and the confdence ntervals narrow farly rapdly. Average Cost (EUR) 190,00 180,00 170,00 160,00 150,00 140,00 130,00 120,00 110,00 100,00 90, Number of replcatons Cumulatve mean Lower nterval Upper nterval Fg. 9. Cumulatve mean and 95 % confdence nterval for average cost. Average Inventory (psc.) 110,00 105,00 100,00 95,00 90, Number of replcatons Cummulatve mean Lower ntrval Upper nterval Fg. 10. Cumulatve mean and 95 % confdence nterval for average nventory. Servce Level (%) 110, , ,000 95,000 90, Number of replcatons Cummulatve mean Lower ntrval Upper nterval Fg. 11. Cumulatve mean and 95 % confdence nterval for servce level. The calculatons and ther graphcal nterpretaton allow concludng that for average nventory and servce level the devaton s less than 5 % at three replcatons. The nterval narrows a bt slower for the average cost and the devaton s less than 5 % after the 5 th replcaton. Thus, t s possble to confrm the convergence of the scenaro modellng results. Smlar calculatons are performed for a perodc replenshment polcy. Therefore, the scenaro can be proposed for the mplementaton of the regular play of the ECLIPS game. V. CONCLUSION Recent advances n nformaton technology have had a great mpact on the process of the development and applcaton of management smulaton games. The research results provded n ths paper show that the complexty of such games can be overcome by ntroducng a formal game scenaro concept. The proposed scenaro development approach allows generatng and smulatng scenaros of games n order to evaluate ther relevance to learnng objectves defned by the game manager. Despte the fact that some of the descrbed procedure steps are not yet automated, the provded examples of scenaro development for ILMG and ECLIPS games show the procedure advantages for game managers. Currently, t s not possble to completely avod the nterference of an expert (game manager or nstructor) n the process of scenaro development; however, further research drecton may be related to the elaboraton of method for generatng values of varous scenaro parameters, as well as to the use of more sophstcated agents for scenaro smulaton. REFERENCES [1] G. Merkuryeva, Y. Merkuryev, J. Bkovska, J. Pečerska, J. Petuhova, Actve Learnng Logstcs Management Through Busness Gamng, n Proceedngs of 4th Internatonal Conference on Interdscplnary n Educaton, pp , [2] A. J. Fara, D. Hutchnson, W. J. Wellngton, Developments n Busness Gamng: a Revew of the Past 40 Years, Smulaton and Gamng, vol. 40, no. 4, pp , [3] G. Merkuryeva, J. Bkovska, T. Ören, An Agent-drected Multsmulaton Framework for Smulaton Games Management, Internatonal Journal of Smulaton and Process Modelng, vol. 7, no. 3, pp , [4] M. Jarke, X. T. Bu, J. M. Caroll, Scenaro Management: An Interdscplnary Approach, Requrements Engneerng, vol. 3, ssue 3 4, pp , [5] D. A. Kononov, V. V. Kulba, S. S. Kovalevsky, S. A. Kosjachenko, Development of scenaro spaces and the analyss of dynamcs of behavour of socal and economc system. Preprnt, Moscow, (In Russan). 148

6 2014 / 17 [6] V. L. Shulc, V. V. Kulba, et. al., Models and methods of analyss and synthess of scenaros of soco-economc systems. Book 1, Nauka, Moscow, (In Russan). [7] L. Ylmaz, T. Ören, Intellgent agents, smulaton, and gamng, Smulaton & Gamng, vol. 37, no. 3, pp , [8] M. Remondno, A Web Based Busness Game Bult on System Dynamcs Usng Cogntve Agents as Vrtual Tutors, n Proceedngs of the Tenth Internatonal Conference on Computer Modelng and Smulaton, pp , [9] R. W. Grubbström, G. Merkuryeva, J. Bkovska, J. Weber. ILMG: Learnng Arrangements and Smulaton Scenaros, n Proceedngs of 19th European Conference on Modelng and Smulaton, pp , [10] J. Bkovska, G. Merkuryeva, The Internatonal Logstcs Management Game: An Innovatve Busness Envronment for Tranng, Producton- Economc Research n Lnköpng, pp l, [11] Y. Merkuryev, G. Merkuryeva, J. Bkovska, J. Hatem, B. Desmet, Busness smulaton game for teachng mult-echelon supply chan management, Internatonal Journal of Smulaton and Process Modelng, vol. 5, no 4, pp , [12] Y. Merkuryev, J. Bkovska, Busness Smulaton Game Development for Educaton and Tranng n Supply Chan Management, n Proceedngs of Asa Modelng Symposum (AMS2012), the Sxth Asa Internatonal Conference on Mathematcal Modelng and Computer Smulaton, pp , Jana Bkovska s a Researcher and Lecturer wth the Department of Modelng and Smulaton, Rga Techncal Unversty, Latva. Her professonal nterests are n the feld of supply chan management, management smulaton games, scenaro approach n complex systems wth applcaton to smulaton games. Her teachng experences nclude the followng management games: ILMG, New ORSIAM Int., ROFA, ECLIPS Game, and Beer Game. E-mal: Jana.Bkovska@rtu.lv Jana Bkovska. Vadību mtējošo spēļu scenārju vedošanas peeja Rakstā r apskatīta vadību mtējošo spēļu vedošanas peeja. Vadīšanas mtējošās spēles gūst azven lelāku populartāt, tā r kā alternatīva peeja mācībām, kas nodrošna vrtuālo bznesa vd vadības lēmumu peņemšanas emaņu trenēšana. Mūsdenās, patecotes nformācjas un komunkācjas tehnoloģju attīstība, paplašnās arī mtējošo spēļu funkconālās espējas, kas paaugstna to efektvtāt un maksmāl petuvna reālās dzīves stuācjām. Līdz ar to rodas nepecešamība zstrādāt tādas metodes, peejas un rīkus, kas nodrošna spēles kompleksās bznesa vdes pārvaldību. Šm nolūkam tek evesta vadību mtējošo spēļu scenārju koncepcja, kura kalpo par pamatu rakstā pedāvātaja spēļu scenārju vedošanas procedūra. Šī procedūra paredz varāku soļu zpld, la zvedotu tādu vadību mtējošas spēles scenārju, kas atblst apmācības mērķem. Dažus no šīs procedūras soļem zplda spēles vadītājs (nstruktors), bet daž r automatzēt. Pemēram, scenārju parametru vērtības tek ģenerētas automātsk, kā arī zvedoto scenārju modelēšana notek automātsk, pemēram, peletojot aģentus, kas azstāj reālus spēles dalībnekus. ILMG un ECLIPS spēles tka zvēlētas, la parādītu, kā zstrādātā peeja tek praktsk realzēta. ILMG r unversāla, nternetā bāzēta, tešsastes vadību mtējoša datorspēle, kas nodrošna vrtuālo bznesa vd lēmumu peņemšanas emaņu plnvedošana tādās sfērās kā starptautskā loģstka, ražošanas vadība, trgznība un fnanses. ILMG spēles scenārju parametru skats r dezgan lels, un scenārja manuāla vedošana un modelēšana r pārāk laketlpīga. Līdz ar to, peletojot zstrādāto scenārju vedošanas procedūru, zdevās kretn samaznāt darba apjomu un patērēto laku. Savukārt ECLIPS r funkconāla galda spēle, kuras mērķs r snegt prekšstatu par pegādes ķēdes darbības pamatprncpem un dažādu krājumu vadības stratēģju evešanas prekšrocībām un trūkumem. ECLIPS spēle tka zstrādāts elektronsks analogs, kas ļauj ģenerēt, modelēt un novērtēt dažādus spēles scenārjus, kas arī būtsk atveglo spēles vadītāja (nstruktora) darbu. Яна Биковска. Подход для разработки сценариев в деловых имитационых играх В статье рассмотрен подход для разработки сценариев в деловых имитационных играх. Использование деловых имитационных игр становится все более популярным, так как обеспечивает альтернативный подход к обучению, предоставляя виртуальную бизнес-среду для освоения навыков принятия управленческих решений. Благодаря современному уровню развития информационных и коммуникационных технологий, расширяются и функциональные возможности деловых имитационных игр, что значительно повышает их эффективность и максимально приближает к ситуациям реальной жизни. В этой связи появляется необходимость в разработке таких методов, подходов и средств, которые обеспечивают управление комплексной бизнес-средой этих игр. Для этого вводится концепция сценария имитационной игры, которая служит основой для предлагаемой в статье процедуры разработки сценариев деловых имитационных игр. Эта процедура предполагает выполнение нескольких этапов с целью разработки такого сценария деловой имитационной игры, который соответствует конкретной цели обучения. Некоторые этапы этой процедуры исполняет инструктор, другие - автоматизированы. Например, значения параметров сценария генерируются автоматически, а также моделирование сценария происходит автоматически, например, с использованием агентов, которые замещают реальных участников игры. Для демонстрации практической реализации разработанного подхода используются две игры: ILMG и ECLIPS. ILMG представляет собой универсальную онлайн деловую имитационную игру, которая обеспечивает виртуальную бизнес-среду для освоения навыков принятия решений в таких сферах, как международная логистика, управление производством, маркетинг и финансы. Количество различных параметров сценария в ILMG игре достаточно большое, следовательно, разработка и моделирование сценария в ручном режиме занимает много времени. Применив разработанную процедуру разработки сценария, удалось значительно сократить объем работы и потраченное время. В свою очередь, игра ECLIPS представляет собой функциональную настольную деловую мимтационную игру, цель кторой дать представление об основных принципах работы цепи поставок, а также об основных преимуществах и недостатках различных стратегий управления запасами. Для ECLIPS игры бал разработан электронный аналог, который позволяет генерировать, моделировать и оценивать различные сценарии игры, что значительно облегчает работу инструктора. 149