An Experiential Learning System for Production Planning and Control C. Basnet J.L. Scott Department of Management Systems University of Waikato Private Bag 3105, Hamilton 2020 New Zealand c11uda(a!wa i k ato.ac.nz i 1s (a!wa i k ato,ac.n z

Abstract We present a spreadsheet-based simulation game for teaching / learning production management concepts of forecasting, material requirements planning, order review and release. In this game the student plays the role of a production planner managing two products, for which customer orders are placed in variable quantities throughout the week. The student builds forecasting and material requirement planning systems to help them in the tasks of production and vendor order release. In parallel with this, we have run a small, learning awareness programme, to test and stimulate the skill of reflection. Initial student responses to the game have been favourable, but the proportion of time spent on reflection is low. Contemplated refinements to the game are presented.

1 Introduction Simulation games are activities designed to mimic the reality of the external world, within the classroom, with the goal of instruction. The learning is intended to be experiential - the student experiences the studied phenomenon and learning proceeds inductively. Besides simulation games, there are other means of providing the experience of reality to students - case study, role-playing, in-basket method, and incident process. The main advantage o f simulation games over these alternates is the dynamic nature of the games - the incorporation of the time element, imitating the passage of time. Students have to live with the results o f their past decisions - the effects of these decisions persist into the future in the game. Another advantage is the verisimilitude offered - some games are able to provide a high level of make-believe and fantasizing. The strong interest that is aroused in the subject matter is itself of pedagogical value. A simulation game can be restarted with a new strategy for playing the game, but a case study can only be used once (Gilgeous and D'Cruz, 1996). The model of experiential learning provides the theoretical underpinning of simulation games as a learning/teaching tool. Kolb's (1984) experiential learning model is shown in Figure 1. According to this model, concrete experience of a phenomenon in

233

the real w orld triggers the learning cycle. This event is observed/experienced, and causes/encourages reflection in the student. The student form s/uses abstract concepts and m odels/hypotheses to m ake sense o f reality. This leads to experim entation and hypothesis testing that provides concrete experience, which starts the cycle again. Sim ulation gam es provide the concrete experience needed in Kolb's m odel, and are a good platform for stim ulating learning aw areness in students, encouraging them to better understand their own learning processes (Scott, 2002).

Figure 1. Experiential learning (EL) cycle model There are m any production m anagem ent gam es available for educational purposes, but the num ber has not grow n in keeping with the grow th in num bers for top m anagem ent gam es and for m arketing gam es. The available games, such as Joblot (C hurchill, 1970), PRO SIM (M ize et a l , 1971), DECID E-P/O M (Biggs, 1987) provide an understanding o f production system s m ore at the strategic level than at the tactical level, as discrete-event dynam ic systems. There is a dearth o f gam es designed to teach specific technical skills in production m anagem ent. Lane (1995) in a pedagogical review o f sim ulation gam es, has supported such a sim ple game serving a specific learning objective against a com plex game satisfying a num ber o f objectives. There is also a lack o f gam es that enhance detailed m odelling and decision m aking capability. The prim ary goal o f the research presented in this paper is to develop a sim ulation gam e that m eets these voids, and to explore the learning im plications o f the game. Our paper presents a gam e w ith the specific objective o f learning about order release in production m anagem ent. A feature o f this gam e is that students build their own decision support system (D SS) to play the game. Building a DSS provides the gam e players w ith a detailed m odelling and decision m aking capability (Y eo and N ah, 1992). This DSS is based on a specific m odel - m aterial requirem ents planning (M RP); teaching M RP is a goal o f the game. In this paper we present a spreadsheet-based production-planning sim ulator called M R P-SIM designed to m eet the above objectives. The next section presents a description o f the game. This is follow ed by a discussion o f the learning aw areness program m e we ran in parallel w ith the game. Then we present student evaluations o f the game. Finally, concluding rem arks are presented.

234

2 2.1

The Game Objectives

Beginning students o f production planning and control (PPC) often struggle with the technical concepts in PPC such as bill o f m aterials, order review and release, and action buckets, to nam e ju st a few. The students need to see how forecasting leads to m aster production schedule, and to m aterial requirem ents planning, and finally to order release. An im portant objective for the students is to appreciate the variability and dynam ics o f the production environm ent, w here for exam ple even the forecast is not a static input to production planning. The gam e is designed to enhance the understanding o f PPC concepts such as bill o f m aterials, routing, order review and release, priority setting, queuing, forecasting, m aster production scheduling (M PS), m aterial requirem ents planning (M RP), and capacity requirem ents planning (CRP). These concepts are usually treated in isolation as discrete concepts. The sim ulation brings out their interactions in a simple, yet realistic setting.

2.2

The Scenario

In this game the students play the role o f the production planner o f a m anufacturing company. They m anage two products for w hich custom er orders are placed on the com pany in variable quantities throughout the week. These products are m ade up o f com ponents, some o f w hich are produced w ithin the com pany, and others are sourced from vendors. In carrying out the production, the parts are routed through processing machines w ithin the com pany w here processing tim e is spent, and queues are built up. The production planning is done on a w eekly basis. A t the beginning o f the week, inventories are checked and orders are released both w ithin the com pany and to the vendors. Through the week the processing takes place. Custom er orders arrive based on a stochastic process that sim ulates seasonality, trend, and random ness. The param eters o f this process are o f course unknow n to the students. As custom er orders arrive, the orders are filled from inventory on-hand. Custom er orders may be filled partially if there is insufficient on-hand inventory for the w hole order. Unfilled custom er orders are placed on file and filled w hen the product is available. The w eekly cycle o f activities is: ■ release production and vendor orders, ■ decide on overtim e for the processes, ■ sim ulate for a week (during this tim e these events occur: production in the facility, order fulfilm ent from the vendor, order arrival from the custom er, and order fulfilm ent to the custom er), and ■ m onitor the situation. The planner participates in this process beginning from week 20 and the gam e ends after w eek 32. A sim ulator w ritten in V ISU A L B A SIC ™ and incorporated in an EXCEL ™ spreadsheet sim ulates this scenario. The students interact w ith the sim ulator in the spreadsheet envirom nent. Profits are accum ulated for every item in the custom er order that is filled. For every item in the custom er order that is late (filled after the day the order arrives), a penalty is charged per day. There are also costs associated w ith holding inventory (both finished and work in process) and with overtim e work.

235

2.3

Student Task

To play the gam e, the students only need to m ake decisions on order release and overtim e on a w eekly basis. The objective o f the students is to m axim ise their total profits at the end o f week 32. A fter playing the gam e for a w hile, students fm d out that ordering on an ad-hoc basis leads them to financial ruin! They are asked to use past data to develope a forecast, w hich, through the m aterial requirem ents planning process, should help them in deciding how m any parts / products to order and when. A capacity requirem ents planning m odule (that they develop) can help them decide how much overtim e to order. Their specific assignm ent is to create a decision support system (D SS) for order release using these concepts. They then use this DSS to play the game and see for them selves how M RP w orks to facilitate accurate order release, in synchronisation w ith the forecast dem and (and to increase their financial perform ance). They create the D SS within the spreadsheet environm ent o f M RP-SIM . Students don't need to do program m ing in V ISU A L BASIC ™ , but they need to be proficient in using spreadsheet softw are. Learning to use spreadsheet softw are is an additional goal o f this assignm ent.

2.4

User Interface

The main screen o f M RP-SIM is show n in Figure 2. The upper left com er shows the products currently being processed by the m achines and their queues (for exam ple, the process M u is currently processing 1200 units o f Delta). It also shows orders placed w ith the vendor. A

B

C

D

F

E

[ ' - 1200

'

H

1

J

E - 3000

L

K

F in a n c ia l P erfo rm a n c e : In ven to ry O ver

1 jS ta tu s an'd W IP of processes:

v

M

L ate

Profit

Vendor

H olding

T im e

D e liv e ry

from

Orders

Cost

Cost

P e n a lty

S a le s

Net P rofit

C - 3000

T his w e e k

0

0

0

0

0

G - 1000

T o ta l

0

0

0

0

0

VlewDedaons

VtewBill o f Materials

Forecast Alpha

|

^

Fast

) A n im a tio n Speed C u rren t W eek

In ven to ry P o sitio n

Day 1

21

VlewMoctel Detail-: In v e n to ry Position:

9000 n

C urrent

8000 7000

P ro d u ct

6000

A lpha

5000 4000

Beta E2 Current

3000

Inventory

2000 1000

0

il ,1

I

in f

□ W o rk in progress

c?

In ven to ry progress 0 700

Custom er Orders °i 0:

300

0

Com m a

1200

3000

0

Delta

2500

1200

0

Epsilon

2500

3000

0

Fi

8500

0

'300

1000

0 0

Gam rna

0 C ustom er Orders

Figure 2. The main screen

236

W o rk in

The low er part o f the screen shows the current inventory position. Current pending custom er orders are also shown. To play the game, the button labelled “Initialise G am e’' is used to initialise the system . This causes a history o f dem and to be created up until week 20. The students can view the current inventory, w ork in progress, and vendor orders at this time. N ext they need to decide on the orders to place for next w eek, their priorities, and overtim e to authorize for the next week. Once they have m ade the decisions, they com m unicate it to the sim ulator by clicking on the “M ake / A lter Decisions” button. Once the decisions are entered, they press the “ Sim ulate!” button to let the production for the week to begin, and to let the tim e advance to the next w eek. The queues o f the m achines, the finishing o f the work, the inventory position, the arrival o f custom er orders, and the filling o f the custom er orders are anim ated on the screen. Profits are accum ulated for every item in custom er orders that is filled. For every late item a penalty is charged per day. There are also costs associated with holding inventory and with overtim e work. The details o f the model m ay be view ed by pressing the “View M odel D etails” button. At the end o f the week the current and cumulative financial perform ance is shown at the top right o f the screen. The students then m ake decisions for the next w eek and repeat the cycle.

2.5

The Assignment

Students are asked to play the gam e in an ad-hoc m anner, w ithout any decision support, to fam iliarize them w ith the sim ulator and to see how well they can perform w ithout the forecasting, M RP, and CRP m odels. They are then asked to build a DSS consisting of these m odels to help them play the game. In building the DSS they can create initial forecasts from the 20 weeks o f historical data. The forecasts and the M R P need to be updated as new data becom es available. All this is done within the spreadsheet environm ent. The students are asked to hand in their DSS (in a diskette) and a sem i­ structured reflective essay on the game, their experiences, and their understanding o f the concepts. The assignm ent consists o f three steps: Step 1. Fam iliarisation with the sim ulator. The students play the gam e in an ad hoc m anner, guessing the decisions. Step 2. Playing the gam e on a reorder point basis. The students try different levels o f reorder points and fixed order quantities. Step 3. Playing the gam e w ith a DSS, built by the students. To do this, they use a forecasting m odel to forecast dem and o f the finished products. This is fed into the m aster production schedule (M PS), w hich is exploded into the M RP for the com ponents. The students then create the CRP m odel from the M RP model. This com pletes the DSS, w hich suggests the order quantities for all the items, and the overtim e to order for all the processes. Formal assessm ent for this assignm ent consists o f a reflective essay (50% o f the assignm ent m arks) and the DSS (50% ). How ever, this assignm ent had a w eight o f 4% in the overall course assessm ent, thus there was a risk that students could choose to give it minim al attention. The assessm ent criteria were reflection breadth (num ber o f activities/item s discussed) and reflection depth (thoroughness o f discussion, depth being more im portant than breadth), and com pleteness/correctness o f the DSS work.

237

3

Learning and the Simulation

The EL cycle (Figure 1) underpins m any processes (Scott, 1990), including sim ulation. In the cycle, reflection is the stage that separates a sim ulation experience from tying it to m odels and concepts used to im prove perform ance. W hile our students have some reflection skills, encouraged by som e fam iliarity in our first year course with simple concepts such as Positive, M inus, Interesting (PM I) o f Edward de Bono, and Single­ loop (SLL) versus Double-loop Learning (D LL) (Figure 3) o f Chris Argyris (A rgyris,1977 ), reflection still appears to be the w eakest link in their EL cycle, for w hich no form al instruction is given. Also, sim ulation models can be seen as devices to support reflection before action. We therefore chose reflection as a parallel, learning focus for this study.

Single Loop:

Action

Desire/ Expectation

t Double Loop:

Result/Consequence

Compare

Desire/ Expectation

Action

I A A A

Result/Consequence

Compare Expectation appropriate? Action best available ? Assumptions reasonable?

Figure 3. Single-loop versus double-loop learning W e decided to: ■ track the m ain activities students used before and during each major step within the M R P-SIM assignm ent, using activity logs, ■ create tim e m aps from the activity logs, ■ record the levels o f reflection w itnessed, and ■ stim ulate learning aw areness by discussing, in class, the results with the students, before they finalized their reflective essays, w hich were part o f the assessm ent scheme. Tracking the m ain activities was done using a sem i-structured, one-page questionnaire w e called an activity log - one for each m ajor engagem ent w ith the software. Each log asked students to record the tim e spent, and thinking behind, each significant task their group had undertaken. C reating tim e m aps o f the steps was simply done by graphing tasks versus (proportion of) tim e spent, over the engagement. One graph for each significant M R P-SIM engagem ent was created, using different colours for each student group. With the M R P-SIM assignm ent requiring students to set levels o f production, to observe the result/outcom e and to revise their estim ates, potential levels o f reflection w ere clearly related to Single-loop and D ouble-loop Learning. Four levels were set: (0)

238

trial and error, to no plan, (1) discussed/predicted what would happen before num bers w ere entered (Sll), (2) questioned m odel or process being used to predict the result/outcom e (Dll), and (3) questioned the process being used in step (2).

4

Results

The assignm ent was distributed in class and the first lab session held, 14 days later. As “entry tickets", the activity logs for ad-hoc and re-order point running o f the sim ulation w ere collected. The second lab session was held, 2 days later. The tw o lab sessions were essentially help sessions, offering individual help to the class m em bers in creating their DSS. Students can often be unsure o f the concepts em phasised in this assignm ent, or even the purpose o f the DSS. M any students struggled to acquire the level o f spreadsheeting skills needed for the assignm ent. The two lab sessions offered instructions in these m atters. The spreadsheet files and the DSS activity logs w ere then subm itted for assessm ent. Discussion o f the results and feedback was given in class, 21 days before their reflective essay on the assignm ent was subm itted. The reflective essay was for three pages o f thoughtful responses to reflective questions we had provided. The students were given three activity logs to fill in, corresponding to three w ays, or steps, o f playing the g am e:- in an ad hoc way, with reorder points for each item s, and with the help o f the DSS created by them . Students filled in the tim e they spent on various activities. All possible activities w ere identified in the logs, including the four dealing with reflection, but the students w ere not told the learning levels o f the logged items until the discussion o f results and feedback session. This session was deliberately held before the final reflective essays were subm itted. W hen presented w ith the learning levels o f their activities and the tim e they spent on activities, most o f the students were surprised at the very high tim e they spent on level 0 activities (trial and error or w ithout a plan), as against single loop or double loop learning.

5

Student E valuation

This game and assignm ent com bination was used in an operations m anagem ent course offered at the final year o f a B achelor o f M anagem ent Studies degree at the University o f W aikato in Ham ilton, N ew Zealand. The assignm ent was a part o f the m odule covering production planning and control. The students were given the assignm ent after covering the concepts o f m aterial requirem ent planning in the class. A t the end o f a 1998 course, 10 students evaluated the gam e anonym ously; in 2002, 6 students. A fter a 2002 course, 5 students did a sim ilar evaluation. Their average rating on a scale o f (1 = strongly agree, 5 = strongly disagree) is presented below in Table 1. Table 1. Student evaluation o f the game

Average agreement rating 1998

Statement Sim ulation gam ing was instructive for learning production planning.

2.0

I enjoyed playing the sim ulation game.

1.9

239

Average agreement rating 2002 1.2

1.6

I put a lot o f effort in playing the sim ulation game.

2.1

1.4

The sim ulation gam e represents fairly well the decision m aking faced by real production planners.

2.1

1.8

I found the sim ulation gam e challenging.

1.8

1.4

There w as a strong sense o f make believe in playing the game.

2.2

2.0

I felt the gam e enhanced my understanding o f planning and control.

1.2

The gam e helped me im prove my use o f spreadsheet software.

1.4

The gam e assignm ent should be retained for next year.

1.8

1.0

(t N ot included in the 1998 evaluation) G enerally, the students found the assignm ent quite challenging. But they felt that they learnt the M RP concepts pretty well. Since the students had some fam iliarity w ith spreadsheets, the spreadsheet form at o f the gam e helped in gaining student acceptance. Their evaluations bore testim ony to this. One student com m ented: “ ... Therefore, using the DSS helped me play the game better and also my profits had increased as well (i.e. started m aking profits rather than losses). I believe the reason for im proved results was due to taking more accurate and precise figures into account when planning the future productions. However, with DSS, decisions w ere still based on using my own judgm ent and the gam e was still played with great deal o f uncertainty due to reliance on forecasting figures. But to m ake it m ore realistic in the essence to m ake the user o f the M RP Sim ulation believe it is the real w orld and to be able to im agine them selves in that environm ent, I think adding more financial data such as how m uch overtim e and late delivery is really costing for each com ponent (not ju st the total), and more costing infonnation would enrich a person’s m ind.” It is clear from student com m ents and evaluations (Table 1) that the students did achieve a good understanding o f the production planning concepts, w hich was the m ain goal o f this exercise. Some students also got a feel o f the uncertainties involved and gained an appreciation o f the fact that a DSS provides decision support, but does not supersede hum an decision m aking. It w ould be stating the obvious to say that the spreadsheet w ork ranged from overw helm ing to underwhelm ing, depending on the students’ previous experience. However, m ost students found that this assignm ent took up m uch m ore tim e than they would have expected or w ished, relative to the 4% assessm ent w eighting it carried.

240

Suggested im provem ents included m aking a com petition out o f it, w ith the highest profit m aking students gaining rew ards and high m arks. A nother student (accounting) suggested more detailed reporting o f the financial outcom e o f the game.

6

C onclusion

In this paper we presented a spreadsheet-based sim ulator for teaching / learning production planning and control concepts such as forecasting, m aterial requirem ents planning, order review and release. The game received a favourable response from the students. The spreadsheet fonnat helped gain acceptance. A lthough the students were not judged on their fm ancial perform ance, they did develop a rivalry to gain the highest profit. This substantially increased the m otivation in playing the gam e. A focus on levels o f reflection attained added a parallel learning focus. This gam e enhances the understanding o f PPC concepts as well as provides the students with an opportunity to build a decision support system that provides the gam e players with detailed m odelling and decision m aking capability, and to learn the difference between ad-hoc decisions and m odel-based decisions. Learning spreadsheet m odelling is an additional educational benefit from the game. Vaszonyi (1993) and Plane (1994) present forceful arguments for using spreadsheets in m anagem ent science / operational research. The low proportion o f tim e spent on reflection - m aking sense o f action, connecting theory and planning for further action - supports the view that our students should benefit from greater understanding o f reflection and its praxis before playing the gam e. There are four aspects we have identified that could be added to the learning aw areness aspect in the future: ■ investigating if tim e spent developing spreadsheeting skills correlates w ith reflection patterns witnessed, ■ recording the order in which tasks w ere done, although this w ould com plicate recoding and analysis, ■ relating time m aps and ordering o f tasks to learning styles, ■ requiring m ore specific reflection on the reflective aspects in the final essay, and ■ discussing w ith future classes taking the assignm ent, the outcom es from this year, thereby seeding interest and aw areness o f reflection pre-assignm ent. (The sim ulation m odel and the related assignm ent m ay be obtained by w riting to the authors).

R eferences Argyris, C., 1977. “D ouble loop Learning in O rganizations.” H arvard Business R eview , Sept-Oct 115-125. Biggs, W.D., 1987. “Functional B usiness G am es.” Simulation and Games 18:242-267. Churchill, G.. 1970, JobJot: A Production M anagement Gam e, The M acm illan Co., U.K. Gilgeous, V., and M. D'Cruz, 1996. “A Study o f B usiness and M anagem ent G am es.” Management Developm ent Review 9:32-39. Kolb, D.A., 1984, Experiential Learning: Experience as the Source o f Learning and Developm ent, Prentice Hall, U.S.A. Lane, D.C., 1995. “On a R esurgence o f M anagem ent Sim ulations and G am es.” Journal o f the Operational Research Society 46:604-625.

241

Mize, J.H, B.E. Herring, C.L. Cook, M.S. Chun, and C.R. White, 1971, Production System Sim ulator (PROSIM V): A User's M anual, Prentice-H all, U.S.A. Plane, D.R., (1994). “ Spreadsheet Pow er.” OR/MS Today 20:32 38. Scott, J.L., 1990. “ OR M ethodology and the Learning C ycle.” OMEGA International Journal o f M anagem ent Science 18:551-553 Scott, J.L., 2002. “ Stim ulating A w areness o f Actual Learning Processes.” Journal o f the Operational Research Society 53:2-10 Vaszonyi, A., 1993. “ W here We Ought to be Going: The Potential o f Spreadsheets.” Interfaces 23:26-39. Yeo, G.K. and F.H. Nah, 1992. “A Participants' DSS for a M anagem ent Game w ith a DSS G enerator.” Simulation and Gaming 23:341-353.

242