MODELING &
7TH INTERNATIONAL WORKSHOP ON ApPLIED SIMULATION
H S ~
2008
MECHANICAL DEPARTMENT, UNIVERSITY Of CALABRIA
M&S CENTER· LABORATORY Of ENTERPRISE SOLUTIONS DIPTEM, UNIVERSITY OF GENOA
~
llOPHANT SIMULATION
ISBN 978-88-903724-1-4
THE 7TH INTERNA.TIONAL WORKSHOP ON MODELING Ii ApPLIED SIMULATION SEPTEMBER 17-192008 CAMPORA S. GIOVANNI (AMANTEA, (5), ITALY
EDITED BY AGOSTINO BRUZZONE PRISCILLA ELFREY ENRICO PAPOFF MARINA MASSEI ISTVAN MOLNAR
PRINTED IN RENDE (CS), ITALY, SEPTEMBER
2008
THE
7TH INTERNATIONAL. WORKSHOP ON M.ODELING a ApPLIED SIMULATION CAMPORA S. GIOVANNI (AMANTEA), ITALY SEPTEMBER 17-19, 2008
ORGANIZED BY MECHANICAL DEPARTMENT, UNIVERSITY OF CALABRIA MSC-LES, MODELING & SIMULATION CENTER, LABORATORY OF ENTERPRISE SOLUTIONS
DIPTEM - UNIVERSITY OF GENOA
LIOPHANT SIMULATION
SPONSORED BY ACADEMIC, INSTITUTES AND SOCIETIES SPONSORS IEEE - INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS SCS - SOCIETY FOR COMPUTER SIMULATION INTERNATIONAL MISS - McLEOD INSTITUTE OF SIMULATION SCIENCE
••
1:5
M&SNET - MODELING & SIMULATION NETWORK IMCS - INTERNATIONAL MEDITERRANEAN & LATIN AMERICAN COUNCIL OF SIMULATION NASA - KENNEDY SPACE CENTER
INDUSTRY SPONSORS MAST - MANAGEMENT AND ADVANCED SOLUTIONS AND TECHNOLOGIES TONNO CALLI PO lIQUIRIZIE AMARELLI GIULIO BARCA PRODonl IN PELLE
III
Index Inventory Management Simulation A STUDY OF MATERIALS MANAGEMENT SYSTEM IN A LARGE MINING ORGANIZATION, Sharif MODELLING AGRICULTURAL PRODUCTION SYSTEMS USING AN "ACTION-FLOW-STOCK" ONTOLOGY, Franc;ois Guerrin
7
STOCHASTIC LEADTIMES IN A ONE-WAREHOUSE, N-RETAILER INVENTORY SYSTEM WITH THE WAREHOUSE CARRYING STOCK, Adriano Solis, Charles Schmidt
13
WAREHOUSE INVENTORY MANAGEMENT BASED ON FILL RATE ANALYSIS, Antonio Cimino, 23
Duilio Curcio, Giovanni Mirabelli, Enrico Papoff
EFFECT OF NEGATIVE PARAMETERS IN INVENTORY MODELS PERFORMANCE, Miguel Cezar 31
Santoro, Gilberto Freire
THE MODELLING AND CONTROL OF THE AGRICULTURAL SET DRIVER, Jacek Kromulski, Tadeusz Pawlowski
37
WAREHOUSE AND INTERNAL LOGISTICS MANAGEMENT BASED ON SIMULATION, Enrico Bocca, Duilio Curcio, Francesco Longo, Alberto Tremori
MODELING
&
ANALYSING INVENTORY MANAGEMENT PERFORMANCE BY SIMULATING SUPPLY CHAIN MANAGEMENT STRATEGIES, Gokhan YUzgaLec, Markus Witthaut, Bernd Hellingrath
41
49
Logistics, Manufacturing & Supply Chain Management SPS TOOLS FOR CAPITAL PLANNING PROJECT ANALYSIS, Nathan Boskers, Simaan AbouRizk
55
USE OF DISCRETE EVENT SIMULATION FOR A LONG RANGE PLANNING OF AN EXPEDITON SYSTEM, Leonardo Chwif, Afonso Celso Medina, Jose Arnaldo Barra Montevechi, Marcos Ribeiro Pereira Barretto
63
SUPPLY CHAIN PERFORMANCE UNDER TRANSIENT DEMAND INCREASES: A CASE STUDY SUPPORTING SUPPLY CONTRACT NEGOTIATION, Ivor Lanning, Cathal Heavey
68
WORKLOAD FORECAST ALGORITHM OPTIMIZATION FOR RE-ORGANIZING RETAIL NETWORK, Agostino Bruzzone, Simonluca Poggi, Enrico Bocca, Francesco Longo, Francesca Madeo, Sabrina Rondinelli
77
A SIMULATION-BASED METHOD FOR THE DESIGN OF SUPPLY STRATEGIES TO ENTER DEVELOPING MARKETS, Christian Schwede, Yu Song, Brian Sieben, Bernd Hellingrath, Axel Wagenitz
83
Decision Support Systems Applications DECISION SUPPORT SYSTEM FOR UREA SYNTHESIS SYSTEM OF A FERTILIZER PLANT, Sunand Kumar, Sanjeev Kumar, Puran Chandra Tewari
IX
93
THE INDUSTRIAL INNOVATION PROCESS BY A NETWORK APPROACH, Nelson Ebecken, 1e
Francisco Moreira
DECISION SUPPORT SYSTEM APPLIED TO COMBINED FREIGHT TRANSPORT, Nicolas Rigo,
1C
Alassane Balle Ndiaye
A SIMULATION-BASED DSS FOR FIELD SERVICE DELIVERY OPTIMIZATION, Mario Rapaccini, Filippo Visintin, Alessandro Sistemi
11
ADVANCED MODELS FOR INNOVATIVE DECISION SUPPORT SYSTEMS IN BROADCASTING SCHEDULE PLANNING & MANAGEMENT, Agostino Bruzzone, Marina Massei, Luca Pierfederici
12.
CONSIDERATIONS ON THE PARTICULAR FEATURES FOR PROCESS AND WORKFLOW MODELING, Victoria lordan, Alexandru Cicortas
He
FORMAL METHOD OF FUNCTIONAL MODEL TRANSFORMATIONS, Janis Grundspenkis, leva Zeltmate
14C
BUILDING
BASED
ON
GRAPH
ANALYSIS OF LOGISTICS IN SUPPORTOF A HUMAN LUNAR OUTPOST, William Cirillo, Kevin Earle, Kandyce Goodliff, J.D. Reeves, Mark Andraschko, Gabe Merrill, Chel Stromgren
148
A DECISION SUPPORT METHODOLOGY FOR PROCESS IN THE LOOP OPTIMISATION, Dan Gladwin, Paul Stewart, Jill Stewart, Rui Chen, Edward Winward
Advance in Information and E-Integration Healthcare Systems and Management
158
in
ORGANISATIONAL ANALYSIS OF DIFFERENT MODALITIES IN DRUG ADMINISTRATION, Lucio BUffoni, Libero Ciuffreda, Antonio Di Leva, Salvatore Femiano
164
A REACTIVE SCHEDULING FOR INTESIVE CARE UNITS, Erhan Kozan
170
MODELLING AND DESIGN OF HOSPITAL DEPARTMENTS BY TIMED CONTINUOUS PETRI NETS, Mariagrazia Dotoli, Maria Pia Fanti, Agostino Marcello Mangini, Walter Ukovich
175
NETWORKS OF QUEUES WITH MULTIPLE CUSTOMER TYPES: APPLICATION IN EMERGENCY DEPARTMENT, Jihene Jlassi, Abderrahman Elmhamedi, Habib Chabchoub
181
,
CRITICAL NEWBORN TRANSPORT IN VENETO REGION: MODELS AND SIMULATION, Francesca Bortolato, Anna Ferrante, Giorgio Romanin-Jacur, Laura Salmaso
187
Industrial Engineering Modeling and Simulation A NON-NEWTONIAN FLOW MODEL FOR SIMULATING THE FABRIC COATING PROCESS, Xiaoming Zhao, George Stylios
193
APPLICATION OF SIMULATION MODELLING TO THE SHIP STEAM BOILER SYSTEM, Enco Tireli, Josko Dvomik, Srdan Dvomik
198
CONTROLLER DESIGN USING COMBINATION OF SYMBOLIC AND NUMERIC CALCULATIONS IN MAPLE, Pavol Bistdk, Peter fapak
204
x
BEZIER FITTING TO ALMOST OVAL GEAR DEVICES, Karol Gajda, Piotr Krawiec, Adam Marlewski
210
MODELING AND SIMULATION NEEDS IN FUSION ENERGY RESEARCH, Gabor Veres
217
REAL TIME COLLISION DETECTION FOR COMPLEX SIMULATIONS BASED ON HYBRID MULTI RESOLUTION APPROXIMATION OF CAD MODELS, Ilario Francesco Ceruti, Giovanni Dal Maso, Diego Rovere, Paolo Pedrazzoli, Claudio Roberto Boer
225
OPTIMIZING A HIGLY FLEXIBLE SHOE PRODUCTION PLANT USING SIMULATION, Fred Voorhorst, Antonio Avai, Claudio Boer
233
SIMULATION CALCULATION OF TRACTOR-POTATO PLANTER COMBINATION MODEL, Jan Szczepaniak
240
THE WEB WINDING SYSTEM CONTROL BY THE BACKSTEPPING METHOD, Nabila Rabbah, Bahloul Bensassi
246
SIMULATION STUDY OF AN AUTOMATED AIR CARGO TERMINAL, Bill Chan, Henry Lau, Steve Chan
252
A SEQUENTIAL HEURISTIC PROGRAMMING APPROACH FOR A CORRUGATED BOX FACTORY: TRADEOFF BETWEEN SETUP COSTS AND TRIM WASTE, P. Zouein, J. Diab
261
WORKSTATION PRODUCTIVITY ENHACEMENT WITHIN HYDRAULIC HOSES MANUFACTURING PROCESS, Antonio Cimino, Duilio Curcio, Francesco Longo, Enrico Papo!!
268
INTEGRATED PLANNING AND CONTROL ON RO-RO TERMINALS, Bernd Scholz-Reiter, Felix Bose, Michael Teucke, Jakub Piotrowski
275
PROJECT OF AN AGV TRANSPORT SYSTEM THROUGH SIMULATION TECHNIQUES, Vincenzo Duraccio, Domenico Falcone, Alessandro Silvestri, Gianpaolo Di Bona
281
MODELING AND SIMULATION OF' BEHAVIORAL SCENARIOS BY USING COUPLED DEVS MODELS, Mamoun Sqali, Lucile Torres
285
Agent and Service based on Modeling & Simulation WSDL-BASED DEVS AGENT FOR NET-CENTRIC SYSTEMS ENGINEERING, Saurabh Mittal, Bernard Zeigler, Jose Risco-Martin, Jesus de la-Cruz
292
MODELLING AND SIMULATION USING STATECHART-BASED ACTORS, Franco Cicirelli, Angelo Furfaro, Libero Nigro
301
MULTI-AGENT BASED SIMULATION OF LARGE RANDOM BOOLEAN NETWORK, Pham Dang Hai
308
MAS General Session SIMULATION OF THE IMPACT OF THE ENERGETIC CHARACTERS OF TRACTORS AND MACHINES ON THE WORKING EFFICIENCY OF THE SOIL TILLAGE UNITS, Arvids Vilde, Edmunds Pirs
314
A SHORT GUIDE TO CLIENT'S SATISFACTION FROM A SIMULATION MODEL, Michal Stec
321
XI
SIMULATION OF THE GOMS KEYSTROCK LEVEL MODEL USING DEVS, Ali Mroue, Jean Caussanel
326
SIMULATION OF NATURAL PHENOMENA BY CELLULAR AUTOMATA WITH THE L1BAUTOTI LIBRARY; AN APPLICATION TO GEOLOGICAL MODELLING, WiWam Spataro, Giuseppe Spingola, Giuseppe lito, Donato D'Ambrosio, Rocco Rongo, Maria Vittoria Avolio, Salvatore Di Gregorio
331
Author's Index
340
XII
A SIMULATION-BASED DSS FOR FIELD SERVICE DELIVERY OPTIMIZATION Mario Rapaccini(a l, Filippo Visintin(bl , Alessandro Sistemi(cl
(",b,c)
University of Florence, Dipartimento di Energetica "Sergio Stecco", Via Cesare Lombroso 6/17 - 50134, Firenze (al
[email protected], (b)
[email protected],
(c)
[email protected]
It is in fact tremendously expensive and time
ABSTRACT This paper aims at presenting the preliminary results of a research targeted at developing a Decision Support System (DSS) for the Field Service Delivery System (FSDS) design. The paper is organized as follow: firstly, we illustrate what a FSDS is; secondly, we identifY the variables to consider in order to design a FSDS and the relationships between them; thirdly, we describe how a DSS supporting the FSDS design should be developed; and finally, we show the result of of a pilot experiment in which a DSS has been developed and applied to a real case study.
consuming, to modifY the FSDS configuration ex-post The aim of this paper is thus to discuss how a Decision Support System (DSS) for designing the FSDS could be developed and to illustrate a first example of a DSS, The paper is therefore organized as follows: firstly, we define and describe a generic field service delivery process; secondly, we identifY all the variables that service managers should take into account to design the FSDS; thirdly, we assess if and where the information relevant to these variables can be found in the most common Enterprise Information Systems; fourthly, we describe how the discrete event simulation can be successfully used to create such a DSS; and finally, we show the result of a pilot experiment in which a DSS has been developed and applied to a real case study.
Keywords: field service delivery system (FSDS), decision support system (DSS), discrete event simulation, after sales service 1. INTRODUCTION This paper aims at describing a tool that can support the management of a Field Service Delivery System (FSDS), that is, the company's function devoted to deliver services - such as the product's installation and maintenance - directly at customer's site. The rationale of the paper lies in the fact that, despite the increasing interest raised by field service in many industries and the large number of applications supporting the field service operations management, there is a lack in models/devices supporting the design of the FSDS (Visintin 2007), A great deal of information is usually available in the Enterprise Information Systems, but generally there are not tools allowing service managers to fully and properly utilize these information, in order to understand the effects that different managerial policies can have on the overall system performance (Agnihotri and Karmarkar 1992). Moreover, the FSDS design activity is a very complex task. It requires, in fact, to forecast when and where the service requests will arise, to figure out what skills and parts will be required in order to fulfill the service requests, and to decide what criteria should be followed to dispatch the technicians. In addition to that, service managers are asked to achieve increasingly high performances, both in terms of customer satisfaction and costs. Because of the complexity and the uncertainty characterizing the FSDS design there is a strong need of simulation-based tools supporting service managers.
2. THE FIELD SERVICE DELIVERY SYSTEM A FSDS is made ofa set of technicians, each mastering a given set of skills and covering a given geographic area, that are remotely dispatched at the customers' sites to fix the customer's problems upon request (Blumberg 1991; Visintin 2007). A typical field service process can be subdivided in three main activities:
L 2. 3.
help desk; dispatching; service delivery.
Figure 1 describes a typical field service delivery process. The Help desk is the company interface with the customer, so it is devoted to receive the incoming service request, to identifY the customer and to offer (when it is possible) a remote assistance, that is, to avoid the service delivery on-field. If the call avoidance does not succeed, then comes the need of selecting and dispatching a field engineer to the customer (Agnihotri and Mishra 2004). Finally, the selected technician gathers all the needed technical data and actually visits the customer In designing and managing this process, service managers need to take into account several variables
116
(M~mUlll
and Chakravarty 2005), that we present in next paragraph.
•
the technicians' Dispatching policy, that is, how a service request is assigned to a field engineer; • the technicians' Scheduling policy, that is, how the amount of service requests is planned to be served by the field engineers throughout the working day; • the Districting policy, that is, the criteria to follow in order to assign each technician a territory; • the Cross-training policy, that is, the criteria to follow in order to assign each technician a given set of skills (Agnihotri, Mishra and Simmons 2003; Upton 1994); • the Spare parts management policy, that is, the way spare parts are stored, reordered and delivered to the customers. The independent variables, are those out of the service manager's control, while the dependent variables, are functions of the independent and the control variables. In Table 1 the aforementioned variables are described in detail, while Figure 2 shows the delivery system design process" As we see, the independent variables represent the input for such a process, because they identify the context in which managers have to work and that they cannot control. The values of the independent variables are totally known before taking any system designing decision. The control variables, instead, represent - to some extent - the output of the decision making itself. They are, in fact, the aspects that managers have to decide about, in order to design a FSDS. We can therefore say that the values ofthe control variables are the results of the decision making process, Finally, the dependent variables are those measuring the outcome and thus the effect of the decision taken by the service manager. We can identity them with the actual system's operational and financial performances (Agnihotri 1989)"
Figure 1: Field service delivery process description 3. FSDS DESIGN PROCESS VARIABLES The design of a FSDS is a decision-making process characterized by several independent, dependent and control variables (Agnihotri, Narasimhan and Pirkul, 1990). The control variables are those that managers can manipulate in order to obtain, in a given context, a desired outcome. They concern:
•
the system Capacity, that is the overall number of technicians;
Figure 2: Delivery system design process 117
. process . bl e~ III the derIvery system deSlgll ·Table 1: Independent, contro an dd ependent vana Control variables:
Independent variables: 1. Geographical distribution ofthe demand for skills.. It depends on: a. the number of products to serve b. the Mean Time Between Failure of each type of product (and in case the different failure modes) c. the skills required to fix each type of product (and in case the skill required for each failure mode) d. the products' location 2. The current geographical distribution of the supply of skills. It depends on: a. the number of field technicians available b . the locations (home address) of the field technicians available c.. the skills mastered by each of field technician
_.,,_........ ..............._"" ,,~
Dependent variables
4. Operational perfonnance:
a. Capacity: number offield technicians to employ
a. b. c. d. e. f. g. h.
b. Cross-training: skills to impart to each field technician
c. Districting: territories to assign to each field technician
d. Dispatching: algorithm to use to dispatch field technicians
e. Scheduling:
is..
downtime response time travel time answering time first contact resolution rate SLA compliance resource utilization etc...
Financial perfonnance : a. revenues b. costs c.. cash flows d. etc ....
algorithm to use to schedule field technicians f.
Spare parts management policy: location and quantity of spare parts to keep in inventory
3. Geographical distribution of the demand for spare parts.. It depends on: a. number oftasks requiring spare parts b.. spare parts/store location c.. availability of spare parts
4. The target operational and
.......
_..
\
financial performance that the delivery system is supposed to meet . .........."..._. . _..... ....
_-_.
__ __ __._
_
_.__._..... ................ "
__._
4. INFORMAnON NEED As can be noticed the independent variables are a set of data that should be easily retrieved from the Enterprises' Information System (EIS). As a matter of the fact: I.
2.
-"......
...................
3
all the data regarding the installed-base (i. base), the contracts, the service requests, the technicians and the spare parts are usually available in the companies' Enterprise Resource Planning system (ERP system); the i-base is usually geo·referred with Geographical Information Systems (GIS) and the same instruments allow to define the areas the field technicians are assigned to;
_ _ ...
""~
......
...... " ....,,_.....'
the product performance in terms of reliability and availability can be easily obtained through statistical tools, starting from the data available in the ERP system, evaluating in particular the functioning time intervals of the machines as the difference between two consequents service requests regarding the same item
These three sources should provide all the needed data. The information flows that allow the DSS (represented by a simulation tool) to simulate the real situation are shown in Figure 3.
118
discrete event simulation (Chung 200.3; De Felice 2007; Law and Kelton 2000). The use of discrete-event simulation, instead of other techniques such as the queuing theory, is fully justified by the number of variables to consider and the randomness of the phenomena to model (Banks 1998; Kelton and Sadowski 200.3; Perros 2007). A representation of the conceptual model that we have developed is shown in Figure 4. In the model the service requests are entities characterized by several attributes:
• Figure.3: Information flows
•
5. THE MODEL A DSS to support the FSDS design should therefore be able to retrieve the required data from the companies' information systems and to perform simulations to test the effectiveness of different managerial policies" Such a tool has thus to allow to:
•
• •
• •
product type (and thus the needed technical data); problem type (and thus the needed skills ofthe technician); installed base localization (and thus the needed location ofthe technician); service level agreement (and thus the performance constraints); need/or spare parts
While the process is in progress the data about the problems occurred are computed by the statistical tool and the values ofthe operating reliability are updated" Each incoming entity has to be assigned to a technician and could require some spare parts, Both technicians and spare parts are resources with limited capacity, characterized by attributes such as the geographical location and the service requests they can be used for.
evaluate the variables that are actually relevant in the FSDS designing; measure - ex-ante - the effects that the typical decisions taken by service managers could have over the delivery system performance.
Due to the complexity characterizing the field service environment, the methodology adopted for modeling the field service system should be the
Figure 4: Conceptual model
119
These resources' attributes depends on the chosen managerial policies. Different policies lead to different values of: • • • •
the statistical significance of the effects of the control variables over the overalI system's performance, In order to build and apply the model to the case study, we performed:
overalI number offield technicians; dimension of the area the technicians have to cover; skills distribution over the technicians; spare parts location and quantity.
• • • • •
The model logic, based on the managerial policies, assign the entities to appropriate resources, creating an entities' flow. Fundamental element of the simulation model is the dispatching algorithm (implemented in the model logic), that is, the working rule of the model. This algorithm depends on the values of the control variables and regulates the entities' flow, considering resources' attributes and availability. It's therefore clear that different managerial policies lead to different algorithms and so to different results of the service delivery process. The outcome of the process can be finalIy calculated in terms of: • • • • • •
The field service delivery process we identified totalIy folIows the model we showed above. We have been able of retrieving all the needed data from the company information system. In particular, the data we considered are the folIowing: • • • •
resources utilization, lead time, downtime, percentage of fixed requests, constraints compliance, etc...
• • • •
The DSS we describe has to be suitable for every kind of company that has to cope with field service delivery problems. Certainly, each specific application needs some efforts to effectively be able of representing the situation. These efforts aim at analyzing the business processes, evaluating the data availability, defining the rules and the algorithms that are suitable for the case (Pidd, 1992; Ross, 1990). In general, to create a functional DSS, we have to: 1.
2.. 3.. 4.
business processes analysis, conceptual model ofthe system, data colIection and manipulation, actual creation of the simulation model, experimental design..
• • • •
• •
develop and verify reliable, flexible and parametric models, able to reproduce "in vitro" different field service delivery systems; validate the models making use of the real data; use the validated models to carry-out scenario analysis and experiments; develop and test algorithms for the scheduling, dispatching and districting optimization by means of the simulation models themselves..
service request arrival date and time, geographical location of the products (i-base localization), characteristics of the products requiring support (product type), type of the problems that can occur to each type of products (problem type), failure modes and possible solutions, needed skills (in dependence of the problem type), needed spare parts (in dependence of the problem type), skills profiles, availability and work areas of the field technicians, instantaneous position of the technicians, spare parts availability and location, average travel time and work time required to fix the problem, target operational and financial performance contractually defined (e.g" Service Level Agreement - SLA), service contract validity, failure rates characterizing the i-base (operating reliability)"
The simplified tool we developed is not able of retrieving automatically the required information from the EIS. We therefore manualIy extracted the data from the ERP system and the GIS trough worksheets, we performed reliability analysis with a statistical tool and we used the obtained data as an input for the simulation model we created with Rockwell Arena. After having created the simulation model, we found its stability parameters (Guttman, Wilks and Hunter 1971; Montgomery 2002) and then we performed a Design Of Experiment (DOE) (Box and Hunters 1978; Mood, Graybill and Boes 1974) selecting three of the control variables described above: dispatching, cross-training and districting. What we did with this variables is perfectly extendible to the complete system, it's just a matter of data amount, time and computational power,
6. CASE STUDY As a pilot experiment we tried to model the FSDS of a big multinational company that manufactures and services office imaging products. Through the simulation tool (developed making use of the RockwelI Arena © suite) we have assessed
120
...
. J..
A 2k DOE analysis showed the influence that changes in the control variables (and in their interaction) give to the overall system's response. We defined 2 different levels for each control variable and we ran the model in each of the 8 different possible combinations showed in Table 3.
In Figure 5 and 6 the comparison between the different possibilities is shown, considering the mean downtime of the products to serve as a performance indicator. A treatment is the variation that is felt while passing from the "-" level to the "+" level in each of the three variables (1 - dispatching, 2 - cross-training and 3 - districting), or contemporaneously in more of them (12, 13,23, 123) (Rotondi 2005). The proportional variation is the effect that treatments causes on the overall system response.
Table 2: Control variables considered in the case study
High flexibility
Low flexibility
4 zones
2 zones
Table 2 shows the levels we defined for the control variables. For the dispatching, the levels are: •
Shortest Time in Queue (STQ) policy, that is, the entity is assigned to the resource with the minimum expected waiting time; Shortest Number in Queue (SNQ) policy, that is, the entity is assigned to the resource with the minimum queue length.
•
For the cross-training, the levels are: •
high flexibility policy, that is, all technicians are able of fixing 3 problem types; low flexibility policy, that is, all technicians are able of fixing 2 problem types.
•
2
+3.8%
12
+5.0%
Figure 5: Performance variation due to the treatments the field different
Mean downtime ~ i/········ ..··_· .. ·~~····
the field different
45
Average
4
35 3 Z.!>
For the districting, the levels are: •
narrow districting policy, that is, a territorial partition in 2 zones; broad districting policy, that is, a territorial partition in 4 zones.
•
lim Meandowntime
2 U 1
05
o 3
12
13
Z3
113
Figure 6: Evaluation of the impact of the considered factors on the performance
Table 3: The 8 variables combinations ofthe 2k DOE
The analysis of variance we performed on the simulation model's results, demonstrates that the control variables are statistically significant with a confidence interval of 95%. This says that all the selected factors and their interactions give a significant impact on the performances of the FSDS, so it is correct to analyze them. In this case the most important element to set seems to be an appropriate dispatching policy (1), followed by a correct districting (3) and then by a proper training and skills spreading policy (2). We also notice (13) that the performance improvement given by a change in the dispatching policy is much more marked if it is applied together with a broad distr'icting policy (more dispatching alternatives imply higher performance differentials)..
H
121
AUTHORS BIOGRAPHY . Mario Rapaccini took his Laurea Degree with honors in Mechanical Engineering (5-yr-course) at Florence University in April 1996. He is a professional engineer since 1996. In may 2000 he achieved PhD.. (3-yr-course) discussing the thesis "Advanced tool for configuration and impact assessment of Integrated Municipal Solid Wastes Management Systems". Currently, he's assistant professor in SSD ING-IND/35. Research topics covered are: managerial economics and business organisation, ICT, simulation modeling and analysis (SM&A), business process re-engineering (BPR). He's fellow of AiIG, EurOMA, ANIMP, AIRO andANIPLA.
REFERENCES Agnihothri,S., 1989. Interrelationships Between Performance Measures for The Machine Repairman Problem, Naval Research Logistics, VoL 36, No.3. Agnihothri S., Chakravarthy S., 2005. Optimal Workforce Mix in Service Systems with Two Types of Customers, Production and Operations Management, Vol. 14, No.2, pp. 218-231. Agnihothri, S., U. Karmarkar, U., 1992. Performance Evaluation of Service Territories, Operations Research, Vol. 40, No.2. Agnihothri, S., Mishra, A., Simmons, D., 2003. Workforce Cross-Training Decisions in Field Service Systems with Two Job Types, Journal oj the Operational Research Society, VoL 54, No.4, pp.. 410-418. Agnihothri, S., Mishra, A., 2004. Cross-Training Decisions in Field Services with Three Job Types and Server-Job Mismatch.. Decision Sciences, vol. 35, n. 2, pp.. 239-257. Agnihothri, S., Narasimhan, S., Pirkul, H., 1990. An Assignment Problem With Queueing Time Cost, Naval Research Logistics, Vol. 37, No.2. Banks J., 1998. Principles ofSimulation, Handbook of Simulation. John Wiley & Sons, Inc.. Blumberg, D.. F., 1991. Managing Service as a Strategic Profit Center. McGraw-Hill. Box, G. E. P., Hunter, J. S., 1978. Statistics for Experimenters.. John Wiley & Sons, Inc. Chung, C. A, 2003. Simulation Modeling Handbook. CRC Press.. De Felice, F., 2007. Applied Simulation and Modelling. Acta Press. \ Guttman, t, Wilks, S. S., Hunter, J. S., 1971. Introductory Engineering Statistics. John Wiley & Sons, Inc. Kelton, W. D., Sadowski, R. P., 2003. Simulation with Arena.. McGraw-Hill. Law, A. M., Kelton, W. D., 2000. Simulation Modeling and Analysis.. Mc Graw-HilL Montgomery, D.. c., Runger, G. c., 2002.. Applied Statistics and Probability for Engineers. John Wiley & Sons, Inc. Mood, A. M., Graybill, F. A., Boes, D. C., 1974. Introduction to the theory of statistics. McGrawHilL Perros, H., 2007 Computer Simulation Techniques. the definitive introduction. Computer Science Department, NC State University, Raleigh, NC Pidd, M, 1992. Computer Simulation in Management Science. John Wiley & Sons, Inc. Ross, S. M, 1990. A course in Simulation. Macmillan Rotondi, A., Pedroni, P., Pievatolo, A., 2005. Probabilita, Statistica e Simulazione. Springer Verlag Italia. Upton, 0. M, 1994. The management of manufacturing flexibility, California Management Review, voL 36, pp. 72-89, Winter. Visintin, F., 2007. Designing after sales service strategy, operations and information technology. Thesis (PhD).. University of Florence..
Filippo Visintin was born in Prato on 1979.. In 2003 he graduated with honors in Management and Production Engineering at University of Bologna.. In 2004 he was admitted to attend the doctoral program in Industrial Engineering of the Department of Energetics "Sergio Stecco", Florence University.. In 2006 he was visiting research scholar at the School of Management of the State University of New York at Binghamton.. In 2007 he took a Ph.D in Industrial Engineering defending a thesis titled "Managing after sales services: strategy, operations and information technologies".. From December 2007 he is researcher and assistant professor (SSD ING-IND/35) at Florence University, Faculty of Engineering.. His current research interests are in the service operation management area. He's fellow of AHG, POMs and EurOMA. Alessandro Sistemi was born in Pisa in 1983. He took his Laurea Degree with honors in Management and Production Engineering at Florence University in December 2007 discussing the thesis "Field service delivery optimization for industrial and professional systems". Since February 2008 he is research fellow at the Department of Energetics "Sergio Stecco", Florence University.
122
"I
