Content uploaded by Mike Steglich

Author content

All content in this area was uploaded by Mike Steglich on Dec 16, 2021

Content may be subject to copyright.

LogisticsLab: An academic software for decision-making in logistics

Mike Steglich

University of Applied Sciences Wildau

Hochschulring 1

15745 Wildau, Germany

mike.steglich@th-wildau.de

KEYWORDS

logistical decision-making, transportation problems, net-

work flow problems, traveling salesman problems, Chi-

nese postman problems, vehicle routing planning prob-

lems, location problems, problem-based learning and

teaching

ABSTRACT

Logistical decision problems are a part of many courses

in the field of logistics, management and operations re-

search. It makes sense to illustrate these optimisation

problems using case studies, which can be reproduced by

students using suitable software. Often, solver add-ins in

spreadsheets programs or general optimisation software

are used, which on the one hand requires a high level of

knowledge in Operations Research and on the other hand

does not always allow an intuitive approach. This article

describes the academic software LogisticsLab with

which the distributors tie in with the idea of interactive

decision support systems to systematically combine the

experiences and intuitions of human decision-makers

with the possibilities of computer-assisted modelling and

optimisation of a wide range of logistical decisions.

INTRODUCTION

Students studying logistics management, supply chain

management, etc. are faced with decision problems in

transporting and storing of goods. Typical problems are

transportation problems, network flow problems, travel-

ing salesman problems, Chinese postman problems, ve-

hicle routing planning problems and location problems

(Rushton, Croucher, & Baker, 2017, p. 3ff). The students

need to understand the problem itself and to learn how to

formulate the problem mathematically, to solve it and to

interpret the solution. A problem-based learning ap-

proach seems to be a suitable learning and teaching

method for such tasks. Using real problems or realistic

case studies, students should improve their problem-solv-

ing skills, their ability to independently acquire theoreti-

cal knowledge and develop critical thinking (Guimarães

& Lima, 2021). In this context, suitable software is

needed to solve the discussed problems.

Simulation software is an approach often used in Oper-

ations Management and Logistics and Supply Chain

Management (LSCM) courses. A prominent example is

the beer distribution game originally invented by For-

rester in the 1960s (Dizikes, 2015) for which several soft-

ware packages exist (Beergame.org, 2021). It deals with

the dynamics within a supply chain with one homogene-

ous good including the bullwhip effect (Edali &

Yasarcan, 2016). Other examples of teaching and learn-

ing approaches using simulation software for LSCM are

Ștefan, Hauge, Hasse, & Ștefan (2019), Kanet & Stößlein

(2008), Angolia & Pagliari (2018) and the Fresh connec-

tion simulation (Inchainge, 2021). In most of these simu-

lations, students have to make and enter various decisions

within the supply chain, although there seems to be no

direct software support for this. Therefore, simulation ap-

proaches for these logistical decisions often also require

an optimisation software package.

A frequently used approach to solving logistical deci-

sions are spreadsheet programs with add-ins for optimi-

sation (Winkels, 2012). However, this approach is more

suitable for small problems due to the limitation of the

invoked solvers and its size restrictions. Furthermore, the

students must have a high level of knowledge in opera-

tions research techniques to formulate, solve and inter-

pret the problems where such programs do not allow an

interactive and intuitive way to do it (Mason, 2013).

Another option are optimisation environments, espe-

cially mathematical programming languages such as the

commercial AMPL (AMPL, 2021), GAMS (GAMS,

2021), etc. or open-source packages like GMPL (GMPL,

2021), CMPL (CMPL, 2021) or Pyomo (Pyomo, 2021).

There are also packages that implement routines for se-

lected logistical decisions with mathematical languages.

For example, Matlog implements routines for transport

problems, vehicle routing and network optimisation in

Matlab (Kay, 2016). Furthermore, the Google OR-Tools

are very interesting, which offer application program-

ming interfaces (APIs) for various programming lan-

guages for various optimisation problems including se-

lected logistical decision problems (Google, 2021b). As

with spreadsheet solver add-ins, students need a high

level of knowledge in operations research, which is not

always the case, especially in management courses.

(Grasas & Ramalhinho, 2016, p. 378).

Another possibility is the use of commercial logistics

software in university courses. It is obvious that the use

of such software gives students a realistic insight into the

management of supply chains. In addition, the acquired

skills have a positive impact on their future position in

the labour market (Campbell, Goentzel, & Savelsbergh,

2000). However, commercial logistics software is often

too expensive to purchase and maintain (Sweeney,

Campbell, & Mundy, 2010, p. 298f). Some manufactur-

ers offer cheaper or free versions of their software for ac-

ademic purposes, although these are often limited in their

functionalities or the size of the problems. For example,

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 42

anyLogistix offers a commercial „supply chain analytics

software to design, optimize and analyze a companies’

supply chain“ (anyLogistix, 2021a), for which a free ver-

sion is available for academic purposes that does not in-

clude some of the commercial functionalities and is also

limited in the number of customers and distribution cen-

tres and factories (anyLogistix, 2021b). If these limita-

tions are not too restrictive for a particular course, then

the use of free versions of commercial software makes

sense. Another problem with commercial software pack-

ages can be that they do not offer the full range of deci-

sion problems discussed in university courses. A good

example is Graphhopper (Graphhopper, 2021), which of-

fers an excellent API for vehicle routing. However, it

only supports vehicle routing. If other problems are to be

discussed, additional software is needed.

Academic logistical software, which offers a wide

range on supported logistical decision problems, is rather

rare. As an example, the Toolbox of the Technical Uni-

versity of Dresden can be mentioned (Toolbox, 2013). In

addition to problems in procurement logistics and pro-

duction logistics, shortest-path problems, network flow

problems, transport problems, vehicle routing problems

and location problems can also be solved. However, the

software is only available in German, seems rather orien-

tated on techniques and does not support all relevant as-

pects of logistical decision making. Another example is a

web-based decision support system for vehicle routing

problems developed by Grasas & Ramalhinho (Grasas &

Ramalhinho, 2016). Embedded in a problem-based learn-

ing approach, the tool is intended “to better understand

the solution method and to show students the need for

decision-making software in complex problems” (Grasas

& Ramalhinho, 2016, p. 389). The combination of the

easy-to-use web-based tool and the didactic approach is

very interesting. However, this tool can only be used for

vehicle routing problems.

The motivation for the development of LogisticsLab

was, considering all the possibilities of using software for

teaching logistical decision problems, to create an aca-

demic software that is easy and intuitive to use and offers

a wide range of supported logistical decision problems.

This paper describes LogisticsLab. In the first section,

an overview of the software including the main function-

alities and the file formats are given. The following sec-

tion is intended to illustrate how students can solve real-

istic case studies by using LogisticsLab. The described

case studies have been used in several courses at Bache-

lor level in German and international universities. The

paper ends with a summary.

LOGISTICSLAB

Overview

LogisticsLab is an academic software for decision sup-

port in logistics. This software can be classified as a prob-

lem-oriented decision support system that can be used to

model and solve transportation problems, network flow

problems, traveling salesman problems, Chinese post-

man problems, vehicle routing planning problems and lo-

cation problems in different variations.

LogisticsLab was originally developed in the 1990s un-

der the name EUS-Lehrsoftware by Dieter Feige for lo-

gistics courses at the Friedrich-Alexander University of

Erlangen-Nuremberg and fundamentally revised and ex-

panded by the author of this article.

LogisticsLab is provided under the LogisticsLab Aca-

demic License. It is available for academic use (such as

research and teaching or for reproducing case studies in

academic textbooks), free of charge and without war-

ranty. It can be downloaded at http://logisticslab.org and

is available for Microsoft Windows. No installation is re-

quired. The user only has to unzip the downloaded zip

file and the contained executables can be used immedi-

ately.

Main functionalities

After unpacking the downloaded distribution file, the Lo-

gisticsLab folder contains several binary files with spe-

cific names that provide the following functionalities.

TPP

LogisticsLab/TPP is a software that can be used to solve

classical transport problem (Hillier & Lieberman, 2015,

p. 319ff) and bottleneck transport problems (Garfinkel &

Rao, 1971). Minimisation and maximisation (or Mini-

Max and MaxiMin for bottleneck problems) can be se-

lected for the objective function sense. It is also possible

to invoke step-fixed costs for source-destination-rela-

tions. In addition, various constraints can be taken into

account (capacities, single-sourcing, supply or demand

surplus, etc.).

NWF

LogisticsLab/NWF is intended for solving network flow

problems, which can be modelled as min-cost flow prob-

lems (Ghiani, Laporte, & Musmanno, 2013). One of the

most important problems to be solved with this model is

the transhipment problem (Hillier & Lieberman, 2015,

p. 401).

TSP

LogisticsLab/TSP is used for solving traveling salesman

problems (Applegate, Bixby, Chvátal, & Cook, 2006)

and open traveling salesman problems (Steglich, Feige,

& Klaus, 2016, p. 312ff) for symmetrical and asymmet-

rical distances between the nodes to be approached. For

open traveling salesman problems, four different types

can be specified (fixed start and end node, fixed start

node and free end node, fixed end node and free start

node, free start and end node). Symmetric distances can

be calculated as airline distances using the coordinates of

the nodes as Euclidian distances, Manhattan distances

and great circle distances. It is also possible to define di-

rected or undirected edges with given distances. In addi-

tion, a distance matrix containing distances obtained by

APIs of geographic information systems like Google

Maps (Google, 2021a), Bing Maps (Microsoft, 2021) or

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 43

OpenStreetMap (OpenStreetMap, 2021b) can be im-

ported via LogisticsLab’s file format.

CPP

LogisticsLab/CPP is intended for solving Chinese post-

man problems in undirected, directed and mixed net-

works. Starting from a start node, each edge is to be trav-

ersed at least once and returned to the start node at the

end of the round trip. The edge-oriented round trip to be

found should minimise the total distance (or other ade-

quate evaluation) (Steglich, Feige, & Klaus, 2016, p.

320ff).

VRP

LogisticsLab/VRP is a software with which vehicle rout-

ing problems in different variants can be solved. The ca-

pacitated vehicle routing problem (VRP) aims to find a

set of routes at a minimal cost level, beginning and end-

ing the routes at a depot, so that the demands of all nodes

(customers, cities, etc.) are satisfied. Each node is visited

at least once and each vehicle has a limited capacity

(Williams, 2013, S. 198ff). With VRP, a user is able to

solve such problems in symmetric graphs using airline

distances (Euclidian, Manhattan and great circle dis-

tances), distances defined for edges or distances obtained

by geographic information systems (e.g. Google Maps,

Bing Maps, OpenStreetMap). Several additional con-

straints can be invoked into the decision (route duration,

route lengths, etc.)

WLP

Warehouse Location Problems are discrete location prob-

lems where both fixed location costs and variable

transport costs as well as the capacities of the potential

sources and the demands of the destinations are included

in the location decision. The objective is to minimise the

sum of the transportation costs and the fixed costs of

building and running warehouses by deciding which

warehouses are established and which customer is deliv-

ered to by which warehouse (Fernández & Landete,

2015, S. 47ff). LogisticsLab/WLP is designed as a soft-

ware for solving single-stage capacitated and uncapaci-

tated Warehouse Location Problems, always assuming

that a demand node can only be supplied from one loca-

tion at a time (single-sourcing).

CLP

LogisticsLab/CLP is a software to solve continuous (and

discrete) median and centre problems. Continuous loca-

tion problems determine the best possible greenfield lo-

cation. The point where a planned logistics node can be

optimally placed is to be found within a certain area on a

flat or spherical surface. In contrast, in discrete location

problems, only existing logistics nodes can act as loca-

tions (Steglich, Feige, & Klaus, 2016, p. 394ff). The fol-

lowing two approaches are distinguished: The median

problem aims at minimising the sum of the (weighted)

distances between the sources and the destinations and

consists in finding the optimal locations of the sources

and an optimal assignment of the destinations to the

sources. Depending on the number

𝑝

of locations sought,

these problems are also called

𝑝

-median problems

(Neema, Maniruzzaman, & Ohgai, 2011). The centre

problem aims at minimising the maximum (weighted)

distance over all relations between the locations of the

sources and the assigned destinations. Depending on the

number

𝑝

of locations to be determined, these problems

are also called

𝑝

-centre problems (Calik, Labbé, &

Yaman, 2015).

User Interface

The user interface of each part of LogisticsLab follows a

uniform structure, as shown in Figure 1 for Logis-

ticsLab/VRP as an example. In addition to the usual ele-

ments, the user interface has a network area on the left-

hand side of the application window with the graphical

representation of the network and a data area on the right-

hand side in which several areas can be selected via tabs

and filled with the necessary data. Generally, interactive

data entry and modification takes place in the data area,

but it is also possible to modify some of the data in the

network area interactively. The network and the data area

can be spread over the entire window.

File Formats

The data of the programs are stored in tab separated text

files with a specific format. To distinguish the data for

the different programs, the files are given the program

abbreviation as file extensions, i.e.: TPP, NWF, TSP,

CPP, VRP, WLP and CLP.

The following principles apply to data files:

• Each file begins with the keyword for the program for

identification. This keyword can be followed in the

same line by information on the date and time.

• The next line contains a comment.

• For some of the applications, a section for options fol-

lows. This begins with the keyword OPTIONS, after

which the corresponding options are listed in the fol-

lowing lines.

• The problem data is divided into data sections. Each

section begins with a line in which the corresponding

keyword (e.g., SOURCES, DESTINATIONS) for the

respective data must appear. In most cases, the size

specifications for the data section (e.g., number of

sources and destinations) follow in a subsequent line.

The following lines contain the corresponding data.

The records are defined line-wise as ASCII text. The

data entries are separated with tabulators.

• The end of file is expressed with the keyword EEE in

the last line.

The data files can be created via the menu function Save

Problem or Save Problem as. However, it is also possible

to edit and maintain an existing data file (as well solution

files) with a spreadsheet software (e.g., Microsoft Excel

or LibreOffice/Calc). Finally, the data must be saved as a

pure text file with data separation by tabs.

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 44

SELECTED CASE STUDIES

This section is intended to illustrate how students can

solve realistic case studies by using LogisticsLab. Both

case studies have been used in courses at Bachelor level

at several European universities, often using a problem-

based learning and teaching approach (Hung, 2011)

(Grasas & Ramalhinho, 2016).

In a first step, the students, organised in groups, have to

analyse a given realistic problem and work out the theo-

retical basis of the problem in order to formulate a suita-

ble mathematical model. The mathematical model helps

the students to understand and formulate the objective

and the constraints of the problem and what data is

needed to solve the case. In addition, students have to ob-

tain the problem data from various sources. Afterwards,

the problem, including the obtained data, has to be en-

tered into LogisticsLab and solved with it. The last but

not least task is the interpretation and presentation of the

solution found.

Street condition survey in Manhattan

In this case study, the students were asked to find the

shortest route to examine the condition of part of the

streets in Manhattan, NY. As shown in Figure 2, it is the

area bounded on the northwest by the 10th Avenue/West

57th Street intersection and on the southeast by the 6th

Avenue/East 42nd Street intersection. All streets in this

area are to be driven at least once.

This problem can be solved as a Chinese postman prob-

lem which can be understood as an edge-oriented round-

trip problem. Consider a network

𝐺 = (𝑁, 𝐸),

where

𝑁

is

a set of nodes and

𝐸

is a set of edges joining pairs of

nodes. Each edge has to be used at least once, whereby

the tour starts and ends at a specific node. The total dis-

tance of this route is to be minimised (Ahuja, Magnanti,

& Orlin, 2013, S. 740) (Steglich, Feige, & Klaus, 2016).

Figure 2: Area in Manhattan (OpenstreetMap, 2021a)

Figure 1: User interface of LogisticsLab/VRP

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 45

In this case, the nodes are defined by the street intersec-

tions and the edges by the streets, considering the direc-

tions of the streets. The coordinates of the nodes and the

distances of the edges can be obtained by using the APIs

of geographical information systems like Google Maps,

Bing Maps or OpenStreetMap.

After starting the application, the user interface shown

in Figure 3 appears, which, like all other components of

LogisticsLab, consists of a network area and a data area.

Figure 3: User interface of LogisticsLab/CPP

The left side of the CPP window is used to visualise the

network graph depending on the data entered. If the net-

work graph contains connections between the nodes,

these are shown as edges. Directed edges are represented

by arrows. Undirected edges do not have arrowheads.

The data area on the right-hand side of the CPP user in-

terface contains four worksheets for entering data and

outputting optimisation results (Problem, Nodes, Edges,

Solution).

The first step using LogisticsLab/CPP is to generate a

new problem. To do this, the menu item File → New

Problem or the New Problem button in the toolbar is se-

lected. In addition to a comment, the number of nodes (in

this case 91 nodes) and the maximum distance between

the nodes (Max. distance) must be entered (Figure 4).

The latter is used when generating randomly based coor-

dinates (Coordinates → Randomly). The generated coor-

dinates can be edited subsequently.

Figure 4: Creating a problem in LogisticsLab/CPP

After generating the problem, the data can be entered in

the data area. The nodes can be edited in the Nodes tab

(Figure 5).

A user can either enter the number of nodes or add each

node individually to the list with the Add button.

Figure 5: Nodes tab in LogisticsLab/CPP

Each node is specified by a selection (Active - selection

for a calculation), a unique node ID (ID), a node name

(Name) and an X and a Y coordinate. Geographical coor-

dinates can also be used instead of simple X and Y coor-

dinates. In this case, the longitude must be entered as the

X-value and the latitude as the Y-value.

An easier way is to open the generated CPP file in a

spreadsheet software, edit the names and copy and paste

the coordinates found with Google Maps, Bing Maps or

OpenStreetMap into the file (Figure 6).

Figure 6: Editing a CPP file in Microsoft Excel

The edges have to be defined in the Edges tab (Figure 7).

To do this, the node from which the edges start must be

specified in the From node field. The edges are defined

in the To nodes list by entering the name, length and type

of the edge. The type can be U for an undirected edge or

D for a directed edge.

Figure 7: Nodes tab in LogisticsLab/CPP

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 46

For large problems, it is also recommended for the edges

to open the CPP file in a spreadsheet program and copy

and paste the distances into it.

After entering all the data, the problem can be solved

by selecting either the Optimisation → Start Optimisa-

tion menu or the Optimise button in the toolbar. To solve

a Chinese postman problem, the following two steps are

carried out by LogisticsLab/CPP automatically (Steglich,

Feige, & Klaus, 2016, p. 320 ff):

1. Cost- or distance-minimal extension of the net-

work into a Euler network,

2. Determination of the Euler tour.

The cost- or distance-minimal extension of the existing

network is solved on the basis of an integer linear opti-

misation model (Winkels, 2012, p. 590f) (Steglich, Feige,

& Klaus, 2016, p. 325f), the results of which are used by

a recursive algorithm to determine an Euler tour

(Mattfeld & Vahrenkamp, 2014, p. 223f) (Steglich,

Feige, & Klaus, 2016, p. 322ff) and thus to solve the Chi-

nese postman problem.

After solving the problem, the graphical representation

of the solution appears in the Network area (Figure 9) and

the details in the Solution tab (Figure 8). In the Network

area (Figure 9), the thickness of the edges indicates

whether they have to be traversed several times.

Figure 8: Solution tab in LogisticsLab/CPP

Figure 9: Network area in LogisticsLab/CPP

The best solution found has a length of 31.886 km (Figure

8). Both the graphical solution and the list of the entire

route can be printed. It is also possible to save the solu-

tion in a solution file which can easily be imported into a

spreadsheet program.

Planning three disaster relief centres in the state of

Brandenburg, Germany

In this case study, students were asked to find the optimal

locations of three disaster relief centres for the state of

Brandenburg. The task of the centres is to send technol-

ogy and emergency forces to the towns and municipali-

ties as fast as possible in the event of a disaster. The max-

imum distance of the centres to the potential places of

operation is to be minimised. Since cities and municipal-

ities with many inhabitants are to be given priority in the

event of a disaster, the distances are to be weighted ac-

cording to the number of inhabitants. The coordinates

and the number of inhabitants of the cities and munici-

palities can be taken from data of the German Federal

Statistical Office (Statistisches Bundesamt, 2020). The

locations of the disaster relief centres have to be planned

on greenfield sites. The problem is a continuous

𝑝

-centre

problem without considering capacities and with de-

mand-weighted distances, which can be formulated as

follows (Drezner, 1984) (Drezner, 2011):

Indices and sets

𝑖 ∈ 𝑆

sources

𝑗 ∈ 𝐷

destinations

Parameters

𝑏!

demand of destination

𝑗

𝑥!,𝑦!

coordinates of destination

𝑗

Function

𝑑

(

∙

) distance function

Variables

𝑅

maximum demand weighted distance

𝑥"!

assignment variable whether source

𝑖

serves

destination

𝑗

𝑢!,𝑣!

coordinates of source

𝑖

According to the expressions (1) and (2), the maximum

demand-weighted distance

6𝑅

between the sources and the

destinations is to be minimised. These distances are to be

determined for all combinations of sources and destina-

tions with a suitable distance function

𝑑

(

∙

). These dis-

tances are only relevant if the corresponding assignment

variables

𝑥"!

are equal to the value one, i.e., destination

𝑗

is assigned to source

𝑖

. Expression (3) ensures that a des-

tination is assigned to exactly one source (Steglich,

Feige, & Klaus, 2016, p. 411).

𝑅 → 𝑚𝑖𝑛!

𝑠.𝑡.

(1)

𝑏!∙𝑥"! ∙𝑑>(𝑢",𝑣"),(𝑥",𝑦")?≤ 𝑅66; 𝑖 ∈ 𝑆,𝑗 ∈ 𝐷

(2)

B 𝑥"!

!∈# = 1;𝑗 ∈ 𝐷

(3)

𝑥"! ∈{0,1}66666;𝑖 ∈ 𝑆,𝑗 ∈ 𝐷

(4)

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 47

The problem can be solved with LogisticsLab/CLP,

which, like all other parts of LogisticsLab, has a user in-

terface consisting of a network area and a data area

(Figure 10).

Figure 10: User interface of LogisticsLab/CLP

The first step is to create the problem file and to collect

the problem data. Since the state of Brandenburg consists

of 417 towns and municipalities, a problem with 417 des-

tinations is to be created via the menu item File → New

Problem or the New Problem button in the toolbar, as

shown in Figure 11.

Figure 11: Creating a problem in LogisticsLab/CLP

Afterwards, the data of the generated problem can be en-

tered in the data area (Figure 12). For all destinations, the

names, the coordinates and the demands have to be en-

tered, whereby for this problem the demands are defined

by the number of inhabitants of the corresponding town

or municipality. Since geographical coordinates are used,

the longitude must be entered as the X-value and the lat-

itude as the Y-value. The names, the coordinates and the

inhabitants of the towns and municipalities can be taken

from an Excel file of the German Federal Statistical Of-

fice (Statistisches Bundesamt, 2020). However, it is rec-

ommended to copy the mentioned data from this Excel

file and paste it into the CLP file opened in a spreadsheet

program (Figure 13).

Figure 12: Destination tab in LogisticsLab/CLP

Figure 13: CLP problem file in Excel

The Sources tab (Figure 14) contains the details of the

locations whose positions are to be determined by the op-

timisation. By entering the number of sources, the size of

the input sheet is adjusted automatically. The names and

the capacities (if needed) can be edited in this tab.

Figure 14: Sources tab in LogisticsLab/CLP

Once all the data has been entered, the problem can be

solved. To do this, the problem type centre of gravity and

the distance function to be used (Distance → Great Cir-

cle) must be selected in the Problem tab (Figure 15).

Since the number of inhabitants of the destinations influ-

ence the priority in the event of a disaster, it must be spec-

ified that the distances are weighted with the demands of

the destinations (checkbox Demand weighted).

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 48

Figure 15: Problem tab in LogisticsLab/CLP

Optimisation is started by selecting either the Optimisa-

tion → Start Optimisation menu or the Optimise button

in the toolbar. LogisticsLab uses state of the art heuristics

to solve continuous location problems. As is usual for

heuristics, the solutions may vary from run to run.

The solution of the problem is displayed in both the net-

work area and data area. In the network area (Figure 16),

the locations found for the three disaster relief centres and

additionally the assignments of the towns and municipal-

ities as potential places of a disaster event are shown. The

exact locations of the disaster relief centres can be found

in the Sources tab (Figure 17) in the columns X-Pos and

Y-Pos. Besides to the number of assigned destinations

(Destinations), the nearest destinations to the greenfield

locations of the sources are shown.

Figure 16: Network area in LogisticsLab/CLP

Figure 17: Solution in the Sources tab

The objective was to minimise the maximum of the de-

mand-weighted distances between the locations to be

found and the destinations. As shown in Figure 18, the

objective function value equals 922,193.6 person-kilo-

metres (similar to passenger-distance). This value is the

minimum Maximum Costs of all relations between the

new locations of the disaster relief centres and the as-

signed destinations. It results from the multiplication of

the number of the inhabitants (Demand) and the dis-

tances. The cost factor is set to the value one and can be

ignored. This indicator is reasonable because it is as-

sumed that an increasing number of inhabitants also leads

to the increase of the amount of technology and emer-

gency forces to be sent to the location of a disaster event.

Figure 18: Key indicators in the Problem tab

The assignments of the destinations to the newly found

locations of the sources can be found in the Destinations

tab (Figure 19). The relation with the maximum inhabit-

ant weighted distances is between the destination Bernau

(Nr. 6) and the first source which is located nearby Neu-

hardenberg (Figure 19 and Figure 17).

Figure 19: Solution in the Destination tab

The objective of this case study can be changed if it is

assumed that the amount of technology and emergency

forces is the same for all events and therefore independ-

ent of the number of inhabitants. In this case, only the

maximum over all distances between the locations and

the potential disaster event places is to be minimised.

This can be achieved simply by not activating the option

Demand weighted in the Problem tab (Figure 15).

After running the optimisation again, a new solution

can be found where the locations of the three disaster re-

lief centres are not attracted to destinations with high

populations. This can be seen in the graphical solution in

the network area (Figure 20) and on the newly deter-

mined coordinates of the sources in the Sources tab

(Figure 21).

Since a heuristic is used, the solutions may be different

for different optimisation runs. The minimal maximum

distance found equals 73.2 kilometres (Figure 22). That

means that the teams have to travel a maximum of 73.2

kilometres over all relations between the newly deter-

mined locations of the disaster relief centres and the po-

tential places of a disaster event. This is less than the min-

imal maximum distance of the original problem which is

equal to 128.5 kilometres (Figure 18 → Distances →

Maximum).

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 49

Figure 20: Graphical solution of the unweighted case in

the network area in LogisticsLab/CLP

Figure 21: New locations of the unweighted case in the

Sources tab in LogisticsLab/CLP

Figure 22: Key indicators in the Problem tab for the un-

weighted case

However, the locations found with a continuous p-centre

problem cannot be implemented in reality in every case.

So, it could be possible that the three locations nearby

Havelaue, Schoenwalde and Hohenfinow (Figure 21)

cannot be realised for many reasons. If only existing

nodes in a network are available as locations for the

sources, a discrete

𝑝

-centre problem has to be solved.

In difference to a continuous problem, the coordinates

of the locations are not variables, because only prede-

fined nodes of the set

𝑆6

of potential sources can be used

as locations. Additionally, all distances between the po-

tential sources

𝑖 ∈ 𝑆

and the destinations

𝑗 ∈ 𝐷

are

known. Regarding the mathematical model, please refer

to the literature (Calik, Labbé, & Yaman, 2015, p. 83f).

LogisticsLab/CLP is mainly intended to solve continu-

ous

𝑝

-median and

𝑝

-centre problems. However, it offers

also to solve both as discrete location problems using a

heuristic. To do so, a user only has to choose the option

Discrete in the Problem tab (Figure 15) and to start the

optimisation again. After selecting this option, a new so-

lution can be found for the given unweighted case. It

should also be mentioned for the discrete problem that

the solutions of different optimisation runs may differ

due to the non-deterministic nature of the underlying heu-

ristic. The discrete solution is shown as graphical solu-

tion (Figure 23) in the network area and the coordinates

are shown in the Sources tab (Figure 24).

Figure 23: Graphical solution of the discrete case in the

network area in LogisticsLab/CLP

It is interesting to note that the new locations for the dis-

aster relief centres in Ziethen, Luebben and Havelaue are

partially unequal to Havelaue, Schoenwalde and Hohen-

finow, which are the closest destinations to the locations

found for the continuous problem. Moving the continu-

ous locations to the closest destinations may lead to

suboptimal solutions.

Figure 24: New locations of the discrete case in the

Sources tab in LogisticsLab/CLP

The minimal Maximum Distance found for the discrete

problems is equal to 75.6 kilometres (Figure 25). This is

only 2.4 kilometres more for this more realistic solution

compared to the solution of the continuous problem.

Figure 25: Key indicators in the Problem tab for the un-

weighted case

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 50

SUMMARY

Students who are confronted with logistical decision

problems in several courses need appropriate software to

solve and interpret realistic case studies. The frequently

used solver add-ins in spreadsheet programs or general

optimisation software require a high level of knowledge

in operations research and do not always allow intuitive

access. Commercial logistics software and the rather rare

academic software often offer only a smaller selection of

supported logistical decisions.

This article described the academic software Logis-

ticsLab, which supports a wide range of logistical deci-

sions. The idea of the software is the systematic combi-

nation the experiences and intuitions of human decision-

makers with the opportunities of computer-assisted mod-

elling and optimisation in the form of an interactive deci-

sion support system. Two realistic case studies, used in

courses at several European universities, demonstrate

how LogisticsLab can be used to gain a better under-

standing of logistical decision-making in an interactive

and intuitive way.

ACKNOWLEDGEMENTS

The author would like to thank Dieter Feige for the many

years of cooperation and for the opportunity to continue

the LogisticsLab project. Unfortunately, Dieter Feige

passed away in 2019, so the author has lost not only a

colleague but also a friend. This article is dedicated to

him.

REFERENCES

Ștefan, I. A., Hauge, J. B., Hasse, F., & Ștefan, A. (2019). Using

Serious Games and Simulations for Teaching Co-

Operative Decision-making. 7th International

Conference on Information Technology and Quantitative

Management (ITQM 2019) (pp. 745-753). Procedia

Computer Science.

Ahuja, R. K., Magnanti, T. L., & Orlin, J. B. (2013). Network

Flows: Theory, Algorithms, and Applications. Harlow:

Pearson New International Edition.

AMPL (2021). Retrieved February 2021, https://ampl.com.

Angolia, M. G., & Pagliari, L. R. (2018). Experiential Learning

for Logistics and Supply Chain Management Using an

SAP ERP Software Simulation. Decision Sciences

Journal of Innovative Education, 16(2), 104-125.

anyLogistix (2021a). anyLogistix - Features overview.

Retrieved April 2021, https://www.anylogistix.com/alx-

features/.

anyLogistix (2021b). anyLogistix - Personal Learning Edition.

Retrieved April 2021, https://www.anylogistix.com/

personal-learning-edition/.

Applegate, D., Bixby, R., Chvátal, V., & Cook, W. (2006). The

traveling salesman problem: A Computational Study.

Princeton: Princeton University Press.

Beergame.org (2021). Versions. Retrieved April 2021,

https://beergame.org/the-game/versions/.

Calik, H., Labbé, M., & Yaman, H. (2015). p-Center Problems.

In G. Laporte, S. Nickel, & F. Saldanha da Gama (Eds.),

Location Science (pp. 79-92). Cham et al.: Springer.

Campbell, A., Goentzel, J., & Savelsbergh, M. (2000).

Experiences with the use of supply chain management

software in education. Production and Operations

Management, 9(1), 66-80.

CMPL (2021). Retrieved February 2021, http://coliop.org.

Dizikes, P. (2015). The Many Careers of Jay Forrester.

Retrieved April 2021, https://www.technologyreview

.com/2015/06/23/167538/the-many-careers-of-jay-

forrester/.

Drezner, Z. (1984). The p-Centre Problem-Heuristic and

Optimal Algorithms. The Journal of the Operational

Research Society, 35, 741–748.

Drezner, Z. (2011). Continuous Center Problems. In H. Eiselt,

& V. Marianov (Eds.), Foundations of Location Analysis

(pp. 63–78). Springer US.

Edali, M., & Yasarcan, H. (2016). Results of a beer game

experiment: Should a manager always behave according

to the book? 21(S1), 190-199.

Fernández, E., & Landete, M. (2015). Fixed-Charge Facility

Location Problems. In G. Laporte, S. Nickel, & F.

Saldanha da Gama (Eds.), Location Science (pp. 47-78).

Cham et al.: Springer.

GAMS (2021). Retrieved February 2021, https://www.gams

.com.

Garfinkel, R., & Rao, M. (1971). The bottleneck transportation

problem. Naval Research Logistics Quarterly (18), 465–

472.

Ghiani, G., Laporte, G., & Musmanno, R. (2013). Introduction

to Logistics Systems Management (2. ed.). Chichester:

Wiley.

GMPL (2021). GNU MathProg modeling language. Retrieved

February 2021, https://www.gnu.org/software/glpk/.

Google (2021a). Distance Matrix API. Retrieved February

2021, https://developers.google.com/maps/

documentation/distance-matrix/overview.

Google (2021b). Google OR-Tools. Retrieved September 2021,

https://developers.google.com/optimization.

Graphhopper (2021). Products. Retrieved February 2021,

https://www.graphhopper.com/products/.

Grasas, A., & Ramalhinho, H. (2016). Teaching distribution

planning: a problem-based learning approach. The

International Journal of Logistics Management, 27(2),

377-394.

Guimarães, L., & Lima, R. (2021). Changes in teaching and

learning practice in an undergraduate logistics and

transportation course using problem-based learning.

Journal of University Teaching & Learning Practice,

18(3), 11-27.

Hillier, F., & Lieberman, G. (2015). Introduction to Operations

Research (10. ed.). New York et al.: McGraw-Hill.

Hung, W. (2011). Theory to reality: a few issues in

implementing problem-based learning. Education Tech

Research Dev(59), 529-552.

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 51

Inchainge (2021). Discover The Fresh Connection. Retrieved

April 2021, https://inchainge.com/business-games/tfc/

Kanet, J. J., & Stößlein, M. (2008). Using a Supply Chain Game

to Effect Problem-Based Learning in an Undergraduate

Operations Management Program. Decision Sciences

Journal of Innovative Education, 6(2), 287-295.

Kay, M. G. (2016). Matlog: Logistics Engineering using

Matlab. Journal of Engineering Sciences and Design,

4(1), 15-20.

Mason, A. J. (2013). SolverStudio: A New Tool for Better

Optimisation and Simulation Modelling in Excel.

INFORMS Transactions on Education, 45-52.

Mattfeld, D., & Vahrenkamp, R. (2014). Logistiknetzwerke:

Modelle für Standortwahl und Tourenplanung (2. ed.).

Wiesbaden: Springer Gabler.

Microsoft (2021). Distance Matrix API. Retrieved February

2021, https://www.microsoft.com/en-us/maps/distance-

matrix.

Neema, M., Maniruzzaman, K., & Ohgai, A. (2011). New

Genetic Algorithms Based Approaches to Continuous p-

Median Problem. Networks and Spatial Economics(11),

83–99.

OpenStreetMap. (2021a). Retrieved January 2021,

https://www.openstreetmap.org/relation/8398124#map=

16/40.7632/-73.9837.

OpenStreetMap (2021b). OSRM API Documentation.

Retrieved February 2021, http://project-

osrm.org/docs/v5.5.1/api.

Pyomo (2021). Retrieved Februray 2021, http://www.pyomo

.org.

Rushton, A., Croucher, P., & Baker, P. (2017). The Handbook

of Logistics & Distribution Management (6. ed.). London

et al.: KoganPage.

Statistisches Bundesamt (2020). Alle politisch selbständigen

Gemeinden mit ausgewählten Merkmalen am 31.12.2020

(4. Quartal). Retrieved February 2021,

https://www.destatis.de/DE/Themen/Laender-Regionen/

Regionales/Gemeindeverzeichnis/Administrativ/Archiv/

GVAuszugQ/AuszugGV4QAktuell.html

Steglich, M., Feige, D., & Klaus, P. (2016). Logistik-

Entscheidungen: Modellbasierte Entscheidungsunter-

stützung in der Logistik mit LogisticsLab (2nd completely

revised and expanded edition ed.). Berlin et al.:

De Gruyter Studium.

Sweeney, D. C., Campbell, J. F., & Mundy, R. (2010). Teaching

Supply Chain and Logistics Management Through

Commercial Software. The International Journal of

Logistics Management, 21(2), 293-308.

Toolbox (2013). Retrieved February 2021, https://tu-

dresden.de/bu/wirtschaft/bwl/log/studium/materialien/

toolbox.

Williams, H. (2013). Model Building in Mathematical

Programming (5. ed.). Chichester: Wiley.

Winkels, H.-M. (2012). Modellbasiertes Logistikmanagement

mit Excel: Lösungen von Problemen in der Logistik unter

Verwendung der Tabellenkalkulation. Hamburg: DVV.

Anwendungen und Konzepte der Wirtschaftsinformatik

(ISSN: 2296-4592) http://akwi.hswlu.ch Nr. 14 (2021) Seite 52