Content uploaded by Alfred Teischinger
Author content
All content in this area was uploaded by Alfred Teischinger on Sep 27, 2022
Content may be subject to copyright.
MASS CUSTOMIZATION IN THE WOOD-
WORKING INDUSTRY: SIMULATION
BASED RESEARCH OF NEW
PRODUCTION CONCEPTS
Hans Häuslmayer*, Manfred Gronalt* and Alfred Teischinger**
Abstract
We propose new ideas for applying mass customization concepts in the production of wooden
floorings, especially the parquet industry. Due to market requirements this industry is obliged to
observe special quality standards during production. Mass Customizing parquet flooring could
open new possibilities for the industry. To offer customers the possibility of individual composed
floors, several types of basis modules instead of currently produced floor boards are considered
for manufacturing. Our research explores these new production concepts in a case study
approach with an Austrian manufacturer of parquet flooring, using simulation as a method to test
new production environments and different product/order mixes. We analyze several possible
product-variety/customer-order mixes and their distinctive production outcomes when producing
different basis modules of floorboards. These results are then compared to the status quo of
parquet manufacturing at the case study company. We show how reconfigured production
processes affect storage levels of semi finished products, service rates and thus customer
satisfaction.
Keywords
mass customization, decoupling point, parquet industry, process redesign, product configurator
Häuslmayer, H./Gronalt, M./Teischinger, A.: „Mass Customization in the Wood-Working Industry: Simulation based Research of New Production
Concepts“. In: Blecker, T./Friedrich, G./Hvam, L./Edwards, K. (Hrsg.), Customer Interaction and Customer Integration Series on Business Informatics
and Application Systems, Berlin: GITO mbH – Verlag für Industrielle Informationstechnik und Organisation 2006 (Customer Interaction and Customer
Integration Series on BusinessInformatics and Application Systems 2), S. 183–198
1 Introduction
We propose new ideas for applying mass customization concepts in the production of wooden
floorings, especially the parquet industry. Due to market requirements this industry is obliged to
observe special quality standards during production. Wood with certain natural characteristics
like colorings or knots is not accepted in higher quality classes and thus has to be sorted out. As a
matter of fact the effective utilization of the natural resource wood and the price of the final
product are negatively influenced as downgrading of material affects the generated value-added.
The customer has only a choice between a narrow range of patterns or quality. Individual
arrangements or a fully customized floor is today, with the current production processes, nearly
impossible to supply. We assume that in the parquet industry, unlike other industries, the product
the customers buys, only consists of prefabricated parts, namely the floor boards, as the actual
product, a customer wants to buy, is the pattern or the impression and other characteristics of his
individual floor.
Until now, examples and applications of mass customization and postponement strategies as well
as customized product configuration in the woodworking industry can mainly be found in
furniture production. Nevertheless, the ability to produce customized furniture with industrial
processes will be the key aspect of survival of furniture producing companies in the future
according to industry consultants. A Delphi-study of the wood-working-industry resulted in the
prognosis that by 2010 over 25% of furniture will be produced in a lot size of one piece on
industrial scale. Working solutions for customized furniture design and production, as well as
individualized, configurator-based kitchen production are no new challenges anymore for the
wood-working industry.
Innovations in production processes of parquet floorings in the last 30 years have mainly taken
place on the technical machinery equipment side. State-of-the-art descriptions of production
processes from the 1970ies do not differ very much from analyses in the 1990ies or our
observations of production processes in a case study in 2005. Individualization concepts in the
flooring industry are starting to become popular, especially concerning artificial floorings like
laminate or easy to re-design natural materials like cork. Also, product presentation or
configuration software, through which customers can dimension the rooms they want to furnish
and experiment with different colors and wood types, is available from manufacturers. Some
manufacturer’s websites even offer online testing possibilities of the noise behavior of floors.
Mass Customizing parquet flooring could open new possibilities for the industry. To offer
customers the possibility of individual composed floors, several types of basis modules instead
of currently produced floor boards are considered for manufacturing. In the production process
several possibilities to modularize floorboards arise. We present and evaluate these possibilities
concerning the implications for the current production process. The remainder of the paper is
structured as follows. In chapter 2 we give a review of related works. Chapter 3 describes the
current production concept in the parquet industry as explored in a case study approach at an
Austrian manufacturer of parquet flooring while chapter 4 explores the new production concepts
proposed. We also present a product configurator for customized floorings already in use and
possible extensions for our new concept. Chapter 5 presents the research approach with process
simulation as the method to test these new production environments and different product/order
mixes. In a simulation study, set up with real-life production data from over 13 months, we
analyze several possible product-variety/customer-order mixes and their distinctive production
outcomes when producing different basis modules of floorboards. In chapter 6 our results are
then compared to the status quo of parquet manufacturing at the case study company. We show
how reconfigured production processes affect storage levels of semi finished products, service
rates and thus customer satisfaction. The paper concludes with the identification of further
research issues.
2 Related Works
The term Mass Customization has been first coined in the late 1980ies and has become subject to
research concerning operations management since Pine (1993). McCarthy (2004) defines Mass
Customization as the capability of companies to produce a relatively high volume of product
options for a relatively large market demanding customized products without tradeoffs in cost,
delivery and quality. Swaminathan (2001) categorized modularization concepts of products and
processes as strategies to achieve mass customization. One of these strategies is postponement, a
strategy that postpones order fulfilment processes until customer demand is known. The
postponement/speculation matrix of Pagh und Cooper (1998) offers a taxinomy of
postponement-strategies. According to the level of planning and/or speculation in the production
process and in the distribution process they distinguish four different postponement strategies. A
comprehensive literature review of postponement as a production process strategy and its
different classifications can be found in van Hoek (2001). The point in the production process
when customer demand becomes known, is known as the Customer Order Decoupling Point
(CODP) or production decoupling point (Meyr 2003). It has been subject to research
synonymously as order-penetration-points (Olhager 2003), freeze-points, product fan-out point,
push-pull point (Schmidt 1999) or level of prefabrication (Piller 1998). Work on the aspects of
mass customization concerning interaction with the customer, especially on product
configuration programs and Web-based manufacturing has been done for example by Blecker
(2003).
Implications of postponement on the supply chain concerning information and material flows
and the connection between product types and the decoupling point have been researched for
example by Yang and Burns (2003). Postponement strategies can also have effects on customer
satisfaction. Reiner (2005) shows for an example of a supplier of the telecommunication
industry, that postponement strategies such as labelling imply a substantial potential for
customer-oriented improvements. Postponement strategies are levers to improve the performance
of business processes. Jammernegg and Reiner (2004) show in a case study of a two-tier supplier
for the telecommunication and automotive industry through process simulation that
reorganisations from Make-To-Stock (MTS) to Assemble-To-Order (ATO) production processes
combined with coordinated inventory and capacity management result in improved delivery
performance at lower total costs.
Examples and applications of mass customization and postponement strategies as well as
customized product configuration in the woodworking industry can mainly be found in furniture
production (Sigg & Jonas 2003). According to a Delphi-study of the wood-working-industry
undertook by the Fraunhofer Institute for Manufacturing Engineering and Automation over 25%
of furniture shall be produced in lot size 1 on industrial scale (ProWood 2004). Working
solutions for customized furniture design, production and ordering processes, like the famous
IKEA kitchen configurator are no new challenges for the wood-working industry anymore.
Concerning production processes of parquet flooring in the last 30 years innovations have mainly
taken place concerning the technical machinery equipment side. State-of-the-art descriptions of
production processes from the 1970ies (Kisseloff 1974) do not differ very much from analyses in
the 1990ies (Hamberger 1995) or our observations of production processes in a case study in
2005.
3 Current Production Process of Parquet
Analyzing the relevant business and production processes we identify the parameters currently
faced in the manufacturing of parquet flooring. The case study company produces floorboards in
5 different qualities, 19 different wood sorts and allows for several different finishing
possibilities, like lacquer or oiling, which the customers can choose freely. Production of parquet
is done in two stages: First, the (visible) upper layer is produced in a make-to-stock environment,
followed by a buffer in which conditioning processes of the floorboards according to the wood
sort, take place. This buffer is also functioning as CODP. In the second step of the production a
make-to-order environment can be stated. Floorboards are retrieved from the conditioning
chamber, and then finished according to customer orders. Customers can order their desired
flooring only in the distinctive qualities. Due to its character as a natural resource wood is subject
to several anomalies and different quality patterns. The company is faced with uncertainties and
cannot ever be sure to produce the quality desired by the customer with the raw material input.
Thus a lot of work-in-progress of different product variants has to be kept in stock in the
conditioning chamber, which is one of the main bottlenecks of the production process.
A multi layer 3-strip parquet production processes in general can be divided into four sub-
sections (Kruse et al. 2003):
• Manufacturing of the upper layer (the one visible to the customer),
• Manufacturing of the middle layer or support layer and undercoating (a process not
considered and described here, as this is a standardized semi finished product),
• Press sizing of upper layer and support layer to a floorboard and finishing operations of the
surface of the floorboard with lacquer or oil, and
• Finishing Processes like milling of tongue and groove, quality end control and packaging.
3.1 Manufacturing of the Upper Layer
The production processes of the visible, upper layer of the flooring board are essential to the
whole production flow and crucial to a company’s success, as during this stage of production the
value added to the end product is defined through quality sorting.
The raw material is cut to a predefined length and planed to a predefined thickness. In a second
step the resulting boards are each cut into several lamellae of equal length and thickness. After
the cutting process these lamellae are very thin, so that from 1cbm of solid wood up to 130sqm
of wooden floor can be produced. These lamellae are afterwards subject to a sorting process,
which can be automated by means of surface scanning or manual. This sorting process is crucial
to the whole production process, as through the examination of the knots and discolorations, the
sorting class and hence quality and price class is defined. Downgrading a lamella from A to B
means to get out a lower value added from one block of wood, then from maybe the next block.
Lamellae are then processed further in only one and the same quality. In this part of the
production process the single lamellae are glued together through press sizing to form the Upper
Layer of a floorboard. After this process these layers have to be stored in a conditioning chamber
for 4-7 days, until they can be further processed.
As it is important for our further considerations, the process of press-sizing the upper layer shall
be examined more closely. The pressing machine consists of conveyor belts, each equipped with
feeders in which the stacks of lamellae, all of the same quality, are filled and refilled by a worker.
Physical impact (air or a mechanic device) shoots out full and cut-in-half lamellae from the
feeders. The lamellae are then glued and pressed together on the front and side ends. Figure 1
gives an impression of the machine and its workflow. For generating one layer of parquet in our
reference application 17 complete and two half-size lamellae are used.
Figure 1: Workflow of Upper Layer Pressing Machine
3.2 Decoupling Point and Finishing Processes
After the storing of the upper layers in the conditioning chamber for a certain time, they are
retrieved according to customer orders and as a next step, press-sized with the middle layer. The
retrieval process marks the CODP of the process. It is here where each floorboard is assigned to
a specific customer requirement. One of the main difficulties in developing mass customization
concepts for wooden-flooring productions lies within the discrepancy of the point in time when a
product becomes unique and distinguishable from other products and the point where an order is
assigned to it. The former is called the Product Differentiation Point (PPD), the latter is the
Customer Order Decoupling Point (CODP). Figure 2 shows the current production processes of
parquet and the two points mentioned, which differ from another. New production concepts have
to take this fact into account as it represents one of the major difficulties in introducing mass
customization concepts in the parquet flooring industry.
After the press-sizing processes the floorboards have to be conditioned again for 6 to 7 days in
order to stabilize the amount of water in the wood and glue to a low level. Afterwards the boards
are sanded and either lacquered or oiled. In the next steps tongue and grooves are milled in the
board, they are packaged and labeled. Storing in the finished goods warehouse concludes the
production cycle from solid wood to 3-strip multi layer parquet.
Figure 2: Production Processes including PPD and CODP
4 New Production Concepts
As mass customization processes are still underdeveloped in the wood-working industry, most of
our research has been exploratory in nature. We developed ideas how a new production concept
of customized wooden flooring could be set up. These production concepts comprise several
sub-features which should ideally work together. Some of these features, like a product
configurator for parquet flooring, database solutions and RFID-logistics we assume as
technically possible and thus given. We will only describe shortly the concept of these IT-Tools.
The production parts of the new concept will be explained more detailed, as the developed
combinations and rules of designing new mass customized floorings are then subject to
simulation experiments, the results of which we present in the last section. Figure 3 gives an
overview of the elements of new production concepts.
Figure 3: Elements of the New Production Concept
4.1 Modularized Floorboards
4.1.1 Concept 1 – Sorted Modularization
As described above, customers can order packages of parquet flooring only in the distinctive
qualities. Our new approach is, instead of 5 qualities, to assemble 3 different types of floor
modules (in the following called M-Boards): With these M-Boards each customer should be able
to design his own specific and unique floor. In order to standardize the M-Boards, different
sorting algorithms and combinations of characteristics have to be implemented in the production
process. M-Board A has very even and uniform features, M-Board C has a very wild appearance
resulting from its components and thus, M-Board B is a mix of all other characteristics, resulting
in a standardized appearance. In difference to current production and ordering processes
customers are allowed to place orders for every type of M-Board in every wood sort they want
and thus over former categorical quality borders. Figure 4 gives an example of our idea. When
defining for example category B of our concept, companies are then able to use much more
grading flexibility as customer requirements are changing.
A
A
A
A
A
A
A
A
A
A
A
A
AAA
BBB
BBBB
CCC
BCCB
CCC
BCCB
BBB
BCCB
AAA
BBBB
Figure 4: Example of a customized floor from M-Boards
Several different CODPs are possible when implementing this production concept. For example,
the CODP of this new production concept could be postponed until the final packaging and
shipping processes, as the information technologies yet to be described could allow a constant
production flow and thus help in smoothing out some of the difficulties like buffer overflow in
today’s production environments. Figure 5 shows the new concept and its possible CODPs.
Figure 5: Process Flow and possible CODPs of Concept 1
4.1.2 Concept 2 – Unsorted Modularization
A second approach we tested with simulation experiments is to assemble M-Boards WITHOUT
the cost intensive step of pre-sorting the lamellae. M-Boards are produced in a first-come-first-
serve production flow; several different characteristics of wood can by chance alone be mixed on
one M-Board. When press-sized together the M-Boards are afterwards classified and put into one
of the three categories A, B and C. The remaining parts of the production processes are left
untouched. We undertook research in the feasibility of the concept and especially the question,
which classification algorithms results in how many and what sort of M-Boards produced. As
this concept could never be tested in a real-life-environment due to cost reasons, we also used the
method of simulation to test, if such a concept could be a feasible idea of mass customizing
parquet flooring. Figure 6 shows the idea.
Figure 6: Process Flow and possible CODP of Concept 2
4.2 Product Configurator, Database and RFID support
Product configurators are very helpful tools in implementing mass customization concepts. They
allow the customer to create, design or build their individual product at home via web-based
software solutions. Automotive and computer industries have been the first innovators in
enabling their customers to design their personal car or PC. Nowadays product configurators are
also common tools in the furniture retail business (e.g. the famous IKEA kitchen configurator).
We now present a working solution of a configurator for wooden floors. The software tiloVCAD
by an Austrian producer of natural floorings already has nearly all functionalities to enable
customers to design and help with the ordering processes of customized parquet floorings. The
tool can be easily obtained at the manufacturer’s website and is easily and intuitively usable even
by customers without much experience. It even offers the possibility to code individual patterns
or designs of floorings via simple ASCII-Files. Customers can draw their desired room or
surface area and place modules, representing all possible wood types and qualities the company
offers, in the drawing. Complex layings or edges in every desired length and width can be
created with every product variant. Photo-realistic images of the results help customers in
visualizing their individual floor. The software also offers an automatic calculation of a bill of
material (BOM) needed, and creates automatically company offers with actual prices, stored in a
database. If the tool is used by commercial users like distributors they can individualize the
software and offers with their own brand, logo, name etc. in order to give out legally binding
offers. Figure 7 shows examples of test-configurations with the tool described (cf.:
www.tilovcad.com)
How could now a product configurator like tiloVCAD be more integrated than today in a mass
customization concept of parquet flooring? When press-sizing the upper layer of the board
(which is, remember, the PPD of the whole process) digital high-resolution images of every
floorboard produced could be taken and stored in a database.
Figure 7: Screenshots of a configuration process with tiloVCAD
Each image has a timestamp so that the database does represent the current inventory of semi
finished and finished products in real-time. Intelligent surface recognition which nowadays is
working in the sorting process as well can be used to differ between idealistic representatives of
every wood type and quality. A database of all work-in-process goods and finished goods would
thus not contain several millions of records, but only one per representative. When customers
apply the configurator they have online-access to this database and can thus design their
individual floors with products actually on inventory or in process. The time-based storage of the
datasets enables the company to be much more precisely in their promised delivery time;
extrapolation from production know-how together with current material requirement planning
allow to give order promises for the future for items not currently on inventory. Another
advantage for the company would be smoothed inventories, as it only offers items currently
available instead of trying to produce for the customer, while at the same time the customer has
through the online configuration process the sensation of considerable more choice possibilities
then in today’s conventional order processes. When press-sizing the upper layer, RFID tags
could be also attached to every floorboard. Every single board would be thus traceable in the
whole production process and in the stock. This should not be a technical problem as RFID
enables every producer worldwide more than 1 quadrillion (a 1 with 15 zeros) identification
numbers for his products, which is equivalent to numbering every individual rice grain on earth
for more than 88 years (Karp and Springer 2005). The RFID chips enable the AS/RS system of
the inventory to pick and batch the single modules for every customized floor ordered. Beyond,
the individual laying plan for every floor created with the configurator and ordered is stored on
the RFID tags of every floorboard, so that while the retrieving processes take place, individual
detailed laying plans can be printed and commissioned too. When a customer orders different
floor modules, different laying plans can be printed, each one highlighting the zone of the
surface, in which the floorboard has to be put in order to create the individual effects desired. As
every RFID chip also stores the time of the production of the board it belongs to, FCFS-policies
can be ensured. Other possibilities could include standardised interfaces for RFID with laptops
and PDAs which enable customers to display their laying plan on their Computer or TV while
doing their home improvement. As RFID tags are energized via passive readout processes they
will lose their functionality once they are glued to the floor. Costs do also not seem to be a
counter-argument. At an estimated retail price of 50 Euro per square meter of parquet flooring,
the price of an RFID Chip (ca. 0.02 Euro) is not even the one-tenth of a percent. Of course such a
system would have considerable set-up and running costs.
5 Simulation Study
We tested the above described new production concepts with the method of process simulation
for validity and applicability. Using the case study we built a complex model of the whole
production process. This model was then validated with the known data before we tested several
possible production process reconfigurations. The most relevant scenarios and their results were
then set in comparison the validated model of current processes. In the current state of research
we only test the possibility of adaptation for mass customization of the production processes,
measured by the ability to fulfill customer’s wishes at least to the same extent as current
processes. Related financial issues like cost or revenue management will be topics for further
research.
5.1 Model Environment
In order to conduct the simulation experiments for testing modularization strategies in the
parquet industry we implemented real-life production data from more than 14 months with the
simulation software Arena®. The real-life data are available for a production line which
produces floorboards in 19 different types of wood and up to five different qualities. Customer
orders for the same time frame where implemented in the simulation as well. Floorboards are
produced according to our specifications and customers are able to order also according to these
specifications. The results of simulation runs of the current production processes were used as
references to compare the performance of the two new modularization concepts.
Table 1 shows the input-percentages of the different wood types; Table 2 gives an overview of
the matrix of the original quality outputs per wood type. It gives also a good idea of the
difficulties caused by inconstant quality of the raw material and its diverse output per wood type.
Table 1: Wood Types Table 2: Output Quality per Wood Type
Wood
T
yp
eqA qB qC qD qE
1 4,47% 61,56% 24,61% 8,15% 1,21%
2 7,72% 52,36% 30,50% 2,39% 7,04%
3 0,00% 100,00% 0,00% 0,00% 0,00%
4 0,06% 93,21% 0,06% 6,66% 0,01%
5 0,00% 64,62% 0,06% 35,29% 0,03%
6 6,83% 67,25% 10,77% 14,97% 0,19%
7 5,15% 57,88% 27,07% 4,63% 5,26%
8 0,00% 96,84% 0,00% 3,16% 0,00%
9 5,70% 54,79% 25,16% 14,22% 0,13%
10 0,00% 100,00% 0,00% 0,00% 0,00%
11 9,72% 72,38% 16,06% 1,83% 0,00%
12 0,00% 79,76% 0,24% 20,00% 0,00%
13 0,00% 56,07% 0,00% 43,93% 0,00%
14 0,00% 80,26% 0,00% 19,74% 0,00%
15 0,00% 98,70% 0,00% 1,30% 0,00%
16 0,00% 90,66% 0,00% 9,34% 0,00%
17 0,00% 100,00% 0,00% 0,00% 0,00%
18 0,00% 100,00% 0,00% 0,00% 0,00%
19 0,00% 100,00% 0,00% 0,00% 0,00%
Wood
T
yp
eInput
Percent
6 19,84%
9 19,35%
1 14,21%
7 11,04%
5 6,97%
2 6,02%
16 4,31%
13 3,03%
12 2,64%
4 2,64%
11 2,38%
8 2,36%
15 1,81%
14 1,40%
17 0,70%
10 0,54%
3 0,44%
19 0,25%
18 0,07%
5.2 Simulation Scenarios
5.2.1 Scenario Status Quo
As already mentioned, this simulation run is the core model of all tested production
environments and used as a reference to compare the new modularization concepts. One product
line is produced in the mentioned 19 different wood types in 5 different qualities. Customers can
place their orders in the same distinctive categories. The CODP is located at the retrieving
process of the upper layers from the conditioning chamber. 18 lamellae of the same quality and
wood type are glued together to form one floorboard, mixing of qualities is forbidden, thus five
different M-Boards per wood type are produced. Customer orders in this simulation run
correspond exactly to the observed real-life data, the demand for M-Boards d(M) corresponds
exactly to the demand for the five qualities qA, qB, qC, qD and qE for every single wood type.
Table 3 shows the parameters of the described model.
Concept Status Quo
Sorting step included Yes
CODP Retrieving process
Different qualities qA, qB, qC, qD, qE
Wood types 19
Lamellae/floorboard 18
Different qualities per M-
Board? No
MA = 18*qA
MB = 18*qB
MC = 18*qC
MD = 18*qD
ME = 18*qE
d(MA) = d(qA)
d(MB) = d(qB)
d(MC) = d(qC)
d(MD) = d(qD)
d(ME) = d(qE)
Possible M-Boards and their
composition
Customer orders or demand
d(M)
Table 3: Parameters of simulation scenario Status Quo
5.2.2 Simulation Scenario Sorted Modularization
When testing our production concept of sorted modularization of the floorboards, the
compositions of the M-Boards change. Table 4 shows the parameters of version 1.0 of this
concept. As before, we do not allow for a mix of different qualities on a single M-Board, thus
MA consists only of lamellae of the quality qA while MB can either be composed of quality qB
or quality qC and MC can either be composed of quality qD or quality qE. Customer orders are
modelled analogously, thus for example demand for floorboards MB equals the sum of the
current part demands for qB and qC.
Version 1.1 of the concept of sorted modularization differs again in the combination possibilities
of M-Boards. Now a mix of different qualities of lamellae on sinlge floorboards is allowed. This
does not cause any difference in composing MA, but for MB, which now can be composed of any
combination of the qualities qB and qC, as long as the restriction of 18 lamellae per M-Boards is
satisfied. The same is true for the composition of MC. Customer orders are analogous to the
former model (see also Table 5). Of course these modularization metrics can and should be
adapted overtime and from every producer, in order to reflect his actual input quality distribution.
Concept Sorted Modularization Sorted Modularization
Version Version 1.0 Version 1.1
Sorting step included Yes Yes
CODP Retrieving process Retrieving process
Different qualities qA, qB, qC, qD, qE qA, qB, qC, qD, qE
Wood types 19 19
Lamellae/floorboard 18 18
Different qualities per M-Board? No Yes
MA = 18*qA MA = 18*qA
MB = 18*qB OR 18*qC MB = x*qB AND y*qC;
x+y=18; 0≤x≤18, 0≤y≤18
MC = 18*qD OR 18*qE MC = x*qD AND y*qE;
x+y=18; 0≤x≤18, 0≤y≤18
d(MA) = d(qA) d(MA) = d(qA)
d(MB) = d(qB) AND d(qC) d(MB) = d(qB) AND d(qC)
d(MC) = d(qD) AND d(qE) d(MC) = d(qD) AND d(qE)
Possible M-Boards and their composition
Customer orders or demand d(M)
Table 5: Version 1.0 and Version 1.1 scenario Sorted Modularization
5.2.3 Simulation Scenario Unsorted Modularization
The above described simulation scenarios all include the cost-intensive production process step
of pre-sorting the lamellae into different qualities before press sizing the upper layers. We
simulated a production environment, in which M-Boards are produced in a first-come-first-serve
production flow. As a consequence, different characteristics of wood can by chance alone be
mixed on one M-Board, which is classified after the press-sizing into one of the three categories
A-C. The remaining parts of the production processes are left untouched. The feasibility of the
concept and especially the question, which classification algorithms results in how many and
what sort of M-Boards produced are tested in the simulation models, as a real-life-test could
never be carried out due to cost reasons.
Table 6 shows the parameters implemented in the model. As one can see, no sorting step in the
production process is included. In order to identify the produced floorboards and to determine
which category of M-Board is applicable, classification rules are implemented too. In our
example this means that a board MA must contain at least 6 lamellae of quality qA but at most 12
of quality qB. No other qualities are allowed on this M_Board. MC contains at least in sum 9
lamellae of quality qD and qE, other qualities are possible as they are not influencing the
impression of the board negatively. All other combinations, which are thought to be by far the
biggest output and thus the standard board produced, are labelled as board MB, which
accordingly is set to ELSE.
The table does not include the formulation of customer orders and demand, for a simple reason:
according to our experiments such a production concept of unsorted modularization would be,
although revolutionary, simple unfeasible. The output of the above described model, fed with
real-life data from our case study of a whole production year, resulted in approximately 13 pieces
of MA produced 867000 boards MB und 19000 of MC. As the composition rules are very strict,
modularization via unsorted FCFS- production is not an option to mass customize parquet
flooring.
Concept Unsorted Modularization
Sorting step included No
CODP Retrieving process
Different qualities qA, qB, qC, qD, qE
Wood types 19
Lamellae/floorboard 18
Different qualities per M-Board? Yes
MA = x*qA AND y*qB; x+y=18; 6≤x≤18, 0≤y≤12
MB = ELSE
MC = x*qD AND y*qE AND z*(qA,qB,qC); x+y+z=18;
x+
y
≥9
;
0≤x≤18
,
0≤
y
≤18
,
0≤z≤9
Possible M-Boards and their
composition
Table 6: Scenario Unsorted Modularization
Other experiments searching for a feasible solution resulted in composition rules including at
least 14 lamellae of qA and qB and 2 lamellae of qD when putting together a board MA. This
may be mathematically correct in order to solve for the implemented customer demand, but from
the design point of view of the impression the board would create, this composition would be
nonsense. Other simulation experiments confirmed our observations: the higher the number of
different characteristics of wood, the smaller is the probability of a cluster of characteristics
forming the impression and M-Board intended. But a FCFS-policy implies no waiting, storage
and sorting steps in order to make up for the small probability of effectively combining unsorted
raw material to different, modularized floorboards. To implement mass customization production
concepts in the parquet industry, this concept is neither applicable nor feasible.
5.3 Results and Discussion
All relevant simulation scenarios resulted in different, but robust outputs, both in produced
variants and in parameters such as throughput time or customer satisfaction. This allows a
comparison between the models is applicable. Robustness was tested in several simulation runs
of the same model subject to by-chance variation of time and quantity in both raw material input
and customer demand. To compare the output of each model its order fulfillment results were set
into relation to the distinctive customer orders. Also benchmarks of the storing process at the
CODP have been measured. Table 7 gives an overview of the results. According to reason
mentioned above concept 2 (unsorted modularization) is not included in the discussion of the
results. All benchmarks are given in both the median and the mean, as the results of the (up to
95) product variants differ on a broad scale.
The Order Fulfilment Rate is the relation of floorboards of all possible product variants (19
wood type times 3-5 different qualities or modules) produced to the floorboards ordered by
customers. The Rate of Numbers of Fulfilled Orders has a similar application, but it takes the
number of fulfilled orders, not produced pieces of floorboards into account. These two
benchmarks can differ as terminating conditions of the simulation model can set in during the
production of orders. Thus, the order is marked as unfulfilled, but the floorboards already
produced are counted into the Order Fulfilment Rate. An interpretation of the results is simple:
The higher one of these two benchmarks, the better customer demand can be met by the
simulated production concept. The other parameters take space resources into account. They are
measures of the usage of storage capacity in the distinctive production concepts. The observed
inventory is the conditioning chamber, which is also the decoupling point of the production
process and thus one of the major bottlenecks. These benchmarks do give an idea how the
different product variants claim finite resources and how big the potential of fulfilment of
customer orders is, independent from real orders.
A mean value of 138,46% CODP Stock Level/Orders in the simulation model Status Quo
means that, over the whole observed production period of 14 months, at the decoupling point
semi-finished products are stored which exceed more than 1.3 times the customer demand. But at
the same time, taking a look at the Order Fulfilment Rate only 85% of the demand can be
really fulfilled with this stock level. This gives an impression of the problems connected with the
natural raw material wood. Customers order for other or more of the qualities than the raw
material input can supply. If we take a look at the median of the same model and parameters it
shows, that not all wood types and qualities are subject to this phenomenon, as both benchmarks
are nearly the same. Of course, estimated customer behaviour and thus orders is the other
variable in our simulation model, which also explains the higher percentages of the Stock
level/Orders benchmark of the new concepts. The other two parameters measured (CODP Stock
Output/Input and Stock End-Level/Start-Level) seem to be self-explicable. The higher the
output of the conditioning chamber and the lower the end level of stock at the CODP, the better
the simulation scenarios use the restricted resources.
mean Status Quo Concept 1.0 Concept 1.1
Order Fulfilment Rate 80,02% 84,11% 84,42%
Rate of Number of
Fulfilled Orders 85,53% 89,62% 89,67%
CODP Stock Level
Orders
CODP Stock Output
CODP Stock Input
CODP Stock End-Level
CODP Stock Start-Level
median Status Quo Concept 1.0 Concept 1.1
Order Fulfilment Rate 93,72% 92,45% 94,69%
Rate of Number of
Fulfilled Orders 98,01% 97,42% 98,22%
CODP Stock Level
Orders
CODP Stock Output
CODP Stock Input
CODP Stock End-Level
CODP Stock Start-Level
93,16%
11,85% 7,30% 10,08%
92,94% 96,29%
26,72%
98,55% 94,86% 98,07%
34,57% 25,63%
156,22%
81,42% 82,63% 80,32%
138,46% 168,31%
Table 7: Outputs of the tested simulation environments
6 Conclusion
We have shown that new production concepts in the parquet industry can deliver better results
than current production processes. Over all simulation runs Concept 1.1 seems to be slightly
more effective in meeting customer demand than Concept 1.0. One of the reasons for this can be
the possibility on purpose grouping of similar wood characteristics on one floorboard. We have
also shown that, the higher the number of different characteristics of wood, the smaller is the
probability of a cluster of characteristics to form the impression intended. A production concept
of parquet flooring without the step of sorting and clustering wood qualities or characteristics
(like concept 2) is thus neither applicable nor feasible.
Interpreting the results one finds other interesting angles as well. Current production processes of
parquet show already a very high level of customization. Differences to our concepts proposed
and tested lie within the restrictions customers face when ordering their individual floorboards.
Current processes only allow and promote only ordering for one quality and wood type, while
our proposed new concepts make way for configuration and ordering possibilities over former
category bounders and restrictions while similarly holding benchmarks or even outdoing them
when compared to conventional processes.
Nevertheless we must state need for further research into these concepts and its processes in
order once to be able to set up such a parquet factory of the future. From today’s point of view
implementing new IT concepts in current production processes and enabling customers with the
tools and knowledge about the individualisation possibilities already offered seems a more
promising way to achieve the benefits of mass customization in the flooring industry than
establishing new production, sorting and combination processes of the individual boards. As up
to now our research has been mostly exploratory in character (answering the question: How
could a mass customization concept work?) future research has to take strong look financial
aspects such as cost or revenue management into account. Another important direction of future
research is under which circumstances such a production concept could also be applicable for
other wood-processing industries.
References
Blecker, T. (2003): Web-based Manufacturing. Ansatz eines betriebswirtschaftlichen Konzepts
einer internetbasierten Produktion. Diskussionsbeiträge des Instituts für
Wirtschaftswissenschaften der Universität Klagenfurt Nr. 2003/01, Universität Klagenfurt.
Hamberger, P. (1995): Die Parkettindustrie in Westeuropa. Eine Branchenanalyse unter
besonderer Berücksichtigung des österreichischen und deutschen Marktes. Dissertation,
Universität für Bodenkultur, Wien.
Jammernegg, W. Reiner, G. (2004): Performance improvement of supply chain processes by
coordinated inventory and capacity management. Paper presented at the 13th International
Symposium on Inventories in Budapest, August 22-27, 2004.
Karp, Alexander, Springer, Gudrun (2005): Jedem Ding seine Identität? Austria Innovativ Nr. 4,
Wien. 44-46.
Kisseloff, P. (1974): Die moderne Parkettherstellung – Stand der Technik. Holz als Roh- und
Werkstoff, 32. 127-134.
Kruse, K. / Teischinger, A. / Emmler, R. / Tobisch, S. (2003): Wood based Floor Coverings, in:
Johansson et al. Wood adhesion and glued products, State of the art Report, Cost Action E13.
McCarthy, I.P. (2004): Special issue editorial: the what, why and how of mass customization.
Production Planning & Control, 15, 4. 347-351.
Meyr, H. (2003): Die Bedeutung von Entkopplungspunkten für die operative Planung von
Supply Chains. Zeitschrift für Betriebswirtschaft, 73, 9. 941-962.
Olhager, J. (2003): Strategic positioning of the order penetration point. International Journal of
Production Economics, 85. 319-329.
Pagh Janus D. / Cooper Martha C. (1998): Supply chain postponement and speculation
strategies: How to choose the right strategy. Journal of Business Logistics, 19, 2. 13-33.
Piller, F. (1998): WISU: Begriffe, die man kennen muss….Vorfertigungsgrad. Das
Wirtschaftsstudium (WISU), 10. 1058.
Pine B., J. II. (1993): Making Mass Customization Happen: Strategies for the New Competitive
Realities. Planning Review, 21, 5. 23-24.
ProWood Stiftung. (2004): Presseinformation „Holzbearbeitungstechnologien 2015“. Retrieved
March 15, 2005 from http://www.prowood.org/PDF/PDF166.pdf
Reiner, G. (2005): Customer-oriented improvement and evaluation of supply chain processes
supported by simulation models. International Journal of Production Economics 96, 381-395.
Schmidt, G. (1999): Mass Customization. Teaching Note, Georgetown University.
Sigg, A. / Jonas, U. (2003): Der standardisierte Sonderwunsch. möbelfertigung, 2. 131.
Swaminathan Jayashankar M. (2001): Enabling Customization Using Standardized Operations.
California Management Review, 43, 3, 125-135.
van Hoek R. I. (2001): The rediscovery of postponement: A literature review and directions for
research. Journal of Operations Management, 19, 2. 161-184.
Yang, B. / Burns, N. (2003): Implications of postponement for the supply chain. International
Journal of Production Research, 41, 9. 2075-2090.
