a profit and loss analysis for make-to-order versus make-to-stock policy—a supply chain case study
TRANSCRIPT
This article was downloaded by: [North Dakota State University]On: 14 October 2014, At: 18:50Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK
The Engineering Economist:A Journal Devoted to theProblems of Capital InvestmentPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/utee20
A Profit And Loss Analysis ForMake-To-Order Versus Make-To-Stock Policy—a Supply ChainCase StudySameer Kumar a , Daniel A. Nottestad b & John F.Macklin ba Opus College of Business , University of St.Thomas , Minneapolis, Minnesota, USAb 3M Company , St. Paul, Minnesota, USAPublished online: 06 Jun 2007.
To cite this article: Sameer Kumar , Daniel A. Nottestad & John F. Macklin (2007)A Profit And Loss Analysis For Make-To-Order Versus Make-To-Stock Policy—a SupplyChain Case Study, The Engineering Economist: A Journal Devoted to the Problems ofCapital Investment, 52:2, 141-156, DOI: 10.1080/00137910701328953
To link to this article: http://dx.doi.org/10.1080/00137910701328953
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness,or suitability for any purpose of the Content. Any opinions and viewsexpressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of theContent should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of theContent.
This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan,sub-licensing, systematic supply, or distribution in any form to anyone isexpressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
The Engineering Economist, 52: 141–156Copyright © 2007 Institute of Industrial EngineersISSN: 0013-791X print / 1547-2701 onlineDOI: 10.1080/00137910701328953
A PROFIT AND LOSS ANALYSIS FOR MAKE-TO-ORDERVERSUS MAKE-TO-STOCK POLICY—A SUPPLY CHAINCASE STUDY
Sameer Kumar
Opus College of Business, University of St. Thomas, Minneapolis,Minnesota, USA
Daniel A. Nottestad and John F. Macklin
3M Company, St. Paul, Minnesota, USA
The make-to-order (MTO) or make-to-stock (MTS) decision is importantfor contract manufacturers supporting product supply chains. This casestudy provides an integrated profit and loss investment analysis for MTOversus MTS policy while also quantifying factory cost. The use of discrete-event simulation integrated with Excel provides a proactive decision sup-port application to predict lead time and profitability of an extruded partwithin a manufacturing supply chain at 3M Company headquartered inMaplewood, Minnesota. The analysis presented predicts the conditionswhere a make-to-stock policy is better than a make-to-order policy in termsof operating income for a single SKU (product) in a large multinationalmanufacturing company acting as a contract manufacturer. We define aninventory to order quantity (IOQ) ratio and use this metric with scenarioanalysis to maximize operating income. The IOQ ratio showcased in thisstudy is applicable for supply chains with predictable customer demand.
INTRODUCTION
Outsourcing non-core competency manufacturing to contract manufac-
turers is commonplace in today’s marketplace. The study explores is-
sues associated with a make-to-order (MTO) versus make-to-stock (MTS)
Address correspondence to Sameer Kumar, Opus College of Business, University ofSt. Thomas, Mail # TMH 343, 1000 LaSalle Avenue, Minneapolis, MN 55403-2005. E-mail:[email protected]
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
142 S. Kumar et al.
decision faced by a contract manufacturer in a product supply chain. In-
ventory reduction is a key initiative in improving financial performance at
many companies, including 3M. Interestingly, the analysis presented in this
article provides the financial justification for holding inventory based on the
conditions examined. A 3M division, acting as a contract manufacturer and
using proprietary processes, produces an extruded semi-finished product
that the hiring firm further processes into a finished good using their own
proprietary processes. A computer-based decision analysis model provides
the 3M business team with the ability to maximize operating income within
the semi-finished product’s supply chain.
With the market pressures for attaining zero inventory, many world-class
manufacturers may have to determine when it is optimal to hold a finished
goods inventory and when it is not. In general, when a finished (in this case,
semi-finished) goods inventory is held for a product at one (or more) supply
chain node(s), we consider the product produced with a MTS policy;
otherwise, we consider that the product is produced with a MTO policy.
Designating the product as MTO solely indicates stockless production, and
it does not imply that the product is made to order due to customized spec-
ifications. This problem is commonly faced by firms in many industries,
possibly with some variations. There are a number of analytical treatments
of the MTO versus MTS decision models in the inventory management
literature (Popp, 1965; Williams, 1984; Li, 1992; Arreola-Risa, 1996;
Sox et al., 1997; Federgruen and Katalan, 1995, 1999; Arreola-Risa and
Decroix, 1998; Carr and Duenyas, 1998; Rajagopalan, 2002).
The literature on the issue of MTS versus MTO goes back to the 1960s
when Popp (1965) presented a simple single-item stochastic inventory
model with zero lead time production or replenishment of an item. Simple
cost comparisons of making the item to stock versus making it to order were
presented. A few papers appeared on this topic during the 1960s, 1970s,
and 1980s, with the most relevant one being Williams (1984). He presents
a model similar to Rajagopalan (2002); he also considers a multi-item envi-
ronment with limited capacity and significant changeover times, stochastic
demands, and processing times. However, there are important differences.
Williams assumes that lower demand items are MTO and higher demand
items are MTS. So, the items are ordered based on ascending demand and
the decision is to identify a threshold item such that items with lower de-
mand follow MTO policy while the others follow MTS policy. Rajagopalan
does not make such an assumption, allowing factors other than demand to
impact MTO/MTS decisions.
Li (1992) investigates the optimality of MTO/MTS in a multi-firm mar-
ket where companies compete for customers based on delivery time for
orders. He considers the one-product case and assumes that unit manufac-
turing times are exponentially distributed. The inventory cost function is a
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 143
present value where the interest rate is compounded continuously. Arreola-
Risa (1996) contemplates a production inventory system for a company
producing multiple heterogeneous products at a shared manufacturing fa-
cility where manufacturing times for the different products are identical
and gamma distributed, and backorder costs are incurred dollars per unit
per unit time. He determines the value of the squared coefficient of vari-
ation of manufacturing time at which the firm is indifferent to produce a
product with either MTO or MTS policy. Arreola-Risa and Decroix (1998)
extend Arreola-Risa’s (1996) production inventory system with manufac-
turing times being general independent and identically distributed random
variables. In addition, different products may have different manufacturing
time distributions. They derive optimality conditions for MTO and MTS
policies and also study the impact of several managerial considerations
on the MTO versus MTS decision. They ponder a company that produces
multiple products facing random demands.
Federgruen and Katalan (1995, 1999) address a similar problem, but their
focus is different. The MTO items in their work, unlike in Rajagopalan’s
(2002), “are not primarily distinguished by. . . zero or small inventories,
but rather by the type of priority they are given in the overall production
strategy” (1995, p. 2). In fact, they define two classes of items, A and B
(equivalent to MTS and MTO, respectively), with B items “interrupting”
the production of A items. The primary focus of their paper is on different
strategies for deciding whether and when to interrupt the production of A
with B items and providing related insights. In contrast, the primary focus
of Rajagopalan’s paper is on whether an item should be MTO or MTS;
accordingly, his formulation and analysis are different.
Carr and Duenyas (1998) also consider a system with some items MTS
and others MTO, but their focus is on whether to accept or reject orders
for MTO items and deciding contractual quantities for MTS items. Sox et
al. (1997) ponder a similar problem where some items are MTS and some
MTO, with demand for MTO items to be satisfied within a certain time
window. They consider more complex scheduling policies. However unlike
in Rajagopalan’s (2002) paper, these papers do not take into consideration
setup times, lot sizing, and related tradeoffs.
A few recent computer-based inventory management models similar to
the one proposed here are the ones developed by Kumar and Meade (2006)
and Boyd (1999). Kumar and Meade (2006) have developed a multi-period
simulation model of a factory with lean manufacturing wherein they study
operational performance trends over a 12-month period based on the in-
ventory reduction policy chosen. Boyd (1999) develops computer-based
decision models to investigate effects on management decision-making
and firm performance studying production planning and control, and cost
accounting systems. The simulation-based decision model developed for
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
144 S. Kumar et al.
the contract manufacturer in our study is unique in many respects relative
to the production planning and inventory management decision models
reported in the literature. The proposed model quantifies supply chain per-
formance and presents profit and loss analysis to estimate the profitability
and factory cost of a successful production run based on information from
the model. The case study reported also shows that a blanket “hold no
inventory” policy can be counterproductive as net income may actually
increase in a MTS environment.
EXAMPLE CASE STUDY
In this section, we illustrate the development of a dynamic decision support
tool for a 3M business team, acting as contract manufacturer, to predict the
performance of a supply chain. An integrated approach using discrete-event
simulation and Excel with Visual Basic for Applications (VBA) provides
the foundation for this application. The analysis predicts the conditions
where MTS policy is better than MTO policy in terms of operating income
for a single SKU (product) in a large multinational manufacturing company
that is acting as a contract manufacturing supplier.
This case study results from a request for simulation services to pro-
vide a decision support tool to quantify the profitability of a semi-finished
extruded part within a 3M supply chain. A simulation model using WIT-
NESS software is developed that links an Excel front-end to WITNESS.
Key information for each manufacturing process step is recorded in Ex-
cel, including cycle time variation, process parameters, lot size, production
yields and product routings. The simulation model predicts the lead time
to complete a single order. Also, a financial profit and loss (P&L) model
is developed in Excel that estimates the profitability and factory cost of
a successful production run based on information from the model. Multi-
ple replications are run and charted to quantify variation and understand
sensitivity within the supply chain.
PROBLEM STATEMENT
Consider the manufacture of an extruded part in a high fixed factory cost
environment following a make-to-order policy instituted from an inventory
reduction initiative. The problem then is to determine the conditions, if any,
where profitability is improved by changing to a make-to-stock policy. The
supply chain for the extruded part following an MTO policy is straight-
forward. An order is received from the customer triggering the production
planning process to begin. Raw materials are procured and the order is
scheduled on the production line. After run planning is complete, parts
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 145
are extruded internally before shipping via truck to a local supplier for the
first of two converting operations. All of the parts required for the order
are shipped in a single shipment for batch processing at the converting
vendor. Next, the entire batch of parts is shipped back to the factory for
an internal coating operation. After coating is complete, the entire order is
shipped via truck to a vendor on the East Coast for final converting. Finally,
the customer (hiring manufacturer) takes ownership of the parts after the
converting vendor has finished processing the order.
The make-to-stock supply chain is similar to the make-to-order supply
chain. Under the MTS policy, semi-finished parts from extruding opera-
tions in excess of the customer order quantity are stored in a warehouse
where they are held until the customer places another order. If sufficient
inventory exists to fill a customer order, extruding operations are bypassed
and coating and converting operations start immediately. Otherwise, semi-
finished extruded parts need to be produced by extruding operations prior to
coating and converting operations. Rather than defining the semi-finished
inventory on a per unit basis, we use an inventory-to-order quantity (IOQ)
ratio (see Equation (1)). This ratio allows the results of the analysis to be
applicable regardless of order quantity and is simply equal to the desired
semi-finished inventory units held divided by the typical order quantity
units. As an example, an IOQ ratio = 2.0 implies that the number of semi-
finished parts produced for inventory is exactly equal to twice the forecasted
order quantity (that is, the number of semi-finished parts required to fill
an order). A detailed explanation of IOQ ratio is located in Analysis and
Results in the Scenario Analysis section.
IOQ ratio = Desired Inventory Units Held
Typical Order Quantity (based on forecasted annual demand)
(1)
The supply chain under both MTS and MTO policies is represented in
Figure 1.
MODELING METHODOLOGY
Variation is present in every process within the supply chain. Discrete-
event simulation is an excellent tool for handling this variation. Model
inputs are obtained from an Excel spreadsheet that includes information in
the following areas:
� Machine setups: changeover frequency and duration� Machine mean time between failures (MTBF): breakdown frequency
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
146 S. Kumar et al.
Figure 1. Supply chain for an extruded part.
� Machine mean time to repair (MTTR): repair times� Defect rates (yield): bad quality parts and failed production run� Transportation times: from plant to supplier (process “A” to process
“B”)� Waiting times: typical time spent waiting for work to begin� Process parameters: component life, part attributes, process rates� Supplier cycle times: time required to process parts at the supplier’s
facility� Material availability: time required to acquire raw material� Equipment availability: time spent waiting for time on process equip-
ment
An example of the spreadsheet containing the above is located in the
Appendix (Table A1). Note that all time is in minutes. Areas shaded in
gray are changeable by the process engineer (end user) and are loaded into
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 147
the simulation model at runtime. For change control, a users initials and
the date of change are entered in the “INIT” and “Date” columns.
KEY MODELING ASSUMPTIONS
Efficient and effective modeling of the supply chain requires that assump-
tions be made. These typically involve resource allocations, shift patterns,
and so on. A proper understanding of these assumptions is important in
analyzing the simulation results. Model assumptions follow:
� Labor is not a constrained resource.� Raw materials are not stocked.� Raw materials are ordered only after a customer order has been re-
ceived.� Model starts with clean extruders (initial clean time = 0 minutes).� Failed runs require waiting time for both process equipment and raw
materials.� Probability of a failed run does not decrease following a failed run.� Batch quantity is set equal to the total number of parts to complete an
order.� Making follows a 24-hour, 7-day shift pattern.� All other operations follow an 8-hour, 7-day shift pattern.
MODEL DEVELOPMENT
WITNESS simulation software is used to model the supply chain. The
model is programmed in minutes. The sequential process steps within the
supply chain are relatively simple to model, as routings are straightfor-
ward for this particular supply chain. At model runtime, the inputs from
the Excel spreadsheet are uploaded. The production equipment produces
a number of different products. Consequently, wait times of up to three
weeks to get on the production schedule are common. Extruding opera-
tions begin as soon as all raw materials are available and the extruding line
is available. The extruders require initial setup including a heating stage,
startup stage and optimization stage. After the setup of extruders is com-
plete, production begins and parts are made at a specified rate and yield.
As soon as extruding operations are complete, WITNESS determines if the
production run was good based on historical yield data. If the production
run was deemed “good,” the batch of parts required for the order is sent to
a local supplier for converting. The number of additional parts produced
for inventory is recorded. However, if the production run is deemed “fail,”
all parts created during the run are scrapped and run planning begins anew.
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
148 S. Kumar et al.
Once the local supplier receives the shipment, a waiting time is calculated
using historical data. After the wait time is over and the local supplier has
finished converting, the batch of parts is shipped back for coating. Again,
there is a wait time before coating begins. Once coating is complete, a truck
transports the parts to a supplier on the East Coast that converts each part
into a salable part. After yet another waiting period, the parts are converted
into finished goods. The customer takes ownership of the parts from this
point thus completing the supply chain. At the end of the simulated run,
data such as cycle times and production statistics are sent to Excel.
MODELING VARIATION
The variation within each process step is modeled using built-in distribu-
tions from WITNESS (Anon, 2005). Triangle distributions have been used
for processes lacking historical data. The triangle distribution includes a
minimum time, most likely time, and a maximum time. As more and more
production runs are completed, data can be collected and fitted to a “better”
distribution that reflects reality. Mean time to repair (MTTR) is modeled
using the Erlang distribution with a mean value and shape parameter. Mean
time between failures (MTBF) is modeled using the negative exponential
distribution with a specified mean. Extruder optimization is modeled using
the lognormal distribution with a specified mean and standard deviation.
Dedicated random number streams are assigned to each random variable
being modeled. Expert Fit software is used to determine the best distribu-
tion fit for various random variables based on historical data.
Sensitivity analysis is performed through an experiment in WITNESS.
The same requirements (order quantity) are run using different random
numbers for each trial. Random number seeds are handled by the WITNESS
software. No warm-up period is modeled as we account for the waiting
time to get on the production line and wait for raw materials to arrive at
the plant. Simulation duration is not fixed; instead, the simulation trial is
finished once order processing is complete. Throughput varies for each
trial as setup times and yields vary from run to run. Some trials have a
failed production run, resulting in a considerably longer lead time and a
considerable increase in factory cost. Net income also varies from trial to
trial.
FINANCIAL MODELING
Factory cost and profitability change from trial to trial due to variation
within the supply chain. Costs are categorized in an Excel P&L financial
model that helps track factory cost and profitability for each simulated
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 149
production run. An example P&L statement is located in the Appendix
(Table A2). Simulation output data used in the P&L is shaded in gray.
Discrete-event simulation facilitates activity-based costing. Variable raw
material costs are segregated into setup and production costs with usage
data provided by the simulation. Internal burden rates are also determined
from the usage statistics available in the simulation. Supplier costs are
variable and based on the order quantity. Transportation and freight costs
are fixed and charged per trip. Production run planning is an overhead cost
that is allocated per run. Inventory and carrying costs for raw materials,
work-in-process, and finished goods are determined based on an internal
holding cost factor and average holding period. The average holding period
depends on the IOQ ratio in the MTS case and is zero in the MTO case.
Sales are simply the order quantity multiplied by the price per part. Each
cost category is summed to obtain total factory cost. Operating income is
reported both before and after taxes.
SCENARIO ANALYSIS
The goal of the scenario analysis is to use data from a WITNESS simulation
model to decide whether a 3M business should be run using a make-to-
stock or a make-to-order policy. The analysis shows that in some cases
the business would see an increase in operating income (OI) using an
MTS policy. Furthermore, the analysis highlights the inventory level that
maximizes OI. This inventory level is characterized by the IOQ ratio. The
relationship of cost and OI relative to the IOQ ratio holds true regardless
of the order quantity.
Key Scenario Assumptions
The following are the assumptions for various scenarios analyzed in this
study:
� Customer behavior is predictable—we assume the ability to accurately
forecast annual customer demand. Our customer typically places or-
ders for a single product once per quarter in amounts that seldom
vary.� Orders are placed four times per year and are evenly spaced every 3
months.� Order quantity remains constant throughout the year.� Cost and operating income figures are totals for one year, or four
orders.� Each year starts with no inventory.
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
150 S. Kumar et al.
� No initial inventory requires production in the first quarter of the year.� One production run per year is a failed run.� Inventory costs associated with inventory generated during the fourth
quarter are not accounted for in the analysis.� Inventory is held after extruding operations and requires additional
processing.
One feature of the simulation model in this study is the built-in flexibility
to adapt to changing assumptions. The Excel interface facilitates changes in
assumptions or input parameters such as cycle times, yields, burden rates,
and the like.
Analysis and Results
To confirm that operating income trends for MTS policy at varied IOQ
ratios held true regardless of order quantity, the simulation was run using
two different order quantities: 35,000 finished parts and 50,000 finished
parts. Operating income is calculated for the MTO policy case and used as
the baseline in this analysis. Figure 2 records the results.
There are several interesting points to note regarding Figure 2. First,
notice that the shape of both curves is the same. This indicates that the
relationship between after-tax OI and the IOQ ratio holds true regardless
Figure 2. Change in make-to-stock operating income vs. make-to-orderoperating income at varying inventory order quantity ratios.
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 151
of order quantity. Also note the discontinuity in OI at IOQ ratios near 0.35,
1.0, and 3.0; at these IOQ ratios, the inventory being held allows the busi-
ness to skip one or more production runs. Each gap up in OI correlates
to one additional run being skipped resulting in a significant reduction in
raw material (startup), burden and freight costs. As the IOQ ratio increases
between these gaps, however, OI decreases. This is due to the increase in in-
ventory holding costs without the benefit of skipping additional production
runs. The IOQ ratio correlates to the run sequence for one year as follows:
• IOQ ratio = 0.0 (make to order)
• Run sequence = run, run, run, run
• Line turned on four times
• IOQ ratio = 0.35
• Run sequence = run, run, run, skip
• Line turned on three times
• IOQ ratio = 1.0
• Run sequence = run, skip, run, skip
• Line turned on two times
• IOQ ratio = 2.0
• Run sequence = run, skip, skip, run
• Line turned on two times
• IOQ ratio = 3.0
• Run sequence = run, skip, skip, skip
• Line turned on one time
The majority of the costs in the manufacturing process occur in extruding
operations, which produces the semi-finished part that is inventoried. The
most significant portion of these costs is due to the burden rate on the line.
Skipping a run eliminates the high production costs.
RESULTS SUMMARY
We summarize our findings from the analysis in terms of the following
major outcomes:
� An MTS policy is better than a MTO policy in some cases because
of the ability to eliminate production costs by skipping one or more
production runs.� Those IOQ ratios that yield higher operating income than the MTO
base case are shown in Figure 2.� The minimum IOQ that shows a benefit over a MTS policy is equal to
order quantity/3 (IOQ ratio ≈ 0.33) rounded up to the nearest whole
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
152 S. Kumar et al.
semi-finished part. This allows one skipped run in four, or once per
year.� IOQ = 1.0 is optimal in terms of cost and operating income, allowing
extruding to be turned on for every other order. This is true regardless
of order quantity.� An IOQ ratio = 3.0 produces lower OI than an IOQ ratio = 1.0, even
though the line only needs to be run once during the year. This is
because the additional inventory costs outweigh the benefit of running
only once per year.� For a given IOQ ratio, the order quantity may determine whether an
MTS policy is better than a MTO policy. For example, an IOQ = 3.0
produces better results than the MTO case for a 35,000 but not for a
50,000 finished part order quantity.
The analysis shows that the business should produce enough parts to
meet an IOQ ratio = 1.0. Operating income will be maximized, produc-
ing better results than a MTO policy even with the additional inventory
holding costs. This will require running the extruding line twice per year,
which will allow the technical aspects of the process to remain fresh for
the manufacturing team. Additionally, some level of safety stock, above an
IOQ ratio = 1.0, could be held and still produce better results than running
in MTO mode. These additional semi-finished parts could be used to ac-
count for order variation. The amount of additional safety stock that can
be produced while still producing better results than MTO is dependent on
the order quantity. For example, an IOQ ratio = 1.50 provides more OI
for a smaller order size than a larger order size (as shown in Figure 2). An
analysis would need to be performed for each specific order quantity to
determine if holding safety stock above an IOQ ratio = 1.0 produces better
results than a MTO case.
CONCLUSIONS
The WITNESS simulation model quantifies the variation in lead time and
profitability of a supply chain producing an extruded part. An Excel front-
end facilitates data entry into the model and an Excel back-end receives
data from the simulation. Sensitivity analysis is easily carried out giving
valuable insight into the performance of the supply chain. Business poli-
cies and system improvements can be modeled to understand the effect on
a supply chain in a simulated world without disturbing the existing sys-
tem. This model evaluates two basic inventory policies. A make-to-stock
business strategy may result in fewer production runs and lower total fac-
tory cost over a given period of time as the extra parts produced can be
stored until the next order arrives. The savings in factory cost may more
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 153
than offset the inventory holding costs. This is especially true when the
probability of a failed run is high, as is currently the case with this product.
The IOQ ratio showcased in this study is a useful metric that is applica-
ble in those instances where customer demand is predictable or able to
be forecast. If the customer’s ordering pattern (frequency and quantity)
is understood and predictable, the model can be used to quantify the dif-
ferences in an MTO policy versus an MTS policy. An optimal inventory
level can be determined using the simulation application, thereby providing
value-added decision support on an ongoing basis. Finally, if product costs
are allocated differently—for example, inexpensive extruding operations
or storing inventory in a finished goods state or at a different node in the
supply chain—the profit-maximizing strategy may switch from MTO to
MTS or vice versa.
REFERENCES
Anon. (2005) Learning WITNESS: Users manual. Lanner Group, Houston, TX. Avail-able at: http://www.lanner.com
Arreola-Risa, A. (1996) On the existence of a manufacturing randomness threshold.Working Paper, Department of Information and Operations Management, LowryMays College and Graduate School of Business, Texas A&M University.
Arreola-Risa, A. and DeCroix, G.A. (1998) Make-to-order versus make-to-stock in aproduction-inventory system with general production times. IIE Transactions, 30,705–713.
Boyd, L.H. (1999) Production planning and control and cost accounting systems:effects on management decision making and firm performance. Doctoral Disser-tation, UMI Co., Ann Arbor, MI.
Carr, S. and Duenyas, I. (1998) Optimal admission control and sequencing in a make-to-stock/ make-to-order production system. Working Paper, Department of Indus-trial and Operations Engineering, University of Michigan, Ann Arbor, MI.
Federgruen, A. and Katalan, Z. (1995) Make-to-stock or make-to-order: that is thequestion; novel answers to an ancient debate. Working paper, Graduate School ofBusiness, Columbia University, New York.
Federgruen, A. and Katalan, Z. (1999) The impact of adding a make-to-order itemto a make-to-stock system. Management Science, 45(7), 980–994.
Kumar, S. and Meade, D. (2006) Financial models and tools for managing lean manu-facturing. Boca Raton, FL: Auerbach Publications.
Li, L. (1992) The role of inventory in delivery-time competition. Management Science,38(2), 182–197.
Popp, W. (1965) Simple and combined inventory policies, production to stock or toorder? Management Science, 11(9), 868–873.
Rajagopalan, S. (2002) Make to order or make to stock: Model and application.Management Science, 48(2), 241–256.
Sox, C.R., Thomas, L.J. and McClain, J.O.(1997) Coordinating production and in-ventory to improve service. Management Science, 43(9), 1189–1197.
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
154 S. Kumar et al.
Williams, T.M. (1984) Special products and uncertainty in production/inventory sys-tems. European Journal of Operational Research, 15, 46–54.
APPENDIX
Table A1. Input worksheet for supply chain model
Uploaded to WITNESS at model runtime
INIT Date Setups Distribution Minimum Most likely Maximum
Clean time Fixed 4320Heat time Fixed 720Start up Triangle 600 780 1440Optimization Lognormal∗ 600 240
INIT Date Breakdowns Distribution Mean Shape
Extrusion mean time to repair Erlang 10.0 3Coating mean time to repair Erlang 10.0 2Extrusion MTBF Negexp 500.0Coating MTBF Negexp 500
INIT Date Defects Percentage
Defective parts: extrusion 5.50Defective parts: coating 4.25Failed production run 12.12
INIT Date Transportation Distribution Minimum Most likely Maximum
Plant to local converter Fixed 60Local converter to plant Fixed 60Plant to East Coast converter Triangle 1200 1440 1680Wait time at local converter Triangle 420 840 1260Wait time at plant Triangle 240 360 480Wait time at East Coast converter Triangle 240 360 480
INIT Date Process parameters Value Units
Process rate: extruding 60 Parts/minProcess rate: coating 45 Parts/minCritical component life 4800 Min
INIT Date Supplier cycle times Distribution Minimum Most likely Maximum
Local converter Triangle 50 60 70East Coast converter Triangle 235 260 285
INIT Date Model initialization Distribution Minimum Most likely Maximum
Production line availability Triangle 30,240 34,560 43,200Raw material procurement Triangle 30,240 34,560 40,320
∗Mean ∗SD.Time in minutes.
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
Make-to-Order versus Make-to-Stock 155
Table A2. Financial (P&L) worksheet for the supply chain model
Variablecost Rate
A Raw material costs Fixed cost ($/lb) (lb/h) Usage (h) Totals
1 Resin usage—setup1a Resin ABC — $1.75 75 44.20 $5,801.251b Resin XYZ — $1.25 75 44.20 $4,143.752 Resin
usage—productionSubtotal $9,945.00
2a Resin ABC — $1.75 75 45.80 $6,011.252b Resin XYZ — $1.25 75 45.80 $4,293.75
Subtotal $10,305.00
VariableB Production line burden
rateFixed cost cost Per unit Usage (h) Totals
1 Making — $1,500.00 hour 90.00 $135,000.002 Coating — $ 750.00 hour 100.00 $75,000.00
Subtotal $210,000.00
VariableC Suppliers Fixed cost cost Per unit Usage Totals
1 Local converter — $300.00 Batch 100 $30,000.002 East Coast converter — $5.00 Part 50,000 $250,000.00
Subtotal $280,000.00
VariableD Transportation/
freightFixed cost cost Per unit Usage (h) Totals
1 Plant to first supplier $100.00 — Trip NA $100.002 First supplier to plant $100.00 — Trip NA $100.003 Plant to second
supplier$2,000.00 — Trip NA $2,000.00
Subtotal $2,100.00
VariableE Overhead costs Fixed cost cost Per unit Usage (runs) Totals
1 Production runplanning
— $2,000.00 Run 2 $4,000.00
VariableF Inventories & carrying
costsFixed cost cost (annual) Per unit On hand Totals
1 Raw materials — $1.50 lb 10,000 $15,000.002 Work in process — $2,000.00 Batch 100 $50,000.003 Finished goods — $50.00 Part 0 —
Subtotal $65,000.00(Continued on next page)
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4
156 S. Kumar et al.
Table A 2. Financial (P&L) worksheet for the supply chain model(Continued)
G Sales $1,000,000H Total costs $581,350 Sales informationI Operating
income$418,650 Parts ordered 50,000
J Percent to sales 41.87% Price per part $20.00K Income tax @
37%$154,901
L After-taxoperatingincome
$263,750 Imported from simulationmodel
BIOGRAPHICAL SKETCHES
DR SAMEER KUMAR, P.E. ([email protected]) is currently a professor of decision sciences
and Qwest endowed chair in global communications and technology management in the
Opus College of Business, University of St. Thomas, Minneapolis, Minnesota. Prior to this
position he was professor of engineering and technology management at the University of St.
Thomas. Before joining the University of St. Thomas, Dr. Kumar was professor of industrial
engineering in the Department of Industrial Management, University of Wisconsin – Stout.
His major areas of research interests include optimization concepts applied to various aspects
of global supply chain management, information systems, technology management, product
and process innovation and capital investment justifications (over 100 refereed papers and 7
books). He earned a BS in mathematics, physics and chemistry and an MS in mathematics
from the University of Delhi in 1970 and 1973, respectively; an MS computer science from
the University of Nebraska in 1980; and an MS in industrial engineering and operations
research and a PhD in industrial engineering from the University of Minnesota in 1990 and
1994, respectively.
DANIEL A. NOTTESTAD is an engineering specialist at 3M Company in St. Paul, Minnesota. He
provides manufacturing and business systems analysis for 3M Divisions and 3M Corporate
Quality initiatives. Dan has worked in a variety of engineering positions since graduating
with a bachelor’s degree in mechanical engineering from the University of Wisconsin at
Madison in 1991. In addition to engineering analysis, work experience includes design,
project and plant engineering. Dan completed his masters degree in Manufacturing Systems
Engineering in 2002 from the University of St. Thomas, St. Paul, Minnesota.
JOHN F. MACKLIN is a Six Sigma Black Belt at the 3M Company in St. Paul, Minnesota.
He drives continual improvements in operational efficiency for one of 3M’s divisions by
employing various Six Sigma tools. He has worked in a variety of engineering positions
since graduating with a bachelor’s degree in mechanical engineering from Worcester Poly-
technic Institute, and a masters degree in mechanical engineering from the University of
Minnesota. In addition to his current role, his work experience includes design engineering
and manufacturing and business systems analysis.
Dow
nloa
ded
by [
Nor
th D
akot
a St
ate
Uni
vers
ity]
at 1
8:50
14
Oct
ober
201
4