You are on page 1of 25

Accepted Manuscript

Comparison between economic order/manufacture quantity and just-in-time


models from a thermodynamics point of view

Mohamad Y. Jaber, Maurice Bonney, Hussam Jawad

PII: S0360-8352(16)30327-8
DOI: http://dx.doi.org/10.1016/j.cie.2016.08.023
Reference: CAIE 4451

To appear in: Computers & Industrial Engineering

Please cite this article as: Jaber, M.Y., Bonney, M., Jawad, H., Comparison between economic order/manufacture
quantity and just-in-time models from a thermodynamics point of view, Computers & Industrial Engineering (2016),
doi: http://dx.doi.org/10.1016/j.cie.2016.08.023

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The 2015 International Conference on Computers & Industrial Engineering (CIE45)

Computers & Industrial Engineering (Special Issue on New Trends in Computers and
Industrial Engineering and their Applications)

Comparison between economic order/manufacture quantity and


just-in-time models from a thermodynamics point of view

Mohamad Y. Jabera*, Maurice Bonneyb, Hussam Jawada


a
Department of Mechanical and Industrial Engineering
Ryerson University
Toronto, ON, Canada M5B 2K3

b
Nottingham University Business School
Nottingham NG8 1BB, UK

Acknowledgment: The first and third authors thank the Natural Sciences and Engineering
Research Council of Canada (NSERC) for supporting this research.

*Corresponding author: mjaber@ryerson.ca; myjaber@gmail.com

Page 1 of 24
Comparison between economic order/manufacture quantity and
just-in-time models from a thermodynamics point of view

Abstract
Just-In-Time (JIT) suggests that inventory is a waste and should be reduced. However, under some
conditions, smaller and more frequent shipments can generate more waste, consume more
resources and increase congestion in a supply chain rendering the JIT policy less effective. This paper
first modifies the standard economic manufacture quantity (EMQ) models by including within them
the costs of transportation, worker stress, process quality, energy, and greenhouse gas emissions.
Then, it uses the second law of thermodynamics to calculate the entropy generated in the modified
EMQ and JIT models, using entropy as a measure of system sustainability. To make the comparison
meaningful, the EMQ model is modified to capture some of the costs that will work for and against
EMQ and JIT. The modification then adds an entropic component to the cost functions of EMQ and
JIT to capture the costs of disorder (entropic) that classical inventory analyses seldom include. The
results show that when worker stress and entropy costs are considered, a JIT policy can be more
expensive to operate than an EMQ policy.

Keywords: EMQ/JIT, emissions, energy, worker stress, entropy, sustainability

1. Introduction
The economic order/manufacture quantity (EOQ/EMQ) model developed by Harris (1913) is the
earliest reported scientific treatment of inventory systems. The model minimizes the sum of
inventory costs, ordering/setup and holding costs. Its simple mathematics and ease of calculation is
perhaps what contributed to its popularity with practitioners and academicians (e.g., Fulbright,
1979; Osteryoung et al., 1986). Many researchers (Andriolo et al., 2014; Drake and Marley, 2014;
Glock et al., 2014; Bushuev et al. 2015) continue to celebrate the EOQ/EMQ model and the model’s
popularity overshadows the critical voices (e.g., Selen and Woods, 1987; Woolsey, 1990; Weiss,
1990). For example, Woolsey (1990) argued that the assumptions of the EMQ/EPQ model are never
met, and concluded: “If you continue to love and use the EOQ without knowing what it is costing you,
I can only say that you deserve each other.”

Over the years, many ways to determine ordering and manufacturing quantities appeared, of which
just-in-time (JIT) is one (e.g., Andriolo et al., 2014). In contrast to the classical EOQ/EMQ model,

Page 2 of 24
just-in-time (JIT) produces a product when needed and delivers it when wanted by customers. JIT
considers inventory as waste that can be reduced by producing in smaller lots more frequently, so
that fewer items are in the pipeline (Bonney and Jaber, 2011). Planning for JIT also improves quality,
productivity, responsiveness, machines availability, workforce flexibility, space and equipment
utilization, flow, and relationship with suppliers (Cavinato, 1990; Jones, 1990; Mehra and Inman,
1992). JIT uses continuous improvement of which reducing setup time (cost) and saving space are
primary drivers (Zangwill, 1987; Jones, 1990; Cavinato, 1991). Readers are referred to Sohal et al.
(1989), Golhar and Stamm (1991), and Goyal and Deshmukh (1992) for a review of JIT systems. Of
course, a reduced setup cost in the model also reduces the calculated values of the EMQ.

The implementation of JIT has not been without challenges. Crawford et al. (1988) who conducted a
survey of early implementers of JIT reported that the problems included not meeting schedules, lack
of cross-trained workers, poor quality, supplier performance, and forecasting, inadequate
performance measures, machine breakdowns, problems with accounting, company resistance to
change, and interference between old and new methods. Serving customers with smaller lots makes
distribution channels busier requiring real-time monitoring by management, which is costly
(Cavinato, 1991). Wijewardena and De Zoysa (1999) reported that the increase in transportation
costs was significant when compared to the reduction in inventory costs. Some studies showed JIT
practices may increase labor intensity, characterized by workload, stress, and fatigue (Godard,
2001). Recently, Koukoulaki (2014) conducted an extensive review of the literature on lean
production environments and concluded that lean practices, such as JIT, that intensify workloads can
have a negative effect on workers’ health and safety. Some of the risk factors and health outcomes
reported in the study from different industries of Koukoulaki (2014) include, but are not limited to,
reduced rest breaks, musculoskeletal disorders, job stress levels, increasing workload, fast work pace
and long hours, and job satisfaction. On the other hand, Conti et al. (2006) concluded that lean
production is not inherently stressful and that stress levels are related to managerial decisions in
designing and operating the production systems.

The EOQ/EMQ and JIT models are not much different (Zangwill, 1987; Jones, 1990) but they reflect
different views of the same problem. JIT re-arranges the EOQ/EMQ formula by setting the lot size to
a desired value (e.g., demand per day) and then progressively reduces the setup cost; however,
achieving zero inventories is misleading (Zangwill, 1987). Quality improvements coupled with setup
reduction moves the EOQ/EMQ closer to JIT (Porteus, 1986). Chyr et al. (1990) revised the EOQ/EMQ
formula where they quantified the setup and the holding costs with setup time and damage cost,

Page 3 of 24
respectively, then compared the modified model to JIT, and showed that when the damage rate is
considered the EOQ/EMQ performed better. However, JIT often eliminates high damage cost and
other situations e.g. large set up times that encourage making items in large batches. Cao and
Schniederjans (2004) contended that the EMQ models in the literature that have considered the
impact of JIT manufacturing did not include the relevance of facility cost reduction. Their results
showed that when using space cost saving is small the revised EMQ/JIT model that includes facility
space reduction costs is superior to the classical EMQ model; the revised production model would be
superior to the EMQ model in most situations except for extreme demand levels. Another paper that
compares EMQ and JIT models is that of Moily (2015), who provided a theoretical economic analysis
of the EMQ model as a basis for linking to JIT, where a stable operating level is a key way to improve
productivity and quality. Moily (2015) also showed that it is possible to maximize profits for a facility
in an environment where demand is relatively sensitive to price and production level is high. The
above studies on EMQ and JIT, either by indication or implication, noted that good accounting,
mainly of holding and setup costs, is important to produce realistic results.

The holding and setup costs are aggregated costs that are difficult to estimate (Waters, 2008). To
investigate this, Jaber (2009, Ch.9) suggested using the law of thermodynamics to represent the
hidden costs inherent in inventory systems that result from inefficiencies in the production inventory
systems. Jaber (2009, Ch.9) collected some of these costs from the literature. They arise from
queues of materials waiting to be processed, machines waiting to be serviced, finished products
waiting for delivery to customers, disposal and waste, product returns, packaging, loss, damage,
storage related, and environmental protection. Sauers (1986) was another author who suggested
adding a further cost component to the EMQ cost function in the form of a “customer
dissatisfaction” cost to the EMQ cost. That cost may be caused by factors like stock-outs, product-
quality problems, and excessive manufacturing lead times, among others.

In recent years, a few researchers have called for non-classical thinking of inventory management
(Jaber, 2009, Ch. 9), i.e. to identify a range of problems not covered in the classical analysis of
inventory systems, like the need to design environmentally responsible inventory systems (Bonney
and Jaber, 2011). The interest in addressing environmental issues in inventory management pushed
research towards closing the sustainability circle by considering social performance along with the
traditional economic and the emerging environmental performance. These matters are reviewed in
Andriolo et al. (2014) and Bushuev et al. (2015). The concept of entropy as a measure of
sustainability is not far from this thinking. Industrialization increased the flows of energy and matter

Page 4 of 24
through society, which lead to a progressive depletion of available resources or to an increase in
entropy generation that result in natural disorder (Norde, 1997). However, reducing inefficiencies in
an industrial process reduces entropy and improves its sustainability and indirectly that of society.
This paper continues this theme of comparing the sustainability of the EMQ and JIT models. It
examines the performance of JIT and EMQ systems in order to help the system designers of an
organization to decide whether or not to implement JIT. To provide this guidance, the paper
investigates a range of situations using the models developed for JIT and EMQ in different situations.
It then derives and incorporates entropy cost functions in both models. The next section, Section 2,
defines classical models of the two approaches and identifies additional cost components, which
include transportation cost, rework cost, worker stress cost, energy cost, and emissions costs from
transportation and production, that can be added to the two different system models to make them
more realistic and comparable. The aims of the paper are to: (1) illustrate that EMQ systems can
under some circumstances be more economic than JIT systems, (2) illustrate how the performance
of the systems change for a range of variables related to the environment, transport, staff and other
parameters, (3) investigate the sensitivity of the results to changes in the system parameters, ,and
hence identify when it would be appropriate to use JIT, and (4) illustrate how the results could be
generalized. It also states the assumptions, lists the notations, and develops the mathematics for the
EMQ and JIT models. Section 3 presents, discusses and interprets the numerical results derived from
the models. The last section presents a summary and conclusions.

2. Mathematical Modeling
This section starts by presenting the new cost terms that will be added to the classical inventory
costs, which are the holding and setup costs. The modified mathematical models for EMQ and JIT
that follow this introduce the assumptions and notations when they are used. However, for the
convenience of the reader, a nomenclature list is provided next. The following section is for laying
the background to the modified EMQ and JIT models presented in sections 2.3 and 2.4 respectively.
The table below compares the assumptions used for calculating the various cost terms for EMQ and
JIT.

Page 5 of 24
Assumption EMQ JIT
Defective items  x
Demand is deterministic and constant  
Emissions from production and transportation  
Energy form producing items  
Entropy cost  
Infinite planning horizon  
No shortages allowed  
Rework cost  x
Transportation cost  

Yes (); No (x)

2.1 Nomenclature

d demand (unit/year)
P unit price ($/unit)
P0 unit market price ($/unit)
elasticity parameter (units/year /$)
t time
production rate (unit/year)

total cost ($/year)


Q order quantity for EMQ model (unit/cycle)
q order quantity for JIT model (unit/cycle)
setup/order cost ($/order)
holding cost ($/unit/year)
up unit production cost ($/unit)
binary parameter
transportation cost ($/year)
fixed cost of transportation ($/shipment)
variable cost of transportation ($/unit)
quality cost ($/year)
unit cost to rework a defective item ($/unit)
probability of the process going out of control
cost of emissions from transportation ($/year)
ce cost of a ton of CO2 emissions ($/ton‐CO2)
greenhouse gases emissions (tons‐CO2/trip)
energy costs per cycle ($/year)

Page 6 of 24
energy cost per unit ($/kwh)
energy per unit time when manufacturing is idle (kwh/year)
energy to manufacture one unit (kwh/unit)
cost of emissions from production ($/year)
a, b and c positive parameters of the emissions function
cost of work related stress ($/year)
fraction of the unit production cost
unit entropy cost ($/unit)
entropy generated by time t
Z cost between EMQ and JIT models ($/year)

2.2 Background
The classical EMQ model produces the quantity Q of a single product in a time period of Q/ρ and
consumes it over Q/d, where  (units/year) and d (units/year) are the production and demand rates,
respectively, with  > d. During production, inventory builds up at a rate of and production
ceases once an inventory level of is reached. The process of production, inventory
accumulation and depletion of items repeats itself indefinitely. The EMQ model assumes no
shortages. The total cost per unit of time, is:

(1)

,where A is the setup cost ($), h is the unit holding cost ($/unit/year), up is the unit production cost
(in this paper, it represents the cost of material and labor only, in $/unit). The other cost
components that will be added to Eq. (1) are:

1. Transportation cost: , where f is a fixed cost ($/shipment) and f0 is a


variable cost($/unit).
2. Defective items that are reworked, , where cr is the unit rework cost
($/unit), is the probability of the processes going out of control and start producing
defective items (Porteus, 1986). Note that this approximation provided by Porteus (1986)
works for very small values of ; Khouja (2005) discusses this point. A JIT system allows
operators to interrupt a production process for corrective purposes, thus reducing the
number of defective items to almost zero (Khouja, 2005). This corroborates the empirical

Page 7 of 24
finding that reject rates reduce proportional to the reduction in lot sizes (Inman, 1994) and
the efforts that JIT implementers place on improving quality.
3. Emissions from transportation, , where  (tons-CO2/trip) is the greenhouse
gases emissions per round-trip (for a fixed distance), and ce is the cost of a ton of CO2
emissions ($/ton-CO2).
4. Emissions from production, , where a, b and c are positive
parameters (Jaber et al., 2013). Minimum emissions occur when .

5. Energy costs per cycle where ceng is the unit energy cost

($/kWh), k0 is in kWh/unit, and k is in kWh/year representing costs of power per unit and
idle power, respectively (Gutowski et al., 2006).
6. Work related stress , where is a percentage of the production cost.

Several studies have discussed that workers stress level is much more in a JIT system than it
is in a traditional system (Inman and Brandon, 1992; Schultz et al.,1998; Landsbergis et al.
1999; Godard, 2001; Conti et al., 2006). Since JIT is associated with producing in smaller lot
sizes more frequently than the EMQ system, then it is reasonable therefore to assume that
reducing the lot size, increases the pace of setup and speeds production and put more
pressure on workers to be perfectionists. So, in this regard and without loss of generality, we
hypothesized that as Q approached , the ideal JIT lot size, stress level
increases.. So, as Q approaches q, the unit production cost increases by to account for
worker related health costs.

Jaber et al. (2004) introduced the concept of “entropy cost” into the EOQ model. They postulated
that the flow of a commodity (demand rate) and its price in an inventory system are structurally
similar to the flow of heat and its temperature in a thermodynamic system. That is, heat flows from
a high temperature reservoir to surroundings of lower temperature during which some of the heat
that is transferred is lost due to entropy, known as the Newtonian heat conduction. This is based the
work of Andresen and Gordon (1992) who examined different heating and cooling strategies.
Analogously, a firm can heat up its market by lowering its price below the market price. Jaber et al.
(2004) expressed the demand rate as d(t) to emphasize the time dependence, which can be written
as:

(2)

Page 8 of 24
, where, P0(t) is the market equilibrium price at time t ($/unit) and P(t) is the product price at time t
($/unit); where >  t > 0. The term K is analogous to thermal capacity in thermodynamics
and represents the increase in demand of a product for the reduction of its price by one monetary
unit. The negative sign in Eq. (2) indicates that products flow from the system to the surroundings.
This equation and Eq. (3) below are very similar to Eqs. (1) and (2) in Andresen and Gordon (1992, p.
295). Readers may refer to the work of Jaber et al. (2004, pp. 5168-5173) who provided an in-depth
explanation of the analogy between a production-inventory system and a thermodynamic system.

Jaber et al. (2004), who suggested using the second law of thermodynamics (entropy law) to account
for inefficiencies in a production-inventory system, calculated the total entropy generated by time t
as:

(3)

Jaber (2004; 2009, Ch.9) presented a concise and focused background to the principles and laws of
the thermodynamics, and the development of Eqs. (2) and (3). If we assume for simplicity, and in line
with the EMQ and JIT assumption of constant demand, that P(t) = P and P0(t) = P0 where P < P0,
then Eq. (2) becomes .

2.3 EMQ model


We define as the unit entropy cost generated to produce and sell an item to the
market, where:

(4)

The entropy cost for a lot of size Q is . The modified cost function in Eq.(1)

is . The modified
EMQ cost function is:

Page 9 of 24
(5)

Eq. (5) is convex in Q; i.e. > 0  Q > 0 since all the

parameters are positive. The modified economic manufacture quantity, Q*, that minimizes Eq. (5),
can be obtained by setting its first derivative with respect to Q equal to zero and solving for Q to
get:

(6)

Where  = 0, 1, and  =1 indicates that entropy is considered and =0 that entropy is excluded
when calculating the lot size in Eq. (6) and the cost in Eq. (5). Note that Eq. (5) is convex since its
second derivative is positive for all values of Q > 0. The classical EMQ that minimises the sum of

holding and setup costs is . Since , then we have


, from which we have .

2.4 JIT model

The review papers of Andriolo et al. (2014) and Bushuev et al. (2015) did not report any study that
investigated JIT for environmental or other factors. The JIT norm has been to produce and deliver
items as needed in smaller lot sizes more frequently. Potentially, this could congest supply chains
and traffic and generate more waste and could have a negative environmental impact; because of
e.g. packaging material, CO2 emissions, landfill, etc. (Bonney and Jaber, 2011). For example,
Mollenkopf et al. (2010) showed a conflict between JIT and green strategies, and suggested that
firms that wish to reduce the negative environmental impact of JIT may have to modify their
production and inventory activities to achieve both lean and green objectives. Very recently, Ugarte
et al. (In press) showed that JIT is associated with higher greenhouse gas emissions than the
EOQ/EMQ due to the significant increase in transport frequency.

The JIT norm is repetitive production particularly of consumer durable goods in small lots. The
behavior of the model here is to produces q units in units of time. Note that the cycle time is
q/d where the average inventory per cycle is qd/2. Although some researchers advocate a zero

Page 10 of 24
inventory level (Cao and Schniederjans, 2004), other researchers argue that zero inventory is
impossible theoretically (Zangwill, 1987).

In a JIT system, the setup cost is much less than that of EMQ, AJ << A. For example, Jones (1991)
reported (from a study by Suzaki) that in Toyota, the setup cost was reduced from 4 hours to 3
minutes and in Yanmar Diesel from 9.3 hours to 9 minutes.

Note that the cycle time is q/d; however, in JIT, we assume that production starts at time , so

we assume that entropy generation in this case occurs during the production and delivery of
products (activities being performed), and that the time segment has no entropy generation.

So, it is reasonable to assume that the entropy cost per cycle for the JIT model is proportional to that

of the EMQ model and is computed as . The annual entropy

cost for the JIT model is

As explained earlier, , which is consistent with JIT generating zero

defective units. This was explained mathematically in (Khouja 2005; Jaber, 2006). The annual JIT cost
function is written in a similar manner to Eq. (5) as:

(8)

Eq. (8) is convex in q; i.e. > 0  q > 0 since all the parameters
are positive. The value of q that minimizes Eq. (8) is determined in a similar manner to Eq. (6), and is
given as:

(9)

Where  = 0, 1, and  =1 indicates that entropy is considered and  =0 indicates that entropy is
excluded when calculating the lot size in Eq. (9) and the cost in Eq. (8).

Page 11 of 24
3. Numerical Results
In this section, we compare the performance of the two systems by using the variable Z, which is the
difference between the costs per unit of time of operating the EMQ system compared with
operating the JIT system. The initial comparison is made using only the classical (production, setup
and holding) costs in the defined models. Subsequently, six other categories of cost components are
introduced, namely: (1) Transportation costs, (2) Rework of defective items, (3) Costs of emissions
from transportation, (4) Costs of emissions from production, (5) Energy costs, and (6) Costs arising
from workers stress. The parameters used are taken from Jaber et al. (2004), Jaber et al. (2013) and
Bazan et al. (2015). They are:  =2000 unit/year (produces minimum emissions), P0=$105, P =$100
and K = 200 units/year/$ corresponding to d=1000 unit/year from Eq. (2), up=60 $/unit, h =12
$/unit/year (i.e. 20% of up ), f =$200, f0=1 $/unit, ce = 20 $/unit, cr = 12 $/unit (or 20% of up),  =
0.0005 ,ceng = 0.1 $/kWh, k0 = 100 kWh/unit, and k = 125000 kWh/year, a = 3.00E-7 ton.year2/unit3,
b =12.00E-04 ton.year/unit2 and c =1.4 ton/unit (production rate that minimizes CO2 emissions
occurs at p = b/2a= 2000 units/year),  = 0.1, A = $500/setup, and AJ = $25/setup (5% of A). The
emissions per trip,  = 1 ton.CO2 per trip (Bazan et al., 2015).

First, the analysis compares the EMQ and JIT models using the classical holding, setup and
production costs. Secondly, the analysis adds the environmental (CO2 emissions and energy) and
social (worker stress/fatigue) costs and compares the EMQ and JIT models for the classical and
modified lot size quantities. Thirdly, the analysis investigates the EMQ and JIT models when entropy
is introduced. Finally, the EMQ and JIT models are optimized for price and production rate.

EMQ vs. JIT with classical costs


The EMQ cost function in Eq. (5) that includes just the classical costs (production, setup, and holding

costs), reduces to whose solution is

, where the cost function could be written as

. Similarly, for the JIT model, the lot size and the closed form cost

function are and . The results are =408.25 units, =


$62,449.49, = 91.28 units, and = $60,547.72. The difference in cost is Z = =
$1,902, or a reduction in unit cost of $6 (1901/(408.2591.28)). This suggests that a firm with this
data using EMQ, has to cut its unit production cost by $6 or 10% to be as competitive as a firm using
JIT, which may not be easy for some firms. This reduction is insensitive to changes in AJ. However,
demand plays an important role in maximizing Z. We have optimized for different values of d and
found that the Z is a maximum when d = 781.79 (P =101.09), where Z = $1,962.26 with a unit

Page 12 of 24
reduction of $8.32 (1962.26/(327.04 91.29)), about 39% more than when demand was 1000. The
minimum value of Z occurs when d is close to .

Adding the classical fixed and variable transportation costs (no.1 from the list following Eq. (1)),
alters the EMQ and JIT formulas to become ,

whose solution is , and

, , respectively, with =

483.05 units, = $63,898.28, = 273.86, = $62,643.16, and Z = 1,255.11. Accounting for


the transportation costs reduces the cost difference between the EMQ and JIT models. The EMQ
model is less sensitive to transportation cost than the JIT model. The production quantity for the
EMQ model increased by 18.3% (from 408.25 to 483.05) and the cost by 2.3% (from $62,449.49 to
$63,898.28), whereas for JIT, the production quantity increased by 200% (from 91.29 to 273.86) and
the cost by 3.5% (from $60,547.72 to $62,643.17), so the transportation cost would not be of
concern. This suggests that for JIT systems the suppliers, manufacturers and the market should be
geographically close. If we use = 408.25 and q0 = 91.29, the costs are increased by 2.4% (from
$62,449.49 to $63,939.39) and 5.3% (from $60547.72 to $63738.61), respectively. The changes in
cost are minimal, suggesting that firms may continue using the classical equations without disturbing
their schedules, as that may bring additional costs.

Production processes are never perfect and defective items are inevitable. Porteus (1986) suggested
producing in smaller lots to improve quality, because in a JIT system, the number of items needing
rework is virtually zero. So, after adding the cost component 2, the EMQ formula becomes

, whose solution

is . The JIT cost function and lot size formulas are the
same as before and ; i.e and . Accounting for transportation and
defectives reduced the EMQ quantity by 16.3% (from 408.25 to 341.57) and increased the cost by
4.2% (from $62,449.49 to $65,098.78), where Z = $2,455.61. Using the classical EMQ
quantity, where = 408.25, the cost and difference in cost are = $65,164.13 and Z =
$1,425.52, respectively. The results in this section show the robustness of the EMQ model for
varying the lot size quantity and adding different costs. The EMQ/EOQ literature implies that
transportation and quality costs are factored into the unit production cost. That is, a firm may adjust
its unit production cost for the EMQ model, , say by 4.34% ($65,164.13$62,449.49)/ $62,449.49)
so it becomes = $62.6 and continue using the classical EMQ formula/model.

Page 13 of 24
The JIT model has an additional advantage, which is to free space (Cao and Schniederjans, 2004). For
the EMQ model, the maximum inventory level is , and for the JIT model it is
. If space annually costs $50,000, then using the JIT model will reduce it by about 55% (from
= 204 to = 91), saving $27,500 annually. Table 1 summarizes the results of this sub-section.

Table 1. Summary of results for the EMQ and JIT


i Z
0 408.25 91.28 - - $62,449.49 $60,547.72 $1,901.77
1 - - 483.05 273.86 $63,898.28 $62,643.16 $1,255.12
1 408.25 91.28 - - $63,939.39 $63,738.61 $200.78
2 - - 341.57 273.86 $65,098.78 $62,643.16 $2,455.61
2 408.25 91.28 - - $65,164.13 $63,738.61 $1,425.52
(0) holding, setup and production costs, (1) Scenario 0 + transportation cost, (2) Scenario 1 + quality costs. Note that Scenario “0,0” is “0”.

EMQ vs. JIT with classical, environmental and social costs


New results in Table 2 are now obtained by gradually adding the environmental cost components
(Emissions (3, 4) and Energy (5)) and social component (work stress/fatigue (6)) to and .
With emissions from production and transportation and energy consumption, the EMQ and JIT cost

functions become (from Eqs. (5) and (8))

and

.
The production quantities that minimize and are computed from Eqs. (6) and (8),
respectively, with  = 0 and  = 0, and found to be = 346.41 ( $85,406.92) and =
285.77 ( = $82,964.64). If we assume that the two firms continue using the classical approach to
compute their production quantities, i.e. =408.25 and = 91.28, then the total annual costs for
the EMQ and JIT systems are $85,463.12 and $84,207.70, respectively, where Z = =
$1255.42. We notice that the EMQ model is less sensitive to changes in the lot size than the JIT
model. For example, in Table 2, producing off the optimal point, 408.25 rather than 346.41 units,
results in an increase in the cost for the EMQ model of 0.07%, which is negligible. The JIT model is a
bit more sensitive to changes in q. A breakdown of the costs for each model and scenario is provided
in Table 2.

Page 14 of 24
Table 2. A breakdown of the costs components for the modified and classical approaches

Modified Classical
EMQ JIT EMQ JIT
Production quantity 346.41 285.77 408.25 91.29
Setup Cost 1443.38 87.48 1224.74 273.86
Production cost 60000.00 60000.00 60000.00 60000.00
Holding cost 1039.23 857.32 1224.74 273.86
Transportation cost 1577.35 1699.85 1489.90 3190.89
Quality cost (rework) 1039.23 0.00 1224.74 0.00
Emissions from transport 57.74 69.99 48.99 219.09
Emissions from Production 4000.00 4000.00 4000.00 4000.00
Energy costs 16250.00 16250.00 16250.00 16250.00
Total $85,406.92 $82,964.64 $85,463.12 $84,207.70

Now adding the cost of worker welfare to the examples in Table 2, i.e.,
and , increases the costs to = $86,915.98 ( = 459.66)
and = $88,964.64 ( = 285.77), where Z = = $2048.66 making JIT more
expensive. If the firms use the classical equations to calculate the production quantities, i.e.
=408.25 and = 91.28, then Z = 3402.94. So far, the JIT performed better than the EMQ model
except when workers’ welfare costs are considered. This can be offset by savings from using less
space for example. However, accounting for worker stress, as assumed here, increases the unit
production cost if we factor it in up; i.e., $66/unit. One may argue that the holding cost in a JIT
system is slightly higher; from 12 to 13.2 (10% higher), which reduces = $89048.33 ( =
272.47).

In the previous examples, the production rate that minimizes CO2 emissions is  = 2000, but not
necessarily so for the firm when all costs are considered. The cost functions and are
optimized for the production rate and the breakdown of the costs are summarised in Table 3. The
results show that the JIT model pushes for faster production than the classical model to minimise
costs. That is operating slightly off the optimal point. The EMQ continues to be less sensitive to
changes than the JIT system; (but has insignificantly higher costs than the modified models by about
0 and 1% respectively). However, the classical EMQ and JIT models remain robust and easier to use
by managers than the other models.

Page 15 of 24
Table 3. A breakdown of the costs components for the modified and classical approaches

Modified Classical
EMQ JIT EMQ JIT
Production rate 2132.15 2238.13 2142.31 2256.08
Production quantity 455.87 302.31 395.33 96.96
Setup Cost 1096.81 82.70 1264.77 257.85
Production cost (including worker stress) 61427.54 66000.00 61433.93 66000.00
Holding cost 1452.37 810.43 1264.77 257.85
Transportation cost 1438.72 1661.58 1505.91 3062.80
Quality cost (rework) 1367.60 0.00 1185.99 0.00
Emissions from transport 43.87 66.16 50.59 206.28
Emissions from Production 4104.79 4340.22 4121.52 4393.47
Energy costs 15862.61 15585.03 15834.82 15540.57
Total $86,794.32 $88,546.12 $86,662.29 $89,718.83

EMQ vs. JIT with entropy cost


This paper has proposed using entropy as a measure of sustainability. The examples in Table 3 are
replicated by setting  = 1 in Eqs. (5), (6), (8) and (9). The results in Table 4 show that by accounting
for entropy, i.e. using the modified models, the production quantities for the EMQ and JIT models
increase significantly, with the JIT higher than the EMQ quantity. If, for simplicity, a manager uses
the classical equations, the entropy costs for both models increase as more cycles occur in a year
and generate entropy more frequently. The entropy costs for the EMQ and JIT models are 5143.93
and 11502.17, respectively, making the JIT policy more expensive, Z = $9761.19. The classical EMQ
and JIT models were optimized for the production rate. The results are summarized in Table 4. JIT
favors a higher production rate than the EMQ model, where the cost of the JIT model is reduced by
3.3% (from 101,709.8 to 98,317.93), while the cost of the EMQ remains almost the same (reduction
of 0.02%).

Page 16 of 24
Table 4. A breakdown of the costs components for the modified and classical approaches when
entropy is considered.

Optimum
Minimum emissions production rate
EMQ JIT EMQ JIT
Production rate 2000.00 2000.00 2056.97 2579.27
Production quantity 408.25 91.29 402.71 103.67
Setup Cost 1224.74 273.86 1241.59 241.16
Production cost (including worker stress) 61341.64 66000.00 61379.33 66000.00
Holding cost 1224.74 273.86 1241.59 241.16
Transportation cost 1489.90 3190.89 1496.64 2929.24
Quality cost (rework) 1224.74 0.00 1208.13 0.00
Emissions from transport 48.99 219.09 49.66 192.92
Emissions from Production 4000.00 4000.00 4019.48 6013.34
Energy costs 16250.00 16250.00 16076.89 14846.33
Entropy Cost 5143.93 11502.17 5214.68 7853.78
Total $91,948.69 $101,709.88 $91,927.98 $98,317.93

Sensitivity analysis
We will consider the classical models since our results so far have shown that they are robust, and
because simple formulas are more appealing to managers.

In Table 2, which does not account for the costs of worker stress, = 0, and entropy,  = 0, we
notice that the costs of energy and emissions from production are the same for the EMQ and JIT
systems for the modified and classical models. Table 2 shows that the JIT is more economical than
the EMQ model because the savings from holding and setup costs are more than the additional cost
that JIT incurs from transportation. Also, the EMQ model is disadvantaged as it incurs quality costs
while the JIT does not. The savings are $1255.42 ($85,463.1$84,207.70) which could be calculated
by subtracting Eq. (5) from Eq. (8), to get

Page 17 of 24
Since  >> Aj and d < , increasing/decreasing f, , h, cr will increase/decrease the value of  with
the JIT system always cheaper than the EMQ. Since the behaviour of the models is predictable for f,
, h, cr, they will be left out. The focus will be on the entropy cost and the cost of worker stress
per unit as they significantly increase the cost of the JIT system, making the EMQ a cheaper
alternative.

Table 4 is replicated for different values of P and P0, i.e., PP0 = 5 which keeps demand constant at
1000, while optimizing for the production rate, . Table 5 summarizes the
results showing only the prices (P and P0), the production rates (), the production quantities (
and ), total costs and entropy costs (E.C.) for the EMQ and JIT models. The results show that as P
and P0 decrease Z decrease. For example, when P = 100 and P0=105, Z = 6389.95 (91,927.98
98317.93), reducing their values to P = 70 and P0 = 75, Z =4922.93 (89,307.26 94,230.19 ). The
reduction in Z mainly comes from lower entropy cost. The results in Table 5 show that JIT favors
faster production rates than the EMQ model. So, the JIT has lower inventory cost and energy cost
than the EMQ model but higher production emissions costs. We replicated the numerical examples
in Table 5 while ignoring worker stress, for the same instances of P and P0 noted above, Z reduced
from 1785.61 (90532.32  92317.93 ) to 346.08 (87884.1188230.19). The results may suggest
that inexpensive items are cheaper to control since their flow generate less entropy. They also
suggest that reducing the cost of stress and bringing other improvements to bring down the costs
may make a JIT system more economical.

Table 5. The production policies for the classical models for decreasing prices (P and P0)

EMQ JIT
P P0  Q0 Cost E.C.  q0 Cost E. C.
100 105 2056.98 402.71 91,927.98 5214.68 2579.27 103.67 98,317.93 7853.78
95 100 2065.84 401.89 91,430.84 4727.62 2554.90 103.18 97,564.64 7207.73
90 95 20674.01 401.15 90,957.64 4262.79 2530.89 102.69 96,838.78 6579.54
85 90 2081.78 400.46 90,508.53 3820.63 2507.26 102.21 96,141.34 5970.34
80 85 2088.92 399.83 90,083.68 3401.47 2484.11 101.74 95,473.37 5381.32
75 80 2095.53 399.25 89,683.22 3005.64 2461.50 101.27 94,835.94 4813.78
70 75 2101.64 398.72 89,307.26 2633.43 2439.52 100.82 94,230.19 4269.12

The numerical examples in Table 5 are replicated for an increasing demand, i.e. P reduces from 100
to 70 while P0 = 105. The results show that for higher demand rates, the JIT becomes more
economical. For example, when d = 1000, Z = 6389.95 (91,927.98  98,317.93) that becomes a
savings in favor of the JIT model as P approaches 70, where Z = 114,814.48 (1,764,408.40
1,649,593.92). This suggests that JIT is more suitable for items with high demand. Note that the

Page 18 of 24
bulk of the costs (about 65%) for this case are from production emissions since the production rates
are much higher than the optimal,  = 2000 units/year.

Table 6. The production policies for the classical models for decreasing commodity price, P, while
P0 = 105).

EMQ JIT
P d  Q0 Cost E.C.  q0 Cost E. C.
100 1000 2056.97 402.71 91,927.98 5214.68 2579.78 103.68 98,317.93 7851.47
95 2000 2299.52 1131.2 173,694.53 1763.65 2567.09 103.42 195,847.59 15028.54
90 3000 3183.20 2084.2 284,395.23 906.82 3001.01 111.82 297,671.95 16896.14
85 4000 4156.0 2980.2 462,647.13 599.0 4001.01 129.12 454,779.73 13821.32
80 5000 5143.7 3862.2 744,716.90 435.0 5001.01 144.35 708,667.29 11635.86
75 6000 6136.6 4739.2 1,166,634.98 332.3 6001.01 158.13 1,094,992.39 9958.66
70 7000 7132.0 5614.1 1,764,408.40 261.8 7001.01 170.79 1,649,593.92 8605.58

For a firm that wishes to use the classical JIT system, the costs arise mainly from the inefficiencies in
the system and the additional management control costs compared with EMQ systems to control a
smooth flow. A firm may focus on continuous improvement efforts to reduce entropy costs, for
example, by using better operational methods to reduce machining times and reducing the
associated inefficient energy consumption and the emissions generated (Mouzon et al. 2007).
Better workers’ training and rest schedules reduce stress and fatigue and thus improve their
productivity (Givi et al., 2015). This reduces the production cost, as part of it is labor. The reduction
in unit production cost and subsequently the selling price, reduced the system cost. It has also been
documented in the literature that unit production cost and inefficiencies in a system can be reduced
by investment that improves learning (Jaber, 2011). Firms can also consider reducing emissions from
transportation by using green vehicles (Golicic et al., 2010; Gurtu et al., 2015).

4. Summary and Conclusions


The examples shown above apply to situations when one could use either the EMQ or JIT production
systems to produce the products. Typically this could be when production is sufficiently stable and in
sufficient volume to justify investment in JIT but not sufficient to justify dedicated lines; in other
words, there is still the need to setup and probably use the line for (a range of) similar products; the
products are not the typical dedicated JIT product lines (e.g. to make cars and domestic products)
that possibly use conveyors, robots, etc. neither are they short product life or intermittent irregular
production products that commonly produces using EMQ or other systems. In this situation, an
obvious general conclusion is that improvements should be made to whatever system is being used.

Page 19 of 24
The paper has illustrated some of the ways in which one can extend the classical models of EMQ and
JIT by including other parameters such as transportation, defective items that need rework, work
related stress/fatigue, energy and greenhouse-gases emissions costs from transportation and
production systems. Demand was modelled similarly to heat flow in thermodynamic systems using
price as an analogue to temperature. The analysis also used models that estimate the entropy
(disorder) cost using the second law of thermodynamics. Entropy is commonly used as a
sustainability measure for mechanical systems but it is not usually used for analyzing inventory
systems. The results of the analysis showed that JIT, a lean practice, which produces products in
small lots more frequently, incurs lower costs than the EMQ model when stress related and entropy
costs were not considered. However, the inclusion of these stress related and entropy costs can in
some circumstances make JIT more expensive than EMQ systems. Although a JIT policy reduces
some costs, e.g. because there are almost no defective items to rework, smaller batches reduce
inventory holding costs and investment reduces setup costs, the cost of controlling the flow of items
from the manufacturing system to the market is more expensive than the EMQ policy. In addition,
the investment required in training, in factory layout, in quality, in setup reduction, can be
considerable (e.g., Wemmerlöv and Hyer, 1989).

An implication of the paper is that an obvious way to improve the performance of JIT systems is by
reducing the inefficiencies that arise from wasted energy, excess emissions, poor worker training,
and heavy workloads. Of course, reducing these same inefficiencies also would improve the
performance of EMQ systems. In addition, reducing setup cost similar to JIT systems could further
improve EMQ systems. The many variables analyzed, meant that at this stage the emphasis focused
on finding the possible implications of the variables rather than on producing an exhaustive and
complete analysis. Indeed, more exciting than the direct comparison of a range of situations for the
specific systems, is that the investigation shows how an analytical approach, subject to the
availability of representative data, could be used to select an appropriate system (EMQ or JIT) to
produce specific products that takes account of demand and its variability and could also include
other variables e.g. investment appraisal along with the parameters that this paper examined.

References

Andriolo, A., Battini, D., Grubbström, R. W., Persona, A., Sgarbossa, F. (2014). A century of evolution
from Harris‫ ׳‬s basic lot size model: Survey and research agenda. International Journal of Production
Economics, 155, 16-38.

Page 20 of 24
Andresen, B., Gordon, J. M. (1992). Optimal paths for minimizing entropy generation in a common
class of finite-time heating and cooling processes. International Journal of Heat and Fluid Flow, 13(3),
294-299.

Bazan, E., Jaber, M. Y., Zanoni, S. (2015). Supply chain models with greenhouse gases emissions,
energy usage and different coordination decisions. Applied Mathematical Modelling, 39(17), 5131-
5151.

Bonney, M., Jaber, M. Y. (2011). Environmentally responsible inventory models: Non-classical


models for a non-classical era. International Journal of Production Economics, 133(1), 43-53.

Bushuev, M. A., Guiffrida, A., Jaber, M. Y., Khan, M. (2015). A review of inventory lot sizing review
papers. Management Research Review, 38(3), 283-298.

Cao, Q., Schniederjans, M. J. (2004). A revised EMQ/JIT production-run model: An examination of


inventory and production costs. International Journal of Production Economics, 87(1), 83-95.

Conti, R., Angelis, J., Cooper, C., Faragher, B., Gill, C. (2006). The effects of lean production on worker
job stress. International Journal of Operations & Production Management, 26(9), 1013-1038.

Cavinato, J. (1991). Logistics tools: Lowering set-up costs. Distribution, 90(6), 52-53.
Chyr, F., Lin, T. M., Ho, C. F. (1990). Comparison between just-in-time and EOQ system. Engineering
Costs and Production Economics, 18(3), 233-240.

Crawford, K. M., Blackstone Jr, J. H., Cox, J. F. (1988). A study of JIT implementation and operating
problems. International Journal of Production Research, 26(9), 1561-1568.

Drake, M. J., & Marley, K. A. (2014). A Century of the EOQ. In Handbook of EOQ Inventory Problems
(pp. 3-22). Springer US.

Fulbright, J. E. (1979). Advantages and disadvantages of the EOQ model. Journal of Purchasing and
Materials Management, 15, 8-10.

Givi, Z. S., Jaber, M. Y., Neumann, W. P. (2015). Production planning in DRC systems considering
worker performance. Computers & Industrial Engineering, 87, 317-327.

Glock, C. H., Grosse, E. H., Ries, J. M. (2014). The lot sizing problem: A tertiary study. International
Journal of Production Economics, 155, 39-51.

Godard, J. 2001. High performance and the transformation of work? The implications of alternative
work practices for the experience and outcomes of work. Industrial & Labor Relations Review, 54(4),
pp 776-805.

Golhar, D. Y., Stamm, C. L. (1991). The just-in-time philosophy: a literature review. The International
Journal of Production Research, 29(4), 657-676.

Golicic, S., Boerstler, C., Ellram, L. (2010). 'Greening'the Transportation in Your Supply Chain. MIT
Sloan Management Review, 51(2), 47.

Page 21 of 24
Goyal, S. K., Deshmukh, S. G. (1992). A critique of the literature on just-in-time manufacturing.
International Journal of Operations & Production Management, 12(1), 18-28.

Gurtu, A., Jaber, M. Y., Searcy, C. (2015). Impact of fuel price and emissions on inventory policies.
Applied Mathematical Modelling, 39(3), 1202-1216.

Gutowski, T., Dahmus, J., Thiriez, A. 2006. Electrical energy requirements for manufacturing
processes, 13th CIRP International Conference of Life Cycle Engineering, Lueven, Belgium, May 31 –
June 2, pp. 623–628.

Harris, F.W., 1913. How many parts to make at once, Factory. The Magazine of Management, 10
(135–136), 152.

Inman, R. A. (1994). The impact of lot-size reduction on quality. Production and Inventory
Management Journal, 35(1), 5.

Inman, R. A., Brandon, L. D. (1992). An undesirable effect of JIT. Production and Inventory
Management Journal, 33(1), 55-58.

Jaber, M. Y. (2006). Lot sizing for an imperfect production process with quality corrective
interruptions and improvements, and reduction in setups. Computers & Industrial Engineering,
51(4), 781-790.

Jaber, M. Y., ed. (2009). Inventory management: non-classical views. CRC Press, FL: Boca Raton.

Jaber, M.Y. (Ed.) (2011). Learning Curves: Theory, Models, and Applications, CRC Press: Boca Raton.

Jaber, M.Y., Glock, C.H., El Saadany, A.M.A. 2013. Supply chain coordination with emissions
reduction incentives, International Journal of Production Research, 51(1), 69-82.

Jaber, M.Y., Nuwayhid, R.Y., Rosen, M.A. 2004. Price-driven economic order systems from a
thermodynamic point of view, International Journal of Production Research, 42(24), pp 5167-5184.

Jones, D. J. (1991). JIT & the EOQ model: Odd couple no more! Strategic Finance, 72(8), 54.

Khouja, M. (2005). The use of minor setups within production cycles to improve product quality and
yield. International Transactions in Operational Research, 12(4), 403–416.

Koukoulaki, T. (2014). The impact of lean production on musculoskeletal and psychosocial risks: An
examination of sociotechnical trends over 20 years. Applied Ergonomics, 45(2), 198-212.

Landsbergis, P. A., Cahill, J., Schnall, P. (1999). The impact of lean production and related new
systems of work organization on worker health. Journal of Occupational Health Psychology, 4(2),
108.

Mehra, S., Inman, R. A. (1992). Determining the critical elements of just‐in‐time implementation.
Decision Sciences, 23(1), 160-174.

Moily, J. P. (2015). Economic Manufacturing Quantity and Its Integrating Implications. Production
and Operations Management, 24(11), 1696-1705.

Page 22 of 24
Mollenkopf, D., Stolze, H., Tate, W., Ueltschy, M. 2010. Green, lean, and global supply chains,
International Journal of Physical Distribution & Logistics Management, 40(1/2), pp 14-41.

Mouzon, G., Yildirim, M. B., Twomey, J. (2007). Operational methods for minimization of energy
consumption of manufacturing equipment. International Journal of Production Research, 45(18-19),
4247-4271.

Norde, W. (1997). Energy and entropy: a thermodynamic approach to sustainability.


Environmentalist, 17(1), 57-62.

Osteryoung, J.S., Nosari, E., McCarty, D.E., Reinhart, W.J. (1986). Use of the EOQ model for inventory
analysis. Production and Inventory Management, 27(3), 39-46.

Porteus, E. L. (1986). Optimal lot sizing, process quality improvement and setup cost reduction.
Operations Research, 34(1), 137-144.

Sauers, D. G. (1986). Analyzing inventory systems. Management Accounting, 67(11), 30-36.

Selen, W.J., Wood, W.R. (1987). Inventory cost definition in an EOQ model application. Production
and Inventory Management Journal, 28(4), 44-47.

Schultz, K. L., Juran, D. C., Boudreau, J. W., McClain, J. O., Thomas, L. J. (1998). Modeling and worker
motivation in JIT production systems. Management Science, 44(12-part-1), 1595-1607.

Sohal, A. S., Keller, A. Z., Fouad, R. H. (1989). A review of literature relating to JIT. International
Journal of Operations & Production Management, 9(3), 15-25.

Ugarte, G. M., Golden, J. S., Dooley, K. J. (In press). Lean versus green: The impact of lean logistics on
greenhouse gas emissions in consumer goods supply chains. Journal of Purchasing and Supply
Management, DOI: 10.1016/j.pursup.2015.09.002

Weiss, E.N., (1990). Lot sizing is dead: Long live lot sizing. Production and Inventory Management
Journal, 31(1), 76-79.

Wemmerlöv, U., Hyer, N. L. (1989). Cellular manufacturing in the US industry: a survey of users. The
International Journal of Production Research, 27(9), 1511-1530.

Wijewardena, H., De Zoysa, A. (1999). A comparative analysis of management accounting practices


in Australia and Japan: an empirical investigation, The International Journal of Accounting, 34(1), 49-
70.

Woolsey, A. (1990). Requiem for the EOQ: An editorial. Hospital Material Management Quarterly,
12(1) 82-90.

Zangwill, W. I. (1987). From EOQ towards ZI. Management Science, 33(10), 1209-1223.

Page 23 of 24
Highlights

 Modifies EMQ to include some operational, environmental and social costs.

 Compares EMQ and JIT models using the second law of thermodynamics.

 JIT is more expensive when worker stress and entropy costs are considered.

Page 24 of 24

You might also like