You are on page 1of 7

Int. J.

Production Economics 140 (2012) 579–585

Contents lists available at SciVerse ScienceDirect

Int. J. Production Economics


journal homepage: www.elsevier.com/locate/ijpe

Optimal budget planning for investment in safety measures


of a chemical company
Yuji Sato n
Graduate School of Policy Science, Mie Chukyo University, 1846 Kubo, Matsusaka, Mie 515-8511, Japan

a r t i c l e i n f o abstract

Article history: The objective of this study was to develop a framework for optimizing budget planning for investment
Received 14 April 2010 in safety measures in a chemical company. Decision making for strategic investment for safety is
Accepted 10 May 2012 complicated due to the intangible factors that enter into both the evaluation and choice of appropriate
Available online 15 June 2012
safety measures, and the individual situation of plants. This study addressed these issues by combining
Keywords: Linear Programing and Analytic Hierarchy Process, showing how to quantify inherent risks within
Budget planning chemical plants and how to optimize budget planning for safety within a chemical company.
Investment for safety & 2012 Elsevier B.V. All rights reserved.
Quantification
Intangible factor
LP
AHP

1. Introduction Given the need for necessary and sufficient risk management
within chemical plants, a variety of approaches have been
With growing interest in global environmental issues, chemi- proposed. However, the number of studies on budget allocation
cal companies need to take responsibility for reducing plant- for safety measures within chemical companies is limited because
based risks such as fires, explosions or leakages, given their of inherent plant-based risks, which results in a lack of a
potentially devastating human and environmental consequences. ‘‘standard’’ framework of optimization for the allocation. Chemi-
In addition to taking responsibility, companies need to adopt cal companies surrounded by uncertainties, such as unexpected
strategic investments for risk management, which is inseparable losses from disasters and envisaged economic effects of safety
from profit generation, due to the increasing focus on account- measures, have therefore tried to optimize budgeting as the case
ability to stakeholders. may be by, for instance, consulting professional analysts for
Decision making for strategic investment for safety is compli- safety. While from the perspective of the uncertainty of the
cated, particularly in chemical companies, where the introduction of results of the investment, such as potential benefits or unex-
plant safety measures is critical to the success of risk management. pected costs of early stage manufacturing technology, budget
In optimizing investment in safety measures, safety managers must planning for investment in safety measures is similar to that for
evaluate and choose safety measures when making decisions about new technologies in manufacturing (Tan et al., 2011), which
strategic investment, which are usually costly and surrounded by serves as a useful reference for this issue.
uncertainty over the likely effects of the different safety measures As Beaumont (1999) ascertained, the criteria that firms use to
available. The difficulties arise mainly from intangible factors, such make investment decisions in manufacturing technology are:
as the safety manager’s judgment of criteria, that enter into the how firms manage the introduction of new technology, whether
evaluation and choice of appropriate safety measures, particularly firms experience unanticipated effects from new technology, and
given the rapidly changing technological environment. Thus, deci- what factors impede or assist its implementation. Some of these
sion-making regarding budget allocation, relying heavily as it does factors can only be clarified once the investment is implemented.
on experience, knowledge, as well as intuition, means that the Although decision makers for investments are not completely
evaluation and choice of appropriate safety measures often lacks ignorant of what the future might be, investment decisions are
transparency and traceability. made under conditions of uncertainty. Frank (1998) considered
the nature and acceptable level of risk, together with manage-
ment’s personal attitude to risk. O’Brien and Smith (1993) also
noted that investment in advanced manufacturing systems must
n
Tel.: þ81 598 29 1122; fax: þ81 598 29 1014. be made while taking into account factors that are difficult to
E-mail address: ysatoh@1988.jukuin.keio.ac.jp predict. In the paper, they discussed how a decision process might

0925-5273/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.ijpe.2012.05.030
580 Y. Sato / Int. J. Production Economics 140 (2012) 579–585

be designed and managed, and proposed the application of the environmental protection issues as well as cost minimization.
Analytic Hierarchy Process (AHP). Most previous studies, includ- Although many approaches to budget allocation in risk manage-
ing Frank (1998) and O’Brien and Smith (1993), however, focused ment have been previously proposed, most have neglected
solely on the efficacy and efficiency of each safety measure in intangible yet critical factors within the allocation or just focused
conjunction with cost minimization, without taking into account on cost minimization in risk management. Global environmental
the sustainability of the company. protection, not just straightforward actions such as CO2 controls
On the other hand, quantitative definitions of risk have been but also implicit support for preserving the environment, must be
established (Kaplan and Garrick, 1981; Apostolakis, 1990), and taken into account. Seen in this light, the approach proposed in
Probabilistic Safety (Risk) Assessment methods have been heavily this paper is critical to the sustainability of the company.
used in risk analyses in the nuclear and aerospace sectors, as well In order to optimize a budget allocation for risk management,
as in chemical companies (Garrick and Christie, 2002; Davies, chemical companies must consider not only necessary and
2002; Labeaua et al., 2000). Based on the solid quantitative model sufficient expenditures but also their social responsibility so as
of risk assessment, the application of these methods to the to avoid accidents that might seriously damage the environment.
optimization of budget planning for investment in safety mea- That is, optimizing budget planning for investment in safety
sures might be an option. In the application, however, the measures entails not only minimizing costs for risk management
assumptions of modeling are often hard coded, which means that but also integrating the economic, legal and social engineering
not only the validity of the model but also modeling flexibility perspectives within the framework. In this paper, LP is applied to
and modularity are harmed (van der Zee.,2004). In addition, how settle the issue of the optimization. Linear functions describing
to quantify intangible factors in the evaluation and choice of the relationship between expenditures for safety measures and
appropriate measures is a critical issue in the application. the achievement record of each safety measure in comparison
This paper aims to address these issues. Dealing with intangi- with the initial estimate each year is formulated. Then by solving
ble factors in the evaluation and choice of appropriate safety the LP, the budget planning for safety within a chemical company
measures, a framework for optimizing budget planning for is optimized.
investment in safety measures is proposed. Unlike previous The rationale for the linear approximation is the difficulty in
studies, the framework is based on the actual situation of a identifying a function representing the relationship between the
chemical company, and the optimization process proposed in this expenditures and the achievement record. The relationship might
paper considers not only cost minimization but also global be a non-linear function (most likely, a concave function), and
environmental protection, which offers quantitative justification even a non-linear problem could be solved mathematically;
for the budgeting for safety measures. however, it is not easy to precisely identify a function. On the
The optimization of budget planning is accomplished by combin- other hand, every function can be represented as a linear function
ing two methods of operational research. One is a traditional by piecewise linear approximation, which implies the segmenta-
technique for optimization, Linear Programming (LP), which is used tion of risk reduction plans (processes) in the optimization of the
to optimize the planning for investment in safety measures in terms budget planning. In addition, dealing with each plan (process)
of the criteria of global environmental protection and cost mini- based on many years of experience is likely to be more familiar to
mization. The second method is an innovative decision support practitioners than precisely identifying the function with abstruse
method, the AHP, which is able to incorporate quantitative and mathematics. Thus, the linear approximation of the relationship
qualitative judgments into evaluations and can then be used in and the use of the function in optimizing budget planning in a real
coming up with the parameters employed in the LP. case could be justified. These approaches would be easily under-
Section 2 describes what the problem is in a chemical stood by practitioners, thus enhancing the applicability of the
company, how to solve the problem, and what the solution for approaches.
the problem would be. In Section 3, an LP model optimizing On the other hand, the LP has budget constraints as well as
budget allocation as well as the application of the AHP deriving non-negative conditions, which imposes a ceiling on the expen-
parameters employed in the model are shown. Section 4 intro- ditures for safety measures. In the event of accidents, companies
duces a case study verifying the approach proposed in this paper, would fully expense what they must cover for restoration to the
and Section 5 concludes this study and proposes future research. original state in order to assume responsibility for the damage
they had caused. The budget ‘‘plan’’ for risk reduction, however,
includes preventive measures for safety as well as countermea-
2. Problem description sures based on the estimation as to how much they should
expense for future reliability, which are not necessary expenses
The goal of this section is to develop a framework for for a particular budgeting. Furthermore, companies need to invest
optimizing budget planning for investment in safety measures not only for safety but also for items such as R&D and welfare for
in a chemical company. In developing the framework, an addi- employees; they must ‘‘balance’’ their budget allocation for each
tional review of past accidents within the plant of the company field they are responsible for. Thus, optimization of the allocation
should be undertaken by building on the chemical company’s would have upper-limit constraints.
existing accident response manual. In order to support the In the framework, the achievement record, called ‘‘percentage
process, interviews with the organization maintaining statistics complete,’’ is employed as one of the parameters in this paper.
in this field, such as the Ministry of Economy, Trade and Industry The percentage complete is an original measure predicting the
in Japan, would be an option. A focus group with safety managers state of risk in the chemical company, which has actually been
in the plant is then held to capture the depth and richness of risk employed for many years in the company surveyed in this study.
within the plant. In this process, the issue of risks within the plant The concept of the percentage complete is drawn from the idea of
would be discussed with the professional analyst for safety. ‘‘plan-do-see cycle’’ management. During the ‘‘plan’’ phase, inves-
Therefore, the framework proposed in this paper is based on a tigation of the plant based on the actual number of incidents of
safety and reliability study of the chemical plant carried out by past years is carried out by safety managers and professional
safety managers. analysts for safety, and a risk reduction plan as well as
The framework reflects intangible factors in the evaluation and budget allocation for investment in safety is designed. In this
choice of appropriate safety measures, while it considers global phase, what the actual number of incidents during the target
Y. Sato / Int. J. Production Economics 140 (2012) 579–585 581

period will be is unknown; therefore, predicting index of the state fact that the model studied in this paper considers the predicted
of risk following the ‘‘do’’ phase is necessary. After the phase, the risk level based on the evaluation of each safety measure and its
company will confirm the actual state of risk and will be able to expected effect on risk reduction. In (1), the weights of each
evaluate how each safety measure will have functioned. The dimension and safety measure, di and mij, are quantified through
company will then design a budget allocation for the subsequent discussion among those in the focus group within the plant,
year at the ‘‘see’’ phase. On the other hand, the percentage reflecting the consensus of safety managers. The percentage
complete used to be estimated based on safety managers’ intui- complete pij, on the other hand, is estimated by linear function
tions or many years of experience in the company. The estimation based on the given resources.
process is thus a black box process, which is not traceable nor Let Imij denote the expenditure for a safety measure mij.
leads to systematic optimization of budget planning. Therefore, Suppose the percentage complete pij can be approximated by a
the company needs to clarify the estimation process in a linear function of the budget amount for each safety measure
P
transparent way. i,j LðImij Þ, then the problem of ‘‘maximizing the degree of risk
reduction’’ with budget constraint for each safety measure would
be formulated as the linear problem shown below.
3. Optimization of budget planning nX X o
Max i j
wdi
wm ij
LðI mij
Þ , ð2Þ
3.1. LP model
X
The optimization of the budget planning for investment in s:t I
j mij
r Di ðbudget constraint of dimension di Þ, ð3Þ
safety measures is carried out using the following three steps to
evaluate the efficacy of safety measures and formulate LP, which 0 r Imij ð4Þ
optimizes the planning for safety within a chemical company.
The objective function (2) of this LP consists of the degree of
Step 1. Propose risk reduction scheme within a chemical plant in a risk reduction, which is the product of the weight of each
sequence of transparent decisions dimension wdi , that of each safety measure wmij and the linear
An evaluation of the efficacy of safety measures for the approximation of the percentage complete of each safety measure
chemical plant is carried out based on the safety and reliability LðImij Þ. The linear constraints (3) and (4) are the budget constraint
study, and safety managers then propose a risk reduction plan for for each dimension and non-negative condition. This LP has, by
the chemical plant. This step clarifies the scheme of risk reduction nature, a convex feasible region, and thus derives optimal solu-
in a transparent way. tions, which in turn optimizes the budget planning for investment
in safety measures for the chemical plant.
Step 2. Describe the relationship between given budget amounts for
safety measures and the total degree of risk reduction within the
3.2. Quantification of intangible factors
plant as a linear function
The relationship between expenditures and the degree of risk
The budget optimization model clarifying how to allocate
reduction for each safety measure is clarified. In this process, the
limited resources (e.g., people, goods and capital) to precaution-
degree of importance of each safety measure is quantified by the
ary safety measures includes the intangible factors, wD and wM, as
AHP, and the percentage complete of each safety measure is
the parameters of the LP. The LP, therefore, needs to quantify
approximated by the linear function, which calculates the degree
these factors in implementing optimization.
of risk reduction. This step enables systematic optimization of the
One of the tools for analyzing the decision-making process for
budget planning.
supporting multi-criteria decision making is the Analytic Hier-
archy Process (AHP), developed by Saaty (1980). The AHP takes
Step 3. Optimize the budget planning for investment in safety the subjective judgment of each decision-maker as input and the
measures quantified weight of each alternative as output. Therefore, not
An LP model optimizing budget planning for investment in only objective issues but also more subjective issues without
safety measures is formulated. The LP consists of an objective theoretical values can be easily quantified. The AHP has thus been
function maximizing the degree of risk reduction with given widely used for decision-making (e.g., economic problems, policy
resources (i.e., cost minimizing of budget amounts for safety evaluation and urban planning) (Vargas, 1990; Saaty, 1994).
measures with allowable risk level) and of constraints that The advantage of the application of the AHP is to be able to
include not only budget constraints but also the constraint of make the evaluation and choice of appropriate safety measures
global environmental protection. transparent and traceable, even though the safety manager’s
judgment relies heavily on not only experience and knowledge
Let wD ¼{wdi } and wM ¼{wmij }, respectively denote the weight but also intuition. The quantification of wD and wM is undertaken
of dimensions and that of safety measures for a risk reduction using questionnaires consisting of the AHP-formatted questions.
plan elicited by safety managers, where diAD and mijAM show a Omitting the details of the procedure here, safety managers were
set of dimensions and safety measures, respectively. Further, let required to answer a series of questions formatted by the AHP to
P¼{pij} denote the percentage complete of each safety measure. derive the weights for the dimensions and safety measures of the
Then the degree of risk reduction can be defined as an index risk reduction plan, since pairwise comparison functions better
representing how high the risk level of the plant would be one than traditional questionnaires in quantifying human perception
year later: (Sato, 2004). They conducted pairwise comparisons of all possible
XX
Degree of risk reduction ¼ w w p ð1Þ combinations of dimensions, such as ‘‘Which safety measure do
i j di mij ij
you think is more important for the risk reduction of your plant,
As shown in (1), the degree of risk reduction is calculated as a measure A or measure B?’’ This process facilitates clarification of
ratio of risk level, which is not based on the number of past the priority of the different dimensions and competing safety
accidents within the plant. The rationale for the definition is the measures of the risk reduction plan.
582 Y. Sato / Int. J. Production Economics 140 (2012) 579–585

4. Case of a Japanese chemical company To reduce risks, the company undertook case studies for share-
holders, and consulted a professional analyst for the safety of
This section introduces the procedure and the result of the chemical plants. A focus group with safety managers in the plant
case study verifying the approach proposed in this paper. The case classified risks within the plant based on three dimensions as follows:
study was carried out at a Japanese chemical company (whose Equipment (hazardous object facilities, poisonous object facilities,
name cannot be disclosed due to a confidentiality agreement). utility facilities, construction); Human (education and training, unsafe
actions, security); Regulation and others (compliance, design review,
4.1. Risk reduction scheme of a chemical company inspection system). Based on the classification, concrete safety
measures for risk reduction consisting of company-wide projects
Although the chemical company has tried to reduce the and activities within the plant were proposed, and a risk reduction
number of accidents, the current safety status of the plant is not scheme was developed as illustrated in Fig. 2.
satisfactory. Each division causes some type of accident, and Within the chemical plant, the safety managers were first
facility error has become a major cause of accidents. What is required to evaluate the degree of importance of dimensions
worse is that the number of accidents increased 80% from 2007 to wD ¼{wdi } and that of safety measures wM ¼ {wmij } (i¼1, y, 3;
2008, after they had succeeded in reducing the number of j¼1, y, 7) for risk reduction. Table 1 summarizes the results
accidents in 2007. The bar chart in Fig. 1 shows the changes in obtained from the questionnaire formatted by the AHP. The
the number of accidents since 2000. These results require that the numbers in the wD and wM columns represent the weights for
company take preventive action immediately. safety measures normalized by the l1-norm within each dimen-
sion. In terms of the importance of dimensions, some managers
emphasized the importance of Equipment, and others that of
Human. In the aggregate, the safety managers weighted Human
16 90
Leakage (wd2 ) most, 0.390. For the safety measures, tank preservation
14 Fire 80
(wm13 ) ranked highest, 0.201, among the safety measures for
number of observation

12 Total 70 Equipment; OHSMS (Occupational Health and Safety Manage-


cumulative total

60 ment System; wm25 ) ranked highest, 0.303, among those for


10
50 Human; and fire code observance (wm32 ) ranked highest, 0.329,
8
40 among those for Regulation.
6
30
4 20
2 10 4.2. Linear approximation of the percentage complete (example)
0 0
2000 2001 2002 2003 2004 2005 2006 2007 2008 year In this section, the relationship between expenditures for
safety measures and the degree of risk reduction within the plant
Fig. 1. Changes in the number of accidents within the company. is clarified. In the process, how to approximate the percentage

Dimensions Measures

Projects Activities

Hazardous object facility Emergency inspection


Electricity intentional preservation

Poisonous object facility Static electricity measure


Equipment Environmental risk hedge activity
Utility facility Tank preservation

Incinerator abolition
Constructions FMEA
Risks within the plant

Education, training Superintendent arousal Natural calamity measure

Human Unsafe action License institution

Security Security intensification


OHSMS

Design review Equipment Management System


Zero-emission activity

Regulation Compliance Fire code observance

5S-activity
Inspection system Inspection system establishment

Fig. 2. Scheme of risk reduction.


Y. Sato / Int. J. Production Economics 140 (2012) 579–585 583

Table 1 Percentage complete


Degree of importance of dimensions and that of safety measures for risk reduction.
Estimated percentage complete
Dimensions wD Measures wM
Pipeline network
Equipment wd1 :0.365 Emergency inspection wm11 :0.161
Static electricity measure wm12 :0.135
Tank preservation wm13 :0.201
FMEA wm14 :0.114
Effluent treatment facility
Electricity intentional preservation wm15 :0.190 Electricity facility
Environmental risk hedge activity wm16 :0.124
Incinerator abolition wm17 :0.0745

Human wd2 :0.390 Superintendent arousal wm21 :0.190


Security intensification wm22 :0.186 Simple approximation
Natural calamity measure wm23 :0.161
License institution wm24 :0.161
OHSMS wm25 :0.303 0 Expenditures
Regulation wd3 :0.245 Equipment management system wm31 :0.196
Fig. 3. Conceptual diagram of simple and piecewise linear approximations.
Fire code observance wm32 :0.329
Inspection system establishment wm33 :0.128
Zero emission activity wm34 :0.130
‘‘5S’’ activity wm35 :0.217
necessary in every application of the approach proposed in this
paper. Table 2 summarizes the results of the linear approximation
of pij. A relationship matrix R ¼{rst} (s, t¼ m11, y, m35), represent-
ing mutual relationships among safety measures is defined and
complete was discussed by safety managers of the company and employed in the following formulation of the LP, where rss ¼1,
professional analysts for safety because the percentage complete rst ¼1/2 when a safety measure t has a secondary effect to the
is supposed to be linearly approximated. Two issues were raised percentage complete of a primary measure s, otherwise rst ¼0.
in the discussion: one was the justification for the linear approx- The constant bij in the table denotes necessary and sufficient
imation of the percentage complete; the other was the secondary expenditures to achieve 100% complete for each safety measure
effect of the expenditures for related safety measures to the estimated by the safety managers. In this paper, coefficients in the
percentage complete of a primary safety measure, due to mutual linear approximation are assumed to be the reciprocal values of
relationships among different measures. 100% complete, 1/bij. As noted above, the secondary effect of the
In the justification of the linear approximation, simple and expenditures for related safety measures to the percentage
piecewise linear approximations were compared for all safety complete of a primary safety measure is then assumed to be half
measures shown in Fig. 2. For example, a measure ‘‘Emergency of the ratio 1/bij of the primary measure. As a result, the
inspection (m11)’’ consists of three inspections of pipeline net- parameters of the LP optimizing budget planning of the company
work, electricity facility, and effluent treatment facility, each of can be represented as shown in Table 2. For example, the
which is an independent facility within the plant. Fig. 3 illustrates percentage complete of ‘‘Emergency inspection (m11),’’ p11, can
the conceptual diagram of simple and piecewise linear approx- be linearly approximated by 0.00667Im11 þ0.005Im14 , since ‘‘FMEA
imations. As shown in the figure, the degree of concavity of the (m14),’’ relates to ‘‘Emergency inspection (m11),’’ and b11 ¼ 150 and
relationship determines the precision of the approximation: the b14 ¼100. A diagonal matrix E ¼diag(1/bij) representing sufficient
lower the concavity between no expenditures and full expendi- expenditures to achieve 100% complete is defined and employed
tures for measures is, the higher the precision of the approxima- in the following formulation of the LP. The percentage complete
tion is. A linear function obtained from simple approximation did pij can then be described as REI, where I¼{Imij }.
not succeed in approximating estimated percentage complete
precisely; the difference, however, was not significant from a 4.3. Budget optimization within the plant (example)
practical point of view. As a result of the discussion, simple
approximation was confirmed to be sensible and was justified to This section presents an example to illustrate the application
be employed in this case. On the other hand, the discussion noted of the framework optimizing budget planning for investment in
that in cases where the difference between estimated percentage safety measures within the plant. The framework is based on the
complete and simple approximation would be significant, piece- actual case shown in Tables 1 and 2. Based on the tables, the
wise linear approximation should be employed. objective function maximizing the degree of risk reduction can be
In terms of the secondary effect, several ways of approxima- described as (5), where the percentage complete is approximated
tion were proposed and examined. For example, the expenditures by the linear functions of the budget amount for each safety
for the safety measures ‘‘FMEA (m14)’’ has a secondary effect on measure. The linear constraints (6)–(9) are the budget constraints
the percentage complete of ‘‘Emergency inspection (m11),’’ p11, for each dimension and non-negative condition.
which is, therefore, determined by the primary measure m11 and nX o
the secondary measure m14. Suppose p11 can be linearly approxi- Max i
wd i
wmij
REI , ð5Þ
mated by aIm11 þ bIm14 . Safety managers of the company and X
professional analysts for safety recognized that the effect of Im14 s:t: I
j m1j
r 250 ðbudget constraint of EquipmentÞ, ð6Þ
to p11 was less than that of Im11 . Therefore, sensitivity analysis was
X
carried out by perturbing the coefficients a and b in order to I r 50 ðbudget constraint of HumanÞ, ð7Þ
j m2j
determine adequate values of a and b. As a result, b ¼ 1/2 is
concluded to be adequate where a ¼1. Since the value of b, 1/2, X
I
j m3j
r 35 ðbudget constraint of RegulationÞ, ð8Þ
was discovered through the safety managers’ many years of
experience but having no empirical evidence, the discussion
emphasized that a sensitivity analysis of coefficients would be 0 r Imij ,ði ¼ 1,. . .,3; j ¼ 1,. . .,7Þ, ð9Þ
584 Y. Sato / Int. J. Production Economics 140 (2012) 579–585

Table 2
Linear approximation of the percentage complete.
P
Dimensions D#
i Measures #
bij 1/bij Related measures pij ¼ i,j LðI mij Þ

Equipment D1 ¼ 250 Emergency inspection (m11) 150 0.00667 m14 p11 ¼0.00667 Im11 þ 0.005 Im14
Static electricity measure (m12) 75 0.0133 – p12 ¼0.0133 Im12
Tank preservation (m13) 80 0.0125 – p13 ¼0.0125 Im13
FMEA (m14) 100 0.0100 – p14 ¼0.01 Im14
Electricity intentional preservation (m15) 110 0.00909 m12, m14 p15 ¼0.00665 Im12 þ0.005 Im14 þ 0.00909 Im15
Environmental risk hedge activity (m16) 50 0.0200 m11, m14 p16 ¼0.00333 Im11 þ0.005 Im14 þ 0.02 Im16
Incinerator abolition (m17) 150 0.00667 m13, m14 p17 ¼0.00625 Im13 þ0.005 Im14 þ 0.0667 Im17

Human D2 ¼ 50 Superintendent arousal (m21) 50 0.0200 – p21 ¼0.02 Im21


Security intensification (m22) 75 0.0133 – p22 ¼0.0133 Im22
Natural calamity measure (m23) 50 0.0200 m21, m22 p23 ¼0.01 Im21 þ 0.00665 Im22 þ 0.02 Im23
License institution (m24) 25 0.0400 m21, m22 p24 ¼0.01 Im21 þ 0.00665 Im22 þ 0.04 ub¼Im24
OHSMS (m25) 50 0.0200 m21, m22 p25 ¼0.01 Im21 þ 0.00665 Im22 þ 0.02 Im25

Regulation D3 ¼ 35 Equipment management system (m31) 15 0.0667 – p31 ¼0.0667 Im31


Fire code observance (m32) 10 0.100 m14 p32 ¼0.005 Im14 þ 0.1 Im32
Inspection system establishment (m33) 18 0.0556 – p33 ¼0.0556 Im33
Zero emission activity (m34) 14 0.0714 m31, m33 p34 ¼0.0333 Im31 þ0.0278 Im33 þ 0.0714 Im34
‘‘5S’’ activity (m35) 13 0.0769 m31, m33 p35 ¼0.0333 Im31 þ0.0278 Im33 þ 0.0769 Im35

#: million JPY.

where, wdi and wmij are those listed in Table 1, E ¼diag(1/b11, safety managers of the chemical company. Based on the retro-
1/b12, y, 1/b35), I ¼{Im1j , Im2j , Im3j }, and spective interview focusing on the differences between the actual
m11 m12 m13 m14 m15 m16 m17 m21 m22 m23 m24 m25 m31 m32 m33 m34 m35 and estimated budget allocation and the change in the state of
2 3
risk at the company following the analyses, both the quantifica-
m11 1 n tion of the degree of importance, wD and wM, and the estimation
m12 6
6 1 7
7 of the percentage complete of each safety measure, Pn, are
6 7
m13 6 1 7 persuasive, resulting in a consensus of the allocation for the
6 7
m14 6
6 1 7
7 company. Table 4 summarizes the feedback obtained: each row
6 7
m15 6 n n 1 7 represents how 14 safety managers of the company thought
6 7
m16 6
6n n 1 7
7 about the results of each quantification. As the table shows, more
6 7
m17 6 n n 1 7 than 70% of the respondents thought that the outputs of the
6 7
m21 6
6 1 7
7 framework proposed in this paper were true or rather true.
6 7
R ¼ m22 6
6 1 7
7
m23 6
6 n n 1
7
7
6 7
m24 6 n n 1 7 5. Conclusions
6 7
m25 6
6 n n 1
7
7
6 7
m31 6 1 7 This paper focuses on the optimization of budget planning for
6 7
6 7 investment in safety measures of a chemical company, and the
m32 6 n 1 7
6 7 correlation between safety measures and the degree of risk
m33 6
6 1 7
7
6 7 reduction. Inherent risks at the chemical plant are quantified
m34 4 n n 1 5
m35 n n 1
based on both concrete safety measures for risk reduction and the
consensus of safety managers in the plant. In addition, the degree
n ¼1/2 and blanks are 0. of risk reduction is evaluated based on the importance and the
Table 3 summarizes the optimal solution derived from the LP. percentage complete of safety measures for risk reduction.
The numbers in the Inmij and Pn columns respectively represent the Furthermore, not only the cost effectiveness of investment in
optimal expenditures for each safety measure mij and its esti- safety measures but also their effects on global environmental
mated percentage complete. In terms of the risk level, the degree issues are considered, thus optimizing the budget planning for
of risk reduction from the year 2008 (set 1) to 2009 under the safety measures.
obtained budget allocation is 0.588, which implies the estimated Two open-ended questions remain. First, how to enhance the
risk level one year later would be reduced from 1 to 0.412. This precision of the linear approximation of the percentage complete
result just shows an example of the optimization of budget of each safety measure. The relationship between budget
planning. The process, on the other hand, is based on the actual amounts and the percentage complete of each safety measure
case at a Japanese chemical company, which could easily be just might not lend itself to linear approximation. In such cases,
applied to an actual planning after the linear approximation of the the process of a safety measure needs to be segmented so as to
percentage complete of each safety measure. formulate an LP model. Second, how to establish a feedback
procedure to refine the safety managers’ subjective judgments
4.4. Feedback from the chemical company in the evaluation of safety measures. Since the quantification
process of the degree of importance of safety measures developed
The results derived from the approach proposed in this paper here is based on safety managers’ intuitions or many years of
give quantitative justification for budget planning for investment experience, improving their judgments taking on broad objective
in safety measures and a rational basis for decision making for plant data is crucial.
Y. Sato / Int. J. Production Economics 140 (2012) 579–585 585

Table 3
Results of the optimization.

Dimensions D#
i wD Measures wM Imn ij Pn wDx wMx Pn

Equipment D1 ¼250 wd1 :0.365 Emergency inspection wm11 :0.161 30.4 0.355 0.0209
Static electricity measure wm12 :0.135 30.1 0.402 0.0198
Tank preservation wm13 :0.201 69.3 0.867 0.0635
FMEA wm14 :0.114 15.3 0.153 0.00638
Electricity intentional preservation wm15 :0.190 49.6 1.00 0.0694
Environmental risk hedge activity wm16 :0.124 50.0 1.00 0.0452
Incinerator abolition wm17 :0.0745 5.29 0.0947 0.00258
sub total 0.228

Human D2 ¼50 wd2 :0.390 Superintendent arousal wm21 :0.190 0.000 0.000 0.000
Security intensification wm22 :0.186 0.000 0.000 0.000
Natural calamity measure wm23 :0.161 5.54 0.203 0.0127
License institution wm24 :0.161 16.5 0.937 0.0587
OHSMS wm25 :0.303 27.9 1.00 0.118
sub total 0.190

Regulation D3 ¼35 wd3 :0.245 Equipment management system wm31 :0.196 1.66 0.111 0.00532
Fire code observance wm32 :0.329 10.0 1.00 0.0806
Inspection system establishment wm33 :0.128 0.000 0.000 0.000
Zero emission activity wm34 :0.130 10.3 1.00 0.0318
‘‘5 S‘‘ activity wm35 :0.217 13.0 1.00 0.0532
sub total 0.171

Degree of risk reduction: 0.588


Estimated risk level 1 year later: 0.412

#: million JPY.

Table 4
Feedback from the chemical company.

Items True (%) Rather true (%) Neutral (%) Rather false (%) False (%) DK (%)

Quantification of wD and wM 28.6 42.9 14.3 7.1 7.1 0.0


Estimation of the percent complete P 21.4 50.0 7.1 0.0 14.3 7.1

N ¼14.

Acknowledgments Kaplan, S., Garrick, J.B., 1981. On the quantitative definition of risk. Risk Analysis 1,
11–27.
Labeaua, P.E., Smidtsb, C., Swaminathanb, S., 2000. Dynamic reliability: towards an
The author is grateful to the anonymous referees and the
integrated platform for probabilistic risk assessment. Reliability Engineering
discussants at the 16th International Working Seminar on Pro- and System Safety 68 (3), 219–254.
duction Economics for their valuable comments that have sig- O’Brien, C., Smith, S.J.E., 1993. Design of the decision process for strategic
nificantly improved this paper. investment in advanced manufacturing systems. International Journal of
Production Economics 30-31, 309–322.
Saaty, T.L., 1980. The Analytic Hierarchy Process. McGraw-Hill, New York.
References Saaty, T.L., 1994. Highlights and critical points in the theory and application of the
analytic hierarchy process. European Journal of Operational Research 52,
426–447.
Apostolakis, G.E., 1990. The concept of probability in safety assessment of
Sato, Y., 2004. Comparison between multiple-choice and analytic hierarchy
technological systems. Science 250, 1359–1364.
process: measuring human perception. International Transactions in Opera-
Beaumont, N.B., 1999. Investment decisions in Australian manufacturing. Tech-
novation 18 (11), 689–695. tional Research 11 (1), 77–86.
Davies, L.P., 2002. Risk assessment in the UK nuclear power industry. Safety Tan, K.H., Noble, J.S., Sato, Y., Tse, Y.K., 2011. A marginal analysis guided
Science 40 (1-4), 203–230. technology evaluation and selection. International Journal of Production
Frank, L., 1998. Approaches to risk and uncertainty in the appraisal of new Economics 131 (1), 15–21.
technology capital projects. International Journal of Production Economics Vargas, L.G., 1990. An overview of the analytic hierarchy process and its applica-
53 (1), 21–33. tions. European Journal of Operational Research 48, 2–8.
Garrick, B., Christie, R.F., 2002. Probabilistic risk assessment practices in the USA van der Zee, D.J., 2004. Modeling decision making and control in manufacturing
for nuclear power plants. Safety Science 40 (1-4), 177–201. simulation. International Journal of Production Economics 100 (1), 155–167.

You might also like