Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
ScienceDirect
Available ScienceDirect
online atonline
Available www.sciencedirect.com
at www.sciencedirect.com
Procedia CIRP 00 (2019) 000–000

ScienceDirect
Procedia CIRP 00 (2019) 000–000
ScienceDirect
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Procedia CIRP 00 (2017)

Procedia 000–000
CIRP 83 (2019) 398–403
www.elsevier.com/locate/procedia

11th CIRP Conference on Industrial Product-Service Systems

11th CIRP Conference on Industrial Product-Service Systems
How the Imperfect E-diagnostics Impacts the Capacity Planning and
How the Imperfect E-diagnostics Impacts theNantes,
Capacity France Planning and
Pricing28th CIRP
Decisions Design Conference,
of a Maintenance May 2018,
Service Provider
Pricing Decisions of a Maintenance Service Provider
A new methodology to analyze Mingyao the Sun,
functional
Feng Wu*and physical architecture of
existing products for an assembly Mingyao Sun, orientedFeng Wu* product family identification
The School of Management, Xi’an Jiaotong University, The Key Lab of the Ministry of Education for Process Control & Efficiency Engineering,
The School of Management, Xi’an Jiaotong University, The Key Lab of the Ministry of Education for Process Control & Efficiency Engineering,
Xi’an, 710000 ,China
Paul Stief *, Jean-Yves Dantan, Alain Etienne, Ali Siadat
Xi’an, 710000 ,China
* Corresponding author. Tel.:+8613038596905; E-mail address: fengwu830@126.com
École Nationale Supérieure d’Arts et Métiers, Arts et Métiers ParisTech, LCFC EA 4495, 4 Rue Augustin Fresnel, Metz 57078, France
* Corresponding author. Tel.:+8613038596905; E-mail address: fengwu830@126.com

* Corresponding author. Tel.: +33 3 87 37 54 30; E-mail address: paul.stief@ensam.eu

Abstract
Abstract
Providing online monitoring (E-diagnostics) has become a value-added practice for equipment maintenance service providers to improve
the service quality
Providing
Abstract andmonitoring
online efficiency. (E-diagnostics)
However, the E-diagnostics
has become aisvalue-added
always imperfect,
practicewhose accuracy maintenance
for equipment depends on how much
service components
providers of the
to improve
equipment
the service are monitored
quality (namely,However,
and efficiency. E-diagnostics depth in ourisresearch).
the E-diagnostics This research
always imperfect, whoseanalyzes
accuracythe optimal
depends oncapacity
how much planning and pricing
components of the
problem
equipment by taking
are into
monitored account the
(namely, imperfect E-diagnostics.
E-diagnostics depth in In
our addition,
research).we consider
This the
research E-diagnostics
analyzes the accelerate
optimal effect
capacity
In today’s business environment, the trend towards more product variety and customization is unbroken. Due to this development, the need of to the
planningservice
and process
pricing
and attempt
problem
agile and by to find into
taking howaccount
reconfigurable the effect
theinfluences
production systems the
imperfectemerged service
E-diagnostics.provider’s
to cope Inwith decisions.
addition, Our results
weproducts
various consider reveal that
theproduct
and E-diagnostics theaccelerate
families.optimal pricing
To design andand
effect capacity
tooptimize
the serviceplanning
process
production
decisions
attemptaretodependent
find how on
the customers’
effect failure
influences thecost and
service service provider’s
provider’s compensation
decisions. Our results cost.
revealSpecifically,
that the if customers’
systems as well as to choose the optimal product matches, product analysis methods are needed. Indeed, most of the known methodsplanning
and optimal pricing failure
and cost
capacity isaim
larger
to
than service
decisions are provider’s
dependent compensation,
on customers’ the service
failure costprovider
and would
service improve
provider’s the price
compensation as the E-diagnostics
cost. Specifically,depth
if increases;
customers’
analyze a product or one product family on the physical level. Different product families, however, may differ largely in terms of the number otherwise,
failure cost the
is price
larger
and
should
than
nature ofbecomponents.
lowered
service as the
provider’s ThisE-diagnostics
compensation,
fact impedes depth
the increases.
service
an In addition,
provider
efficient wechoice
would improve
comparison and findthe
asof
the
priceaccelerate effect becomes
as the E-diagnostics
appropriate product increasingly
depth
family increases; obvious,
combinations otherwise,customers’
the price
for the production
expected
should bewaiting
lowered time
as increases
the but
E-diagnosticsthe service
depth price decreases.
increases. In addition, we find as the accelerate effect becomes increasingly
system. A new methodology is proposed to analyze existing products in view of their functional and physical architecture. The aim is to cluster obvious, customers’
expected
these waiting
products timeassembly
in new increasesoriented
but the product
service price decreases.
families for the optimization of existing assembly lines and the creation of future reconfigurable
© 2019
assembly
© 2019 TheThe Authors.
systems. Based
Authors. Published
on Datum
Published by Elsevier
Flow Chain,
by Elsevier B.V.
B.V.the physical structure of the products is analyzed. Functional subassemblies are identified, and
Peer-review
a©
Peer-review
functional under
2019 Theanalysis
under responsibility
Authors. isPublished
responsibility
performed.by
ofofthe
thescientific
scientific
Elsevier
Moreover, B.V. committee
acommittee
hybrid of of
thethe
functional11th11th CIRP
CIRP
and Conference
Conference
physical on Industrial
on Industrial
architecture Product-Service
graph Product-Service
(HyFPAG) Systems.
Systems
is the output which depicts the
Peer-review under responsibility of the scientific committee of the 11th CIRP Conference on Industrial
similarity between product families by providing design support to both, production system planners and product designers. Product-Service Systems.An illustrative
Keywords: Imperfect E-diagnostics; capacity planning; pricing; E-diagnostics depth; accelerate effect
example of a nail-clipper is used to explain the proposed methodology. An industrial case study on two product families of steering columns of
Keywords: Imperfect E-diagnostics; capacity planning; pricing; E-diagnostics depth; accelerate effect
thyssenkrupp Presta France is then carried out to give a first industrial evaluation of the proposed approach.
© 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the 28th CIRP Design Conference 2018.
1. Introduction The main idea of the E-diagnostics is to extract useful
1. Introduction
Keywords: Assembly; Design method; Family identification
The main idea
information of the the
that reveals E-diagnostics
equipment’sis condition
to extractfrom useful
the
Nowadays, customers always purchase the “after sales” information
available datathat reveals the
monitored equipment’s
by multiple condition
sensors, and thenfrom usethe
it
Nowadays,
service in terms customers
of service always purchase
agreements fromthethe“after sales”
equipment available
to predictdatafailures
monitored in by multiplewith
advance sensors,
the and thenmining
data use it
service in terms
manufactures in of service
order agreements
to keep from the in
the equipment equipment
normal to predict failures in advance with the data mining
technologies.
1.manufactures
Introduction
condition [1]. inTherefore,
order to many keep the equipment
equipment in normal of
manufacturers, the product
technologies.
While the range
benefitandofcharacteristics
E-diagnosticsmanufactured
is believed and/or
to be
condition [1]. Therefore,
especially those many equipment
with high complexity including assembled
manufacturers,
and precision, While in
substantial the
bythisbenefit
system.
both ofIn E-diagnostics
industry this practitioners
context, theismain challenge
believed
and to be
researchers in
Due to
especially
General those
Electricthe withfasthigh
Co., development
complexity
Siemens AG and inASML
and the Holding
precision,domain
includingof
N.V, modelling
[5][6][7], and
substantial by analysis
almost both is now
no industry
effort not onlydevoted
haspractitioners
been to cope
andtowith single
researchers
determine
communication
General have
etc.[2], Electric and
Co.,an
transformed ongoing
SiemenstheirAG trend
and ASML
business of digitization
models Holding and
N.V,
from being products,
[5][6][7], a almost
quality-speed limited product
no
trade-off effort range
has or
strategies beenexisting
devoted
together product
with families,
to determine
E-diagnostics
digitalization,
etc.[2], have
providers manufacturing
transformed
of products enterprises
their
to becoming business are
providers facing
models
of both important
from being but
products also tobecause
quality-speed
accuracy be able to
trade-offanalyze
customers and
can to
strategies compare
together
only products
with
experience to quality
define
E-diagnostics
a low
challenges
providers
and services ofin[3].today’s tomarket
products becoming environments:
providers of aboth continuing
products new product
accuracy
product families.
because
with a high It can be
customers
production canobserved
only thatThe
speedexperience
[8]. classical
a lowexisting
quality
quality-speed
tendency
andThe towards
services [3]. reduction
development of smart of manufacturing
product development times and
technologies, like product
trade-offfamilies
product with are regrouped
a high
in service systems inhas
production function
speed of clients
[8].
aroused The or features.
the quality-speed
attention of
shortened
condition product
The development lifecycles.
monitoring, of smart Inmanufacturing
addition,
information theretechnologies,
is an increasing
transformation like However,
and data researchesassembly
trade-off inin service
recent oriented
systems
years product
has families
aroused
[9][10][11]. Theare
the hardly
optimal to
attention find.
of
service
condition
demand
mining, ofetc.,monitoring,
customization,
have enhanced information
being at the
the servicetransformation
same time in
provider’s and data
a global
capabilities researches
Ondecision
rate in of
the product recent years
family
maintenance [9][10][11].
level, products
service The
differoptimal
provider mainly
has service
in two
studied by
mining,
competition
to monitoretc.,the have
with enhanced the
competitors
equipment’s allservice
over the
condition provider’s
world.by
remotely, capabilities
This trend,
means of main rate
somedecision of maintenance
characteristics:
researchers but(i)they
the do service
number of provider
not take account has
components studied
and
of the by
(ii) the
quality-
to
whichmonitor
various issensorsthe equipment’s
inducing condition
the development
mounted on the equipment remotely,
from macro
[4]. by to
This means
microof type
capability some
speedofresearchers
components
trade off and but they
(e.g. do notthe
mechanical,
consider takeE-diagnostics
account electronical).
electrical, of theis quality-
perfect
various
is oftensensors
markets, referredmounted
results on the lot
inasdiminished
E-diagnostics equipment
sizes [4]. to
due
by industry This capability
augmenting
practitioners. speed trademethodologies
Classical
[12][13]. off and consider the E-diagnostics
considering mainly singleisproducts
perfect
is oftenvarieties
product referred(high-volume
as E-diagnostics by industry
to low-volume practitioners.
production) [1]. [12][13].
or solitary, already existing product families analyze the
To cope with
2212-8271 © 2019 this
Theaugmenting variety
Authors. Published as wellB.V.
by Elsevier as to be able to product structure on a physical level (components level) which
Peer-reviewpossible
identify
2212-8271 ©under
2019responsibility
Authors. of
The optimization the scientific
Published committee
potentials
by Elsevier in ofthe
B.V. the 11th CIRP Conference
existing causeson Industrial
difficultiesProduct-Service
regardingSystems.an efficient definition and
doi:10.1016/j.procir.2017.04.009
Peer-review under responsibility of the scientific committee of the 11th CIRP Conference on Industrial Product-Service Systems.families. Addressing this
production system, it is important to have a precise knowledge comparison of different product
doi:10.1016/j.procir.2017.04.009

2212-8271©©2017
2212-8271 2019The
The Authors.
Authors. Published
Published by Elsevier
by Elsevier B.V. B.V.
Peer-reviewunder
Peer-review underresponsibility
responsibility
of of
thethe scientific
scientific committee
committee of the
of the 28th11th
CIRPCIRP Conference
Design on 2018.
Conference Industrial Product-Service Systems.
10.1016/j.procir.2019.04.093
 Mingyao Sun et al. / Procedia CIRP 83 (2019) 398–403 399
2 Author name / Procedia CIRP 00 (2019) 000–000

In this paper, we consider that E-diagnostics is imperfect 3. Analysis

and may leave out some failures, whose accuracy is
determined by how many indicators (namely, E-diagnostics Because of the imperfect E-diagnostics, the service
depth in our paper) are monitored. In addition, the information provider has to make a promise to compensate for those
collected by E-diagnostics may be beneficial to improve the missed failures, which also appears in the equipment
service rate but guarantee the service quality (namely, maintenance contract design literatures [2][16]. We use
accelerate effect in our paper); we explore how the effect denote the compensation cost. Therefore, the service
impacts the service provider’s strategies. Figure 1 shows the provider’s objective function can be described as
flow schematic of conducting maintenance service with
( ) ( ) ( ) ( )̂
imperfect E-diagnostics. Our research focuses on these ( )
problems. ( )(
Failure
)
predicted
( ) ( )
Patient arrivals Imperfect E- Maintenance
diagnostics service

Failure
The following proposition derives the optimal solutions of
missed the above problem.
Proposition 1: Considering a maintenance service
Customer
departures
provider with fixed E-diagnostics depth , a unique market
equilibrium exists in which
(i) the optimal service rate is equal to
Fig 1. Flow schematic for maintenance with imperfect E-diagnostics ( )
and the corresponding optimal
2. Model of Customer Decision ( )
price is √ ;
Customers’ perceived service value consists of two parts. (ii) the expected service value that customers perceived at the
( )
First, E-diagnostics can predict the failure in advance and help equilibrium state is ( ) ( ) ;
customers avoid the useless production cost, which is called (iii) the equilibrium arriving rate is equal to
failure cost and denoted as . The E-diagnostics accuracy ( )
is dependent on E-diagnostics depth , that is, . √ .
Therefore, the service value of this part is ( )
How E-diagnostics depth ( ) affects the service
( ) . Second, the regular maintenance service
provider’s decisions? Is it beneficial if the exogenous E-
provides necessary repair or replacement of defective items diagnostics depth is high? The following corollary implies
for customers, the value of which can be represented by that the customer failure cost saving ( ) and the service
( ) ( ) because of the quality-speed trade compensation cost ( ) play important roles in the mentioned
off consideration. As a result, customers’ perceived service problems.
value is equal to
Corollary 1: As the E-diagnostics depth increases:
( ) ( ) ( ) ( ) (i) the optimal service rate and the equilibrium
arriving rate both increase in ;
Customers’ utility is dependent on three crucial factors: (ii) if , the expected service value for customers
service value, waiting time and price. Without loss of ( ) ( ) and the optimal price increases in ; if
generality, we assume that customers obtain zero utility from , ( ) ( ) and decreases in .
balking. A unit-time waiting cost is incurred for each
customer staying in the system. We assume that the service We note that the pricing policy and the expected service
time follows an exponentially distribution and thus the service value are hugely affected by the failure cost saving and the
system corresponds to an M/M/1 queue, which is extensively service compensation cost . When customers are more cared
adopted in service operation research [14][15]. Hence, the net about the E-diagnostic accuracy(i.e. ), our result
utility derived by any customer at the arrival rate ( ) is indicates that as E-diagnostics depth increases, the
improvement of the service value in testing resulting from the
( ) ( ) ( ) increase of dominates the decline of service value in
( ( )) processing resulting from the decrease in service time.
Therefore, the customers’ expected service value is increased,
And customers will continuously join the service until the and thus the service provider would improve the price
value ( ) is driven to zero. Therefore, the equilibrium without reducing the effective demand. However, when the
arrive rate is given as service provider is more concerned about the E-diagnostic
accuracy (i.e. ), the result is totally different.
( )
( ) ( ) 4. Benefits of the E-diagnostics
The model developed in the previous section only takes
400 Mingyao Sun et al. / Procedia CIRP 83 (2019) 398–403
Author name / Procedia CIRP 00 (2019) 000–000
the E-diagnostics as a process to identify equipment’s status.
In this section, we consider the possibility that the information
collected from E-diagnostics process is beneficial for the
subsequent maintenance service. The forecasted service
requirement by E-diagnostics can also reveal some specific
resources (i.e. spare parts, technical workers, et al.) that
needed to complete the service, which can help the service
provider reduce the time for preparing the service. Therefore,
the service rate is improved but the service quality does not
decline.
Customers are divided into two types by the imperfect E-
diagnostics, indexed by , is the customer type that
is accurately diagnosed by E-diagnostics (failure predicted
customers) and is the customer type that is misdiagnosed
(failure missed customers). Upon the service, type
customers can experience a faster service process than type
customers as the service provider can prepare the specific
maintenance resources for the accurately diagnosed customers.
That is, in order to achieve a service depth ( ) , the
workload size for type customers follows an exponential
distribution with mean while for type customers follows
an exponential distribution with mean , where is a
measure of the useful information that is obtained by the E-
diagnostics, namely “accelerate factor”. In this case, the
expected waiting time of the service system is [11]
( )
( ) [ ][ ] depth increases. In addition, we reveal that as the accelerate
( )
Lemma 1:Consider a maintenance service provider with
E-diagnostic and the accelerate factor is , the total expected
waiting time ( ) is convex decreasing in and convex
increasing in .
Lemma 1 shows that the waiting time is strictly
increasing in total arriving rate , thus, for each level of price
, the market equilibrium demand condition uniquely defines
a arriving rate ( ). The objective function for the service
provider can be obtained as follows by substituting from the
market equilibrium demand condition
( ) [ ( ) ( ) ( )]
Proposition 2: When service rate differentiation is
incurred due to the accelerate effect of the E-diagnostics, as
the value of accelerate factor increases,
(i) the service provider admits more customers into the
service (i.e. increases in ) but works slower (i.e.
decreases in );
(ii) the expected waiting time increases in ;
(iii) the equilibrium price decreases in .
Proposition 3(ii) indicates that customers encounter an
increasingly congested service system as the accelerate factor
increases. It is rather counterintuitive because one can easily
consider that the E-diagnostics information could ease the
system congestion if the accelerate effect exists. The
explanation is as follows. As the accelerate factor increases,
the service provider admits more customers but works with a
3
low service rate (proposition 3 (i)). Though the E-diagnostics
can improve the average service rate, the induced congestion
level increases because of the increase in arrival rate and
decrease in service rate at equilibrium.
It is also interesting to find that the service provider
charges a lower price as the accelerate factor decreasesat
equilibrium. The underlying intuition is as follows. Combing
the decreased service rate with the increased expected waiting
time, we observe that the induced net service value for
customers decreases in essentially, and thus the equilibrium
price should be decreased.
5. Conclusions
This paper was initially motivated by an observational study
in a semiconductor backend industry. The diagnosis of the
high precision equipment is always imperfect. In order to
improve the diagnostic accuracy, quantities of sensors are
mounted on equipment. However, it may be wasteful since
much information monitored is useless. In this paper, we
study the optimal capacity planning integrated with pricing
problem of a maintenance service provider.
Our model reveals that the customers’ failure cost and
service provider’s compensation cost affect the pricing policy
significantly. Specifically, if customers’ failure cost is larger
than service provider’s compensation, the service provider
would improve the price as the E-diagnostics depth increases;
otherwise, the price should be lowered as the E-diagnostics
effect becomes increasingly significant, (i) the equilibrium
arrival rate increases; (ii) the optimal service rate decreases;
(iii) the service provider’s revenue increases. It is
counterintuitive to find that the expected waiting time does
not decline but improves despite the existence of the
accelerate effect.
Acknowledgements
This work was supported by the National Natural Science
Foundation under Grant No. 71871177, 71471144, National
Key R&D Program of the Ministry of Science and
Technology: 2018YFB1703000 and 2018YFB1703001.
Appendix A.
We provide some key proof of our results.
A.1. Proof of proposition 1
. We solve the service provider’s utility maximization
problem in two steps. First, we find the optimal price ( )
for a given service rate . Then we can obtain the optimal
service rate by plugging ( ) into the objective function.
First, we show ( ) is concave in price . Taking the
first and second derivative of , we get
 Mingyao Sun et al. / Procedia CIRP 83 (2019) 398–403 401
4 Author name / Procedia CIRP 00 (2019) 000–000

(ii) Taking the derivative of and ( ) ( ) with

( ) ( ) respect to , we can have the following results:
[ ] ( ) [ ( ) ( )]
( ( ) ( ) )
[ ] ( )
( ( ) ( ) ) ( ( ) ( ) )
( ) is concave in price can be proved since . By Therefore, if , it is obvious that
[ ( ) ( )]
solving the first order condition, we can get the optimal , and thus the optimal service price
service price for any value of service rate : ( ) and induced service value increases in E-diagnostics depth .
( ) √ ( ( ) ( ) ) . As a result, the However, if , the result is the opposite.
service provider’s revenue function is equal to: (iii) The service provider’s revenue function in the short
run problem is:
( ( )) [ √ ][ ( ) √ ( )] ( ) ( ) ( )
( )
The expected profit from each customer
where ( ) ( ) ( ) . Taking the first ( )
√ , which increases in obviously.
derivative of ,
In addition, we have confirmed in (i) that the equilibrium
( ( ))
( ) demand increases in . As a result, we can confirm that the
revenue increases in .
( ( ))
√ [ ( )] ( )
A.3. Proof of lemma 1
As a result,
The formulation of the waiting time is
( ( ) )[ √ ]
( ) ( )
( ) [ ][ ]
( )
When √ ( )
, the equilibrium demand is equal to
and we define
from lemma 1. Thus, we only consider the effective
( )
case that √ ( )
, which indicates that the sign of ( ) ( )

is the same with ( ) . Because the sign of ( ) is

first positive and then negative in , we can know that
( )
( ( )) is concave in . Solving the first order condition,
we can get the optimal service rate: It is obvious that , and , .
( )
(i) Taking the first order derivative of with respect to , we
can get
which in turn yields
( )
√
Taking the first order derivative of with respect to , we
( ) can get
( ) ( )
( )
√
(ii) Taking the second order derivative of with respect to ,
we can get
A.2. Proof of corollary 1

(i) Taking the derivative of and with respect to ,

we can have the following results:
( ) ( )

Therefore, we can get the conclusion that optimal service rate

and the equilibrium arrival rate both increase in .
402 Mingyao Sun et al. / Procedia CIRP 83 (2019) 398–403
Author name / Procedia CIRP 00 (2019) 000–000 5

Similarly, we can also confirm .

Because , we can get

[
( ) ( )
the result .
( ) ( )( )
][ ]
( )

( )[ ( ) ]
[ ][ ]
( )
( )
Similarly, we can get

( )( )[ ( )] ( (

))

As a result, .
Taking the second order derivative of with respect to
, we can get

( ) ( ) ( )
)
( )

( )
( )
( )
As (the queue is limited), we can get the
result .
Similarly, we can get

[ ( )]
[ ( )]

As a result, .

As mentioned in lemma 1, , thus,

we can obtain
( )

Further, we can get

(
( )( ) ( ) ( )( )
)
( )

Keep in mind that .

 Mingyao Sun et al. / Procedia CIRP 83 (2019) 398–403
6 Author name / Procedia CIRP 00 (2019) 000–000
References
[1] Vardar C , Gel E S , Fowler J W . A framework for evaluating remote
diagnostics investment decisions for semiconductor equipment suppliers.
European Journal of Operational Research, 2007, 180(3):1411-1426.
[2] Chan T H, De Véricourt F, Besbes O. Contracting in medical equipment
maintenance services: An empirical investigation. Management Science,
2018.
[3] Zhang X , Han X , Liu X , et al. The pricing of product and value-added
service under information asymmetry: a product life cycle perspective.
International Journal of Production Research, 2015, 53(1):25-40.
[4] Topan E , Tan T , Van Houtum G J , et al. Using Imperfect Advance
Demand Information in Lost-Sales Inventory Systems with the Option of
Returning Inventory. IISE Transactions, 2018:246-264.
[5] Hung M H, Chen K Y, Ho R W, et al. Development of an e-
diagnostics/maintenance framework for semiconductor factories with
security considerations. Advanced Engineering Informatics, 2003, 17(3-4):
165-178.
[6] Muller A, Marquez A C, Iung B. On the concept of e-maintenance:
Review and current research. Reliability Engineering & System Safety,
2008, 93(8): 1165-1187.
[7] Locy M. The impact of e-diagnostics-one year later. 2001 IEEE
International Symposium on Semiconductor Manufacturing. ISSM 2001.
Conference Proceedings (Cat. No. 01CH37203). IEEE, 2001: 435-438.
[8] Gupta S, Rajagopalan S. Managing operations and customer demand at
Coverking. Case study, Marshall School of Business, University of
Southern California, Los Angeles, 2012.
403
[9] Anand K S, Paç M F, Veeraraghavan S. Quality–speed conundrum:
Trade-offs in customer-intensive services. Management Science, 2011,
57(1): 40-56.
[10] Li X, Guo P, Lian Z. Quality ‐ speed competition in customer ‐
intensive services with boundedly rational customers. Production and
Operations Management, 2016, 25(11): 1885-1901.
[11] Dai T, Akan M, Tayur S. Imaging room and beyond: The underlying
economics behind physicians’ test-ordering behavior in outpatient
services. Manufacturing & Service Operations Management, 2016, 19(1):
99-113.
[12] Kurz J, Pibernik R. Flexible Capacity Management with Future
Information. Working paper，2016.
[13] Kurz J. Capacity planning for a maintenance service provider with
advanced information. European Journal of Operational Research, 2016,
251(2): 466-477.
[14] Kostami V, Rajagopalan S. Speed –quality trade-offs in a dynamic
model[J]. Manufacturing & Service Operations Management, 2013, 16(1):
104-118.
[15] Ni G, Xu Y, Dong Y. Price and speed decisions in customer-intensive
services with two classes of customers[J]. European Journal of
Operational Research, 2013, 228(2): 427-436.
[16]Wang W, Zhao F, Peng R. A preventive maintenance model with a two-
level inspection policy based on a three-stage failure process. Reliability
Engineering & System Safety, 2014, 121(1):207-220.