outsourcing analysis in closed-loop supply chains for

8
Outsourcing Analysis in Closed-Loop Supply Chains for Hazardous Materials Víctor Manuel Rayas Carbajal Tecnológico de Monterrey, campus Toluca [email protected] Marco Antonio Serrato García Tecnológico de Monterrey, campus Toluca [email protected] Abstract In recent years, environmental issues have become a main topic worldwide. Governments around the world have established laws and policies to reduce the impact of industrial activity, forcing companies - especially those who produce or manage hazardous materials- to satisfy specific requirements on their supply chain systems. This is why many companies consider outsourcing as an option for these functions. Through this research, a Markov decision models are developed to support outsourcing decisions in a closed-loop supply chain system for hazardous materials. The models are based on the risk levels and sales behavior of the product considered. An optimal monotone nondecreasing policy is identified, which provides valuable insights for decision-makers involved in such systems. Keywords: Closed-loop supply chain; Markov Decision Model; Hazardous materials; Outsourcing. 1 Introduction Due to laws, international agreements, pressure from society, among other factors, human safety and health, environmental protection and security concerning hazardous materials supply chain are main topics for many countries, industries and organizations around the world (Mullai and Larsson, 2008). Hazardous materials are defined and regulated by a number of agencies around the world, whom establishes the regulations that concern the handling, storage and distribution of hazardous materials (Murray, 2013). For this reason, the shipping of hazardous materials can only be performed by carriers that are registered and only when the material is properly classed, described, packaged, marked, labeled, and in condition for shipment. Integrating environmental concerns into supply chain management has become increasingly important for manufacturers to gain and maintain competitive advantage (Zhu et al., 2008). As more executives adopt environmental practices, supply chain strategies will only increase in importance. As companies focus more tightly on their core competencies, they will rely more heavily on their suppliers for non-core activities such as the transportation, recovery and disposal of their products (Handfield et al., 2005). The characteristics of each product and activity suggest specific strategies. Low-value activities require little attention and might even be completely outsourced (Handfield et al., 2005). With an outsource strategy, companies can improve benefits while they are focusing on their core activity (Boyson, et al., 1999).

Upload: others

Post on 27-Apr-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Outsourcing Analysis in Closed-Loop Supply Chains for

Outsourcing Analysis in Closed-Loop Supply Chains for Hazardous

Materials

Víctor Manuel Rayas Carbajal

Tecnológico de Monterrey, campus Toluca

[email protected]

Marco Antonio Serrato García

Tecnológico de Monterrey, campus Toluca

[email protected]

Abstract

In recent years, environmental issues have become a main topic worldwide. Governments around the

world have established laws and policies to reduce the impact of industrial activity, forcing companies -

especially those who produce or manage hazardous materials- to satisfy specific requirements on their

supply chain systems. This is why many companies consider outsourcing as an option for these functions.

Through this research, a Markov decision models are developed to support outsourcing decisions in a

closed-loop supply chain system for hazardous materials. The models are based on the risk levels and

sales behavior of the product considered. An optimal monotone nondecreasing policy is identified, which

provides valuable insights for decision-makers involved in such systems.

Keywords: Closed-loop supply chain; Markov Decision Model; Hazardous materials; Outsourcing.

1 Introduction

Due to laws, international agreements, pressure from society, among other factors, human safety and

health, environmental protection and security concerning hazardous materials supply chain are main

topics for many countries, industries and organizations around the world (Mullai and Larsson, 2008).

Hazardous materials are defined and regulated by a number of agencies around the world, whom

establishes the regulations that concern the handling, storage and distribution of hazardous materials

(Murray, 2013). For this reason, the shipping of hazardous materials can only be performed by carriers

that are registered and only when the material is properly classed, described, packaged, marked, labeled,

and in condition for shipment.

Integrating environmental concerns into supply chain management has become increasingly important for

manufacturers to gain and maintain competitive advantage (Zhu et al., 2008). As more executives adopt

environmental practices, supply chain strategies will only increase in importance. As companies focus

more tightly on their core competencies, they will rely more heavily on their suppliers for non-core

activities such as the transportation, recovery and disposal of their products (Handfield et al., 2005).

The characteristics of each product and activity suggest specific strategies. Low-value activities require

little attention and might even be completely outsourced (Handfield et al., 2005). With an outsource

strategy, companies can improve benefits while they are focusing on their core activity (Boyson, et al.,

1999).

roger
Typewritten Text
R. Z. Ríos-Mercado et al. (Eds.): Recent Advances in Theory, Methods, and Practice of Operations Research, pp. 41-48, UANL - Casa Universitaria del Libro, Monterrey, Mexico, October 2014.
Page 2: Outsourcing Analysis in Closed-Loop Supply Chains for

2 Environmental Risk

The use of hazardous materials can cause unintentional accidents. Incidents have occurred in every

system of the hazardous materials supply chain, including platforms, all modes of transport, chemical

plants, terminals and storages (Mullai and Larsson, 2008). Managers have come to realize that a large and

increasing amount of environmental risk can be found in nearly every company’s supply chain, increasing

the importance of the decisions in this area (Hanfield et al., 2008). This risk implies that the companies

must be more specialized on each one of these activities or outsource some of them in order to focus on

their core activity (Zhu et al., 2008).

A risk/cost framework for the hazardous materials management system must include an assessment of the

risk due to storage, transportation, treatment and disposal. The risk cost calculation may vary by the type

of activity involved, but it must be according to the accident rate and possible affected population

(Killmer et al., 2002). Only for the case of petroleum products, from 1990 to 2000, 36 accidents were

reported in the management of these materials, which resulted in more than 2200 deaths and about 3,000

injured people (Alcantara and Gonzalez, 2001).

3 Closed-Loop Supply Chains and their Outsource

The research of the supply chain management has passed through various stages, from the individual

activities optimize to an entire analysis of the whole chain. This is the case of the closed-loop supply

chain management, which is define as “the design, control, and operation of a system to maximize value

creation over the entire life cycle of a product with dynamic recovery of value from different types and

volumes of returns over time” (Guide and Van Wassenhove, 2009).

The major difference between CLSC and traditional supply chains is for a forward supply chain, the

costumer is at the end of the process, and for a CLSC, there is value to be recovered from the costumer or

end-user. The value to be recovered is significant, only in the United States is over $50 billion in annual

sales of remanufactured products (Guide and Van Wassenhove, 2003). Although, there are some

complicating characteristics for planning and controlling a supply chain with remanufacturing of external

returns. Some of them are the requirement for a reverse logistics network, the uncertain timing and quality

of cores, the uncertainly in material recovered from cores, the problem of stochastic routings for materials

and highly variable processing times and the need to balance returns of cores with demands for

remanufactured products (Guide, 2000).

Supply chain management is recognized as a strategy for improving competitiveness by improving

customer value and reducing cost (Mentzer, 2004). Given the logistics costs that are implied in this

activities and the customers’ demands for shorter order cycles, some companies consider outsourcing

these activities to third party logistics (3PL) providers. Warehouse, distribution and reverse logistics are

the most common activities to outsource (Arroyo et al., 2006). Also, it is common to outsource multiple

logistics services, but just a few companies outsource the manufacture or production activity (Lieb and

Bents, 2005). Boyson et al. (1999) find that firms can improve customer service and reduce costs by

outsourcing packages of functions and suggest as the main benefits of 3PL: cost savings, operational

efficiency, flexibility and improved customer service. Generally, outsourcing logistics functions is a long

term decision (Serrato, et. al. 2007). This is consistent with the survey by Boyson et al. (1999), where the

respondents who have outsourced an activity in their company, only 4% reported that they stopped

outsource this activity.

4 Problem Description

Page 3: Outsourcing Analysis in Closed-Loop Supply Chains for

Because some activities of the closed-loop supply chain for hazardous materials do not represent the core

business of the company, one of the most important decisions for any organization is which activities

should be outsourced to a 3PL. Such decision considers not only whether or not outsource, but also when

they should be outsourced, in order to minimize the total expected cost for the whole cycle and reduce the

risk cost associated, this by a Markov Decision Model (MDM) which is ideal for stochastic problems.

The models are based on the Risk Levels, Sales and Returns behavior of the hazardous material

considered. For this model it is assumed that the CLSC consists of six main activities (figure 1). Lieb and

Bentz (2005) indicate that in their surveys conducted in 2004, 67% of companies outsource distribution

activities, warehousing activities 46% and 33% reverse logistics (RL) activities. According to this, it is

assumed that these activities can be outsourced, while the production is a core activity and never will be

outsourced and the market and re-use/disposal activity are probabilistic (with a known probability

distribution) and the company does not directly control.

Figure 1. Closed-loop supply chain activities.

5 Markov Decision Model Developed

We define the follow notation for the MDMs developed:

Sets

Subactivities or expenses,

{

Parameters

Cost of subactivity or expense j

Unit shortage cost

Environmental risk cost,

L Length of the product life cycle

W Time length defined by the firm to continue managing the returns for the product analyzed

T Length of the study horizon, T = L + W

t Decision epoch, , t={1, …, T-1}, where decision epoch t represents the end of period t. Time T

corresponds to the end of the problem horizon, where no decision is taken.

Expected sales for period t

Rate for sales increase,

Page 4: Outsourcing Analysis in Closed-Loop Supply Chains for

Rate for devolutions,

Random variables

Amount of units sold by the firm during period t

Cumulative sales experienced by the firm from period 1 through the end of period t, ∑

Number of units returned in period t

Cumulative number of units returned from period 1 to the end of period t, ∑

State variables

Amount of units sold by the firm during period t

Number of units sold and not returned at the end of period t,

Capacity at period t

Model assumptions

Returns are a function of the number of units previously sold but not yet returned. Each unit has a binomial distribution probability of being returned (Serrato, et al., 2007).

Sales are assumed to be distributed Poisson with mean λ. Average sales change at a known rate in each period (Chen, 2014).

There is a cost per unit for the collection and handling of a returned unit, which is considered less than the savings generated by remanufacturing one unit (Savaskan et al., 2004).

Given that warehouse, distribution and RL does not represent a core activity, it is also assumed that once the outsourcing decision is taken, it remains in place for the rest of the problem horizon as an absorbing state (Serrato, et al., 2007).

If any activity is still done internally and it incurs in a shortage, then there would exist a cost associated to meet this demand. If the activity is outsourced, the 3PL would always have enough capacity to satisfy such a demand (Serrato, et al., 2007).

Because of the last assumption, each activity can be considered independent such that the decision will be

made with a different and independent model, being analyzed together as a final stage in this research.

5.1 Markov Decision Model for Warehouse and Distribution

Due to the similarities of the operations and parameters in the warehouse and distribution activities, both

models shares the states, action, transition probabilities and rewards definitions, but there is a difference

on the calculation of the environmental risk cost associated. The models are defined by:

States. The system state at each decision epoch t is defined as { }, for t=1, …, T. At decision

epoch 0, the system state is { }, where . Also, .

Actions. Given the purpose of the MDM, we assume that two actions are available:

Continue performing the activity internally, by updating the firm’s capacity to the expected

amount of sales in the next period, i.e., [ ] [ ] .

Adopt an outsourcing strategy for the activity by having a 3PL perform such activity and taking

the firm’s activity capacity to zero; i.e., .

Transition probabilities. As the sales at each period follow a Poisson distribution, the transition

probabilities among states are defined as [( )| ]; i.e.,

[( )| ] {

[( )| ] {

[ | ]

Page 5: Outsourcing Analysis in Closed-Loop Supply Chains for

[ | ]

Rewards. The following reward structure is defined for actions a=0 or 1.

[ ] ( ) ( )

[ ]

[ ]

Where denotes . Environmental Risk Cost for Warehouse. This cost is defined as (Killmer et al., 2002).

Where:

Probability of an accident occurs

Cost per person affected

Affected area in case of accident

Population density of the area affected

Environmental Risk Cost for Distribution. This cost is defined as (Killmer et al.,

2002). Where:

Probability of an accident occurs

Cost per person affected

Population of the destiny

Travel distance

Distance between the route and the destiny

5.3 Markov Decision Model for RL

The MDM is defined by:

States. The system state at each decision epoch t is defined as { }, for t=1, …, T. At decision

epoch 0, the system state is { }, where .

Actions. Given the purpose of the MDM, we assume that two actions are available:

Continue performing the activity internally, by updating the firm’s capacity to the expected

amount of returns in the next period, i.e., [ ] [ ] .

Adopt an outsourcing strategy for the activity by having a 3PL perform such activity and taking

the firm’s activity capacity to zero; i.e., .

Transition probabilities. As the returns in each period follow a binomial distribution,, the transition

probabilities among states are defined as [(( ) ) | ]; i.e.:

[(( ) ) | ] {(

)

[( )| ] {(

)

Rewards. The following reward structure is defined for actions a=0 or 1.

[ ] ( ) ( )

[

]

[ ]

[ ]

Where denotes . Environmental Risk Cost. This cost is defined as (Killmer et al., 2002). Where:

Probability of an accident occurs

Cost per person affected

Affected area in case of accident

Page 6: Outsourcing Analysis in Closed-Loop Supply Chains for

Population density of the area affected

5.4 System Dynamic

For all the models, the system follows the dynamic presented at figure 2 for each period t. At the end of

the last period of production and returns, all the capacity remaining in the system is sold.

Figure 2. System dynamic.

The action decided for each period is defined by the maximum reward earned by continuing optimally

from state onwards for each independent activity at each period t. The optimal policy for each

activity can be obtained by solving recursively:

{ [ ] ∑ [ | ]

[ ] ∑ [

| ] }.

The optimal policy will be the result of the decisions taken for each activity for each period.

6 Conditions for a Monotone Optimal Policy

In principle, this problem can be solved recursively backwards from period T to identify an optimal action

for each possible state. However, depending on the conditions of the problem, the number of states to

evaluate could grow very large (Serrato, et al. 2007). For this reason, it is desirable to identify a simple

form for an optimal policy. This policy corresponds to a threshold (in terms of the sales and cumulative

returns given a particular capacity level), beyond which the outsourcing action a=1 is optimal.

Sets of conditions exist that ensure that optimal policies are monotone in the system state (Puterman,

1994). One set of conditions stated for the existence of a monotone optimal policy is:

1. [ ] is nondecreasing in for { }.

2. [( ) ] is nondecreasing in for all and { }.

3. [ ] is a superadditive function on .

4. [( ) ] is a superadditive function on .

5. [ ] is nondecreasing in Where

( ) ∑

When all of these conditions are satisfied, there exists a monotone non-decreasing policy that is optimal.

7 Numerical Ilustration

To demonstrate the behavior of the decisions for the MDMs for warehouse and distribution, consider a

particular scenario defined by the parameters:

Page 7: Outsourcing Analysis in Closed-Loop Supply Chains for

L = 5 W = 1

By solving recursively this model, we have the results shown at table 1. For this case, the optimal policy

is to perform the activity internally until period t=5. This example confirms that under certain condition,

there exists a monotone optimal policy. Table 1. Optimal policy for the MDMs for warehouse and distribution

( ) ( )

1 -581.2 -586.6 -581.2 0

2 -497.1 -502.8 -497.1 0

3 -396.5 -401.4 -396.5 0

4 -280.4 -283.4 -280.4 0

5 -151.3 -149.5 -149.5 1

6 0 0 0 1

For the MDM for RL, consider this scenario:

L = 5 W = 1

By solving recursively this model, we have the results shown at table 2. For this case, the optimal policy

is to outsource the activity internally since period t=1. This scenario also shows that the outsource

decision results in an absorbing state. Table 2. Optimal policy for the MDM for RL

( ) ( )

1 -474.8 -461.2 -461.2 1

2 -452.5 -436.0 -436.0 1

3 -400.4 -381.4 -381.4 1

4 -314.0 -293.2 293.2 1

5 -178.5 -167.2 -167.2 1

6 0 0 0 1

8 Conclusions and Future Work

The importance of the CLSC for hazardous materials and the outsourcing as a strategy in order to

minimize costs and focus on the core business has been stated in this study. Also, a MDMs to support the

outsource decision in a CLSC for hazardous materials was developed. The models considers several

elements that are critical in defining the characteristics of an CLSC for hazardous materials, such as

expected sales, uncertainty in the return volume, capacity, operating, shortage and environmental risk

costs.

Some sufficient conditions for the existence of an optimal monotone nondecreasing policy have been

generally described. The existence of an optimal monotone non-decreasing policy implies the presence of

a threshold above which it is optimal to follow an outsourcing strategy for the RL system; otherwise, to

continue performing the RL activities internally. This threshold is defined in terms of a partial ordering

for the system states, where given a fixed capacity at a decision epoch, the states are ordered according to

the sales and cumulative returned units, such that if that volume goes above a particular level, then it is

optimal to follow an outsourcing strategy and take advantage of the economies of scale implied by

involving a 3PL.

Page 8: Outsourcing Analysis in Closed-Loop Supply Chains for

As a future research, the conditions for the existence of an optimal monotone nondecreasing policy must

be verified and proven, as well as an extended numerical evaluation for a case of study. Also, many other

important characteristics of the system could be analyzed in order to take a better decision, as the life

cycle length.

References

1. Alcantara, M., and Gonzalez, T. Modelacion de radios de afectacion por explosions en instalaciones de gas.

CENAPRED. 2001. Mexico.

2. Arroyo, P. Gaytan, J. and Boer, L. A survey of third party logistics in Mexico and a comparison with reports on

Europe and USA. International Journal of Operations and Production Management. 26(6): 639-667, 2006.

3. Boyson, S., Corse, T., Dresner, D. and Rabinovich, E. Managing effective third party logistics relationships:

What does it take? Journal of Business Logistics, 20: 73–100, 1999.

4. Chen, Y. An Inventory Model with Poisson Demand and Linear Lost Sales, International Journal of Service

Science, Management and Engineering. 1(1): 17-20, 2014.

5. Guide, V. D. and Van Wassenhove, L. Special section on closed-loop supply chains. Interfaces 33(6): 1–2,

2003.

6. Guide, V.D. and Van Wassenhove, L. The evolution of closed-loop supply chain research. Operations

Research. 57(1): 10-18, 2009.

7. Handfield, R., Sroufe, R. and Walton, S. Integrating environmental management and supply chain strategies.

Business Strategy and the Environment. 14: 1-19, 2005.

8. Killmer, K. A risk/cost framework for logistics policy evaluation: Hazardous waste management. The Journal

of Business and Economics Studies. 8(1): 51-65, 2002.

9. Lieb, R. and Bentz, B. The use of Third-Party Logistics Services by Large American Manufacturers: The 2004

Survey. Transportation Journal. 44(2): 5-15, 2005.

10. Mentzer, J.T. Fundamentals of Supply Chain Management, Sage, United States. 2004.

11. Mullai, A. and Larsson, E. Hazardous material incidents: Some key results of a risk analysis. WMU Journal of

Maritime Affairs, 7(1): 65-108, 2008.

12. Murray, M. Federal Hazardous Materials Transportation Regulations. About.com. Retrieved February, 2013,

from http://logistics.about.com/od/legalandgovernment/a/Federal-Hazardous-Materials-Transportation-

Regulations.htm

13. Puterman, M.L., Markov Decision Processes: Discrete Stochastic Dynamic Programming, John Wiley & Sons.

New York, NY. 1994.

14. Savaskan, R., Bhattacharya, S. and Van Wassenhove, L. Closed-Loop Supply Chain Models. Management

Science. 50(2): 239-252, 2004.

15. Serrato, M. A., Ryan, S. M. and Gaytán, J. A Markov decision model to evaluate outsourcing in reverse

logistics. International Journal of Production Research, 45(18): 4289 – 4315, 2007.

16. Sherbrooke, C. C. Metric: A Multi-Echelon Technique for Recoverable Item Control. Operations Research,

16(1): 122-141, 1968.

17. Zhu, Q., Sarkis, J, and Lai, K. Confirmation of a measurement model for green supply chain management

practices implementation. International Journal of Production Economics, 111: 261-273, 2008.