Greenhouse gas penalty and incentive policies for a joint economic lot size model with industrial and transport emissions

Article history: Received October 27 2016 Received in Revised Format December 22 2016 Accepted March 1 2017 Available online March 2 2017 This paper presents a joint economic lot size model for a single manufacturer-a single buyer. The purposed model involves the greenhouse gas emission from industrial and transport sectors. We divide the emission into two types, namely the direct and indirect emissions. In this paper, we consider the Government’s penalty and incentive policies to reduce the emission. We assume that the demand of the buyer is normally distributed and partially backordered. The objective is to minimize joint total cost incurred by a single manufacturer-a single buyer and involves the transportation costs of the freight forwarder. Transportation costs are the function of shipping weight, distance, fuel price and consumption with two transportation modes: truckload and lessthan-truckload shipments. Finally, an algorithm procedure is proposed to determine the optimal order quantity, safety factor, actual shipping weight, total emission and frequency of deliveries. Numerical examples and analyses are given to illustrate the results. © 2017 Growing Science Ltd. All rights reserved


Introduction
Global warming as an indicator of climate change occurs as a result of increasing greenhouse gasses (GHGs).Human activities produce the increasingly large amount of GHGs, particularly CO2, which is accumulated in the atmosphere.GHG reduction, an especially CO2 emission reduction is the only way for human survival in facing global warming.The Kyoto Protocol is issued and signed in 1998 by the members of the United Nations (UN) and the European Union (EU), aiming for all participating countries to be committed to reducing the GHG emission amount by 5% to the 1990 level.As a result, many countries have ratified the protocol and have enacted regulations to reduce carbon emission.Policymakers designed regulations such as carbon caps, carbon tax, carbon cap and trade or carbon offset (Benjaafar et al., 2012).An example of the emission standards for diesel engines implemented by EU is that it will be given penalties for vehicles that do not meet minimum standards (Piecyk et al., 2007).Carbon emission can be incurred at various activities.
Freight transportation and manufacturing industry are viewed as leading sectors in economic development.These sectors are the major factors in emission sources and energy consumption.For instance, GHG emissions from transport and industry in the US accounted for 26% and 21% of the total in 2014, respectively (www.epa.gov).GHG emissions from transport sector come from burning fossil fuels for trucks, cars, ships, trains and planes.Meanwhile, GHG emissions from industry come from fossil fuels for energy to produce products from raw materials.The energy consumption of transport and industry sectors is affected by direct and indirect emissions.Direct emissions are the emissions produced from the activities controlled by the companies that are directly related to GHG emissions, such as controlled boilers (generators), furnaces, vehicles, production process and equipment (forklifts) etc. Indirect emissions are the emissions resulted from company activities but are produced by the sources beyond the company.Indirect emission is associated with the amount of energy used and the utility supplying it such as purchased electricity, heat, steam, and cooling.The classification of emissions in this article is shown in Fig. 1.

Fig. 1. The classification of emissions in this article
There are three common carbon policies, namely: carbon emission tax, inflexible cap, the cap-and-trade (Hua et al., 2011;Benjaafar et al., 2012;Hoen et al., 2014).Policymakers can also provide penalties and incentives to reduce emission or impose costs on carbon emissions.A firm can reduce its carbon emission by changing its production, inventory, warehousing, logistics and transportation (Hua et al., 2011;Benjaafar et al., 2012).For more details, the firm can use less polluting generators (boilers), machines or vehicles (direct emissions).While the firm can reduce their carbon emission by using cleaner or environmentally friendly energy sources for indirect emission (Helmrich et al., 2015).This paper developed a mathematical model of a supply chain, i.e.GHG emissions from transport and industrial sectors.The objective was to minimize the integrated costs of supply chain and total emissions produced by these sectors.Subsequently, we analyzed of how imposing on carbon emission tax, penalty and incentive policies impacts the optimal decision variables.
The rest of this paper is organized as follows.The existing literature is reviewed in Section 2. Section 3 describes the problem description, notation, and assumptions.Section 4 develops two mathematical models (with and without penalty and incentive policies) and solution algorithms.Sections 5 and 6 contain numerical example; analysis and discussion, and section 7 concludes the paper.

Literature review
In recent years, research dealing with supply chain inventory management system has attracted attention many scholars.One of the first works that studied the Joint Economic Lot Size model (JELS) was conducted by Goyal (1977).Banerjee (1986) relaxes the assumption of lot-for-lot policy and infinite

Freight Transport Sector Direct Emission Indirect Emission
Manufacturing Industry Sector

Direct Emission
Indirect Emission production rate.Goyal (1988) developed a model with Lu (1995) relaxed the assumption of Goyal (1988) and specified the optimal production and shipment policies when the shipment sizes are equal.Goyal (1995) then developed a model where successive shipment sizes increase by a ratio equal to the production rate divided by the demand rate.Later, Hill (1997) considered the geometric growth factor as a decision variable and he suggested a solution method based on an exhaustive search for both the growth factor and the number of shipments.Based on previous researches, Hill (1999) developed a general optimal policy model.Most of these coordinated models assume as deterministic demand.In fact, the buyer has usually faced lead time and demand uncertainties.Liao and Shyu (1991) developed an inventory model with probability in which lead time is the unique variable.Later, Ben-Daya and Raouf (1994) extended Liao and Shyu's model (1991) model with lead time and ordering quantity as decision variables.Ouyang et al. (1996) generalized Ben-Daya and Raouf's model (1994) model by considering shortages.Moon and Choi (1998) and Hariga and Ben-Daya (1999) further improved and revised Ouyang et al.'s model (1996) by optimizing the reorder point.The integrated inventory models under stochastic environment were developed by Ben-Daya and Hariga (2004), Ouyang et al. (2004) and Jauhari et al. (2011).
Pioneering research works on carbon emission models can be found in Hua et al. (2011) and Wahab et al. (2011).Hua et al. (2011) adopted the emission constraints into classical EOQ model, i.e. carbon emission through a cap-and-trade system under the assumption that carbon emission is linear with the order quantity.Wahab et al. (2011) developed mathematical models: a domestic and an international supply chains that took the environmental impacts.Benjaafar et al. (2012) and Chen et al. (2013) developed emission constraints to a single-level lot sizing (EOQ) and an integrated lot sizing models with the dynamic demand under different carbon emission policies (carbon emission tax, inflexible cap, the cap-and-trade) and analyzed the trade-off between costs and emissions.Jaber et al. ( 2013) developed a mathematical model for a two-level supply chain with incorporating carbon emission tax and penalties to reduce emission amount.This model takes into the emission amount as a function of the production rate.Setup cost, holding cost and emission cost are involved in determining the optimal production rate.Hoen et al. (2014) studied the problem of transportation model selection with carbon emission regulations and stochastic demand.Helmrich et al. (2015) introduced integrated carbon emission constraints in lot sizing problems.The main difference among of the models of Helmrich et al. (2015) with Benjaafar et al. (2012) and Chen et al. (2013) is the type of emission constraints, that their functions of emissions are sensitive with setups and holding cost.Xu, et al. (2015) derived the optimal total emission and production quantities of products overall levels of the cap and analyzed the impact on these optimal decisions.Zanoni et al. (2014) extended the model of Jaber et al. (2013) with Vendor Managed Inventory and Consignment Stock system (VMI-CS).Bazan et al. (2015a) extended and compared the works of Jaber et al. (2013) and Zanoni et al. (2014) by developing the mathematical model for a two-echelon supply chain system that considered the energy used for production.Bazan et al. (2015b;2017) extended their previous work and investigated a reverse logistic model and considered emissions from manufacturing, remanufacturing and transportation activities.
The above-mentioned papers mostly focus on the single-echelon system or two-echelon system without incorporating transportation costs.The inventory-theoretic model with transportation and inventory costs was first introduced by Baumol and Vinod (1970).Lippman (1971) assumed transportation cost with a constant cost per truckload.Langley (1980) considered actual freight rates function into lot sizing decision.Carter and Ferrin (1996) developed a lot-sizing model using enumerations techniques that consider actual freight rate schedules to determining the optimal order quantity.Swenseth and Godfrey (2002) proposed a method to approximate the actual transportation cost with actual truckload freight rates.Abad and Aggarwal (2005) involved transportation cost into inventory model and determining lotsize and pricing decision with downward sloping demand.Nie et al. (2006) and Ertogral et al. (2007) presented an integrated inventory model with transportation cost.Ben-Daya et al. (2008) presented joint economic lot sizing models with different shipment policies.Mendoza and Ventura (2008) presented an algorithm based on a grossly simplified freight rate structure for truckload (TL) or least-then-truckload (LTL) shipments.Rieksts and Ventura (2008;2010) considered a combination of two different modes of transportation: LTL and FTL (full truckload).In the field of supply chain coordination, researches such as Viau et al. (2009) and Kim and Goyal (2009) focused on the integration of inventory and transportation decisions.Yildirmaz et al. (2009) considered joint pricing and lot-sizing decision with transportation.Leaveano et al. (2014aLeaveano et al. ( , 2014b) ) extended Nie et al's model (2006) with distance parameter.Gurtu et al. (2015) developed the inventory models with involving the fuel price.
Addressing the gap between the studies, this paper developed JELS model by incorporating FTL and TL carriers, GHG emission and stochastic demand for a two-level supply chain between a manufacturer and a buyer.We assume that GHG emissions are produced by direct and indirect emissions of industrial and transport sectors.The Government can provide penalties and incentives to reduce emissions.Therefore, we developed a JELS model involving the penalty, incentive and industrial and transport emissions.

Problem description
This paper studied a supply chain system and GHG emission.The GHG emission is one Key Environmental Performance Indicator (KePI) used as a tool to measure a company's sustainability performance of environmental aspect.The GHG Protocol defines direct and indirect emissions as follows (www.ghgprotocol.org): 1. Direct GHG emissions are the emissions from the sources owned or controlled by the reporting entity.2. Indirect GHG emissions are the emissions as the consequences of the activities of the reporting entity but occur at the sources owned or controlled by another entity.
The GHG Protocol has been defining of how the companies should manage and establish three categories of emissions as shown in This paper considered a two-echelon supply chain system consisting of a manufacturer and a buyer.The buyer sells items to the end customers whose demand follows a normal distribution with a mean of D and standard deviation of σ.The buyer orders the item at a constant lot of size Q from the manufacturer.Once an order is placed, a fixed ordering cost Sb incurs.The manufacturer produces the product in a batch size of Qn with a finite production rate P (P > D) with a fixed setup cost Sm.The manufacturer also produces the indirect (EI1) and direct (EI2) emission quantities to the atmosphere from its production facilities.Indirect emission is consumed by electricity (eco), steam (sco), heating (hco), cooling (cco) and loss of energy to produce a production quantity.While boiler (generator) directly produces direct emission to the atmosphere and also produces a production quantity.The manufacturer will pay the cost of emissions corresponding to the number of emissions produced and the Government's carbon emission taxes (CGHG).During the production period, when the first Q units have been produced, the manufacturer may schedule to the third party (freight forwarding services) to pick-up its product.In this policy, the freight will give surcharge per shipment (θ) to the manufacturer and the manufacturer will send the invoice as freight costs to the buyer.The surcharge may consist of the setup cost for the fleet and material handling costs (www.fedex.com).In this policy, the manufacturer will not pay the transportation cost.As a consequence of the pick-up policy, the distance from the location from the freight to the buyer is 2 . We assume the distance between these parties is linear.The freight cost also influences the fuel prices (δ) and fuel consumed by diesel truck (γ).The freight rate, Fx is charged to the buyer.The buyer pays the freight rate to the freight for each shipment weight (Wx) which is scheduled by the freight.In this activity, the freight will produce the transport indirect emission quantity (ET1).The buyer will receive the lot size of Q with average every D/Q unit of time the inventory level until to zero.The buyer will produce the transport direct emission quantity (ET2) in which the emission comes from the material handling process, such as fuel of forklift, etc.Similarly, the buyer also pays these quantity emissions with the carbon emission tax (CGHG).The Government made a penalty (ρ) and incentive (η) policies to reduce direct and indirect emissions from manufacturing industry and freight transport.The penalties are given if total emissions have exceeded the Emission Limit Value (ELV), otherwise, if total emissions are below the ELV then the incentive will be provided so that it can be derived using improvement activities.The partial backorder (πx) and lost sales (π0) are permitted.The system description is illustrated in Figure 2.

Fig. 2. The overview of problem in this paper
The following parameters and decision variables notation are listed below:

Assumptions
In addition, the following assumptions are made in deriving the model: 1.The model assumes a single item with a single-vendor and a single-buyer inventory system and involves a single freight provider.2. We consider the pick-up policy which is offered by the freight provider.The product will be picked by the freight and delivered from the manufacturer's location to the buyer's location.In this policy, the buyer will be charged an additional charge (surcharge) with θ (in dollar) by the freight.3. The product is manufactured with a finite production rate of P, where P > D. 4. The buyer orders a lot of size Q and the manufacturer's produce nQ with a finite production rate P in one setup, but ship quantity Q to the buyer over n times.The vendor incurs a setup cost Sm for each production run and the buyer incurs an ordering cost Sb for each order of quantity Q. 5.The demand X during lead time L follows a normal distribution with mean DL and standard deviation √ .6. Shortages are allowed with partial backorders and lost sales.7.All items are purchased Free On Board (F.O.B) origin.The buyer incurs all the freight costs.

Model
In this section, we formulate an integrated inventory model with GHG emissions penalty and incentive policies, emission from transport and industry sectors, and stochastic demand.

Buyer's total cost per year
The total cost of the buyer is composed of ordering cost, holding cost, shortage cost, freight cost, surcharge cost and carbon emission cost.These components are evaluated as following: 1. Ordering cost.

The ordering cost per year
(1) 2. Holding cost.The expected net inventory level just before receipt of an order is , and the expected net inventory level immediately after the successive order is .Hence, the average inventory over the cycle can be approximated by 2 ⁄ . Therefore, the buyer's expected holding cost per unit time is 2 ⁄ . Using the same approach as in Montgomery et al. (1973), the expected net inventory level just before receipt of a delivery is 1 .The expected shortage quantity at the end of the cycle is given by √ , where, 1 , and Ø, Φ denote the standard normal density function (p.d.f) and c.d.f., respectively.Where, √ .
The holding cost per year = √ 1 √ (2) 3. Shortage cost.As mentioned earlier, the lead time demand X has a c.d.f. with finite mean DL and standard deviation √ .Shortage occurs when X > r, then, the expected shortage quantity at the end of the cycle is given by .Thus, the expected of backorders and lost sales per order is and 1 , respectively.
The shortage cost per year = 1 √ (3) 4. Freight cost.We extended the work of Swenseth and Buffa (1990) and Gurtu et al. (2015) to determine the freight cost.Using the notations: dm is the distance from the manufacturer to the freight (in miles), db is the distance from the freight to the buyer (in miles), δ is the fuel price ($/liter) and γ is the fuel consumption (liters/mile).Let Fx set the lower bound when shipping weights .Hence, the freight rate (Fy) for the actual shipping weight (Wy) is as follows: (4) The freight cost rate per pound per mile can be represented by: Subject to 0 1 Upon substitution of Eq. ( 4) into Eq.( 5) and simplifying, the resulting unit rate is (6) By substituting the Eq. ( 6) into the total freight, we have: (Swenseth & Buffa, 1990;Swenseth & Godfrey, 2002) where the actual shipping weight (Wy = Qw) and 2 represents a pick-up policy from the freight to the manufacturer and from the manufacturer to the buyer.
The freight cost per year = 2 1 2 (8) 5. Surcharge cost.In this policy, we assume that the freight offers pick-up services (by on call) and the products will be picked from the manufacturer and delivered to the buyer with the surcharge per shipment, θ (in dollar).This fee includes ordering cost by phone call, material handling cost, labor cost, wooden pallet collars etc.
The surcharge cost per year (9) 6. Carbon emission cost.As described in the problem description, transport GHG emissions are divided into two parts: indirect and direct transport emissions, with the notations: ∆ is transport indirect emission factor (ton CO2 per liter), γ is the fuel consumption (liters per mile), dm is the distance from the manufacturer to the freight (in miles), and db is the distance from the freight to the buyer (in miles).
Transport indirect emission quantity ∆ 2 (10) For the transport direct emission quantity, we use the notations: ∆ is transport direct emission factor (ton CO2 per lb), w is the weight of unit part (lbs per unit) and Q is order quantity (units).
Transport direct emission quantity ∆ (11) The carbon emission tax (CGHG), hence total transport emission cost per year with indirect and direct emissions is given by: Total transport emission cost per year = (12) The Eq. ( 12) can be rewritten into: Total transport emission cost per year = ∆ ∆ (13) Finally, the total cost for the buyer per year without penalty and incentive policies can be formulated by considering Eqs. ( 1)-(3), Eq. ( 8), Eq. ( 9) and Eq. ( 13).The total cost for the buyer (TCb1).One has:

Manufacturer's total cost per year
Total cost for the manufacturer consists of setup cost, holding cost and carbon emission cost.These components are evaluated as following: 1. Setup cost.The manufacturer produces nQ in one production run time.Therefore, the setup cost per year (15) 2. Holding cost.The manufacturer's inventory per cycle can be calculated by subtracting the buyer's accumulated inventory level from the manufacturer's accumulated inventory level.Hence, the manufacturer' average inventory level per year is given by = 1 1 .
The manufacturer's holding cost per year is = 1 1 (16) 3. Carbon emission cost.As the same describe in the buyer's carbon emission, industrial GHG emissions are divided into two parts: indirect and direct emissions.We used the notations: ∆ is industrial indirect emission factor (ton CO2 per Kwh), eco is the electricity energy consumption (Kwh), sco is the steam energy consumption (Kwh), hco is the heating energy consumption (Kwh) and cco is the cooling energy consumption (Kwh) and Lr is energy loss rate (%).

Industrial indirect emission quantity ∆ (17)
We use the notations: ∆ is industrial direct emission factor (ton CO2 per unit), nQ is production quantity (units) to determine the industrial direct emission quantity.
Industrial direct emission quantity ∆ (18) Hence, total industrial emission cost per year with indirect and direct emissions is given by: Total industrial emission cost per year = (19) The Eq. ( 19) can be rewritten into: Total industrial emission cost per year = ∆ ∆ (20) Finally, the total cost for the manufacturer per year without penalty and incentive policies can be formulated by considering Eqs.(15-16) and Eq. ( 20).The total cost for the manufacturer (TCm1).One has: Accordingly, the integrated total cost for a single manufacturer and a single buyer inventory system without penalty and incentive policies is the sum of the Eq. ( 14) and Eq. ( 21).One has:

Penalty and incentive policies
To formulate the penalty and incentive policies, the Government sets the overall limit on emission (also called "cap") as a basis value at first. Figure 3 illustrates the penalty and incentive policies.

The transport emission model with the penalty and incentive policies
Fig. 3 describes that if total transport emission exceeds the ELVT, ∑ , the buyer would have to pay an exceed emissions (penalty and loss of incentive) from the gap of ∑ and ELVT.Otherwise, if total emission is lower than the ELVT, ∑ then the buyer will receive the Government's incentive and benefit of the penalty.Furthermore, the transport emission model with the penalty and incentive policies is given by: The industrial emission model with the penalty and incentive policies The penalty and incentive from industry sector will be paid by the manufacturer.Similarly, the model of the industrial emission with the penalty and incentive policies is given by: (24) Total transport and industrial emissions with the penalty and incentive policies Accordingly, the formula of transport and industrial emissions with the penalty and incentive policies is the sum of the Eqs.(23-24).One has: (25) By substituting the Eqs.(10)(11)(17)(18) to Eq. ( 25), the Eq. ( 25) can be rewritten into: Finally, adding the Eq. ( 26) to the Eq. ( 22), we find the integrated total cost for a single manufacturer and a single buyer inventory system with the penalty and incentive.One has:

Solution procedure
Our objective is to find the optimal decision variables which minimize the above functions.For fixed n, we take the partial derivatives of Eq. ( 22) with respect to Q and k, respectively.The results for the first model, we obtain: By setting Eq. (27-28) equal to zero and solving for Q and Φ(k).One has: and In order to examine the effect of n on , , , we take the first and the second partial derivatives of Eq. ( 22) with respect to n.One has: This show that , , is a convex function in n, for fixed , .Thus, the search finding the optimal number of deliveries, n * is reduced to finding a local optimal solution.In the same way the first model, to obtain the minimum of Eq. ( 27), take the first partial derivatives of , , with respect to Q and k and setting them to zero.One has: and

Solution algorithm
The following algorithm is developed to find the optimal values for order quantity, safety factor, total emission, and the optimal number of deliveries in one production cycle.
Step 7 Repeat Step 2 -6 until no change occurs in the values of Q and k.The result is denoted by (Q * , k * ) for both models.
Step 8 Compute the cost functions.
Step 9 Set n = n + 1, repeat step 2 for both models.
Step 10 Check and evaluate the cost function., then * , * , * is a set of optimal, therefore the optimal of total emission is * ∆ 2 ∆ ∆ ∆ * * .

Numerical example
This section demonstrates of the models to obtain the optimal solution.Table 2 shows the set data for our example.Implementing two models by optimizing the cost function, we found the optimal solutions as given in Table 3.The minimum joint total cost for the first model (without penalty and incentive policies) is $95,998.58/yearwith the optimal order quantity, Q * = 677,67 units; actual weight, Wy * = 14,908.77lbs; safety factor, k * = 2.25; number of delivery, n * = 3 times, total transport emission quantity, ET * = 37.67 ton CO2 and total industrial emission quantity, EI * = 107.10ton CO2.The emissions results show that the total transport emission quantity is below the ELVT (50 ton CO2) and the total industrial emission quantity is higher than the ELVI (100 ton CO2).With the same parameters, the optimal solutions of the second model (involving the penalty and incentive policies) yields a minimum joint total cost of $92,586.91/year with Q * = 438,05 units; Wy * = 9,637.17lbs; k * = 2.42; n * = 4 times, ET * = 24.50ton CO2 and EI * = 104.39ton CO2.Due to the transport emission quantity below the ELVT and the industrial emission quantity above the ELVI, then the impact is the provision of incentives and penalties by the Government to the buyer and the manufacturer, respectively.So, the penalty and incentive policies have contributed in reducing emissions.The second model gives an impact to a decreased transport emission quantity with a saving of 13.18 ton CO2 (34.98%).It is affected from the actual weight on the second model that is smaller than the actual weight on the first model (9,637.17< 14,908.77),because the actual weight is direct emission (Wy * = Q * w).The saving of the industrial emission comes from production quantity (direct emission) on the second model Q2 * n2 * = 1,752.20units that are smaller than the first model of Q1 * n1 * = 2,033.01units with a saving of 2.71 ton CO2 (2.53%).The saving of total emission quantity is 15.89 ton CO2 (10.97%).The joint total cost saving on both models is $3,411.67(3.55%).
We compared and analyzed the results of independent and integrated policies.Table 4 shows that the total emission for an integrated policy is higher than the independent policy on the first model.A high value on total emission on an integrated policy is contributed from increasing of transport and industry decision variables simultaneously.It can be understood that the optimal actual weight transported on integrated policy (14,908.77units) is higher than independent policy (12.903.51 lbs) in which both are derived from the optimal order quantity.It is the same in production quantity of integrated policy that is higher than independent policy.The integrated policy generates higher order quantity, aiming to reduce joint total cost.The total cost of independent and integrated policies are $96,784.60/yearand $95,998.58/year,respectively.Therefore, the integrated policy gives total cost saving of $786.03/year or 0.81%.The second model has a similar discussion with cost saving in the amount of 0.24%.

Table 4
The comparison of independent and integrated policies for model 1 and model 2 In the next case, we included the freight schedule data and illustrated the above solution procedure.Table 5 presents the actual freight rate schedule data by considering shipping weight and distance.The data were adopted from Swenseth and Godfrey (2002) and Leaveano (2014b).The freight rates were redefined from the freight rate per pound to freight rate per pound per mile.For instance, we assume FTL can be delivered 600 miles with weight equal and more than 18,257 lbs and freight cost is a constant charge ($1,110/shipment).Therefore, the freight rate per pound per mile is obtained by dividing freight rate per shipment with the highest break point and distance.Unlike the case of 10,000 -18,000 lbs, the freight rate per pound is a variable rate based on the load transported by the LTL.The freight rate per pound per mile can be obtained by dividing freight rate per pound with distance.

Table 5
The freight rate schedule data We discussed the effect of various full truckload capacities (Wx) from 25,000; 20,000, 15,000; 10,000; 7,500 and 5,000 lbs.The results for this example are summarized in Table 6.The results obtained for various values of the full truckload capacity are lower total emission, especially emission from the transport sector.However, the opposite effect on the total cost will increase.The increase in the total costs is due to an increase in the manufacturer's setup and emission.If the full truckload capacity is decreased gradually then it will give an opportunity for the manufacturer to increase its production setup.
For instance on the 2nd model, Wx = 7,500 and 5,000 lbs, the results n * = 5 times and 8 times, respectively.Hence, the impact is the manufacturer's emission also increased (103.93 to 105.02 ton CO2).Because the emission has exceeded from ELVI, then the penalty will be given to the manufacturer.

Analysis and discussion
In this section, we studied and analyzed the effect of various parameters to determining the optimal decision variables such as the optimal order quantity, safety factor, the number of deliveries and total emission and subsequently on joint total cost for both models.The parameter dividing 3 categories, there are the general parameter, transport sector and industrial sector (Table 7).

Effect of changing in the Government's penalties and incentives
We analyzed the effect of changing in the penalty on total emissions and total costs.From the assumptions used in the development model that has been described earlier, the Government's penalty should be larger than the Government's incentive.We use incentive $125/year/ton CO2.So, the penalties used in this analysis are 125,150,200,250,350,500,700,1000,1200 and 1500 ($/year/ton CO2).The effects of changing in the penalty on the decision variables and total costs are shown in Table 8.

Table 8
The effect of changing in penalties and incentives on decision variables and total cost Penalty Model 1 Model 2 Saving  In Fig. 4, it can be observed that the penalty policy can reduce total emission and total cost on the second model.As we know, the first model is not considered as the penalty policy; hence the results show that the total emission and total cost are the constant rates.The comparison of these models produces total emission saving from 5% to 20% and total cost saving from 1% to 33%.So, we conclude that the Government's penalty will give impact on the reduction in total emission and total cost of the parties.Table 8 shows the effect of varying the incentives to determine the decision variables and total costs.In contrast to the penalty policy, the incentive should be lower than the penalty.The penalty, in this case, is $300/year/ton CO2.Hence, the incentives used in this analysis are 5, 10, to 300.Fig. 5 shows that the second model produces the total emission, and the total cost is lower than without penalty and incentive policies.Thus, we conclude with the same penalty conclusion that the Government's incentive can reduce the total emission and the total cost with saving on total emission and total cost of 9% -13% and 1% -6%, respectively.

Effect of changing in carbon emission tax
Carbon emission tax has an impact to determining the amount of transport and industrial emissions and total costs on both the models.The parties will pay a carbon emission tax ($/ton CO2) on how much GHG emission produced by transport and industrial sectors.In this case, we used the initially of a carbon tax of $20/ton CO2 and total ELV of 150 ton CO2 (ELVI = 100 ton CO2 and ELVT = 50 ton CO2).The behavior of the total emission and total cost for both models is shown in Figure 6.In this experiment, the carbon tax is associated with the optimal order quantity, and it affects the total cost and total emission.In the case when carbon taxes are more $20/ton CO2, the impact is the increase in the total emission and total cost, simultaneously.The carbon tax is $55/ton CO2 will cause the second model to be inefficient compared to the first model.Furthermore, the increase in carbon tax also leads to a large gap of the total emission between the first and second model (saving of 11% -17%).It is affected by penalty and incentive imposed on the second model.Finally, we noted that the carbon emission tax influences the total emission and total cost.Fig. 6.The effect of changing in carbon emission tax on total cost and total GHG emission

Effect of changing in Emission Limit Value
Here, we studied the effect of varying ELV from transport and industry sectors.In this case, we set total ELV by 150 ton CO2.The values of ELV transport start from 10 to 100 ton CO2 (low to high) and contrary to the values of ELV industry start from 140 to 50 ton CO2 (high to low). Figure 7 portrays a simple relationship between total cost and two ELVs.Based on the ELV transport low value and the ELV industry high value, the impact is the total cost of the buyer and the manufacturer will be increased and decreased, respectively.However, for a high value of ELV transport and low value of ELV industry, the impact is the total cost of the buyer and the manufacturer will be decreased and increased, respectively.There is a

Effect of changing in production rate
In this example, we discussed and investigated the varying values of production rate.As shown in Figure 8, the increase in production rate will make the total cost of the both models increase.Likewise, the total emission will also increase.The increase in the total emission is influenced by a large increase in the transport emission while the industrial emission declines.It affects to an increase in the buyer's holding cost, carbon emission cost, and penalty, but the manufacturer's holding cost and carbon emission cost and penalty will be decreased.

Effect of changing in average and standard deviation of demand
In the Fig. 9 and Fig. 10 show the behavior of the stochastic environment.We may see that if the average and standard deviation rises, then the total cost will also elevate in both models.If the demand average increase, the buyer's holding cost, carbon cost and safety stock on the both models will also increase.In contrast to demand standard deviation, having more demand standard deviation will affect to reduce the buyer's holding and carbon costs, and safety stock is pretty much to handle demand variation.   . (11-15) show the effect of varying parameters from the transport sector, such as fuel price and consumption, distances, transport direct and indirect factors on total emissions and total cost of a supply chain.In the second model, the penalty and incentive policies are considered.The graphs show that the cost can be significantly increased by distances and direct emission factor and the second model can reduce total cost and total emission.The increase in total cost comes from the buyer's cost, otherwise the manufacturer's cost.

Sensitivity parameter from the industry sector
The observations of the varying parameter from industrial parameters, an especially loss rate of energy (electricity, steam, heating, and cooling), industrial indirect and direct emissions factors on total cost and total emission are illustrated in Figs.(16-18.In this case, the total cost is affected by the increase in the optimal production quantity and total industrial emission; hence the impacts on the total cost of the manufacturer are the manufacturer's holding cost, setup cost, carbon emission cost and penalty cost.As we can see, the total cost can be significantly increased by the industrial direct emission factor.Finally, we concluded that these parameters contributed to the increase in total emission and total cost.

Conclusion
In this study, we formulated a supply chain inventory model considering carbon emission tax, the penalty and incentive policies.Here, we also considered stochastic demand and industrial and transport GHG emissions.The numerical example and analysis showed that these policies and stochastic environment can influence the decision-makers in determining the optimal order quantity and reduce total emission resulting from transport and industrial sectors.We also examined the relationship between relevant parameters of these sectors and the total emission associated with total cost.Significant cost savings on total cost of the entire supply chain can also be achieved by considering the penalty and incentive policies.Our findings provided some useful insights to practitioners.This paper contributed to an integrated inventory literature with GHG emission.
The proposed models have limitations.The proposed model in this paper could be extended in various directions.Future research may consider multi-manufacturer and multi-buyer.The other indirect emission may be involved such as waste disposal.In addition, other policies to reduce GHG emissions can be incorporated into inventory models such as investment cost of emission reductions, recycling, remanufacturing, cleaner production or green manufacturing etc.

Fig. 3 .
Fig. 3.The illustrated of penalty and incentive formulas

3
Evaluate the optimal Q * .(Step 3.1) For the first model, find * by substituting ψ(k0) into Eq.(30) (Step 3.2) For the second model, find * by substituting ψ(k0) into Eq.(34).Step 4 Calculate actual shipping weight, * .If is satisfied go to Step 5. Otherwise, go to Step 6 if truckload constraint is not satisfied .(Step 4.1) For the first model, Revised the optimal lot size and go to Step 6. (Step 5.1) For the first model, * .(Step 5.2) For the second model, * .Step 6 Determine of Φ(ki * ) then find ki from Φ(ki) by checking the normal table.
go to Step 9, otherwise go to step 11 on the second model.Step 11 The optimal decision variables.(Step11.1)For the first model, * , * , * * , * , , then * , * , * is a set of optimal, therefore the optimal of total emission is *

Fig. 4 .
Fig. 4. The effect of changing in penalty on total cost and total emission

Fig. 5 .
Fig. 5.The effect of changing in incentive on total cost and total GHG emission the trade-off among total cost of the buyer and the manufacturer and the values of ELV.

Fig. 7 .
Fig. 7.The effect of changing in ELV on total cost and total emission

Fig. 8 .
Fig. 8.The effect of changing in production rate on total cost and total emission

Fig. 9 .
Fig. 9. Effect of changing in demand average on total cost and total emission

Figs
Figs.(11)(12)(13)(14)(15) show the effect of varying parameters from the transport sector, such as fuel price and consumption, distances, transport direct and indirect factors on total emissions and total cost of a supply chain.In the second model, the penalty and incentive policies are considered.The graphs show that the cost can be significantly increased by distances and direct emission factor and the second model can reduce total cost and total emission.The increase in total cost comes from the buyer's cost, otherwise the manufacturer's cost.

Fig. 16 .Fig. 18 .
Fig. 16.Effect of changing in loss rate of electricity, steam, heating and cooling on total cost and total emission

Table 2
Parameters and values for numerical example

Table 3
The comparison of model 1 and model 2

Table 7
Categories of sensitivity parameter