In the traditional inventory system, it was implicitly assumed that the buyer pays to the seller as soon as he receives the items. In today’s competitive industry, however, the seller usually offers the buyer a delay period to settle the account of the goods. Not only the seller but also the buyer may apply trade credit as a strategic tool to stimulate his customers’ demands. This paper investigates the effects of the latter policy, two-level trade credit, on a retailer’s optimal ordering decisions within the economic order quantity framework and allowable shortages. Unlike most of the previous studies, the demand function of the customers is considered to increase with time. The objective of the retailer’s inventory model is to maximize the profit. The replenishment decisions optimally are obtained using genetic algorithm. Two special cases of the proposed model are discussed and the impacts of parameters on the decision variables are finally investigated. Numerical examples demonstrate the profitability of the developed two-level supply chain with backorder.
1. Introduction
Since the introduction of the classical economic order quantity (EOQ) model by Harris [1], many researchers have extended it in several ways. One of the discussed issues in this area is including delay in payment, as an incentive system, in the EOQ or economic production quantity (EPQ) models [2]. According to Piasecki [3] and Molamohamadi et al. [4], different types of delay in payment can be classified as (1) pay as sold, (2) pay as sold during a predefined period, (3) pay after a predefined period, and (4) pay at the next consignment order.
In the first type of delay in payment, so-called consignment inventory, the buyer defers paying for the items till they are sold to the customers. The second type refers to the case that the buyer pays off as soon as he sells the items to the customers during a predefined period. At the end of this period, he can either pay for the remaining items in his stock or return the unsold items to the vendor. According to the third type of delay in payment, which is known as trade credit in the literature, the buyer must pay to the vendor at the end of a predetermined period. During the credit period, the buyer sells the items to his customers and accumulates revenue and earns interest. After this period, however, he would be charged a higher interest if the payment is not settled. Based on the fourth type, the payment for each order would be settled at the time of the next replenishment order. Therefore, there is one replenishment cycle delay for each received order in this type. The advantage of delay in payment contract to the buyer is obvious; he does not need to invest his capital in inventory and can earn interest for the items he sells. Moreover, the vendor can apply this agreement as a sales promotional tool for attracting new buyers and selling new and unproven products.
As this paper focuses on the third type of delay in payment, we review the literature related to trade credit (please refer to Seifert et al. [5] and Molamohamadi et al. [6]). Goyal [7] presented an EOQ mathematical model for determining the economic order quantity where the supplier offers a fixed credit period to the retailer to settle the account. His paper is the infrastructure for its following studies. Aggarwal and Jaggi [8] extended Goyal [7] by considering deterioration rate and assuming that the customer accumulates the sales revenue and earns interest during the credit period and beyond it. Jamal et al. [9] included shortages in the proposed model by Aggarwal and Jaggi [8] to generalize it. Teng [10] modified Goyal’s [7] model by distinguishing between the unit purchase cost and the selling price. By applying an EPQ model, Chung and Huang [11] further developed Goyal [7] by assuming finite replenishment rate. Huang [12] considered a two-level trade credit and deduced Goyal [7] as a special case of his research. In a two-level trade credit, not only does the vendor offer trade credit to the buyer, but the retailer also provides a credit period to his customers.
Huang [13] investigated the retailer’s inventory policy under two-level trade credit with unequal selling and purchasing prices and extended Teng [10] and Huang [12] by considering the retailer’s limited storage space. Teng and Goyal [14] complemented the shortcoming of Huang [12]’s model in calculating the earned interest from the time the retailer is paid by his customers, not from time zero. They further extended his paper by relaxing the limitations on the selling and purchasing prices, as well as retailer and customer’s credit periods. Huang [15] established an economic order quantity model in which the supplier provides the retailer partially permissible delay in payment for the order quantities smaller than a predetermined quantity and offers him complete trade credit otherwise. Huang [16] differentiated between the purchase cost and the selling price and presented an EPQ model under two levels of trade credit to generalize Chung and Huang [11] and Huang [12]. Teng and Chang [17] reformulated Huang’s [16] model by calculating the retailer’s earned revenue from the time he is paid by the customers and further extended his model by assuming that the customer’s credit period is not inevitably smaller than the retailer’s delay period. Su [18] developed a supplier-buyer inventory model in which the supplier’s selling price is dependent on his productions cost. He further assumed that the latter is affected by the market demand and production rates, and the production rate is sensitive to the price dependent market demand. He finally obtained the optimal pricing, ordering, and inventory decisions of a profit maximizing system under trade credit contract.
Dye and Ouyang [19] proposed an EOQ mixed-integer nonlinear programming model under two levels of trade credit for deteriorating items with time-varying demand and applied a traditional particle swarm optimization (PSO) algorithm to determine the optimal selling price and replenishment policy. Dye [20] applied PSO algorithm to obtain the optimal replenishment decisions of an EOQ model with price and time dependent demand, partially backlogged items, and deterioration under two-level trade credit policy. Mahata [21] presented a generalization of Goyal [7], Chung and Huang [11], Huang [12], and Huang [16] where an economic production quantity model is formulated for exponentially deteriorating items under two levels of trade credit with the assumption that the customer’s partial credit period is not necessarily smaller than the retailer’s complete credit agreement. Lou and Wang [22] formulated an EPQ inventory model for defective items under two independent levels of trade credit to extend some of the previous studies including Goyal [7], Teng [10], Huang [12], and Teng and Goyal [14].
Having applied cuckoo search algorithm, Molamohamadi et al. [23] solved an EPQ model of an exponentially deteriorating item with price-sensitive demand under trade credit contract and allowable shortages. Chern et al. [24, 25] formulated a supply chain under trade credit financing with noncooperative Stackelberg and Nash equilibrium solutions, respectively. Chen and Teng [26] determined the retailer’s optimal cycle time by developing an EOQ model under trade credit policy for continuously deteriorating items with maximum lifetime. Chen et al. [27] reformulated Mahata’s [21] proposed model by calculating the earned and paid interest based on the facts that (i) the retailer earns interest from the time he is paid by the customers and (ii) the retailer’s interest payable must be calculated based on the total items in stock, not only on the unsold finished products. Some of the previous models such as Goyal [7] and Teng [10] are mentioned as special cases of their proposed model. Chen et al. [28] complemented some shortcomings of Huang’s [15] mathematical expressions and figures and proposed a simple method to solve the inventory problem.
Reviewing the literature clarifies that trade credit has received great attention of the researchers, while it has still outstanding space for further studies. For instance, it is mostly assumed that the demand rate is constant. However, recently, Teng et al. [29] developed an EOQ inventory model under trade credit contract in which demand has an increasing function of time. Although, their model can be considered as a generalization of its preceding studies, it has great potential for further extension. As it is stated in Jamal et al. [9], when delay in payment is assumed, shortages are more important as they affect the quantity ordered to benefit from the delay in payment. Moreover, in practices, not only does the supplier propose a delay period to the retailer, but the retailer also allows his customers to defer their payment. Thus, considering two levels of trade credit contributes to practical situations.
Considering the gaps in the literature, we extend the proposed model of Teng et al. [29] to the case of backorder and two-level trade credit. It is assumed here that the retailer’s credit period offered by the supplier is greater than the customer’s delay period offered by the retailer. The proper replenishment policy and the maximum profit of the retailer are then obtained by applying genetic algorithm (GA). It is finally deduced that the inventory system of Teng et al. [29] and the traditional inventory system are special cases of our proposed model and the results obtained in this paper are compared with these cases.
The rest of this paper is organized as follows. Section 2 lays out the notations and assumptions used in the modeling of the problem. The model is formulated in Section 3 and the two special cases of the presented model presented are discussed in Section 4. The genetic algorithm, used for solving the model, is described in Section 5. Regarding the numerical examples of Teng et al. [29], Section 6 provides some numerical examples and the conclusion is finally discussed in Section 7.
2. Notations and Assumptions
The following notations and assumptions are used in this paper.
2.1. Notations
ordering cost per order,
unit purchasing cost,
unit selling price (with s>c),
unit stock holding cost per unit of time (excluding interest charges),
unit backorder cost of retailer per unit of time,
interest which can be earned per $ per unit of time by the retailer,
interest charges per $ in stocks per unit of time by the supplier,
the retailer’s trade credit period offered by supplier in years,
the customer’s trade credit period offered by retailer in years,
the net profit of the retailer per unit of time,
the inventory cycle time,
the inventory cycle time with positive stock,
the retailer’s order quantity,
the inventory consumed in T1.
2.2. Assumptions
The demand is assumed to have an increasing function of time and is defined by D(t) as follows:
(1)Dt=a+bt,
where a and b are nonnegative constants and t is the growth stage of the product life cycle.
Shortages are allowed and completely backordered.
The lead time is zero.
The retailer is offered a delay period (M) by the supplier and provides the customers with a shorter credit period (N). The retailer pays off to the supplier at the end of the credit period (M) and pays for the interest charges on the remaining items in his stock with rate Ic during M,T1 if T1≥M. When the credit period is greater than the positive-stock replenishment cycle, the retailer would not be charged by any interest after settling the account.
The retailer accumulates revenue and earns interest with rate Ie from N to M.
3. Mathematical Formulation
According to the notations and assumptions discussed in previous section, the retailer’s inventory system is depicted in Figure 1 and can be explained as follows. The retailer orders and receives Q units of items at time zero and sends the previously backlogged orders to the customers immediately. The remaining inventory (Q1) depletes gradually due to the customers’ demand and becomes zero at time t=T1. From T1 to T there is no inventory on hand and the arriving orders would be backlogged to the next cycle.
Inventory level at the manufacturer from 0 to T.
Now the inventory level at time t can be described by the following differential equation:
(2)dI(t)dt=-Dt=-a+bt,0≤t≤T,
with the boundary condition I(T1)=0. Therefore, the solution to (2) would be
(3)It=aT1-t+12bT12-t2.
According to (3), the inventory levels at the beginning and at the end of the replenishment cycle are as (4) and (5), respectively:
(4)Q1=I0=aT1+12bT12,(5)IT=Q1-Q=aT1-T+12bT12-T2.
Since the retailer is encountered with backorder during T1 to T, and as T1<T, (5) is smaller than zero. So, the retailer’s order size per cycle time can be obtained as follows:
(6)Q=I0+IT=aT+12bT2.
The retailer’s net profit consists of the following elements.
(a) The ordering cost =A.
(b) The inventory holding cost excluding interest charges =h∫0T1I(t)dt=h((1/2)aT12+(1/3)bT13).
(c) The backorder cost =-cb∫T1TI(t)dt=cb(1/2)a(T-T1)2+(1/6)b(T3-3TT12+2T13).
(d) The purchasing cost =cQ=caT+(1/2)bT2.
(e) The sales revenue =sQ=saT+(1/2)bT2.
(f) Interest payable: for calculating the interest payable, two cases must be considered.
Case 1 (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M55"><mml:msub><mml:mrow><mml:mi>T</mml:mi></mml:mrow><mml:mrow><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msub><mml:mo>≤</mml:mo><mml:mi>M</mml:mi></mml:math></inline-formula>).
Since the delay period is greater than the cycle time in this case, the retailer does not have any stock on hand at the time of paying to the supplier. So, the interest charged in this case equals zero.
Case 2 (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M56"><mml:msub><mml:mrow><mml:mi>T</mml:mi></mml:mrow><mml:mrow><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msub><mml:mo>≥</mml:mo><mml:mi>M</mml:mi></mml:math></inline-formula>).
In this case, the retailer would be charged for the on-hand inventory between M and T1 with the rate Ic. Therefore, the interest payable is
(7)cIc=∫MT1I(t)dt=cIc12aT1-M2+12bT12(T1-M)cIcIc-16bT13-M3.
(g) Interest earned; according to assumption 5, the retailer would start earning interest for the items sold, from the time they are being paid by the customers (N) until M, when the retailer must pay to the supplier. Hence, there are two cases to discuss.
Case 1 (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M63"><mml:mi>M</mml:mi><mml:mo>-</mml:mo><mml:mi>N</mml:mi><mml:mo>≤</mml:mo><mml:msub><mml:mrow><mml:mi>T</mml:mi></mml:mrow><mml:mrow><mml:mn>1</mml:mn></mml:mrow></mml:msub></mml:math></inline-formula>).
The retailer receives the payment for Q-Q1 items from the customers at time N and then the payment will be settled with the rate a+bt for the rest items. This case is illustrated in Figure 2.
Supposing the revenue function as R(t), for N≤t≤T1+N, we have
(8)dR(t)dt=a+bt,N≤t≤T1+N,R(N)=0.
The solution to this equation is
(9)Rt=at-N+12bt-N2.
So, the function for N≤t≤T1+N is
(10)Rt=at-N+12bt-N2.
Now the interest earned per cycle when M-N≤T1 can be computed as
(11)sIe∫NMR(t)dt+(Q-Q1)(M-N)=sIea2M-N2+b6M-N3=sIe-e+aT-T1+12bT2-T12M-N.
Accumulation of interest earned when M-N≤T1.
Case 2 (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M76"><mml:mi>M</mml:mi><mml:mo>-</mml:mo><mml:mi>N</mml:mi><mml:mo>≥</mml:mo><mml:msub><mml:mrow><mml:mi>T</mml:mi></mml:mrow><mml:mrow><mml:mn>1</mml:mn></mml:mrow></mml:msub></mml:math></inline-formula>).
According to Figure 3, the retailer’s earned interest in this case is
(12)sIe∫NT1+NR(t)dt+(Q-Q1)(M-N)+Q1(M-T1-N)=sIea2T12+b6T13+a(T-T1)+b2(T2-T12)=sIe=×M-N+aT1+b2T12M-T1-N.
Accumulation of interest earned when M-N≥T1.
Based on the charged and earned interests, three general cases must be considered for modeling the retailer’s net profit per unit time (NP(T)) which is calculated as (revenue − ordering cost − purchasing cost − holding cost − backordering cost − interest payable + interest earned)/T:
(13)NP(T1,T)=NP1(T1,T),M≤T1NP2(T1,T),T1≤M≤T1+NNP3(T1,T),M≥T1+N,
where
(14)NP1(T1,T)=s-cTaT+12bT2-AT-hT12aT12+13bT13-cbT12aT-T12+16bT3-3TT12+2T13-cIcT12aT1-M2+12bT12(T1-M)-16bT13-M3+sIeTa2M-N2+b6M-N3+sIeT-++aT-T1+12bT2-T12M-N,NP2(T1,T)=s-cTaT+12bT2-AT-hT12aT12+13bT13-cbT12aT-T12+16bT3-3TT12+2T13+sIeTa2M-N2+b6M-N3+sIeT=++aT-T1+12bT2-T12M-N,NP3(T1,T)=s-cTaT+12bT2-AT-hT(12aT12+13bT13)-cbT12aT-T12+16b(T3-3TT12+2T13)+sIea2T12+b6T13+a(T-T1)+b2T2-T12+sIe=+×M-N+aT1+b2T12M-T1-N.
Based on these models and according to (13), it can be verified that, in T1=M, NP1(T1,T)=NP2(T1,T), and T1=M-N implies that NP2(T1,T)=NP3(T1,T).
Thus, our problem is
(15)MaximizeNP(T1,T)SubjecttoT1≤T.
This is a nonlinear maximization problem which is going to be solved to obtain the optimal values of T1 and T and compute Q1 and Q accordingly based on (4) and (6).
4. Special Cases
This section discusses the two special cases of the inventory system proposed in the last section, the model of Teng et al. [29] and the traditional EOQ model with backorder.
Case 1.
Setting N=0, cb=0, and T1=T, the following result is achieved:
(16)NP1(T1,T)=s-cTaT+b2T2-AT-hTa2T2+b3T3-cIcTa2T-M2+b2T2T-M-b6T3-M3+sIeTa2M2+b6M3,M≤T,(17)NP3(T1,T)=s-cTaT+b2T2-AT-hT(a2T2+b3T3)+sIeTa2T2+b6T3+aT+b2T2M-T,+sIeTa2T2+b6T3+aTaT+b2T2+M≥T,
which are equal to ∏2(t) and ∏1(t) in Teng et al. [29], respectively.
Moreover, it can be certified that when M=T, NP1(T1,T)=NP2(T1,T)=NP3(T1,T).
Case 2.
Setting the values of M and N to zero, we would have the traditional inventory system with backorder where the payments are settled without delay. In this case, Ic refers to the opportunity cost the retailer incurs for keeping the inventory in his stock during the positive inventory time. So, the model for traditional inventory system with backorder is
(18)NP1(T1,T)=1Ts-caT+b2T2-A-h+cIca2T2+b3T3.
5. Solution Procedure
Since analytical solving of the nonlinear formulated model in Section 3 is difficult, we have employed a metaheuristic algorithm to find the proper values for T1, T, Q1, and Q, while the net profit is maximized. Recently, genetic algorithm has attracted considerable attention and has been used successfully to the supply chain related problems (see [30, 31]). So, we apply genetic algorithm here to determine the proper replenishment strategy.
GA starts with a population of a set of randomly produced representative solutions to the problem going to be solved. Each individual is called a chromosome and consists of a string of symbols. These chromosomes evolve through succeeding iterations, known as generations, and their fitness is evaluated in every generation. Then, genetic operators including crossover, mutation, and selection are used to create offspring, population of the next generation. Crossover combines the information of the two candidate solutions of the parents by exchanging alternate pairs of their accidentally selected crossing sites to produce two offspring. In mutation, the values of a chromosome’s genes would be changed randomly and selection reproduces the new population by the most highly ranked chromosomes of the existing generation. Using these evolutionary operators, GA is able to modify and pass on the best solutions to the next population. Hence, each generation yields improved offspring and generally the population moves to better solutions, ideally to the global optimum [32]. Pseudocode 1 illustrates the pseudocode of the proposed algorithm.
<bold>Pseudocode 1: </bold>The pseudocode of the applied genetic algorithm.
Algorithm: GA (PopSize, PC, PM, Genes)
//Initialize generation 0:
k=0;
Pk = a population of “PopSize” randomly-generated individuals;
Do
{
//(1) Crossover
Select “PC × PopSize” members of Pk randomly;
Pair them off to produce offspring and save them as C1;
//(2) Mutation
Select “PC × PopSize × Genes” genes of Pk randomly;
Mutate them and save the changed individuals as C2;
//(3) Selection: Creating generationk + 1
Compute fitness (i) for each i∈ (Pk∪C1∪C2);
Select the “PopSize” best of them for the next generation;
//Increment:
k=k+1;
}
While the number of generations is less than a hypothetical max generation;
Return the fittest individual from Pk;
The parameters of the applied algorithm are set to the following values. Population size, probability of mutation, and probability of crossover are equal to 150, 0.02, and 0.8, respectively. For assuring that T1 is smaller than T, penalty method with the value 1015 is used. The algorithm continues till the termination criterion is met; it is met when the number of the iterations reaches 1000. Moreover, for restricting the values of the variables (T1 and T) and getting better results, the maximum value is considered 0.2. Compared with the cases of higher limits, this value resulted in better net profit for the applied examples in Section 6.
6. Numerical Examples
Following the four examples provided in Teng et al. [29], we present some numerical examples in this section to show the results of the applied genetic algorithm, examine whether the proposed EOQ model under two-level trade credit with backorder is profitable to the retailer, test the sensitivity of the model to the changes of the input parameters, and compare the results with the model of Teng et al. [29] and the traditional EOQ inventory system with backorder.
Example 1.
Suppose that a=3600 units, b=2400, M=1/12 year, s= $1 per unit, c= $0.5 per unit, A= $10 per order, h= $0.5 per unit per year, Ic= $0.155 per year, and Ie= $0.08 per year. Applying genetic algorithm, the proper replenishment policy and net profit of the retailer for different values of N and cb are calculated and shown in Table 1.
Obtained replenishment policy for Example 1.
cb
5
10
15
20
25
N=1/15
T1
0.1410
0.1402
0.1400
0.1399
0.1399
T
0.1562
0.1478
0.1450
0.1437
0.1429
Q1
531.4042
528.2017
527.4393
527.1146
526.9379
Q
591.7789
558.1185
547.3368
542.0218
538.8565
NP
1690.9209
1683.4107
1680.8347
1679.5318
1678.7452
N=1/14
T1
0.1411
0.1403
0.1401
0.1400
0.1400
T
0.1563
0.1479
0.1451
0.1438
0.1430
Q1
532.0320
528.8021
527.9499
527.6089
527.4223
Q
592.1829
558.6094
547.7684
542.4564
539.2929
NP
1690.6480
1683.2006
1680.6468
1679.3554
1678.5758
N=1/13
T1
0.1413
0.1401
0.1402
0.1401
0.1399
T
0.1564
0.1476
0.1452
0.1439
0.1429
Q1
532.5723
528.0781
528.3549
527.9951
527.1334
Q
592.4417
557.6453
548.0746
542.7680
538.9143
NP
1690.3864
1683.0136
1680.4869
1679.2091
1678.4376
In the whole table NP=NP3.
Special Cases. (i) Example 1 presented in Teng et al. [29] is a special case of this example in which N=0, cb=0, and T1=T. The result of GA for this case is T=0.1340, Q=503.7677, and NP1=1682.7105 which was also obtained in Teng et al. [29].
(ii) Zero values for M and N leads this example to the conventional inventory system with shortage in which no delay in payment is assumed. The results of this special case are shown in Table 2.
Obtained replenishment policy for the traditional case of Example 1.
cb
5
10
15
20
25
T1
0.1338
0.1334
0.1333
0.1333
0.1334
T
0.1492
0.1411
0.1385
0.1371
0.1365
Q1
503.0591
501.4132
501.3869
501.1447
501.7134
Q
563.9292
531.6733
521.5435
516.2486
513.8079
NP
1674.7187
1666.6643
1663.8890
1662.4829
1661.6331
Example 2.
Considering the second example of Teng et al. [29], we have D=3600+2400t units, M=1/12 year, s= $1 per unit, c= $0.5 per unit, A= $10 per order, h= $1 per unit per year, Ic= $0.13 per year, and Ie= $0.08 per year. Table 3 shows the solutions obtained for this example when backorder and customers’ trade credit are assumed in the model.
Obtained replenishment policy for Example 2.
cb
5
10
15
20
25
N=1/15
T1
0.0822
0.0839
0.0846
0.0850
0.0851
T
0.0989
0.0924
0.0903
0.0893
0.0886
Q1
303.8926
310.4945
312.9991
314.6264
315.1490
Q
367.6386
342.9968
334.8354
331.0722
328.3378
NP
1605.1604*
1590.9873
1585.8947
1583.2705
1581.6701
N=1/14
T1
0.0822
0.0839
0.0846
0.0850
0.0868
T
0.0989
0.0924
0.0903
0.0893
0.0903
Q1
304.0934
310.4480
313.2649
314.7091
321.3629
Q
367.5960
342.7112
335.0242
331.1502
334.7725
NP
1604.7161*
1590.6403
1585.5848
1582.9800
1581.3497
N=1/13
T1
0.0824
0.0841
0.0847
0.0850
0.0853
T
0.0989
0.0926
0.0904
0.0893
0.0887
Q1
304.6805
311.2158
313.6500
314.7649
315.7399
Q
367.9643
343.4857
335.3263
331.0772
328.8301
NP
1604.2874*
1590.3300
1585.3184
1582.7368
1581.1625
*NP=NP2. For the rest, NP=NP1.
Special Cases. (i) One of the special cases of this example when N=0, cb=0, and T1=T reports the same results of the second example of Teng et al. [29] as T=0.0823, Q=304.2236, and NP3=1586.6884.
(ii) The results of traditional inventory system with allowable shortages of Example 2 when M=N=0 are presented in Table 4.
Obtained replenishment policy for the traditional case of Example 2.
cb
5
10
15
20
25
T1
0.0785
0.0804
0.0811
0.0819
0.0876
T
0.0952
0.0890
0.0869
0.0863
0.0913
Q1
290.0597
297.2694
299.9274
302.9720
324.4927
Q
353.7538
329.8431
321.8196
319.5002
338.7249
NP
1595.8190
1581.0432
1575.7132
1572.9602
1570.7055
Example 3.
Suppose that a=3600 units, b=2300, M=1/7.5 year, s= $1 per unit, c= $0.5 per unit, A= $10.2278 per order, h= $0.5 per unit per year, Ic= $0.17 per year, and Ie= $0.08 per year. With different values assigned to N and cb, the solutions obtained for this example by GA are demonstrated in Table 5.
Obtained replenishment policy for Example 3.
cb
5
10
15
20
25
N=1/15
T1
0.1440
0.1445
0.1453
0.1443
0.1443
T
0.1597
0.1523
0.1506
0.1483
0.1474
Q1
542.3339
544.1482
547.5471
543.6133
543.3533
Q
604.1778
575.0968
568.2234
559.0530
555.6941
NP
1694.5580
1686.7691
1684.0872
1682.7359
1681.9173
N=1/14
T1
0.1455
0.1450
0.1459
0.1448
0.1448
T
0.1613
0.1528
0.1511
0.1487
0.1479
Q1
548.2020
546.0887
549.6271
545.5131
545.4294
Q
610.4544
577.0097
570.3747
560.9377
557.7669
NP
1693.8558
1686.1022
1683.4340
1682.0905
1681.2763
N=1/13
T1
0.1461
0.1451
0.1454
0.1449
0.1458
T
0.1618
0.1529
0.1506
0.1487
0.1490
Q1
550.3455
546.6795
547.6956
545.6541
549.4377
Q
612.5216
577.4625
568.2460
560.9213
561.8045
NP
1693.0960
1685.3898
1682.7439
1681.4036
1680.5950
In the whole table NP=NP1.
Special Cases. (i) Considering zero values for N and cb and setting T1=T, the optimal replenishment policy determined by GA is as T=0.1333, Q=500.4455, and NP1=NP3=1692.8885 which is identical to the results of Teng et al. [29].
(ii) Table 6 presents the obtained ordering policy for the case that no delay in payment is assumed to show the conventional inventory system with backorder.
Obtained replenishment policy for the traditional case of Example 3.
cb
5
10
15
20
25
T1
0.1308
0.1308
0.1310
0.1309
0.1307
T
0.1461
0.1384
0.1361
0.1347
0.1337
Q1
490.6748
490.4136
491.4703
490.7945
490.0092
Q
550.6563
520.3205
511.4378
505.7431
501.9317
NP
1668.1513
1660.1071
1657.3286
1655.9201
1655.0680
Example 4.
Suppose the fourth example of Teng et al. [29] in which a=3600 units, b=1200, M=1/12 year, s= $1 per unit, c= $0.5 per unit, A= $10 per order, h= $0.9 per unit per year, Ic= $0.16 per year, and Ie= $0.08 per year. The solutions of this data set, when N and cb are included in the model, are represented in Table 7.
Obtained replenishment policy for Example 4.
cb
5
10
15
20
25
N=1/15
T1
0.0795
0.0820
0.0841
0.0834
0.0837
T
0.0941
0.0895
0.0892
0.0872
0.0867
Q1
289.9633
299.2573
306.8219
304.4183
305.3441
Q
343.9493
327.0806
325.8445
318.5614
316.6962
NP
1591.4617*
1580.4222*
1576.4328
1574.4111
1573.1648
N=1/14
T1
0.0796
0.0821
0.0829
0.0835
0.0838
T
0.0942
0.0896
0.0880
0.0873
0.0868
Q1
290.4614
299.6770
302.5949
304.7477
305.7191
Q
344.2597
327.3980
321.2614
318.8368
317.0280
NP
1591.0349*
1580.0848*
1576.1503*
1574.1244
1572.8887
N=1/13
T1
0.0798
0.0820
0.0831
0.0836
0.0838
T
0.0942
0.0894
0.0881
0.0873
0.0869
Q1
290.9395
299.0856
303.2486
305.0162
305.9744
Q
344.5000
326.5366
321.8018
319.0373
317.2285
NP
1590.6303*
1579.7860*
1575.8928*
1573.8874
1572.6645
*NP=NP2. For the rest, NP=NP1.
Special Cases. (i) When N=cb=0 and T1=T, the fourth example of Teng et al. [29] is a special case of this example and the equivalent solutions obtained here by GA are T=0.0810, Q=295.5996, and NP3=1579.7113.
(ii) The results of the traditional inventory system with backorder for Example 4 are shown in Table 8, when M and N are equal to zero.
Obtained replenishment policy for the traditional case of Example 4.
cb
5
10
15
20
25
T1
0.0754
0.0780
0.0789
0.0794
0.0797
T
0.0902
0.0856
0.0840
0.0833
0.0828
Q1
274.9067
284.2815
287.7354
289.5310
290.6344
Q
329.5883
312.5334
306.7940
303.9124
302.1825
NP
1580.7096
1568.8370
1564.5468
1562.3322
1560.9801
Example 5.
Suppose that a=3600 units, b=1200, M=1/2 year, N=1/5 year, s = $1 per unit, c = $0.5 per unit, A= $10 per order, and Ie= $0.08 per year. The replenishment policies for different values of h, Ic, and cb are shown in Table 9.
Obtained replenishment policy for Example 5.
Ic
h
cb
5
10
15
20
25
0.13
0.9
T1
0.0758
0.0784
0.0792
0.0801
0.0801
T
0.0907
0.0860
0.0844
0.0840
0.0832
Q1
276.4047
285.7800
288.8577
292.1710
292.1543
Q
331.3114
314.1828
307.9850
306.7559
303.7719
NP
1668.4121
1656.4728
1652.1599
1649.9317
1648.5752
0.16
1
T1
0.0711
0.0737
0.0747
0.0751
0.0755
T
0.0864
0.0816
0.0800
0.0791
0.0787
Q1
258.9127
268.4857
272.0984
273.6748
275.1391
Q
315.5950
297.8718
291.9541
288.5273
287.1792
NP
1657.2806
1643.8682
1638.9849
1636.4554
1634.9095
In the whole table NP=NP3.
Special Cases. (i) When N=cb=0 and T1=T, Teng et al. [29] is a special case of this example and the solutions found are the following.
For Ic= $0.13 and h=0.9, T=0.0815, Q=297.5441, and NP3=1701.3369.
For Ic= $0.16 and h=1, T=0.0770, Q=280.7654, and NP3=1686.8285.
(ii) The results of the traditional inventory system for Example 5 are shown in Table 10, when M and N are equal to zero.
Obtained replenishment policy for the traditional case of Example 5.
Ic
h
cb
5
10
15
20
25
0.13
0.9
T1
0.0762
0.0787
0.0796
0.0801
0.0804
T
0.0909
0.0863
0.0848
0.0840
0.0835
Q1
277.8044
287.1020
290.5287
292.3346
293.3977
Q
332.2228
315.2022
309.4800
306.6227
304.8784
NP
1582.4520
1570.7982
1566.5921
1564.4219
1563.0974
0.16
1
T1
0.0707
0.0733
0.0743
0.0748
0.0755
T
0.0859
0.0813
0.0797
0.0789
0.0788
Q1
257.3913
267.2165
270.8610
272.7613
275.2462
Q
313.7748
296.4620
290.6188
287.6816
287.2915
NP
1569.6400
1556.2955
1551.4351
1548.9178
1547.3752
Comparing the results of the examples for the proposed inventory system with two-level trade credit and allowable shortages with the results obtained by Teng et al. [29] illustrates that, for some values of backorder cost, the proposed model in this paper is more profitable to the retailer. It is profitable to the point that the effects of not paying the inventory holding cost and the charged interest during the backorder period outweigh the cost of shortages. As it is expected, the examples show that the retailer’s net profit decreases with the increase of the backorder cost.
Moreover, the net profit earned by the retailer in the two-level trade credit is more than that earned by the same model without delay in payment which is because of the fact that the retailer does not tie up his capital in inventory and can earn interest when the payment is made with delay.
7. Conclusion
For matching the real world inventory systems, this paper extends the model proposed by Teng et al. [29] to the case of two-level trade credit with backorder. The formulated model is then solved by applying genetic algorithm and its validity is proved by solving the same examples reported in Teng et al. [29]. Comparing the results of this paper with that of Teng et al. [29] demonstrates that it can be more profitable when backorder cost is smaller than or equal to a specific value. Moreover, comparing with the traditional inventory system with backorder, the trade credit, both the model of Teng et al. [29] and the one proposed in this paper, would increase the profit of the retailer.
The presented inventory model of this paper can be extended in several ways. For example, deteriorating items can be considered. Moreover, the system may be generalized for partial backlogging shortages. In addition, developing a two-echelon supply chain inventory system consisting of a supplier and a retailer would be of great research interest.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgment
This research is supported by the University Putra Malaysia.
HarrisF. W.How many parts to make at onceGlockC. H.GrosseE. H.RiesJ. M.The lot sizing problem: a tertiary studyPiaseckiD.MolamohamadiZ.RezaeiahariM.IsmailN.Consignment inventory: review and critique of literatureSeifertD.SeifertR. W.Protopappa-SiekeM.A review of trade credit literature: opportunities for research in operationsMolamohamadiZ.IsmailN.LemanZ.ZulkifliN.Reviewing the literature of inventory models under trade credit contactGoyalS. K.Economic order quantity under conditions of permissible delay in paymentsAggarwalS. P.JaggiC. K.Ordering policies of deteriorating items under permissible delay in paymentsJamalA. M. M.SarkerB. R.WangS.An ordering policy for deteriorating items with allowable shortage and permissible delay in paymentTengJ.-T.On the economic order quantity under conditions of permissible delay in paymentsChungK.-J.HuangY.-F.The optimal cycle time for EPQ inventory model under permissible delay in paymentsHuangY.-F.Optimal retailer's ordering policies in the EOQ model under trade credit financingHuangY.-F.An inventory model under two levels of trade credit and limited storage space derived without derivativesTengJ.-T.GoyalS. K.Optimal ordering policies for a retailer in a supply chain with up-stream and down-stream trade creditsHuangY.-F.Economic order quantity under conditionally permissible delay in paymentsHuangY.-F.Optimal retailer's replenishment decisions in the EPQ model under two levels of trade credit policyTengJ.-T.ChangC.-T.Optimal manufacturer's replenishment policies in the EPQ model under two levels of trade credit policySuC.-H.An integrated supplier-buyer inventory model with conditionally free shipment under permissible delay in paymentsDyeC.-Y.OuyangL.-Y.A particle swarm optimization for solving joint pricing and lot-sizing problem with fluctuating demand and trade credit financingDyeC.-Y.A finite horizon deteriorating inventory model with two-phase pricing and time-varying demand and cost under trade credit financing using particle swarm optimizationMahataG. C.An EPQ-based inventory model for exponentially deteriorating items under retailer partial trade credit policy in supply chainLouK.-R.WangL.Optimal lot-sizing policy for a manufacturer with defective items in a supply chain with up-stream and down-stream trade creditsMolamohamadiZ.ArshizadehR.IsmailN.An EPQ inventory model with allowable shortages for deteriorating items under trade credit policyChernM.-S.ChanY.-L.TengJ.-T.GoyalS. K.Nash equilibrium solution in a vendor-buyer supply chain model with permissible delay in paymentsChernM.-S.PanQ.TengJ.-T.ChanY.-L.ChenS.-C.Stackelberg solution in a vendor-buyer supply chain model with permissible delay in paymentsChenS.-C.TengJ.-T.Retailer's optimal ordering policy for deteriorating items with maximum lifetime under supplier's trade credit financingChenS.-C.TengJ.-T.SkouriK.Economic production quantity models for deteriorating items with up-stream full trade credit and down-stream partial trade creditChenS.-C.Cárdenas-BarrónL. E.TengJ.-T.Retailer's economic order quantity when the supplier offers conditionally permissible delay in payments link to order quantityTengJ.-T.MinJ.PanQ.Economic order quantity model with trade credit financing for non-decreasing demandChenX.WanW.XuX.Modeling rolling batch planning as vehicle routing problem with time windowsParkY.-B.A hybrid genetic algorithm for the vehicle scheduling problem with due times and time deadlinesGenM.ChengR.