The Effectiveness of the Clarke & Wright Savings Algorithm in Determining Logistics Distribution Routes (case study PT.XYZ)

. A good logistics system is one factor that affects the smooth running of company activities. Currently, transportation costs account for about two-thirds of total logistics costs, so to increase efficiency, companies try to maximize the use of transportation equipment and personnel by determining the optimal distribution route. Route optimization can be done using the Vehicle Routing Problem and the Clarke-Wright Method. PT. XYZ is a company engaged in maintaining ATM machines (Automated Teller Machines). The form of this activity can be in the form of installing air conditioners, TVs and repairing ATM machines. This company faces problems in the logistics section, where the installation target is often not achieved from the specified target. This is because the distance from one point to another is coupled with the congestion level in big cities, so this causes the company to pay large operational costs. Therefore, a distribution route is needed to save time and distribution costs. This study uses the Clarke and Wright Savings Algorithm to solve the problems faced by companies. The results showed that after using the Clarke and Wright Savings Algorithm, the total distance and shipping costs were reduced by saving 97.9 Km of distance and saving costs.


Introduction
Business competition is getting tougher, causing many consequences.Companies are required to improve their competitiveness continuously.In addition, they must be able to change themselves to become agile so as to improve their performance.One of the factors that influence the smooth running of the company's activities is the role of good logistics.In general, logistics relates to planning and controlling the flow of goods or materials.
According to the Council of Supply Chain Management Professionals in [1] logistics is the process of planning, implementing, and controlling so that transportation procedures can run effectively and efficiently and are connected from the starting point to the consumer, whereas according to [2] the process This consists of planning, regulating and controlling the flow of goods, from upstream to downstream.In carrying out its duties, the logistics unit is usually assisted by several sections, such as analysis systems, warehouses, purchasing, transportation, and distribution.Logistics is one of the foremost parts of a business.
Following are some of the contributions of logistics in companies according to [3] where there are optimal logistics costs to encourage company performance, which is marketing-oriented because logistics tasks play an important role in satisfying customers and achieving profit for the company.[4] Based on Figure 1, The Transport and Logistics Industry is the backbone of any economy, and the driving force behind all sectors, be it agriculture, manufacturing, or services.Reducing transportation costs is the number one challenge.This occurs because transportation costs account for around 30% of the total shipping costs.However, if there is an increase in fuel prices, it can increase to 50%.This is evidenced by a study conducted by [5] regarding the main challenges that occur in the logistics sector.While the number one challenge in the logistics sector is reducing transportation costs.According to [6], transportation costs account for about a third of the total logistics costs, so companies must plan and determine optimal distribution routes.Optimizing routes to reduce costs is called VRP or Vehicle Routing Problem.VRP is a solution for finding the shortest path with limited vehicle capacity, while the Clarke-Wright Method solves route optimization problems [7].Clarke and Wright discovered this method in 1964, and it was later published as an algorithm that is used as a solution to vehicle routing problems where a set of routes at each step are exchanged to get a better set of routes this method is used to solve a fairly large problem [8].One of the problems that could be solved is many routes.The essence of this method is to calculate savings that are measured by how much distance and time can be reduced by associating existing nodes and making it a route based on the largest saving value, namely the distance travelled between the source node and the destination note.A study from [9] has conducted previous research with this method to minimize the distribution costs of postal shipments by the Czech Republic post office.From this research, it was found that this research succeeded in reducing the distance by 5.8 km from the original total distance of 75.8 km, resulting in savings of 7.65% km.Another research was also carried out by [3] using the same method to solve problems with the distribution of bakery products in the city of Malang, with a savings of 21.3 km of the total mileage of 74.9 km and savings in distribution costs of Rp. 16,293 of the total cost of Rp. 57,300 in a day.According to [10], some of the logistical challenges in Indonesia include actors' understanding of Supply Chain Management (SCM), where inadequate infrastructure and data and logistics information systems have not been integrated.
One of the businesses affected by this problem is PT.XYZ, which is located in West Jakarta.This company has been established for over 5 years and faces problems in the logistics section, such as targets for the maintenance of ATM machines are often not achieved.This is due to the large number of ATM machines that have distances from one ATM point to another far apart ATM point plus the level of traffic density during working hours, so the problem causes large operational costs that need to be incurred by the company.Based on the background of the problems described above, the authors would like to identify the following problems: Is the distribution route carried out by PT.XYZ effective and efficient?How do the distribution route optimization results apply using the Clarke and Wright Savings Algorithm approach?In this study, the authors limit the scope of research that will be discussed by using the Clarke and Wright Savings algorithm to optimize PT.XYZ so that it affects profit and the level of service to customers.From the case study above, the author uses the Clarke and Wright Savings Algorithm to solve the problems faced by companies with the aim of getting the best route that the company can apply to deliver goods according to predetermined targets.
The first novelty of this research is providing source code for the Clarke and Wright Savings algorithm.The Clarke and Wright Savings algorithm currently faces criticism from a scholar [11] because previous research did not provide source code for the algorithm.The second novelty shows that Clarke and Wright Savings algorithm could reduce carbon emissions from transportation costs.Previous research showed that carbon emissions could be reduced by redesigning the logistic systems [12].

Literature Review 2.1 Operational Management
A series of activities that create value in the form of goods and services by converting inputs into outputs [13].The purpose of operations management is to use resources so that the production process can be carried out effectively and efficiently.Operations management has five benefits: Improved efficiency -Innovation & Competitiveness -Better use Of Resources -Create Value for Customers -Improved Product Quality.It ensures the products are of the highest quality standards to meet the target customer's expectations [14].

Supply Chain
A management system for various activities in order to obtain raw materials, followed by transformation activities so that they become products in process, then become finished products and are forwarded to delivery to consumers through the distribution system [13].In this process, various problems are often encountered [14]: Producers are overwhelmed because consumers want better products without seeing producers' difficulty in processing products according to consumer wishes.The decline in the production process causes the company to overcontrol inventory because it takes a certain amount of time.The number of requests from customers who order without looking at product standards.The existence of Fragmentation of Supply Chain Ownership results in increasingly complicated SCM processes because it covers a wide area.There is often miscommunication between consumers and fellow staff and uncertainty in demand.

Transportation Management
Transportation is an attempt to move inventory from point to point in the supply chain [13].Transportation has several roles in the supply chain, including because products are rarely produced and consumed in similar locations, shippers require product movement, and carriers move or transport products [15].Distribution is the process of channelling goods and services from producers to target consumers [16].Effective distribution will facilitate the flow or access of goods so that consumers can obtain their products conveniently.

Distribution Strategy
For most modes of transportation, there are two types of delivery methods, namely Direct Deliveries and Milk-Run Deliveries [13] are: a) (Direct Deliveries), where direct deliveries are made from the location of origin to the location of the recipient.With this shipping method, the problem is simply choosing the shortest route from the two locations.The advantage of this method lies in the simple operation and coordination of shipments.
b) (Milk-Run Deliveries) Is a shipment that brings products from one location of origin to several receiving locations or shipments that bring products from several origin locations to one receiving location.The advantage is that it can be used more efficiently, and the cost of receiving shipments is lower because the recipient location is able to get fewer or larger shipments at the same cost.

Vehicle Routing Problem
A search for a solution includes determining the number of routes and means of conveyance that starts and ends at the depot of origin.According to [14]

Research Method
In this study, researchers used a quantitative method, with the explanatory research type.The unit of analysis in this study is PT XYZ.The steps taken in analysing the data have been collected using the Clarke and Wright Savings Algorithm approach [9] where: 1. Through a transport network S = (V, H), where V is a set of nodes in the network while H is a set of starting point of the network, and nodes V1 to Vn are the nodes that need to be served 2. Create a distance matrix D = {d (i, j)} for the transportation network S by specifying its delimiters through parameters.3. Based on matrix D, there is a set of basic routes (V0 -Vi -V0) for all network nodes presented.4. Make a benefit coefficient matrix Z = {zij} by using the relation: is the distance between the initial node and vertex i, is the distance between the initial vertex and j vertex, and distance between vertices i and j. 5. Choose the highest score ------proceed to Step 7. 6.Using the value deleting the combined routes.7. If algorithm will end.

Distribution Route
Using the Clarke and Wright Savings method, customers will be manually grouped into certain location zones.After the grouping is complete, determine several stopping points and transport with a number of shipments so that it can be known optimally.
From the distribution route, it will be generated through this method which later is expected to include the number of trips, the distance travelled, and the shipping costs.This distribution route is the initial solution to the VRP problem.After processing the data and getting the results, it can be compared with the conditions applied to this company.From the results of manual processing, shipping routes are obtained based on the area (one of the routes is for the West Jakarta area).Based on Figure 2, the shipping data will be translated at PT. XYZ as code A0/B0/C0/D0/E0/F0.Data on the distance that vehicles need to travel to distribute goods in Kilometres (Km) taken from one of the West Jakarta 1routes: Table 1 showed the code for each destination location, for example showing A0 -A1, which means the required distance from PT. XYZ to somewhere 19 km away, and so on.For A0 -A0 is the distance from PT. XYZ to PT. XYZ, which explains the letter X, means there is no distance to be calculated.

Data Analysis
After performing calculations using the VRP Solver software, the distribution has been divided based on the initial delivery route and per region, using an average speed of 65 km/hour with normal road conditions or no traffic jams.The following is the distribution route used by PT.XYZ to work on delivery services: From table 2, it can be concluded that the shipping route used for each car with sequential conditions starting from PT. XYZ to a destination is obtained for 2 hours, so the total time required, if combined, is obtained for 8 hours at the four points of interest.After performing calculations using the VRP Solver software, the following results are obtained based on the distribution is: After performing calculations using the VRP Solver, the results for the new distribution route for the West Jakarta 1 area are obtained as follows: The table 3 shows that the distance described by the VRP Solver software suggests new routes in the West Jakarta 1 area, where the first destination point needs to be taken after taking the required equipment at PT. XYZ is the driver who can go to Indomaret -Joglo A2.After that, the second destination point can be continued to Indomaret-Rawa Belong.
If the work at the second point has been completed, the driver can continue his journey to the third destination, namely Indomaret -Tanjung Duren Barat.Furthermore, Indomaret -Panjang Kedoya Utara is the final destination for the services that need to be carried out by the driver.After finishing working on the fourth destination point, the driver can return to PT. XYZ to do administrative reporting.From the route suggested by VRP Solver, the total distance travelled by the driver is 47.3 km, whereas PT.XYZ followed this route, so PT.XYZ can save a distance of 10 km from the original route.

Analysis Results Based on All Areas
After performing calculations using VRP Solver, the results of the new distribution route are obtained as follows: Table 4 shows that the VRP Solver software recommends using a new route if all routes are scrambled and also to see if the total distance can be reduced compared to routes per area.

Calculation of Shipping Costs
Based on data on fuel costs and toll and parking fees, the authors know the costs incurred by the company in making one delivery.The following are the shipping costs for each vehicle: Table 5 -7 show that toll and parking fees are one of the operational costs that the company must incur.The total expenditure for shipping costs on the initial route is Rp.928,317, -while the total cost of sending routes per area that the Clarke and Wright Savings Algorithm have processed is Rp.884,600, -and for the total cost of sending random routes that have been processed by the Clarke and Wright Savings Algorithm, which is Rp.860,280, -.Thus, there is a in total delivery costs between the initial route and the route per area of Rp. 43,717, -for one trip.Meanwhile, the difference in total delivery costs between the initial and random routes is Rp.68,037, -.

Analysis of Proposed Alternative Routes in the Distribution of Goods
The Clarke and Wright Savings Algorithm is used in order to be able to find routes with shorter distances compared to the routes currently used so as to reduce operational costs in terms of shipping goods.This route is obtained from the results of dividing consumers into travel routes based on the largest saving order.The table 8 and 9 below is a comparison of the results of the initial route calculation and the proposed alternative routes that have been carried out using the Clarke and Wright Savings Algorithm:  Table 10 shows that after using the Clarke and Wright Savings Algorithm, the company can save 62.9 km of mileage per area and save Rp. 43,717, -every month.If this route is carried out for up to one year, PT.XYZ was able to save Rp. 524,604, -.However, the random route calculation is said to be superior to the route per region because it uses the company's random route.

Conclusions
Based on the results of the calculations and analysis above, several conclusions can be drawn as follows: 1) PT.XYZ currently used cannot be said to be effective and efficient because of PT.XYZ only trusts the intuition of the driver to go from the closest consumer to the farthest.2) When using the new route and schedule, the areas will be reduced from 472.4 Km to 374.5 Km and PT.XYZ was able to save 97.9 Km in one working day.3) If PT XYZ applies the proposed alternative route.
then the cost will be reduced from IDR 928,317 to IDR 860,280 and PT.XYZ was able to save Rp. 68,037 -in one working day.

Table 4 .
Delivery Sequence Results in All Areas Based on VRP Solver

Table 1 .
West Jakarta Consumer Mileage Data

Table 2 .
Initial Delivery Route per Region

Table 3 .
The results for the new distribution route

Table 5 .
Total Initial Route Shipping Cost

Table 6 .
Total Route Shipping Costs per Region processed by Clarke and Wright Savings

Table 8 .
Comparison of Initial Route Delivery Results

Table 9 .
Comparison of Delivery Route Results by Alternative Regions by the Clarke and Wright Savings Algorithm

Table 10 .
Proposed Alternative Routes per Area by Clarke and Wright Savings Algorith