Optimizing Distribution Route of Packed Drinking Water with The Clarke and Wright Savings and Nearest Neighbor Methods (Case Study of PT. GSI)

In the business, distribution has a very important role. The distribution process must be able to deliver products on time while it also reduces the transportation costs which can consume around 50% of the company's total logistics costs. If distribution costs can be reduced, it can indirectly increase the company’s profit. PT. GSI, located in Yogyakarta, is a distributor of packed drinking water company with brand “PLG”. In determining the distribution route, PT. GSI does not use any scientific method and only uses the driver’s intuition. The purpose of this study is to optimize the distribution route of the “PLG” packed drinking water delivery to obtain the minimum mileage by using the Clarke and Wright Savings and Nearest Neighbor methods. The Clarke and Wright Savings method and the Nearest Neighbor method are used because the customer locations are far from each other and there are also customers whose delivery locations are centered on one delivery area, so that the calculation of distance savings is needed. Then the evaluation is done to determine which method is better. The comparison of the solution for data on 1 st ,2 nd ,3 rd ,4 th , and 6 th March 2021 showed that the Nearest Neighbor method proposed best solution while data on 5 th March 2021 showed that Clarke and Wright Savings giving the best solution. The results show that the Nearest Neighbor method provides better result in determining the distribution routes of PT.


INTRODUCTION
Companies that want to win the business must improve the customer satisfaction. One thing that can improve the customer satisfaction is though providing on time delivery in the distribution. In product distribution, the product delivery can have several destinations or customers then result in several routes. If the distribution routes are not well managed, this can cause higher distance and longer time for distribution then result for higher distribution costs and waste the energy and time (Muhammad et al., 2017).
By optimizing the distribution route, the total travel distance and travel time can be minimized (Pujawan, I Nyoman, 2010). The optimal method is a method that produces the shortest total mileage, short distribution time, 8. Periodic VRP, where service to consumers can be carried out at some time during the planning horizon where the planning horizon is only valid on one day. 9. VRP Heterogeneous Fleet of Vehicles (VRPHFV), in which the characteristics and types of vehicles used by the companies are vary. 10. Dynamic VRP (DVRP), which is if there is a new customer in creating a route, the new customer is inserted into an additional route when creating the main delivery route. 11. Stochastic VRP (SVRP), in which the number of customers, service time, and the customer requests are uncertain. Every customer has the possibility of not being visited every day.
CVRP is one of the most common optimization problems in transportation, distribution, and logistics. The purpose of the CVRP is to determine the total travel time while at each route, it is not allowed to exceed the vehicle capacity. In this case, the aim is to obtain the best method for determining the shortest distribution route [12].
The Clarke and Wright Savings algorithm is the heuristic algorithm that is most often used in the VRP. There are several steps in the Clarke and Wright Savings method (Clarke & Wriht, 1964 S ij = C 0i + C 0j -C ij C 0i = distance from depot to node i C ij = distance from node i to node j S ij = value of distance savings from node i to node j 4. Create a savings matrix. 5. Selects a cell in which two routes can be combined into a single route. Nearest Neighbor is one of the algorithms included also in the heuristic method. The purpose of this method is to find the best route by distributing products based on the closest distance to the last vehicle location (Pujawan, I Nyoman, 2010). There are several steps in the solution using the Nearest Neighbor method.
1. Set the origin point of departure. 2. Find the shortest distance from the origin to another point (destination point) and connect the two. 3. The currently selected destination point will be the point of origin for the next departure. 4. Checking whether all points have been visited or not. a. If there is a point that has not been visited, repeat step 1. b. If all points have been visited, go to step 5. 5. Route is formed. 6. Finished.

RESEARCH METHODS
The object of this research is the distribution route of PT. GSI, starting from PT.GSI's warehouse to the several agents (consumers) who ordered "PLG" Packaged Drinking Water. The study is limited to the distribution areas in West Sleman, North Sleman, North Bantul, and West Yogyakarta. The data required in this study is primary and secondary data. The primary data required are vehicle distribution system, the type of vehicle, vehicle capacity, distance between warehouses to the consumers and distances between consumers. The secondary data used in this study are consumer's demand data, the delivery date and the agent's address data. This study uses interview and observation for data collection.
There are several steps used to determine the distribution routes using the Clarke and Wright Savings and Nearest Neighbor methods. The first step is identifying the problem, conducting a literature study, and collecting the data. Then data processing is carried out using the Clarke and Wrights savings method which begins with identifying the distance matrix, determining the savings matrix, allocating consumers to the routes, and then sorting the distribution routes. After that, validation is carried out using the Nearest Neighbor method which begins with grouping based on the previous method, namely the Clarke and Wright Savings method and then determining the distribution route. After getting the results of data processing then the two methods are compared. The best method is method that has the least total travel distance. Figure 1 shows the research method implemented in this research.

Distribution Route Analysis
The existing distribution route of PT. GSI is still not optimal as shown in Table 2. The current distribution route in the areas of West Sleman, North Sleman, North Bantul, and West Yogyakarta is determined only based on the driver's intuition. The product delivery at PT. GSI is carried out every day and the product are delivered next day after the consumer places an order. The following Table 3 is the distribution route as the solution proposed by the Clarke and Wright Savings method. Meanwhile, in the Nearest Neighbor method, the data processing is done by sorting the route based on the minimum distance. The method results in six routes carried out for six days. The following Table 4 is a route distribution as the solution proposed by the Nearest Neighbor method.

Proposed Distribution Route
In this research, the solutions proposed for the distribution route of the "PLG" packaged drinking water distribution indicate that the Nearest Neighbor method can reduce more the total travel distance compared with the existing route and the Clarke and Wright Savings method, as shown in Table 5.
Based on the Table 5, it can be concluded that the Nearest Neighbor method produces a delivery route that is shorter than the initial route (driver's intuition) and the Clarke and Wright Savings method. The Nearest Neighbor method can be used as a better method for determining the distribution route of PT. GSI for "PLG" packaged drinking water distribution in the areas of West Sleman, North Sleman, North Bantul, and West Yogyakarta.

CONCLUSION AND FUTURE RESEARCH
1. On March 1, 2021, the selected route is the route that uses the Nearest Neighbor method with a total distance of 63.2 km, so that a total distance savings of 28.2 km is obtained. On March 2, 2021, the route chosen was the route using the Nearest Neighbor method with a total distance of 91.34 km, so that a total distance savings of 99.66 km was obtained. On March 3, 2021, the selected route is the route that uses the Nearest Neighbor method with a total distance of 119.075 km, so that a total distance savings of 47.475 km is obtained. On March 4, 2021, the selected route is the route that uses the Nearest Neighbor method with a total distance of 53 km, so that a total distance savings of 17 km is obtained. On March 5, 2021, the selected route is the route that uses the Clarke and Wright Savings method with a total distance of 85.6 km, so that a total distance savings of 23.7 km is obtained. On March 6, 2021, the selected route is the route that uses the Nearest Neighbor method with a total distance of 105.016 km, so that a total distance savings of 82.584 km is obtained.