Warehouse Specification Proposition for Urbanis (Urban Farming Company) Using Discrete Event Simulation Method

_____________________________________________________________________________________________________ Two percent of the world's surface use for cities, yet it consumes 75 percent of its resources. Urban farming is an emerging alternative food network that could supply some of the food needs in cities with less emission, healthier food, and the environment. Urbanis is a company that likes to contribute to the acceleration of urban farming, especially in Indonesia, by utilizing vacant land and labor. In 2021, Urbanis plans to scale up the production capacity to 10 tonnes per month or 400 kg per day. It requires us to have a warehouse to store the food product that has not been absorbed by the market. The purpose of this study is to find warehouse specifications for Urbanis and the amount of labor and rack inside the warehouse alongside capital and operational expenditure. This research uses a layout with an area of 5x14 meters for experimental design. The model then translated into a discrete-event simulation model named Anylogic. The results show, for each amount of arrival, the number of labor that utilizes effectively are two labors with a maximum number of rack 50. Given these results, the author conducted operational and capital expenditure, which consist of variable analysis and additional variables such as a table, fan, and chair. The result is Urbanis need Rp 50.738.000 for capital expenditure while Rp 10.871.337 for operational expenditure.


Previous Research on Food Logistics
The food industry is becoming interconnected from farmer to customer. This relationship is called the Food Supply Chain Network (FSCN). There are 2 FSCN, fresh agriculture products, and processed food products. Fresh agricultural products are more vulnerable to the condition due to their freshness. Meanwhile, processed food products generally have been through a process that makes it hardly perishable (Vorst et al., 2009).
To deliver the product to the customer rightly, this requires the management of logistics in all nodes. Logistic management is a movement of goods to the customer with its objective to deliver the right product, in the right condition, with the right quality, in the right place, at the right time, and cost (Aghazadeh, 2004).
In research by Aghazadeh (2004), to make the food supply efficient, there should be a sharing of information electronically from the retail, distributor, and manufacturer. It helped all the chains could order or manufacture the "enough" amount of food, which led to a lower cost.
There are many KPI that could be used to see the effectiveness and efficiency of logistics. Table 1 KPI for logistics service provision. www.msocialsciences.com

Food Warehouse Simulation
Inventory cost could be as high as 50% of the purchase cost in many companies (Hodson, 1992). Thus, many companies are trying to be as efficient as possible in storing material or products in the warehouse.
Food warehouses have different challenges compared to other products. Food is a perishable product that requires special treatment such as temperature control, freshness date, and many more. Failure to provide this gives risk to the product and will be waste due to the quality (Stragas & Zeimpekis, Unknown). There are many types of food warehouses; cross-docking, 3PL, and private or lease warehouses. This type of warehouse has its function -See table 2. There are many KPI to measure the efficiency of the warehouse. Previous articles have been write regarding the use of simulation as a tool in warehouse design (Senko and Suskind, 1990). More examples were written of articles discussing models of order picking systems (Daniels et al . 1998, Kim et al . 2002, cost models for inventory and inventory sizing models (Cormier and Gunn 1996), warehouse capacity expansion models (Cormier and Gunn 1999), storage allocation (Macro and Salmi, 2002).
One of the most common methods of analyzing warehouse operations has historically been a simulation. Through seminal work on the simulation ability in warehouse management (see, for example, Bafna, 1973 andBiles 1977) to recent publications, the simulation was mainly used to validate the efficiency of warehouse structures or material handling equipment.

Order Picking
In the internal distribution center operation, order picking could occupy up to 50% of the overall cost (Hsieh & Tsai, 2006). In order to minimize the cost, four typical strategies could be used. They are (1) layout design, (2) picking policies, (3) storage assignment policies, and (4) routing policies (Chan & Chan, 2010).
According to Ashayeri (1985), the warehouse layout design goal is minimum building cost or material handling. Meanwhile, storage assignment policies aim to reduce travel time. Generally, the storage assignment policies are as follows: random storage, classified storage, fixed storage, volume-based storage, etc. (Hsieh & Tsai, 2006). On the other hand, routing goals are to reduce unnecessary routes of picking orders. Lastly, picking policies determined how the order would be grouped (Chan & Chan, 2010).
According to Krauth (2016), one of the warehouse's measurements is labor utilization and fixed asset costs. For labor utilization, a success indicator of labor utilization could see if the utilization is going up. In contrast, the fixed asset cost measurement will be a success if it's going down. Hence, this research will be focusing on the measurement of labor utilization and fixed cost (rack and building).

Methodology
The method used in this research is the discrete-event simulation. The research will use the framework suggested by Banks (2014)

Problem Formulation
In early 2021, Urbanis will increase its production capacity up to 20 times its current capacity. Aside from that, the market will shift into community-supported agriculture (CSA), which will change the business model into subscription healthy food production. Due to various types of food products, the companies need to have a place where it can store and combine the food produce. These necessities www.msocialsciences.com could be done in the warehouse. With high investment up front to make a warehouse, the company decided to measure the layout, consisting of the amount of rack and the utilization of labor.

Setting of Objective and Overall Project Plans
This research aims to find the amount of rack used by the companies when the production capacity increases. Meanwhile, it also measures the utilization of labor that works to store and prepare the food produced. From this measurement, then, the author will give the estimation of capital and operational expenditure that the company will be planned.

Model Conceptualization
The model will start with the truck coming to the warehouse. The supplier will unload the food products into the unloading area, from the truck, which will then be seized by the labor to put in the rack area. While putting in the rack area, the labor will then wait for the order.
After the order is received, the labor will collect the food products to be combined and packed. Then, it waited for the courier to come to the warehouse to bring the food products to the customer.

Data Collection
This research will use a quantitative approach to answering the problem. The data collected on this research will vary, ranging in observations, secondary data, and primary data (internal company data). In this research, the observations data will be the speed of pick-up car, labor, courier, and how many kgs could be brought per order by the courier.
The observation data will be collected by observing the speed of pick-up car, labor, courier, and how many kgs could be brought per order by the courier. Aside from that, it will also be strengthened by interviewing people who already work in similar situations. Secondary data will be pick-up car size, labor size, motorcycle size, and box size. The secondary data will be collected from official website companies or third party commerce such as tokopedia and rumah123. Lastly, primary data collected from internal companies data, which is the future projection of production capacity and schedule of food produce arrival per day in the warehouse.

Model Translation
Because most real-world programs result in models needing much information storage and computation, the model must be entered in a computer-recognizable format. In this research, the model will be computerized using a discrete-event simulation software application. The author chose Anylogic since this software is used by some Fortune 500 companies.

Verified
Verification applies to the simulation model. Will the machine system run properly? Since it is challenging, if not impossible, to effectively turn a concept into its model without any testing. When the functional and feedback parameters. The system structure is described correctly on the machine, and verification is done. www.msocialsciences.com

Validated
Validation is the determination that a model represents the actual system accurately. Validation is typically accomplished by model validation, an iterative method that matches the model against real system actions and utilizes the inconsistencies between the two against refinement of the model and the observations obtained. This cycle is replicated before the consistency of the model is considered appropriate.

Experimental Design
In this research, the model will have six scenarios. This scenario differs in the supplies arriving in the warehouse and the amount of labor working inside the warehouse. From these differences, the author will then choose one scenario suitable for Urbanis in case of the maximum amount of equipment (rack) and average utilization of labor.

Production runs and analysis
The analysis of the simulation will measure the performance that has been selected in the research objective. In this research, two performances should be measured, maximum rack, and utilization of labor. After collecting the data from 6 scenarios, the researcher will choose which is the most suitable one.

More runs?
Based on the analysis of completed runs, the analyst will determine if additional runs are needed and how those additional experiments should be designed.

Documentation and Reporting
After the simulation has been done, the researcher will make a documentation of the program and progress that will be put on this research. This documentation will consist of the maximum amount of rack used, the average utilization of labor, and total product shipped monthly.

Implementation
Lastly, the researcher in this research in which management of the company will make the implementation plan according to the analysis of the simulation. The implementation will consist of the company's timeline and resources to implement the business solution suggested by this research.

Model Conceptualization
The flow chart of the simulation will start with the pick-up car coming with the food produced. Then, it unloads all the food produced into an unloading area, which will then be loaded to the rack by the labor inside the warehouse. After it is loaded, the labor then waits for the order by the customer. After the order is received, the labor then brings the food products to the packing station to be prepared. Lastly, the order produced will then be given to the courier who will ship it to the customer. www.msocialsciences.com

Layout Design
In this research, researchers will experiment with a layout with a total length and width of 14 and 5 meters, respectively. In front of the building, there will be parking space totaling 10m2 (2x5m). Inside the building, there will be three areas; unloading area (2,5x2m), packing and loading area (2,5x2m), and rack area (5x8m). The rack area will then be divide into two categories based on the food produced; bok choy and water spinach. The rack could be level up to 3 levels, with each linear rack could store up to 60 boxes. In each rack, it will have an aisle with a width of 1,3 meters.

Production capacity
Future projection of the production capacity of urbanis warehouses will be divide into two scenarios.
The first scenario will start in the first quarter of 2021, which will use total production capacity (8 tonnes/month) that partnered with DISPANGTAN. In the second quarter, we will add the production capacity with the Rumah Sakit Jiwa Provinsi Jabar, which will add up to 2 tonnes per site. The amount of food arrive at the warehouse will be the division between total production and the average number of working days (6 days per week) per month (25 days) which estimated will be: The proportion of the food produced will be 50% each, which will be scheduled to be received 50% at 9 A.M. and 12 P.M on the working day (Monday -Saturday).

Market
As our company is planning to focus on Community Supported Agriculture (CSA) using a membership model, there will be a significant change in the percentage of food produced being absorbed by each channel. The strategic management projection will be:

Local Resident 15%
As our members will mostly be people who are interested in sustainable products, most of the B2C market will be people who have middle-high income. The amount of product shipped by Urbanis will be linear to the amount of food available in the warehouse.

Secondary Data
The secondary data is collected in the retail price of plastic rack and minimum wages in Bandung (Upah Minimum Kota Bandung). The plastic rack's retail price will be collected via e-commerce, while minimum wages in Bandung will be referred to as Surat Edaran Nomor: 561/75/Yanbangsos. The price and wages will be shown in table 8. Meanwhile, the size of the agent (labor, pick-up car, courier (motorcycle), and box) will be collected by looking at the official site or third-party commerce such as tokopedia. The size of the stakeholders will be in table 9.

Observational Data
Observational data collected will be the speed of the agent in the warehouse, labor, pick-up car, and courier. This data collected by observing the agent then strengthens by interviewing people who work or have responsibilities in similar situations, such as Faza and Ryan, as CEO of Urbanis and Head of Logistics in Harvest Supply, respectively.

Model Translation
The conceptual model then translated into a simulation software application named Anylogic. In this model, the researcher mimics the real-world event based on the previous part's data.

Figure 4: Business Translation
While testing in the simulation software, Anylogic, this model is verified to run well. Aside from that, this model is also validated to mimic real-world events inside the warehouse. Hence, this model could be used to simulate the Urbanis warehouse.

Experimental Design
In the 1st quarter and 2nd quarter of 2021, there will be three scenarios conducted for each quarter. Each quarter, the scenarios will be the amount of food arrive per schedule (Table 11) and the labor that works inside the warehouse. All the scenarios will be shown in Table 11.

Documentation and Reporting
From the simulation conducted, the data shown for each scenario will be shown in table 12 and 13.

Analysis
Based on the data collected, it will then be calculated by using the average cost per kg for each scenario. Since in scenario 1, the rack reached full capacity and could not keep up to manage upcoming supply for the next day, by then, it was eliminated from the option. Table 14. below is the analysis for the cost. As the results show that, the least cost by each quarter is using two labors. Therefore, the author suggested using two labors for both the 1st quarter and 2nd quarter. www.msocialsciences.com

Building
For warehouse purposes, the author suggested finding a building that fits the amount of rack needed by the warehouse. Based on Rumah123.com, rent for the building is to pay for a minimum yearly. For this reason, we used the maximum amount of rack used in the 2nd quarter since it has a higher amount.
In the second quarter, the amount of rack needed will be in total is 50. Divided into three levels of rack, the amount of size that needed by the company will be 16,67 rack or rounded to 17 racks. With this amount, the size for the rack only will need around 40,8m2.
Gathering the data from Rumah123.com, the researcher suggested one suitable place that could be used by the company.

Figure 5: Building Suggestion
In the figure above, the building will have a total surface area of 100m2, with a total building area of 95m2. The area's length and width are 5x20m, which is larger than the layout conducted in the simulation. In rumah123.com, the yearly rent for the building is Rp 45.000.000/year.
By having a larger area, this building could become a warehouse and office space for Urbanis. Meanwhile, the left area could also become the input space in the model, such as the toilet.

Additional Warehouse Equipment
There is a lot of material that needs to be in the warehouse yet not measured in the simulation. Such as a chair for the labor to sit or take a rest.  Table  3 Chair 5 Fan 2 The table will be used for multi-purpose purposes such as packing, eating tables, etc. Meanwhile, chairs will be used for labor, courier, and supply to take a rest. Lastly, the fan is used to make sure the place is well ventilated.

Additional Warehouse Human Resource
With labor focus on the hard work inside the warehouse, there is a need to have people manage the warehouse data. Aside from that, this person will be responsible for supervising the labor while receiving guests. Hence, the company hypothetically needs one admin to control and manage the warehouse.

Capital Expenditure
The equipment price is collected via Tokopedia (e-commerce), while building data is collected in Rumah123.com.

Operational Expenditure
On admin wages, the author assumed the wages are similar to labor wages. Hence, the operational expenditure will be wages multiplied by three.

Conclusion
On answering the research problem, the author conducted discrete-event simulation via application software, Anylogic. The data collected comes from 3 resources, such as primary data (Urbanis internal data), secondary data (online data collection), and observation data (Interview and observation). Then, the data will be analyzed using Anyogic to find how many people and equipment (rack) need by the Urbanis to provide food products in the future. The discrete-event simulation results will be the basis for answering the problem statement of this research. The problem is satiated as follows: Urbanis need to have a warehouse with a maximum rack dimension of 50 racks while having two labor to work on food produce storing and picking. Looking at rumah123.com, the layout suitable for urbanis is locate in Bandung with a minimum land area requirement of 5x14 meters. Aside from that, the author acknowledges there is a need for additional equipment not yet analyzed in this research, such as furniture, detailed order picking equipment, table, chair, fan, etc.
It shows that for both the 1st quarter and 2nd quarter warehouse scenario, urbanis need a similar amount of labor, which is two manpower for order picking activities. For the rack, it has differences in both quarters, in the first quarter, with the labor used 2, the maximum racked used will be 32. Meanwhile, in the second quarter the maximum rack used is 50.