Archive for category Optimization

Container Terminal Integrated Planning: Balancing Schedule and Cost

hamburg-2103261_640Containerized sea-freight transportation has grown significantly these recent year, brought an increasing interest on the optimization on container terminal operations. Recent research trends nowadays leads to the integration of container terminal planning. As the container terminal operations itself consists of some intertwined systems, so the integration between them becomes critical.

Operational efficiency and productivity become the key issues in the development of container terminal operations management. As the service provider, container terminals are expected to satisfy the shipping lines schedule related to the vessel arrival and departure time. In other hand, container terminals are also in the urge of maintaining their operational cost in order to compete with other terminal. Somehow at the container terminal, efficiency and productivity will depend a lot on the operational planning held by the terminal manager.

In the daily practice, terminal managers determine the berthing position of each vessel before they arrived at the terminal, due to the information given by the shipping line and also plotted the handling time for the vessel’s operation. This usually called as tactical berth allocation problem. Then, they also determine the certain position for each container in the yard storage area according to some rules adapted in the terminal, this case called tactical yard allocation problem. In order to attain a good state of container terminal operational planning at tactical level, terminal operators are faced with the challenge to both minimize the total violation time of vessel’s arrival and departure time, and minimize the yard transportation distance in a balanced way.

As one of the Indonesian highest-throughput container terminal, PT. Port of Tanjung Priok is experiencing the throughput escalation, particularly from 2015 to 2016, and confront those tactical problems on berth and yard allocation on their daily practice. Unfortunately, they still encounter the violation of vessel arrival and departure time, which undermining the shipping line schedule. This might happening as the result from the possibility of poor operational planning at tactical level and the lack of integration between berth and yard allocation planning.

To cope with this issue, we are trying to get an optimal model for integrated operational planning for ocean going terminal at PT. Port of Tanjung Priok, by adopting a mixed-integer programming model proposed by (Liu et al. 2016). We are developing the model with the consideration of the actual condition on the terminal as the parameter in order to generate a reliable output. The actual information provided by the terminal manager that counts into the consideration of the model are:

  1. Vessel arrival order, vessel length, and number of containers loaded into and discharged from the vessel;
  2. Total length and characteristic of terminal’s quay;
  3. Number of equipment (quay cranes) available in the terminal;
  4. Technical constraints determined by the terminal manager;
  5. Layout and capacity of terminal yard storage area.

Some integer variables are set as the major output, assisted with some other minor variables as the supporting of integration process. As well as the parameter, some constraints also generated based on the actual condition of the terminal. The model then solved with a bi-objective optimization method called ɛ-constraint method, thus we set two objective function which are i) minimize the total violation of actual and scheduled vessel arrival & departure time; and ii) minimize the total yard transportation distance representing the terminal’s operational cost.

In the daily practice of PT. Port of Tanjung Priok within the planning horizon of one week, noted that the current condition conform the small class according to generation by (Liu et al. 2016). Since the class is small-scale, it is possible to get an exact global optimum result from the model. Model will generate the optimal number of the two objective functions and the solution set of major variables generated by the model, which are:

  • the berthing position of each vessel in the planning horizon along the quayside of terminal;
  • number and scenario of equipment will be used at each vessel operation;
  • start and end time of vessel operation, representing the arrival and departure time;
  • storage position of each container in the yard area.

Then as the final result, we transformed the output into the berth and yard allocation map separately, for the ease of the practical use of the terminal. As the model solved with a bi-objective optimization method, output of the model includes the Pareto Frontier that represents the relation between the two objectives function. It resulted that total violation time has a contrast relation with the total yard transportation distance.

Since the model is considering the actual condition of the terminal, the result has high possibility to be applied in the actual practice at PT. Port of Tanjung Priok. Along with that, the Pareto Frontier could be used as decision support tool to determine the set of scenario that suit the terminal’s particular condition, related to the time violation and yard transportation distance. But, regarding to some sort of limitations within the development of the model, this projects is still needs further improvements in order to get more reliable result.

This research is conducted by Rizka and Komarudin

, , ,

No Comments

Workshop on Data Analytics and Visualization

Data Modeling and Visualization is one of the branches in our optimization research. Our new research groups, Signifier Analytics, has been established to foster the development of knowledge and experience on how data can turn to insights. Two of our prominent researchers, Komarudin and Aziiz Sutrisno, shared their experience on a half-day workshop. Pak Komarudin talked about his experiences as a data scientist for a mining company in Indonesia, where he must predict the weather for the following weeks of operations based on previous year data. Pak Aziiz shared his experiences in analyzing social media for trends and how to best visualizes these data for giving insights.

42592177_10217179288620471_4252236335659614208_o

The workshop is conducted in two waves, one for our students and one for external publics. The external workshop is conducted at Kekini Co-Working Space, Cikini, Jakarta

, , , , ,

No Comments

Reducing Cost of Road Maintenance for the Resources Industry

production-1891426_640Haul road in open mine has short durability. It is because haul road is constructed without asphalt/concrete pavement and passed by big vehicle with heavy load. There is some kind of decreasing quality of haul road such as improper cross section, inadequate roadside drainage, corrugations, potholes, ruts, and loose aggregate. Poor haul road quality will impact on increasing production costs and decreasing mine productivity. Usually open mine use motor grader to maintain the quality of haul road. Way of working of motor grader is to scrap the inadequate haul road surface.

There are some differences among the haul road segments such as characteristic, traffic density, kind of decreasing quality, durability, etc. Therefore, systematically grader route and schedule is needed to minimize the delay of haul road maintenance. Usually grader route and schedule just based on grader’s operator experience. There is no specific approach that can be used in grader route and schedule.

This research focused on grader route and schedule optimization in coal haul road maintenance. Optimization model in this research is designed using Bandit Algorithm. The objective of the optimization model is to minimize the maximum penalty. In this case, penalty is used to describe amount of loss that is caused by maintenance delay on each haul road segment. Grader start from the initial point to a road segment and moves over and over to the other road segment until working hour is over. Determination of he next road segment is based on weight of maintenance delay on each road segment. Greater the weight of the maintenance delay of a road segment, greater the probability of that road segment to be addressed by grader. Grader scraps if the road segment is late maintained and just passes if otherwise. When the working hour is over, grader stops moving and optimization model calculates the objective and records the route as a new solution. The steps are done again as many as have been determined (iteration). Solution with the best objective is chosen as the final solution.

With the probability, grader is not directly addressed to the road segment with the greatest maintenance delay weight to allow for the other road segments to be the next grader destination. This is because short term solutions have effect on long-term solution (whole solution); maybe the best short-term solution is not the best long-term solution. An example in a simpler problem is: we must determine route from city A to city D with 2 possible route that are A-B-D and A-C-D. With the distance between A-B < A-C, A-B is the best first movement. But for the overall movement, maybe A-B-D is not the closest route. Although the distance between A-B < A-C, distance of A-C and C-D can be closer than A-B and B-D.

Optimization model showed a significant cost savings for the mining operations by creating a more effective roads maintenance with reduce cost. With the pressure of low prices in the resources industry, a simple but yet complex optimization can help them stay more competitive.

This research is conducted by Denni and Dr. Komarudin.

,

No Comments

This is What You Should Do When You Have No Doraemon’s Magical-Anywhere-Door

There are times when we wish Doraemon’s magical-anywhere-door really does exist. If so, we can reach our destination without having to travel miles in a long time. But, since Doraemon does not exist, fortunately there is something we can do to at least save the mileage and time we sacrificed to travel from one to another places. Especially in case of post men or delivery couriers who have to visit a lot of destinations in such limited amount of time and capability.

Recently, a research assistant from Systems Engineering, Modeling, and Simulation Laboratory from Industrial Engineering major, Universitas Indonesia is conducting a research about Vehicle Routing Problem with Time Windows to optimize distribution route and schedule. It was first inspired by Indonesia’s current logistic condition which is still not optimal. It is proven by the decreasing index of Indonesia’s logistic performance during the past five years. One of the reason is the high Indonesia’s logistic cost which can be considered as the highest logistic cost in the world. And the one that contributes almost half of the logistic cost is transportation cost.

On the other side, customer needs are rising annually. Moreover, the growing online shopping market create an increasing demand of same-day delivery service. Based on McKinsey survey, online retailers as the main originator of B2C shipments, have a large interest to reduce delivery time in order to foster the products sale. Therefore, we need a better planning of distribution route and schedule, especially for delivery service providers and courier companies.

The purpose of research conducted at SEMS Laboratory about Vehicle Routing Problem with Time Windows (VRPTW) is to find the most optimum distribution route with lowest total distance yet still manage to fulfill all demand and considering the constraints of vehicle capacity and customers’ time windows.

Since VRPTW belongs to NP-hard optimization problems, our researcher used heuristic method which is translated to Netbeans 8.1 software in C++ programming language. They also use local search to perform simple iterations to produce fairly accurate solutions. Local search methods which are used in this study are exchange, two-opt, and insert. They can be applied to customers in the same route (intra-route) and customers in different routes (inter-route). Basically, these methods perform some iterations of moves that determine the most optimum combination and sequence of customers to visit. This will be finalized by using Lin Kernighan Helsgaun algorithm. The iteration will stop once it can not generate a better solution.

Fig 1. Illustration of 2-opt intra-route move

Fig 1. Illustration of 2-opt intra-route move

The resulting improvements will ultimately result in reduced transportation costs. Thus, vehicle routing problem with time windows can be a solution for urban logistic problems. So, when you do not have any Doraemon’s magical door, make yourself sure that at least you have a well-planned distribution route and schedule!

This Research is Conducted by Vincencia Sydneyta and Dr. Komarudin

, , ,

No Comments

How to optimize unequal area facility layout to maximize storage

work-1713103_640Facility layout problem, especially with the unequal departmental area (UAFLP), is one of the problems studied in combinatorial optimization and has received the attention of many researchers in the past decade. The goal of UAFLP is to allocate departments into a facility to obtain the most efficient arrangement. The UAFLP study has a final objective to minimize the total cost of material handling between departments.

Competition in today’s business world is inevitable. Increasing the quality of productivity becomes one of the keys to success in facing competition with business management effectively and efficiently. This can happen by maximizing existing resources ie employees, machines, and other facilities. Therefore, the industry needs to be able to optimize production capability and effectiveness to face competitors. The production process becomes the key that needs to be managed effectively to minimize production costs with higher effectiveness.

Facility layout design has a close reinforcement to the size of the physical arrangement of elements in a manufacturing and service system, such as department, machinery, operational tools, and so on. The purpose of the facility layout design is the design with the minimum material handling cost. With proper facility layout, material handling costs can be reduced. In general, material handling contributes about 20-50 percent of the total. Reduction in the company’s operational costs, along with the increased efficiency of the production system becomes a necessity that every industry needs to do.

The problem of facilities layout that is often the researcher’s attention is Unequal Area Facility Layout Problem (UAFLP). Initially, UA-FLP was developed by Armor and Buffa (1963). They explain that there is a rectangular facility with a fixed Width and Height and several (n) departments that need to be allocated to the facility. There are some problems that have not found the optimal solution and require a long computation time. The goal is to reduce non-feasible solutions to reduce the complexity of possible solutions.

This research will develop a mathematical model using Mixed Integer Programming method based on Flexible Bay Structure. Some additional constraint functions will be attempted to be added to the model. Testing is done by comparing the effect of each additional constraint function that has different approaches in cutting the complexity of possible solutions. The comparison result of the combination of the constraint function used indicates which constraint function has a major influence in reducing the computation time of the model.

This research concludes that all simulated problem sets using additional constraints in the model can provide better computation time than before. The effect that occurs can be the value of the solution becomes not optimum or remain optimum. All computational time gives significant improvement.

This research is conducted by Randa Adi Saputra and Komaruddin

, ,

No Comments