INFORMATION TECHNOLOGY: USE OF SIMULATION TO OBTAIN IMPROVEMENTS IN LOGISTICS OPERATIONS

It is well known that companies that stand out for excellence in logistics intensively adopt new information technologies. A subset of these technologies are decision support systems. These are applications that assist management in identifying, evaluating and comparing operational alternatives. Among the various existing applications, computational simulation has emerged as one of the tools of increasing use in modern management, particularly in the areas of logistics and operations.

Observing the growing interest in this technique, we seek here to provide a succinct but comprehensive view of this powerful methodology. After a brief conceptualization, we answer the most frequent questions asked by professionals interested in its use, such as:

• What is simulation?
• When and where to use it?
• What can simulation provide us with as an answer when studying a problem?
• What computational resources are needed?
• What technical training is required?

In addition, some of the applications developed at the Center for Studies in Logistics will be reported, illustrating the potential of simulation as a tool to support decision-making. Finally, we intend to show that simulation can indeed become a powerful ally in the search for greater operational and logistical efficiency in the increasingly competitive world in which we live.

1. A NEW STAGE OF THE COMPUTERIZED ENVIRONMENT

There is no doubt that modern logistics has been heavily influenced by the evolution of information technology. This technological evolution provided advantages for logistical operations that became faster, more reliable, cheaper and more efficient. Another important contribution of this computerized environment was also the greater availability of information about processes, and the possibility of analyzing such information using more sophisticated quantitative tools that until recently were the privilege of a few large organizations.

Simulation thus appears as a powerful operational research tool that, although known since the early 50s, has only now become more accessible to a much larger audience.

The first applications of simulation in operations and logistics were in the areas of mining, steel and maritime transport. Today, strongly influenced by technological advances, new opportunities have emerged, covering virtually all links in the Supply Chain.

In fact, logistic systems are complex dynamic systems, involving several elements interacting with each other and influenced by effects of a random nature. Situations like this pose serious difficulties for an analytical study of the problem, making computational simulation a strong ally, if not the only one, for the design and analysis of logistics systems. In addition, as a strong incentive, we have increasingly powerful computers in our work environment, almost immediate availability and access to information and a range of software capable of assisting us in different decision-making situations.

2. WHAT IS MEANT BY SIMULATION?

Without worrying about a precise definition of simulation, we can understand it as the use of models for the study of real problems of a complex nature, through computational experimentation. Thus, simulation consists of the process of building a model that replicates the operation of a real or idealized system (still to be built!) and conducting computational experiments with this model in order to better understand the problem under study, test different alternatives for its operation and thus propose better ways of operating it.

In this way, observing Figure 1, we can summarize the main steps in a practical application of simulation in:

1. model building
2. Transformation of this conceptual model into a computational model specific to the experimentation process;
3. Experimental testing of action alternatives to choose the most appropriate ones;

2.1. model building

One of the fundamental steps in the simulation process is a good understanding of the problem under study and the construction of a model that best represents its operation. Although there are specific tools and approaches to the modeling process, this will always be a mixture of art and science. Initially, this model will be logical in nature, graphical representations on paper with numerous annotations. Afterwards, depending on the computational resource to be used, this logical model will be translated into a simulation program, also called computational model. At this point, the main objectives of the problem to be studied must be determined and what response the model should give to decision makers.

2.2. Computational modeling

We understand computational modeling as a set of actions aimed at translating the logical/conceptual model into an operational model. As such, this modeling covers three fundamental and also laborious steps of the simulation process:

• Data collection and statistical modeling;
• Programming, using software appropriate to the nature of the problem;
• Verification and validation.

In this way an operation or system is translated in terms of rules, actions and process times.

2.3. Experimentation

Once the computational model has been built and duly validated, we move on to the experimental phase in which the various alternatives under consideration will be tested. In addition, through simulation, we can carry out sensitivity and what-if analyzes to assess the effect of possible changes before they actually occur or are implemented.

As examples, we might be interested in assessing whether the effect of implementing a new picking system on vehicle loading time in a Distribution Center would be worth the cost of its acquisition. Likewise, we could assess the effect of different inventory policies on the level of service provided to customers, in terms of product availability and cost of inventory.

Using statistical experimentation processes and supported by consistent analysis of the results obtained, we will be one step closer to suggesting the best alternatives to follow or recommendations that may be necessary. This phase consists of testing different alternatives, using the representation of the real system, which is the simulation model.

3. WHERE TO USE THE SIMULATION?

Evaluating action alternatives has never been an easy task. Especially when the results of choosing a particular alternative are not fully predictable.

Before implementing a new process, we need to have an early idea of ​​its possible results, either to confirm our expectations regarding the benefits sought or to identify possible side effects. When we look at most logistics operations, we have the articulation of various supply chain functions and their inherent complexities. When purchasing, production, storage, replacement policies, material handling, physical distribution decisions must be taken in an integrated manner, simulation is a suitable tool to quantify the potential gains between each alternative and the effects of their interrelationships.

Thus, simulation is indicated for systems where the consequences of the relationships between its various components are not known a priori and difficult to translate in an analytical way.

Assume a simple example of sizing a logistics system for freight forwarding. The objective would be to quantify the necessary number of forklifts and workers per shift to handle a certain volume of cargo. We will usually have a high number of products, with different loading times, and sometimes shared resources (workers and forklifts being allocated to activities other than loading). Through simulation, a model of this dispatch system is replicated and tests with different numbers of forklifts and workers can be performed, revealing to the analyst the best decision between investment and productivity benefit.

In general terms, the simulation applies to types of problems where you need:

• Provide a better understanding of the nature of a process. With this, new ideas usually arise aiming at greater productivity.
• Identify specific problems or problem areas within a system, in particular bottlenecks, over-optimal buffer stocks, and potentially idle resources.
• Help us establish future investment strategies for an existing system, better showing when and how much there is to gain at each new stage.
• Test new concepts before their implementation and without interfering with the operation of a system currently in progress.
• Evaluate the benefits of new investments before there is an actual commitment of a company's resources.

For the specific case of applications in logistics operations, we highlight:

• Dimensioning of loading and unloading operations: determination of the number of docks, number and type of forklifts, cargo preparation area, etc.
• Inventory sizing: determination of safety stock and base stock in multi-link systems (central and regional distribution centers), considering uncertainties in the supply of raw materials and in the demand for products and their consequence on the level of service provided: Where should stocks be located? Centralized or distributed at the ends. What is the cost of serving our customers with 95% product availability? And 98%?
• Material handling study: evaluation of the cost/benefit ratio of implementing new equipment and new technologies such as conveyor belts, stacker cranes, automatic picking systems, etc.
• Transport system: determination of the ideal fleet in terms of number and size of vehicles, considering the profile of orders to be delivered, the duration of trips and loading and unloading times and the result on the use of vehicles, service time, etc.
• Production Flow: sizing of equipment and workstations. Evaluation of different configurations of resources: production cells, specialized lines, etc.
• Customer service in general: such as the number of POS's in supermarkets, customer service boxes in banks, etc.

4. APPLICATION EXAMPLES

The application of simulation as a decision support tool can become clearer through examples of research projects developed at the Center for Studies in Logistics in partnership with national companies. A model developed for port operations and another for loading fuel at distribution bases will be presented.

4.1. container terminal

This study was developed by the Center for Studies in Logistics together with Integral Transportes, which is part of the Lachmann group. Its main objective was the development of a tool to assist in decision-making regarding the sizing of resources at its Terminal located in Santos.

The use of this terminal is strongly linked to streamlining the port and retroport operation in the clearance of cargo carried out by the Federal Revenue Service. This type of terminal stores the cargo while the product nationalization process takes place.

Efficiency in system operation is associated with the correct balance between the number and type of resources used and the input and output flow of containers at the Terminal. What is intended is to identify the lowest cost configuration that is able to meet customer requirements in a given maximum time.

Several pieces of equipment are used in handling loads, loading and unloading trailers, including reach-stackers (Figure 2), gantry cranes and large forklifts. Equipment can cost from $350 to a few million dollars. These resources must have characteristics that allow the handling of 20 and 40-foot containers, which represent most of the cargo handled by the Terminal.

The use of the simulation model made it possible to analyze all the operations that the resources perform in the terminal, thus making it possible to generate different configurations of the number of resources (exclusion of stackers, inclusion of one more tower forklift, exchange for a more productive resource, etc. ).

Another advantage provided by the simulation was the possibility of varying arrival and departure rates of containers at the terminal, thus simulating different medium and long-term scenarios in terms of demand behavior. In this way, it was possible to evaluate the best configuration of the terminal for the current situation and also the future expansion needs in case of growth in demand.

4.2. Fuel Distribution Base

Fuel distribution begins at refineries, with products transferred and stored at the Distribution Bases, where tank trucks are supplied and fuel is mixed with the company's own products. From the Distribution Base, the products are sent to the company's final customers, such as service stations, large consumers and wholesalers. Figure 4 illustrates this distribution system.

A Distribution Base, in a simplified way, is composed of fuel storage tanks and bays for loading tank trucks. In each bay, which could be compared to a loading dock, there are nozzles for supplying each type of fuel.

Dimensioning a Distribution Base means determining the number of bays, the mix of fuel nozzles in each of these bays, that is, what type of fuel the nozzle must carry, its loading flow, the space required for parking the vehicles (proportional to the queue waiting for loading) among other aspects.

Bad sizing, in this case, is critical, since the installation of a base involves large investments, related to the fixed structure characteristic of this type of operation. Thus, excess capacity would imply very high operating costs and lack of capacity would imply long waiting times for vehicles to be loaded, which would lead to delays in deliveries and high transportation costs due to low fleet utilization.

To aid in the planning and dimensioning of this system, the Center for Studies in Logistics and Companhia Brasileira de Petróleo Ipiranga developed a simulation model. This model takes into account several characteristics, such as compartmentalization of tank trucks, different truck arrival rates and the possibility of testing different loading configurations.

Through the model, it was possible to test several configurations of the resources, such as operating with a greater number of bays and lower loading flow and vice versa. Different ways of constructing pens were tested, with different compositions of nozzles per pen. The operation of the base in different shifts was also analyzed: 24, 16 and 12 hours.

Some parameter settings yielded 40% reductions in load time. With this, a greater number of customers could be served or a reduction in the size of the fleet could be carried out. These benefits could then be accurately compared with the necessary investments.

Such tests would be unfeasible in the real system, as its cost is very high, demonstrating one of the great benefits of this tool in approaching such complex problems in the logistics sector.

These results guided the company in its decision making regarding new investments to improve loading time and, in the future, it is intended to apply this methodology in the planning of new distribution bases.

5. NECESSARY TECHNICAL SKILLS

The execution of a simulation project requires different technical skills. Due to the complexity, some projects may even require the participation of specialists. In fact, some companies, because they use this technique intensively, have created specialized teams. However, as simulation software is becoming more and more user-friendly, its learning is easier, allowing its use by an ever-increasing number of users.

In general terms, these skills are:

• Good computer base: necessary for learning the software and modeling the problem in question. However, this does not imply the indication of a professional in the field of information technology, but rather a person with a user profile, accustomed to the development of projects using software.
• Reasonable knowledge of statistics: necessary for the intensive use of data characteristic of simulation studies. Statistical knowledge is necessary for an adequate treatment of input data (statistical modeling) and a correct interpretation of the results that the model can generate.
• Knowledge of process analysis techniques: the professional or team must have mastery over all relevant details of the system, the relationships between its components and must have the ability to translate them into a set of logical rules. Of course, a good dose of sensitivity to the problem in question cannot be lacking. A simulation project with little involvement of people who work in practice with the system has a high probability of not achieving the desired objectives.

6. NECESSARY COMPUTATIONAL RESOURCES

The growing popularity of using simulation as a modeling and problem analysis tool has resulted in a vast and growing availability of simulation software on the market.

Today, with a micro Pentium, in a standard configuration (32Mb of RAM with 133 MHz), we already have a machine capable of processing quite complex and previously unimaginable applications. However, software has come to represent a crucial factor in the use of simulation.

The biggest acquisition cost does not reside in the computational platform, as seen, but in the cost of the software itself. We indicate a consultation with the main manufacturers, bearing in mind that prices are customized for each type of customer need. However, these prices are falling every year, being very affordable even for small and medium-sized companies.

Table 1 presents the main software and their suppliers. The omission of some products is not intentional, the objective was just to present the main software available on the market. The table indicates whether there are representatives of the software in the Brazilian market.

7. INFORMATION SOURCES

For those interested in going deeper into the subject and obtaining more information about this powerful tool, I suggest visiting the following sites below: