Bahan Ajar Minggu Ke-2 Simulasi Sistem

Bahan Ajar Minggu Ke-2

Tujuan Instruksional Umum Setelah menyelesaikan mata kuliah ini mahasiswa semester
6 mampu menganalisis output simulasi dengan Promodel
dari sistem nyata.

Tujuan Instruksional Menjelaskan pengertian sistem, model, dan simulasi.

Khusus

Introduction (2/2)

1. Simulation Costs And Risks

Simulation modeling, while generally highly effective, is not free. The main
costs incurred in simulation modeling, and the risks attendant to it, are listed
here.

Modeling cost. Like any other modeling paradigm, good simulation modeling
is a prerequisite to efficacious solutions. However, modeling is frequently
more art than science, and the acquisition of good modeling skills requires a
great deal of practice and experience. Consequently, simulation modeling can
be a lengthy and costly process. This cost element is, however, a facet of any
type of modeling. As in any modeling enterprise, the analyst runs the risk of
postulating an inaccurate or patently wrong model, whose invalidity failed to
manifest itself at the validation stage. Another pitfall is a model that
incorporates excessive detail. The right level of detail depends on the
underlying problem. The art of modeling involves the construction of the
least-detailed model that can do the job (producing adequate answers to
questions of interest).

Coding cost. Simulation modeling requires writing software. This activity can
be errorprone and costly in terms of time and human labor (complex software
projects are notorious for frequently failing to complete on time and within
budget). In addition, the ever-present danger of incorrect coding calls for
meticulous and costly verification.

1
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

Simulation runs. Simulation modeling makes extensive use of statistics. The

analyst should be careful to design the simulation experiments, so as to
achieve adequate statistical reliability. This means that both the number of
simulation runs (replications) and their length should be of adequate
magnitude. Failing to do so is to risk the statistical reliability of the estimated
performance measures. On the other hand, some simulation models may
require enormous computing resources (memory space and CPU time). The
modeler should be careful not to come up with a simulation model that
requires prohibitive computing resources (clever modeling and clever code
writing can help here).

Output analysis. Simulation output must be analyzed and properly interpreted.

Incorrect predictions, based on faulty statistical analysis, and improper
understanding of system behavior are ever-present risks.

2. Example: A Production Control Problem

This section presents a simple production control problem as an example of

the kind of systems amenable to simulation modeling. This example illustrates
system definition and associated performance issues. Consider a
packaging/warehousing process with the following steps:

1. The product is filled and sealed.

2. Sealed units are placed into boxes and stickers are placed on the
boxes.
3. Boxes are transported to the warehouse to fulfill customer demand.

These steps can be combined into a single processing time, as depicted in the
system schematic of Figure 1.

Figure 1. Pictorial Representation Of A Packaging/Warehousing System

2
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

The system depicted in Figure 1. is subject to the following assumptions:

1. There is always sufficient raw material for the process never to starve.
2. Processing is carried out in batches, five units to a batch. Finished
units are placed in the warehouse. Data collected indicate that unit-
processing times are uniformly distributed between 10 and 20 minutes.
3. The process experiences random failures, which may occur at any
point in time. Times between failures are exponentially distributed
with a mean of 200 minutes. Data collection also showed that repair
times are normally distributed, with a mean of 90 minutes and a
standard deviation of 45 minutes.
4. The warehouse has a capacity (target level) of R 500 units.
Processing stops when the inventory in the warehouse reaches the
target level. From this point on, the production process becomes
blocked and remains inactive until the inventory level drops to the
reorder point, which is assumed to be r 150 units. The process
restarts with a new batch as soon as the reorder level is down-crossed.
This is a convenient policy when a resource needs to be distributed
among various types of products. For instance, when our process
becomes blocked, it may actually be assigned to another task or
product that is not part of our model.
5. Data collection shows that interarrival times between successive
customers are uniformly distributed between 3 and 7 hours, and that
individual demand sizes are distributed uniformly between 50 and 100
units. On customer arrival, the inventory is immediately checked. If
there is sufficient stock on hand, that demand is promptly satisfied;
otherwise, customer demand is either partially satisfied and the rest is
lost (that is, the unsatisfied portion represents lost business), or the
entire demand is lost, depending on the availability of finished units
and the loss policy employed.

This problem is also known as a production/inventory problem. We mention,

however, that the present example is a gross simplification. More realistic
production/inventory problems have additional wrinkles, including multiple

3
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

types of products, production setups or startups, and so on. Some design and
performance issues and associated performance measures of interest follow:

1. Can we improve the customer service level (percentage of customers

whose demand is completely satisfied)?
2. Is the machinery underutilized or overutilized (machine utilization)?
3. Is the maintenance level adequate (downtime probabilities)?
4. What is the tradeoff between inventory level and customer service
level?

The previous list is far from being exhaustive, and the actual requisite list will,
of course, depend on the problem to be solved.

3. Project Report

Once a system has been modeled, simulated, and analyzed, the study and its
conclusions are often written up as a project report. A well-thought-out and
properly written report is essential to the success of the study, since its value
lies in its conclusions and their dissemination to technical and management
personnel. The goal of this section is to aid the reader in structuring a generic
report for a performance analysis study by outlining a generic skeleton.

A project report should be clearly and plainly written, so that nontechnical

people as well as professionals can understand it. This is particularly true for
project conclusions and recommendations, as these are likely to be read by
management as part of the decision-making process. Although one can hardly
overemphasize the importance of project reporting, the topic of proper
reporting is rarely addressed explicitly in the published literature. In practice,
analysts learn project-writing skills on the job, typically by example.

A generic simulation project report addresses the model building stages

consists of (at least) the following sections:

Cover page. Includes a project title, author names, date, and contact
information in this order. Be sure to compose a descriptive but pithy project
title.

4
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

Executive summary. Provides a summary of the problem studied, model

findings, and conclusions.

Table of contents. Lists section headings, figures, and tables with the
corresponding page numbers.

Introduction. Sets up the scene with background information on the system

under study (if any), the objectives of the project, and the problems to be
solved. If appropriate, include a brief review of the relevant company, its
location, and products and services.

System description. Describes in detail the system to be studied, using prose,

charts, and tables. Include all relevant details but no more.

Input analysis. Describes empirical data collection and statistical data fitting.
For example, the Arena Input Analyzer provides facilities for fitting
distributions to empirical data and statistical tests.

Simulation model description. Describes the modeling approach of the

simulation model, and outlines its structure in terms of its main components,
objects, and the operational logic. Be sure to decompose the description of a
complex model into manageable-size submodel descriptions. Critical parts of
the model should be described in some detail.

Verification and validation. Provides supportive evidence for model goodness

via model verification and validation to justify its use in predicting the
performance measures of the system under study. To this end, be sure to
address at least the following two issues: (1) Does the model appear to run
correctly and to provide the relevant statistics (verification)? (2) If the
modeled system exists, how close are its statistics (e.g., mean waiting times,
utilizations, throughputs, etc.) to the corresponding model estimates
(validation)?

Output analysis. Describes simulation model outputs, including run scenarios,

number of replications, and the statistical analysis of simulation-produced
observations. For example, the Arena Output Analyzer provides facilities for

5
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

data graphing, statistical estimation, and statistical comparisons, while the

Arena Process Analyzer facilitates parametric analysis of simulation models.

Simulation results. Collects and displays summary statistics of multiple

replicated scenarios. Arena's report facilities provide the requisite summary
statistics (numerous reports are displayed in the sequel).

Suggested system modifications (if any). A common motivation for modeling

an extant system is to come up with modifications in system parameters or
configurations that produce improvements, such as better performance and
reduced costs. Be sure to discuss thoroughly the impact of suggested
modifications by quantifying improvements and analyzing tradeoffs where
relevant.

Conclusions and recommendations. Summarizes study findings and furnishes

a set of recommendations.

Appendices. Contains any relevant material that might provide undesired

digression in the report body.

Needless to say, the modeler/analyst should not hesitate to customize and

modify the previous outlined skeleton as necessary.

4. Simulation Examples

Here are some example of sumulation.

6
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

7
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

8
 Bahan Ajar Minggu Ke-2 Simulasi Sistem

Sumber:

1. Altiok, T. and Melamed, B. (2007): Simulation Modeling and Analysis

with Arena, Elsevier Inc.