Background of the study:

The Fast-Moving Consumer Goods (FMCG) industry faces challenges in

meeting customer demand while minimizing costs and maximizing
productivity. Production scheduling is a critical aspect of manufacturing in
the FMCG industry, and it involves allocating resources to tasks in a certain
sequence and over given periods of time. The FMCG industry faces particular
challenges with production scheduling due to high volume, short lead times,
and high customization. Manufacturers must optimize their production plans
and implement effective production scheduling strategies to respond quickly
to changing customer demand patterns. Challenges include high demand
variability, product complexity, and coordinating production activities across
the entire supply chain. Addressing these challenges requires innovative
approaches to production scheduling, including specialized software tools
and approaches that generate optimal production schedules based on a
range of criteria.

Problem statement:

 Unilever Vietnam faces difficulties in production scheduling during fluid

container manufacturing due to the nature of the manufacturing
system.
 Major obstacles include the requirement for machine changeover
between different product types, which leads to increased downtime
and reduced production efficiency.
 Fluctuating demand for various product types is another significant
challenge that needs to be considered in production scheduling to
ensure the right products are produced at the appropriate time while
minimizing production costs.
  The problem is considered as a Hybrid Flowshop Scheduling Problem
(HFSP) in operations research.
 The proposed scheduling problem is Flexible & Hybrid Flowshop
Scheduling Problem with Demand-varying (FHFSP-DV) and deals with
demand allocation in a planning horizon.

Objectives:

 Develop and implement a production scheduling model for Unilever

Vietnam's manufacturing system
 Optimize production schedule to maximize resource efficiency,
minimize downtime, and reduce production costs
 Consider machine changeover times, demand variability, and
production costs
 Ensure adaptability to changing market demands and production
capacities
 Improve overall efficiency and profitability of Unilever Vietnam's
manufacturing system
 Address complex production scheduling challenges during fluid
container manufacturing.

Unilever Bottle Manufacturing Procedure:

This passage describes the plastic bottle manufacturing process, which

involves transforming raw materials into finished plastic bottles through
steps such as resin preparation, bottle molding, trimming, and finishing. The
two main workshops proposed to resolve scheduling operations are bottle
molding and labeling. The molding process can be done through either blow
molding or injection molding. After molding, the bottle may undergo
additional finishing processes, such as polishing or labeling. Labeling
involves applying a label or graphic to the bottle, typically including product
information such as the product name, ingredients, and manufacturer
information.

Literature Review:

(Đọc table trên slide)

Methodology:

Exact methods are optimization techniques utilized in operations research

that aim to find the exact solution for a given problem. These methods are
grounded in mathematical programming models that ensure the optimality
of the solution. Linear programming, integer programming, and mixed-
integer programming are the most commonly used exact methods in
operations research.

Recently, constraint programming (CP) has emerged as a potent exact

method in operations research.

 Constraint programming (CP) is a powerful exact method in operations

research that can handle complex constraints and non-linear
relationships.
 CP provides guaranteed optimal solutions to problems, which is useful
in situations where suboptimal solutions come at a high cost.
 The balance between model complexity and solution quality is crucial
when using CP to solve problems.
 The selection of appropriate search algorithms is important in
achieving good solution quality and computational time.

(Constraint programming (CP) is an indirect exact method in operations

research. This means that instead of searching the entire solution space, CP
uses constraints to prune large portions of the search space and guide the
search towards a solution. CP relies on an underlying search algorithm, such
as depth-first search or branch-and-bound, to find the optimal solution.)
Approximate approaches are often utilized in operations research to solve
optimization problems where finding the exact optimal solution is either
infeasible or too time-consuming. In such cases, approximate methods that
provide an acceptable solution within a reasonable amount of time are
preferred. Metaheuristics are a type of approximate approach that are widely
used in operations research to solve complex optimization problems.

Metaheuristics are iterative algorithms that aim to improve a set of

potential solutions to find a good solution, but not necessarily the exact
optimal one. They are advantageous in handling complex problems that
exact methods may struggle to solve, easy to implement and adaptable, and
can be combined with exact methods to improve solution quality. However,
selecting the appropriate algorithm and tuning its parameters can be
challenging, as performance can be affected by factors such as problem size,
structure, and complexity, and different algorithms have varying strengths
and weaknesses depending on the problem.

The combination of Constraint Programming and Metaheuristics

In fact, FHFSP is a well-known NP-hard problem that arises in manufacturing

systems. This problem is challenging due to its combinatorial nature and the

presence of various constraints. Constraint programming (CP) is a powerful

technique used to model and solve combinatorial problems. On the other

hand, metaheuristic algorithms are effective optimization methods that can

be used to solve complex problems such as the FHFSP. A recent approach

that has gained significant attention in solving the FHFSP is the combination

of CP and metaheuristics.
In this approach, CP is leveraged to formulate the problem model and its

constraints. The CP model is used to generate feasible solutions that satisfy

the constraints. Then, metaheuristic algorithms are used to improve the

quality of the solutions generated by the CP model. This approach allows for

the use of metaheuristics to explore large search spaces and find high-

quality solutions while taking into account the constraints of the problem.

Conceptual Design Model:

Phase 1: Problem Identification and Methodology Selection

In this phase, the researcher identifies the problem to be solved and selects
the appropriate methodology to address the problem. The researcher should
consider the scope of the problem, research questions, data availability, and
research objectives. This phase includes the following activities:

- Literature review: This activity involves reviewing the existing

literature to identify gaps in knowledge and to understand the current
state of research in the field.
- Problem identification: This activity involves identifying the problem
to be addressed and defining the research question.
- Methodology selection: This activity involves selecting the
appropriate methodology to address the research question. The
researcher should consider the strengths and limitations of different
methodologies, such as surveys, case studies, experiments, and
simulations.

Phase 2: Data Processing

In this phase, the researcher collects and processes the data required for the
study. The data processing phase includes:

- Data collection: This activity involves collecting data using various

techniques, such as surveys, interviews, and observation.
- Data cleaning: This activity involves cleaning the data to remove
errors, inconsistencies, and missing values.
- Data transformation: This activity involves transforming the data
into a suitable format for modelling. This may involve aggregating or
disaggregating the data, standardizing the data, or converting the data
into a different format.
- Data integration: This activity involves integrating data from
different sources to create a comprehensive dataset.

Phase 2: Modeling

This phase involves developing a mathematical model that represents the

problem to be solved. There are steps to follow:

- Objective determination: This activity involves defining the

objective of the model. The objective could be to minimize or maximize
a particular quantity, such as cost, profit, or efficiency.
- Constraints building: This activity involves defining the constraints
that must be satisfied by the model. Constraints may be related to
resources, capacity, time, or other factors.
- Model development: This activity involves developing the
mathematical equations that represent the problem. The researcher
should consider the appropriate level of detail required for the model
and the appropriate mathematical techniques to use.
- Model testing: This activity involves testing the model using sample
data to ensure that it produces reasonable results.
 - Model modification or finalization: Based on the test results, the
researcher may need to modify the model to improve its accuracy or
efficiency. Alternatively, the model may be considered final and ready
for analysis.

Phase 4: Result Analysis

In this phase, the researcher analyzes the results generated by the model
and draws conclusions based on the analysis following the basis of analysis
structure:

- Solution generation: This activity involves generating solutions using

the model. The researcher should consider the appropriateness of
different solution techniques, such as exact methods, metaheuristics,
or hybrid methods.
- Sensitivity analysis: This activity involves analyzing the sensitivity of
the model to changes in the input data or model parameters. This can
help identify critical factors that affect the model's performance.
- Scenario analysis: This activity involves analyzing the impact of
different scenarios on the model's performance. Scenarios may involve
changes in external factors, such as market demand or resource
availability.
- Performance evaluation: This activity involves evaluating the
performance of the model based on various criteria, such as solution
quality, computational efficiency, and practical feasibility.
- Conclusion drawing: This activity involves drawing conclusions
based on the results and analysis. The researcher should consider the
implications of the results for theory, practice, and policy, and identify
avenues for future research.