Home > Research > The analysis of work period and wages cost for interior finish work of mansion construction by using agent-based modeling (Torigai, 2011)

The analysis of work period and wages cost for interior finish work of mansion construction by using agent-based modeling (Torigai, 2011)

In recent years there is no significant difference among Japanese general contractors' building techniques, and along with the globalization process they cannot avoid conflicts with the emerging countries, which have strong competitive price. As a result the most important thing is management skill. One important indicator they have to consider is the cost reduction, and it is necessary to consider the wages payment since outsourcing is prevalent in this kind of business. Another indicator is the shortening of construction period. But traditionally there is a trade-off between these two indicators in the sense that improving one of them will make the other one worse.

In this study, in order to achieve Pareto optimization that improves both indicators, we only increase the number of workers of several selected jobs. In order to optimize the indicators during the simulation, we should solve the integer programming problem by exploring the cases that increase the number of workers from 0 to 10, for each of the selected jobs. However this problem is too complex to be solved efficiently, so we simplify the problem to two steps. By applying this simplified method, we represent the complexity of interior finish work of apartment, and propose a production plan to improve both indicators mentioned above. SOARS is used to construct this model.

Daisuke Torigai, Master of Engineer, 2011)

Details

1. Background and Purpose

The scheduling methods that consider the wage for workers is important for general contractors in Japan. It is because general contractor must pay the wages for workers and the ratio is really large. Therefore, we make a agent-based model (ABM) that can analyze process control of general contractor and analyzing by using it. A real data come from Takenaka Corporation is used in this model.

This research focused on two aspects of process control. First one is the complexity of interior finish work of mansion. It is said that the interior finish work has lots of process and the relation between works is complex, thus the management of the progress is difficult. The second one is trade-off between work duration and cost. In the projects for general contractor, shortening work duration and reduction of wage cost are important. But the relation between work duration and wage cost is trade-off like Fig.1. Therefore, we propose the method for analysis of the number of input worker which can improve trade-off index like work duration and wage cost under complex interior finish work.

2. Methodology

2.1. Previous Research

There is several previous research like analyzing process control of construction work by Ueda et al. This approach, however, doesn't focus on interior finish work, thus there is no dynamical process and complexity of the relation among workers. There is also an approach called critical-path method. The critical-path method isn't enough when there is resource, which means workers, competition.

2.2 Apply of Agent-Based Modeling

In this research, we apply agent-based modeling to make a model which represent the characteristics of interior finish work. Agent-based modeling is a method in which individual agents behave according to their decision making criteria and we can analyze some aspects under restriction and index. We can analyze lots of things in process planning under construction restriction by agent-based modeling.

4. Model

This model considers some real work data that comes from Takenaka Corporation. This model is constructed by using social simulation language SOARS.

4.1. Model Overview

In ABM, the design of each agents and the environment is needed. A worker is agent and the environment is consisted from workspace, site office and worker's station. The workspace represents the room in mansion. The site office represents the spot that has information about progress and room state. The worker's station is a spot for worker waiting to next work. The worker represents actor who implement work in workspace. Each worker has job and decide action according to job. The simulation step is 6 min.

4.2 Detail Design of Model

We design the model about worker and workspace in detail here. The workspace, which represents room, has a variable called finished-work-list. The finished-work-list is a set that contains finished work of each room.

Each worker has a variable called work-list that is defined by job. The work-list is a set that contains works the job can implement. A variable called precede-work-list is also set to each work. Each worker compares their precede-work-list and finished-work-list, then if the room is the one to be worked, the worker moves to the room and start to work. If the worker working in the room finished the work, he goes back to site office and update finished-work-list. Each worker repeats this series of work.

We also define the term of wage cost and worker's operation rate as evaluation index mathematically. We can analyze the trade-off by evaluating costs and efficiency quantitatively.

5. Proposal of Searching Method for Worker Input Combination

We propose a method which can improve trade-off relation like work duration and wage cost by changing some parameters search good input solution. In this model, there are 42 types of job. Thus, for example, if we want to consider changing input parameter from 0 to 10, we have to search 10^42 parameter state when we attempt to get the best solution. This is not realistic method.

Therefore, we propose the method searching Pareto optimum set in the case of 42 job combination D = (d1,d2,・・・,d42) and analyze using it. The method is as follows.

  1. Restrict jobs by following 2 steps,
    1. Scoring the importance of jobs by using some evaluation index and reducing 42 jobs to 15,
    2. Identify the most influential job in 15 jobs for evaluation index by using design of experiment.
  2. As to identified job, we search Pareto optimal set by changing the number of worker.

6. Simulation Result and Consideration

By using this model, we have taken simulation result and used the method for worker input combination. Consequently, we can find 15 influential jobs by using (1)1-1 step. We also get the most influential job by using (1)1-2 step. The job is worker of floor heating. We can get the evaluation index through the result of (2) step like Fig.2 and Fig.3. The intuition of this graph shows that if we change the number of worker of floor heating from 2 to 3, the index that contains trade-off can get Pareto improvement.

7. Conclusion

In this research, we build a model that can represent the structure of wage cost and work duration of the interior finish work. We also propose a new method that investigates good worker input combination. Therefore, The index like wage cost and work duration, which have been considered as trade-off relation, can be improved simultaneously by using this method. This method is practical in terms of calculation because it can reduce the combination of worker input.

8. Future Work

General contractor cannot manage the schedule of workers in detail in the real world. The general contractors have to change their strategy that can manage the schedule of workers in detail in order to compete with global firm. The research in response to this phenomenon will be important in the future world.

to Page top
Copyright (C) Deguchi Lab. Since 2002. All rights reserved.
degulab[at]cs.dis.titech.ac.jp