Deguchi Lab. LogoDeguchi Lab. Sub Logo

Home > Research > Model building of collective housing interior work in which valuation analyses of work period and cost are possible (Igarashi 2012)

Model building of collective housing interior work in which valuation analyses of work period and cost are possible (Igarashi 2012)

In interior work of collective housing construction, there are various employment types and work relations, and traditional management method has its limits. From such a background, the method of the process control, which contain complicated constraints and could be used to manage individual worker is needed. In this study, based on the case study in previous research, the model for process control applicable to collective housing interior finishing work of various forms is built. In addition, in the previous work, the search method of the worker number which narrowed down the scope by using an experiment design method was proposed. In this study, in order to examine the validity and usefulness of the method, milestone interval analysis method with local search is compared with the previous one. Social simulation language, SOARS, is used for implementation.


For construction work, due to the mobilization characteristic of workers, and the divergent employment patterns, the production leveling of various jobs on construction site is very difficult. In order to improve the situation, different project planning methods are deployed during the construction work. However, with the popularity of high-rising buildings, extant methods cannot satisfy the increasing requirements of project planning. Therefore, for collective housing construction corporations, how to construct the collective houses efficiently is becoming a challenging problem.

Collective housing construction work usually starts from constructing the frame of the building, followed by interior finish work. For frame construction, limited types of workers are involved, thus the construction progress is very clear to understand and the management is comparatively easy. However, in contrast the interior finish work requires variant types of workers and the number of scheduled jobs are large. Besides, the scheduling order of jobs is complicated and the construction progress might vary from site to site. As a result, the management is becoming extremely difficult and challenging.

From 2008, Komatsu and Torigai from our lab started to work on the collective housing construction field trying to tackle problems rising from the involved interior finish work by applying agent-based approach. Following their works, this studying is collaborated with Takenaka corporation, and continued to build agent-based model (ABM) which could be used to analyze the project management problems.


Starting from Komatsu’s research work, agent-based modeling is applied to model the interior finish work involved in the collective housing construction. Applying agent-based modeling could capture the employment discretion and types, which is hard to achieve by previous research methods. In addition, by considering the influences of changing the number of different types of workers, we could examine the resulted total work period of interior finish work with accordingly changed working efficiency and operation ratio. The simulation result is out of expectation in the sense that with the increment of workers, the work period is not decreased linearly; rather the decreasing rate is also slowing down. Therefore, there exists a trade-off between the cost and the work period. If the number of workers is increased, although the work period is shortened, the cost will increase accordingly. On the other side, if the cost is controlled and shrunk, it is very possible that the work period will extend as a result.

Although this kind of trade-off commonly exists in construction work, due to the complexity involved in interior finish work, there is no analyzing method in literature yet. In order to understand this trade-off further, based on Komatsu’s agent-based model, Torigai improved the model with empirical data collected from a real case. This agent-based model reflected the interior finish work construction process with complex constraining factors considered, and proposed a way to manage each individual worker, which was never handled by previous research work. From the simulation result with empirical data collected from the construction site, by understanding the influence of increment of a particular type of works on the status of other types of workers, we could find Pareto efficient solutions to improve the trade-off indications - cost and work period - simultaneously with particular compositions of different worker types required. Furthermore, he proposed an experiment design to explore the efficient combination of different types of workers, and identified the particular worker type which should be increased. However, since this model was based on a particular completed mansion construction to show the improvement of both cost and work period, whether other cases could be integrated to this model directly, and whether this model was suitable for other cases were still not clear.

Research objective

The objective of this studying is to build a general agent-based model to analyze the construction management in collective housing interior finish work, based on Torigai’s case study. Specifically, the model could analyze the management in collective housing interior finish work, and find the optimized number of workers needed from examining the trade-off indicators, taking data about the building, workers, and tasks as input, and the working period and cost as output. Finally, this model should be able to support the decision making of general contractors when deciding the number of workers required for collective housing interior finish work.


This work aims at constructing agent-based models which could be used to evaluate the management of collective housing interior finish work. The overall structure of the model is as shown in Fig. 1. The model treats workers with decision making ability as agent, the working site of workers as spot. The social simulation language, SOARS, is used for implementation.

In the work spaces, all the rooms to be constructed are shown. In addition, the information such as the working progress table and the jobs which the workers should perform are recorded in the office as well.

We consider different types of workers as the production capability who will work on the scheduled tasks. Accroding to the worker type, workers might work on variant types of jobs. At each day, if there is any available job, then the workers will start to work and go to the office to check the progress table first. If there is any work left from yesterday, or there is any new work available, they will go the correponding work spaces. If there is no work at all, then they will step into the next stage to search for available works. When they go the work spaces, if there is no other workers in the room or the work has not been completed, then they will start to work. If there is other workers working in the spaces, or the work has been completed, then they will go to other spaces to work. If there is no available work at all, then they will go back to the office to search for other available works. If all the works are completed, they will go back to the office to update the progress table and go back home.

Comparison of explorative methods

By applying the aforementioned model, this research work discusses the explorative methods to find the optimized composition of workers. Torigai’s studying proposed a method of experiment design to identify the optimized composition of workers. However, the usefulness of this method has not been discussed and verified yet. Therefore, this studying aims to examine the usefulness of the previous experiment design method. In order to achieve this, we will compare the following two methods in terms of the improvement rate of the two indicators from the simulation result.

  1. Experiment design method on the overall construction work (Torigai’s research)
  2. Milestone interval analysis method on each individual term of the construction work (This research)

From the simulation result as shown in the following table, we could see that both of the methods improve the two indicators, but the experiment design method is better in the sense that the resulted improvement rate is more significant.

Conclusion and future work

In this work, based on the previous case study model, we construct an agent-based model applicable to variant types of collective housing interior fninsh work considering the complicated contraint factors. In addition, we examine the usefulness of the experiment design method proposed in previous reserach work to identify the opmitized composition of different types of workers. From the simulation result, we could see that although our proposed milestone interval analysis method improves the two indicators, in terms of the improvement rate, the experiment design method performs better. It indicates that the experiment design method is more effective in identifying the types of worker whose number should be increased.

In future work, we could compare the experiement design method with other explorative methods which aims to find the optimized composition of workers to further examine its usefulness.

to Page top