Martin Fischer, Oussama Khatib, Rui Liu
Careful and detailed observation of construction work showed that a lack of context information for construction crews can lead to rework, delays, change orders, or loss of productivity.
Primary Research Objective:
This research will formalize a context information model and research to what extent context information can be generated automatically.
Potential Value to CIFE Members and Practice:
- Help understand what context information is important for construction crew members and jobsite tasks
- Improve production management and control based on thorough analysis of context information
- Finally, help improve productivity by reducing rework and wasted resources (time, material, etc.)
Research and Theoretical Contributions
- Understand Context for Construction Crews at the Workface
- Build a Context Information Model for Construction Crews
- Develop an Algorithm for Retrieving Context Information
- The result of this research can be implemented to help construction crews acquire necessary context information so that they can improve the productivity for their tasks at the workface.
Industry and Academic Partners:
Research Updates & Progress Reports
To understand what context information is important for construction crew members, we have been working with Bouygues to conduct a field study on a jobsite in Paris.
Our collaboration with Bouygues includes two stages. For the first stage, we spent one week exploring the jobsite to get an overview of the project. The project is currently at the its very early stage.
In between the two stages, we developed a jobsite tracking form for field engineers to track the context information that is important for construction crew members. Thereafter, once a good amount of data has been collected, we may perform analysis over the data and explore what context information has been neglected by field crew members and what is the impact of missing the context.
For the second stage, we will research to what extent the context information can be automatically captured.
Progress Report - April 2020
1 Research Problem
According to the value stream mapping (VSM) theory, construction crew members at the workface are at the very last stage of the value flow to deliver the final building product (Arbulu et al. 2003). Hence, the quality of the finished work is largely dependent on the way construction crew members perform their tasks. In order to quickly and safely deliver a building product that satisfies the client’s requirements, construction crew members at the workface should be able to acquire necessary context information that is of interest to their tasks. However, this is not the case in reality. Even though there is no consensus on the meaning of context, for now, it can be simply considered as the relevant information for construction crew members.
To further investigate this problem, we conducted a first-run study on a precast parking structure project. More specifically, we recorded time-lapse videos for workface operations for the installation of precast building components (e.g., shear walls, columns). We also took notes of
problems that took place during installation. Thereafter, we analyzed the cycle time for each installation. A comparison was made between the average cycle time for precast elements with which problems were observed versus those with which no problem was observed. Figure 1
shows the average cycle time for installing shear walls with/without problems observed.
Figure 1 shows that shear walls with such problems took, on average, more than twice as long to install than those without such problems. The variation indicates that workface operations can be improved to reduce cycle times of the installation. Moreover, as we recorded the problems associated with each installation cycle, we found that the delays could be mainly attributed to the inability to synthesize or update the context information at the workface. For example, a mismatch of the rebars beneath the shear wall with the cast-in-place cups can cost a lot of time for the installation, as shown in Figure 2.
Since this analysis was made only based on shear walls on the ground level, we extended the analysis to include other shear walls and precast columns, beams, and spandrels. We found that the installation of about 40% of the shear walls, 33% of the columns, and 25% of the spandrels
were affected negatively by the lack or change of context information at the workface.
2 Theoretical and Practical Points of Departure
To address the problem stated above, it is important to first understand the meaning of context for construction crews so it can be implemented for construction information management.
2.1 Definition of Context
There have been many efforts towards understanding, modeling, capturing, and implementing context (Abowd et al., 1999; Ryan et al., 1999; Zimmermann and Oppermann, 2007). However, the definition of context is still vague and researchers have not yet reached a consensus. The
difficulty with defining context is that context, in essence, is a subjective concept. Simply put, context can be considered as the information of interest to an entity. For different research fields, the meaning of context may vary. In order to understand the context in the construction industry, especially for construction crews , the researchers found the following definitions for context in the literature:
- based on enumerating examples (Brown et al., 1997; Gross and Specht, 2001);
- based on providing synonyms (Brown, 1995; Zimmermann and Oppermann, 2007);
- based on categorizing information of an entity (Schilit et al., 1994; and Ryan et al., 1999); or
- based on analyzing the constraints for an action (Devlin, 2005).
Definition (1) is incomplete and definition (2) is too generic. For definition (3), a widely recognized definition is given by Abowd et al. (1999), who state that context is “Any information that can be used to characterize an entity.” This definition points out that context should be associated with an entity. In addition, definition (3) suggests that context should be categorized into different context types; some categorizations could end up being too specific. According to definition (4), context should be the information that constrains and impacts an action.
In addition, the above-discussed definitions of context are mostly based on a static view, whereas context is always subject to uncertain changes. Given the existence of uncertain changes, Adomavicius, et al. (2011) classified context information into two dimensions: contextual
variables and changing contextual variables. This classification indicates that recognizing uncertain changes is equally important as the variables themselves. Therefore, to utilize context information appropriately, it is necessary to account for the uncertain changes and imperfection of context information.
To summarize, context should be closely attached to an entity. An entity can be a person, an object, a place, which is determined based on the purpose of applying context. Furthermore, it is preferable that context consists of information that is of interest to the entity; when there is
some action taken by this entity, context will have impact on the process and the outcome of this action. To make the definition more specific, context can be categorized into different context types. Lastly, it should be noted that context is subject to changes over time. Hence, any context information should be evaluated to check whether it is correct for a certain state, and it will be helpful to have a sense of both the present and future context.
2.2 Construction Information Management
Having recognized the importance of context for construction crews, the researchers further investigated the information that constitutes context for crews. Kunz and Fischer (2009) discuss the use of POP (Product, Organization, and Process) models to represent and manage
information of the three major aspects of a project. POP models complement 3D product models with process and organization models to become a more comprehensive method for Virtual Design and Construction (VDC). Those three models are inter-connected and changes in one model often lead to changes in the other two. To generate work instructions that include context information, the information in POP models should be at the construction part level so that the information can be executable at the workface.
Moreover, Mourgues et al. (2008) developed the AROW (Activity, Resource, Object, and Work area) model to represent the information needed by construction crews. Based on this AROW model, Mourgues and Fischer (2008) developed a field instruction template to deliver taskspecific
information to construction crews. An example of the instruction is shown in Figure 3. As can be seen from this instruction, it contains different sections as work area, drawing, instruction, bill of materials (BOM), and equipment and tools. Nevertheless, this instruction does not
adequately and quickly update and transmit the context information for crews.
The AFM (Activity Flow Model) by Garcia-Lopez and Fischer (2016) uses the concept of activity flow to explain the reasons for a delay of a certain activity. Therefore, this model can be utilized to indicate the status of an activity and show the project schedule performance. However, the AFM model does not delve into reasons accounting for the schedule variability and does not sufficiently consider context information.
To sum up, this research uses the POP, AROW, and AFM models as the points of departure. A context information model for construction crews can be built based on those existing knowledge models for the construction industry.
3 Research Questions
This research investigates two research questions.
3.1 What constitutes context for construction crews on the jobsite?
This question asks about the meaning of context for construction crews. To answer this question, we need to first understand what context means given that there is no consensus on the definition of context yet.
3.2 How to automate the generation of context information?
This question investigates how to generate context information automatically. To do this, we will explore, develop and test algorithms to retrieve information from a context information model.
4 Current Progress
The researchers have built a context information model which consists of different context types and defines the relationships between those context types based on literature review. Further, the researchers conducted field studies in five projects, and observed 60 field tasks.
Currently, the researchers are performing analysis to evaluate whether the context information model can explain what has been observed from the sites. Thereafter, the researchers are planning to implement the triangulation method by conducting formal interviews with industry
experts to evaluate whether the context information model sufficiently captures context information for construction field crew members.
- Abowd, G.D., Dey, A.K., Brown, P.J., Davies, N., Smith, M. and Steggles, P., 1999, September. Towards a better understanding of context and context-awareness. In International symposium on handheld and ubiquitous computing (pp. 304-307). Springer, Berlin, Heidelberg.
- Adomavicius, G. and Tuzhilin, A., 2011. Context-aware recommender systems. In Recommender systems handbook (pp. 217-253). Springer, Boston, MA.
- Arbulu, R., Tommelein, I., Walsh, K. and Hershauer, J., 2003. Value stream analysis of a reengineeredconstruction supply chain. Building Research & Information, 31(2), pp.161-171.
- Brown, P.J., 1995. The stick-e document: a framework for creating context-aware applications. Electronic Publishing-Chichester-, 8, pp.259-272.
- Brown, P.J., Bovey, J.D. and Chen, X., 1997. Context-aware applications: from the laboratory to the marketplace. IEEE personal communications, 4(5), pp.58-64.
- Devlin, K., 2005. Confronting context effects in intelligence analysis: How can mathematics help. Center for the Study of Language and Information, Stanford University.
- Kunz, J. and Fischer, M., 2009. Virtual design and construction: themes, case studies and implementation suggestions. Center for Integrated Facility Engineering (CIFE), Stanford University.
- Garcia-Lopez, N.P. and Fischer, M., 2016. A construction workflow model for analyzing the impact of in-project variability. In Construction Research Congress 2016: Old and New Construction Technologies Converge in Historic San Juan, CRC 2016 (pp. 1998-2007).
- Gross, T. and Specht, M., 2001. Awareness in context-aware information systems. In Mensch & Computer 2001 (pp. 173-182). Vieweg+ Teubner Verlag.
- Mourgues, C. and Fischer, M., 2008. A work instruction template for cast-in-place concrete construction laborers. Center for Integrated Facility Engineering. Stanford University. Working Paper, 109.
- Mourgues, C., Fischer, M. and Kunz, J., 2008. Method to Produce Field Instructions from Product & Process Models for Cast-In-Place Concrete Operations (Vol. 110). Working Paper.
- Ryan, N., Pascoe, J. and Morse, D., 1999. Enhanced reality fieldwork: the context aware archaeological assistant. Bar International Series, 750, pp.269-274.
- Schilit, B., Adams, N. and Want, R., 1994, December. Context-aware computing applications. In Mobile Computing Systems and Applications, 1994. Proceedings., Workshop on(pp. 85-90). IEEE.
- Zimmermann, A., Lorenz, A. and Oppermann, R., 2007, August. An operational definition of context. In International and Interdisciplinary Conference on Modeling and Using Context (pp. 558-571). Springer, Berlin, Heidelberg.
Original Research Proposal