Generating Bills of Materials for Daily Work Orders Consisting of Made-to-Stock Materials
Project Team
Prof. Martin Fischer, Min Song
Overview
Based on the findings, we made a prototype that operates with LOD400 objects and SWFBs in a lab environment. However, the intended user of the prototype is foremen, so the prototype developed in a lab environment needs to be implemented in a field setting with foremen. By carrying out the field experiments in this proposal, we hope to iteratively formalize an ontology and methodology for rapidly generating a daily BOM for MTS types of materials.
Project Background
Research Motivation
ETO materials are materials that are typically engineered and fabricated at an upstream stage of a supply chain. The engineering and fabrication are usually done at a prefabrication shop - examples of the materials that fall under the category of ETO are ductwork, pipe spools, external cladding, and HVAC units. These types of materials tend to have a unique shop drawing and a unique BOM at one corner of the shop drawing. This BOM information is often stored in the computers of a manufacturer who operates a prefabrication shop. If one wishes create a system that rapidly generates a daily BOM for ETO materials, it can be done by utilizing this information.
MTS materials are materials that are typically stocked at a construction site and further processed at the site like cutting, fabricating, or spraying/applying - examples include drywall track and studs, sheetrock, and on-site cut pipes or conduits. These types of materials, unlike ETO materials, do not typically have a product-based, unique shop drawing with BOM. The work involving MTS materials tend to have a shop drawing for a typical part of a design. Since ready-made and product-based BOM information is typically unavailable for MTS materials, extra time and effort are needed to generate a daily BOM in comparison to ETO materials.
Industry Example
We visited five construction sites where daily work orders were manually made (without BOM information included in them). From those sites, we collected 831 cards (daily work orders) and conducted a study on whether a system for rapidly generating a daily BOM can be created or not. It turned out that this can be done for 244 cards (out of 831 cards) using the pre-existing information at the sites like shop drawings or construction documents. These were cards related to ETO materials. The remaining cards that could not be made to do so were the cards related to MTS materials (except for a card that did not have any materials related, i.e., a card for concrete surfacing).
Another study we conducted with the 831 cards was related to the level of development (LOD) defined by the American Institute of Architects (AIA). One observation was that all the five construction sites did not have LOD400 BIM and this made us wonder how many of 831 cards would have corresponding objects in BIM had the LOD400 BIM been available at the sites. The study showed that 97.6% (811 out of 831 cards) of the cards would have corresponding objects in BIM if the sites had LOD400 BIM. The intuition we got from this study was that LOD400 perhaps can serve as a basis for creating a system that helps foremen rapidly generate daily BOMs, even for MTS materials, since most of the cards would have corresponding 3D objects when LOD400 BIM is used.
Prototyping
With the intuition, we first created a prototype that imports LOD400 objects (related to MTS materials) and allows a foreman to select 3D objects with a touch interface. However, we quickly found out that there had to be a better method for selecting and scoping 3D objects since the number of objects is unmanageably large in LOD400 BIM. Hence the question arose how to aggregate LOD400 objects into a unit of work for a faster generation of a BOM. The theories discussed by prior researchers (Hopp and Spearman 1996; O’Brien 1998; Morkos 2014) shed light on this question, which showed the benefits of creating a small batch size in production. Building on the theories and our subsequent case study at a residential project (which observed the smallest batch size for 46 tasks), we started to test a concept we refer to as the smallest workface boundary (SWFB) that encapsulates LOD400 objects in the smallest workable unit and incorporated the concept into the prototype.
Research Objectives
This research aims to apply the concept of a SWFB developed in a lab environment to field environment, and formalize it through iterative field experiments. The formalization involves iteratively developing the concept of SWFB so that a foreman at a site can rapidly generate a daily BOM that includes MTS materials information on a fabrication level (i.e., LOD400). At each site, we ask foremen to generate a daily BOM with SWFBs and if a shortcoming is found, we address this and the improved version of the prototype is tested at the next site. Process metrics, such as the time it takes to generate a daily BOM, are logged to make comparisons between field experiments and different versions of the prototype.
Field Experiments (in progress)
As of March 2018, we have conducted two field experiments. The first experiment was in Abu Dhabi and the second one was in Stockholm. In the first experiment, three foremen were involved in the experiment in a span of three weeks and they generated 46 daily BOM cards for mechanical, glass, and ceiling frame trades using the prototype (Figure1). In the second experiment, also three foremen were involved in a span of three weeks and they generated 33 daily BOM cards for facade, planter box, and protective concrete trades using the prototype (Figure2). In all experiments, the trades tested were the ones that worked with MTS materials. In the first experiment, the speed of generating a daily BOM card with LOD400 was 2 minutes 40 seconds on average using the prototype. A shortcoming was found in the prototype and we are in the process of addressing this issue: as of March 2018, the speed of generating a daily BOM card has improved to 43 seconds on average - the observation from the second experiment.
Acknowledgements
We thank CIFE members CCC and Oscar Properties for the access to their projects and supporting the research.
References
- Hopp, W. J., and Spearman, M. L. (1996). "Push and pull production systems." Factory physics - Foundations of manufacturing management, Richard D. Irwin, Chicago, IL.
- O’Brien, W. (1998). "Capacity costing approach for construction supply-chain management." Ph.D. thesis, Stanford Univ., Stanford, CA.
- Morkos, R. (2014). "Operational efficiency frontier - visualizing, manipulating, and navigating the construction scheduling state space with precedence, discrete, and disjunctive constraints." Ph.D. thesis, Stanford Univ., Stanford, CA.