Improving planning and estimating process for scaffolding operation in heavy industrial construction projects

重工业建设项目通常涉及复杂的结构，大规模的场地，以及来自土木，机械和化学等不同学科的大量工人。这些学科可能需要类似或完全不同的脚手架系统，以便工人不仅可以进入他们的工作区域，而且还可以水平和垂直移动材料。在这些复杂的场地上，不同学科所要求的脚手架系统的可变性可能是增加项目成本的主要原因，因为一般建筑承包商需要在现场安装大量的脚手架系统。然而，无论是在学术界还是在实践中，脚手架工程的规划和评估都是一个相对较少受到关注的话题。在实践中，脚手架工程是根据专家经验、历史数据和公司的风险政策，以一种临时的方式进行估计和计划的，尽管脚手架作为临时工程占项目总成本的60%以上，或约占直接工作总工时的30-40%。为了应对这一挑战，本研究提出了以下任务：(i)确定影响脚手架活动工时的因素（例如，学科和工作分类）；（ii）计算搭架的间接工时与直接工时的比率（按学科计算）；（ii）将现场脚手架工程的工时划分为分区（例如模组、船只等）；（iii）使用机器学习技术开发预测模型，以提高脚手架估计过程的准确性和效率；（iv）开发后数据挖掘三维可视化，用于项目管理，如调度冲突，监控施工现场脚手架工作的严重程度和利用情况，与项目进度相对应。因此，这项研究将为建造业带来以下好处：(i)提高预估脚手架工时的准确性和效率，从而减少经济损失；（ii）提高脚手架规划的准确性；（iii）提高加拿大建筑承包商在国际市场上的竞争力。

Heavy industrial construction projects generally involve complex structures, large-scale sites, and large numbers of workers from different disciplines such as civil, mechanical, and chemical. These disciplines may require similar or completely different scaffolding systems in order for workers to not only access their working areas but also move material horizontally and vertically. The variability of scaffolding systems required by the different disciplines operating on these complex sites can be a primary cause of increased project costs since general construction contractors need to have large volumes of scaffolding systems on site. However, planning and estimating of scaffolding works is a topic that has garnered relatively little attention either in academia or in practice. In practice, scaffolding works are estimated and planned in an ad-hoc manner which depends on expert experience, historical data, and a company's risk policy, even though scaffolding as temporary works account for more than 60% of the total project cost or approximately 30-40% of the total man-hours of direct work. To address this challenge, this research proposes following tasks: (i) identification of factors (e.g., disciplines and work classifications) affecting man-hours of scaffold activities; (ii) computation of ratio between scaffolding indirect man-hours and direct man-hours (per discipline); (ii) mapping man-hours of scaffolding works into sub-areas (e.g., modules, vessels, etc.) on-site; (iii) development of a predictive model using machine learning techniques to improve accuracy and efficiency in the scaffolding estimation process; and (iv) development of post-data mining 3D visualization for the purposes of the project management such as scheduling conflicts and monitoring severity and utilization of the scaffolding work on the construction site corresponding to project progress. As a result, this research will provide to the construction industry the following benefits: (i) improving the accuracy and efficiency of estimating scaffolding man-hours which can reduce the economic loss; (ii) improving the accuracy of scaffolding planning; and (iii) enhancing competitiveness in the international markets for construction contractors in Canada.