Bentley STAAD YII2023

Automated design of civil and architectural structures with STAAD API  Featured

*Design of civil or architectural structures with STAAD API. Credits: Hyundai Engineering*

Hyundai Engineering, winner of Bentley Systems' 2023 Going Digital Awards in Structural Engineering, used STAAD and Artificial Intelligence (A.I.) to automate shelter and pipe rack designs to respond more effectively to frequent design change to the construction planning of chemical and electrical power plants. The result was an error-prone resistant, integrated solution of 3D modeling with intelligent digital workflows that provided accurate design information and accelerated front-end engineering design by at least 30%. A database of 1,680 scenarios was created, generating 27 million prediction models, that can automatically convert design information into a 3D model for real-time synchronization of maintenance prediction and improvement work of buildings in similar projects in the future.

Designing process with STAAD API

In plant engineering, the workflow necessitates the cohesive integration of diverse disciplines for the design, construction, and maintenance of industrial plants. The collaboration among piping, electrical, instrumentation, and mechanical disciplines is crucial for the seamless execution of projects. The information derived from these disciplines substantially contributes to the structural design aspects of civil or architectural infrastructure projects. Furthermore, the incorporation of automation, prediction, and optimization processes into plant engineering practices enhances operational efficiency, safety, and overall project success, aligning with the evolving demands of contemporary industries.

A piperack is a raised structure used in industrial plants and facilities to support pipes, conduits, and cable trays. In fact, piperacks play a critical role in industrial plant design by providing an organized and elevated pathway for utility lines. From a structural point of view and to succeed accurate analysis outcomes and an efficient and cost-effective design one should start from the basic 3D model information in collaboration with automated piperack modeling techniques. The latter is achieved using commercial Finite Element software, e.g. the STAAD that offers notable advantages such as automated model generation, corrective functionalities, and the automated application of piping and friction loads.

Learn more about STAAD

In addition, the design procedure encompasses the structural components, including the column, main girder, beam, and sub-beam, while adhering to various constraints such as girder width being less than column width and beam depth being less than girder depth (Fig. 1). Another significant point of the design process involves the structural members connection.

Figure 1: Schematic representation of some restrictions that are considered in the design procedure. Slide from Hyundai Engineering presentation at the The Year in Infrastructure and Going Digital Awards (YII 2023)

For the A.I.-based structural design, the process initiates with automated shelter modeling, progresses to database construction, and culminates in A.I.-based recommendations. Databases can be tailored based on structural dimensions, leading to the generation of thousands of results. These results are amalgamated with diverse regressors, including Decision Tree Regressor, XGB Regressor, Gradient Boosting Regressor, and Random Forest Regressor. In the final model, the Support Vector Regressor is employed for the ultimate prediction, providing recommendations for rolled beams, trusses, and/or column distances/bays.

The utilization of the STAAD API presents several advantages, notably a substantial reduction in analysis and design time (30-60%) and a corresponding decrease in outsourcing costs. The estimated outsourcing cost reduction of $1,931,811 has been calculated for 50 bidding projects, 25 FEED projects, and 20 execution projects over a 5-year period, aligning with guidance from Bentley applied to the overall development process. It is imperative to note that three key stages of A.I. are of primary interest in this procedure: the descriptive, predictive, and optimal A.I. stages. In essence, the automation of data recognition precedes structural quantity prediction, culminating in optimal recommendations for the structural system in the third stage.