Martin Fischer, Eduardo Miranda, Ram Rajagopal, Filippo Ranalli
The scope of this project is to design a robust, scalable and generalized algorithm that can act as a design aid by simultaneously optimizing the member sizing, detailing and topology of conventional steel building structures for cost, while guaranteeing all strength/sizing/constructability requirements per code.
The research approach will focus on the further development of computationally-efficient optimization methods, based on an existing topology and sizing optimization framework. Validation will be performed using industry case studies on conventional moment-frame building structures.
In modern structural design practice, the design space is explored and iterated-through manually to find solutions that are compliant with strength and drift requirements, per AISC. However, without the aid of computational design tools, manual exploration can be insufficient, and the design sub-optimal and time-consuming. Furthermore, when searching for the optimal design, conventional optimization algorithms directly associate cost to weight, leaving out fabrication and erection. Finally, ensuring the design meets constructability requirements can be challenging if done manually.
We will run the algorithm on retrospective or current full-scale projects from the industry, where a previously-completed design is examined with the improved MDO
framework, and the cost and buildability performance are compared on a fair basis. The current test case is a 3-story steel moment-frame building in San Diego, featuring a high roof structure, composite slabs and wind/earthquake loading.
1. Develop a detailing engine that will instantiate individual connection details based on the local node geometry and load demands.
2. Develop constructability checks both on the local node/member scale, and on the global sub-assembly scale. These checks will beguided by a designer-specified rule-set.
3. Develop a granular connection cost model, based on fabricator/erector guidelines or collected data-sets on which to perform statistical learning.
4. Enhance existing sizing algorithm to effectively minimize cost for strength and stiffness compliance, rather than solely weight.
5. Allow for connection optimization in post-processing, to ensure the appropriate distribution of moment fixity throughout the structure.
6. Automatically generate the topology design space (ground structure) from the BIM or analysis models, using the designer-input parameters.
7. Enhance the topology algorithm through the introduction of a guided probabilistic exploration schedule, in order to search for better-than-greedy local minima.
8. Develop a composite floor optimization algorithm that will find the cheapest feasible beam/girder design considering camber and full/partial composite action.
As of this proposal, all the logic for first phase described above (composite floor and lateral system sizing and costing) has been developed and tested on smaller benchmark structures. A newly-erected 3-story building has been chosen as the first industry retrospective case study. The comprehensive analytical model of the case study has been assembled from the structural drawings and the steel coordination model. Furthermore, an algorithm to evaluate the material and fabrication cost of the members and connections of the existing design has been developed, and has shown good accuracy. So far, about 13% savings in material, erection and fabrication have shown to be possible within the gravity system, which makes up the vast majority of the elements in the structure. The runtime is currently under one hour. The expected progress by the end of Summer 2019 is to have fully debugged, generalized, refined and run the automated member design logic on the case study above, to have shown major cost savings in the optimized solution and to have a preliminary draft to be published in a structural engineering or construction journal.
Future progress (after summer 2019) will be focused on integrating the frame/composite member optimization above with a new detailing engine and cost model, in order to achieve higher savings in fabrication and erection. The optimization may then be directed at optimizing the foundation type and layout in order to minimize material and excavation while guaranteeing local strength and global settlement requirements.
Original Research Proposal
Seed Proposal PDF