C+S September 2023 Vol. 9 Issue 9 (web)

Job Plans standardize the steps involved in completing a maintenance task or repair. They ensure a uniform process that promotes efficiency, improves communication, comprehension, and compliance throughout an organization. Safety Plans, on the other hand, lay down the necessary measures to minimize risk to personnel involved in maintenance work and prevent any inadvertent damage to the physical assets. Such a framework, while undeniably beneficial, also requires a significant initial investment in creating standards and ensuring their quality. One such project undertaken by a public transportation agency serves as a case in point. The objective was to create tasks for Job Plans for a set of assets and assign safety considerations where necessary, culminating in a unified resource that would efficiently manage work and risk across an asset portfolio.

Plan was assigned safety precautions. Another prompt was employed for this purpose to determine relevant safety considerations for each Job Plan Task, accounting for areas where the identified OSHA hazards were likely to be encountered. The system utilized a sequence of meticulously engineered prompts to generate JPSPs for asset classes. It identified potential hazards, generated general Job Plan Tasks that covered all subcategories within a given asset class, and assigned safety precautions to each Job Plan Task, all with remarkable efficiency and accuracy. The project then moved into a second phase, where ensuring the accuracy and usefulness of the AI-generated JPSPs became the focus. Through a careful review process involving two separate engineering groups, issues were identified, and improvements were made, leading to high-quality results. This project marks a significant shift in the dynamics of the workforce, with AI taking

Credit: ​Nicholas Russo, EAM GIS Analyst, Symetri

over tasks traditionally performed by developers, and SMEs assuming a central role in fine-tuning the outputs. As AI systems continue to advance, SMEs with their deep domain knowledge and expertise are becoming pivotal in ensuring the accuracy, relevance, and overall quality of AI-generated outputs. In this new era, a synergistic relationship is emerging between AI, SMEs, and developers. The workload is intelligently distributed, allowing developers to focus on strategic initiatives, while SMEs contribute their specialized knowledge to refine the AI outputs. This evolution heralds a future where the collaborative efforts of AI and SMEs become paramount, paving the way for unparalleled productivity and innovation. Dr. Jeff Chen, Ph.D., LEED AP is Director of Digital Transformation, Symetri. Dr. Chen leads digital technology integration services for all aspects of client busi - nesses to drive efficiency, reduce environmental impacts, and increase sustain - ability. He has directed Enterprise Asset Management, (EAM), Enterprise Project Management (EPM), Building Information Modeling (BIM) and data asset validation projects for some of the nation’s largest and most respected organizations and public agencies, including a bi-state transportation agency, Columbia University, and the University of Pennsylvania Health System. Nicholas Russo is an Enterprise Asset Management GIS Analyst with the Symetri Asset Management, Strategic Advisory Service Team. He specializes in AI integration, data management and governance, digital twin solutions, GIS cloud integration and application development. Nick has led the development of AI implementations for a bi-state transportation agency, and has designed and implemented scalable GIS-BIM information systems.

The ambitious project was to develop generic Job Plan and Safety Plan (JPSP) templates for nearly 500 asset classes. These templates could be tailored to unique requirements by field experts. However, the project ran into a significant challenge during the proof of concept stage, involving the need for multiple subject matter experts (SMEs) to review and refine the templates due to the diversity of the asset classes. OpenAI GPT came to the rescue, streamlining the process by generating draft JPSP templates based on asset classification names. This breakthrough demonstrated the power of AI when it is integrated with human expertise, transforming the traditional ways of working and propelling asset management into new frontiers. A precisely engineered system of prompts was devised to generate a JPSP for asset classes, ensuring their seamless alignment with a predefined structure. The first prompt was designed to generate a comprehensive list of potential hazards that workers may encounter while servicing the asset, using authoritative standards established by OSHA as a reference. Subsequently, another prompt was created to generate a general set of Job Plan Tasks whose scope encompassed all applicable subcategories within the given asset class. For instance, the procedural steps designed for an asset class such as Pump were carefully crafted to seamlessly extend to Metering Pumps, Booster Pumps, and other related classifications, ensuring maximum efficiency and consistency in output. Following the definition of OSHA hazards and generation of Job Plan Tasks, an iterative process was initiated, wherein each step of the Job

21

September 2023 csengineermag.com

Made with FlippingBook Annual report