In going from careful use of floppy disks to now regularly filling terabyte SSDs to capacity with a single project, BDI is no stranger to the rapid growth of data sizes in civil engineering applications over the last few decades. The company has advanced right alongside this growth during its history, rising to meet the challenge posed by our clients to offer higher resolution results, greater varieties in service offerings that mean greater varieties in data, and turnaround times that allow for more timely and cost-effective decision making. With the rise and hype of AI platforms, ML models, and various other automated solutions in recent years, BDI is also no stranger to the questions of how and where these technologies fit into our processes and approaches, having the benefit of this long history to offer us answers.
We believe that to Do The Right Thing when it comes to the projects we undertake and clients we support, the expertise, experience, and due diligence of an engineer or analyst cannot be replaced with AI solutions. Work in our industry starts with people, and at BDI, it ends with professionals who recognize they are accountable for the results they deliver, not the tools they use to develop them. In this same spirit of accountability, security and sensitivity are paramount to maintaining the trust of our clients and safety of the public, viewing ourselves as stewards of the data we are asked to collect, analyze, and manage. This stewardship means that we provide regular data security training to staff at every level, actively engage best practices for data storage and transmission, and take great care to ensure that client data only interacts with BDI-developed software or software developed by comparably accountable vendors.
Bearing these values in mind, doing the right thing does also mean ensuring our clients benefit from the best-in-class technology available, so we also believe that automated and semi-automated solutions represent critical tools for accelerating and standardizing the generation of data-driven results in processes directed and reviewed by accountable professionals.
Acceleration & Standardization
Imagine you have to report on the post-tensioned (PT) duct condition of 3 tendons in 8 girders for 10 spans with 2 faces, resulting in a total of 160 individual CAD sheets like the one presented above. Each sheet itself contains 3 different NDE condition overlays (Impact Echo, Ultrasonic Tomography – MIRA, IE+MIRA) presenting roughly 1,000 classification points each and presenting 110 unique values and measurements in each title block – that’s a total of 160,000 points and 17,600 values to present.
Now, imagine you’ve been asked to shift and invert the origin of the coordinate system for your CAD, which means updating the position of those 160,000 points.
Now, imagine you’ve gathered additional validation data and can further refine your classification model, resulting in those 160,000 points and 17,600 values needing to be updated.
Now, imagine those requests coming at 4:13pm on a Thursday with the hope of getting updated mappings out by close of business Friday at the latest.
Finally, imagine doing these updates manually – carefully doing column flips and subtracts in a spreadsheet for position changes while making sure that 17,600 values are hand-entered exactly as the statistics you re-calculated are presenting.
That’s the sort of challenge BDI regularly experiences in the course of its data-driven work. Automation at BDI isn’t necessarily about saving time or driving down costs so much as making more of the time we have on a given project. Developing and deploying automated workflows and tools overseen by accountable analysts through regular quality control test points, accelerating traditionally slow and repetitive tasks and standardizing those tasks that tend to be prone to transcription or mis-click errors. All in service of ensuring that human engineering and analytical intelligence can engage the more nuanced and complex challenges that projects present, challenges that would be otherwise inappropriate or ineffective for AI or other automated tools to address.