You've heard about agile software development, right? The web 2.0 manifesto usurped the upper echelons of high technology and overthrew the waterfall crowned kings of the 1990's internet.
I'd like to share an example of what we can call agile infrastructure development. Here I don't mean servers or data warehouses or any backend bits or bytes. I mean moving actual atoms. Specifically essential infrastructure and more specifically a stormwater recharge facility in my hometown, a humble ditch with native plants with the fancy technical name of a bioswale.
Water processes will be familiar to any public works project. Whether a street resurfacing or a stormwater project, typically those bits of physical infrastructure involve incredibly thorough identification of requirements, design and then development. It’s a linear, sequential, process with each step proceeding the next.[1]
Agile software development processes by contrast involve a continuous feedback loop and emphasize “responding to change over following a plan.” This approach is particularly apropos for living infrastructure that functions as part of a dynamic, constantly evolving living ecosystem like the example stormwater bioswale. The diagram below provides the initial design of the project.
Previously the water flowed directly to the channelized concrete flood control basin. The diagram above shows the initial project design, which was simplified for the first construction phase.
This project had a dirt simple motivation. Let the water slow down and recharge into the groundwater rather than flow into the ocean. Of course, much in California is not as simple as it may seem on its face. In that small space, the water crosses three jurisdictional boundaries — the Arroyo and Foothills Conservancy (AFC), LA County and the CV Water District.
Waterfall processes proved impractical for the process. The initial project estimated cost from the CVWD engineering department was close to a million dollars, mostly due to anticipated studies and design. Some also questioned the project performance, asking how much water the project would actually recharge. Such questions are on their face reasonable but involve unreasonable costs to answer. A formal recharge study could easily cost $50-100k plus.
The actual volumes of water range vary based on precipitation and the water year though on an order of magnitude basis are in the single digits of acre feet. The value of that water to CVWD thus ranges up to 10 AF recharged * $1450 / AF (marginal cost of imported water) or up to $14,500 per annum. In many years that might be much less.
Acquiring more precise estimates of that water recharge volume doesn’t actually change the decision calculus. Instead it just adds to the project soft costs and also critically would slow down the speed of the project.
By contrast, actual construction involved in-house Crescenta Valley Water District resources — one day with four field crew and District equipment. Moreover, the construction provided a great field training day as part of the ongoing professional development of the CVWD team to improve their construction and project management skills.
The marginal cost to the district was thus zero dollars. In addition, in the agile spirit emphasizing “simplicity—the art of maximizing the amount of work not done”— the project design was downscaled to just a single initial trench for this first water year, greatly simplifying the project approval process by only engaging in construction on AFC property.
Here your author participates in the CVWD staff training exercise as part of the bioswale construction.
The fine grading of the bioswale and planting is being done on an ongoing basis by AFC volunteers, including yours truly. Talk about creating something out of nothing! Further the project will include a camera on the bioswale to provide an ongoing coarse grained estimate of the water flow and additional recharge. Those data will provide invaluable information to course correct as needed for the next water year and rainy season.
This iterative approach is not unique in physical infrastructure development. Design-build methodologies can involve modular design and iterative construction processes — retrofitting one wing of a hospital or one piece of a treatment plant before design is complete on another portion of the project. That enables faster delivery and also lessons learned to be shared from one phase to the next.[2]
Today physical infrastructure processes can implicitly assume a waterfall based approach for everything. Building physical infrastructure, particularly with the complexities today around permitting, can involve exquisitely intricate plans and extremely long documentation before anything actually gets built.
With climate change, plans with long lead times can become obsolete upon arrival as weather patterns and other demands upon physical infrastructure dynamically evolve. Agile approaches provide a pathway to implement, adjust and course correct — critical components of effectively adapting to situations involving deep uncertainty, such as the escalating climate crisis.
Lastly as the worlds of bits and atoms increasingly interweave and intersect, agile development practices provide more than a metaphor for a path to build physical infrastructure. Such approaches hint at how we might pioneer adaptive, integrated systems that enable greater symbiosis between the organic and the synthetic.
The bioswale in action!
Notes
[1] Here is a typical waterfall software development example project courtesy of Chat GPT:
Building a Payroll Management System for a Company
Requirements Gathering: Define functionalities like employee data storage, salary calculation, tax deductions, and report generation.
Design: Develop detailed flowcharts, database schemas, and system architecture diagrams.
Implementation: Code the backend for calculations, front-end for user interaction, and integrate them.
Testing: Perform unit tests on each module and system-wide integration testing.
Deployment: Install the system on company servers or provide it as a software package.
Maintenance: Address user feedback, resolve bugs, and add features if needed.
This approach works well for projects with stable and well-understood requirements but can struggle in dynamic or fast-changing environments.
These steps parallel the step by step, sequential approach typical for physical infrastructure. Those can include feasibility studies, requirements gathering, engineering design, permitting approvals, construction, inspection and testing and then finally maintenance.
[2] There are also famous historical examples. For example, building the Colorado River Aqueduct involved pioneering modular construction methodologies where different sections of the canal were built concurrently. Lessons from innovative uses of precast concrete and novel workforce organization could be shared across the project.
That project also involved advanced surveying through aerial photography and surveying, technologies that were still in their early days. That provided an overall project master plan that then could be modified and adapted within modular construction sections. Within a software development metaphor, the project could perhaps be considered akin to a modified agile framework whereby implementation sprints operate within a longer term strategic roadmap.
