In January of 2021, our team of Graduate and Undergraduate engineers at UC Berkeley began a partnership with the Aloha Spark Innovation Group at Joint Base Pearl Harbor-Hickam.

The current process for preparing valuable air assets for impending hurricanes was inefficient, expensive, and disorganized. By improving this process, we aim to increase the number of mission-capable assets available immediately after a disaster strikes.

Having more assets on base and fully operational after a hurricane allows for faster threat response, expanded resilience to assist with post-disaster Search and Rescue and better support of America's warfighting capability.

To better understand our problem space, we dove right into the user research. We initially interviewed 5 weather ops, 8 maintainers, and 12 people from upper level command, all of which are highlighted in the following outline.

Based on these user group’s roles, responsibilities, and knowledge areas, we narrowed down our focus on the maintainers as it was clear that they ultimately make the decision in hangaring air assets. We gained valuable insight, including one particular interview where we witnessed maintainers fighting with each other on whether or not a C-17 was more important to hangar first than an F-22. This shed light on the root of the problem: decision making. Unfortunately, the air force does not have a clear and detailed plan on how to hangar aircraft before a hurricane. Their current solution is using a PowerPoint that is updated only twice a day by one user (multiple users cannot edit at the same time). As a result, it does not account for changing weather, aircraft maintenance or other issues happening on the fly. The outline below summarizes our key insights and display this problem as a cyclic problem.

In viewing the problem as a system, we could take a step back and view the multiple components feeding into this ongoing issue. The need to make a decision quickly is dependent on relaying information through the PowerPoint slide deck, which is not getting updated as circumstances change in real time. Understanding this, we came to our final problem frame:

Based on our final problem frame, we came up with five prototypes and chose the best three of them to show to our key stakeholders. Then, through a series of interviews, we got feedback on our prototypes, as well as the suggestions for our next step.

1. Hangar Extension

It is to build additional shelters outside the hangar to protect more air assets. The main problems include how to choose the materials and design standards to make the structure strong enough. Also, the cost of implementing it would be very high.

2. Portable Turntable

This idea is to place the aircraft on the turntable to increase the flexibility of the direction and position of the aircraft in the hangar. However, it would be difficult to make the turntable fit each individual aircraft of different size unless we design separate turntables for each aircraft, which would be expensive. Meanwhile, It will be a problem with storing it when not in use.

3.Tetris Grid

It is a planning system facilitated by cameras in hangar to help decide on the optimal location of air assets. According to the interviews, this solution is the best one. However, the problem is that in reality, it is hard to paint anything in the real hangar, and it is difficult to rotate and twist the aircraft so that it is exactly on the specific grid points.

Therefore, the suggestion we've got is to turn this idea into software that can freely position aircraft and other air assets based on the virtual grids on a computer or tablet. Also, another software's advantage is that it can be updated in real time on multiple clients to ensure the tasks can be delivered quickly.

The suggestion for our next step is making 3D models of hangars and aircrafts to help implement the software. On the one hand, through the 3D model, we can understand the constraints and the rules for a software algorithm. On the other hand, our stakeholders also hope to get such a series of models to assist them in making relevant decisions.

We scheduled demos with the software providers of existing companies on the market: NAVFAC spiders and STAX, both of which help planners arrange sea and air assets. However, neither were able to fully address automating the hangaring process.

We were introduced to the US Navy’s agile software development group, TRON, who took the initiative to lead the technical development of an original program to address our automation need.

So, our focus shifted to the key unknown to bring that software to life: what rules and guidelines would an automated algorithm need to follow? Put another way, could we develop heuristics about how maintainers and planners make decisions today, that we could then automate? To answer this question, we developed a set of experiments to help identify these rules, using a 1:160 scale mockup of the hangar space and air assets.

The first step was to understand more closely how the maintainers go through the process of placing the aircraft by creating a 3D scaled representation of the aircraft and hangars. Then we ran through two mock evacuation scenarios with the maintainers to derive algorithm rules and constraints. We created a 3D representation of the hangars and aircraft using cad from our sponsors.

During the exercise, each maintainer gave helpful feedback that we otherwise would not have been able to obtain without hardware, for example: minimum wing-to-wing clearance and towing operations. These findings will shape the needs of the software tool employed by the military.

Our small-scale hardware is available to the maintainers at any time for mock evacuation exercises. The maintainers also have the option of pursing either Stax or SPIDERS 3D. The value in this solution is applicable for all organizations sheltering air assets, across both the DoD and the commercial sector. Our value sources include more space available to the maintainers for hangaring, mobilizing assets on a short weather notice, and less maintenance downtime.

We’d like to thank our sponsors, Major Votipka and Lt. Lam, as well as our project supervisor, Professor Vivek Rao.