Several factors influenced my decision to intern at GE Gas Power in the summer of 2020. I was pursuing a minor in energy engineering and considering a career in the energy industry, and after interning at a tiny automation company the previous summer, I wanted to see what it was like to work in a much larger organization. GE certainly checked off both boxes, and even though I didn’t have any personal ties to the company, I applied, interviewed, and committed early in the year. I planned to borrow my dad’s car, rent an apartment with some fellow interns, and move to Schenectady, the site of GE’s steam power organization (which is part of the larger GE Gas Power). In fact, I had already printed out a lease to review and sign when I received word that GE would be honoring their internships remotely due to concerns over COVID-19.

At that point, the pandemic was pretty new. Although I’d already been told not to come back to school after spring break, it still stung to realize I’d be spending the next ten weeks glued to a computer screen for eight hours a day at a desk in my brother’s bedroom. One reason I’d chosen to become a mechanical engineer was that I wanted to build things and work with my hands, not just do computer work; in a matter of weeks, this plan and many others were either completely foiled or in jeopardy. Part of me doubted a virtual internship could be a good experience at all.

Immediately Impressed

Work started on June first with a minor tech crisis: dozens of interns from across the country needed help connecting to even the most basic of GE’s systems. It was the first time mass remote onboarding had happened for any of us, so issues were almost inevitable. I was impressed, however, that IT help was prompt and effective, contrasting almost every other tech support experience I’ve ever had. Maybe GE could make a virtual experience work after all.

In the afternoon, my assignment leader introduced me to our team, and engineers from across the organization reached out to welcome me and offer whatever help they could. Throughout the following weeks, the people I spoke with always took the time to explain not only what I needed to know to get my work done but also relevant background, history, and more, regardless of how little time they had in their schedules. This became a consistent theme throughout the summer: engineers, HR, and even corporate leaders took time to build up our knowledge and skills even when there wasn’t a clear benefit to themselves or the company. Interviewing engineers from across GE’s steam turbine organization, learning from recent hires in GE’s rotational program, and attending resume, interview, and networking sessions were all significant and valuable parts of my intern experience.

Let’s Get Technical

Of course, I spent the bulk of my time on technical work, and I wouldn’t want it any other way. My first major project was to reformat and draft updates for Steam Turbine Design Practice 021, a document with key information about GE’s standards for steam turbine instrumentation. Fortunately, I was assigned to work with an excellent engineer who knew a great deal about the topic and could refer me to all right people for information updates. After learning the steam turbine basics from him, I spent several weeks conducting meetings with subsystem experts and codifying their input to make the Design Practice as useful and concise as possible.

At the end of week four, it was time for a new challenge. Even before I’d started work, my assignment leader decided to break up my ten-week internship among several teams in order to expose me to more than one of the many aspects of engineering a steam turbine. A recent graduate of GE’s rotational program welcomed me to the buckets team (a steam or gas turbine’s rotating airfoils are often called buckets), where my impact for the summer would end up being greatest. After acquainting me with the basics of what his team did, he presented me with the problem I was to help solve: buckets take too long to design.

Almost all steam turbines are unique, with their specifications tailored even to the outside temperatures at their installation sites. Modern steam turbines contain dozens of rings of buckets on their rotors, and each ring’s bucket design is unique. When I arrived, many parts of the design workflow had already been automated using Excel spreadsheets and macros written in NX, but there were a few parameters that still required several human hours of work to fully specify. Several hours spent on each of several dozen buckets adds up quickly; a tool that could automatically determine these parameters was predicted to save two or more weeks of engineering time per requisition steam turbine. My job was to lay some of the groundwork for this tool, proving that it could work by finding models capable of empirically predicting the desired parameters in seconds.

To get me started, my assignment leader introduced me to some data analysis tools and showed me where to find the parameters of interest for several hundred already-designed buckets. He and another engineer had already consolidated lots of data for me to use, and he suggested a few independent parameters that might be especially influential in the predictive model. Beyond that, I had the liberty to do as I saw fit. Looking back, I believe that freedom enabled me to do my best work. I quickly did as I was told, implementing the suggested model with data he’d gathered. Then I collected much more information from the database and identified influential factors my assignment leader hadn’t thought of. Using knowledge I’d gained in my statics and controls classes, I also created a second predictive model using log-log transformed data, which I believed would better represent the twist mechanics in the buckets. Of course, I reported on what I was doing and immediately received approval to keep exploring these new ideas.

Eager to demonstrate my log-log model’s superiority, I spent several days testing it and researching how to make it better. Committed to objectivity, however, I made sure to spend an equal amount of effort working on my assignment leader’s model, taking insights I’d gained while working on one model and applying them to the other as well. Throughout the process, I recorded performance metrics to quantify the effects of the changes I made and to provide an objective basis for inter-model comparison. It soon became obvious that although my preferred model met the standards for success, it wasn’t actually outperforming the other. Once I knew that, it didn’t make sense to keep working on both, so I set my preferences aside and committed to the original model.

Overall, the most time-consuming part of my work was figuring out which independent variables could best predict the parameters of interest. To make the process as efficient as possible, I brainstormed a few algorithms for solving the problem, tested their efficacity, and built a standard workflow around implementing the best one. This made applying it across a variety of bucket types and materials much quicker and easier. By the end of my time at GE, I had more than a dozen such models working accurately enough to be implemented, but I knew that alone wasn’t enough. To make sure my work could be interpreted and used after I left, I added thorough summaries with performance comparisons to the front of all my documents and took advantage of my standard work approach to organize my input and output data in a logical scheme. After reviewing these in detail with my assignment leader and formally presenting the results to his team, I knew I’d added real value in my short time at GE.

Reflecting

Even when it meant spending days crunching numbers at a desk in the basement, I enjoyed analyzing the bucket data and took pride in the knowledge that my work would actually help build the foundation for a major time-saving tool. It’s true that very little went as planned that summer; even during the internship, my assignment leader and I decided to shift focus from experiential breadth to depth. Come august, however, I realized thankfully that my initial apprehensions about virtual work were unwarranted. The summer, though more constrained than normal, had many silver linings, and I am a better professional, collaborator, and engineer for having gone through it.

Pitch-Out

All EID interns at GE give a final “pitch-out” presentation near the end date of their assignment. This helps clarify to stakeholders in their work what has been done and provides an opportunity for interns to reflect and thank the many people that helped them during their assignment. I cannot share the actual slide deck I used for my pitch out because it contained proprietary information, but I’ve re-created a similar deck without any of the sensitive content. View it here!

External Links

Read my LinkedIn post about the experience here

Read my final performance review from GE here