It’s simple time management. If making the best BBQ requires a 16-hour smoke, and you want to eat dinner at 6pm, you need to start cooking at 3AM.
‍
Our team would wake up every Saturday morning to kick off a cooking session so that we would end up with the perfect BBQ. But this did not happen overnight.
‍
I remember one time where we messed up all the timing. The meat was cooked at the wrong temperature, we applied too much rub, and the meat didn’t get enough time to rest properly. Compared to a professional chef, it was an incredibly unimpressive brisket.

But being hungry college students, we didn’t really care.

Before cooking, we interviewed a number of food scientists and chefs that explained the science of BBQ to us. They taught us about reaction curves, how meat cooks, and best tips to look out for.
‍
To turn those concepts into tangible experiments, we laced our meat with at 10 different sensors including temperature, humidity, and gas volatility. At this moment in the class, I was in charge of project management, time schedules, and coordinating cooking shifts.

Tender, love, and care. And also a bunch of science. We ran 7 different experiments trying to isolate and codify the different variables involved in barbecue.

During one experiment, I built a small condensation hood that would collect the fumes from the smoker, and then automatically condense it into liquid utilizing the zero-degree Boston weather. The goal of the experiment was to scientifically analyze and compare how different types of wood and charcoal would affect the byproduct of smoke.

We used a quantitative method to find the correlation between a qualitative outcome - taste. After a course of our experiments, we found that there were still 2 key factors that affected taste beyond all other: temperature control and heat distribution.

Temperature is by-far the most important variable to control during a smoke. If you’re able to control the temperature, then you’re able to accurately predict the cooking time and rate for your meat.
‍
You may ask: why can’t you just set it like an oven? Our smoker is a little bit different because one of our design constraints was to use a charcoal-only wood smoker. For some reason, charcoal-only smokers also tend to make better bbq.

After lacing our meat and sensors with thermistors, I created a MATLAB script that would analyze and plot the overall data to create a model based on accuracy.

The second most important variable is heat distribution on your cooking surface. Because meats come in different shapes and cuts, you need to be able to accurately predict how your meat will cook.
‍
One time we had to lop off half of our brisket, just so the other side wouldn’t overcook. During our experimentation, we found out that there are lots of things that affect heat distribution such as where you , oxygen flow, and random sparks or flame-ups.

We had two main hypotheses that came out of our experimentation. First, we needed a way to seamlessly control the temperature. Second, we needed something that created a uniform cooking surface temperature. Our early concepts were super crude and simple. They started out as hand-sketches, and then were digitized as we debated the physics behind them.
‍
We learned that having an electronically control fan-based system would allow us to control the oxygen flow, which affected temperature. We also learned that a hyperbloid shape caused the heat to choke up at the middle of the smoker, but then would evenly disperse the heat evenly across a cooking surface.

I created the 2D models that we would later use in presentations.

After we settled on one design, we needed to find a quick, cost-effective way to test the concept. Why not do it electronically with COMSOL?

We created 3 different models to test: the Big Green Egg (industry leader), a another off-the-shelf smoker, and our concept. Those comparisons gave a good baseline to see whether or not our models were accurate, and an easy way to evaluate the efficiency. We accounted for as many environmental variables as we could - outdoor temperature, fire blaze consistency, wind flow, and smoker material composition.

During this phase, I left it to the chemical and bio engineers to work on simulating environments. I only helped with communications, timelines, presentations, and occasionally running to the cafeteria to get food.

Despite all this experimentation, we learned that there was another half that we were missing in the equation. What happens before you even put the meat in the smoker?

There were so many steps that were involved in meal prep. We had to go buy the meat with the right quality, create a rub and texture, thaw out the meat, trim the fat off, and so much more. If we didn’t prep the meat properly, then it didn’t matter how good our smoker was - you need amazing meal prep and a great smoke to get the perfect bbq.

This is where most of my work began. As the project manager and designer on the team, I led the research effort on the overall cooking experience to focus on the various aspects of meal prep.

In addition to cooking 16 hours every single Saturday, there were a number of us who would go into the lab and start meal prep the Friday afternoon before our big smoke.

Meal prep is an equally important part to the cooking process. If you don’t have the right ratio of seasoning, don’t let your meat thaw properly, or leave too much excess fat, your meat could turn out less than perfect.

I helped with a few parts in meal prep,  but as many of my friends know, I am a very bad cook. So I left it up to my classmates who were far more skilled.

Another big design challenge was to get a live-update of how we were cooking. We were cooking outside during the dead of winter. It was 15 degrees and we had to be outdoors monitoring our lab-based scientific equipment.

For the first few smokes, we had to run back and forth to check on our progress, which you can imagine quickly became a problem area to fix. So, we wanted to way to get remote access to the smoke where our professor could even get live updates while we were cooking.

I built a cloud-based prototype on a Raspberry Pi that could give us live feeds and visualizations of our smoke.

The application stack included a basic python script that interfaced with our thermistors, a Parse backend, and a web client that utilized d3.js for live visualizations.

So, here’s what we learned so far: we needed the perfect hardware to control temperature and heat distribution, we needed an easy way to know how to prep for a smoke, and we needed a convenient way to get live updates.

The best way to do that was to create a simple, mobile experience that assisted you through your barbecuing experience.

I created a number of user journey maps and data flow maps to figure out how we could capture a new BBQ experience, while focusing on integrating it with our corporate partners.

After a number of rough prototypes, ugly wireframes, and clunky designs, it was time to start dialing up the fidelity. We validated some of our assumptions based on the stages of Meal Prep and cooking but also left some new hypotheses in the high-fidelity prototype.
‍
Because Williams Sonoma was our Corporate Partner, it was important for us to brand the app to be aligned with their business purposes. We included a “Shop” section within the app that let users browse the Williams Sonoma catalog and suggest ingredients for the perfect BBQ.

I used Illustrator to create the design language, and then a great prototyping too called Proto.io to create gestures and interactions. The prototype included animations, swipe gestures, transitions, and more.

A few weeks leading up to the end of the semester, we were in heads down more. I primarily acted as a project manager and worked with subgroups within our team to divvy up all the work.
‍
We had one team hand build a full-scale and a half-scale functional prototype of the smoker. The electronics team and I worked on ways to create a cloud-enabled monitoring system. And another team worked on the logistics for our overall presentation; they had to get food-serving certified in order to comply with Massachusetts state law.

The culmination of our semester ended in a Demo Day, where we presented our research, findings, and final prototypes to an audience of professors, Williams Sonoma leadership, reporters, and fellow students. At the end of our presentation, we had a super-duper scientific “tasting” session that featured our smoked brisket with a side of slaw and some biscuits.

After a successful presentation, a group of students went to incubate the concept at the Harvard Innovation Lab. They eventually became a VC-backed startup, and are continuing to build the perfect BBQ dream today.

The final prototype was a barbecue smoker that included a refueling chute, automated fans, and a Raspberry Pi device - which allowed for perfect temperature control. After experimental validation, the hyperbloid shape ended up producing the perfect heat distribution.
‍
What does that all mean?

We were able to scientifically predict how meat cooks. Isn’t that a great superpower?

In addition to the perfect hardware, we had to have the perfect software that would complement it.

The App features five main section: Discover, Prepare, Cook, Share, and Shop. Those 5 steps were all part of the holistic cooking experience.

By designing an all-inclusive experience, we’re able to get amateur chefs - like a bunch of Harvard engineering students - one step closer to being professional pitmasters.