Blog Post #4: How I Prepared For The 2023 Season

The 2023 season is finally over, and as fruitful as it was, the focus has already shifted to next year. Therefore, now’s an appropriate time to talk about how I prepared for the 2023 season, in hopes that this information aids in your preparation for next season.

When I sat down in January of 2023 to plan out my tasks it quickly became evident that many of them were dependent on one another. To sort out where to begin, keep track of deadlines, and record who I needed information from, I created a “2023 Masterplan” Excel sheet. I always find that laying out all my tasks in written form helps me visualize the path to completing them most efficiently. From there, I sat down and began to chip away at my to-do list, and here’s where I started.

Documentation and Organization

Race engineers like to harvest as much data as possible because it helps to develop a complete understanding of the vehicle’s performance for a desired instance in time. Therefore, the more sources you can acquire data from, the more likely you’ll have the tools needed to diagnose a certain issue, whether it’s outright performance or balance-related. When speaking about data here, it can refer to anything from the logged data from sensors on the car measuring certain parameters, to a session’s metadata such as ambient and track temperatures, as well as setup/set down numbers. While logged data from the race car is reliably recorded and assigned to the session it was logged in, desired auxiliary information from that same session that is recorded by a human may not be so consistently written and stored. Therefore, one of the first challenges I faced in 2023, was to develop a system where the information needed for engineering work could be reliably accessed and be used effectively.

Keeping detailed run sheets, setup/set down sheets, and tech numbers for one car can already be an organizational challenge of its own. In Ferrari Challenge, where one team often runs multiple cars, the reality is that one engineer cannot be present at all times to keep notes of all the vital information. This is where using the mechanics to help gather the necessary data you need to make decisions becomes critical. First created in 2020, then refined in 2023, I developed physical run sheets that the lead mechanic of each car could use to document the desired data from each session. The sheets are straightforward, and include a space to record lap times, write the tyre set number, session name, starting cold pressures, target tyre pressures, tyre surface temperatures, driver comments, setup changes, brake wear, and fuel pumped out at the end of the session. The ability to have these recorded without myself present in pit lane allows me to look at the sheet at the end of a session, and at a glance, understand what happened. The mechanic run sheets are then stored in a binder so that every sheet from that weekend is accessible in one spot. Although I like to be in pit lane as much as possible, I’m often present at technical inspection, speaking with the tyres and fuel team to share information and plan ahead, otherwise I’m working with driver coaches and mechanics on data analysis and setup respectively. Things get hectic during a race weekend, and as much as you would love a few clones of yourself to be present in every situation, the next best thing is to have those around you take on some of the work by keeping up documentation when you can’t do so. On top of the run sheets, the mechanics are also responsible for filling out their parts usage sheet which helps to accurately bill customers, as well as a tech inspection sheet to write down tech numbers when I’m not present. Amongst all improvements, I believe my greatest achievement this year was being able to coordinate the race team to have them provide me with the information I needed, reliably, quickly, and without taking up much extra time from the mechanics.

 The Madness Behind The Math – Spreadsheets

With race weekends being busy enough, the last thing a race engineer wants is self-doubt about tyre pressure and fuel calculations. Having robust and reliable math with solid methodologies to back it all up gives you confidence in your work, and when you trust your numbers, you perform better as an engineer. With that said, you should always have someone double-check your work, because everyone is prone to making mistakes no matter how much experience they have. For items such as qualifying and race fuel loads, I consult the data engineer to ensure my numbers make sense. I created my race engineering spreadsheet for R. Ferri Motorsport back in 2020, but I created a completely new one for 2023 to include all the new things I’ve learned over the past few years. The best time to create one is in the off-season. Important items to include in these sheets are the max/min values for technical inspection, a space to make notes on various investigational findings from setup changes, tyre set tracking, tyre pressure calculations, fuel load calculations, and weather condition tracking. Having a single file to work out of simplifies the engineering workflow process, maximizes organization, and reduces the time spent searching for information over multiple files. Marking your sheets with a version number is also important as it allows you to make iterative changes to your spreadsheets as you see fit, without the worry that you may be using an outdated copy. I constantly find myself tweaking calculations, adding sheets in a workbook, or removing outdated information that isn’t useful anymore, so storing old Excel files helps build a library of progress.

Lap Time Simulation – Point Mass Models

Over the past six years, if there is one thing I learned about simulation and modelling it’s that, “All simulations are wrong, but some simulations are useful”. By nature, engineers take pride in the complexity of their work, and this also applies to the simulation industry. Whether it’s a CFD, FEA, or lap time simulation model, there’s always something exhilarating about creating a model with very high fidelity. The reality is, that high-fidelity models require lots of information that is usually not available to racing teams running in categories below F1, the Hypercar class, or a factory GTE team. Transient or quasi-steady state lap time simulation models have their place in motorsport, but when such fundamentals as the tyre model are guesses, it can be easy to mislead yourself. Another common saying in the simulation industry is “Garbage in, garbage out”, and the less garbage you feed into your model, the more useful the results will be. Your goal as an engineer is to put together a model which is the most complete picture of your vehicle, without making many additional guesses that can skew the model’s results. Most of the vehicle information needed for a basic point mass model can be found in the race car’s user manual, but oftentimes, some parameters have to be estimated using logged data. In Ferrari Challenge, rough lift and drag coefficients for the 488 Challenge Evo are provided, but I’ve gone and estimated them from logged data. You’ll find that the information provided in the manufacturer manuals is decently accurate, but best practices suggest to go and verify those figures yourself. A good example of this is wheel-to-damper motion ratios. Some manuals include a single value, while some include the motion ratio in its non-linear form, as a function of wheel position. Furthermore, figures in the manual are typically theoretical and are not representative of how certain devices perform when they’re under load. Using a kinematics and compliance rig to determine your actual installation ratios is always desirable, but none of this is relevant to developing a point mass model. This is just to make a point that it's healthy to be sceptical of what data is presented to you in the manual. Regardless, performing your tests and gathering your data is a good sanity check to ensure the values you’re using in your simulation represent reality.

With the vehicle data sorted, track generation is the next step. The point mass software I used was OptimumLap, and this already had some of the tracks I needed. With that said, I found much better correlation when I created the tracks myself using logged data. While easy to use, one of the downsides of using OptimumLap was having to manually enter circuit altitude points. To properly capture the effects of track elevation at places such as COTA, Road Atlanta, Sonoma, and Road America, the software needed many points, and this was probably the most time-consuming activity. The ease of working with software is critical, but software selection and what to look for in a lap time simulation will be reserved for another blog post. From here, I ran the first simulations and began to correlate each one with logged data. By exporting each sim file to a CSV, I was able to compare a simulated lap of Homestead with an actual lap of Homestead. This allowed me to tweak the tyre model – which comprises lateral and longitudinal scaling factors – aero coefficients, as well as global grip scaling and power factors. After doing this for every track, I kept a database of scaling factors I needed to use that gave me good correlation while keeping my tyre model constant. A quick note on correlation, this will be covered in a future blog post, but this is arguably one of the most critical steps in building a lap sim model of any kind. Some important items to start with are matching the simulated mid-corner speeds, and top speeds, to reality. These are primarily a function of vehicle mass, lateral tyre performance, engine power, and drag coefficient. For low downforce vehicles such as a Ferrari Challenge car, the lift coefficient is still heavily tweaked to match reality, but its influence on mid-corner performance is quite minimal concerning other fundamental aspects of the model. Selecting a criteria which determines when a model is “correlated to reality” is also an in-depth topic for another post, but focusing on matching vehicle mid-corner and top speeds is a good start. With the vehicle model, track profiles, and correlation sorted, the generation of lap time sensitivities is the next step.

The desire to build a point mass model developed from the fact that this year, I’d be going to tracks I’d never been to before. I wanted to create a tool that could guide me to make more informed decisions on setup items before the weekend started, as well as during sessions. The model would also allow the team to compare one track to the next, aiding in the understanding of how we’d expect not only the vehicle to perform but also correlating a driver’s performance with circuit characteristics so we’d know if a “strong” weekend was expected or not. The primary reason for creating a point mass model is to run sensitivity studies and calculate what parameters matter most to the vehicle performance at each circuit. Some of the performance items we’re interested in are the lap time influence of; lateral and longitudinal tyre performance, vehicle mass, downforce and drag, power, global circuit grip, and global aero scaling (air density). With a more complicated simulation model, one could evaluate tyre energy per corner, sensitivity to vehicle pitch, roll, and yaw, and then back-calculate a host of other vehicle parameters to compare to reality. Critically, for the results to be of much use, they need to be normalized by lap distance so that one track can be compared to the next.

Chart of setup parameter sensitivities obtained from the point mass model for every track on the 2023 Ferrari Challenge calendar.

With the values obtained from the simulations, you can put together a chart similar to the one above. I must explicitly state that the values above have been randomized to prove a point, but the focus on what kind of values we’re interested in evaluating remains the same. What I particularly like about my metrics are the sensitivity ratios. The aero sensitivity returns the ratio of lap time gained per increment of the rear wing’s downforce to drag. The larger the number, the more incentive there is to run more rear wing. Montreal’s Circuit Gilles Villeneuve has a number below 1.0, which suggests that you gain lap time for every increment of rear wing you remove. The TC (Traction Circle) Sensitivity ratio depicts the trade-off in prioritising either lateral or longitudinal performance. The larger the number, the more incentive the team has to run maximum camber. There is not one track on the 2023 calendar which dipped below zero, but Homestead was close and suggested that, if there was a track to sacrifice lateral grip in favour of longitudinal grip to minimize lap time, it would be at that track. From there, the rest of the sensitivities are normalized and ranked. The rankings are then included in my pre-event report to the team so that it not only justifies the starting setup but also gives people an idea of what changes we’ll be making. The rankings also power the discourse in my pre-event reports, as I prep the mechanics on what tools to have ready in pitlane, with the changes that we’re most likely expecting. I like to create a visual such as the graph below, which presents a lot of information about each vehicle but is very easy to digest. Each track is a different colour, and their sensitivities to each input are easily comparable. The graphs below made their way into every pre-event report I wrote this year.

Another great feature of the sim model is the ability to test setups in isolation. I often get asked by the drivers and driver coaches what the top speed difference would be between a maximum and minimum downforce spec package. This year, I was able to quantify that for every event using my simulation model, which meant I not only had an answer for everyone who asked, but if it was significant, I likely mentioned it in my pre-event report. It also lets you explore different setups that, at first glance, you may not have thought would be viable. At one event in 2023, the theoretical lap time for a high and downforce package was nearly identical. I suggested that we run a low downforce spec to start the weekend and prepare for the race, and then trial a high downforce spec to run in qualifying. This was contrary to what we ran at this track in the past, but we never had any other data to push us in a new direction. During the weekend, we ran that low downforce specification which the driver liked more than the high downforce spec, and we ended up bagging a podium as we had great race pace, and could not easily be passed with our top speed advantage.

Finally, one of the most useful outputs from the lap time simulation model is the time lost per every additional 10kg in mass the vehicle carries. Before each weekend, one of my pre-event analysis points would be about pace and tyre degradation, and my prediction about how bad it would be. Looking at the data from the year prior, I could calculate the pace degradation of the best car over a stint, and knowing the fuel consumption per lap, I could figure out the tyre degradation in seconds per lap by removing the fuel effect. This is also something to be saved for a future blog post, but pace degradation is the summation of the pace reduction due to tyre degradation, and the fuel effect. With pace degradation and fuel effect known, tyre degradation can be calculated. This is only true for a set of track and ambient temperatures, as tyre degradation is heavily reliant on these factors, and it also assumes tyre degradation is linear, which in some cases it isn’t. Regardless, the ability to understand the magnitude of tyre degradation at each circuit led the team down some setup avenues that potentially minimized tyre degradation, or counteracted the effects of the vehicle balance shift over a stint. To summarize, the ability to generate sensitivity metrics, trial setups in isolation, and be more informed of the behaviour of the track and vehicle combination before a weekend began, are all what made my point mass models so useful to the team this year.

Pre-Event And Post Event Engineering Reports

The best-performing teams are cohesive and communicate effectively within themselves. One of the biggest improvements in communication this year at R. Ferri Motorsports came with the implementation of pre and post-event engineering reports. I’ve already touched on pre-event reports briefly, but more formally, they aim to inform the team about important aspects of the weekend ahead and to keep the engineering team aligned for setup decisions and calculations. The first page includes various statistics that are needed for basic calculations such as the expected number of race laps and expected fuel consumption. It also shares average race pace, and previous pole position times, which helps the drivers understand what pace is expected of them. After the simulation metrics are reported, the next page dives deeper into the setup of the vehicle, and why certain setup decisions were made. From there, I also comment on the expected setup direction and speak about any test items we’re looking to run through during the weekend. These comments are good because they inform the team about what they need to have ready in pitlane to make these changes. As an example, if you go into a weekend where you’re expecting lots of low-speed oversteer due to historical data, and you already have no margin to soften the rear anti-roll bar or stiffen the front, then the mechanics would know to be ready with ride height adjustment tools. The last page has the starting setups for each car, as well as ABS and traction control settings used in previous years, which provides the drivers and their coaches with a good baseline on where to start. By reading the pre-event report, everyone on the team starts the event fully informed, which improves working efficiency within the team and also reduces the number of questions people ask me.

The post-event engineering reports are arguably more important, as it’s a chance to examine things from the weekend that you maybe didn’t have a chance to do in the moment. The reports typically start with a general weekend summary, coupled with any deficiencies we noticed that need to be addressed before the next event. This is also a good place to write about any novelties that were noticed, such as the vehicle’s GPS not functioning correctly due to the track being situated near a military base (Homestead). From there, the report covers the driver and vehicle performance for each session and looks to quantify the effects of setup changes. Understanding how vehicle performance and balance can migrate with a specific setup change is critical, and documenting it so that the learning is applied to future events is even more important. The race pace is then analyzed, and pace/tyre degradation is compared to what was predicted before the weekend. The end of the engineering report is filled with vehicle and performance metrics generated from the data logged from that weekend. This is a very vital section of the report, as it allows you to not only draw conclusions from performance data but also aid in diagnosing vehicle reliability problems. Metrics deserve their own blog post, but by looking at lap average (or minimum or maximum, or at a single point) data, it can be easy to spot when certain values are trending in a certain direction, and to compare them to previous events. A good engineering report is the foundation of a team’s knowledge gained in that season and helps the race engineer better understand the car's behaviour. There were many times throughout 2023 where, before making a setup change or choosing to run an experiment, we considered the results from the last time we made that particular change. Though report writing can be tedious, it is an absolute necessity for any team looking to seriously challenge for a championship, and be renowned in their series as a well-oiled operation.

This blog describes how I prepared for 2023, but preparation for everyone is different. With that said, the fundamentals remain the same, and being well-organized, well-informed, and well-documented is always a recipe for success.