Plan one-variable aero tests
Generated from
content/lms/race-aerodynamics/06-measure-and-visualize-performance/01-plan-one-variable-aero-tests.md; edit the source file, not this page.
Source path: content/lms/race-aerodynamics/06-measure-and-visualize-performance/01-plan-one-variable-aero-tests.md
Course: Engineer downforce you can actually use
Module: Measure and visualize performance
Estimated duration: 60 minutes
A one-variable aero test is a discipline problem before it is an engineering problem. You are not trying to prove that a part is clever. You are trying to learn whether one defined change made your car behave measurably better, measurably worse, or not measurably different under the conditions you actually tested. That sounds simple, but it is where a lot of club-level aero work fails. The car gets a splitter extension, a wing angle change, a blocked grille opening, and a ride-height tweak all at once. The driver goes out, feels more planted in one corner, loses speed at the end of the straight, and comes back with a story instead of a result.
This lesson teaches the planning habit that prevents that. The bonded material is blunt about the reason. Aerodynamic work is hard to generalize. A change that works on one car may not work on another similar car, and trial and error are part of the development process. That is not permission to guess. It is the reason you test in a way that keeps the guessing small. A good one-variable test gives you a clean comparison: same car, same measurement method, same basic conditions, one intended aero change.
This lesson sits after the lessons on measuring forces, normalizing runs, making airflow visible, and spotting separation. Those skills matter here, but this lesson is narrower. It is about the test plan. Before you tape yarn to the car, plumb a pressure line, change a wing setting, or interpret a speed trace, you decide what question the test must answer, what counts as evidence, what must stay fixed, and what would make the run untrustworthy.
The core rule: choose the variable before choosing the tool
Start every aero test with a one-sentence question. If the sentence contains more than one change, the test is not ready. If the sentence does not say what you will measure, the test is not ready. If the sentence is really a hope about lap time with no mechanism attached, the test is not ready.
A clean question sounds like this: does the intermediate rear wing height produce more useful downforce than the lowest wing height without enough added drag to hurt the lap? Another clean question: does the new outlet louver keep flow attached over the bodywork behind the opening? Another: does sealing this gap increase measured panel pressure difference in the direction we expect? These are not the same test. One is primarily a force and performance compromise. One is a flow-visualization question. One is a pressure question. The part may be the same physical area of the car, but the planned evidence is different.
The weak version sounds like this: try the new aero package and see if it feels better. That is not a test. The source material warns against aerodynamic modification by feel, intuition, or copying what others have done. The comparison to engine tuning is useful: you would not tune air/fuel ratios without measurements, and you should not treat aero differently. Feel matters after you have evidence, because the driver has to operate the car, but feel cannot be the first and only instrument.
The variable must be the thing you deliberately change, not everything that changes during the day. If wing height is the variable, then wing height is the variable. Do not also change angle, endplate, gurney, tire pressure target, ride height, cooling opening, or driving line during the same comparison and then pretend the result belongs to height. If you are testing a splitter extension, the extension length is the variable. If you are testing tuft behavior behind a vent, the vent state is the variable. If you are testing straight-line drag, the hardware state and speed range are the variable boundaries. The rest of the plan exists to defend that definition.
Why one variable matters more in aero than it seems
Aerodynamics punishes lazy comparisons because many devices interact. A wing does not only create a force in isolation. It can affect the flow arriving at another part of the car. An underbody change can alter the car's pitch sensitivity. A cooling opening can change drag and local pressure. Even the air around the car changes when another car is close. The corpus gives the important warning in several forms: competition cars interact when they run close together, straight-ahead testing is not the same as cornering attitude, and the downforce-to-drag compromise can dominate the performance result.
That means a successful test is not just a part swap. It is a controlled comparison. You are trying to prevent three common sources of false learning. First, you do not want a result from a driver adaptation to look like a result from the part. Second, you do not want a result from conditions to look like a result from the part. Third, you do not want one aero effect to hide another because you changed several things at once.
This is also why a slower result is not a failure if the test was clean. The McBeath material is clear that trial and error are essential parts of development. A slower run that tells you a change added drag without enough useful downforce is useful. A faster run that you cannot explain because five things changed is not useful. You are trying to remove some of the guesswork, not create better-sounding guesswork.
Build the test around the evidence type
The bonded sources support several evidence types that are practical outside a professional wind tunnel. You can observe airflow behavior, measure panel pressures, measure airflow speed, measure lift or downforce, and measure drag. You can also use data logging and performance simulation when the car, track, and model are sufficiently known. The important planning move is to pick the evidence type that matches the question.
Use airflow visualization when the question is whether the air is attached, separated, or being guided the way you think. Wool tufts are the lowest-cost example in the corpus. The behavior is specific enough to be useful: attached flow makes tufts line up in orderly rows with only small flutter, while separated flow makes them whirl and point randomly. That does not directly tell you lap time. It tells you whether the air near the surface is doing what the part requires it to do.
Use pressure measurement when the question is about local pressure differences on panels, intakes, outlets, or body surfaces. The corpus does not give a full hardware procedure in the bonded chunks, so do not invent one here. The planning point is enough for this lesson: pressure work answers a different question from tuft work. Tufts show behavior at the surface. Pressure readings can show whether a surface or opening is seeing the pressure environment your design assumes.
Use lift, downforce, and drag measurement when the question is about the primary force result. The sibling lesson on instrumentation belongs here: if you need a force number, use the relevant method and calibrate it. The planning lesson is that you decide before the run whether the test is trying to find more downforce, less drag, a different compromise, or a visible flow improvement that may deserve a later force test.
Use data logging and simulation carefully when the question is performance. The corpus describes performance prediction as a model that needs a track map, car mass and dimensions, roll stiffness, an aerodynamic model, power and torque, gearing, braking, and tire behavior. That is a lot of inputs. It also says downforce and drag values can be related to predicted lap time. The lesson for an intermediate driver is not that you must build a professional simulator. It is that aero performance is not one isolated number. A change that adds downforce may still be wrong for a particular circuit if the drag cost is too large, and a low-drag change may be wrong if it gives away too much cornering or braking capacity.
Write the test card before the first run
A one-variable aero test should have a test card. It can be a notebook page, a shared spreadsheet, or a printed sheet on a clipboard. The form matters less than the discipline. Write it before the car moves, because after the session everyone remembers what they wanted to see.
The first line is the question. The second line is the baseline configuration. The third line is the changed configuration. The fourth line is the measurement method. The fifth line is the run condition you are trying to reproduce. The sixth line is the rejection rule: what makes the run unusable.
For a wing height test, the card might say that the baseline is the lowest wing height and the change is the intermediate wing height. The measurement method might include logged straight speed, driver comments limited to stability under braking and corner entry, and any available downforce or drag estimate. The rejection rule might include traffic, missed shift, obvious driver error, or a run where the car was not in the planned configuration. The point is not that those exact rejection rules appear in the corpus. The point is that the corpus demands careful, rigorous testing and warns that careless recording or calculation can make testing worthless. A rejection rule is how you protect the result from wishful thinking.
For a tuft test, the card looks different. The variable might be vent open versus vent blocked, or splitter fence installed versus removed. The measurement is video, not lap time. The rejection rule might be poor camera view, wet tufts, loose tape, or a run where the car never reaches the target speed range. The success criterion is not that the driver likes the car. It is that the video clearly shows whether the tufts line up or whirl randomly in the area you are testing.
For a drag test, the card looks different again. The corpus specifically warns that skipping two-way averages can make testing worthless. If your drag method uses road or straight-line passes, plan the passes in both directions where that method applies, and write down how many you need before you start. The purpose is to avoid treating one direction, one gust, or one surface condition as truth. The sibling lesson on normalizing the run covers the correction details. Here the planning rule is simple: if your chosen method requires normalization, the normalization is part of the test, not something you add afterward if the result is confusing.
Baseline, change, and sanity check
The simplest structure is baseline, changed configuration, and if time allows, a return to baseline. The return is not magic, and the bonded chunks do not prescribe a named A-B-A protocol. It is a practical way to obey the broader rule that all testing requires concentration and care. If the car returns to the baseline configuration and the baseline measurement also returns near the original result, your confidence improves. If the return baseline does not resemble the first baseline, you do not know whether the changed part caused the difference.
Do not make the plan so ambitious that you cannot execute it. A club test day with limited sessions does not support six devices, three wing heights, two rake settings, and four cooling states. The professional world may use CFD and wind tunnels to work through many configurations, then validate solutions. A track-day driver or club racer usually needs a smaller plan. One baseline and one change done carefully teaches more than a menu of rushed changes.
If you have only one session, use it to collect baseline evidence or a visualization result. The corpus says flow visualization can be used during testing or even in competition when test time is hard to come by. That does not mean every competition run should become an experiment. It means some evidence can be gathered in limited time if the plan is modest. A camera looking at a tufted panel may give you useful airflow behavior without asking the driver to change the car's configuration mid-day.
Control the run condition
Aero is speed-sensitive and attitude-sensitive. The bonded chunks do not provide a complete equation lesson here, but they do provide the practical planning warning: straight-ahead testing is good for basic mapping, optimizing drag, and understanding braking from high speed in a straight line, while more lap time is generally spent in corners. Professional teams test yaw angles representative of cornering because aerodynamic performance can differ between straight-ahead and cornering attitudes.
Your plan must therefore say whether this is a straight-ahead test, a braking-phase test, a cornering-attitude test, or a visual flow test in a specific region. If you test only on the straight, do not claim you proved the device improves the car in a long corner. If you test only in a cornering condition, do not claim you proved the straight-line drag cost. If the part is supposed to work in yaw, plan a way to observe or measure it in a yaw-relevant condition, or state that the test is only a first straight-ahead screen.
This is one of the biggest intermediate-level upgrades. Novices often ask whether the part worked. Intermediate drivers learn to ask where it worked. A splitter, wing, duct, or underbody change can have different value in a straight braking phase, a high-speed corner, and a low-speed corner where aero load is much smaller. The test plan should not overclaim beyond the condition it actually sampled.
Keep other cars out of your result
The corpus is direct that aerodynamic interactions are unavoidable when cars run close together. From the following car's point of view, losses of downforce and grip can be mitigated by lateral offset in the case study described, but the broader lesson for testing is cleaner: do not let another car become an unrecorded variable.
If you are comparing two aero configurations, traffic is not just an inconvenience. It can change the air arriving at your car. It can also change the driver's throttle, braking, and line. A run in clean air and a run tucked behind another car are not the same test. Plan for spacing. If you cannot get clean spacing, mark the run as compromised instead of forcing it into the comparison.
There is a racecraft exception, but it belongs in the plan. If your question is specifically how the car behaves in another car's wake, then the wake is the variable environment you are testing. In that case you still need structure. You would compare a defined car position, spacing, and offset as best you can, and you would not mix those results with clean-air aero development. Clean-air development and traffic-interaction development answer different questions.
Do not judge the part by one number
Aero parts trade benefits. The bonded material points to the downforce-to-drag compromise as a major part of performance analysis, especially in high-downforce and efficiency-sensitive cars. That compromise is why a test plan should define the decision metric before the run. More downforce is not automatically better. Less drag is not automatically better. Better tuft behavior is not automatically faster. The right question is whether the measured change helps the car in the performance context you care about.
For a long straight and modest cornering demand, a drag increase may cost more than a downforce gain returns. For a circuit or section dominated by high-speed corners and braking from high speed, the added downforce may be worth the drag. The corpus supports this way of thinking through the performance-simulation discussion: the car, track, power, gearing, braking, tires, and aerodynamic model all influence the lap-time result. Even if you do not run a simulation, you should think like the result depends on context.
This also keeps you from overreading driver confidence. A car that feels calmer may be faster, but it may also be calmer because it is slower at the end of the straight. A test plan that records speed, force, flow, or pressure evidence keeps the driver comment in its proper place. The driver's words help interpret the evidence. They do not replace it.
Calibration cues: what better testing feels like
A better aero test does not necessarily feel dramatic from the seat. It feels boring in the paddock because everyone knows the plan. The car goes out in the baseline state. The evidence is captured. The car comes back. One thing changes. The car goes out again. The same evidence is captured. The notes are written immediately. Nobody has to reconstruct the configuration from memory.
For flow visualization, improvement looks like clearer behavior, not nicer paintwork. In attached flow, the tufts line up in neat rows and only flutter a little. In separated flow, they whirl and point randomly. A good visualization test produces video clear enough that you can classify the region without arguing from hope. If the tufts are unreadable, the result is not neutral. It is unusable.
For drag or force testing, improvement looks like repeatability. The corpus warns that mistakes in writing down or calculating results can make testing worthless, and specifically names skipped two-way averages as an example. A good test gives you enough disciplined passes that the direction of the result survives the normalization method. If one pass says yes and the next says no, you have not earned a setup change yet.
For performance interpretation, improvement looks like a result that matches the question. If you tested straight-ahead drag, you can talk about straight-ahead drag. If you tested cornering yaw attitude, you can talk about that attitude. If you only filmed tufts, you can talk about flow behavior. That restraint is a skill. The test plan is doing its job when it keeps you from claiming more than you measured.
Worked example: rear wing height without moving the target
The bonded chunks include a simple wing-height comparison reference: the lowest wing height and the intermediate wing height. Treat that as the model of variable definition. The variable is height. If you use it as a one-variable test, you do not also change wing angle, flap state, or endplate shape. You ask a constrained question: does the intermediate height produce a useful performance change compared with the lowest height under the condition tested?
The plan should name the expected mechanism. A wing depends on attached flow and pressure behavior to produce useful force. Another bonded chunk notes that wing twist could be altered so flow across the span remains attached longer, delaying large-scale separation and stall. You are not testing twist in this example, but the mechanism matters: a wing setting is only useful if the flow remains organized enough for the device to work. If the height change exposes the wing to cleaner air, or changes interaction with the body, your evidence should show either a force or performance change, and ideally a flow-behavior clue if you can safely film it.
A disciplined test would run the lowest height as baseline, the intermediate height as the change, and then avoid claiming that the car's entire aero package has been solved. If straight-line speed falls and the car gains stability under high-speed braking or corner entry, the result may still be good at one circuit and bad at another. If lap time improves but traffic changed between runs, the result is suspect. If the driver likes the car but the measured speed loss is large and no cornering benefit appears, you have a different decision to make. The part is not right or wrong in the abstract. The comparison tells you what happened in the tested condition.
Worked example: sports prototypes at small yaw angles
The sports-prototype yaw material is a clean example of choosing the condition before interpreting the result. Straight-ahead testing is useful. It can map basic behavior, optimize drag, and help understand braking from high speed in a straight line. But the source also points out that more lap time is generally spent in corners, and that teams test yaw angles representative of cornering because aerodynamic performance differs between straight ahead and cornering.
So suppose your club prototype, time-attack car, or fast track car gets a new front dive plane or underbody fence. A straight-line run may show little drag penalty or a small speed change. That does not prove the device is useful in a fast corner. It also does not prove it is useless. The one-variable plan must state that the first test is straight-ahead screening, and the later test must sample the cornering attitude if the part is meant to work there.
For an intermediate driver, the key is humility in the conclusion. After a straight test, you may say the change did or did not alter the straight-line evidence enough to matter. You may not say it solved yaw behavior. After a yaw-relevant test, you may compare the cornering-attitude evidence, but you still keep the variable clean. The test earns a narrow conclusion. Narrow conclusions are what make the next test better.
Worked example: a road-based car with measured aero changes
The Edgar bonded material describes a Honda Insight with extensive aerodynamic modifications that reduced drag, produced measured downforce, and improved straight-line stability. The useful lesson is not that you should copy an Insight. The source explicitly warns against copying another car as a substitute for testing. The useful lesson is that a passenger-car-based platform can be measured with low-cost methods, and that drag, downforce, and stability are related but distinct outcomes.
If you were planning a one-variable test on a similar road-based track car, you might start with the easiest observable. If the question is whether a rear body change keeps flow attached, use tufts and video. If the question is whether the change reduced drag, plan the drag method and two-way averages. If the question is whether downforce changed, use the downforce or lift method rather than driver confidence alone. The same car can need several test types over time. The one-variable discipline keeps each test from becoming a vague verdict on the whole build.
Common mistakes
The first mistake is copying a fast car without testing your own. The corpus says what works on one car may not work on another, even if the cars seem similar. Good looks like using another car as an idea source, then writing a test for your car.
The second mistake is changing an aero package instead of a variable. A splitter extension plus wing angle plus cooling change plus ride-height change may produce a lap-time difference, but you will not know why. Good looks like one hardware change, one measurement plan, and one conclusion.
The third mistake is treating feel as proof. The driver may be right that the car feels calmer, but the source material warns against modifying by feel or intuition without measurement. Good looks like pairing driver comments with the relevant evidence: flow video, pressure, drag, lift, downforce, or logged performance.
The fourth mistake is skipping the normalization the method requires. The bonded material specifically names not performing two-way averages as an example of a skip that can make testing worthless. Good looks like planning the averages before the run and rejecting incomplete data instead of massaging it later.
The fifth mistake is using straight-line evidence to make a cornering claim. Straight-ahead testing has value, but yaw and cornering attitudes can differ. Good looks like limiting the conclusion to the condition tested, then planning the next test around the missing condition.
The sixth mistake is testing in traffic and pretending the air was clean. Cars running close together interact aerodynamically. Good looks like clean spacing for clean-air development, or a separate wake-interaction test if that is the actual question.
Drill: the two-session one-variable test card
At your next event, choose one aero question that can be answered without compromising safety or session flow. Do not choose your most complicated idea. Choose one variable you can change repeatably in the paddock: a wing height, a vent state, a small removable trim piece, a taped opening, or a tufted visualization area.
Before the first session, write the test card. Name the baseline configuration, the changed configuration, the evidence method, the run condition, and the rejection rule. In session one, capture only the baseline evidence. Do not chase lap time. Do not adjust the part. After the session, write notes immediately while the configuration is still certain.
Before session two, make only the planned change. Capture the same evidence in the same planned condition. If the session is compromised by traffic, a missed configuration, unreadable video, or incomplete required averages, mark it compromised. Your success criterion is not that the new part wins. Your success criterion is that, at the end of the drill, you can make one of three honest statements: the change improved the measured evidence in the tested condition, the change worsened the measured evidence in the tested condition, or the test was not clean enough to decide.
If you have a third session, return to baseline and repeat the evidence capture. That return run is a confidence check. If the baseline does not come back toward the original evidence, do not force a conclusion. The lesson from the corpus is that testing must be careful and rigorous. Knowing that your data is not trustworthy is better than making a confident wrong setup decision.
When to stop and re-plan
Stop if the test question changes mid-day. Stop if the car cannot be put back into the documented configuration. Stop if your evidence method cannot actually see the thing you care about. Stop if you keep finding yourself explaining away bad data. Stop if the result depends on a traffic interaction you did not plan. Stop if you are about to combine a flow-visualization result, a driver feeling, and an unnormalized speed trace into one big conclusion.
Aero development is full of blind alleys. That is not a reason to be cynical. It is a reason to make each alley short. The value of a one-variable test is that it gives you a clean next move. Keep the change, reverse the change, or plan the next test. If you go quicker, the evidence should tell you why. If you go slower, change back and try something else. That is not failure. That is the process working.
Worked example: rear wing height without moving the target
Use the wing-height comparison as a discipline exercise. The bonded material names the lowest wing height and the intermediate wing height, which is enough to define a clean variable. The baseline is the lowest height. The change is the intermediate height. The test is no longer clean if you also alter wing angle, endplate detail, gurney state, or another bodywork setting. Plan the evidence before the run: if the question is performance, look at the force, drag, speed, or lap-time evidence appropriate to your setup; if the question is flow quality, add visualization that can show whether the flow remains organized. The conclusion must stay narrow. You may learn that the intermediate height was better, worse, or inconclusive in the tested condition. You have not proved every wing setting, every track, or every yaw state.
Worked example: sports prototypes at small yaw angles
The sports-prototype yaw material is the best warning against overclaiming a straight-line test. Straight-ahead testing is useful for basic mapping, drag optimization, and high-speed braking knowledge, but the same source explains why teams also test yaw angles that represent cornering. If your aero change is intended to help in a fast corner, the plan cannot stop at a straight. Use the straight test as a screen, then plan a separate condition that represents the cornering attitude. Keep the variable the same between those tests. Otherwise you will not know whether the different result came from yaw, from the part, or from an unrelated change.
Worked example: road-based car with measured drag and downforce
The Edgar material describes a Honda Insight with extensive aerodynamic modification, reduced drag, measured downforce, and improved straight-line stability. The lesson is not to copy the Insight. The same source warns against copying other cars instead of measuring your own. The lesson is that a passenger-car-based track platform can be tested with practical methods, and that different evidence answers different questions. Tufts can show whether airflow is attached or separated. Pressure methods can address local panel behavior. Lift, downforce, and drag methods address the primary forces. A good one-variable plan chooses one of those questions at a time instead of treating the whole car as one vague aero package.
Common mistakes: six ways to ruin the comparison
The common errors are predictable. Copying someone else's car ignores the warning that what works on one car may not work on another. Changing an entire package hides the cause of the result. Trusting feel alone repeats the poor approach of modifying by intuition without measurement. Skipping required two-way averages attacks the validity of the calculation itself. Using straight-line data to claim cornering behavior ignores the yaw warning. Testing in traffic ignores aerodynamic interaction between nearby cars. Good work looks quieter: one variable, one evidence type, one planned condition, documented configurations, and a conclusion that does not reach beyond the data.
Drill: two-session one-variable test card
At your next test day, run a two-session drill. Before session one, write a card with five lines: question, baseline configuration, changed configuration, evidence method, and rejection rule. Session one captures only baseline evidence. Between sessions, change only the planned aero variable. Session two captures the same evidence in the same condition. The drill succeeds if you can honestly classify the outcome as improved, worsened, or inconclusive in the tested condition. If you have a third session, return to the baseline and repeat the capture as a sanity check. Do not measure success by whether the new part wins. Measure success by whether the comparison stayed clean enough to trust.
When this principle breaks down
One-variable discipline does not mean every aero system is independent. The corpus repeatedly points toward interaction: devices affect flow around other areas, cars interact when running close together, and straight-ahead behavior can differ from cornering attitudes. The principle breaks down only if you pretend interaction does not exist. The response is not to abandon discipline. The response is to plan smaller tests. First isolate the clean-air effect. Then test the yaw-relevant condition. Then, if racecraft demands it, test the wake or following-car situation as its own question. Each test earns one conclusion.
Author Review
No quiz questions are attached to this lesson.
Sources
| # | Document | Chunk | Pages | Score | Collection |
|---|---|---|---|---|---|
| 1 | uio julian edgar car aero testing | 06a78fd3-5941-96b8-badf-a5ccb37cf731 | 6 | 1 | uio_books_raw_v1 |
| 2 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 17fd5a9b-5fdf-ead1-ff69-572014594b23 | 477 | 1 | uio_books_raw_v1 |
| 3 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 90b5a640-d9b2-b0ef-2f6e-f9a0dadce5aa | 411 | 1 | uio_books_raw_v1 |
| 4 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 893cce66-5e94-8af0-6d98-00acc7cbd324 | 383 | 1 | uio_books_raw_v1 |
| 5 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 9f0edfc1-9e8c-3a96-a48d-b0d658513db3 | 385 | 1 | uio_books_raw_v1 |
| 6 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 6edca499-2988-7702-ccc8-3d17b516edff | 385 | 1 | uio_books_raw_v1 |
| 7 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 2abb3a1a-1abc-3549-8f79-9fce704061d6 | 334 | 1 | uio_books_raw_v1 |
| 8 | uio julian edgar car aero testing | 8a8a57a0-d10b-6d9e-7ee7-a345aca95a3c | 4 | 1 | uio_books_raw_v1 |
| 9 | Competition Car Aerodynamics 3rd Edition McBeath Simon | d788f877-dfdc-2c41-96e0-e6a0de38e907 | 412 | 1 | uio_books_raw_v1 |
| 10 | Competition Car Aerodynamics 3rd Edition McBeath Simon | d59eb8fd-22e2-e843-aecb-2b23f2efcbda | 213 | 1 | uio_books_raw_v1 |
| 11 | Competition Car Aerodynamics 3rd Edition McBeath Simon | cd94958f-1042-ceff-8d99-06fa06ac633b | 504 | 1 | uio_books_raw_v1 |
| 12 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 61068e74-0999-1e25-03bd-8c545f352d25 | 26 | 1 | uio_books_raw_v1 |
| 13 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 9e3001fd-e626-5565-9b11-bc3cab151d27 | 281 | 1 | uio_books_raw_v1 |
| 14 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 09774fa8-5f4f-bd8e-8c79-d57ffe9e2cf2 | 9 | 1 | uio_books_raw_v1 |