Use segment reports to isolate variance
Generated from
content/lms/data-interpretation-ii-advanced/06-metric-driven-analysis/01-segment-reports.md; edit the source file, not this page.
Source path: content/lms/data-interpretation-ii-advanced/06-metric-driven-analysis/01-segment-reports.md
Course: Read the data your hands can't feel
Module: Build metrics that survive the noise
Estimated duration: 65 minutes
The skill: turn a lap-time problem into a location problem
A segment report is not the final answer. It is the filter that tells you where to start looking. When you come in from a session and say the car felt worse, the lap felt messy, or the stopwatch says you were down, the segment report turns that vague complaint into a smaller question: which part of the lap changed, and did it change once or keep changing?
That is the heart of the skill. You do not open every trace at once and stare at the whole lap. You first break the lap into meaningful pieces, compare those pieces across laps or against a reference, and identify the section where the largest useful variance lives. Then you open the channels that can explain that section: speed, throttle, brake pressure, steering, RPM, gear, lateral and longitudinal G, GPS line, and any other channel your logger gives you. The segment report tells you where the time went. The traces tell you why it went there.
For an intermediate driver, this matters because most lap-time loss does not feel clean from the seat. A small hesitation at throttle pickup can feel like a corner-exit problem, but the segment report may show that the real loss began with a weak brake release or a missed minimum-speed phase. A lap that feels bad overall may only contain one poor segment caused by traffic. A lap that feels smooth may still be slow because every corner entry is slightly conservative. Segment reports keep you from diagnosing the whole lap from memory.
The rule is simple: isolate variance before you interpret variance. First find the segment that is different. Then decide whether the difference is driver execution, traffic, car behavior, conditions, or a measurement artifact. Only after that do you set the next-session objective.
What the segment report is actually measuring
Lap timing starts with a beacon. The logger records when the car passes the lap beacon, whether that beacon comes from an infrared pulse, a GPS point, or a manually entered point in the data. A segment report adds more beacons around the track. Those added points are usually virtual beacons, placed at fixed distances from the start-finish reference. The software then measures elapsed time between consecutive beacons. Each row or column in the report is a question about one section of track.
That mechanical fact matters. A segment is not a corner unless you place its boundaries so that it behaves like one. If you put one beacon before a braking zone and the next far down the following straight, the report blends braking, cornering, throttle pickup, gear choice, exit speed, and straight-line acceleration into one number. That can be useful as a coarse first pass, but it cannot tell you which phase caused the loss. If you need to investigate cornering performance, the corner can be its own segment. If you need more detail, the corner can be divided into entry, apex, and exit.
Good segment placement begins with the question you are asking. If you want to evaluate engine performance, use an accelerating straight where the car is pointed and the driver is not asking for much lateral grip. If you want to evaluate cornering performance, put boundaries around the corner or split the corner into phases. If you want to understand whether an aero or drag change helped, compare the places where that change should show up: faster corner sections for downforce, straight-line sections for drag, and then the total elapsed result to see whether the trade was worth it.
The report is only as useful as those boundaries. A bad boundary hides cause and effect. A good boundary exposes the phase where the driver or car is inconsistent.
Why distance alignment matters
Once a segment report identifies the section, compare traces by distance, not by time. The reason is practical. Two laps are rarely at the same place on track at the same elapsed time once a difference has begun. If one lap brakes earlier, turns sooner, lifts, or exits faster, the time-based traces start to drift away from the physical track. You may think you are comparing brake release at the same corner, but one trace is already farther down the road.
Distance alignment lets you compare the same place on the track. The speed at the 300-foot board, throttle at the apex zone, steering at the same curb, and RPM at the same exit all line up as track locations instead of elapsed-time coincidences. That is why the speed trace is normally the first follow-up after the segment report. Speed is the result channel. It shows where the car was faster or slower. Once you know where the speed difference appears, you inspect the driver channels that produced it.
Treat the segment report as the table of contents. Treat the distance-aligned traces as the chapter.
The triage pass: do this before deep analysis
Start with the whole session, not only the single fastest lap. Fastest laps are useful, but they can hide the real learning opportunity. One lap may be helped by a draft, harmed by traffic, or saved by one excellent corner while the rest of the lap is ordinary. If you only compare fastest lap to fastest lap, you can miss the consistency problem that is actually costing the weekend.
After the car comes in, download the logger every time. Before chasing performance, glance at the vital channels if you have them: temperatures, pressures, battery voltage, and any obvious sensor problems. If the data stream itself is suspect, a segment report can lead you into a false diagnosis. Then listen to the driver notes. Did traffic appear? Did the car feel loose only in one corner? Was there a missed shift? Was the lap interrupted? Driver memory is not enough to prove the cause, but it is useful for knowing what to inspect.
Now open the segment report. Do not start by asking which lap was best. Ask which segment shows the largest useful shortfall. Useful means it is not obviously traffic, not a beacon glitch, not a warm-up or cool-down artifact, and not a one-off that cannot be repeated or practiced. Highlight the fastest and slowest segment times, compare each lap against the best lap or a reference lap, and look for two things: the largest deficit and the largest spread.
The largest deficit tells you where your current lap is losing the most time against the reference. The largest spread tells you where your driving is least repeatable. Those are related but not identical. A segment can be consistently slow against a better driver, which means you have a technique gap. A segment can be sometimes quick and sometimes poor, which means you already have the pace in that section but cannot reproduce it. The first problem asks for a model. The second asks for repeatability.
This is why the segment report belongs before the trace dive. It keeps you honest about priority. If one section is costing half a second and another section is costing five hundredths, the half-second section gets first attention unless there is a safety reason to do otherwise.
From where to why: the channel order
Once the report points to a segment, open the speed trace for that section. Speed is not always the only important channel, but it is the cleanest first view because lap time is made from speed over distance. Ask where the speed delta begins. Does the slower lap arrive slower before braking? Does it brake earlier? Does it reach a lower minimum speed? Does it stay slow after apex? Does it recover on the straight? The first visible speed separation often tells you which phase to inspect.
Then add throttle. Look for coasting, hesitant application, early throttle followed by a lift, or lifts in fast corners. A segment report may show that a corner-exit section is slow, but throttle may reveal two different causes. One lap may wait too long to apply throttle. Another may apply too early, force a lift, and lose more time even though the first throttle movement happened sooner. The segment number alone cannot separate those mistakes.
Then add brake pressure. Look at the shape of the initial application, the trail, and the tail. A light, long brake event can make the segment slow in a different way than a hard, short event. Inconsistent pressure may show why the same corner alternates between quick and poor. A long brake tail into a section that should be released sooner can delay rotation and throttle. A sudden pressure release may show up as entry instability or a low minimum speed. Again, the segment report tells you where to inspect; the brake trace gives you mechanism.
Then add steering, RPM, and gear. Steering helps you see whether the driver is adding angle suddenly, carrying excessive total steer, correcting after throttle, or turning in differently lap to lap. RPM and gear show whether a segment loss is actually a shift timing or gear-choice problem. A segment bounded around a straight can make an engine or shift issue obvious, but the segment time cannot tell you whether the loss came from throttle hesitation, a short shift, a wrong gear, or a poor exit speed feeding the straight.
If you have G channels, use them to verify the hypothesis. Lateral and longitudinal acceleration show how the tire is being asked to work. Combined G or G-sum can help you see whether the driver is blending brake, turn, and throttle well or leaving grip unused. GPS line can confirm whether a different path caused the speed difference. A throttle histogram can support a broader session pattern by showing whether time at full throttle, part throttle, or no throttle changed across runs. These are supporting channels, not replacements for the segment-first process.
Reading the patterns: what variance means
Not all variance is the same. Your job is to name the pattern before you prescribe the fix.
The first pattern is an execution spread. Your best segment is close to the target, but several laps in the same section are much slower. That means you have already produced the time at least once. The next objective is not to discover a new technique; it is to reproduce the clean version. In the traces, execution spread often appears as inconsistent brake pressure, different throttle timing, changing steering shape, or different minimum speed from lap to lap. The next-session goal should be one repeatability cue, not a whole new driving style.
The second pattern is a reference gap. Every lap is slow in the same segment compared with the instructor, target lap, or previous outing. That means your current model for that section is probably conservative or technically different. You may be braking too early, carrying too little entry speed, giving up exit, or using a line that makes the whole section longer or slower. The report only identifies the gap. The traces and, if available, GPS line show which part of the model differs.
The third pattern is a contaminated lap. One lap is slow in one segment, but the rest of the lap is normal and the driver remembers traffic or a mistake. Do not build a theory around that lap. Mark it, exclude it from performance judgment, and keep it in mind only as traffic evidence. Traffic can have a considerable influence on lap time, and a segment report is one of the best ways to avoid blaming technique for a lap that was simply blocked.
The fourth pattern is a setup or condition signature. After a setup change, the affected segments should make sense. If you added downforce, you would expect the clearest difference in corners where downforce matters, not necessarily in a slow hairpin. If you increased drag, straight-line segments should be watched closely. If the driver says the car is not drivable but the speed and driver-activity channels do not show the expected performance change, look for the obvious and check the channels that should respond to that setup change. Do not let a segment report become a story generator.
The fifth pattern is a lap-combination gap. Your theoretical fastest is clearly quicker than your actual best lap because your best segment times are scattered across different laps. That is useful information. It means the car and driver package has shown the pieces, but the driver has not assembled them into one lap. For an intermediate driver, that often points to attention management and objective discipline. You do not need five new techniques; you need to protect the strongest segment while improving the weakest one without creating a new leak elsewhere.
Using theoretical fastest without lying to yourself
The theoretical fastest lap is the sum of the fastest segment times in the outing. It is an estimate of potential, not a promise that the driver could simply decide to run that lap on command. The source example from Istanbul shows the software dividing the track into seven segments, highlighting fastest and slowest segment times, and calculating a theoretical fastest lap that was 0.235 seconds quicker than the actual fastest lap. That small gap tells a different story than a gap of several seconds.
Use theoretical fastest as a consistency gauge. If your actual best lap is close to the theoretical fastest, you are already assembling most of your best segments into the same lap. The next gain probably requires raising the quality of a particular segment. If the theoretical fastest is much quicker than the actual best, the first job is to stop leaking your best work across different laps. Very good drivers tend to get close to the theoretical number because they repeat the strong pieces together.
Be careful with the emotional trap. Theoretical fastest can make a driver think the lap is already available. It is better to treat it as a map of demonstrated pieces. Ask which pieces were real, clean, and repeatable. A best segment created by traffic draft, a missed beacon, or an over-risky entry should not become your standard. A best segment created by a clean brake point, stable release, early committed throttle, and normal line is a valid target.
Fastest rolling and similar software indicators are useful for the same reason: they try to estimate the performance potential of the configuration without pretending that every single best sector happened in one ideal lap. Use them as prompts for investigation, not as trophies.
How to choose segment boundaries
Begin coarse, then split only where the data demands it. A good first report might divide the track into five to seven sections. That is enough to expose the large problem without drowning you in numbers. Once the report identifies the problem section, split that section into smaller phases. A corner can become entry, apex, and exit. A complex can become approach, first corner, connector, second corner, and exit straight. A straight can become launch from the previous corner, mid-straight acceleration, and braking approach if you need to separate exit speed from engine pull and braking point.
Place boundaries where the car changes jobs. Before braking is a different job than braking. Brake release and rotation are different from full-throttle exit. A straight-line acceleration test is different from a high-speed cornering test. Segment boundaries should respect those jobs. If a segment contains too many jobs, it will still tell you that time moved, but it will not show which skill moved it.
Also leave room for causality. If you place the exit segment boundary exactly at the apex, you may miss the fact that a poor entry caused a poor exit. If you place the boundary too far down the straight, you may mix corner exit with engine performance. The segment should be long enough to capture the consequence but short enough to keep the cause visible.
When comparing drivers, make the boundaries fair. The Skip Barber example uses segment-time differences and speed points around the course to identify where the client differs from the instructor. That works because both laps are being compared through the same physical sections. If your logger allows a reference lap in the dashboard, sector or predictive timing can give the driver immediate feedback, but the same rule applies after the session: compare the same section, then investigate the channels that explain the difference.
How to turn the report into a next-session objective
The output of this process should be one objective, not a long list. The data-analysis process is systematic: overview, inconsistencies, details, cross-checks, why, comparison, calibration to your driving, imagined ideal, and next-session objective. Segment reports fit that flow because they keep the objective bounded.
A weak objective says to be faster in Turn 7. A useful objective says to make the Turn 7 to Turn 9 segment repeatable by matching the best lap brake-release shape and avoiding the throttle lift after initial pickup. Another useful objective says to test whether the straight-line deficit is exit speed or acceleration by comparing speed at the exit beacon, mid-straight, and brake marker. Those objectives are small enough to practice and specific enough to verify.
Before you leave the paddock, write down the target segment, the suspected cause, the channel that proves it, and the change you will make. If the target is corner entry, your proof channel may be brake pressure plus speed at turn-in. If the target is exit, your proof may be throttle shape plus speed at the next beacon. If the target is a setup change, your proof may be the segment where the change should matter, supported by speed, wheel load if available, or objective observation.
The next session should not become a scavenger hunt. You are trying to change one thing in the section that matters most.
Calibration cues: how you know the skill is improving
The first cue is that your segment reports start producing decisions faster. Instead of staring at the whole lap, you can say which section deserves attention and why. You become less impressed by a single fast lap and more interested in repeatable segment quality.
The second cue is a smaller spread in the target segment. If the section used to swing widely from lap to lap and now clusters tightly, you have improved repeatability. The fastest lap may not immediately change if you are practicing a conservative repeatable version, but the report will show a narrower band. That is useful progress.
The third cue is a better relationship between actual best and theoretical fastest. When more of your best segments occur on the same lap, the gap closes. That means your driving is becoming more assembled. You are no longer borrowing one good corner from lap three, one good straight from lap five, and one good braking zone from lap eight.
The fourth cue is trace agreement. The speed delta appears where you expected. The throttle, brake, steering, gear, or G channels support the explanation. If the report says the segment improved but the traces show a weird one-off or a contaminated lap, you do not trust the improvement yet. If the report and traces tell the same story across several laps, you can trust the change.
The fifth cue is a cleaner driver debrief. You stop saying the whole lap felt off. You can say the car was fine through the fast opening section, the large loss was the Turn 7 to Turn 9 section, and the speed trace shows the deficit began at entry rather than exit. That is the difference between complaining and analyzing.
How this lesson connects to the sibling lessons
Plot the trend before you judge the change belongs before you draw a big conclusion from one segment spike. Segment reports can reveal a one-lap loss, but run charts and session trends help you decide whether that loss is noise, traffic, temperature, fuel load, tire state, or a real change. Use the segment report to locate the section. Use the trend to decide whether the section is behaving consistently over time.
Design a personal metric for your weakest phase belongs after this lesson. Once the segment report identifies your weak phase, you can build a metric around it. If the weak phase is throttle pickup, you may track throttle hesitation or time to full throttle. If the weak phase is braking, you may track brake pressure shape, release tail, or speed at turn-in. If the weak phase is cornering commitment, you may track speed through a bounded corner segment. The segment report tells you where your metric should live.
The full chain is: trend for context, segment for location, trace for cause, personal metric for practice.
Instructor-level discipline: what not to overread
A segment report is powerful because it compresses a session into comparable numbers. That is also why it can mislead you. The number does not know whether you were in traffic. It does not know whether a sensor drifted. It does not know whether you changed line to let a faster car by. It does not know whether your fastest segment was a clean repeatable technique or a messy risk that happened to work once.
Your discipline is to treat every surprising segment as a question. Check the lap context. Check the driver note. Check the speed trace. Check the supporting channels. Ask why until the explanation matches more than one piece of evidence.
That is the difference between using data and worshipping data. The goal is not to have more screens open. The goal is to leave the paddock with one better job for the next session.
Worked example: Skip Barber segment map from Turn 7 to Turn 9
The Skip Barber computer-coaching example is the clean model for this lesson. The instructor first creates a target lap, then compares the client and instructor by segment times and significant speed points around the course. The important move is not that the software can draw a track map. The important move is that the analysis starts with the segment-time differences so the driver does not waste attention on parts of the course that are already close.
In the example, the run from Start-Finish through the exit of Turn 2 is only a tenth apart. That is not the first coaching target. It may still contain details, but it is not the largest shortfall. The analysis then finds larger differences elsewhere and chooses the big problem first: from the entry to Turn 7 to the exit of Turn 9. Only after that priority is set does the coach inspect speed differences through the segment and look for the cause of the speed deficit.
That order is the lesson. If you open the whole-lap overlay first, you may get pulled into every small wiggle. If you start with the segment report, you can say that Turn 7 to Turn 9 is the main work area. Then the speed trace tells whether the client arrives slow, brakes too early, gives up minimum speed, hesitates on exit, or loses speed at multiple points. Throttle, brake, steering, RPM, and gear then explain the speed picture.
Notice also that the example compares speeds at significant points, not just total segment time. Segment time says the section is slow. Speed points show how the slowness is distributed. If most of the difference is at corner entries, the coaching target is different than if the loss grows only after throttle pickup. The segment report narrows the field; the speed points and traces define the correction.
At your next event, copy the method even if your software is simpler. Pick a reference lap, split the track, find the largest clean segment difference, and refuse to chase a smaller section until you understand the big one.
Worked example: using split segments to judge a wing change
The aerodynamics example teaches the same habit from the car-development side. If you change a rear wing, a total lap time alone is not enough. The car may gain speed in a fast corner and give some of it back on the straight. The right question is where the change should matter.
If the goal is improved cornering speed from added downforce, time the car through corners where that downforce should make a difference. If the worry is drag, inspect straight-line segments. Then relate those section times back to the total lap result. This prevents a common setup mistake: feeling one section improve and assuming the whole lap improved.
The source example also points out the value of objective observation. A trusted observer at an important track section may report that the car looked planted through Turn 2 compared with other cars. That observation does not replace the segment time, but it helps interpret it. If the Turn 2 split improves and the observer saw a more stable car, the story is stronger. If the split does not improve even though the driver reports more confidence, you have a different question. Maybe the wing helped balance but added drag, maybe the driver did not exploit it, or maybe the segment boundary is not capturing the right place.
For a driver, this matters even when you are not changing wings. If you make a technique change that should improve corner entry, do not judge it by the whole lap first. Judge the entry segment, then see whether the exit and following straight paid for it or punished it. If you brake later but lose exit speed, the total section from entry to the next straight will tell the truth. If you enter a little more conservatively but get to throttle cleanly and carry more speed to the next beacon, the segment report will show that too.
Drill: three-session segment variance loop
Use this drill at your next data-capable event. The count is three sessions. The working time is ten minutes after each session, plus the normal time you need to download data. The success criterion is simple: by the end of the third session, you can name one target segment, show whether its lap-to-lap spread improved, and point to the speed plus one driver channel that explains the result.
Session 1 is the baseline. Download the data as soon as the car comes in. Check that the basic channels are believable. Mark warm-up, cool-down, obvious traffic, and missed-shift laps. Build a coarse segment report with roughly five to seven sections around the lap. If your software already has sectors, use them as a starting point, but adjust if they hide the question you care about. Compare all clean laps, not only the fastest lap. Circle the largest useful segment spread and the largest useful reference gap. Choose one target segment for Session 2.
Before Session 2, open the distance-aligned speed trace for that target segment. Add throttle if the loss appears at or after exit. Add brake pressure if the loss begins at approach, braking, or rotation. Add steering if the line or correction pattern looks suspicious. Write one objective in plain language. For example: make the entry-to-apex portion repeatable by matching the best-lap brake-release shape, or protect exit speed by applying throttle once and not lifting. Do not write three objectives.
Session 2 is the test. Drive three laps with the target segment in mind, but do not sacrifice safety, flags, traffic judgment, or the rest of the lap. Your job is not to force a hero segment. Your job is to produce comparable data. After the session, repeat the report. Did the target segment spread shrink? Did the best time improve? Did the adjacent segment get worse? Open the same speed and driver-channel traces. If the trace does not support the improvement, treat the result as unproven.
Session 3 is the confirmation. Keep the same target unless Session 2 proves that the original diagnosis was wrong. Make one refinement, not a new plan. If throttle was the issue, refine throttle timing. If brake-release shape was the issue, refine release. If traffic contaminated the laps, collect cleaner laps before changing the objective. After Session 3, compare the target segment across all three sessions and compare actual best lap to theoretical fastest. Success is not only a quicker lap. Success is a narrower target-segment spread, a trace-supported explanation, and one clear next objective.
Common mistakes: when segment reports lie to you
Mistake 1: treating the fastest lap as the only lap. A fastest lap may contain one excellent section and several ordinary ones. Good analysis looks across all clean laps and checks consistency and inconsistency. What good looks like: you compare the fastest lap, the theoretical fastest pieces, and the lap-to-lap spread before deciding what to practice.
Mistake 2: using segments that are too broad. If one segment includes braking, turn-in, apex, throttle pickup, a shift, and half a straight, the report can tell you that time was lost but cannot tell you which job caused it. What good looks like: start coarse, then split the problem area into entry, apex, and exit or into the specific phases that match the question.
Mistake 3: overlaying by time after the segment points to a problem. Time alignment lets the traces drift away from the track location once the laps differ. What good looks like: compare by distance so the same point on the track lines up across laps.
Mistake 4: diagnosing traffic as technique. A slow car, point-by, local yellow, or compromised pass can ruin one segment without saying anything about your skill. What good looks like: mark contaminated laps and do not build a coaching plan around them.
Mistake 5: stopping at the segment table. Segment time is a where tool, not a why tool. What good looks like: after the report identifies the section, open speed first, then the driver and vehicle channels that can explain the speed difference.
Mistake 6: chasing the smallest visible difference because it is interesting. Data makes every corner tempting. What good looks like: choose the biggest useful shortfall first, just as the Skip Barber example prioritizes the section with the meaningful segment-time gap instead of obsessing over a tenth in a section that is already close.
Mistake 7: believing theoretical fastest as a guaranteed lap. Theoretical fastest is assembled from best segments, and those segments may have happened on different laps under different circumstances. What good looks like: use it as a consistency gauge and ask whether the best segments are clean, repeatable, and compatible with each other.
Mistake 8: making a setup conclusion from the wrong segment. If a change should affect high-speed cornering, but the only gain is in a slow straight-adjacent section, the explanation may be driver execution or noise. What good looks like: inspect the segments and channels where the change is expected to show up, then check the total elapsed effect.
Mistake 9: leaving without a next-session objective. Analysis that does not change the next session is only entertainment. What good looks like: you leave with one target segment, one suspected cause, one proof channel, and one driving change.
When this principle breaks down
Segment reports are less useful when the data is not trustworthy. If the beacon point is wrong, GPS alignment is poor, laps are not aligned to the same distance reference, or a key channel is missing or broken, the report can create false precision. In that case, fix the measurement problem first or fall back to carefully recorded split times and observations.
They are also limited when the session is too contaminated. Heavy traffic, repeated flags, passing drills, warm-up laps, cool-down laps, or inconsistent run plans can make the segment table noisy. You can still learn from the data, but you should not treat it like a clean performance comparison.
A third limitation is that segment time can hide tradeoffs across boundaries. A later brake point may improve the entry segment and damage exit speed beyond the next beacon. A more committed fast-corner line may improve one section while increasing risk or tire demand later. That is why you always inspect adjacent segments and the full lap impact after you identify a local gain.
Finally, segment reports cannot replace judgment. The report may say a risky entry was quicker once. That does not mean it is the right target for an intermediate driver. The correct target is the fastest section you can understand, repeat, and integrate into a whole lap without creating a larger safety or consistency problem.
Author Review
No quiz questions are attached to this lesson.
Sources
| # | Document | Chunk | Pages | Score | Collection |
|---|---|---|---|---|---|
| 1 | Analysis Techniques for Racecar Data Acquisition | f781d637-931d-357a-31c8-ef327207ce15 | 7 | 1 | uio_books_raw_v1 |
| 2 | Going Faster Mastering the Art of Race Driving - Carl Lopez | 99cefdf0-340b-dd25-d11e-19ae5c329ccf | 149 | 1 | uio_books_raw_v1 |
| 3 | Analysis Techniques for Racecar Data Acquisition | 48d56105-c3c5-81ff-d649-243a1cd43c35 | 6 | 1 | uio_books_raw_v1 |
| 4 | Analysis Techniques for Racecar Data Acquisition | 298d8370-448d-3f4a-4164-cc740c02801e | 7 | 1 | uio_books_raw_v1 |
| 5 | Data for Drivers | cabda699642b26311b0a7ef998da2c71 | 15 | 1 | uio_books_raw_v1 |
| 6 | Briefing on High-Performance Driving and Event Operations | d91cd52a-c875-8a59-4539-4257a7b4d757 | 1 | 1 | uio_books_raw_v1 |
| 7 | Analysis Techniques for Racecar Data Acquisition | 3516d932-714f-e990-2877-853744eef9e5 | 18 | 1 | uio_books_raw_v1 |
| 8 | Competition Car Aerodynamics 3rd Edition McBeath Simon | 5f8444f1-9007-b180-bf39-4bd1c06b0296 | 342 | 1 | uio_books_raw_v1 |