Shakedown repairs with cheap evidence

Generated from content/lms/race-car-mechanic-reference/05-prove-repairs-with-testing/02-shakedown-with-cheaper-resources.md; edit the source file, not this page.

Source path: content/lms/race-car-mechanic-reference/05-prove-repairs-with-testing/02-shakedown-with-cheaper-resources.md

Course: Service the race car that has to finish

Module: Prove repairs and changes with testing

Estimated duration: 65 minutes

Purpose

A shakedown with cheaper resources is the discipline of proving as much as you honestly can before you spend scarce track time, enter a race, or ask the driver to trust a repair at speed. It is not a magic way to prove endurance without endurance. The corpus is blunt on that point: long durability can only be known by doing long durability. What you can do cheaply is remove uncertainty in layers. You can inspect the parts most likely to hurt the driver or end the race. You can measure the static facts that affect every later comparison. You can run short, controlled sessions instead of aimless laps. You can use the tach, gauges, sector times, observer notes, and driver feedback to decide whether the repair behaves like the known baseline or whether it needs to come back apart.

This lesson sits before full race proof. It also sits beside reversible A/B testing, brake-specific testing, and estimate-based planning. Here the question is narrower: after you fix or change something, how do you shakedown the car without burning your most expensive resource first? The answer is to build a ladder from low-cost certainty to higher-cost exposure. Each rung should answer a real question and either let you continue or force you to stop.

The core principle

Cheap shakedown works when each step has a fixed basis of comparison. You do not learn much from a driver motoring around until something feels better or worse. You learn when you know what condition the car started in, what was changed, how hard the car was asked to work, what the driver did consistently, and what changed in the evidence. That is why baseline matters so much. A change can look positive simply because the driver is improving, the tires are coming in, or the track is getting faster. A change can also look negative when the track, weather, or tires have moved away from the earlier condition. If you cannot return to the original condition, you may not know whether the repair is good, bad, or merely tested under different circumstances.

The cheap part is not carelessness. It is sequencing. The first priority in development is still that the car must finish. The second priority is that time at the track is always in short supply. A low-cost shakedown preserves both. You spend shop time and paddock time on inspections and measurements that would be wasteful to discover at speed. You spend short runs on basic function and repeatability. You reserve longer or harder sessions for the point where the cheap evidence says the car deserves them.

Sub-skill 1: separate safety proof from performance proof

Before you ask whether the repair made the car faster, ask whether it is safe enough to test. Race cars contain a few life-or-death parts, including axles, hubs, spindles, hub carriers, and steering arms. Failure there can create a serious accident. Other highly stressed parts in the engine, transmission, and driveline may mainly cost the race and cause internal damage. Both matter, but the first group controls whether the shakedown should happen at all.

That distinction keeps your cheap test honest. If a hub carrier, spindle, steering arm, axle, brake linkage, or other high-consequence part was disturbed, the shakedown starts with inspection rather than driving. The cost of inspection is small compared with teardown time and the possible loss from failure. When a repair touches safety-critical structure, the cheapest resource is not a short session. The cheapest resource is the time it takes to check the part, the fasteners, the mating faces, and the expected motion before the car moves.

This also prevents a common category error. A car can feel normal for two laps and still have a crack beginning in a highly stressed part. A short run does not replace inspection. The run comes after inspection, and it is used to catch dynamic problems that static inspection cannot expose: heat, load transfer, vibration, driver fit in the moving car, brake behavior, temperature and pressure trends, and whether the repair lets the driver repeat the baseline.

Sub-skill 2: measure the facts that make later evidence cheaper

A cheap shakedown often depends on boring measurements. One example from the corpus is rolling tire diameter. The most accurate static method given is to roll the loaded tire for 10 exact revolutions and divide the distance by 314. That number matters because tire size, gear ratio, and gear selection let you relate tachometer readings to road speed when you do not have expensive logging. The brake-testing corpus gives the same practical idea from another direction: if the car has no speedometer, a speedometer shop can record the miles per hour that correspond to tachometer readings, but you must record gear ratio, driving-wheel tire size, and the gear used.

For a mechanic shakedown, this is not trivia. If you repaired a driveline issue, changed tires, changed gearing, changed a speed sensor, or are trying to compare a car without a calibrated speedometer, the tach can become your cheap repeatable instrument. Bondurant emphasizes that the speedometer is not the important instrument in practice; engine revs are. If you choose a predetermined engine red-line and keep it constant, you can gauge improvement or regression with less noise. You can also avoid turning a repair check into an accidental top-speed contest.

The practical habit is simple. Before the car runs, write down the configuration that changes the meaning of your data: tire size, gear ratio, selected gear for any speed check, intended redline or shift point, brake balance setting if relevant, wing or bodywork state if relevant, and baseline alignment or chassis setting if that is the repair area. If the car cannot be returned to a known state, write that down too. Cheap evidence is only cheap if you can interpret it after the session.

Sub-skill 3: make the first dynamic run boring on purpose

The first moving shakedown should not be a development session. Smith describes the early sequence as getting the driver comfortable in the moving car, fitting the driver to the moving car rather than only the stationary one, putting reasonable tires on the car, and letting the driver drive before making chassis changes. If the car is driveable, the early goal is to settle in, get tires hot, bed pads where needed, adjust brake ratio when appropriate, and establish lap and segment baselines. That is already a lot of work. Adding setup experiments at the same time contaminates the result.

For a repaired car, boring is a feature. Warm the car properly. Warm the tires enough that chassis feedback means something. Run only a few laps. Pick two or three safe places on longish straights where the driver can glance at the tach, oil pressure, and water temperature. The point is not to stare at the dash. The point is to choose the glance points before the car leaves pit lane so the driver is not deciding under load. Bondurant's guidance is practical: a quick gauge glance can warn of problems early enough to save a lot of money.

The driver should also hold the driving demand steady. Use a fixed redline, fixed shift point, fixed brake marker if brake testing is not the subject, and the same line as much as possible. If the driver makes a mistake, the instruction is to correct it smoothly and put it aside until the session is finished. A shakedown lap is not improved by emotional editing inside the helmet. You gather the lap, then evaluate outside the car where perspective changes.

Sub-skill 4: use short sessions as a filter, not as a substitute for thinking

Several chunks converge on the same rhythm: a few laps, then stop and discuss. Bondurant recommends short runs followed by stepping out of the car and thinking about what went right and wrong. Puhn recommends only a few laps and one change when checking brake balance, followed by discussion with observers. McBeath summarizes a wing comparison method in which each configuration ran five laps, only the wing configuration changed, lap-time averages were used, and abnormal laps were discarded.

For a cheap repair shakedown, that rhythm is more valuable than the exact number. A three-to-five-lap run is long enough to warm the car and catch many obvious dynamic problems, but short enough that you can inspect before a small problem becomes expensive. The stop is part of the test, not a delay in the test. When the car comes in, you debrief the driver, inspect the repair area, read the gauges or data you have, look for fresh evidence, and decide whether the next run should be the same, easier, harder, or canceled.

This is where many teams waste money. They get relieved when the car survives the first out-lap, then they keep circulating. The better habit is to stop while the information is still fresh and the car is still intact. Cheap shakedown is deliberately interruptive. It trades uninterrupted seat time for more chances to catch a problem while it is still small.

Sub-skill 5: make only one meaningful change at a time

The cheapest shakedown can still produce expensive confusion if you change too much. Smith's development advice is direct: do not make more than one change at a time, at least in related areas. Puhn says the brake balance test should run only a few laps, make only one change, then discuss. McBeath's aero example also keeps the comparison narrow: each configuration is run, only wing configuration changes are made, and averages are compared.

For a repair shakedown, the single-change rule has a slightly different flavor than performance development. You may not be trying to find the best setup. You are trying to determine whether a repair restored known behavior. That means your strongest comparison is often repaired condition versus known baseline, not repaired condition plus new pads plus new wing angle plus different tires plus a driver who is finally comfortable. If you must change another thing for safety or logistics, write it down and lower your confidence in the conclusion.

There is one exception that is not really an exception. Some changes are part of making the car testable. If pads need bedding, that must happen before brake evaluation. If gearing is wrong for the test, the gear change may be required. If the driver is physically uncomfortable in the moving car, the fit must be fixed before feedback is meaningful. But those are enabling changes. Once the car is testable, the evidence gets cleaner when you change one thing, run briefly, stop, and compare.

Sub-skill 6: prefer cheap instruments that answer the exact question

You do not need a professional data system to learn something, but you do need the right kind of evidence. Bondurant gives a driver-level minimum: tachometer, oil pressure, and water temperature checked at predetermined places. Van Valkenburgh emphasizes consistency and baselining. Puhn adds observers during hard braking and speed or tach calibration. McBeath lists traditional aerodynamic test outputs: lap times, sector times, high-speed corner entry, apex, and exit speeds, straight-line speeds, and driver feedback on handling balance. Smith warns not to trust subjective judgments or lap times alone; take corner and straight times so you know where the gain or loss occurred.

The lesson is not that one data source is always enough. The lesson is that each cheap source answers a different question. Gauges answer whether the car is trying to hurt itself. Tach and fixed redline answer whether the driver is comparing like with like. Segment and corner times answer where performance changed. Observer notes answer what the car did from outside, especially under braking. Driver feedback answers whether the repair changed balance, confidence, comfort, or repeatability. Static measurements answer whether the car was actually in the configuration you thought it was in.

When resources are limited, pick the smallest evidence package that matches the repair. A cooling repair needs temperature trend and inspection, not a wing comparison. A brake-balance repair needs known speed or tach relation, clean or representative surface, driver feel, observer notes, and maximum deceleration if available. A bodywork or aero repair needs straight speed, sector time, high-speed corner behavior, and periodic return to baseline because weather, track condition, and tire deterioration can move the target. A suspension or chassis repair needs warm reasonable tires and segment evidence, not cold-tire impressions.

Sub-skill 7: baseline, rebaseline, and stop pretending conditions stand still

A baseline is not just the first thing you wrote down. It is the reference that lets you interpret later evidence. Van Valkenburgh states that all tests should be baselined because you cannot know whether a change is positive or negative without a known fixed basis of reference. McBeath adds a trackside complication: if weather or track conditions change during a session, it is crucial to return to the baseline setup periodically. Tire deterioration can also change the baseline.

For a low-cost shakedown, this prevents two false conclusions. The first false conclusion is that the repair improved the car because lap time improved after the driver warmed up. The second is that the repair hurt the car because tires went away or the track changed. You protect yourself by returning to the known state when possible, using segment evidence rather than whole-lap emotion, and comparing repeatable laps rather than one heroic lap.

The repair version of rebaselining can be simple. If a previous setup sheet exists, compare the repaired condition against it before the run. If the old part can be reinstalled safely and quickly, keep the option available for an A/B test. If you cannot return physically, return procedurally: same tires if possible, same fuel range if known, same driver, same run length, same shift points, same brake references, same data channels, same debrief questions. You may not remove all noise, but you can stop adding noise yourself.

Sub-skill 8: use the debrief to decide the next cheapest question

The out-of-car debrief is where shakedown becomes engineering instead of hope. Bondurant tells the driver to get out and think about what was right or wrong, where the car could be smoother, faster, more consistent, whether apex areas were hit, whether downshifts and trail braking were right, whether the chassis was set through turns, and whether braking was late enough. For this lesson, you are not using that list to coach ultimate lap time. You are using the same habit to separate driver variation from repair evidence.

Ask the driver what happened before you tell the driver what you saw. Did the car repeat? Did the repaired system feel the same every lap? Did the symptom appear at a specific phase: braking, turn-in, apex, exit, straight, shift, curbing, heat soak? Did the gauges move at a certain point? Did the car need more steering or correction than the baseline? Did the driver feel comfortable enough in the moving car to give useful feedback? Then compare that with the objective record you have: lap or segment times, tach at fixed points, temperatures, pressure, observer notes, and inspection evidence.

The next action should be conservative and explicit. If the car shows a safety problem, stop. If the car shows an expensive mechanical trend, stop and inspect. If the evidence is clean but thin, repeat the same short run before escalating. If the evidence is clean and repeatable, increase the demand in one controlled way: a slightly longer run, higher sustained load, harder braking test, or a formal A/B comparison. Do not let relief become a test plan.

Calibration cues

You are improving at cheap shakedown when the car produces fewer mysteries after each run. The first cue is consistency. One very fast lap among scattered times is not meaningful test evidence. A group of repeatable laps, repeatable sectors, or repeatable tach readings is more useful even if it is not impressive. The second cue is location. You know where time or behavior changed: braking zone, straight, high-speed entry, apex, exit, or a particular segment. Smith's point is practical: if you know where, it becomes easier to figure out why.

The third cue is reversibility or at least return-to-reference. You can go back to the baseline setting, or you can repeat the same procedure closely enough that the comparison is fair. The fourth cue is instrument agreement. The driver says the car is stable under braking, the observers do not see abnormal lockup or attitude, the tach/speed relation is known, and deceleration or braking distance does not contradict the story. Or, in an aero/bodywork check, the driver feedback, sector times, high-speed corner speeds, and straight speeds all point in the same direction.

The fifth cue is that you are stopping for reasons rather than feelings. You do not stop because the session clock ended. You stop because the planned few laps are complete, because the next evidence step is clear, or because the car has told you to stop. That discipline is what keeps cheap shakedown from becoming cheap-looking expensive testing.

Common mistakes

The first mistake is aimless motoring. Smith says much race-car testing wastes time, effort, and money when the team arrives without a plan or sends the driver out to enjoy himself. Seat time can be valid early in a driver's career, but development work requires a purpose. Good looks like a written question for the run, a known baseline, a short run length, and a stop point.

The second mistake is testing on cold or worn-out tires and pretending the result is chassis evidence. Smith specifically warns against evaluating chassis performance on cold or worn tires. Good looks like using reasonable tires, letting them reach a useful state, and not drawing chassis conclusions from conditions that hide the real behavior.

The third mistake is making tiny changes too early. One shock click or a similarly small adjustment may tell you nothing before the car is close to optimum. Good looks like using changes large enough to expose direction when you are early in the process, then narrowing the step size later.

The fourth mistake is relying on whole-lap time or driver feel alone. Lap time is useful, but it can hide where the gain or loss occurred. Subjective feel is useful, but it can be fooled by driver adaptation or confidence. Good looks like combining driver feedback with corner times, straight times, observer notes, tach checks, gauges, and inspection.

The fifth mistake is treating a dirty or wet brake-balance test as proof for a dry race. Puhn warns that brake-balance testing on a dirty or wet surface is useless unless you plan to race on a similar surface. Good looks like testing on a representative surface, or clearly labeling the result as only a crude temporary setting until traction improves.

The sixth mistake is skipping inspection because the car completed a short run. Van Valkenburgh's inspection guidance cuts against that habit. Lightweight, highly stressed cars require frequent inspection for potential failure, especially around critical components. Good looks like inspecting before the run, inspecting after the run, and using the run as one piece of evidence rather than the whole proof.

When this principle breaks down

Cheap shakedown has boundaries. It cannot prove that a car will survive 24 hours, 500 miles, or any other long event simply because it survived a few short runs. It cannot turn bad conditions into good data. It cannot make an uncalibrated speedometer trustworthy. It cannot make a driver with scattered laps into a precise test instrument. It cannot prove chassis performance on cold or worn-out tires. It cannot make multiple simultaneous changes interpretable.

When the boundary appears, the professional answer is not to invent certainty. You either reduce the question to one the available resources can answer, or you escalate to the resource the question actually requires. If the question is endurance, you need longer exposure. If the question is brake balance for a dry race and the surface is wet or dirty, you need a representative surface. If the question is whether a safety-critical part is cracked, you need inspection rather than laps. If the question is where time changed, you need segment evidence instead of general impressions.

The operating rule is simple: use cheap resources to remove cheap uncertainty, then spend expensive resources only on the uncertainty that remains.

Worked example: brake-balance repair with no expensive data system

Suppose you replaced or disturbed the brake-balance mechanism, installed new pads, or repaired a hydraulic issue. The cheap shakedown question is not whether the driver can set a personal best. The question is whether the brake system behaves predictably under hard braking and whether balance changes make sense.

Start before the track session. If speed matters and the car has a speedometer, calibrate it. If it does not, use the tachometer as the speed reference, but only after you have recorded the gear ratio, driving-wheel tire size, and the gear used for the check. If tire size changed, remeasure rather than relying on old notes. If you can measure maximum deceleration, record it. If not, still collect driver feel and observer notes.

Choose the surface carefully. A dirty or wet track makes a brake-balance test useless unless you expect to race on a similar surface. If conditions are not representative but the brakes have never been set, make only a crude safe setting and record what you saw so you know how much to move when traction improves.

Run a few laps, not a session-length outing. The driver changes balance to learn how it feels, observers watch hard braking from outside, and the team makes only one balance change before stopping to discuss. The useful evidence is not a single dramatic stop. It is agreement between the driver, the observers, and the numbers you can collect. If the driver reports stability, the observers see no alarming behavior, deceleration or braking reference is repeatable, and the repair area stays clean and secure, you have evidence to escalate. If the car locks unpredictably, the feel changes lap to lap, or the repair area shows leakage or movement, stop. More laps will not make that cheaper.

Worked example: repaired wing mount or bodywork using traditional aero evidence

Suppose a wing mount, splitter support, body panel, or aero attachment was repaired after an incident. You may not have pressure taps, wind-tunnel time, or a large data system. The cheap question is whether the car still behaves like the baseline in the speed ranges where the aero device matters.

First, do not confuse this with a full aero-development program. McBeath's track-testing discussion says traditional testing can record lap times, sector times, higher-speed corner entry, apex, and exit speeds, straight-line speeds, and driver feedback on aerodynamic handling balance. That is enough to detect a meaningful problem if the test is disciplined. It is also enough to fool you if you let conditions drift.

Use the known baseline setup if possible. Run a short baseline or compare against fresh baseline data from the same configuration. Then run the repaired configuration over a fixed small number of laps. The wing-comparison example cited in the corpus used five laps per configuration, changed only wing configuration, averaged lap times, and discarded abnormal laps. For repair proof, the same pattern applies: hold the variables still, compare sectors and high-speed behavior, and do not let unrelated setup changes into the result.

Return to baseline periodically if weather, track condition, or tire condition may have moved. Tire deterioration can change the reference even when the repair is fine. If straight-line speed is down but high-speed corner behavior is stable, you may have drag or bodywork alignment to inspect. If high-speed entry, apex, or exit speed falls with driver feedback of balance change, inspect the aero repair and mounting geometry before pressing on. If the evidence is mixed, repeat the short comparison rather than declaring victory from a single lap.

Worked example: static tire and tach check before a driveline shakedown

Suppose you repaired a driveline issue, changed final drive, swapped tires, or replaced parts that affect speed indication. Before the car runs hard, make the tach useful. Van Valkenburgh gives a simple method for rolling tire diameter: roll the loaded tire for 10 full, exact revolutions and divide the distance by 314. That gives a more accurate working value than an unloaded tire size because the tire deflects under weight.

With tire size, gear ratio, and selected gear recorded, the tachometer becomes a repeatable shakedown instrument. Bondurant's practice guidance is to care about revs rather than top speed and to keep a predetermined redline constant. That matters after a repair because it reduces the temptation to test by feel. The driver can check the tach at chosen straightaway points, stay within the intended limit, and compare exit or straight readings across runs.

The outcome is a cheaper first track run. If the tach relation is normal, gauges stay normal, and the driver can repeat the run without unusual noise, vibration, or shift behavior, you have enough evidence for the next rung. If the tach relation is wrong, the car does not pull as expected, or temperatures and pressures trend badly, you have caught the problem before turning it into a full-session failure.

Drill: the three-run cheap shakedown ladder

Use this drill at the next event after a meaningful repair. It is designed for one driver, one mechanic or crew lead, and one observer if available. The count is three short runs. Each run is three to five laps unless the car gives you a reason to stop earlier. The success criterion is not lap time. Success is repeatable behavior, normal gauges, no inspection findings at the repaired area, and evidence clear enough to decide the next step.

Run 1 is the function run. Before leaving, write the repair question, baseline configuration, fixed redline or shift point, gauge glance locations, and stop point. The driver warms the car and tires, drives below maximum attack, checks tach and gauges only at the planned places, and comes in after the planned lap count. The crew inspects the repair area and critical adjacent parts. The driver gives feedback before hearing the crew's theory.

Run 2 is the repeat run. Make no performance change. Repeat the same procedure. The purpose is to separate a one-lap impression from a repeatable result. If the car behaves the same way twice and inspection is clean twice, confidence rises. If the story changes, do not escalate. Find out whether the variable is driver inconsistency, track condition, tire state, temperature, pressure, or the repair itself.

Run 3 is the one-variable demand increase. Only if the first two runs are clean, increase one thing. That may be harder braking for a brake repair, a longer sustained high-speed section for an aero or cooling repair, or a slightly longer run for a heat-related repair. Do not combine a demand increase with a setup change. Stop after the planned count and inspect again.

The drill fails honorably if you stop early for a real finding. It fails badly if you keep running because you are short on time. Track time lost to inspection is cheaper than race time lost to failure.

Common mistakes: what bad looks like and what good looks like

Aimless motoring is the most expensive cheap shakedown mistake. Bad looks like sending the driver out because the car is together. Good looks like a written question, a short run, and a defined stop.

No baseline is the second mistake. Bad looks like changing parts and judging the result against memory. Good looks like a known setup, known measurement, or repeatable procedure that lets you compare.

Too many changes at once is the third mistake. Bad looks like repairing the car, changing tires, moving brake balance, altering wing angle, and then trying to explain the lap time. Good looks like one meaningful change at a time and an honest note when logistics force another change.

Cold-tire chassis judgment is the fourth mistake. Bad looks like deciding the car has understeer or oversteer before the tires are in a useful range. Good looks like using reasonable tires, warming them, and holding chassis judgment until the feedback can mean something.

Whole-lap worship is the fifth mistake. Bad looks like calling the repair good because one lap was fast. Good looks like repeatable sectors, corner and straight evidence, driver feedback, observer notes, and inspection.

Skipping critical inspection is the sixth mistake. Bad looks like assuming a short run proved a steering, hub, spindle, axle, or hub-carrier repair. Good looks like inspection before and after, with the run used only to add dynamic evidence.

Cross-references to related skills

Use the reversible A/B test lesson when the question is whether one configuration is better than another and both states can be restored. This shakedown lesson gives you the cheaper precondition: the car must be safe and repeatable enough for A/B evidence to mean anything.

Use the undo-focused planning lesson when the change itself may need to come back out quickly. Cheap shakedown depends on the ability to return to a known condition or at least return to a known procedure.

Use the brake-specific lesson when the repair concerns pads, balance bar, hydraulic work, or braking confidence. This lesson only gives the shakedown structure around that work.

Use the race-proof lesson when the question becomes whether the whole car can survive the event. Cheap shakedown can justify escalation, but it cannot honestly replace race-distance proof.

Use the estimates lesson before you decide where to spend the next session. Van Valkenburgh's handicap-method discussion and Smith's priorities both point in the same direction: resources should go where the potential gain or risk justifies them, not only where the team enjoys working.

Author Review

No quiz questions are attached to this lesson.

Sources

#	Document	Chunk	Pages	Score	Collection
1	Tune To Win Carroll Smith	661f2c93-57bd-f041-90d0-fc9ff0cb634b	160	1	uio_books_raw_v1
2	Tune To Win Carroll Smith	ce81b94c-7b42-8fa1-7e9b-115ac71adcbe	162	1	uio_books_raw_v1
3	Race Car Engineering Mechanics Paul Van Valkenburgh	4a0085b1-a5b6-20ef-c288-ff092fa3e4d9	116	1	uio_books_raw_v1
4	Race Car Engineering Mechanics Paul Van Valkenburgh	6761997c-1267-f401-0671-5bfbf75c8c8d	104	1	uio_books_raw_v1
5	Race Car Engineering Mechanics Paul Van Valkenburgh	d30e9f9a-0bf8-9e64-1cc3-8190d86f09d0	133	1	uio_books_raw_v1
6	Brake Handbook Fred Puhn	e38a4194-d555-2ffd-739b-884f82a25adf	117	1	uio_books_raw_v1
7	Competition Car Aerodynamics 3rd Edition McBeath Simon	4adf8cb4-89c7-1b45-bd4d-9bb03634ecf3	345	1	uio_books_raw_v1
8	Bob Bondurant on high performance driving Bondurant Bob 1933- Blakemore John etc.	904a3b7a-e9c4-fb8a-61a3-305e69dfe439	134	1	uio_books_raw_v1