Engine vital signs & data monitoring
Before looking at performance, confirm the car is healthy: oil and fuel pressure, fluid temperatures, lambda, voltage, rpm. Good alarm channels and trend views catch failures a run before they happen.
Engine vital signs & data monitoring: L1 Foundation explanation
Segers' first rule of data analysis: reliability before performance. A car that stops is infinitely slow, and a car with a hidden problem (low fuel pressure, overheated oil, failing alternator) may simply be down on performance. After every run, check the vital signs before you open the lap comparison.
Core channels and typical healthy values for a racing engine (always check the engine supplier's limits):
| Channel | Typical healthy | Watch for |
|---|---|---|
| Oil pressure (hot, high rpm) | 4–7 bar | drops in long corners (surge), slow decline over run |
| Oil temperature | 90–125 °C | sustained > 130 °C |
| Water temperature | 80–105 °C (pressurised) | > 110–115 °C, rising trend in traffic |
| Fuel pressure (port inj.) | 3–5 bar, steady | drops at low fuel/long corners (pickup), voltage-related drops |
| Fuel pressure (direct inj.) | 100–350 bar | rail pressure not reaching target |
| Lambda | 0.85–0.95 at full load | lean (> 1.0 at full load) → detonation/overheat |
| Battery voltage | 13.5–14.4 V running | below 13 V = charging problem |
| Gearbox/diff oil temp | 80–130 °C | trend over stint |
| Max rpm / over-rev counter | under limit | spikes on downshifts |
Methods (Segers §3.1):
- Statistics table — min/max/mean per run or lap. Quick, but misleading: a 0 bar oil pressure minimum may simply be the car stopped in the pits.
- Whole-run time graph — trends become obvious: temperatures should rise and stabilise.
- Alarm math channels — e.g. low oil pressure only counts when the engine is loaded:
- Driver alarms on the dash, with priorities and the driver knowing what each means and what to do.
Segers' alternator-belt example is the lesson: the voltage was already too low in the run before, nobody noticed, and the car stopped on track.
Engine vital signs & data monitoring: L2 Practitioner explanation
Making the checks robust.
Conditional alarms. Raw thresholds create false alarms (idle, pit lane, cranking). Combine conditions and durations, as in Segers' MoTeC example: oil pressure < 3 bar for > 1 s while rpm > 2500 for > 2 s. Typical set:
- Oil pressure vs rpm: many engines want roughly ≥ 0.7 bar (10 psi) per 1000 rpm up to the relief-valve plateau — use the supplier's pressure-vs-rpm table, not a constant.
- Water temp > limit for > 5 s; separate warning and critical levels.
- Fuel pressure deviation from target (or differential fuel pressure vs manifold pressure in turbo engines) > 0.3 bar.
- Lambda bank-to-bank difference > 0.05 at full load → injector or sensor problem on one bank.
X-Y diagnostics. Plot one channel against its expected driver. Segers shows oil pressure vs lateral g in a GT3 car: points below the normal floor appear only in right-hand corners, increasing with g — oil surge (low oil level or scavenge/tank design). Other useful pairs:
- fuel pressure vs fuel level and lat g (pickup at low fuel),
- water temperature vs vehicle speed (cooling airflow; traffic sensitivity),
- oil temperature minus water temperature (heat exchanger health),
- voltage vs rpm (charging at idle).
Trend across runs. Store per-run max/mean for each vital and plot over the event: a slow rise of max oil temperature at similar ambient = cooling degradation or debris in radiators; increasing crankcase pressure or oil consumption = ring/seal wear.
Cooling decisions. Normalise water temperature to ambient: . If ΔT is 70 °C at 25 °C ambient, expect ≈ 105 °C at 35 °C. That is how you decide radiator blanking (tape) before the session, not after the alarm.
Driver display. Alarm priority matters: Segers' example had a fuel pressure alarm hiding a low-voltage alarm that was the root cause. The driver must also know how to acknowledge an alarm because it can hide other dash data.
Engine vital signs & data monitoring: L3 Advanced explanation
From thresholds to models. Static thresholds only catch gross failures. Better diagnostics compare measured values to a model of expected behaviour:
- Oil pressure model: fitted from healthy runs (pressure rises with rpm until the relief valve, falls with temperature as viscosity drops). Alarm on the residual (measured − expected) rather than on the raw value. A residual drifting −0.3 bar over a weekend is an early bearing/pump/filter indicator long before a 3 bar threshold trips.
- Coolant heat balance: heat rejected ∝ fuel energy rate × (fraction to coolant ≈ 25–30 %). Heat removed ∝ radiator effectiveness × airflow × ΔT. Use rise rate in a known slipstream/traffic phase to estimate radiator margin — tells you how many laps behind another car the engine survives.
- Lambda and knock: knock control retards ignition when knock is detected; a channel of retard per cylinder is both a reliability and a performance channel. Persistent retard on one cylinder = local hot spot, injector or fuel-quality issue; retard on all cylinders on hot days = IAT problem.
Sampling rates. Temperatures: 1–10 Hz is enough. Pressures: 50–100 Hz to see surge in corners. Knock and misfire: ECU-internal high-rate. A 1 Hz oil pressure log will not show a 0.3 s surge in a fast corner.
Sensor faults vs real faults. Segers' alternator example shows temperatures snapping to default values when sensor supply failed — an implausible step (e.g. 92 → 120 °C in one sample) is a sensor/ECU default, not physics. Build plausibility channels: rate-of-change limits, cross-checks (oil temp vs water temp), and redundant sensors where critical. The engineer must decide in seconds on the radio whether to box the car: knowing what a sensor fault looks like versus a real failure is part of the job.
Life tracking. Engines, gearboxes and turbos have mileage/hour limits; track rpm histograms (time above x rpm), over-rev counts and peak temperatures per session in the engine log. These justify (or deny) warranty claims with the engine supplier and decide whether a qualifying engine can still do the race.
Key formulas
Trackside takeaways
Reliability first: check vitals after every run before any performance analysis.
Statistics tables hide context — always look at the whole-run trace and trends.
Make alarms conditional (rpm, duration) to avoid false triggers; prioritise driver alarms.
X-Y plots (oil pressure vs lat g, fuel pressure vs fuel level) expose surge and pickup problems.
Use ΔT to ambient to predict water temperature and decide radiator blanking before the session.
Implausible steps are usually sensor/ECU defaults; know the difference before calling the car in.
On your car
Production-based engines with dry sump; high lateral g (1.8–2 g) makes oil surge checks vs lat g worthwhile, especially at low oil level late in long stints.
Check yourself
A statistics table shows minimum oil pressure 0.1 bar for the run. What is the right next step?
- Segers – Analysis Techniques for Racecar Data Acquisition — §3.1 Check the Car's Vital Signs · statistics table, run trends, alarm channels, oil pressure vs lat g, alternator failure example
- Segers – Analysis Techniques for Racecar Data Acquisition — §16.1 Fuel Consumption · injector/lambda faults detected via fuel accounting