PULSCAR ANALYSIS OF NHTSA PUBLIC DATA — VALVE COVER GASKETS AND OIL LEAKS FROM THE TOP OF THE ENGINE PulsCar analysis of NHTSA public data, September 2026. License: CC BY 4.0. Files in this folder may be reused with attribution to PulsCar (pulscar.io). -------------------------------------------------------------------------- 1. SOURCE FILES -------------------------------------------------------------------------- NHTSA public flat files downloaded from static.nhtsa.gov and loaded into a local SQLite database: Complaints FLAT_CMPL.txt 777,040 owner complaints (ODINO), received 1 January 2015 - 8 September 2026. The set below runs 5 January 2015 - 4 September 2026. Field definitions: https://static.nhtsa.gov/odi/ffdd/cmpl/CMPL.txt Recalls FLAT_RCL.txt one row per make/model/model year per campaign; campaigns counted DISTINCT on the campaign number and the population taken as the campaign maximum. Communications FLAT_TSBS.txt manufacturer communications; counted DISTINCT on the NHTSA document id. Complaints are owner reports. They are not verified repairs, not failure rates, and they carry no denominator of vehicles on the road. Nothing here ranks reliability. -------------------------------------------------------------------------- 2. THE COMPLAINT QUERY (1,310 -> 1,200 -> 1,195 -> 1,172) -------------------------------------------------------------------------- Step 1 - full-text search (FTS5) of the narrative for any of: "valve cover" | "valve covers" | "cam cover" | "camshaft cover" | "rocker cover" -> 1,310 complaints. Step 2 - failure proximity. The phrase must sit within 100 characters of: leak | gasket | seep | drip | oil | crack | warp | fail | replac | smoke | burn | fire | seal | weep | defect | melt -> 1,200. Step 3 - passenger vehicles only. 5 complaints filed under motorcycle, bus and heavy-truck makes (Kawasaki, Suzuki, Yamaha, Freightliner, Blue Bird) were removed. -> 1,195. Step 4 - the access-only exclusion. 23 complaints were removed where every mention of the cover sits in a window naming internal engine work (timing chain, lifter, camshaft, pushrod, rocker arm, VTC actuator, head gasket, cylinder head) and no oil, gasket, leak, smoke or fire word appears near it. On those cars the valve cover is the thing a technician removes to reach something else, not the thing that failed. -> 1,172 complaints. -------------------------------------------------------------------------- 3. SAMPLE READING (30 narratives, seed 20260918) — AND THE NOISE WE KEPT -------------------------------------------------------------------------- Thirty complaints were drawn at random from the step-3 set and read in full. 24 of 30 genuine valve cover or valve cover gasket problems 6 of 30 the cover appears as an access point or as one line inside a larger engine failure: a broken timing chain guide found "after pulling the valve cover", a failed lifter, a VTC actuator rattle, an engine replacement where the gasket was one item on a list Raw false-positive rate: 6 of 30, 20%. Step 4 removes only one of those six, because the other five do mention oil or a gasket somewhere near the cover. We tried a stricter filter that required an oil, gasket or leak word beside every mention. It removed two of the six false positives — and three true positives with them, including a Buick Encore fire and a Chevrolet Trax whose owner had the cover replaced four times. We kept the broader set and are publishing the noise level instead of hiding it behind a cleaner-looking query. READ EVERY COUNT BELOW WITH A CEILING OF ROUGHLY ONE IN SIX. They are counts of narratives that name a valve cover in a failure context, not a census of gasket failures. -------------------------------------------------------------------------- 4. THE FIRE FINDING, AND ITS LIMIT -------------------------------------------------------------------------- The fire flag is NHTSA's own field, not our regex. Fire flag in this set 83 of 1,172 = 7.08% All-complaint baseline 1.82% That is 3.9 times the baseline, and it is the strongest single number in this folder. Twenty of the 83 fire-flagged narratives were read in full: 16 of 20 oil from the cover reached something hot, or the fire started at the cover, or the owner is describing the GM 3.8L valve cover recall directly 4 of 20 the valve cover MELTED in a fire that started elsewhere — the cover is a consequence in the story, not a cause So the direction of the finding survives reading, and its size should be discounted by roughly a fifth. The article says so. The fire-flagged set is also concentrated: 15 Pontiac Grand Prix, 6 Chevrolet Impala, 5 Buick LeSabre, 5 Buick Regal, 2 Chevrolet Monte Carlo — the GM 3.8L V6 family covered by recall 15V-701 and its three predecessors. Published in valve-cover-fire-flagged-complaints.csv. -------------------------------------------------------------------------- 5. MILEAGE -------------------------------------------------------------------------- Reported mileage used only where above 0 and below 400,000. 661 of the 1,172 complaints carry a usable mileage. Median 98,000 Q1 68,000 Q3 130,000 At or under 36,000 miles: 61 of 661 Not a service life: complaints record failures, never survivals. -------------------------------------------------------------------------- 6. MANUFACTURER COMMUNICATIONS -------------------------------------------------------------------------- FTS5 search of communication summaries for "valve cover" returned 110 distinct NHTSA document ids. 48 describe a leak or oil loss, 11 name a PCV valve, crankcase ventilation or a PCV baffle, and 1 names a stop sale. Published by make in valve-cover-bulletins-by-make.csv. -------------------------------------------------------------------------- 7. RECALLS -------------------------------------------------------------------------- Only three campaigns in the file name a valve cover, a cam cover or a rocker cover in the defect text, and one of them is a heavy-truck air valve. The two that matter are enormous: 15V-701 GM, 26 October 2015, 1,003,556 vehicles. 1997-2004 Buick Regal and Pontiac Grand Prix, 1998-1999 Chevrolet Lumina and Oldsmobile Intrigue, 2000-2004 Chevrolet Impala, 1998-2004 Monte Carlo. An improved remedy for vehicles "previously repaired under recalls 08V-118, 09V-116, and 15V-201", because those remedies "did not adequately remove the safety risk". Owners advised to park outside. 14V-415 Hyundai, 8 July 2014, 61,122 vehicles. 2007-2012 Veracruz, oil leaking from the front cylinder bank valve cover gasket onto the alternator. Re-recalled as 20V-705 on 13 November 2020 (60,097 vehicles). 15V-201 GM, 6 April 2015, 330,742 vehicles — the predecessor named inside 15V-701. Campaign text read from the flat file and re-read through NHTSA's public campaign API in September 2026. -------------------------------------------------------------------------- 8. FILES IN THIS FOLDER -------------------------------------------------------------------------- valve-cover-complaints-by-context.csv What the 1,172 narratives describe, with counts, shares, and the NHTSA fire / towed / crash flags against the all-complaint baseline. valve-cover-complaints-top-models.csv The 50 most-reported models with fire-flag counts, median reported mileage, the share naming a recall, and how often the alternator and the spark plug well appear. valve-cover-fire-flagged-complaints.csv The 83 fire-flagged complaints by make, model and model year. valve-cover-bulletins-by-make.csv The 110 valve cover bulletins by make, with leak and PCV counts. No personal data is published: no VIN, no city, no dealer name, no free text. -------------------------------------------------------------------------- 9. LIMITATIONS -------------------------------------------------------------------------- - Complaints are self-reported, unverified, and have no denominator. - Roughly one narrative in six names the valve cover without the cover being the failed part. See section 3. - The fire flag does not distinguish a fire caused by the cover from a fire that melted it. See section 4. - Filing is driven by publicity: 45.1% of these narratives name a recall or campaign, and the largest clusters are recalled vehicles. - Owner-written dollar figures mix quotes with paid invoices and are not published as a cost distribution in this folder for that reason. ------------------------------------------------------------------------------ CASE SENSITIVITY — SECTION ADDED 19 SEPTEMBER 2026 ------------------------------------------------------------------------------ NHTSA stores complaint narratives in ALL CAPS for records filed through roughly 2020 and in mixed case for records filed from 2021 on. A case-SENSITIVE search of the narrative field loses about 92% of the complaints filed after 2020, so this is the single largest way an analysis of this database can go wrong while still looking finished. Every query behind this file is case-insensitive. SQLite's LIKE and its FTS5 full-text index both ignore case by default, and no query in this work uses GLOB or any case-sensitive comparison. This was not taken on trust. On 19 September 2026 every published folder was re-checked against the same database, comparing the figures here with what a case-sensitive query would have returned. The result: the complaint set published here (1,310) is LARGER than the absolute ceiling a case-sensitive search of the same terms could have returned (667 narratives). A case-sensitive query cannot produce a bigger set than itself, so the bug is ruled out for this folder by arithmetic, not by assurance.