PulsCar analysis of NHTSA public data, September 2026 — "Car squealing when starting" License: CC BY 4.0. Credit "PulsCar analysis of NHTSA public data" with a link to https://www.pulscar.io/blog/car-squealing-when-starting SOURCE FILES (downloaded from https://static.nhtsa.gov/odi/ffdd/ in September 2026) - Complaints: COMPLAINTS_RECEIVED_2015-2019.zip, 2020-2024.zip, 2025-2026.zip (cmpl). Vehicle complaints only, one row per ODINO. Received dates 1 January 2015 to 8 September 2026. Unique complaints in the working table: 777,040. - Recalls: FLAT_RCL.zip (rcl), one row per make/model/year per campaign; counts are DISTINCT campaign. - Manufacturer communications (TSBs, SSMs, customer satisfaction programs, STAR cases): TSBS_RECEIVED_2015-2019, 2020-2024, 2025-2026 (tsbs). Counts are DISTINCT NHTSA document ID. - Investigations: FLAT_INV.zip. Documents quoted in the article were opened as PDF from https://static.nhtsa.gov/odi/tsbs/ and read with pdftotext; recall wording comes from the recall flat file (same text as the api.nhtsa.gov record). 1. SQUEAL COMPLAINTS (squeal_complaints_read.csv, squeal_sources_summary.csv) Step 1 — every complaint whose free text contains the substring "squeal" (case-insensitive): 1,282 of 777,040. Step 2 — candidate set: those 1,282 filtered with a Python regex for engine accessory-drive words (serpentine | drive belt | fan belt | belt (not followed by buckle/retract/light/warning/latch) | tensioner | idler | pulley | alternator | water pump | power steering pump | p/s pump | a/c compressor | compressor | generator), excluding texts whose only "belt" is a seat belt: 102 candidates. Step 3 — we READ ALL 102 in full and coded each by hand: belt_system_squeal Y/N (N = brakes, tires, suspension, audio, or the word appears in an unrelated context; 18 of 102 were N, a 17.6% false-positive rate for the regex); reported_source = what the owner or the shop said was squealing, in the owner's words, or "unresolved" when the report never names it; startup_or_cold_context = the squeal is described at engine start, cold start or a cold morning; part_replaced_without_fix = a part the report says was replaced and the noise came back; shop_could_not_find_or_called_normal; outcome_as_reported = loss of steering assist, loss of brake assist, fire/smoke, stall, towed (from the report text; the towed/crash/fire flags are NHTSA's own fields and are kept separately). The sources summary groups the 84 confirmed reports (grouping rules are in the reported_source column; each report has exactly one group). Mileage: 56 of the 84 give a mileage between 1 and 399,999; median 71,250, quartiles 42,773 and 111,125. What this is NOT: complaints are self-reported, unverified, and not a failure rate. 84 reports in 11.7 years across all makes is a small set; we rank no make or model from it. Owners who filed with NHTSA skew toward safety outcomes (loss of steering, brakes, fire), so mild belt squeals are under-represented. 2. WET-BELT SLIP (wet_belt_slip_complaints.csv) Query: text contains "belt" AND (rain | wet | puddle | splash), then a regex requiring an engine belt word (serpentine | drive belt | accessory belt | fan belt | alternator belt | belt slip/squeal/ squeak/chirp/slipping/slips) AND a wet word (rain | raining | rainy | wet | puddle(s) | splash* | car wash | standing water), excluding seat belt: 47 candidates, all read; 26 describe a belt that slips, squeals or comes off when wet. 13 of the 26 are 1997-2002 Chrysler, Dodge and Plymouth minivans; 4 are 2013-2015 Ford Explorer; 4 are 2005-2009 Kia (Rondo, Rio); 2 quote Mazda recall 14V-173. The other 21 candidates mention belts and water in unrelated ways (water pump, body leaks, a sentence that says "no belt noise"). 3. BULLETINS (squeal_bulletins.csv) FTS5 query on the summary field: (squeal OR squeals OR squealing OR chirp OR chirping OR squeak) AND (belt OR tensioner OR pulley). 67 distinct documents. Each was classified from its summary and, for 13 of them, from the PDF: accessory_drive_belt = Y when the document is about the engine accessory belt, its tensioner, pulleys or a driven accessory; N for timing belt or timing chain chirps, electric steering-gear belts, a mirror trim "belt molding", and PCM flashes that mention a belt in passing. Result: 47 accessory-drive documents, 20 not. A document's presence says a manufacturer wrote about the condition; it does not measure how many vehicles had it. 4. RECALLS (accessory_drive_recalls.csv) Recalls whose defect text contains serpentine | drive belt | belt tensioner | idler pulley | accessory belt | accessory drive, or (added by hand) crankshaft pulley 11V-539 and Winnebago 20V-474, read one by one. Seat belt tensioner recalls (16V-209, 16V-518, 20V-638, 22V-515, 25V-253) were removed. 20 campaigns remain; 13 cover cars, SUVs and one motorhome and 7 cover buses, engines or a recreational vehicle. "vehicles" is NHTSA's potentially-affected count. 5. OTHER COUNTS QUOTED IN THE ARTICLE - GM trucks, "vacuum pump" AND "belt" in text, makes Chevrolet/GMC/Cadillac: 40 complaints, 7 towed, 5 crash flags (regex only, not hand-read beyond the 4 that appear in set 1). - Kia Telluride and Hyundai Palisade complaints containing "squeal": 5, of which 4 describe a cold-start squeal below about 30-40 F that the dealer could not attribute to the belt. - Alternator recalls quoted: 14V-634 (435,081 vehicles), 17V-435 (442,214), 17V-741 (17,116), from the recall file. LIMITATIONS Text search misses complaints that describe the noise without the word "squeal" (screech, whine, chirp were not part of set 1). Complaint texts are truncated by NHTSA in some older records. Model-year counts are not normalized by vehicles on the road.