PulsCar analysis of NHTSA public data, September 2026 Topic: turbocharger failures, coverage documents and recalls, 2015-2026 License: CC BY 4.0. Attribution: "PulsCar analysis of NHTSA public data, September 2026". SOURCE FILES ------------ NHTSA Office of Defects Investigation public flat files, downloaded September 2026 and loaded into a local SQLite database: FLAT_CMPL.txt - owner complaints. 777,040 rows, received 1 January 2015 - 8 September 2026. Field definitions: https://static.nhtsa.gov/odi/ffdd/cmpl/CMPL.txt FLAT_RCL.txt - recall campaigns. 245,578 make/model/year rows. FLAT_TSBS.txt - manufacturer communications (service bulletins, dealer messages, warranty bulletins, special coverage and customer satisfaction programs). Complaint narratives are self-reported. They are NOT a failure rate: no denominator, no verification of the diagnosis, and reporting volume is influenced by forums, class actions and news coverage. Counts describe what owners wrote to NHTSA. COMPLAINT QUERY AND FILTERING ----------------------------- Step 1, full-text search: SELECT c.* FROM complaints c JOIN complaints_fts f ON f.rowid = c.rowid WHERE complaints_fts MATCH 'turbo OR turbocharger OR "turbo charger" OR wastegate'; -> 4,271 complaints. Step 2, keep only reports where the turbo word is tied to a failure. For every match of \bTURBO\s?CHARGER\b | \bTURBOCHARGER\b | \bTURBO\b | \bWASTE\s?GATE\b | \bWASTEGATE\b the 70 characters on each side were tested for \b(FAIL|FAILED|FAILING|FAILURE|WENT OUT|BLEW|BLOWN|SEIZ|LOCKED UP|SHATTER|BAD|DEFECTIVE| REPLACE|REPLACED|REPLACEMENT|LEAK|LEAKING|WHIN|WHISTL|SCREECH|SMOK|BROKE|BROKEN|CRACK| SHAFT|BEARING|OIL STARV|NEEDS? TO BE|WENT BAD|GAVE OUT|DIED)\b Two kinds of match were skipped before that test: (a) engine badges - the 16 characters before the word end in a displacement, e.g. "2.0L ", "1.5 L", "LITER", "LITRE"; (b) model names - the last word of the complaint's own MODEL field appears in those 16 characters, which removes "911 TURBO", "SONATA TURBO" and similar trim names. -> 2,003 complaints, received 1 January 2015 - 8 September 2026. This is the published set. SAMPLE READING (false positives) -------------------------------- 30 of the 2,003 drawn at random (Python random.seed(23), random.sample) and read in full: 25 - the turbocharger or its plumbing is the subject. 3 - a real turbo failure described inside a complaint mainly about something else. 2 - false positives: the word is a trim name that the model-field rule did not catch. False-positive rate found: 2 of 30 (6.7%). An earlier, looser filter (160-character window, no badge or model-name exclusion) kept 3,018 complaints and read 8 of 30 as false positives; it was discarded. PRICES: DELIBERATELY NOT PUBLISHED ---------------------------------- Dollar figures were extracted the same way as on other PulsCar pages: the first amount between $200 and $12,000 whose surrounding 240-character window contains "TURBO" and a cost word (COST|QUOTE|ESTIMATE|PAID|BILL|CHARGE|REPAIR|FIX|REPLACE), excluding windows naming a transmission, CVT, head gasket, suspension, steering rack, purchase, loan or lease. That produced 144 figures with a median of $3,000. Twenty were then read in full and only about half were the turbocharger repair - the rest were engine replacements, diagnostic fees, catalytic converters and unrelated repairs described in the same paragraph. The median was therefore NOT published and no owner-cost CSV is included for this topic. The price ranges in the article come from the estimator and shop-survey sources named in its footer, not from complaint text. CONTEXT CATEGORIES (turbo-complaints-by-context.csv) ---------------------------------------------------- Narratives upper-cased and whitespace-normalized; categories are NOT exclusive. recall/bulletin named (RECALL|TSB|SERVICE BULLETIN|CAMPAIGN) reduced power (REDUCED (ENGINE )?POWER|LIMP|LOSS OF POWER|LOST POWER|NO POWER| WOULD NOT ACCELERAT|FAILED TO ACCELERAT|UNABLE TO ACCELERAT) warranty (WARRANT|GOODWILL|EXTENDED COVERAGE|CUSTOMER SATISFACTION PROGRAM| SERVICE CAMPAIGN|SPECIAL COVERAGE) repeat failure (AGAIN|SECOND TIME|TWICE|3RD|THIRD TIME|MULTIPLE TIMES|4 TIMES|FOUR TIMES) stalled (STALL|SHUT (OFF|DOWN)|CUT OFF|SHUT ITSELF) dollar figure \$\s?[\d,]{3,} engine damage (ENGINE (WAS |)REPLAC|NEW ENGINE|LONG BLOCK|ENGINE FAILURE| ENGINE DAMAGE|SEIZED ENGINE|BLOWN ENGINE) oil (OIL LINE|OIL FEED|OIL SUPPLY|OIL RETURN|OIL LEAK|OIL STARV| LEAKING OIL|LOW OIL|OIL CONSUM) turbo replaced (TURBO\w*\s?(CHARGER)?\s+(WAS |WERE |)REPLAC|REPLACED THE TURBO| NEW TURBO|TURBO REPLACEMENT) fire (FIRE|BURNING SMELL|CAUGHT FIRE|FLAME) no warning light (NO (WARNING |CHECK ENGINE |)LIGHT|WITHOUT (ANY |)WARNING|NO WARNING) smoke SMOK wastegate/actuator (WASTE ?GATE|ACTUATOR|SOLENOID) coolant (COOLANT (LEAK|LINE|PIPE|FEED)|ANTIFREEZE) boost code (P0299|P0234|P2563|P003A|UNDER ?BOOST|OVER ?BOOST) whine (WHIN|WHISTL|SCREECH|SIREN|SQUEAL) Baselines over all 777,040 complaints: towed 5.11%, crash 4.37%, fire 1.82%. Turbo set: towed 147 (7.3%), fire 38 (1.9%), crash 16 (0.8%), injuries 22, deaths 0. MILEAGE ------- Values of 0, blank or 400,000 and above dropped. 993 of 2,003 rows remain. median 74,000 at or under 36,000: 142 (14.3%) at or under 60,000: 342 (34.4%) at or under 100,000: 760 (76.5%) Per-model medians in turbo-complaints-by-make-and-model.csv use the same rule; the mileage_values column gives the n behind each median. Makes with 10 or more complaints and make+model pairs with 12 or more are published. RECALLS (turbo-recall-campaigns.csv) ------------------------------------ SELECT ... FROM recalls WHERE lower(defect||' '||coalesce(remedy,'')||' '||coalesce(consequence,'')) LIKE '%turbocharg%' OR lower(component) LIKE '%turbo%'; Rows are per make/model/year and were collapsed to distinct campaign numbers. 43 distinct campaigns, 474,112 potentially affected units. names fire: \bfires?\b | \bfire\. | catch fire | vehicle fire | compartment fire over defect + consequence -> 36 campaigns, 392,273 units. names a turbo oil line: /oil (supply|feed|line|pipe|return|strainer)/i over defect -> 14 campaigns, 108,319 units. A light_vehicle column marks campaigns whose makes are cars, SUVs, pickups or vans. Excluded from that subset as non-light-vehicle: Triumph motorcycles, John Deere, Orion, New Flyer, Nova Bus, Volvo Bus, ElDorado, Mack, International/Navistar, Autocar, Cummins Westport, ProCharger, Roush and Ford Performance equipment recalls. Light-vehicle subset: 28 campaigns, 457,505 vehicles; 25 name fire (384,931 vehicles); 11 name a turbocharger oil line, feed pipe or strainer (102,668 vehicles). MANUFACTURER COMMUNICATIONS (turbo-warranty-and-coverage-documents.csv) ----------------------------------------------------------------------- SELECT t.* FROM tsbs t JOIN tsbs_fts f ON f.rowid = t.rowid WHERE tsbs_fts MATCH 'turbocharger OR turbo OR wastegate'; Rows are per make/model/year; collapsed to distinct (make, bulletin_no) pairs -> 1,693. Coverage documents are the subset whose summary contains TURBOCHARG or TURBO CHARGER and one of: WARRANTY EXTENSION, EXTENDED WARRANTY, EXTENDS/EXTENDING THE WARRANTY, LIMITED WARRANTY EXTENSION, SPECIAL COVERAGE, SPECIAL POLICY, CUSTOMER SATISFACTION PROGRAM. -> 41 documents, published in the CSV with make, number, NHTSA id, date, models and years. The summary column is NHTSA's abstract truncated to 220 characters, not the bulletin text. PRIMARY DOCUMENT OPENED (September 2026) ---------------------------------------- MINI USA Service Information M01 01 18, February 2018, "N18 Engine Turbocharger: Limited Warranty Extension to 10 Years/120,000 Miles" - NHTSA document MC-10135055, five pages, including the reimbursement section for qualifying prior customer-pay repairs. The other coverage documents listed in the article were read from the federal file's summaries; their PDFs were not served for those document IDs at the time of writing, and the article says so by naming them as bulletins rather than quoting them. LIMITATIONS ----------- 1. Complaint counts are not failure rates and must not be read as model reliability. 2. "Turbo" is both a component and a badge; the exclusions above reduce that confusion but a 6.7% false-positive rate remained in the sample read. 3. The fire share in the recall file describes recall campaigns, not the probability that any individual turbocharged car catches fire. 4. Coverage documents change: extensions are revised, expanded and superseded, and the CSV is a snapshot of the federal file in September 2026. Verify by VIN with a dealer. ------------------------------------------------------------------------------ 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, in two layers: * the complaint set is built with an FTS5 MATCH full-text search, which folds case before matching, so it cannot miss a mixed-case record; and * where the narrative is then tested with patterns, the narrative is upper-cased and whitespace-normalized first, so the patterns compare upper against upper. 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. Re-run here: step 1 is MATCH 'turbo OR turbocharger OR "turbo charger" OR wastegate'; the recall query wraps its text in lower() before comparing.