PULSCAR ANALYSIS - P0449 AND P0457 (EVAP VENT VALVE CIRCUIT, LEAK DETECTED / FUEL CAP) IN NHTSA'S PUBLIC FILES PulsCar analysis of NHTSA public data, October 2026. License: CC BY 4.0 Attribution: "PulsCar analysis of NHTSA public data, October 2026." Published alongside https://www.pulscar.io/blog/p0449-code ------------------------------------------------------------------------------ Prepared October 2026. Source data: NHTSA public flat files downloaded on 7 October 2026 (complaints received 1 January 2015 to 4 October 2026 - the latest received date in the files - 782,908 records; manufacturer communications; recalls since 2010; investigations) and documents downloaded one at a time from static.nhtsa.gov. 1. Complaints (3,501, all read in full or in excerpts around the matching words, two readers) First search: every complaint whose narrative names P0449 or P0457 by any spelling we searched (a letter O for zero, a space, hyphen or period after the P, the suffix 00, or a bare 449/457 after "code", "DTC", "error", "CEL", "coded" or "fault"), names the vent valve or solenoid in owners' words, names the gas cap or filler neck with a warning, light, code, emission or inspection context, names the EVAP system with a leak, code, light or inspection context, names a sibling EVAP code (P0440 P0441 P0442 P0443 P0446 P0447 P0448 P0455 P0456 P0498 P0499), or mentions spiders with EVAP or fuel wording: 2,564 complaints. Completeness audit: before the second batch was coded, an independent completeness search of the same file for what the first search missed proposed 937 complaints (found_by = independent audit in the amounts file and audit in the complaints file); all 937 were added, their relevance decided by the unchanged codebook. Coding: the codebook (codebook.txt) was written, dated and hashed before any complaint was read; an addendum for the second batch was hashed before that batch was read. Two independent classification passes coded all 3,501 separately, each from the codebook only, without seeing the other's codes. One category (N not an EVAP warning event on this vehicle; G gas cap; V vent valve, vent solenoid or vent line; F filler neck, filler pipe, fuel inlet or tank; H other EVAP hardware such as purge valve or canister; S software or campaign procedure only; O another cause; C checked and cleared; U no diagnosis or repair stated) and four flags (cap_tried = the gas cap was tightened, replaced or checked; recall = a recall, campaign, special coverage or bulletin is tied to the event; inspection = a failed or blocked emissions test or inspection; returned = the warning came back or the repair did not hold). Agreement over all 3,501: category 3,202 (91.5%), Cohen's kappa 0.890; cap_tried 0.922; recall 0.901; inspection 0.958; returned 0.891. Relevance (N versus not N): 3,399 of 3,501 agreed. Every disagreement (537) was read in full and settled under written rules R1-R13 (codebook.txt); one consistency sweep after the first batch found that the coders had applied one reading of rule R4 to 132 agreed rows opposite to the one adopted, and the rule was revised and its earlier decisions re-made; seven agreed codes were changed in consistency passes. The resolution and rule_applied columns record all of it. Final: not relevant (N) 998; of the 2,503 relevant: U 1,051, H 755, F 235, V 146, G 106, O 101, C 60, S 49. First search 1,650 relevant, audit batch 853 relevant. Flags among the relevant: cap_tried 409, recall 757, inspection 86, returned 781. 32 complaints print P0449 (18) or P0457 (14) themselves. Makes among the 2,503: Ford 942, Chevrolet 231, Jeep 156, Dodge 116, Volkswagen 109, Toyota 103, Hyundai 93, Honda 92, Kia 87, Chrysler 77. Models: Focus 396, Fusion 170, Escape 161, Edge 73, Grand Cherokee 50, Sonata 39. Median model year 2014. 1,382 complaints give a mileage from 1,000 to 400,000; median 80,000. By year received: 2015 187, 2016 180, 2017 173, 2018 190, 2019 245, 2020 200, 2021 172, 2022 199, 2023 260, 2024 258, 2025 260, 2026 179 (2026 to 4 October). In NHTSA's own fields: fire 6, crash 20, injury 21, towed 55. Mention counts (by script, not coded; case-insensitive, in the relevant narratives): spider 6; tank deformed, crushed, swollen or collapsed 101; nozzle clicking off, spit-back or fuel spewing 36; VIN not in a recall or similar words about 200 (the script gave 203; the patterns were not saved and independent recounts give 159 to 211); refusal or not-covered wording 150; "known issue" or "common problem" 101; capless filler or fuel inlet 95; smoke test 41; stall or no-start around a fill-up 111. Recall numbers named in the relevant narratives (by script): 18V-735 92, 19V-515 35, 15V-793 31. Figures count complaints, not vehicles. 2. Owner-reported amounts (229 figures in 189 complaints) Every dollar figure (and "dollars" or "bucks" amounts) in the relevant complaints was extracted by script (561 candidates, 445 in the first batch and 116 in the audit batch, including mileage written as "NNk", which was excluded) and decided one by one under one rule: keep every figure the narrative ties to the diagnosis or repair of the EVAP event - a diagnosis, inspection or smoke-test fee, a gas cap, a vent part, a purge valve, a canister or its filter, a filler neck or pipe, a fuel tank or tank module, or an EVAP repair whose part is not named - whether paid, billed or quoted; exclude mileages, vehicle prices, payments and down payments, trade-in and loan figures, rentals, towing, repairs of other systems, restatements of a figure already counted, other owners' costs and totals that mix unrelated items (332 exclusions in our working files, each with a reason class). 104 figures are quotes, 62 bills, 57 described as paid, 6 not stated. Groups: diagnosis 44 (median $189.50); gas cap 7 (median $130); vent valve, vent line or canister vent 16 (median $479.04); purge valve, pump or solenoid 25 (median $400); canister or its filter 28 (median $700); filler neck, pipe or fuel inlet 11 (median $500); fuel tank or tank module 35 (median $1,400); EVAP repair, part not specified 48 (median $900); several parts together 15 (median $1,600). 3. Manufacturer documents (319 distinct documents read in full) First search: every manufacturer communication in NHTSA's index (5.3 million index rows, one per model-year, 188,586 distinct communications) whose summary names P0449 or P0457 (by any spelling), a vent valve or solenoid, a canister vent, a vent line or hose, a gas cap, filler cap, filler neck or fuel fill inlet with a warning context, or an EVAP, fuel vapor, vapor canister or charcoal canister wording with a leak, code or light context: 1,434 NHTSA ids. These were split into a core tier (the code, vent wording or cap wording; 379 ids in 257 near-duplicate clusters of 5-word shingles with Jaccard 0.6 or more) and a tier whose summaries only name EVAP leak codes or EVAP wording inside long code lists (530 ids). Downloaded: one representative of each core cluster plus every id whose summary names P0449 or P0457 (268 ids, 264 with a PDF). The EVAP code-list tier and the other cluster members were not read: 173 of them were opened and scanned by script and none prints P0449 or P0457; the remaining ones were not downloaded because the shared NHTSA download queue made it impractical - this is a limit of the study. Completeness audit: an independent search reviewed 1,130 further candidates and proposed 76 ids (172 index rows); 75 had a PDF (one had none on NHTSA) and were read in full. Recalls: the filings of 12 campaigns on the EVAP vent path, filler or tank (08V156, 11V134, 12V063, 14V114, 14V464, 15V793, 18V735, 18V769, 19V515, 20V347, 22V673, 23V071): 61 recall documents read (Part 573 reports, owner letters, dealer letters, acknowledgements, completion reports; other recall papers in those folders, such as quarterly reports and miscellaneous correspondence, were not read). Investigations: PE13-016, DP14-002 and DP17-002, read from NHTSA's investigation flat file (no PDF). 325 files were read: 186 first-search texts, 75 audit texts and 61 recall documents, plus 3 investigation records; six are byte-identical copies (md5), so 319 distinct. 19 distinct documents print P0449 and 13 print P0457 in their own text (21 and 15 file ids; the rest are copies). No recall document prints either code. Pages without a text layer were rendered and read as images (the Mazda CX-30 recall papers, two Mazda and Suzuki letters and two scrambled-layer recall letters). Every verbatim quote in the article was checked by script against the document or complaint text; quotes from image-only pages were checked against our transcriptions. Coverage status in September 2026 is our computation from the terms printed in each document, not a statement by the manufacturer. No document prints an individual car's in-service date. Limits: complaints over-represent failures, refusals and expensive repairs; owners report diagnoses as they understood them; the Ford canister purge valve recalls produce a large share of the complaints (396 Focus complaints) and are a different fault from P0449 and P0457, included because owners' warnings are the same; a narrative that cites only a campaign number is not counted; searches of NHTSA's index cannot see documents whose summary names none of the search terms. Nothing here is a failure rate.