Short version. With only the fuel fault fixed, Mabel is likely to finish the round trip, but not without trouble. The realistic case expects about 3 unplanned stops, a 70% chance of at least one stop that costs a day or more, and roughly a 1 in 8 chance of a trip-breaking failure. Two items carry most of the risk: the cooling system, with an engine-out outcome if it goes, and the front ball joints, with a crash outcome if they go. Most of the open items can be dealt with before you leave. The recommended pre-trip set below cuts the realistic chance of a trip-breaking failure from to , and mechanically caused crash or fire risk from to . A winter return is the largest single source of crash risk in the model.
7,800 km: 2,900 out, 2,000 around Santa Cruz, 2,900 back in winter. Change the trip or the pre-trip work below and every number updates.
An interruption is any unplanned stop of 15 minutes or more. Long delay means a day or more, usually with a shop or parts shipping involved. Major means a week or more, engine, transaxle or structural work, or giving up on the van. Crash or fire counts only events where a mechanical failure is the cause. For scale, the average US vehicle and driver have about a 0.9% chance of a police-reported crash over the same distance, from all causes. Cost and delay percentiles come from 12,000 simulated trips per scenario.
Outbound legs carry multipliers for sustained grades and late-season heat in the Great Basin and Central Valley. Local driving carries more starts per kilometre. See each system for its multipliers.
Bars show the realistic case, split by how bad the outcome is. The thin line runs from the optimistic to the pessimistic estimate. Sorted by the realistic chance of a long delay or worse. Hover a row for numbers.
Read a point as "a y% chance the trip's unplanned repair, tow and recovery bill exceeds x". It excludes lodging, lost time and the van's own value. Log scale.
Each row: chance of any interruption from that system over the trip (optimistic · realistic · pessimistic). Open a row for history, cause, the severity mix, costs and the reasoning behind each scenario's rate.
Brake hoses, vacuum hoses, tank grommets and vapour hoses all failed from age, not wear. The same rubber is still in the coolant hoses, the fuel hoses between tank and pump, CV boots and engine mounts that no one has replaced. That is why cooling and fuel keep non-trivial rates even after their headline fixes.
Horn, right signal, reverse switch terminal, the cluster foil's open pre-exciter, the zamak lock tumblers. Every electrical fault found so far was a connection, not a component. Untouched connectors are the base of the electrical and ignition rates.
A previous owner disconnected the temperature gauge, the oil-pressure switch and the coolant-level sensor. Two of three are back. Working warnings do not prevent failures, but they move outcomes from the major tier to the long or short tier, which the severity mixes reflect.
5%-moisture brake fluid, phosphate coolant, a clogged filter with no history, suspected 15W-40 oil. All renewed. What has not been renewed is the ignition refresh, which has no record.
Model. Each system fails as a Poisson process in distance. Its rate λ (interruptions per 1,000 km) is scaled by segment multipliers for the outbound, local and return legs. Over the trip the expected count is μ = λ × effective distance, and the chance of at least one interruption is 1 − e−μ. Each interruption falls into a severity tier with the shares shown, and causes a crash or fire with probability c. Systems are treated as independent, so trip-level chances pool as 1 − e−Σμ. Costs and delays per tier are triangular distributions, simulated per trip.
Base rates. There is no published breakdown dataset for 43-year-old vans, so the overall rate is anchored by extrapolation and then checked against Mabel's own record.
US roadside data: about 69 million breakdowns a year, 61% of them in the 44% of vehicles that are ten or more years old. That works out to roughly 0.016 breakdowns per 1,000 km for older vehicles, counting lockouts and flats. ADAC uses a stricter count of member assistance calls, and finds 3.1% of ten-year-old cars break down in a year (0.0026 per 1,000 km at typical German mileage). Its age curve rises by about 1.6–1.8× every five years. Carrying that from ten to 43 years gives a factor of about 20, so a range of 0.06 to 0.35 per 1,000 km depending on the definition. This assessment counts every 15-minute stop, which is broader than either dataset, so the prior is set at a mean of 0.25 per 1,000 km, Gamma(α = 2, β = 8), deliberately wide.
Mabel's record. Under this ownership she has covered an estimated 800 km. That is an assumption, because there is no odometer reading after 14 July. She has had two fuel-delivery events in that time (Calmar, 5 October) and no other on-road interruptions; the 3 September stall was a hose knocked off in the garage. The non-fuel posterior is Gamma(2, 8.8): mean 0.23, 90% interval 0.04–0.54 per 1,000 km. Including the fuel events gives Gamma(4, 8.8), mean 0.45, which is why the fuel premise matters so much.
Bottom-up check. The realistic system rates excluding fuel and winter sum to 0.31 per 1,000 km, which sits near the 74th percentile of the posterior. That is somewhat pessimistic against a generic old vehicle. The known red items (ball joints, starter, bearings, unresolved idle, undiagnosed 2nd-gear jerk) justify it: a generic prior does not know about them. With 800 km of history the record is too short to override the prior for any single system, so per-system rates come from mechanism and condition, with fleet shares as a check. ADAC attributes 45% of breakdowns to the battery, 22% to engine electronics, 10% to starter, alternator and wiring, and 9% to tires. Mabel's battery is strong and recently proven, so that share moves toward ignition, cooling and the open chassis items.
Crash shares. In NHTSA's crash causation survey, the vehicle was the critical reason in about 2% of crashes. Within that group, tires were 35%, brakes 22%, and steering, suspension, transmission and engine 3%. NHTSA's tire-ageing review found that 77% of tire claims in one insurer's data came from five hot states, and 84% of those involved tires older than six years. Those figures set the order of the per-event crash probabilities: tires and ball joints high, brakes moderate now that the hydraulics are new, engine faults low. Values of c are judgment, informed by these shares, not measured.
Costs. Ranges come from Edmonton and California shop rates and current parts prices, converted to CAD. Reference points: a Vanagon head-gasket job at about US$1,550–1,900; a GoWesty rebuilt engine at US$4,999 plus a US$1,500 core. Abandonment costs cover recovery and shipping or replacement, not the van's own value.
Limits. Independence understates clustering: a weak cooling system and long grades fail together. The pessimistic scenario is the hedge for that. Rates for ball joints, frame and pop-top are judgment with no fleet data behind them. Driving behaviour (speed, how quickly you stop when the alarm sounds, whether you carry spares) moves results as much as any single part.