What the numbers know—and what they don’t.

Reading the disclosure

Evidence grades

The badge grades the specific step, not the confidence implied by its numerical precision.

• EstablishedDirectly applicable research or a verified technical requirement supporting that claim.
• DerivedExplicit arithmetic or a derivation with stated assumptions. Correct arithmetic alone does not validate a model.
• HeuristicA fitted relationship, planning convention or rule of thumb without direct validation.

Model 01

This is not a measured optimum or a compatibility certificate. Berto’s tire-drop tables inspired the curve, but he did not publish or validate this equation. Non-road use extrapolates beyond that context. Berto’s tables and inflation article are context, not coefficient validation.

1. Load, width and pressure

Units are kg, mm and bar. Reference rim width is 21 mm except MTB, which uses 30 mm. The width adjustment and equal reference load are unvalidated conventions. Enter labeled widths, not a measured width that has already incorporated the rim adjustment. Legacy MTB inputs below 10 mean inches (1.5–4, converted ×25.4); otherwise supported label widths are 15–100 mm.

2. Wheel offsets and setup

All offsets are heuristic—not measured axle loads or validated psi gaps. Pressure guidance provides context for per-wheel tuning. Fixed construction, surface and feel multipliers do not establish speed, rolling resistance or grip benefits.

Breakpoint-pressure discussion explains why more pressure is not always better; it does not validate these multipliers.

3. Estimated lower bounds and equipment limits

These are estimated tuning lower bounds, not proven safe minimums. Within 15% prompts a heuristic caution. Below the bound, raise to it only if it does not conflict with a hard equipment cap.

For non-hookless equipment, an entered maximum is also a hard cap. Check the lowest applicable tire or rim limit; absence of a supplied limit is not evidence of safety. A lower bound above the cap withholds that wheel, never forces a numeric answer. An independently valid wheel may remain visible, but do not ride on a partial plan. Equal bounds give one exact pressure.

Range endpoints round inward to hundredths of a bar. If no hundredth fits, display the exact constraint rather than an inverted interval. Primary bar display also stays within the interval. Rounded psi steps down if the nearest integer would exceed the cap. Bar constraints are authoritative. Small ±0.15 bar tuning changes are a convention and must stay inside equipment limits.

Software input limits: rider 30–250 kg, bike/gear 3–60 kg, internal rim 10–50 mm, optional maximum 0.5–12 bar. These limits do not prove model applicability.

Model 02

For general planning by healthy adult recreational/endurance cyclists—not children, clinical treatment, heat illness or an elite-racing prescription. Simple mode requires no FTP and never infers missing power, carbohydrate oxidation or sweat rate.

1. Duration and perceived effort

Linear interpolation of these anchors is a heuristic interpretation of broad duration guidance, not studied floors, ceilings or minimum needs. Under 45 minutes returns zero: usually no on-bike carbohydrate is needed, but small amounts can help short intense efforts. Exactly 45 minutes / steady gives 20 g/h. The generic curve is flat beyond 6 hours and never recommends above 90 g/h.

Simple defaults to steady effort and a 60 g/h practiced limit, with 90 selectable. Advanced allows an explicit practiced target of 1–120 g/h to replace the duration target, still capped by the selected 60/90/120 limit. Above 60 g/h, use a glucose–fructose blend and practise. Above 90 g/h is specialized trained-rider evidence with uncertain general applicability (Podlogar); 120 is an app bound, not a scientific upper limit. No performance multiplier or estimated burn-rate cap remains.

2. Optional advanced power and energy

Both FTP and average power (or ratio) must be supplied together or left blank. Average-power ratio is not normalized-power intensity factor. Mechanical energy is arithmetic; approximate dietary energy is an efficiency convention, without an individual ±10% guarantee. Neither output influences the carbohydrate or fluid plan.

3. Fluid and measurement

Fluid is a carrying assumption, not a schedule to force down. No lower floor, formula-based sweat prediction, absorption ceiling or confidence interval is claimed. Drink to thirst and individualize; food targets never compel fluid intake. In heat or with high losses, plan cooling and access to fluids and seek individualized advice. Sweat variability, measurement context and drinking-strategy review inform these cautions, not the 75% coefficient.

Use dry body mass and the same scale in comparable conditions; assume approximately 1 kg/L. Account for food and other non-sweat changes on longer rides. The result is condition-specific.

4. Sodium and signed fluid balance

The editable 600 mg/L default is a drink-mix planning convention, not a universal ACSM rule. Volume categories never infer sodium concentration. This is intake, not replacement of all sweat losses or prevention of hyponatremia. Without fluid input, sodium is unavailable.

The water-deficit scenario requires Advanced, measured sweat, numerical intake and body weight. It is not predicted total body-mass change: urine, food, metabolic mass and changing conditions are excluded. Positive values above 2% prompt caution without forcing fluids; negative values flag possible overdrinking/weight gain and are not clamped away. Fluid guidance does not validate this individual prediction.

5. Totals, packing and partial hours

Totals multiply unrounded hourly values by exact ride hours. For example, 2 hours / steady / practiced60 gives 52.5 g/h and 105 g total even when the headline rounds to 53. Each timeline segment uses its actual length, including the last partial hour; at most 24 rows.

Packing is deterministic serving arithmetic, not an optimum diet. Supplied carbohydrate and excess from rounded servings are shown separately. With no fluid plan, bottles, mix and refills are unavailable; bars/gels still cover the carbohydrate total. Refills count individual bottles replenished, not stops. Every total, packing step and timeline segment has calculation-linked working.

6. Supported inputs and restored state

Total duration is greater than zero and at most 24 h; hours and minutes are integers, minutes 0–59. Planned fluid is 0–1.2 L/h; measured sweat 0.1–5 L/h; sodium 0–2000 mg/L. Optional advanced FTP is 50–600 W, power 20–600 W, ratio 0.1–1.5 and weight 30–250 kg. Gel serving 1–100 g, drink/bar serving 1–150 g, bottle 250–1500 mL, cages integer 1–6. These are software limits, not evidence of universal applicability.

Known fields restore defaults, then saved settings, then URL values. Last duplicate query value wins; blank overrides saved values. Unknown keys are ignored and malformed inputs produce errors. Legacy power state without a mode opens Advanced with a migration notice; explicit Simple ignores inactive advanced values. Watts win conflicts in the same source; URL ratio without URL watts overrides saved watts. Old temperature, sweat-category and performance rules are not reinterpreted as measurements. Simple shares/saves active fields only; advanced fields stay in memory for deliberate reactivation. Revised models do not promise historical results.