Bulk fuel and propane delivery is a forecasting business with a routing problem attached. The core question is not "what is the most efficient route" but "who needs fuel before they run out, and how do we serve them economically".
What makes it distinctive
Variable delivery quantity. You deliver until the tank is full or the delivery limit is reached. Quantity is not known in advance, which affects route capacity planning and vehicle utilisation.
Consumption forecasting. Customer usage depends on weather, occupancy, appliance mix and habit. Delivering too early wastes a visit; too late means a run-out, an emergency delivery, a safety procedure and a very unhappy customer.
Metered delivery and pricing. Product is metered at the point of delivery, priced at the time of delivery, and invoiced from meter readings — with legal metrology requirements on the meter.
Safety and regulatory obligations. Hazardous goods transport rules, driver certification, vehicle placarding, tank inspection regimes, leak checks and documented delivery procedures.
Tank ownership. Suppliers typically own the tank at the customer's premises, making it a tracked asset with inspection and maintenance obligations.
Remote tank level monitoring replaces forecasting with measurement for the customers that have it.
Degree-day forecasting
The industry-standard approach for heating fuel:
`` Estimated consumption = K-factor × degree days since last delivery ``
Where the K-factor is the customer's own consumption rate, calculated from their delivery history and the degree days between deliveries. Modern systems refine this continuously as each delivery provides another data point.
Practical requirements:
- Weather data by customer location, historical and forecast
- Automatic K-factor recalculation after every delivery
- Separate handling of non-heating load — cooking, hot water, process use — which is weather-independent
- Confidence bands, so that uncertain customers are visited earlier
- Flags for customers whose consumption pattern has changed, which often indicates an appliance change, a leak, or occupancy change
Routing implications
Bulk fuel routing differs from parcel or case delivery:
- Capacity is measured in product volume, and the vehicle empties progressively through the route
- Delivery quantity is uncertain, so plans need buffer or dynamic adjustment
- Customers are prioritised by urgency, from forecast tank level rather than by order
- Geographic clustering matters enormously, since rural density is typically low
- Will-call and automatic delivery customers are handled differently and often on the same route
- Emergency and run-out deliveries must be inserted into existing plans
Good software plans by combining urgency (days to run-out) with geography, filling routes with customers who are due soon and in the right area — rather than strictly by urgency, which produces long, inefficient routes.
Compliance
Requirements vary by jurisdiction but consistently include:
- Driver certification for hazardous goods transport
- Vehicle inspection, testing and placarding
- Meter calibration and certification
- Delivery documentation and record retention
- Tank inspection and maintenance schedules
- Leak check procedures documented at delivery
- Emergency response procedures and equipment
- Customer premises safety checks
The software should hold these as tracked, expiring records with escalating reminders — driver certifications, vehicle tests, meter calibrations and tank inspections all have dates that must not be missed.
Metrics
| Metric | Why |
|---|---|
| Gallons or litres per stop | Core efficiency |
| Gallons per route hour | Productivity |
| Run-out rate | Service failure and safety measure |
| Forecast accuracy (predicted vs actual fill) | Forecasting model health |
| Percentage of tank capacity filled per delivery | Visit efficiency |
| Emergency delivery rate | Planning quality |
| Cost per delivery | Economics |
| Tank monitor coverage of volume | Model maturity |
Percentage of tank capacity filled is the underrated metric: deliveries that only fill a tank halfway indicate visits that were made too early.
Frequently asked questions
How accurate is degree-day forecasting?
With a well-maintained K-factor and enough delivery history, it is accurate enough to plan against for most residential heating customers. Accuracy degrades for customers with irregular occupancy, mixed-use consumption or recent appliance changes — those are the ones to prioritise for tank monitors.
Are tank monitors worth the cost?
On high-volume, unpredictable or run-out-prone customers, generally yes: they eliminate forecasting error, allow fuller deliveries and prevent the emergency deliveries that destroy route economics. Fitting them universally is rarely justified.
How do we handle will-call customers?
Track them separately, since they generate unplanned demand. Many operations use forecasting to prompt proactive contact with will-call customers approaching a low level, converting an emergency into a planned delivery.
What causes run-outs?
Usually forecasting error from a changed consumption pattern, a missed delivery window, or a will-call customer who did not call. Analyse every run-out for cause — the pattern usually points to a specific correctable failure rather than random variation.
Can general route optimisation software handle bulk fuel?
Rarely well. The combination of variable delivery quantity, urgency-based prioritisation, forecasting and metered invoicing requires purpose-built capability. Products serving this sector exist and are worth the specialisation.
