Spare parts are the line item almost nobody negotiates and the one that decides how long a kitchen stands still. A commercial exhaust system is bought once but repaired for a decade, and the gap between a two-day repair and a three-week repair is rarely the price of the part. It is whether that part was within reach of the site when it failed. This is a buying method for parts, written the way it should be written in the purchase file: by criticality class, by depth, and by the clause that keeps the part available in year five.
A stockout costs something other than the part
Buyers instinctively compare a spare against its own price. That comparison is the wrong one. The relevant number is the cost of the operating hours lost while the part is in transit. A kitchen that closes for a Friday and Saturday service does not lose the part; it loses covers multiplied by average spend, plus labour that still clocks in, plus the bookings that do not come back the following week. A component invoiced at the value of two covers can therefore trigger a loss measured in hundreds of covers.
The arithmetic is easy to run with the buyer’s own figures. Take weekend covers, multiply by average spend, add the wage bill for the two days, then add a retention factor for the bookings that migrate to a competitor. Most operators find that one lost weekend exceeds the annual carrying cost of a full class A parts kit. That single comparison is usually enough to move spares out of the procurement argument and into the operating budget, where it belongs.
Carrying cost is not the purchase price. It is capital tied up, shelf space, insurance, and the risk of a part ageing past its shelf life. For a kitchen estate that figure is typically a small fraction of one lost service, which is why the argument for holding stock is usually won on downtime rather than on discount.
Four classes of spare, four different rules
Treating every part the same produces the two classic failures at once: a warehouse full of parts that never fail, and no high-voltage power supply in the country when one fails. Spares must be sorted before they are counted. The classification below is the one worth copying into a tender annex, because it gives the supplier a defensible basis for what they hold and gives the buyer a basis for what to demand.
| Class | Definition | Typical parts | Holding rule |
|---|---|---|---|
| A – Stop | Kitchen cannot operate, or operates outside a safety or legal requirement | High-voltage power supply, fan motor, drive belt, door interlock switch, control board | One per site on single-cell systems; one per two sites on multi-cell systems |
| B – Degrade | Output falls, odour or grease capture slips, compliance margin erodes | ESP cell, collector plates, ceramic insulators, sealing strips, activated carbon charge | Market level, sized by the formula in the next section |
| C – Convenience | Repair slows the work but the kitchen keeps trading | Wash hoses, nozzles, drain trays, gaskets, lamps, castors | Branch level, small quantities, replenished with consumables |
| D – Never | Too large, too slow-moving, or unique to one site | Hood shells, duct sections, structural frames, canopy panels | Order on failure; hold drawings instead of metal |
The classification has a hard consequence that buyers should write into the contract rather than assume. A class A part that is not held within the promised lead time is a service failure, not a supply delay. If the supplier will not accept a lead-time commitment per class, the classification is decorative.
The four numbers every stocked part needs
Depth is not a feeling and it is not “one of everything”. Four numbers settle it, and all four should appear on the parts schedule at handover. If any of them is missing, the buyer is holding stock on the supplier’s optimism.
- Mean time between replacement, in operating days, taken from field data or from a written supplier statement. Not a warranty period – a replacement interval.
- Replenishment lead time, in calendar days, door to door: production queue, freight, customs clearance and inland delivery. Ex-works dates flatter this number by three to five weeks.
- Minimum viable quantity: how many of this part can fail before the kitchen is out of compliance or out of business. Single-cell systems answer one. Multi-cell systems often answer one and a half.
- Failure cost in operating hours, which decides whether the part justifies air freight when it is finally ordered.
| Part | Replacement interval | Lead time | Min viable qty | Failure cost |
|---|---|---|---|---|
| Activated carbon charge | 90-180 days | 30-45 days | 1 per unit | Odour complaints within days |
| ESP cell | 3-5 years with washing | 30-45 days | 1 per system | Full stop on single-cell units |
| Ceramic insulators | 2-4 years | 30-45 days | 1 cell set per site | Arcing, then shutdown |
| High-voltage power supply | 5-8 years | 30-60 days | 1 per 15 units | Full stop |
| Fan motor or belt | 4-7 years | 20-60 days | 1 per 10 units | Full stop |
| Sealing strips | Every cell service | 30-45 days | 1 set per 5 units | Bypass, capture loss |
Sizing depth with a formula instead of a hunch
The formula is deliberately blunt: stock equals the replenishment lead time plus the inspection interval, divided by the replacement interval, multiplied by the quantity consumed per failure, rounded up. Written out, it says something simple – hold enough to survive one full replenishment cycle plus one inspection cycle, and no more.
Worked example. An estate runs ten ductless units on a 120-day carbon change, carbon arrives 45 days after order, and the maintenance visit happens every 30 days. Lead time plus inspection interval is 75 days; divided by 120 gives 0.625 changes per unit per cycle. Multiplied by ten units that is 6.25, rounded to seven carbon charges. Note what the pooling did: holding one charge per unit would have put ten in the store, three of which would sit until they aged. Pooling across the estate is what makes depth affordable.
The same arithmetic applied per site instead of per estate is the most common stocking error in multi-site groups. Cells, insulators and power supplies are interchangeable across identical units, so they should be counted once at market level and drawn against by any site. Depth belongs to the pool; the response time belongs to the site. Mixing the two produces both over-stocking and stockouts at the same time.
Two caps apply to the formula. Shelf life caps the carbon and any elastomer, because a charge stored for two years is a charge with reduced working capacity. Capital caps the rest: if a part’s annual consumption is below one unit across the whole estate, hold it centrally and accept the freight.
Wear items and the warranty boundary
Most parts arguments start at the wrong end – with a failure that the buyer expects the warranty to cover and the supplier classifies as wear. The boundary is worth fixing in writing at order stage, because after a failure nobody argues about the engineering any more, only about who pays.
The standard position on commercial kitchen purification equipment is a twelve-month whole-unit warranty, with the components that do the actual collecting treated as consumable wear items: the ESP cell, the aluminium collector plates, the sealing strips and the ceramic insulators. That is not an evasion, it is how the equipment works. A cell that has washed grease off a cookline for three years has absorbed a predictable amount of mechanical and thermal cycling, and the failure mode at the end of it is wear, not defect.
The practical consequence for the buyer is that the wear list must be named in the contract, item by item, and the parts schedule must then be budgeted as an operating cost. A buyer who has not done this discovers the boundary at the first claim, and usually discovers it as a dispute rather than a purchase order.
Consumption ratios per ten units in service
Where field data is thin, ratios give a starting schedule that can be corrected after two quarters of records. The figures below are deliberately conservative: they assume proper washing intervals and a normal duty mix, and they should be raised for solid-fuel or 24-hour operations.
| Part | Ratio per 10 units | Why this ratio |
|---|---|---|
| ESP cell | 1 per 5 units, single-cell systems; 1 per 10 units, multi-cell | A single-cell unit is down at the first cell failure, so the spare is a class A part |
| Collector plates | 1 service kit per 10 units per year | Plates are usually replaced with the cell rather than individually |
| Ceramic insulators | 8 per 10 units per year | Tracking and cracking appear at inspection, not at failure |
| Sealing strips | 1 set per 5 units per year | Compression set causes bypass long before the strip splits |
| High-voltage power supply | 1 per 15 units | Low failure rate, but every failure is a full stop |
| Activated carbon | One full charge per unit, plus 25 percent | Change interval is 90-180 days; the extra 25 percent covers early saturation |
Carbon deserves a note of its own, because the depth question is really a weight question. A lightly loaded bed of 20-25 kg saturates faster than a 50 kg bed on the same duty, so two systems that look identical from the outside can need different stocking. The ratio should therefore be expressed in kilograms per unit per year, not in charges, and checked against the change interval the supplier actually recommends for the duty class on site.
Three ways to hold the stock, and who owns the risk
Once depth is known, the commercial structure matters as much as the quantity. There are three arrangements in general use, and they differ in who carries the capital and who carries the obsolescence.
| Structure | Title sits with | Buyer pays | Obsolescence risk | Best for |
|---|---|---|---|---|
| Buyer-owned | Buyer | On delivery | Buyer | High-consumption wear items with known intervals |
| Consignment | Supplier until drawn | On draw | Supplier, if the clause says so | Class A parts that rarely fail but stop the kitchen |
| Supplier pool | Supplier | Per event, at a fixed price | Supplier | Small estates, single sites, new markets |
Consignment is the structure most often offered and least often defined. Four terms decide whether it is genuinely cheaper or merely deferred: the reconciliation cycle and who counts, how shrinkage is split, whether the price is locked when the stock is placed or when it is drawn, and what happens at exit – whether the supplier buys back undrawn stock at invoiced value or the buyer is left holding parts for equipment it no longer runs. Without those four in writing, consignment is a storage arrangement with a friendly name.
The clause that keeps parts available in year five
Availability is the part of the parts argument that buyers forget because it does not bite in year one. It bites when a control board revision is discontinued in year four and the only route is a full controller replacement. The clause below is short enough to survive negotiation and specific enough to be enforceable.
- Parts list at handover: item description, part number with revision suffix, quantity per unit, wear classification, and replacement interval.
- Availability term: listed parts available for five years from delivery of the equipment.
- Discontinuation notice: 90 days written notice before a part is withdrawn, with a last-time-buy offered at the price in force at the date of notice.
- Equivalence: any substitute must carry a signed form-fit-function statement, with dimensional tolerance, electrical rating and material grade addressed explicitly.
- Lead-time commitment per class: five working days for class A from market stock, ten for class B, twenty for class C. A commitment without the class split is not a commitment.
- Spare price basis: fixed for twelve months, then adjusted against a named published index, with no retroactive application.
- Part warranty: twelve months from fitment, with the wear items named individually rather than referred to by category.
- Documentation as a condition: exploded diagrams, wiring schematic, and the cell disassembly sequence, supplied at handover – not on request.
The kit rule, and why a cell on its own is not a spare
A part is only a spare if everything needed to fit it is on the same shelf. A replacement ESP cell without insulators, sealing strips and the correct high-voltage connector is a heavy box that cannot restore service. A fan motor without the matching belt, pulley and terminal block is the same. The kit rule is the cheapest quality control a store can apply: for every class A and class B item, list the mating consumables and count them together.
The second discipline is storage condition. Carbon should stay sealed and dated, because an open bag in a warm store loses working capacity quietly. Ceramic insulators should be kept dry and boxed, since a hairline crack from a loose bin becomes an arc path at ten kilovolts. Cells should be stored flat with the ionizer frame protected and never stacked on their collector plates. None of this is expensive; all of it is invisible until a part that was “in stock” fails on fitment.
Counting follows from the same logic. A quarterly cycle count against minimum and maximum levels, in the maintenance system rather than a spreadsheet on one person’s laptop, is enough. The two figures worth tracking over time are stock turns per class and the number of emergency air freight orders per year. Rising air freight on class B items is the earliest signal that the depth formula was set with the wrong replacement interval.
Where the stock sits: the freight reality
Parts depth is a geography problem before it is a finance problem. Sea freight from an Asian factory to a European or North American port commonly runs four to six weeks door to door, and customs can add a week without warning. A supplier who promises class A parts from factory stock has promised a six-week repair, whatever the invoice says.
The structure that works is three tiers. Site level holds the parts that fit in a service van and cover the first visit: sealing strips, insulators, fuses, sensors, hoses. Branch level holds class A items for the sites it serves, counted as a pool. Central or market level holds the slow-moving, high-value stock – cells, power supplies, blower assemblies – and absorbs the freight. The test of the structure is simple: can every site be back in service within the class A lead time without an air freight order. If not, one of the tiers is empty.
Matching the parts schedule to the architecture
The schedule follows the equipment, not the other way round. Systems that recirculate air inside the kitchen and pair a multi-stage electrostatic cell with a carbon odour stage carry no duct, no fire dampers and no rooftop fan, so their parts profile is narrow and dominated by cells, insulators and carbon. Buyers comparing ductless range hood options should expect a short list, a fast class A response and a carbon line that dominates annual spend.
Ducted systems inherit the whole air path. A ducted ESP purifier installation adds fan bearings or belts, a variable-speed drive, damper actuators and access panel gaskets to the schedule, and the fan becomes the class A item that decides the response time. The parts schedule should grow with the air path, and the lead-time commitment should be demanded for the fan as firmly as for the cell.
Mobile units sit in a third category. A mobile catering cart travels, vibrates and is handled by people who did not install it, so the failure profile shifts towards fixings, castors, electrical connections and impact damage rather than wear. Depth can be shallower, but the kit should include the fasteners and connectors that a travelling unit loses first.
The design response to all of this is not a bigger store, it is a shorter parts list. Equipment built around a small number of replaceable modules, with a documented disassembly sequence and a parts number that survives a revision, is cheaper to keep running than equipment with a long tail of site-specific components. That is the criterion worth adding to the technical evaluation, next to capture efficiency and noise: how many distinct parts does this system need to stay alive for five years, and which of them are on the wear list.
Ten questions to take into the negotiation
- Which parts are classified as wear items, named individually?
- What is the replacement interval for each wear item at our duty class, in operating days?
- What is the door-to-door replenishment lead time for each class, not the ex-works date?
- Which parts will you hold in our market, and at what depth?
- What is the discontinuation notice period, and is a last-time-buy at held price included?
- Will you sign a form-fit-function statement for any substitute part, addressing tolerance, rating and material?
- What is the price basis for spares after year one, and which index?
- On consignment, who counts, how is shrinkage split, and what is the buy-back at exit?
- Are exploded diagrams, the wiring schematic and the cell disassembly sequence included at handover?
- How many emergency air freight orders did this model generate across your installed base last year?
The last question is the one that separates a schedule from a sales answer. A supplier who can quote an air freight rate per installed unit knows their own failure profile. A supplier who cannot has just told the buyer that the parts schedule will be built from somebody else’s experience, after the first breakdown.
