Insights / Process design

How to Size Buffer Tanks and Intermediate Storage

A buffer tank has to do four things: take a whole batch when it is discharged, keep the next stage running while the stage before it is stopped, give the stage before it somewhere to put product while the next stage is stopped, and do all that without holding product longer than its quality limit allows. The first and last can be calculated directly. The middle two depend on how often and for how long each stage stops, which is why tanks sized from a single calculation tend to be either too small to protect output or larger than they need to be.

What a buffer is for

Intermediate storage sits between two stages that do not run in step: a batch cooker and a continuous filler, a fermenter and a bottling line, a moulding line and a packing hall. It does three jobs:

  • Rate matching. Absorbs the difference between a stage that produces in batches and one that draws continuously, or between two stages running at different speeds.
  • Starvation protection. Holds enough product to keep the downstream stage running while the upstream stage is stopped, late or changing over.
  • Blocking protection. Has enough free space for the upstream stage to keep producing while the downstream stage is stopped.

Starvation protection needs the tank to be full. Blocking protection needs it to be empty. A tank that does both has to run at a level somewhere in between, which is why buffer size and the level the schedule aims for are one decision.

The basic calculations

For a first estimate, add up these components in kilograms or litres:

Component Calculation
Heel: product the pump cannot draw From the tank and pump design
Room for a batch discharge One batch volume
Cover for an upstream stop or late batch Downstream draw rate × longest upstream delay you want to cover
Room for a downstream stop Upstream output rate × longest downstream stop you want to absorb
Rate mismatch over a run (Inflow rate − outflow rate) × run length, if one side is faster for long periods
Freeboard Space for foaming, agitation and level control margin

Then check the result against three limits:

  • Hold time. Maximum useful volume is roughly the downstream draw rate multiplied by the longest time product may sit in the tank. Above that, product will regularly exceed its limit whenever the next stage slows down.
  • Cleaning. A tank being cleaned cannot receive product. If cleaning-in-place takes an hour and happens at every product change, either the schedule needs a gap upstream, or the plant needs two smaller tanks used alternately.
  • Product changes. One tank can usually hold one product at a time. Several smaller tanks may serve a varied schedule better than one large one.

A worked example

The figures are illustrative. A sauce plant has a batch cooker feeding a filling line through one buffer tank.

  • The cooker produces 4,000 kg batches, one every 100 minutes when on time. About 30% of batches are late, by up to an hour.
  • The filler draws 2,600 kg an hour when running. It stops on average every two hours, for an average of 20 minutes.
  • The tank heel is 300 kg, and product may not be held for more than four hours.

Both stages average about 2,200 kg an hour once their delays and stops are counted, so neither can make up for the other.

A rule-of-thumb sizing adds the heel (300 kg), one batch (4,000 kg) and an hour of filler draw to cover a late batch (2,600 kg), giving about 6,900 kg. The hold-time limit caps useful size at about 10,400 kg (four hours of filler draw).

A rule of thumb gives one number. It does not say how much output each size protects. To show that, we ran a simple simulation of this example: a year of minute-by-minute running with random late batches and filler stops, repeated eight times for each tank size and averaged.

Tank working capacity Line output Time the filler is waiting for product
5,000 kg 2,063 kg/h 6.1%
6,900 kg (rule of thumb) 2,126 kg/h 3.5%
8,000 kg 2,145 kg/h 2.9%
10,000 kg 2,163 kg/h 2.2%
12,000 kg 2,172 kg/h 1.8%
16,000 kg 2,182 kg/h 1.4%

Output rises with tank size, with each extra tonne of capacity adding less than the one before. Going from 6,900 kg to 10,000 kg adds about 1.7% output. Going from 10,000 kg to 16,000 kg adds less than 1%, and tanks above about 10,400 kg break the four-hour hold limit anyway.

Even the largest tank cannot lift output above what the two stages can make between them. If more output is needed beyond that, reducing the late batches and filler stops is the better target.

Whether 10,000 kg is worth the extra over 6,900 kg depends on the margin on 1.7% more output against the cost of the larger tank, its floor space and its cleaning. That comparison needs the output figure from the table, which a rule of thumb does not provide.

Why real plants need a model for this

The example has one tank between two stages. A real plant has several tanks, several cookers and several fillers sharing them, a schedule that changes product every few hours, cleaning cycles that take tanks out of service, and shelf-life rules that differ by product. The interactions are what decide how much storage is needed, and they change with the schedule.

That also works in the other direction. A better schedule can reduce the storage a plant needs:

  • Pickles and condiments: intermediate tank requirements cut by 40% through better scheduling, as part of a move from 24/7 to 24/5 operation.
  • Vinegar: fewer storage tanks needed while tank utilisation rose to 87%. Some tanks had been masking scheduling problems.
  • Chocolate moulding: 7% less waste from sizing the buffer between moulding and packing in the model instead of by rule of thumb.

Common questions

Is one large tank better than two smaller ones?

One large tank is cheaper per litre and simpler to control. Two smaller tanks allow one to be cleaned or changed to a new product while the other is in use, and keep hold times shorter. For a varied schedule with frequent cleaning, two tanks often protect more output for the same total volume.

How do I account for shelf life in the buffer?

Set the maximum hold time for each product, then make sure the level the schedule runs at, divided by the draw rate, stays below it. A model can track how long each batch actually sits in the tank, which matters when the next stage stops for longer than planned.

What data do I need to size a buffer properly?

Batch sizes and cycle times, draw rates, downtime records for the stages on either side (how often they stop and for how long), cleaning times and frequencies, and hold-time limits by product.

Related guides

Size your storage in a model