CPG Scheduler
All posts

EPEI (Every Product Every Interval): How Often Should Your Line Make Each SKU?

schedulingchangeoverssmedpar levelsinventory planning

Every Product Every Interval (EPEI): How Often Should Your Line Make Each SKU?

Every scheduler knows the line that runs like this: SKU A goes every day, B and C a couple times a week, and SKU F only gets made when we run out. Nobody planned it that way. It just happened, one urgent changeover at a time.

There's a single number hiding underneath that mess, and it's worth naming: how long does it take for the rotation to come all the way back around to each SKU? In lean terms, that's your EPEI — Every Product Every Interval. Get it right and your inventory, your par levels, your stockouts, and your changeover load all fall into place. Get it wrong and you firefight forever.

This post explains what EPEI is, why a smaller one is usually better, what physically limits how small it can go, and how to calculate and shrink it on a real food-and-beverage line.

What EPEI actually is

EPEI is the frequency with which the different SKUs or part numbers on a process get produced. If a machine is sequenced so that every assigned item runs at least once every three days, the EPEI is three days. Run everything once a week, and your EPEI is one week. That's the whole definition — it's a measure of how often the rotation repeats.

The mental model that makes it stick is the production wheel. Picture every SKU the line makes as a slot on a wheel. The wheel turns in a fixed sequence, and one full turn touches every slot once. EPEI is simply how long one full turn takes. A tight wheel spins fast — every SKU comes around every day or two. A wide wheel turns slowly — some SKUs wait a week or more for their slot.

This is a different decision from leveling, even though they're related. Heijunka is about smoothing the sequence and mix so you don't run giant campaigns followed by long droughts. EPEI is about the size of the wheel — how many SKUs are on it and how often the whole thing repeats. You can't level a sequence sensibly until you've decided how big the cycle is, so EPEI is the natural companion to leveling.

Why a smaller EPEI is usually better

The general lean guidance is blunt: it's good for EPEI to be as small as possible. A small interval means you produce a small lot of each item each time it comes around, which minimizes the inventory sitting in the system. Here's why that cascades through a CPG plant:

  • Smaller lots → lower cycle stock. If SKU F runs every two days instead of every two weeks, you only need to cover two days of demand between runs, not fourteen. Cycle stock drops in proportion to the interval.
  • Lower cycle stock → lower par levels and less safety stock. When the next run is never far away, you don't need a mountain of finished goods to bridge the gap. Your par levels and safety stock can come down with the wheel.
  • Fresher product. In food and beverage, inventory isn't free to hold — it ages. A tighter wheel means product spends less time in the warehouse, which matters for shelf life, date codes, and quality complaints.
  • Faster response to demand swings, fewer stockouts. A fast wheel comes back around quickly, so an unexpected spike on one SKU gets a fresh run in days, not weeks. The long gaps between runs are exactly where stockouts hide.

So if smaller is better, why doesn't everyone just run every SKU every day? Because two real constraints push back.

The two things that cap how small EPEI can go

A machine's achievable EPEI depends on two factors: changeover time and the number of SKUs assigned to it. A line with long changeovers — and the large minimum batch sizes that usually come with them — running a wide variety of SKUs will inevitably have a large EPEI, unless you reduce changeover time or cut the SKU count.

The logic is simple arithmetic. Every turn of the wheel has to fit, inside the available production time, both the run time for every SKU and a changeover between each one. The more SKUs you carry, the more changeovers per turn. The longer each changeover takes, the more time gets eaten by setup instead of making product. At some point there isn't enough time left in the cycle to also produce demand — so the wheel has to slow down (bigger EPEI) to amortize all that changeover time over fewer, larger runs.

That's the trade-off curve every scheduler lives on:

  • Push EPEI down and you spend more total time on changeovers, but carry less inventory.
  • Let EPEI grow and you spend less time on changeovers, but carry more inventory and fresher product becomes harder.

The right answer isn't "smallest possible" in a vacuum — it's the smallest wheel your changeover load can sustain while still meeting demand. The whole game is moving that curve, which we'll get to.

How to calculate EPEI on a real line

Here's the back-of-the-envelope version you can do this week. You need five inputs:

  1. Available production time per period (after planned downtime, breaks, sanitation).
  2. Demand per SKU over that period.
  3. Run rate for each SKU (units per hour).
  4. Changeover time between SKUs (use your real numbers, not the spec sheet).
  5. Number of SKUs on the line.

The method: figure out how much time per cycle is consumed by production, then see how much is left over for changeovers. The interval is how long it takes to accumulate enough leftover time to fit all the changeovers in one full wheel.

Let's work a beverage line with 8 SKUs, a single shift of 480 minutes/day, and total daily demand that consumes 360 minutes of run time at current rates. That leaves 120 minutes/day for changeovers.

Now say each changeover averages 30 minutes. One full turn of the wheel needs 8 changeovers = 240 minutes of setup. You only free up 120 minutes a day, so one complete wheel takes 240 ÷ 120 = 2 days. Your EPEI is 2 days — every SKU gets made at least every other day.

Now halve the changeover with a SMED effort to 15 minutes. One turn now needs 8 × 15 = 120 minutes of setup, and you free up 120 minutes a day. The wheel now turns in 1 day. You just cut EPEI in half — and with it, roughly half your cycle stock — without touching demand or adding a shift.

Scenario Changeover (min) Setup per wheel Daily slack EPEI
Baseline 30 240 min 120 min 2 days
After SMED 15 120 min 120 min 1 day
Drop 2 SKUs 30 180 min 120 min 1.5 days

Notice the third row: cutting from 8 SKUs to 6 also shrinks the wheel, because there are fewer changeovers per turn. Changeover time and SKU count are the two levers, and the math shows it directly.

Setting the wheel: fixed interval vs. variable volume

Once you know the feasible interval, you choose how rigid the wheel is:

  • Fixed sequence, fixed interval. Same SKUs, same order, same quantities every turn. Dead simple, highly predictable, great for stable demand. The downside is it ignores week-to-week demand shifts.
  • Fixed sequence, variable volume. Keep the order and the interval, but flex the quantity of each SKU based on current orders and where its inventory sits against its par level. This is the workhorse for most CPG lines: the wheel structure stays stable while the amounts breathe with demand.

This is where the wheel ties straight into your inventory triggers. Each SKU's par level and reorder point decide how much to run when its slot comes up — and whether an SKU that's draining fast needs an extra run before its normal slot. A well-set wheel plus honest reorder points is what keeps you off the panic-changeover treadmill.

Levers to shrink EPEI

When the wheel is too wide, three levers move it:

  1. Reduce changeover time (SMED). This is the biggest and most direct lever — the worked example shows halving changeover time can halve EPEI. Convert internal setup steps to external (done while the line still runs), standardize the procedure, and stage tooling and materials ahead of time.
  2. Rationalize SKUs and group families. Fewer SKUs on the line means fewer changeovers per turn. And grouping SKUs into families that share formats, flavors bases, or cleaning requirements lets you sequence so the expensive changeovers happen less often and the cheap ones absorb the variety.
  3. Dedicate vs. share equipment. A high-runner that's choking the wheel might deserve its own dedicated line or shift, freeing the shared line to spin a tighter wheel for the rest.

The high-mix, low-volume reality check

EPEI is described as the backbone of lean in high-mix, low-volume environments — and that's exactly where it's hardest. In those settings demand is more erratic and shared equipment runs a wide variety of items, so changeovers and demand variation are the core obstacles to flow. That describes a lot of CPG: a co-packer running dozens of private-label variants, or a sauce line cycling through fifteen flavors on shared kettles and fillers.

The lesson translated: you can't make the variety go away, so the wheel becomes the mechanism that creates flow despite it. You pick the smallest interval your changeover load can sustain, group SKUs into families to tame the setups, and let the wheel turn predictably. Erratic demand doesn't break the wheel — it just flexes the quantities inside a stable cycle. That predictability is what turns a chaotic high-mix line into one you can actually plan around.

How CPG Scheduler operationalizes EPEI

Doing this by hand once is educational. Doing it continuously, as demand and changeover times shift, is what the scheduler is for. CPG Scheduler computes the feasible EPEI from your real changeover matrix and current demand, flags when SKU count or creeping changeover times are forcing the wheel wider than it should be, and re-sequences the wheel when demand moves — flexing run quantities against par levels while keeping the cycle stable. You see the trade-off, not just a number.

Operator checklist: find and set your EPEI this week

  1. List the SKUs on one line and pull honest run rates and demand for each.
  2. Measure real changeover times between them — stopwatch, not spec sheet.
  3. Do the math: production time consumed vs. time left for changeovers → solve for the interval.
  4. Compare to today. Is your actual rotation tighter or wider than the feasible EPEI? Where are the long gaps that cause stockouts?
  5. Pick one lever — a SMED target on your worst changeover, or a SKU/family consolidation — and recalculate to prove the wheel gets smaller.

Name the number. Once you can say "our wheel turns every two days," every other conversation — par levels, safety stock, when to run extra — gets a whole lot clearer.

Sources

More from the journal