From Kiosk Punch to Plan Versus Actual Variance in EDGEBIC

Every figure in the plan-versus-actual variance view of EDGEBIC by User Solutions traces back to an operator tap: run and rework punches become daily actual hours, those hours sum per work center, and the report subtracts them from the scheduled hours the planner committed. There is no re-keying, no export, and no intermediate spreadsheet. That short chain is what makes the variance column worth acting on rather than worth arguing about.

This article follows the chain link by link, because knowing where a number comes from is what decides whether you trust it at the Monday meeting.

The Chain in Four Links

Link one: the punch. An operator taps Start Run and a punch opens with a server-side timestamp. They tap Pause, Resume, and eventually Complete Operation, and each tap closes one punch and opens the next. See how the kiosk captures a production punch.

Link two: the daily rollup. Closed run and rework punches are grouped by the date they started on. Each group becomes one row of actual hours and good pieces for that operation on that date. Setup, idle, down, and teardown time is excluded here and never enters the actual hours figure. See how punches roll up into daily actual hours.

Link three: aggregation by work center. The report sums those daily rows for every operation on a work center inside your date range, giving one actual hours total per machine.

Link four: the comparison. Against it sits the scheduled hours the engine allocated to that machine over the same range, plus the available hours its shift calendar provides. Variance is scheduled minus actual, in hours and as a percentage, with the number of days that saw any activity alongside.

The Three Numbers, and What Each Answers

The report puts three different quantities side by side, and confusing them is the most common misreading.

Utilization compares scheduled to available and is forward-looking: it tells you whether next month is buyable. Variance compares actual to scheduled and is backward-looking: it tells you whether your standards describe reality. A machine can be 105 percent utilized (overloaded) and running perfectly to standard, or comfortably at 60 percent while every job on it takes a third longer than quoted. They are different problems with different owners.

The utilization side of that picture, including the rating bands from idle through critical, is covered in how to run the work center utilization report.

A Worked Example

Range: one month. Five work centers with kiosk terminals.

Four readings come straight off this table.

Heat-1, CNC-1, and Finish-1 are fine. Variance inside a few percent on a month of real work is normal noise. The standards on these machines describe what happens.

Mill-1 has a standards problem. Nearly fifteen percent over plan, consistently, across twenty active days. That is not a bad week. Either the routing cycle times on the parts it runs are optimistic, or something structural about the work has changed. Fixing it makes every future quote and every future capacity check on Mill-1 more honest.

QC-1 has a data problem, not a capacity problem. A hundred scheduled hours and zero logged means nobody is punching there. The plan is committing a machine you have no evidence about.

Heat-1 is the bottleneck and it is running slightly hot. Combined with a utilization reading above 100 percent, that is the machine whose standards you most want correct, because every planning decision in the shop keys off it.

Why Excluding Downtime Is the Right Call

It is tempting to want interruptions in the actual hours column so that the machine's whole day is accounted for. Resist it.

The variance column exists to answer one question: does the work take as long as we said it would? Fold a coolant failure into it and the answer becomes contaminated by an availability event that has nothing to do with the routing standard. Mill-1 above would look worse, and you would not be able to tell whether its problem was cycle time or maintenance.

Lost time is not thrown away. It sits on its own punches with a reason code, feeding downtime analysis, the availability factor in OEE, and setup variance where it belongs. Two questions, two numbers, one honest punch stream underneath. See what a pause-with-reason record is worth.

Closing the Loop

Reading variance is only half of it. The same daily hours that feed this report feed the next reschedule, which plans remaining work forward from where the shop really is and never moves completed operations. So a shop that punches consistently gets two returns from the same taps: a plan that reflects reality tonight, and a standards audit at month end. That is the loop described in closing the loop from shop floor to plan.

It is also the mechanism behind results like the User Solutions customer GE Railcar going from 30 percent to 90 percent on-time delivery. Nothing about that came from a cleverer algorithm alone. It came from a plan that knew what was actually happening.

The Bottom Line

Plan-versus-actual variance is a four-link chain from one operator tap: punch, daily rollup, per work center sum, comparison against scheduled hours. Because only run and rework time crosses into actual hours, the column measures work rather than interruptions, and because nothing is re-keyed along the way, the number is as good as the punch discipline behind it. Read the whole loop in the shop floor execution guide, or explore EDGEBIC.