FabTandem Try the live board

Field notes Research

Why busy job shops ship late.

Most late shipments are not caused by lazy people or slow machines. They are created at the front door, when a shop commits to more work than it can actually run. Here is what decades of job-shop research say about it.

More work in the shop does not mean more work out the door

Little's Law ties three numbers together: work in process equals throughput multiplied by flow time. If the laser or the brake can only finish so much per week, then pushing more jobs onto the floor does not raise the number of shipments. It raises the queue, and every job in that queue waits longer.

That is why a shop can feel slammed while shipping the same dollars it did last month. The extra work did not add output. It added waiting.

A little overloaded is a lot late

Congestion is not linear. As a constrained work center gets close to its real limit, waiting time climbs sharply. The last few points of utilization cost far more in lead time and schedule stability than the first ones.

Real capacity is also smaller than machine hours on paper. Setups, absences, rework, downtime, inspection holds, missing material, engineering questions and outside processing all take their share. Custom work adds variability on top: one tricky bend sequence or a poorly understood print can eat far more brake time than quoted and back up everything behind it.

Then the expediting starts

When dates start to slip, shops expedite, break planned sequences, add overtime and reshuffle priorities. Those moves add even more variability, and jobs that were on track become late too. One bad-fit job can push several good ones past their dates.

A customer can plan around a long lead time. They cannot plan around an unreliable one.

A supplier that says six weeks and ships in six weeks is easy to schedule. A supplier that says three weeks and ships in five, seven or nine forces the buyer to chase status, build in padding and, eventually, find a second source.

What the research recommends

The research does not say to turn work away whenever the shop is busy. It says to treat accepting an order as a production decision, not only a sales one.

  • Look at the real workload before accepting. Ebben, Hans and Olde Weghuis (2005) found that, when due dates are tight, acceptance methods that account for the workload already in the shop clearly beat simple rules.
  • Offer a later date before saying no. Moreira and Alves (2006) found that adding a due-date negotiation step before rejecting an order improved almost every workload measure they tested.
  • Count quoted work you expect to win. Thürer, Stevenson, Silva and Land (2013) found the shortest, most reliable due dates came from finite loading that also counted orders quoted but not yet confirmed.
  • Plan acceptance, dates, release and dispatching together. Moreira and Alves (2009) showed these four decisions behave differently when considered together, so fixing scheduling alone cannot undo overload created upstream.
  • Do not release everything at once. Mezzogori, Romagnoli and Zammori (2021) found that workload control, which holds jobs in a pool and releases them as the floor can absorb them, combined with due-date negotiation, cut the number of tardy jobs.

What this looks like on a Monday morning

  1. Know your constraint and its realistic weekly hours, not its theoretical ones.
  2. Measure the backlog in remaining hours at that constraint, not just in dollars.
  3. Before promising a date, check whether those hours are actually open in that window.
  4. When they are not, offer a later date you can keep. Decline the part, not the customer.
  5. Make every expedite name the job it pushes back. No hidden displacement.
  6. Watch on-time delivery, past-due hours, expedite count and how long jobs sit without moving.

Where FabTandem fits

FabTandem does not decide which quotes to accept. It gives you the facts those conversations need: the whole backlog in one place, work-center load from the operations still remaining, past-due and at-risk jobs, and the real reason each blocked job is not moving. Capacity views depend on the hours, rates and routings your shop sets up.

Try the live board Request a demo

Sources

  1. Ebben, M.J.R., Hans, E.W., and Olde Weghuis, F.M. (2005). Workload based order acceptance in job shop environments. OR Spectrum, 27, 107–122. University of Twente
  2. Moreira, M.R., and Alves, R. (2006). Does order negotiation improve the job-shop workload control? FEP Working Paper 213, Universidade do Porto. RePEc
  3. Thürer, M., Stevenson, M., Silva, C., and Land, M.J. (2013). Towards an integrated Workload Control (WLC) concept: the performance of due date setting rules in job shops with contingent orders. International Journal of Production Research, 51(15). Taylor & Francis
  4. Moreira, M.R., and Alves, R. (2009). A methodology for planning and controlling workload in a job-shop: a four-way decision-making problem. International Journal of Production Research, 47(10). Taylor & Francis
  5. Mezzogori, D., Romagnoli, G., and Zammori, F. (2021). Defining accurate delivery dates in make to order job-shops managed by workload control. Flexible Services and Manufacturing Journal, 33, 956–991. Springer, open access
  6. Little, J.D.C. (1961). A proof for the queuing formula: L = λW. Operations Research, 9(3), 383–387.