One shared line or separate lines? Compare the waits

With the same servers and the same demand, one shared line gives shorter average waits than a line per server, because no server sits idle while someone waits in another line. At 80% busy with two servers, the formulas give 4 min for separate lines against 1.8 min for one shared line.

Two separate lines, 80% busy4 minsimulation 3.9 min
One shared line, 80% busy1.8 minsimulation 1.8 min
Wait removed by sharing56%same servers, same demand, same service time

Your service point

Separate lines: arrivals split evenly and nobody switches lines, so each server runs its own single-server queue. Shared line: one queue feeds all servers. Random arrivals and exponential service in both. Starting values are an example.

Result

Busy share, each server80%
Average wait, separate lines4 min
Average wait, one shared line1.8 min
Wait removed by sharing56%
Separate4 min
Shared1.8 min
Chance of waiting and simulated waits, separate against shared
LayoutChance of waitingFormula waitSimulated wait
2 separate lines80%4 mincheck
One shared line71%1.8 mincheck

Same busy share (80%), more servers

Average wait for separate and shared lines as the number of servers grows at the same utilization
ServersSeparateShared
24 min1.8 min
34 min1.1 min
44 min45 sec
64 min26 sec
84 min17 sec

The separate-line simulation runs one of the 2 identical lines, fed 1/2 of the arrivals.

The result

Two servers, 96 arrivals an hour, one minute of service each: 80% busy. Separate lines are two single-server queues each fed 48 an hour. The shared line is one two-server queue. Formulas are exact for random arrivals and exponential service; the simulation is seed 7, 30 replications of 200 hours.

Separate lines against one shared line at 80% busy with two servers
LayoutChance of waitingFormula waitSimulated wait
Two separate lines80%4 min3.9 min
One shared line71%1.8 min1.8 min

Why pooling wins

separate: each line is M/M/1 with ρ = λ/c ÷ μ Wq = ρ ÷ (μ − λ/c) shared: one M/M/c line (Erlang C) Wq = C(c, a) ÷ (c·μ − λ)

Both layouts have the same servers, the same demand and the same busy share. The difference is only who is allowed to serve whom. With separate lines, a server whose line is empty cannot help a long line next door; with one line it can. The averages differ because idle capacity is wasted in one layout and used in the other.

When it does not

  • People choose lines. Customers who pick the shortest line behave partly like a shared line, so the gap is smaller than the worst case shown.
  • Servers are slower when shared. Pooled servers must handle every job type, and each may be slower on the ones they are less practised in.
  • Customers see the queue. Research on queue design has found that people react to how a line looks and how long it appears, so a single long line can be judged worse even when it is faster. Treat the result as the mathematical ideal.
  • Priorities and skills. Priority customers and skills-based routing are different queueing systems and are not modelled by this calculator or by Simflowly.

Cross-training: pooled queue vs shared crew

Two different things share the word "pool". A pooled queue, which this calculator measures, is one line feeding identical servers. A shared crew in Simflowly caps the people working across several stations: three stations that draw from a crew of four can run at most four jobs at once between them. The crew answers "what if my people move between stations"; it is not a single shared line. For pickers who also pack, see the picker staffing calculator.

Questions

Why does a shared line wait less?

In separate lines one server can be idle while customers wait in another line. A shared line sends the next customer to whichever server frees up, so capacity is never stranded. The effect is called resource pooling.

Does pooling always help?

In this model yes, but real systems can differ. People may prefer to pick their own line, servers may be slower or lose specialist skills when they must serve everyone, and seeing a long single line can put customers off. Published research on these behavioural effects shows that the benefit can shrink or even reverse, so the calculator gives the ideal case.

Is a shared crew in Simflowly the same thing?

No. A shared crew in the app is a cap on the number of people working across several different stations at once. A shared line in this calculator is one queue feeding several identical servers. They answer different questions.

What about cross-training?

Cross-training is how an organisation gets pooling: servers who can all do every job can share one line. Estimate the gain here with identical servers, then simulate your real mix with different times per job, because pooled servers that are slower at some jobs give back part of the gain.

How big is the gain with more servers?

It grows. At the same busy share, the table in the calculator shows how far the two layouts drift apart as servers go from 2 to 8.

What does the calculator assume?

Random arrivals, exponential service times, identical servers, no switching of lines and an unlimited line. Separate lines receive an equal share of arrivals. Real service times that are more regular shrink both waits, and the simulation lets you test that in the app.

Keep going

Updated 17 November 2026. Free to use, no account. The calculation runs in your browser; nothing you type is sent to us.