Erlang C queue calculator (M/M/c and M/M/1): waiting time, utilization, servers needed

An M/M/c queue has random arrivals, random service times and c identical servers. Enter the arrival rate, the service rate of one server and c to get utilization, the probability that a customer waits and the average wait, using the exact Erlang C formula. The system must have utilization below 100%.

Your service point

Assumes random (Poisson) arrivals and exponential service times: the standard M/M/c model. Starting values are an example.

Result

Utilization per server80%
Chance a customer waits64.7%
Average wait before service1.3 min
Average time in system2.5 min
Average number waiting2.59
Wait against utilization at your service rate
20%40%60%80%7.3 min0

Average wait (up) against how busy each server is (across). Your point is the lime dot; the shaded band is above 90% busy.

Exact Erlang C results. Utilization = λ ÷ (c × μ).

Open this exact case in Simflowly

The formulas

a = λ ÷ μ (offered load) ρ = a ÷ c (utilization, must be below 1) C(c, a) = [a^c ÷ (c! (1 − ρ))] ÷ [Σ(k=0..c−1) a^k ÷ k! + a^c ÷ (c! (1 − ρ))] Wq = C(c, a) ÷ (c·μ − λ) (average wait in line) Lq = λ · Wq (Little's law)

Worked example

120 customers an hour arrive (λ = 2 a minute), each server handles 50 an hour (μ = 0.833 a minute) and there are 3 servers. Offered load a = 2.4, utilization ρ = 80%.

The probability of waiting is 64.7%, the average wait in line is 1.29 minutes and on average 2.59 customers are waiting. Add a fourth server and utilization drops to 60%, and the wait falls to a fraction of that. These numbers are computed from the formula above; the check button compares them with a seeded simulation.

With 4 servers the same demand gives utilization 60%, a 28.7% chance of waiting and an average wait of 0.22 minutes.

One server: the M/M/1 case

ρ = λ ÷ μ Wq = ρ ÷ (μ − λ) Lq = ρ² ÷ (1 − ρ)

40 customers an hour arrive and one server handles 50 an hour. Utilization is 40 ÷ 50 = 80%, the chance a customer waits is also 80%, the average wait in line is 4.80 minutes and on average 3.2 customers are waiting. Type 1 in the Servers box to reproduce it. Pure M/M/1 and M/M/c both assume random arrivals and exponential service, so treat the result as a first answer and check it with the simulation button.

Questions

How many workers do I need at a service point?

Enough that utilization stays well below 100% and the average wait meets your target. Enter your arrival and service rates, then raise the number of servers until the wait is acceptable. Going from 3 to 4 servers can cut the wait by far more than a quarter.

What is the M/M/1 queue?

M/M/1 is the case with one server: random arrivals, exponential service times and a single line. Utilization is ρ = λ ÷ μ, the average wait in line is Wq = ρ ÷ (μ − λ) and the average number waiting is Lq = ρ² ÷ (1 − ρ). Enter 1 as the number of servers in this calculator to get it.

What is the difference between Erlang C and Erlang B?

Erlang C assumes a customer who finds every server busy waits in line for as long as it takes. Erlang B assumes that customer is turned away and never queues, so it gives the chance of being blocked instead of a wait. Use the Erlang B calculator for lines that cannot queue.

What does M/M/c mean?

The first M says arrivals are Markovian (random, Poisson). The second M says service times are exponential. The c is the number of identical servers sharing one line.

What is the Erlang C formula?

Erlang C gives the probability that an arriving customer finds all c servers busy and has to wait. The average wait follows from it: Wq = C(c, a) ÷ (c·μ − λ).

When is this model wrong?

When service times are not exponential (real ones are often more regular), when customers give up, when arrivals come in waves, or when servers differ. Simflowly simulates those cases; the check button runs a seeded simulation for this exact model.

Why does the wait explode near 100% utilization?

Because there is no spare capacity to absorb random bunching of arrivals. The queue length grows roughly as 1 ÷ (1 − utilization), so going from 90% to 95% busy about doubles it.

Keep going

Updated 13 November 2026. Free to use, no account. The calculation runs in your browser; nothing you type is sent to us.