Lecture 5
HSOM
Variability, uncertainty and flexibility
We are working with people, which means we cannot predict everything upfront like in a
factory. ED gets overcrowded. Waiting times are important in the ED, because it is
inconvenient and unhealthy. In order to understand it we need to know where it is coming
from.
Learning goals
- Distinguish between uncertainty and predictable variability and analyse the
consequences for process performance.
- Use queuing theory to analyse waiting time problems
- Explain the trade-off between variability, excess capacity and waiting times
- Analyse the role of flexibility and pooling in buffering
- Critically reflect on the role of human behavior in service systems
Uncertainty and predictable variability
- Variability refers to deviation from average conditions
- Variability comes in two flavours
o Predictable
o Unpredictable uncertainty
ED – we know from the past and from data that it is more busy during Monday rather than a
Sunday. And we see that it is more busy during the afternoon than the morning. This is
called predictable variability. However, sometimes it is the case in which it happens that
morning/night is more busy than afternoon etc, this is called unpredictable variability.
Predictable variability
- Organisations understand predictable aspects of variability and respond in their
resource allocation
o Fewer doctors and nurses in ED during the times in which it is less busy
usually
- Sometimes called ‘synchronization’ (demand and supply in concert/in synch)
- Unpredictable variability = unsynchronized variability.
Sources of unpredictable variability
- Demand for service:
o Fluctuating patient arrival
o Fluctuating severity
- Provision of service:
o Service times (difference in working rhytm of staff, some work faster some
slower)
o Staffing (absenteeism etc)
- You can’t ensure predictability on the side of demand (patients etc), but also not on
the supply side (staff etc)
1
, Variability principle (aka queuing principle)
- Unsynchronized variability causes queuing and reduces throughput
- Reason: variability leads to occasional idle periods (less busy/empty), which are lost
to the system and cannot be offset or compensated by busy periods
- So can’t compensate for the fact that it is empty sometimes, dependent on the
patient to come, but cant make up for the empty times with busy times
- The more unpredictable variability, the worse it gets
- We wish to measure it, predict it…. But how?
queuing theory – concerned with the mathematical analysis of processes with variation
in demand and service times. How we can measure and predict waiting times for certain
systems. “There is nothing so practical as a good theory”.
Queueing theory aka waiting time models
- Patient (customer) arrival pattern
- Patient (customer) decision:
o Entering the queue/leaving directly (so entering, seeing the queue and either
joining the queue or just leaving immediately)
o Enter the queue within conditions (showing up, and depending on the
conditions of the waiting time, 10 min or 1 hr or uncomfortably busy space,
one can decide to choose otherwise go to GP etc)
o Leaving the queue after some time (showing up, waiting and still deciding to
leave)
Waiting time model – service characteristics
- Service time distribution
- Number of servers (staff present)
- Queue capacity (waiting time area, limited seats etc)
- Queueing discipline (what is the type of order patients are being served)
o First come first served (FCFS)
o Last come first served (this is more about other things, like blood service etc)
o Different priorities (severity e.g.)
Waiting time model – notation (in Fitzsimmons)
A/B/X/Y/Z
- A = arrival distribution
- B = service distribution
- X = number of servers
- Y = system capacity
- Z = queueing discipline
2
HSOM
Variability, uncertainty and flexibility
We are working with people, which means we cannot predict everything upfront like in a
factory. ED gets overcrowded. Waiting times are important in the ED, because it is
inconvenient and unhealthy. In order to understand it we need to know where it is coming
from.
Learning goals
- Distinguish between uncertainty and predictable variability and analyse the
consequences for process performance.
- Use queuing theory to analyse waiting time problems
- Explain the trade-off between variability, excess capacity and waiting times
- Analyse the role of flexibility and pooling in buffering
- Critically reflect on the role of human behavior in service systems
Uncertainty and predictable variability
- Variability refers to deviation from average conditions
- Variability comes in two flavours
o Predictable
o Unpredictable uncertainty
ED – we know from the past and from data that it is more busy during Monday rather than a
Sunday. And we see that it is more busy during the afternoon than the morning. This is
called predictable variability. However, sometimes it is the case in which it happens that
morning/night is more busy than afternoon etc, this is called unpredictable variability.
Predictable variability
- Organisations understand predictable aspects of variability and respond in their
resource allocation
o Fewer doctors and nurses in ED during the times in which it is less busy
usually
- Sometimes called ‘synchronization’ (demand and supply in concert/in synch)
- Unpredictable variability = unsynchronized variability.
Sources of unpredictable variability
- Demand for service:
o Fluctuating patient arrival
o Fluctuating severity
- Provision of service:
o Service times (difference in working rhytm of staff, some work faster some
slower)
o Staffing (absenteeism etc)
- You can’t ensure predictability on the side of demand (patients etc), but also not on
the supply side (staff etc)
1
, Variability principle (aka queuing principle)
- Unsynchronized variability causes queuing and reduces throughput
- Reason: variability leads to occasional idle periods (less busy/empty), which are lost
to the system and cannot be offset or compensated by busy periods
- So can’t compensate for the fact that it is empty sometimes, dependent on the
patient to come, but cant make up for the empty times with busy times
- The more unpredictable variability, the worse it gets
- We wish to measure it, predict it…. But how?
queuing theory – concerned with the mathematical analysis of processes with variation
in demand and service times. How we can measure and predict waiting times for certain
systems. “There is nothing so practical as a good theory”.
Queueing theory aka waiting time models
- Patient (customer) arrival pattern
- Patient (customer) decision:
o Entering the queue/leaving directly (so entering, seeing the queue and either
joining the queue or just leaving immediately)
o Enter the queue within conditions (showing up, and depending on the
conditions of the waiting time, 10 min or 1 hr or uncomfortably busy space,
one can decide to choose otherwise go to GP etc)
o Leaving the queue after some time (showing up, waiting and still deciding to
leave)
Waiting time model – service characteristics
- Service time distribution
- Number of servers (staff present)
- Queue capacity (waiting time area, limited seats etc)
- Queueing discipline (what is the type of order patients are being served)
o First come first served (FCFS)
o Last come first served (this is more about other things, like blood service etc)
o Different priorities (severity e.g.)
Waiting time model – notation (in Fitzsimmons)
A/B/X/Y/Z
- A = arrival distribution
- B = service distribution
- X = number of servers
- Y = system capacity
- Z = queueing discipline
2
