Literature week 4 HSOM
Hummy Song, Anita L. Tucker, Karen L. Murrell (2015) The Diseconomies
of Queue Pooling: An Empirical Investigation of Emergency Department
Length of Stay. Management Science 61(12):3032-3053
Key objective for service organizations: improving efficiency and customer experience.
Queue management decisions are important in this, because they determine how long
customers have to wait for service.
Pooling separate streams of identical customers into a single queue of identical servers is
more efficient than a set of dedicated queues. Leads to shorter waiting times because
customers can be processed by any available server rather than waiting for one to become
available. However it may not always yield to expected performance improvements, e.g.
when you combine customers who have different needs. Also, the servers might be affected
if they get more customers assigned just because they are faster.
LOS = a measure of service time, starts with the time the physician begins delivering care to
the patient and ends with either a bed request for admission to the hospital, or the
discharge of a patient to their home or an outside facility. This includes both the active time
in which patients receive care, and the inactive time.
Round robin (RR) routing policy: assigning patients to physicians in which patients were
evenly distributed across physicians independent of physician speed or idle time. This was
done because physicians had fixed salaries, so no financial incentives. Fairness constraint.
A queueing system with a dedicated queuing system and fairness constraint has less waiting
and LOS time (than with pooling queue). Interviews with physicians suggest that, in this
context, the increased ownership that stems from a dedicated queuing system with a
fairness constraint leads to a situation in which the improvements in service rates due to
better flow management are greater than the variability-buffering benefits of a pooled
queuing system with a fairness constraint. Physicians change their behavior as a response to
their assigned responsibilities and ownership over the work routines and resources needed
to accomplish it.
Pooling is usually a good method to reduce costs, by reducing waiting times even when
work is allocated fairly among servers using a RR routing policy. Moreover, pooling with a
fairness constraint is even better than classical pooling, because when there are a lot of
customers arriving fast servers have an incentive to slow down their service rate in systems
which distribute customers based on server availability (instead of fair distribution).
However, sometimes the behavioural responses of servers and customers can reduce these
expected beenfits. This can happen because servers either speed up their service when
there are a lot of customers, or slow it down as to not have to deal with a lot. Or increasing
service time when there are not a lot of customers. Depends on financial incentives and
responsibilities (ownership) how a server will behave. Employees feel ownership when they
are given resources and responsibilities to manage the complete workflow of a meaningful
task.
Dedicated queuing systems with fairness constraint have higher ownership feeling then
pooling. Thusly they might be more motivated to efficiently manage their workload.
1
, Strategic servers: servers who have multiple options of service to give/ask of the customers,
and their choice of service/request might influence or prolong a customers LOS for example.
They can reduce their own idel times and further increase the flow of patients if they want.
E.g. with patients at the ED, in a dedicated queue a physician would be only focused on their
patients, and in pooled queue physicians and nurses (triade) etc would share patients.
Dedicated queue, with increased ownership and responsibility of managing flow is able to
significantly decrease service time. Because physicians will be more ‘dedicated’ and also
because they get (financial) benefit from it.
Hopp & Lovejoy (2013) Management principles
Principle = comprehensive and fundamental law, doctrine or assumption. Principles are
highly general and stable. Principles relevant to hospital management:
- Flows: entities that move through the system as a consequence of the organisations’
function. (e.g. jobs, customers, transactions, patients etc) Speed and efficiency drive
its performance.
o A sequence of steps through which entities move to provide service.
o Key concerns about a hospital flow: volume (no. of patients), time (how long
is each process) and quality (how effective is the process).
o Capacity: it is the throughput it can achieve provided that it is never starved
for work. The most a person can work is 100% (being busy 100% of their
time), so once that capacity has been met (e.g. 6 patients per hour and
capacity is 6 6/6), that is all the person can handle. Capacity overloads (when
arrival rate exceeds capacity) cannot persist over the long term. This can be a
human resource, but also physical (like a room). However, physical resources
might handle stuff parallel instead of serially, and then you have to multiply
the capacity of an individual unit by the number of units in the resource. To
define the capacity of a complex system like an ED, one must know which
resource limits system capacity.
o Utilization: long-term fraction of time it is busy, given by the average arrival
rate of work at the station (e.g. patients per hr) divided by the resource
capacity over the long term. It is the same calculation as for capacity, but
then for long-term. A system with the highest utilization have the least
capacity to catch up from a capacity overload.
o Bottleneck: the resource in a system with the highest utilization. System
Capacity principle: the bottleneck will be the first one reaching 100%
utilization, and constrains the maximum rate of the system and hence defines
its capacity. The physician is the bottleneck in the example, because he/she
has the smallest capacity therefore the highest utilization. So then the
capacity of that physician will be the capacity of the ED (e.g. 4 patients).
- Information: organisations rely on explicit and implicit data and knowledge to
support operations (e.g. task description, customer orders etc).
- Human behavior: the operations of each organization are defined and implemented
by humans involved in them. Their behavior influences the performance.
2
Hummy Song, Anita L. Tucker, Karen L. Murrell (2015) The Diseconomies
of Queue Pooling: An Empirical Investigation of Emergency Department
Length of Stay. Management Science 61(12):3032-3053
Key objective for service organizations: improving efficiency and customer experience.
Queue management decisions are important in this, because they determine how long
customers have to wait for service.
Pooling separate streams of identical customers into a single queue of identical servers is
more efficient than a set of dedicated queues. Leads to shorter waiting times because
customers can be processed by any available server rather than waiting for one to become
available. However it may not always yield to expected performance improvements, e.g.
when you combine customers who have different needs. Also, the servers might be affected
if they get more customers assigned just because they are faster.
LOS = a measure of service time, starts with the time the physician begins delivering care to
the patient and ends with either a bed request for admission to the hospital, or the
discharge of a patient to their home or an outside facility. This includes both the active time
in which patients receive care, and the inactive time.
Round robin (RR) routing policy: assigning patients to physicians in which patients were
evenly distributed across physicians independent of physician speed or idle time. This was
done because physicians had fixed salaries, so no financial incentives. Fairness constraint.
A queueing system with a dedicated queuing system and fairness constraint has less waiting
and LOS time (than with pooling queue). Interviews with physicians suggest that, in this
context, the increased ownership that stems from a dedicated queuing system with a
fairness constraint leads to a situation in which the improvements in service rates due to
better flow management are greater than the variability-buffering benefits of a pooled
queuing system with a fairness constraint. Physicians change their behavior as a response to
their assigned responsibilities and ownership over the work routines and resources needed
to accomplish it.
Pooling is usually a good method to reduce costs, by reducing waiting times even when
work is allocated fairly among servers using a RR routing policy. Moreover, pooling with a
fairness constraint is even better than classical pooling, because when there are a lot of
customers arriving fast servers have an incentive to slow down their service rate in systems
which distribute customers based on server availability (instead of fair distribution).
However, sometimes the behavioural responses of servers and customers can reduce these
expected beenfits. This can happen because servers either speed up their service when
there are a lot of customers, or slow it down as to not have to deal with a lot. Or increasing
service time when there are not a lot of customers. Depends on financial incentives and
responsibilities (ownership) how a server will behave. Employees feel ownership when they
are given resources and responsibilities to manage the complete workflow of a meaningful
task.
Dedicated queuing systems with fairness constraint have higher ownership feeling then
pooling. Thusly they might be more motivated to efficiently manage their workload.
1
, Strategic servers: servers who have multiple options of service to give/ask of the customers,
and their choice of service/request might influence or prolong a customers LOS for example.
They can reduce their own idel times and further increase the flow of patients if they want.
E.g. with patients at the ED, in a dedicated queue a physician would be only focused on their
patients, and in pooled queue physicians and nurses (triade) etc would share patients.
Dedicated queue, with increased ownership and responsibility of managing flow is able to
significantly decrease service time. Because physicians will be more ‘dedicated’ and also
because they get (financial) benefit from it.
Hopp & Lovejoy (2013) Management principles
Principle = comprehensive and fundamental law, doctrine or assumption. Principles are
highly general and stable. Principles relevant to hospital management:
- Flows: entities that move through the system as a consequence of the organisations’
function. (e.g. jobs, customers, transactions, patients etc) Speed and efficiency drive
its performance.
o A sequence of steps through which entities move to provide service.
o Key concerns about a hospital flow: volume (no. of patients), time (how long
is each process) and quality (how effective is the process).
o Capacity: it is the throughput it can achieve provided that it is never starved
for work. The most a person can work is 100% (being busy 100% of their
time), so once that capacity has been met (e.g. 6 patients per hour and
capacity is 6 6/6), that is all the person can handle. Capacity overloads (when
arrival rate exceeds capacity) cannot persist over the long term. This can be a
human resource, but also physical (like a room). However, physical resources
might handle stuff parallel instead of serially, and then you have to multiply
the capacity of an individual unit by the number of units in the resource. To
define the capacity of a complex system like an ED, one must know which
resource limits system capacity.
o Utilization: long-term fraction of time it is busy, given by the average arrival
rate of work at the station (e.g. patients per hr) divided by the resource
capacity over the long term. It is the same calculation as for capacity, but
then for long-term. A system with the highest utilization have the least
capacity to catch up from a capacity overload.
o Bottleneck: the resource in a system with the highest utilization. System
Capacity principle: the bottleneck will be the first one reaching 100%
utilization, and constrains the maximum rate of the system and hence defines
its capacity. The physician is the bottleneck in the example, because he/she
has the smallest capacity therefore the highest utilization. So then the
capacity of that physician will be the capacity of the ED (e.g. 4 patients).
- Information: organisations rely on explicit and implicit data and knowledge to
support operations (e.g. task description, customer orders etc).
- Human behavior: the operations of each organization are defined and implemented
by humans involved in them. Their behavior influences the performance.
2
