Chapter 5 Simulation study
5.4 Experiments in the simulation
We described the base solution to represent the current situation in Section 5.3.3 and compared this to the historic data in Section 5.3.3. From this point we continue to simulate the alternatives proposed in Sections 4.2.1.1 - 4.2.1.5. In the following sections we describe how we implemented the proposed alternatives in the simulation.
5.4.1 Increase the number of cases scheduled
For this alternative we increase the number of cases scheduled per week by 10 every simulation run and look at the outcome measures. We increase the number from 600 cases scheduled per week to 750, which represents an increase of 25% compared to the current situation. This would be reached after four years in case of an annual growth of 6. A time horizon of four years is long enough, and a lot can change in the meanwhile, also on the demand side. When no change in the current system would occur and demand would grow with six percent annually we can at least predict when the current system will reach its maximum capacity.
One of the assumptions we make is that when growth occurs this happens equally among all the services. Foremost the simulation shows when the system is fully utilized and when the maximum capacity of the 7-day release program is reached with the current constraints.
5.4.2 Alter the release day
The effect of altering the release day can be modelled by changing the day that the rooms are released to the other services. The current situation is 7 days before the day of service, but this can be changed to 5, 6 ,8 9 or 10 days as described in Section 4.2.1.2. We changed this in the simulation program and kept the other constraints the same to show the difference. Since there is an arrival rate of the patients we also have to alter the two groups of patients that are generated, this is done according to Table 7. We did not have to make further assumptions. We only allowed to schedule on the
5.4.3 Vary the request day constraint
The requested day is always honored is the current motto. When the number of cases increase this motto is going to be harder to achieve without scheduling in overtime. In the simulation program we followed the scheduling approach of the current situation. When the case could not be scheduled this alternative is put into effect. So first we try to schedule according to the current situation, and thereafter instead of cancelling the case, the simulation program tries to schedule the case with the options given in Table 8. When the option of one day before and after is scheduled we first look at the day before and thereafter we look at the day after the requested day. With the option of all of the days within the 7-day release program, we look at all of the consecutive days, starting with the first one, when the requested day could not be honored.
5.4.4 Relaxation of the soft constraints of the rooms
In order to show the difference between the different room constraints we ease up these constraints one by one and show the difference per room(s). There might be a difference in the outcomes since the case mix is different for the different
ORs/services. The rooms for which we erase the constraint of not releasing the rooms
to other services (so we release the rooms). We will do this for the following ORs:
VOR 4 Ophthalmology
VOR 8 Neuro Interventional
VOR 31, VOR 32, VOR 33, VOR 34 Cardiac surgery
VOR21 Neurosurgery
VOR25 Burn (Monday Wednesday & Friday)
VOR 12 & VOR 13 Orthopedic Trauma
The restriction will still remain that ophthalmology, neuro interventional, cardiac surgery, neurosurgery, and burn can only schedule in their own rooms. Where orthopedic trauma can also be performed in a different OR. We hard coded this in the Block Schedule in the simulation program or hard coded the options in the scheduling algorithm. We did not change the scheduling procedure described in Section 5.3.3.
5.4.5 Change the scheduling policy
For the different scheduling options we consider to sort the patients every day by case duration before scheduling them. This should lead to a better fit. This is done for both the regular elective cases as for the staged cases.
The other option is to schedule according to the four options for the selection of the best suitable OR, namely: Best Fit (Dexter et al., 1999), First Fit (Hans et al., 2008), Level Fit, and Random Fit. Where Best Fit means the suitable OR with the least time available after the surgery, but not allowing overtime. First Fit, is the first available suitable OR. Level Fit tries to schedule the cases evenly over the ORs to create a utilization of the rooms that is as even as possible among the rooms. The Random Fit, randomly draws a room, and when it fits it is scheduled in this room, otherwise a new room is randomly drawn. We will not simulate the best fit, level fit and random fit last option. Since with the time horizon of a year to schedule the surgeries, this would yield an access time that is worse than the current access time.
As described in Section 4.2.1.5 we also eliminate the 7-day release schedule and don’t allow to schedule the cases outside the dedicated Block Time schedule. This means that in the simulation program we use the first fit schedule and only generate regular elective patients and do not release the rooms to any other service.
Due to time constraints we will not simulate the option of having a waiting list of for example two weeks, although this might be beneficial to VUMC, besides the time constraint we think this would harm the satisfaction of the patient. Because the patient would walk out of the clinic not knowing exactly on which day the surgery is going to be scheduled.
5.5 Conclusion
This chapter describes the conceptual model and how this conceptual model is translated in the simulation program. We demonstrated that with the assumptions ade to model the date we are able to simulate the current situation adequately. The next chapter shows how the alternative solutions are simulated and the results from the simulated current situation and alternative solutions.