Designing a Model in Arena® Simulation Software

by Saurabh Gupta

As part of Simulation Modeling course work for Masters of Science (IT) from Lindner College of Business, University of Cincinnati under Prof. David Kelton

Revisiting

The model was desired to simulate multiple tasks taken up by front desk representatives in a regular patient-care clinic. The person at front desk greets the patients and answers phone calls. Also, performs some scheduled tasks, over the course of the week, like bill preparation, credentialing, Laundry etc. The design goal of the project was to build a model which is able to handle scheduled tasks for the day and also incoming random patient walk-ins and phone calls. Also, one of the other secondary goals were to allow for model to adjust itself to new tasks or resources. All this required applying concepts of programming in a different paradigm.

Situation

Cheryl, Phyllis and Renae have their schedule fixed. Renae is scheduled to work 0900hrs to 1500hrs from Monday to Saturday. Cheryl and Phyllis work alternatively from Monday to Saturday in same time schedule 0900hrs to 1500hrs. This is well defined in table below.

Resource

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Cheryl

o

o

o

o

Phyllis

o

o

Renae

o

o

o

o

o

o

Clinic hours (12:00 to 12:30 break)

0900hrs to 1500hrs

0900hrs to 1500hrs

0900hrs to 1500hrs

0900hrs to 1500hrs

0900hrs to 1500hrs

0900hrs to 1500hrs

Table 1 – Day of Week vs. Resource Matrix

Scheduled tasks such as billing, credentialing, changing beds etc. are owned by each individual. Following are some of the defined tasks and their owners –

Scheduled Task

Cheryl

Phyllis

Renae

Credentialing

o

o

Changing Beds

o

o

Laundry

o

o

Statements

o

Super Bill entry

o

Training

o

Table 2a – Tasks vs. Resource Matrix

However, unscheduled task such as patient walk-ins, phone calls and PT help may or may not come to them.

Unscheduled Task

Cheryl

Phyllis

Renae

Attending to patient walk-ins

o

o

o

Attending to patient call-ins

o

o

o

Helping patients in clinic

o

o

o

Table 2b – Tasks vs. Resource Matrix

Each day, not more than 2 people work at clinic. Renae works everyday while Cheryl and Phyllis alternate among themselves. Following table shows task to be taken by each resource for given working day –

Day of the week Resource 1 Resource 2
Monday Credentialing, Changing Bed, Laundry Statement, Super Bill entry
Tuesday Credentialing, Changing Bed, Laundry Statement, Super Bill entry
Wednesday Credentialing, Changing Bed, Laundry Statement, Super Bill entry
Thursday Credentialing, Changing Bed, Laundry Statement, Super Bill entry
Friday Credentialing, Changing Bed, Laundry Statement, Super Bill entry
Saturday Credentialing, Laundry Training

Table 3 – Day of Week vs. Tasks to be Completed Matrix

Design Challenges

We defined following challenges when beginning to model –

High level challenges –
  1. Develop a flexible task matrix – thus allowing user to add/remove tasks assigned to each resource without modifying the core model
  2. Dynamic resource allocation – Ability to increase number of resources and also change their schedule
Low level challenges –
  1. Design two different processes, which will interact with same set of resources i.e. allocate resources to a task from two different task set , which are of different nature (Table 2a and Table 2b)
  2. Have the model chose the resource based on his/her availability
  3. Assign a task to resources based upon ownership and priority of task from a set (Task set as defined in Table 2a and Table 2b. Day and Priority set as defined in Table 3)
  4. Release and count the entities which were waiting in queue by end of day
  5. Collect metrics like how many tasks were left pending at the end of day, what was the overall utilization of each resource (scheduled vs. instantaneous)

Design Approach

Patient walk-ins and patient calls were two primary tasks for resource at front desk. Both of these entities were created using Non stationary Poisson process – thus simulating real world scenario of patient and phone calls arrival without marked schedule. This task is then assigned to resource based on their availability and also tie-breaker rule.

Two entities, for each resource, were created at the start of the day and was then disposed at the end of the day. These entities were then assigned to respective resource (Phyllis, Cheryl or Renae) as per their availability for that day of the week as defined in Table 1 (Day of Week vs. Resource matrix). Later, this entity was made to go through tasks based on following conditions:

  1. Task ownership as defined in Tasks vs. Resource Matrix (Table 2b)
  2. Tasks to be performed for given work day as defined in Day of Week vs. Tasks to be Completed Matrix (Table 3)
  3. Order of priority as defined in 2nd column of Day of Week vs. Tasks to be Completed Matrix (Table 3)

Priorities defined for Resources in-case of tie

  1. Pick incoming phone call
  2. Greet incoming patients and help patients in treatment if called for
  3. Carry on with daily tasks

Models Designed

Handling Patients and Phone Calls

Patients and phone calls are handled by front desk with priority as

Creation of “Resource” as entity

“Sunrise” entity is split between two Resources are assigned respective tasks for the day (as per the Week day vs. Task list matrix – Table 3). Then they are forwarded, using “Proceed Towards” block, to respective stations as shown in next model (following the one below)

Station for each defined task in

The resource entity arrives at each of these stations as per the Task vs. Resource matrix (Table 2) and Day of Week vs. Task matrix (Table 3)