Due Monday, April 28, in class. .

Continuing with the elevator simulation - a little bit at a time.

For ongoing discussions, be sure to watch the forum and some $CS310/elevators/thoughts.html.

• Begin by writing a short .els file that specifies one elevator in a building with just two floors. Then you don't have to worry about choosing a random number to determine which floor a Passenger leaves at. Hard code the arrival of just a few passengers at specific times. Don't use random number generators to create new PassengerArrival events.

  Choose simple numbers for the travel and door opening and closing and entry and exit times in the input file so that you can work out in advance exactly what the sequence of events should be.

  Imagine and code the simplest possible elevator logic.

• Start instrumenting your simulator so that it produces output statistics when requested. Here's a possible syntax. You can add to it a little at a time while you're thinking about harder problems. But it's important to start it soon so that you can use it for debugging as you develop the simulation.

  Elevator Simulation
  
  (date)
  
  Simulation input parameters:
  [choose some reasonable syntax - perhaps the one from ElsData.toString()]
  
  Trip statistics:
  	Passenger arrival rate      	   average trip time
  floor	expected	actual  	from main	to main
  0	xxx		xxx		xxx		xxx
  1	xxx		xxx		xxx		xxx
  2	xxx		xxx		0		0
  3	xxx		xxx		xxx		xxx
  ...
  
  Simulation statistics:
  Simulated time: xxx
  Elapsed time:   xxx
  CPU time:       xxx
  Number of trips: xxx
  Number of Events scheduled: xxx
  Number of Events created: xxx
  
  

• Introduce randomness into Passenger arrivals. If Passengers are supposed to arrive at the Elevator on a floor at a rate of R/second then the average time between arrivals should be 1/R seconds. Each PassengerArrivalEvent should schedule the next one about that many seconds later. Choose the random number with the right average value from an exponential distribution. We'll discuss exponential distributions in class. You will find code for generating random numbers from such a distribution in $CS680/elevators/Exponential.java.

• Draw conclusions from your simulator. Here are some questions you will want to try to answer:
  • How long (in simulated time) do you need to run the simulation so that the results are reliable - that is, so that repeated runs produce average trip times (for each particular trip) that are close enough to each other so that you can draw conclusions.
  • Do the results from your simulator match those from a classmate's?
  • How well do the results of the simulation match your experience in the Healey Library?
  • When you fix all the parameters except the input rate, how does average waiting time for a particular trip vary as arrival rate varies?
  • What arrival rate saturates the system (so that the number of passengers waiting at each floor just keeps growing without bound and the average trip time is infinite)?
  • Which of the input parameters are the most critical? (Elevator speed? time to open and close doors? time for passengers to enter and leave?)
  • How does the choice of the elevator dispatch algorithm affect passenger trip times?