Spreadsheet Models for Managers
Getting Access to Spreadsheet Models for Managers
If 
you use Excel to model businesses, business processes, or
business transactions, this course will change your life. You’ll learn how to create tools for yourself that will amaze
even you. Unrestricted use of this material is available in two ways.
- As a stand-alone Web site
- It resides on your computer, and you can use it anywhere. No need for Internet access.
- At this Web site
- If you have access to the Internet whenever you want to view this material, you can purchase on-line access. Unlimited usage. I’m constantly making improvements and you’ll get them as soon as they’re available.
To Order On Line
| Order "Spreadsheet Models for Managers, on-line edition, one month" by credit card, for USD 69.95 each, using our secure server, and receive download instructions by return email. |
| Order "Spreadsheet Models for Managers, on-line edition, three months" by credit card, for USD 199.00 each, using our secure server, and receive download instructions by return email. |
| Order "Spreadsheet Models for Managers, downloadable hyperbook edition" by credit card, for USD 199.00 each, using our secure server, and receive download instructions by return email. |
To Order by Mail
Make your check payable to Chaco Canyon Consulting, for the amount indicated:
|
And send it to: Chaco Canyon Consulting 700 Huron Avenue, Suite 19C Cambridge, MA 02138 |
To use the course software you’ll need some other applications, which you very probably already have. By placing your order, you’re confirming that you have the software you need, as described on this site.
Spreadsheet Models for Managers
Demonstrations for Session 12
Service Systems
We have two demonstrations (2007+) for this session:
Modeling Single Server Queues (2007+)
[Sheet: Example1]
This example shows a simple way, using names, to evaluate the formulas we’ve seen. By assigning values to the names ArvRate and SvcRate, which represent λ and μ respectively, we can create readable and understandable models of service systems. Perhaps you’re wondering why we don’t just use the names Lambda and Mu. We could use Lambda — that would be OK. But Mu presents problems, because it could be confused with the label for column MU of the worksheet. That’s why, in this course, we usually use ArvRate for λ and SvcRate for μ.
(a)
For this illustration we suppose that customers arrive at a Poisson-distributed rate of 20 per hour. Service times average 2.6 minutes, and are exponentially distributed. How long is the average waiting line?
This is a straightforward application of the formulas of the slide “Performance Measures for Single Server Systems” of the class notes (Office 2007+), or from the reading on Service Systems. We’re looking for the average length of all lines, even when there is no line. Looking at the readings, we find that the quantity we want is given by
Lq = λ2/(μ (μ - λ))
The formula is:
(b)
Excluding empty lines, what is the average line length?
Now the formula we want is:
La = μ/(μ - λ)
The formula is:
The Poisson Distribution (2007+)
[Sheet: Example2]
This example plots the shape of the Poisson distribution. It’s intended to give you a feel for what the arrival distribution looks like. The plot shows the probability of n arrivals on or before time t. The four series shown depict the distributions of the arrival of the nth customer by time t, for n = 0 to 3.
The plot for n = 0 shows the probability of zero arrivals as a function of time. In other words, for t=5 say, it answers the question “What’s the probability that we have no arrivals?” For high λ, this plot moves in towards 0 — more probability is concentrated toward 0. This happens because of the first factor of the Poisson distribution, (λt)k, which for k = 0 becomes (λt)0. At t = 0, this is 00 = 1. The second factor, e-λt, then prevails.
For other values of n, the two factors compete. Since the first is a simple power, and the second is an exponential, the second always wins, eventually. Thus the shape of the distribution for n > 0 is humped, with the maximum occurring at greater t as n increases.
Last Modified: Wednesday, 27-Apr-2016 04:15:26 EDT
Modeling service systems in general is extraordinarily complex, but as we’ve seen, if we make reasonable approximations, we can gain powerful tools that are very easy to apply. In the case of service systems, we assumed that the system was at equilibrium or close to it. Analogously, we can make simplifying assumptions for many other complex questions. Examples are process control, resource scheduling, resource allocation, cost allocation, vehicle routing, and many more.