Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Application of Queuing Theory to Determine Terminal Scope 1

Status
Not open for further replies.

GaTechTheron

Mechanical
Jan 26, 2006
110
We are expanding a terminal to account for a processing facility that we are building. There are some areas that we can save money by combining long lines to docks/rail/truck loading (with multiple different product loading streams) at the expense of product quality / line cleaning.

To address the scope of terminal expansion requirements, we would like to know if there is a software program out there than can apply queuing theory to tank terminal opertions to determine what is the most effecient scope that should be installed.

Additionally, can such systems do a sensitivity analsysis so that we understand how differences in the simulation inputs, affect terminal operations (i.e. risk of demmurge cost in current design as the campaigns of the processing facility and natural weather and other delays affect operations).

This will be hugely beneficial to understand how this is done to determine terminal scope, manpowe requirements, and ultimately CAPEX & OPEX budgeting for the new facilities.

Thanks,
Theron
 
Replies continue below

Recommended for you

Queuing theory can tell you the number of tanks, rail cars, or trucks, etc. you might need for each pipeline. Running n number of simultaneous multiple pipelines would give you an idea of how many pipelines you would need.

I once considered buying such a program a long time ago and at that time there were one or two programs I found that said that they do it, but I don't see them around now. I decided I could solve the problem myself. In any case you could set this up using a discrete event simulator, of which there are some free open sources for that you can download.

I have found Monte Carlo simulation and genetic algorithms both useful for this type of analysis for multiple product distribution scenarios. Genetic algorithms can determine batching schedules by working out plausible series of pipeline system "states" resulting in best delivery schedules. A number sequence (DNA) representing individual sequential system states, each of a fixed period of time of system operation, where for instance by 1=gasoline, 2=diesel, 3=jet fuel, etc. can be developed by a programed genetic cross-over and selection of the better sequences. Good criteria are those that deliver acceptable volumes of each products in the largest batches, yet fitting into the tanks or vessels available at the time, etc. After running it for awhile time, reasonable solutions can be achieved. The trick is to write good criteria for the selection procedure that results in good solutions.

Monte Carlo simulations can break down probable demand schedules for multiple products, say stated as various normal, or other types of distribution curves of probable demands for each product into confidence levels of passing varying volumes and combinations products through the same, or multiple, pipelines.

If you have only a few products and normal distributions of each product, it's actually pretty easy (based on the central limit theorem) to develop the capabilities without actually doing any GA or MC, as the capacity of each pipe is roughly the sum of all curves. The number of cars you need is the finished curve / volume of a car, etc.

If you don't find a canned program already set up for DT simulation, and you don't want to go DIY, these guys at Lanner WITNESS offer Witness, which is a highly versatile program that is capable of simulating almost any process composed of a sequence of events and will probably be able to set up your terminal model as well, if you have a budget available to do it.

If you want to try DIY, it's probably not as complex as you think, at least the basics, and this paper might get you off to a good start,
 
Thanks BigInch. You never cease to amaze me.

It will take me some time to digest all of this information, and determine if I can relatively easily apply it to this scenario (as DIY). I'll have to revert.

Thanks,
Theron
 
I believe your problem is beyond the level of a simple software analyzing queing. You can prepare a simple spreadsheet for queing, like in a bank line.


However, your problem should be analyzed with 6 sigma. Six Sigma is a set of techniques and tools for process improvement. It was developed by Motorola and is used in many industrial sectors.


The problem involves queing, debottlenecking, production processes, quality, etc., not just queing.
 
bigInch,

I have seen these terminal management softwares (specifically the one that Honeywell sells), but this is not the type of problem I am trying to solve (at the moment). The problem I have now is... do we build 10 new dock lines, or can we get away with 1 dock line, and flush after every load-out? Can I use my swing tank for a new product from a new campaign, or must I build another tank? These are my challeneges that are not 100% clear based on facilities today.

We can opt to spend $30M to expand the terminal with dedicated facilities for every new product, or we can spend $5m to expand the terminal such that products share the facilities (but OPEX is increased from flushing & increase operator presence).

The answer to the question is going to depend on the frequency, size, and location of shipments....

I'll take a look at some of what you send over, but we are in the middle of preping for commissioning another facility at the moment... (just trying to get ahead of the game).

Thanks,
Theron
 
Arena modeling from Rockwell will do what you are describing:




Document68_ipibms.jpg
 
Right. It doesn't take very long at all to pay for dedicated lines when faced with loss of revenue from handling, cutting interfaces and product downgrading. It can depend a lot on the products you want to run, their quality standards and which can be batched together with the smallest loss of value. You might find it works batching a couple of little used compatible products, but then have dedicated lines for the higher volume or incompatible products. In fact those screening factors right there just might reduce the problem to an easy solution.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor