Examining theme park throughput
Since I'm a software guy by trade, I'm generally fascinated with real world problems unlike the conceptual ones that I deal with regularly within the confines of computer memory. I feel pretty strongly if you spend some time learning about the kinds of problems that other engineering types are confronted with and how they go about solving them, you can apply a different thought process to your own work. As I just got back from my annual Walt Disney World vacation, I spent some time thinking about the queuing theory behind the throughput of theme park rides.
Generally speaking, the people at Disney know what they're doing when it comes to theme parks, if for no other reason than they've been doing it longer than everybody else. As such, their statistics tell them that customer satisfaction, not surprisingly, is directly related to the number of attractions a particular guest experiences during their day at a Disney theme park. They have also found that the amount of money someone is willing to spend at the shops on their way out is a function of their satisfaction.
So, if you get on more rides (generally around 7 to 9), you will spend more money on things besides your theme park admission.
How do you increase the chances that someone reaches that magic number and spends more money at the retail shops on the way out? Increase the throughput of the attractions, that's how. Moving more people through the rides means that more people get to experience them on a given day.
Anybody who has spent some time playing Roller Coaster Tycoon, or something similar, is familiar with this problem and the 3 ride styles mentioned below. Toying with details can make a big difference and give each guest a better chance to be satisfied.
Stop and go loaders
Examples: Dumbo or a Carousel
These types of rides tend to have the lowest throughput of them all and that's why you don't usually see to many of them. The limiting factor here is simple: nobody can get on the ride until everybody from the previous cycle has gotten off. Meaning, you can't add riders one at a time, you have to add them in bulk and you can only add them at specific times. This leads to extremely low unload and load times because of the coordination involved.
The loading can be assisted by counting out people in the queue who should be able to get on in the next loading cycle while the previous loading cycle is enjoying the ride, but the bigger limitation of mass load/unload still exists. Alternatively, the other way the throughput of a stop and go loader can be increased is by decreasing the ride time. That translates into more load/unload opportunities per hour, but runs the risk of a guest being dissatisfied with the ride they get.
Interval loaders
Examples: Roller coasters (most of them) and dark rides (most of them)
Most theme park rides are interval loaders. A vehicle comes into a station where the previous riders get off and a new set gets on. While this is going on, other cars are out on other sections of the track enjoying the ride. This is far more efficient than the stop and go loader because the waiting time cost of loading/unloading is spread out to different cars instead of paying that penalty all at once.
The most common optimization is to simply add more cars to the track so as to maximize the number of people on the ride at a time that are not in the loading zone. Another method is to separate the loading station from the unloading station so that the two can happen more easily in parallel. Less obvious is to have a longer track. The more space there is out on the ride experience, the more vehicles (and therefore riders) can be out on the attraction at any given moment. Or, make the vehicles bigger so that they carry more people (think Pirates of the Caribbean or It's a Small World), thereby minimizing the throughput penalty that is paid by the loading/unloading process.
Continuous loaders
Example: Haunted Mansion and Spaceship Earth
What happens if you add enough cars to an interval loader such that there is no space between the vehicles any more? You get a continuous loader. The track of the ride is completely occupied by as many cars as it could possibly carry, maximizing the number of people experiencing the ride at once (albeit riders experience different portions of the ride at the same time).
Disney pioneered this style of ride at Disneyland with its Omnimover system originally used in Adventure Thru InnerSpace (now defunct) and The Haunted Mansion (not only still active, but with a version at almost every Disney resort around the world). The downside of the system is it requires a slower moving ride, since all cars (even the ones being loaded) have to move at the same speed, and typically a more elaborate loading/unloading system so that the cycle of vehicles never has to stop.
At the Haunted Mansion, for example, riders first step onto a moving ramp that is moving at the same speed as the vehicles. Guests have the length of this ramp to get into the assigned vehicle, as opposed to having to quickly jump into a moving one if the ramp were not provided.
Final thoughts
While I'll be the first to admit to being a huge Disney geek, theme park ride throughput presents an interesting problem to solve. You have to provide some pleasant experience, but you need to move as many people as possible through that experience to maximize the number of riders you might have on a single day. In a way, it's not all that different from memory utilization problems that software engineers like myself face (with loading/unloading being analogous to a cache or supplemental swap files out on hard disk) and presents some creative ways to minimize waiting time, increasing the chances that more park guests come away satisfied with their day.
Generally speaking, the people at Disney know what they're doing when it comes to theme parks, if for no other reason than they've been doing it longer than everybody else. As such, their statistics tell them that customer satisfaction, not surprisingly, is directly related to the number of attractions a particular guest experiences during their day at a Disney theme park. They have also found that the amount of money someone is willing to spend at the shops on their way out is a function of their satisfaction.
So, if you get on more rides (generally around 7 to 9), you will spend more money on things besides your theme park admission.
How do you increase the chances that someone reaches that magic number and spends more money at the retail shops on the way out? Increase the throughput of the attractions, that's how. Moving more people through the rides means that more people get to experience them on a given day.
Anybody who has spent some time playing Roller Coaster Tycoon, or something similar, is familiar with this problem and the 3 ride styles mentioned below. Toying with details can make a big difference and give each guest a better chance to be satisfied.
Stop and go loaders
Examples: Dumbo or a Carousel
These types of rides tend to have the lowest throughput of them all and that's why you don't usually see to many of them. The limiting factor here is simple: nobody can get on the ride until everybody from the previous cycle has gotten off. Meaning, you can't add riders one at a time, you have to add them in bulk and you can only add them at specific times. This leads to extremely low unload and load times because of the coordination involved.
The loading can be assisted by counting out people in the queue who should be able to get on in the next loading cycle while the previous loading cycle is enjoying the ride, but the bigger limitation of mass load/unload still exists. Alternatively, the other way the throughput of a stop and go loader can be increased is by decreasing the ride time. That translates into more load/unload opportunities per hour, but runs the risk of a guest being dissatisfied with the ride they get.
Interval loaders
Examples: Roller coasters (most of them) and dark rides (most of them)
Most theme park rides are interval loaders. A vehicle comes into a station where the previous riders get off and a new set gets on. While this is going on, other cars are out on other sections of the track enjoying the ride. This is far more efficient than the stop and go loader because the waiting time cost of loading/unloading is spread out to different cars instead of paying that penalty all at once.
The most common optimization is to simply add more cars to the track so as to maximize the number of people on the ride at a time that are not in the loading zone. Another method is to separate the loading station from the unloading station so that the two can happen more easily in parallel. Less obvious is to have a longer track. The more space there is out on the ride experience, the more vehicles (and therefore riders) can be out on the attraction at any given moment. Or, make the vehicles bigger so that they carry more people (think Pirates of the Caribbean or It's a Small World), thereby minimizing the throughput penalty that is paid by the loading/unloading process.
Continuous loaders
Example: Haunted Mansion and Spaceship Earth
What happens if you add enough cars to an interval loader such that there is no space between the vehicles any more? You get a continuous loader. The track of the ride is completely occupied by as many cars as it could possibly carry, maximizing the number of people experiencing the ride at once (albeit riders experience different portions of the ride at the same time).
Disney pioneered this style of ride at Disneyland with its Omnimover system originally used in Adventure Thru InnerSpace (now defunct) and The Haunted Mansion (not only still active, but with a version at almost every Disney resort around the world). The downside of the system is it requires a slower moving ride, since all cars (even the ones being loaded) have to move at the same speed, and typically a more elaborate loading/unloading system so that the cycle of vehicles never has to stop.
At the Haunted Mansion, for example, riders first step onto a moving ramp that is moving at the same speed as the vehicles. Guests have the length of this ramp to get into the assigned vehicle, as opposed to having to quickly jump into a moving one if the ramp were not provided.
Final thoughts
While I'll be the first to admit to being a huge Disney geek, theme park ride throughput presents an interesting problem to solve. You have to provide some pleasant experience, but you need to move as many people as possible through that experience to maximize the number of riders you might have on a single day. In a way, it's not all that different from memory utilization problems that software engineers like myself face (with loading/unloading being analogous to a cache or supplemental swap files out on hard disk) and presents some creative ways to minimize waiting time, increasing the chances that more park guests come away satisfied with their day.
Labels: General stuff
2 Comments:
My spelling moronics continue. Fixed the title, "troughput", I'm clearly an idiot.
I *was* wondering what 'troughput' was, at first glance. Some seriously technical jargon, I decided.
After realizing what you meant, I figured I would cut you some slack; you were after all on vacation...
Post a Comment
<< Home