• Welcome to SC4 Devotion Forum Archives.

new traffic experiments

Started by ldog, October 23, 2009, 06:16:28 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

ldog

Sorry for the doublepost.
I'm trying to find some discussion about te lots, specifically entry cost and capacity.
I don't want to necro the sticky in the NAM section from Mott, especially since I seem to recall reading far more recent discussion on the topic.

Specificly I've been wondering but too busy to ask; has a consensus been reached on what formulas to use for these values?
Last I remember reading it seemed to be heading that way.

IIRC, I wound up going with .96/speed for in-line stations (el, mono, etc) and I wanna say .02 for off-line (bus, train, etc) which seemed to be what was heading towards becoming the prevailing standard. I haven't touched the RTMT lately since they would throw my tests off (even though I would love to use them...they just make life so much easier). Although since I've got so many of y'all RTMT team members here, let me ask, are y'all satisfied that the stations act identicly to the network segment they are on or is that still an ongoing tweaking process? In other words, that road-bus station, does it impact the traffic up, down or act just like a normal road segment does?

I also remember seeing capacity suggestions but not any kind of formulas or even a good rule of thumb. I've seen a lot of Steve's capacity suggestions but I take them in the context of simulator Z's network capacitys as well; in other words they are too high for my current scenario. As I said I am using default station caps currently for testing, but I'm always considering the future.

Oh, and thanks btw Steve for that confirmation about the trip starting cost stuff :)

z

I did a large series of experiments addressing just this question, and I have been spending the past few days writing them up.  They should appear in my development thread late tonight.  Both of you might want to see what these show before constructing new experiments.

b22rian

Quote from: z on November 06, 2009, 04:42:11 PM
I did a large series of experiments addressing just this question, and I have been spending the past few days writing them up.  They should appear in my development thread late tonight.  Both of you might want to see what these show before constructing new experiments.

Ok great Steve,

Ive been looking forward to this for awhile now.. :satisfied:

ldog

#123
Quote from: jplumbley on November 06, 2009, 04:23:21 PM
Hey Lenny... I am curious about this PH thing... You guys have proved in Simulator Z that it causes abandonment at higher values.  I still do not get abandonment in the way you guys are showing, but of course I am using my Simulator A, is it possible for you to test a City with Simulator A and show this effect happening?

The thing I am wondering is if the "perfect" pathfinding value is effected by other properties as well, which may or may not make this value a value that changes from one Simulator to the next.  The speeds will probably have an effect on this, maybe max commute time as well?  A and Z have some very different values in them, and maybe something in Simulator Z is making the PH value more sensitive, or maybe my play style just fits well with Simulator A, that I simply don't run into the same issue you guys have found.

I just want to understand what the difference is and why it effects you and not me.

Remember, I am not using Z. I am using "L alpha .003" which at the moment is something between vanilla and B but with .003 ph. My speeds are all vanilla, I have a max commute of 24, mt commute of 10. I adjusted the CvS, ICE (intersection congestion effects) and transit type preferences to values that are different than all other simulators but I consider them pretty conservative.

I was not out to prove anything about the PH when I started these tests; I had considered it an issue I was going to put on the shelf for the moment. However observing the abandonment and the inevitable suggestion of lower the ph (remember I started at .009) which I decided to do since I was intrigued at this point, and I figured if nothing else it would stop Steve and Brian from trying to "convert me to .003" :P . I'm pretty confidant if I drop A in I will observe the same effect.

I do have a reason why I agree with Steve about this being a bug AND that I can believe you when you say you don't see abandonment problems with A yourself.
When you have 2 different things that you know both to be true and yet they conflict with each other then you have to bridge the gap.
What you and Steve have been debating are actually 2 seperate issues.

Much like you no doubt, whenever I had abandonment prior I considered it for the same reason; obviously this is a sign from the traffic gods that my network sucks, and I must make amends. So shoring up deficiencys in my network and/or zone placement and the traffic gods smile upon me again; the abandonment resolves itself.  If you don't lay out your city well then you should have problems. Where and when this is the case you and I are in total agreement.

However what Steve has been asserting all this time, and what I am observing at the moment as well has nothing to do with this. My network is adequate for a city easily 10 times the size it is (probably more even). There is NO congestion anywhere. I have highway and rail going from each section. The longest commute distance I would roughly estimate at about 500 tiles (there and back). Just like A my parameters should allow a full 1024 tile commute on an avenue (in fact you should be able to do 967 tiles on an uncongested road), nevermind a highway or rail. Obviously something is amiss here.

Now normally I play in a similar mindframe to you. In building these test cases I am trying to very quickly build very large citys so I am playing in an artificial style (for myself at least). Had I been playing normally I either A: wouldn't have noticed the abandonment or B: written it off as poor citybuilding on my part, but it would never occur to me that the traffic simulator was to blame.

Now it isn't a very serious bug I suppose; as the consensus seems to be a bit of abandonment is tolerable, even if it is for no good you could get banned for thating reason (LOL, I just had to see what the profanity filter was going to do with that one...I admit it is something I have been wondering for weeks now...very clever it is). A bug it is nonetheless. At a minimum it means there are other things in play here that none of us understand. Although lets see what Steve has to show us, maybe his new theory panned out and he figured it out.

If Steve doesn't answer all our questions then I would be happy to try it out with A, Z and whatever else y'all would like. Although it may take me a few days to get to it.

ldog

Well, having read http://sc4devotion.com/forums/index.php?topic=5382.msg286055#msg286055 sheds a bit of light on the situation.
Of course now that leaves us with how to tune according to the newly discovered relationship.
I still think the traffic effect and trip length effect values in the developer exemplars tie in to this somehow, but I am not even about to start playing with them at this point in time.
They have some complicated looking values and I don't have a clue where to start.
It makes sense if you think about it; how can you make the commute time affect desireability without causing abandonment. Also it has to be scaled to the different wealth levels of course.

So today I finally isolated my crash and burn problem to *drum roll* the no-cd crack I was using. Imagine that </sarcasm>. Luckily I always backup the non-cracked exe for times like these. Hasn't crashed once all day. The convenience of not mounting the disc is outweighed by the convenience of not rebooting the computer every 5 minutes. Which was the point it was getting to around 200k pop.

Still getting quite a bit of new abandonment, even with .003 ph but then max commute is 24. Traffic is still not bad enough in my opinion. I expect to see the traffic get far worse before it starts causing abandonment like this. The situation of course is worse since I got a bunch of overeducated shits for sims. My fault for making the city too nice. So off to raise the max commute and see how that goes. Bus and train use is still not high enough (to get the kind of data I am looking for), I have the odd stations that are a bit over 100% use but most of them are very low.

A couple recent thoughts on buses, but still not enough useful data so this is still theory. Steve's new Euro sim and the total lack of change in bus use so far shown by testers got me thinking. If the bus is effected by congestion then the speeds should be set higher...I think somewhere between whatever your car speeds are as the lowend and the original Maxis ratio car/bus speed as the top, especially if you want to get more use. If leaving the bus not effected by congestion, then I think the speed needs to be lowered, max the same as the car, probably lower. This isn't backed up by any kind of testing; it is just what I think is a good set of values to start testing within.

I'm also thinking I need to drop caps back down a bit, at least for testing purposes.

catty

Quote from: ldog on November 07, 2009, 05:31:54 PM
So today I finally isolated my crash and burn problem to *drum roll* the no-cd crack I was using. Imagine that </sarcasm>. Luckily I always backup the non-cracked exe for times like these. Hasn't crashed once all day. The convenience of not mounting the disc is outweighed by the convenience of not rebooting the computer every 5 minutes. Which was the point it was getting to around 200k pop.

I use Alcohol 52% to play SimCity and never have had a problem with it

QuoteThis CD & DVD emulation software allows users to create a virtual CD / DVD drives, and play CDs & DVDs without the need for the physical disc

its €22.30 EUR or 27.00 USD to buy and the home page is

http://www.alcohol-soft.com/

:)
I meant," said Ipslore bitterly, "what is there in this world that truly makes living worthwhile?" DEATH thought about it. "CATS," he said eventually, "CATS ARE NICE.

ldog

#126
Quote from: jplumbley on November 07, 2009, 05:39:30 PM
I was only looking for a test from an unbiased tester is all, and you are that guy.  I did not realize you were using your "Sim L".  There are many reasons why I may not get the same results, one being the way I play, another being that I don't grow to millions of Sims in one city tile, others may be other differences within the Simulator values.  And, since I have problems at the PH value of .009 in achieving the results you guys have found, then I need someone I trust has a good understanding of what is happening.

So please dont take the request as me "doubting" your or Steve's observations.  I am curious, but fail to achieve the same abandonment issue.

I have not been trying to argue Steve's assertions, just trying to understand them more clearly.  It is difficult to get unbiased answers out of him.  I want to understand why I do not find the same problems as he does, not refute his assertions.  To me it seems that play style has a big effect on how the PH affect the pathfinding system since we see two completely different outcomes.  And if that is the case, then obviously there could very well be differing values of the PH which would determine "perfect" pathfinding depending on play styles or maybe even "difficulty".  And again if this is the case, and it does show that more "efficiently" designed networks systems can satisfy higher PHs that maybe this would be what determines the "difficulty" of the Simulator.  But, then since everything effects everything else in the Simulator, maybe there are other things that effect the sensitivity of the abandonment issue you guys have discovered.

The thing is, there were many people that used Simulator A and did not have this problem when using .009, atleast not in the scale Steve showed in his post earlier this week.  One being Nate, who posted his results for Steve.  I think there is still a lot that needs to be discovered on this issue, beyond the fact that it happens, but in the realm of why it happens in one city but not another.  I am just trying to analyse and throw out some ideas on what is occuring.  And one thing that jumps out is the play style.

I understand. I will try to go ahead and run it tomorrow. You should definitly read Steve's big post from last night if you haven't though. Nevermind, you did. Saw your response.

I am sure playstyle has a lot to do with it. Based on our conversations you and I share playstyle "ethics" but remember I have much less experience at playing the game than you or most people. The longer you have been playing this game the less choices you had. You played the game pretty much on vanilla and mods came out slowly I would imagine. When you got new mods you had time to absorb them. Y'all also didn't have all the tools available that we do now. And then of course the current mods are not the mods they were a few years ago. Everything has evolved. This would give you a far different perspective than I have. Such as it is, I have less preconcieved notions based on how I used to play and how things should act. For example, you probably place stations, entrance ramps, bus stops, etc at certain intervals you know work under your usual conditions. Then you can compare the effects of whatever changes you made to how things used to work and how they now work. I am just placing them where I think should be a reasonable distance, not by the way the game works, or has worked, but by the way I am trying to make it work. Also right now since I am trying to very quickly create a big city, I am slapping things down in a somewhat crazy fashion. The city costs about 70k a month to run, it only brings in 20k...you do the math. This is not normal play, not by any sense of the word I can imagine.

Once again you gotta remember, Perfect pathfinding heuristic is not a feel good value, it is a strictly defined absolute value. "There can be only one!". One we still don't know for that matter.
The question you mean to ask is "What is the proper number to use for ph, and in what circumstance?"
Even with Steve proving the effect of the ph and then tieing it to max commute time, and my results own tests validating that (to me at least) that still does not answer that question definitively (again to me at least). Your entire second paragraph is still very pertinent thought process. Right now I am concerned with trying to get together some kind of unified theory of how the fool thing works and what kind of options we really do or don't have with it. Even though I am screaming "BUG! Run for your lives! Aghhhhh!" right now, I haven't ruled out that in the end this may be the only way to set a difficulty level. What sense is there in having all red congestion everywhere and not having to suffer anything from it besides the annoying "a chaos of cars" constantly?

ldog

Quote from: catty on November 07, 2009, 06:24:03 PM
I use Alcohol 52% to play SimCity and never have had a problem with it

its €22.30 EUR or 27.00 USD to buy and the home page is

http://www.alcohol-soft.com/

:)

Daemon tools lite here, free ;)
I alternate between both programs though.

Sometimes I am far too lazy even for that lol. When I built this computer I went a bit nuts. I got 3x 146 GB WD Raptors in a stripe, it cost an arm and a leg compared to probably the terrabytes I could have got in 7200k RPM drive, but I got a need for hd speed. I also try to keep it always under half full so space is always at a premium. The ESATA drive I have for archives tends to overheat and screw up frequently so I don't leave it on when I don't need it.

But yeah, it was a much simpler solution to just throw the isos on my C:\ than deal with all that :D

b22rian

Quote from: ldog on November 07, 2009, 06:37:40 PM

I am sure playstyle has a lot to do with it. Based on our conversations you and I share playstyle "ethics" but remember I have much less experience at playing the game than you or most people. The longer you have been playing this game the less choices you had. You played the game pretty much on vanilla and mods came out slowly I would imagine. When you got new mods you had time to absorb them. Y'all also didn't have all the tools available that we do now. And then of course the current mods are not the mods they were a few years ago. Everything has evolved. This would give you a far different perspective than I have. Such as it is, I have less preconcieved notions based on how I used to play and how things should act. For example, you probably place stations, entrance ramps, bus stops, etc at certain intervals you know work under your usual conditions. Then you can compare the effects of whatever changes you made to how things used to work and how they now work. I am just placing them where I think should be a reasonable distance, not by the way the game works, or has worked, but by the way I am trying to make it work. Also right now since I am trying to very quickly create a big city, I am slapping things down in a somewhat crazy fashion. The city costs about 70k a month to run, it only brings in 20k...you do the math. This is not normal play, not by any sense of the word I can imagine.



Yup this is quite an important point you raised here.. experience playing the game means a lot..
I have improved a lot in many areas of the game , just from playing quite a bit and the old  (trial and error)
system still works believe it or not !

Thanks , brian

ldog

Quote from: b22rian on November 10, 2009, 03:41:25 AM
Yup this is quite an important point you raised here.. experience playing the game means a lot..
I have improved a lot in many areas of the game , just from playing quite a bit and the old  (trial and error)
system still works believe it or not !

Thanks , brian

Yeah, that is why I have been (relatively) quiet lately. I'm spending a lot more time testing than in the reader now.
I just came to the conclusion that transit enabling lots just for the hell of it is pretty assinine. Sure it looks cool to have cars driving in and out of them but it screws with the simulator.
Some people may not care, and that's fine for them. Myself, I'd prefer to have things that look cool but don't break the game. So I've been doing a bit of editing of quite a number of my favorite lots.

What else, what else? I've found settings that I pretty much like so just been trying to put them through some more rigorous testing, and tweaking the odd bit here and there to learn more about how things interact.

Ahhh! I took what Steve did with the clean air ordinance, and I dropped the var into the car smogging ordinance set it to .7 (I started at .9 and then dropped from there a few times) and removed the default effect. It looks promising actually. The overall citywide pollution reduction of the car smogging act remains the same at that level, but it only changes the traffic pollution and not the industrial. It still needs some testing to see that the effect remains consistent but so far I am very pleased with the result. It is subtle but noticeable; it does not remove all the traffic pollution but you can see the colors change to a more muted yellow and the radius decreases a bit. If you had the car smogging ordinance enabled prior to making the change, the air pollution graph stays about the same level (if you didn't have it on then you will of course see a reduction when you do).

ldog

Well in light of Steve's recent experiments regarding the max commute time, I started thinking about how I could jack the commute time up and still keep the same overall balance I was looking for.
So I thought let me try lowering speeds. Talking walking down to 1, we then divide everything else by 3.5 . I then multiply my max commute and mt commute by 3.5 giving me 105 and 70.

Things worked pretty much as expected, there was a slight drop in MT use, but since I rounded everything up to keep whole numbers (train I might have rounded down...am at work), and so highway got a boost to 24 (23.4 was calculated) most of the others came in closer and their rounding only added 0.15, so it was not a surprise.

Now I went back and redid my stations patch, demolished all my stations and replaced them with stations that were updated to new values. MT use plummeted. I was puzzled for a bit but then I realized prior I was using .02 for which is not .96/pedspeed but a value that others (Cogeo?) found works as a compromise. Using .96/1 (which is .96) of course was disastrous. So for tonight I will change it to .07 and try that out, or maybe I will just leave it.

Another thing I am...
... we had a server blowup in the meantime and here it is 2 hours later from where I left off  and I forgot what else I was going on about.
Hopefully this post is coherent, I'm heading home ;)

ldog

#131
So I got home and instead of playing I got to rereading the infamous te switches and you discussion.
I also got to thinking more about the effects of the prefered commute strategys, the trip starting cost modifiers and the max mt commute time.
Instead of doing more testing I wound up opening up Excel and trying to chart out some commutes.

I really wonder that the max mt commute time is not what it has been described as by various people but instead exactly what it says it is; the commute trip max time for an mt prefered sim.
The math proves the above can't be right; the bus especially would never be used over a car, even with the higher default speed ratio; nevermind A & Z with them lower
I have tried it out with keeping the original Maxis ratio (6:4 so 24 and 16, 30 and 20) I have tried setting it the same as my max commute, I have even tried with it set at 10 (when I had regular max commute time of 24 or 30) when I was taking it as Jason had said somewhere that it was "the max time sims will spend walking to stations"...I thought 10 seemed reasonable since it meant they would walk 35 tiles, so divide by 2 for morning/evening, divide by 2 again for each end (home to station, station to work) which meant to me about an 8 tile walk on each leg. In game it has been pretty hard to pin down the exact effect, there are just too many variables in play. Well at least I got it narrowed down to either A. The max distance sims will walk to stations or B. The max amount of time they will spend on the bus/rail.
Having found that commute time calculations only apply to morning commute explains why I could not determine anything consistently. Research on this continues. and on the next page we find out Max MT commute time is an unused value

Now when you consider the default starting trip cost for a car under MT prefered is 1.95, that means in vanilla a sim will go only 3.65 time units by car (don't forget car also always has a .4 overhead tacked on it from another var) so making the MT trip only 4 seems reasonable. The bus even with its higher speed is still slower most of the time on the same network (I figured a 3 TU walk and 1 TU MT ride for average...if we only had to walk 2 TU then the bus wins). It would explain why the sims will not walk far at all in vanilla, especially for the bus. So it would seem an MT prefered sim expects a faster commute. And we also find out all this is incorrect as well, because the values are some kind of relative weight that I am still trying to determine a better explanation of how they work

Now car and fastest prefered will spend longer commuting but then with the penalty to walking for car prefered only being .1 I don't see how it makes much difference between the 2 types.
All things being equal (congestion, etc, etc) on the shorter MT prefered length trips I calculated ave beats rail, highway beats all. I have calculated the longer trips, but once again how many minutes you have to walk is what really makes or breaks MT.

Ok, so who gives a rats ass about the vanilla traffic sim? Where am I going with this? For one thing, my spreadsheet proves to me Maxis designed the traffic sim around the medium tile. Probably for performance reasons. No surprise there, just look at the maps that came with the game. Also it proves the traffic sim needed to be scaled out x4 In light of my recent commute time experiments drop this multiplier to x2 at a minimum to make a city of any size in a large tile viable. No surprise there either. Sims A and B are basicly that,the traffic sim scaled out x4. In most places. Some vars of course were not touched at all. Z on the other hand of course is scaled out even bigger, but virtually nothing is untouched.

Now I even tested for a while with no network preferences whatsoever for any wealth type (everyone was at fastest prefered) and it didn't seem to make much difference so it probably isn't all that important. Hell they might even be more broken vars for all I know. They aren't
Still, I wonder though about the ratios for these other values, if they should be scaled out as well. They shouldn't

A few other unrelated thoughts:

Speeds and commute time, forget reality. We've seen it doesn't apply. Proper ratios to each other are all that matter to give you the effect you want. It isn't like you can actually see the network speeds when playing the game anyway. I don't think the automata are affected :P The commute time graph...well...we all know how buggy it is anyway. So who really cares how long it says the commute is? The map scale really should have been 32M per tile instead of 16 and then things would be a lot more sensible (yeah, yeah, Mott basicly said this too....so far I regurgitate a lot of old information...although I like to think I am expounding on it at least a little  ::) ) Pretending that is what it is and scaling everything else accordingly would work well except for 1 small thing...the network tiles become ridiculously large. A 32M wide road? Preposterous. Even with the sidewalk included.

Network capacitys? Here is something that really drives home to me at least how much playstyle does effect and is effected by the simulator. What determines the needed network capacitys? How dense you pack your peeps...zone density, building sizes (CAM, etc) and here's a big one *drumroll* blocksize! How big you make your blocks, which then sets the ratio of zoned land/network segments is going to have a huge effect on what is a reasonable capacity. So one has to determine what one considers a reasonable population density in order to set network capacitys properly.

So speed and commute time are really questions of scale (and other things as Steve would be quick to point out, but not going to get into those right now but also not discounting their importance) and network cap is a question of playstyle & building capacitys.

Also I know Cathy and probably Brian and others were interested in my findings on station capacity; I haven't given up on that. I just don't have anything conclusive yet and I think these other issues might be related to it as well. So far I have gotten several stations to high 300s but nothing over 400% yet. Running the numbers through excel I am starting to see why I haven't saturated them yet, even after putting the rail capacitys back to double (but leaving the stations at default).

Being as I can't sleep I just keep finding things to add. Still working on that spreadsheet; for those hoping to get some flashy sheet that has it all broken down into formulas, I'm sorry to disappoint but it is already getting quite tedious. I'm getting it close enough to do a little more tuning for my next test (which will be soon). It will be a long time (if ever) before I have anything that could be considered a useful general reference.

One thing however just dawned on me. If we scale the trip starting cost up for car, then we can use higher transit switch entry costs and it should mitigate or even neutralize the penalty. Who cares you say? What's the point if we're just going to make it irrelevant? Because we should then be able to eliminate the pedestrians shortcutting through the TS without causing a detrimental effect on MT use.

b22rian

Quote from: ldog on November 12, 2009, 11:19:57 PM
So I got home and instead of playing I got to rereading the infamous te switches and you discussion.
I also got to thinking more about the effects of the prefered commute strategys, the trip starting cost modifiers and the max mt commute time.
Instead of doing more testing I wound up opening up Excel and trying to chart out some commutes.


Also I know Cathy and probably Brian and others were interested in my findings on station capacity; I haven't given up on that. I just don't have anything conclusive yet and I think these other issues might be related to it as well. So far I have gotten several stations to high 300s but nothing over 400% yet. Running the numbers through excel I am starting to see why I haven't saturated them yet, even after putting the rail capacitys back to double (but leaving the stations at default).



   hey Lenny,

really interesting post here..
Im very glad you had time to read this thread !  .. In my mind the best and most famous thread in
devotion history  :o
I think ive read through it like 6 times  ???
There is just a treasure drove of information in there for those who love traffic !

As to the station capacity maxes..
Yes, that is one of the more interseting topics to be sure.. I know Steve has stated that there is quite a
lot of variance amongst stations as to how far over capacity you have to go for each to render them
over their capacities for their daily commutes.. As i have 3 cities over a million , i really should help you
out with some testing in this areas.. i should be able to come up with something and of course I will tell
you about the results I get..

Hope you will continue to keep this thread active as you continue to enlighten all of us with your findings
and diligent testing you have faithfully done..

Thanks Again, Brian

ldog

#133
And now, the moment you've all been waiting for:
*drumroll*
It's picture time!
In my insanity I decided to start an MD for this, so for your viewing pleasure I present:
http://sc4devotion.com/forums/index.php?topic=9382.msg287921#msg287921

Y'all can skip the theatrics and head down to the 2nd post to see the latest test results.

@Brian
I guess it is required reading for the RTMT team eh.
I think I am catching up to you, that might have been my 6th time reading it as well.

Good stuff, good stuff.
:angrymore:
Reminds me I forgot to mention my MT station patch in the MD, I guess it isn't really relevant.
Speeds did not change from 1 sim to the next this time so I didn't need to redo any TE cost modifiers (yet...)
That will be one of the next things I work on though.

After I rework my spreadsheet  ::) in light of my recent findings.
It's ALL in the math baby.

catty

Quote from: ldog on November 12, 2009, 11:19:57 PM
...Also I know Cathy and probably Brian and others were interested in my findings on station capacity; I haven't given up on that...

:thumbsup:


Something else you might find interesting to read, I certainly did, was the last post Mott made

Quote from: mott on November 12, 2007, 01:58:04 PM
There's a clear explanation of RULs, written by the person who invented them, at:

http://www.mathematica-users.org/webMathematica/wiki/wiki.jsp?pageName=Road_rules

A quick read through this page will explain why it takes so long to get the new networks working properly. ;)


The link is still valid and is on "Utility and Road Network layout in SIMCITY 3000"

:)

I meant," said Ipslore bitterly, "what is there in this world that truly makes living worthwhile?" DEATH thought about it. "CATS," he said eventually, "CATS ARE NICE.

ldog

Quote from: catty on November 14, 2009, 12:34:45 PM
:thumbsup:


Something else you might find interesting to read, I certainly did, was the last post Mott made

The link is still valid and is on "Utility and Road Network layout in SIMCITY 3000"

:)



Fascinating. I have been confining myself to the traffic simulator for the moment, but I'm interested in all aspects of modding (except BATing, I have the artistic skills of a gnat, and that is probably insulting to gnats) Besides, it is always interesting to read when the devs say something that is actually technical instead of just marketing crap (or worse, backpedaling on why they can't fix something). You probably just saved someone in the NAM team from a bunch of stupid questions in the near future  :thumbsup:

Now I just need to find a bootleg of Mathematica  :o
(joking)

catty

Quote from: ldog on November 14, 2009, 01:30:49 PM
Fascinating. I have been confining myself to the traffic simulator for the moment...

One more link I promise and this is from the Transit Switch Entry Cost - Discussion Thread, if you haven't already read it, its worth taking a look especially at this post  :)

Quote from: CLR1SC4D on May 31, 2008, 10:37:41 PM
Sorry about the long time between responses.  Other than the normal reasons like being busy I have been further testing the commute engine to better understand it, and isolate parameters so that I am dealing with only one at a time.

@dedgren: Thank you.  CLR, Chris, or Christopher is fine.

While working with the commute engine exemplar to refine information on "Transit Switch Entry Cost" I have come across some interesting info and discrepancies between what has been posted and what the tests are showing.  If I am missing something, or the test information can be repeated/confirmed it would be good to know.  (If this would be better posted in new thread or in the commute engine tweaking thread started by Mott let me know and I can post there or it can be moved.)

All the information was gathered with experiments where each Sim had the same network travel length by setting up a city in some form as below (except for point 4):

R   C
R   C
B==B==B
R   I
R   I

Where:
"R" = Residential Zone
"C" = Commercial Zone
"I" = Industrial Zone
"B" = Bus station
"=" = Road
In some scenarios the bus stations were replaced with roads or changed in size.  Most scenarios the bus stations did not provide jobs.  Also some experiments had a road looping around the top.

As the bus station in the middle was increased in length, with all other parameters remaining the same, the Commute Time increased proportionally to the increase in number of bus station lot tiles and the "Transit Switch Entry Cost".

Below are the summarized results in no specific order.  Information related to transit switch entry cost has been bolded.

1.) Commute Time displayed can be calculated by:

Commute Time = INT{[ SUM(Tiles Transversed/Transit Speed) + SUM(Transit Station Tiles Transversed/Transit Switch Entry Cost)/0.96 ] * 24}

cont...
I meant," said Ipslore bitterly, "what is there in this world that truly makes living worthwhile?" DEATH thought about it. "CATS," he said eventually, "CATS ARE NICE.

z

Thank you, Cathy for finding yet another important thread.  Chris's experiments were excellent, and really helped show what was going on.  The final equation is where the formula TSEC = .96/speed was derived from.  And the 24 is the adjustment to the commute time graph scaling factor to make the graph correct in cities with no neighbor connection.

ldog

#138
Quote from: catty on November 14, 2009, 02:22:43 PM
One more link I promise and this is from the Transit Switch Entry Cost - Discussion Thread, if you haven't already read it, its worth taking a look especially at this post  :)


THAT is the post I knew I had read but was losing my mind trying to find again a few days ago.

Once again Cathy, thank you very muchly :D

Oh, by the way, if you happen to see my mind floating anywhere among some old post, can you kindly link it back to me  /wrrd%&

[sigh]
This is one of those prime examples of a thread that is confusing, especially when you are new and trying to read every thread you can find that might have some bearing on your interest. Some of the information is good, some is bad, some I still don't know about.

It also ended a year ago, yet it still isn't really clear if a consensus was reached or not.

While I'm not disputing the formula (for the moment, too tired to really evaluate it....although he is missing a parenthesis...typo I assume) I still can't figure out how it was determined. Maybe when I reread the post again in the morning with fresh eyes and mind.

ldog

#139
Since most of last night was spent reading and writing pms, I thought I would start the morning with a bit of testing.
I took my vanilla with .003 ph and added 0/1.3 to the beginning of the CvS. Surprisingly it had very little effect on commute times or abandonment (we do have quite a bit more since we switched from L to V, there is only so much the pathfinder can do; however this city was intended to have problems, so if they all went away then I would know I overshot my goals for the traffic sim)

Since this didn't produce any notable results I decided to throw an extra subway line from the CBD down to the LERS block in the bottom just to see what happened. I did actually get the downtown subway station (shown in yesterdays pics) to reach 515% but it bounced up and down several times. I didn't feel like letting it run long enough to fully stabilize and since adding this extra line changes our baseline for future tests I exited without saving.

I'm going to go back in and try out 600 max commute time (and the 1.3 CvS topend) and see what we get.

And the friggin thing ran for about 3 years on cheetah before it locked my computer up again.  :'(
Removing the no-cd hack lessened the frequency of this happening but did not eliminate it. As I said in another post, my computer while it is getting old (I tend build a new one about every 3 years) I can still pop in any new DX10 game and run it full bells and whistles, even heavy shadows and high antialiasing, without a stutter. Older DX9 games are another story. It wouldn't be so bad if the game just CTD but the locking up is making me nuts. Time to run it in a virtual machine methinks.

So anyways, as expected the commute time started creeping up and the abandonment started going away. Downtown sub station actually hit 545%. Of course I really don't think it is possible to determine if the pathfinder keeps trying to fling people at it or not once it fully saturates. Not in this city anyway. I probably could setup some specialized test scenario and look for abandonment that would probably be caused if that was the case, but for now there are too many other things to experiment with that are more important than finding out if rails could be set to be not affected by congestion and station cap used to control them. I did do some limited tests last week with it turned off that I didn't talk about; results were inconclusive.

I still prefer the bus the way it is. I also think the speed for it needs to be left higher than cars, my spreadsheets show (at least for vanilla and some derivitives I have concocted, I have started putting in values for other simulators as well, but haven't analyzed them closely enough) that the break even point from having to walk to the station actually eats up the speed advantage very easily. Testing also doesn't show absurdly high bus use, however I may place bus stations farther apart than what is considered normal.  Walking speed and max commute time also make a huge difference. So does TS entry cost and starting commute times. Then throw in CvS, all the other possible differences and the whole thing becomes extremely complicated.

Then take the fact that a bus holds how many people? 100? 150? 200? I really can't remember, I haven't rode a bus in many years. In NYC during peak times the buses get crammed as full as possible. When I lived in Seattle I also remember that most of the buses there were the accordian in the middle type...they were like double size buses. I've never seen one packed quite as full as a NYC bus but then if actual safety capacity was not routinely exceeded in NYC I would have to say it was a much higher capacity bus that Seattle had. So if 1 bus is like 100 cars, then how many buses do we expect on a given road during a commute trip in SC4? If we said like 20 buses, at 150 people per, that is 3000 people riding the bus through a given segment. While that would mean there are more bus passengers on the road than cars with low capacity settings, I don't think it is unrealistic, and I have yet to see my bus volume on a given segment go higher than car. Yeah, I said the R word again. Anyway, it is just something to think about, there is no correct or incorrect answer here. This I firmly believe is one of those things that comes down to personal preferences.

Oh, by the way, back to the car smogging ordinance again. I had been playing with the change, when I switched to vanilla for the above tests I didn't add it. Now being as all my industry is HT the only air pollution I have is from traffic. With the car smogging not effectively functioning as emission controls the way Maxis made it, my pollution has of course gone up. All the more reason I think it is very useful to make the change (provided it doesn't break anything). I still don't know what a reasonable reduction % is either, a quick search and skim through some materials on emissions hasn't produced the kind of quick and easy answer that we need. I see they talk mostly about reduction of 1 type of pollutant or another, and the figures seem to be anywhere from 20-80% (it also ranges by which standard you are under and what year that was of course as well). For our purposes all we need to know is the overall reduction in pollution it should grant. Also there needs to be a suitable penalty to enacting it, I don't feel that anything should be a get out of jail free card.

Let me reiterate, even though this is the same parameter Steve added to the clean air ordinance, his was a different reason that has nothing to do with the car smogging. He greatly increased the volume of traffic and the pollution scaled with it. However that pollution didn't scale properly and he felt the need to do something about it. Ideally we should be able to directly change the car pollution levels but I would assume this is something not accessible to us so his solution was to do the next best thing he could find, to add that effect to the clean air ord, something I would assume most people pass sooner or later in all their cities. I feel it was a very good solution.