Perth out of control...

[cac_sabre]

Another Runway 06 Only Day

This Tuesday morning saw Perth Airport operations on RWY 06 due to strong easterlies and excessive crosswind on RWY 03 again.

Based on the experience of the day that caused this thread to start, Perth Tower reduced the number of available slots for departure from 32 to 26 per hour. (For "RWY 06 only" operations)

This meant that the whole morning departure "push" lasted about 20 mins longer than usual and many aircraft were significantly delayed passed their preferred taxy time.

However times for commencement of taxy to departure averaged 10 mins with a 17 min peak.

Instead of having 16 aeroplanes at the holding point and taxiways burning fuel going nowhere, we had a maximum of 8, which was manageable but provided enough airframes to maintain a steady departure flow.

Tomorrow morning looks like another 06 only day, so we will be using the "formula" again to test its validity.

[westausatc]

Oh goody! Lots more fun then!

And thanks to you guys in the TCU for not making us give the 'double' spacing between arrivals! It's a huge help!

[Continental-520]

Yah,

It all seemed to work a bit better on Tuesday. Good work to the crew organising that!

Was the amount of traffic comparable to a few Tuesdays ago when we waited 42mins?


520

[cac_sabre]

About the same numbers

This Tuesday morning's push numbered 58, woulda been more but 2-3 cancelled. "Black" Tuesday was 59 from memory.

Please be aware however that its not a precise "science" every day is different, route mix, aircraft type mix and the order they "present" all are a factor in how smoothly or otherwise it goes.

Other things that help are:

• advise as early as possible if you cant make your slot... we will try and assign it to another aircraft
• if you haven't got the slot you want, listen out after being assigned your clearance and slot and we will call you if an earlier one becomes available due a u/s whatever

The next step that could add an estimated up to extra 10% efficiency would be to get airlines to avoid planning several flights in a "row" that will fly on the same SID. Eg this morning we had 6 Spudo departures in a row, this is a loss of 6 mins (3 slots) in the hour. Haven't had any luck there, the vagaries of airline ops / customer demands make this difficult to achieve, plus it would require coordination and cooperation between competing operators to be most effective.

[Jetpipe2]

Similar SIDS

CAC_sabre

If sequencing 6 aircraft is difficult on the same SID why not give then the Radar Deprture and use that to fix the problem?

Whilst they are all heading in roughly the same direction, they aren't all going to the same spot, so once clear of the crossing tracks and with GPS seperation standards it would be direct to wherever?

After all does ATC asks QF, VB etc not to fly SY - ML at the same time?

[Quokka]

Jetpipe2,

The idea of using RADAR Departures was suggested in 2006... and the answer was no.

[cac_sabre]

Yo Jetpipe

If sequencing 6 aircraft is difficult on the same SID why not give then the Radar Departure and use that to fix the problem?

SIDs are established for a number of reasons including noise abatement and traffic management. In less complicated times approach / departure controllers would shot-gun the morning departure burst through every point of the compass using radar departures, it was efficient and dare I say fun!. As traffic numbers increased, larger aircraft were used and the community started to object to the aircraft "noise nuisance" , Air Traffic Management practices had to change.

I expect this is the primary reason for the Radar Departure not being used more often, maybe the TCU guys could provide more insight into the issue.

[Led Zep]

Perth Ground, morning, 08/07/08

Ok, what was the spanner thrown in the works this morning? Jeeze I'm glad I'm not in air traffic control, hats off to the lady who was on 121.7 this morning...better you than me!
We ended up having a laugh about it, no harm done, etc etc and ask only as we had never heard it so congested, it was bad even for a Tuesday.

[Starts with P]

A girl in Perth Tower.... Probably was "Spanna"

[Skystar320]

on all indication is that she is something to 'gawk' at

[lifestyler]

Whats your "indication" skystar320 ? Do you know the girl ? I do she's a top chick, a great controller and a friend of mine...

Give the girl more respect than that

[Ejector]

But Australia is the lucky country. 3rd world service with 1st word cost.

enjoy it boys and girls

[general disaray]

speaking of which whatever happened to spanna haven't heard her on the radio in a while she quit or get shipped off somewhere else? I say bring her back we need something to brighten up our mornings.......

[ROH111]

On holidays...

Last I knew, Thailand and other random places - hunting!

[andrewr]

In creating the “system” it was calculated that given the enroute restrictions, the maximum achievable departure rate was 30 per hour. We created 32 slots per hour adding the extra two to make sure the ATC system was always under “pressure”

This sounds like a queue designed by an accountant not a mathematician. Anyone who has even a little queuing theory will see that this creates a queue where the wait time increases towards infinity until the arrival (aircraft ready to depart) rate drops below 30. Close to 30 the wait time will still be very long as you have no capacity to catch up after any delay.

As I was reading this I was guessing that maybe 25/hour would be about the limit, if things were well organised - then read further down and found that was indeed the case.

Rather than experimenting, why not get someone who knows queuing theory to do the maths? It is a pretty well studied and understood area. An expert should be able to look at the stats and give you charts showing slots vs. utilization vs. average wait time. They can even show you the differences in wait time with designated slots, clustered arrivals or random arrivals.

[Gundog01]

Sounds like a doctors surgery taking bookings. 10 min per patient = 7 patients per hour?? And i get annoyed waiting for 7 mile ILS traffic to land at TVL. Cant say i miss Perth.

[cac_sabre]

I thought infinity was a big number

If we allocated 25 slots per hour on most mornings we would unnecessarily delay between 3 and 6 flights.
When aeroplanes are servicable, self loading freight has turned up on time, aircraft taxy on time and in a good order, we can punch out 32 per hour. We do reduce the slots available when we are reduced to single runway ops or when weather precludes the use of tower visual separation for departures.
I am enough of a mathematician to be fairly confident that say by having 2 too many slots allocated per hour that in the 2 hour long departure rush, the overflow in the queue will not even come close to infinity... more like 4
The same technique is used in Sydney for the management of CTMS based arrival slots eg in a particular runway configuration and cloud base the achievable acceptance rate is "X" but X plus 2 slots are made available, once again to ensure a slot does not go unfilled.

Anyone have a better management strategy I'd be glad to hear it

[Monopole]

Anyone have a better management strategy I'd be glad to hear it

CTAF(R)

[andrewr]

I thought infinity was a big number

I did say until the rate drops below 30/hour. The point is that delays grow and multiply until there is slack in the system.

I am enough of a mathematician to be fairly confident that say by having 2 too many slots allocated per hour that in the 2 hour long departure rush, the overflow in the queue will not even come close to infinity... more like 4

By your maths, after the first hour, the maximum wait is 4 minutes, after 2 hours the maximum wait is 8 minutes. If you are seeing delays longer than that, maybe the maths is more complicated?

Queue maths is complicated enough that I can't actually remember how to do the equations. I do remember the principles e.g. the general relationship between arrival rate, service rate and wait time. It can also account for the fact that aeroplanes aren't always serviceable, self loading freight don't always turn up on time, and aircraft don't always taxy on time and in good order.

If we allocated 25 slots per hour on most mornings we would unnecessarily delay between 3 and 6 flights.

Not necessarily. You might find you can have exactly the same number of flights depart, but with reduced average waiting time. A queue of 10 doesn't improve the departure rate over a queue of 3. (Of course calculations will probably give a different number, 25 just looked an "about right" number to me.)

[flog]

Google "Erlang".

The principles were developed for call traffic in telecommunications but the same principles apply to ATC - i.e. volume per hour, wait expectations, blocking (clearance not available), etc.

Service levels (i.e. X% of target within Y seconds/minutes) do not fail in a linear fashion, they fall off a cliff. For example you could be happily meeting a service level and then one little delay will cause a complete cluster fluck as it pushes the math over the edge.

Actually thinking about this - does anyone have traffic arrival data for ATC at Perth (not just volumes but actual times and the time from contact to clear of rwy if it exists). I'm going to run a little simulation and apply some of the optimisation algorithims we use here and see what it tells us.