Cost Index, ECON SPD, and MACH/times in BACARS flight plan

[torrence]

Questions for Hardy and probably Gary

I was flying a flight prepared with BACARS and PSX NET Dispatch suite yesterday (used Gary's manual to set it up).  Working great on test flight PHLI - KLAX.  When I started checking actual vs predicted times, speeds, winds etc as I passed waypoints, I noticed small but noticeable differences from flight plan.  I was using ECON SPD with the CI of 53 that flight plan uses resulting in  ~0.82M at FL370.  The flight plan predicts for each waypoint were higher, ~ 0.84M.  I could get these speeds, but had to increase CI to ~290-300. 

Just wondering what the factors affecting this are.  I admit to not paying too much attention to CI and optimum speeds and fuel usage in the past since I didn't have much insight into real world dispatch and planning until BACARS came along.  Are the differences in the airframe in Gary's model for BACARS  a factor - or winds?  Or just realistic variances from the plan?

Cheers,
Torrence

Move this over to the BACARS area if that's where it should go.
T

[Hardy Heinlin]

Hi Torrence,

the CI algorithm in PSX is based on maximum range speed (CI 0) and high speed cruise (CI 9999), with LRC speed at CI 200 with reference to DLH CF6 aircraft. There are certain interpolations between these three points. In real life there are also airline specific factors included. The exact formulas are not available, not even to dispatchers. There are dispatcher programs that output data tables for all kinds of parameter combinations, but they don't tell how the output is calculated. So what you get in PSX is an approximation based on published Boeing data (LRC at CI 200 etc.). There will always be minor differences to other products.


Cheers,

|-|ardy

[torrence]

Thanks, Hardy

I thought I remembered discussions of this being fairly complex - even Bulfer is pretty general and opaque on the topic, other than 'these are a few examples of how it works'.  So, as a practical matter, if I have a BACARS flight plan and schedule, I can adjust CI and/or SEL SPD once I'm at cruise altitude to meet the schedule as long as: 1. I'm cleared for 'speed at my discretion' and 2.  fuel reserves are OK.  After all, I don't have a real life bureaucracy pinching every penny based on today's Jet A prices.  And noting that oceanic speed restrictions, changing weather and other unknowns are why those reserves are there in the first place.

Cheers,
Torrence

[Hardy Heinlin]

... I can adjust CI and/or SEL SPD once I'm at cruise altitude to meet the schedule as long as: 1. I'm cleared for 'speed at my discretion' and 2.  fuel reserves are OK.

Yes, you can.


|-|ardy

[Jeroen Hoppenbrouwers]

Torrence, this CI thing is one of  those cases where a rather oversimplified formula tries to balance the cost of speed (more fuel) versus the cost of delay (more missed connections, hotels, bad press, angry pax, extra crew, ...). A nice ballpark, but like with any other plan, subject to change once wheels up. A bit like, you launch a vehicle to Saturn, and once it is off the launch pad, you turn off the lights and come back in a year or two to look at it, because the CI was calculated just right so it will be okay. Or flying IRS all the way and never making a position fix until you should be on final with runway in sight.

It sort of agrees with my view of project schedules  :-P

Hoppie

[Gary Oliver]

What would be really nice is to be able to provide an airline specific table for the CI calculation.

Based on the numbers of real BA cost index 53, the PSX CDU speeds are way too low because it's based on DLHs 0-9999 figures rather than BAs range which we think is 0-200

I could fudge the CI figure on the bacars flightplan output so we enter say 900 into the PSX FMC and you get roughly BA 53, however that's a major bodge.

Hardy,

What data would you need for a CI table, and if we provided you with one based on real aircraft and fixed based sim data would you be prepared to add an alternate CI calculation based on BA?

Cheers
Gary

[Hardy Heinlin]

Hoppie, that's true, the airline has more kinds of costs to consider (hotel etc.). Add to that the fuel cost difference at origin versus destination. I'm wondering whether dispatchers can input any fuel cost specific data in their programs. If so, the CI is even more complex. But I think such extra parameters (hotel, today's fuel price at destination) are not in the CI but thereafter: They lie in the dispatcher's decision as to what CI value is to be used today on that route. That's actually the point of working with a CI.

Gary, that's impossible. You would need zillions of 3D data records,
at least GW_tons * ALT_thousands * CI_tens = 200 * 30 * 1000 = 6,000,000 records.

Or even coarser:
1. Dimension: GW in 10 ton steps (200 t .. 400 t)
2. Dimension: ALT in 1000 ft  steps (FL150 .. FL450)
3. Dimension: CI in 10 steps (0 .. 1000)

20 * 30 * 100 = 6000 records

According to Boeing, CI 200 is LRC speed, not high speed. I can't imagine BA's max CI is 200?

I think 9999 is 744 specific, and the interpolation from 200 upwards is not linear. There's not much change above 1000.


Cheers,

|-|ardy

[torrence]

A bit like, you launch a vehicle to Saturn, and once it is off the launch pad, you turn off the lights and come back in a year or two to look at it, because the CI was calculated just right so it will be okay.

It sort of agrees with my view of project schedules  :-P

Hoppie

  - Got a real chuckle out of the Saturn mission analogy since that model is exactly what the administrators at NASA use with every new mission to justify an unrealistic cost.  And then they yell and scream about overruns and need for de-scoping when the ops model is reviewed and found to be unacceptably risky.

Cheers,
Torrence

[DougSnow]

I'm wondering whether dispatchers can input any fuel cost specific data in their programs. If so, the CI is even more complex.
|-|ardy

Most US airlines FPS use fuel cost differential to determine economic tankering, but only UA uses fuel price to determine cost index (maybe GTI and UPS does too). Their flight planning systems are more modern and can determine a true minimum cost CI.

Most other carriers with older FPS systems essentially fly LRC but in CI form, so if your CI never changes from day to day, you're essentially using an LRC CI.   

[Hardy Heinlin]

So is the fuel cost differential (and price if possible) a variable inside the FPS core in addition to the usual Mach planning parameters like gross weight, flight level, tailwind, ISA deviation for a given CI? Or is it excluded from the FPS core, i.e. is there an extra application that can recommend a certain CI for a given fuel cost differential (and price if possible), and that recommended CI is then entered into the FPS core? By "core" I mean the traditional main program.


|–|ardy

[DougSnow]

Nothing that is publicly available. PFPX doesnt have the necessary routine to calculate an optimum CI. It can do a CI plan if the aircraft data supports it (I enter CI150 and it can calculate the CI150 fuel data based on manufacturer aircraft performance data).  All commercial flight planning systems can do CI-based flight planning, but the ones that can calculate an Optimum CI based on the per unit fuel price (liters, gallons etc) in addition to the hourly and cycle costs are very few, Sabre FPM, Lido, and FSS (that I am aware of). Jeppesen might be able to but I havent seen it in work.

For any system with Optimum CI capability, for it to work and be accurate, the airline must enter (and regularly maintain) their fuel costs (which is done by the group that buys the fuel), as well as cycle and per hour costs.

[Will]

Doug,


What's the real-world variance in the CIs that dispatchers use? Meaning, will an airline typically use one CI for every flight? Or will they use a small set of CIs to choose from based on desired cost, like maybe 50, 100, 150, and 200? Or do they get very particular, with one day's flight using 199 and the next day's flight using 201?

[DougSnow]

Will

Thats down to the operator and their operational philosophy and the capabilities of their FPS. For example, at work we can do a VCI, variable cost index. However that is really a CI designed to drive an on-time arrival time (since for us achieving scheduled arrival time is paramount), not a true minimum cost CI (which is an optimum CI). Our FPS uses fuel costs to determine when to tanker, but it doesn't drive any decision on CI.

Airlines with a modern FPS simply tell their FPS to plan Optimum CI, and one day the number could be 64, the next day be 132 or whatever (within a range as defined by the FMC on the airplane and the fuel costs and differentials and so on), that is a true minimum cost flight plan. Their FPS can also have the costs for a delay baked into it, and the FPS will calculate a true optimum CI, with not just the hourly costs to operate the airplane, but if they are overblock inbound, the additional costs they could incur (misconnects), and drive the CI higher to avoid those costs. 

For example, our standard default CI for our 757/767 is CI80; for the 777F its CI150. But there is nothing scientific about those numbers, they are just LRC (99% of the range of Max Range Cruise), but in terms of CI. In other words, in the FMC substantiation documents from Boeing, that CI value is roughly parallel to the CI curves on the graph.

Bulfer explains what a true Optimum CI is really well, but the truth is very few airlines can really fly true optimum CI due to a lack of a truly modern flight planning system. For example, one US mAAjor is still on the same flight planning engine since the early 1980s.

Looking at the FMC documents for the 744, a CI of 250 is ROUGHLY parallel to LRC at the higher gross weights for a given altitude for any engine, whether PW, RR or GE.

[torrence]

According to Boeing, CI 200 is LRC speed, not high speed. I can't imagine BA's max CI is 200?

I think 9999 is 744 specific, and the interpolation from 200 upwards is not linear. There's not much change above 1000.


Cheers,

|-|ardy

[/quote]

I finally went back to my latest Big Boeing Bulfer, and CI is indeed discussed in some detail on pp 121-123, with the examples on p 123 having these comments:  "Cost Index of zero = Max Specific Range, a higher fuel cost will produce a lower Cost Index", "Cost Index of 9999 = Minimum Time (non-PIP =999)", and (PIP/747/777=9999).

This is all consistent with comments in this thread but I'm not sure exactly from Gary's comments what BAW's FMC implementation is: just not using CI's above 999 or are they using the non-PIP FMC?  The definitions in Bulfer are: "Valid entries on the non-PIP 757/767 is 0-999; on the PIP 757/767/747-400 it is 0-9999", which implies there is no non-PIP 747-400 version, but company specific implementations are occasionally missed since the authors depend on Wiki-like editing by line pilots for these type of updates.

For what it's worth this link shows a quick plot of ECON SPD VS CI for early cruise at FL370 for my PSX flight noted in my first post above - it's non linear as you note.

https://www.dropbox.com/s/wwgo9ngmga939bx/ECON%20SPD%20VS%20CI%20chart%20TVJ.pdf?dl=0

Cheers,
Torrence

[Britjet]

One suggestion I would make if you want to see how and when the transition from IAS to MACH occurs is to watch the individual readout on a 747 using Flightradar24, or similar programme.
Peter.